INFORMATION SOURCES OF AGRICULTURAL EXTENSION SPECIALISTS IN INDONESIA
by
PAUL GEOFFREY WARING MUNDY
A thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Mass Communications)
at the UNIVERSITY OF WISCONSIN-MADISON 1992
ii
INFORMATION SOURCES OF AGRICULTURAL EXTENSION SPECIALISTS IN INDONESIA Paul Geoffrey Waring Mundy Under the supervision of Professor John Fett ABSTRACT This study aimed to discover (1) where Indonesian agricultural extension subjectmatter specialists (SMSs) obtain technical information, (2) why they use certain sources, (3) how important research-extension linkages are relative to other problems facing extension, (4) how quickly research-derived technologies spread among SMSs, and (5) what the SMSs' unmet information needs are. I conducted two nationwide mail surveys of a stratified random sample of livestock and food-crops SMSs working at province and district-level offices and Agricultural Information Centers. The first survey (response rate = 72%, valid n = 280) focused on the first two questions. The second (response rate = 75%, valid n = 165) focused on the last three. I also conducted personal interviews with 101 SMSs, local extension officials and other persons connected with the Indonesian research and extension systems. Respondents obtained most information from other nearby individuals: field agents, other SMSs, farmers, and superiors. The agricultural press and extension publications also were major sources. Research publications and direct contacts with researchers were relatively unimportant as sources. Multiple regression revealed that SMSs tended to obtain information from a source if (1) the source was familiar, (2) it was perceived to be locally relevant, (3) it was close by and accessible, and (4) it was timely. There was some evidence that SMSs would use a source if (5) it was easy to understand and use, (6) it was credible and complete, and (7) the SMSs thought it their job to use the source. The source type was important in determining which factors influenced the amount of information an SMS obtained from the source. Research-extension linkages are an important problem in Indonesia -- second only to mobility among the problems listed. Information flows are slow: it took about two years for news about a new technology to reach 50% of the SMSs, and about six years for the news to reach 80%. The main unmet information needs were in regional planning, farm systems analysis, post-harvest processing of both crops and livestock, horticultural crops, and livestock feedstuffs.
iii
iv
ACKNOWLEDGMENTS This study was carried out under the auspices of the Small Ruminant-Collaborative Research Support Program (SR-CRSP) Sociology Project, at the Research Institute for Animal Production (RIAP) of the Central Research Institute for Animal Sciences (CRIAS), Bogor, Indonesia. It was partly funded by a grant from the Department of Journalism and Mass Communication, Iowa State University. Many people contributed to this study, and I'm bound to have omitted someone. My apologies to all those whom I have so slighted. Needless to say, any errors remain mine alone. Professor Eric Abbott, my major professor for my Master's program in Journalism and Mass Communication at Iowa State University, has been a guide and inspiration throughout my graduate career. He has been a wonderful support for me and my family, both at Iowa State and since we left Ames. His provision of funding through ISU made this study possible. I owe him much. I have enjoyed an extraordinarily satisfying and educational stay at the University of Wisconsin-Madison. This has been in large part due to the five members of my Ph.D. committee. Professor John Fett's light but deft hand has guided me through my doctoral career. His reactions to my research proposal were prescient of the conceptual and logistical problems this study would encounter. The speed with which he read the drafts of this dissertation was exceeded only by the thoroughness, aptness and humor of his comments. I owe to Professors J Lin Compton and Russell Middleton an appreciation of farmers' situations and of the urgent need for formal institutions to serve these better. Lin has been a leader in pushing for closer research-extension ties and for greater accountability and sensitivity of both research and extension to their rural clients. Russ's concern for the rural poor and the possibilities of helping them organize to improve their lot has been a major influence. Professor Marion Brown has given me an appreciation of the range of problems facing the poor in the developing world, and that lack of information may not be the main one. I especially appreciate the extraordinary lengths he went to in navigating university bureaucracies on my behalf. Professor Albert Gunther's guidance through the conceptualization process was a valuable experience: his fingerprints can be seen most clearly in Chapter 5. While he was not a member of my committee, Professor Richard Powers of UWMadison was an unfailingly good-humored source of statistical guidance. He saved me many hours of sweat.
v
This study would not have been possible without the help of Dr P. Sitorus, director of CRIAS; Dr Sabrani, director of RIAP; Dr Luis Iniguez, then SR-CRSP liaison officer at CRIAS and RIAP; Dr Mike Nolan of the University of Missouri-Columbia, coordinator of the SR-CRSP Sociology Project; and the multifarious contacts and generous assistance of Dr Constance McCorkle, formerly of the SR-CRSP. Drh Marwan Rangkuti, of CRIAS's Research Communication Department, was extremely helpful and provided many insightful comments. Sdri Josefine Wie, administrative assistant with the SR-CRSP, solved many administrative problems with great efficiency. Sdr Hadi Budiman of CRIAS and Sdr Duaman of the Sungei Putih Livestock Research Station accompanied me on my visits to extension institutions in West Java and North Sumatra, assisted in data collection, and provided many useful ideas. Dr Sybille Scholz of the SR-CRSP kindly hosted me and provided me with a vehicle with which to tour North Sumatra. Sdr Inu Gandana Ismail of the SWAMPS-II research project provided similar much appreciated assistance in South Sumatra. I spent much of my time in Bogor using the facilities of my former institutional home, the Central Research Institute for Food Crops (CRIFC). I would like to thank Dr Ibrahim Manwan, Director of CRIFC, for allowing me to use the institute facilities. Drs Mahyuddin Syam, head of CRIFC's Research Communication Department, has been an invaluable friend and guide since I first visited Indonesia. His concern for the practical relevance of research was a major factor shaping this study. I worked closely together with the communication staff of both CRIFC and CRIAS. I greatly appreciated their assistance and friendship during the research. In particular, Ir Adi Widjono made major contributions to the study design. I would also like to acknowledge the assistance and input of Dr Amri Jahi of Bogor Agricultural University and Ir Sulastuti Sophia of the Center for Agricultural Library and Research Communication. Among the many people in the Indonesian extension system who contributed to this study, I would like to single out Ir Erru Getarawan, head of the Ciawi Agricultural Information Center, for special thanks. She spent many hours with me, patiently explaining how the extension system works, suggesting topics for study, and providing several introductions to key ministry personnel in Jakarta. Her colleagues at AIC Ciawi provided numerous comments and much assistance in collecting the data on AIC media outputs and the addresses of respondents. Unfortunately rules of confidentiality mean I cannot reveal the names of the many people I interviewed for this study, in Bogor and Jakarta, and around the provinces of West Java and North and South Sumatra. I also cannot reveal the names of the questionnaire respondents, many of whom added well thought-out comments to the questionnaire. Nevertheless, I thank them. During our stay in Bogor, my family and I stayed with Ir Iis Arifiantini and Ir Prasodjo Waluyo. Isye, Pras, and their children gave generously of their home and their privacy for eight months. We cannot hope to repay their kindness and hospitality.
vi
Finally, I would like to thank my wife, Dr Evelyn Mathias-Mundy, and our son, Oliver. Your love, patience, humor, and hard work have kept me sane. I am fortunate indeed to have two such wonderful people so close. It is to you that I dedicate this volume. Kepada semua teman yang telah membantu saya selama kami berada di Indonesia, saya ucapkan banyak terima kasih. Mudah-mudahan tulisan ini berguna dalam usaha kita melayani para masyarakat pedesaan.
vii
viii
CONTENTS Abstract
iii
Acknowledgments
v
Contents
ix
List of Tables
xiii
List of Figures
xvii
1
Introduction
1
Why study research-extension linkages? Why Indonesia? Purpose of this study Study outputs Assumptions Contents of this volume 2
Research-Extension Linkages: Approaches and Concepts
5
Introduction Agricultural knowledge systems Approaches to studying information flows Summary 3
Agricultural Extension in Indonesia Introduction Indonesia: The setting Agriculture in Indonesia The Ministry of Agriculture Extension
ix
20
Effectiveness of extension The mass media Summary 4
Agricultural Research in Indonesia
52
Introduction AARD and agricultural research Technology recommendations The literature on linkages AARD research communication departments AARD audiences AARD communication activities Constraints to research communication Summary 5
Information Flows and Their Causes
78
Introduction Concepts Information flow Measuring information flows Influences on information flows Source-receiver relationships Summary 6
Methods
113
Timing and location Interviews Mail survey Questionnaire design and pretesting Information Sources questionnaire
x
Publications questionnaire Sampling Data cleaning and additional variables Initial analysis Influences on information sources 7
Specialists' Extension Activities
135
Introduction Presentation of results Response rate Personal characteristics Work time Field vs. central orientation Extension activities Problems faced Problems in obtaining information 8
Specialists' Information Sources and Needs
169
Introduction Information sources Publication readership and usefulness Information diffusion rates Information needs 9
Factors Influencing Information Flow Introduction Index construction Information Flow Familiarity Proximity
xi
194
Structure Capacity Openness Reward Energy Synergy/Timeliness 10
Combined Effect of Factors on Information Flow
252
Introduction Correlations among indices Comparison among source types and institutions Effect of indices on Information Flow Summary 11
Toward Improved Links
271
Introduction Summary of findings Determinants of information flow Source-specific strategies Implications of the FP SCORET model Conclusion References
298
Appendices
320
1
Information Sources questionnaire
320
2
English translation of Information Sources questionnaire
325
3
Publications questionnaire
334
4
English translation of Publications questionnaire
339
5
Frequencies of variables from Information Sources questionnaire
348
xii
6
Frequencies of variables from Publications questionnaire
xiii
395
xiv
List of Tables 2.1
References on agricultural extension and research-extension linkages worldwide
3.1
Food crops and livestock production in Indonesia, 1966 and 1991
3.2
References on agricultural extension in Indonesia
3.3 Numbers of media titles produced by the 28 Agricultural Information Centers, 1984-1990 3.4 Numbers and prices of agricultural books stocked by bookstores in Bogor, West Java, 1991 4.1
References on the Indonesian agricultural research system
4.2
References on research-extension linkages in Indonesia
6.1 Questions in the Information Sources questionnaire used to measure Information Flow 125 6.2 Items in the Information Sources questionnaire and the FP SCORES variables they were intended to measure 6.3
Numbers of specialists sent questionnaires
7.1 Example of presentation of statistically significant differences among subgroups of respondents as identified by analysis of variance 7.2 Sampling and response rates from Information Sources and Publications questionnaires 7.3
Respondents' institutional affiliation
7.4
Specialization of respondents
7.5
Education of respondents by institutional affiliation
7.6
Gender of respondents by institutional affiliation
7.7
Urban/rural background of respondents by institutional affiliation
7.8
Farm background of respondents by institutional affiliation
7.9
Other work of respondents by institutional affiliation
xv
7.10
Wealth, as reflected by number of respondents who own selected items
7.11
Time respondents spent per week on various activities, by institutional affiliation
7.12 Percentage of work time respondents spent solving field problems and providing information from "above" 7.13
Respondents' opinion about central recommendations
7.14 Number of times respondents engaged in extension and information seeking activities, by institutional affiliation 7.15
General problems faced by extension
7.16
Problems respondents face in obtaining information
8.1 Amount of information respondents received from 24 sources, by institutional affiliation 8.2 Top five information sources of respondents at provincial and district offices and Agricultural Information Centers 8.3 Number of respondents naming sources among their five preferred information sources, by institutional affiliation 8.4 Comparison of desirability of 24 sources and actual information flows to respondents from those sources 8.5 Frequency that respondents read selected agricultural publications, and ratings for publication usefulness 8.6 Percentages of respondents specializing in food crops who obtained most information from various sources 8.7 Percentages of respondents specializing in livestock who obtained most information from various sources 8.8
Percentages of respondents who needed information on food crops-related topics
8.9
Percentages of respondents who needed information on livestock-related topics
8.10
Percentages of respondents who needed information on other topics
9.1
Correlations among variables in the FP SCORES indices
9.2 Number of variables, mean correlations, and Cronbach's α of original and revised forms of indices 9.3
Factor loadings of variables intended to measure the FP SCORES concepts
9.4
Loadings of items on factors identified through factor analysis
xvi
9.5 Simple Pearson's correlation coefficients between items in the Information Flow index 9.6
Information Flow scores by source type and institution
9.7
Familiarity scores by source type and institution
9.8 Standardized simple regression coefficients (beta) for Information Flow against Familiarity, by source type and institution 9.9 Maximum and minimum ns in calculating regression coefficients for Information Flow against FP SCORES variables 9.10
Proximity scores by source type and institution
9.11 Standardized simple regression coefficients (beta) for Information Flow against Proximity, by source type and institution 9.12
Structure scores by source type and institution
9.13 Standardized simple regression coefficients (beta) for Information Flow against Structure, by source type and institution 9.14
Capacity scores by source type and institution
9.15 Standardized simple regression coefficients (beta) for Information Flow against Capacity, by source type and institution 9.16
Openness scores by source type and institution
9.17 Standardized simple regression coefficients (beta) for Information Flow against Openness, by source type and institution 9.18
Reward scores by source type and institution
9.19 Standardized simple regression coefficients (beta) for Information Flow against Reward, by source type and institution 9.20
Energy scores by source type and institution
9.21 Standardized simple regression coefficients (beta) for Information Flow against Energy, by source type and institution 9.22
Timeliness scores by source type and institution
9.23 Standardized simple regression coefficients (beta) for Information Flow against Timeliness, by source type and institution 9.24 Standardized simple regression coefficients (beta) for amount of information obtained against Timeliness, by source type and institution 10.1
Correlations among FP SCORET indices
10.2
Information Flow and FP SCORET index scores by source type
10.3
Information Flow and FP SCORET index scores by institution type
xvii
10.4 Standardized regression coefficients (betas) for predictors of Information Flow in simple and multiple regression 10.5 Regression of Information Flow on source type, institution, and FP SCORET indices (by source types) 10.6 Regression of Information Flow on source type, institution, and FP SCORET indices (by institution)
xviii
LIST OF FIGURES 3.1
Provinces of Indonesia
3.2 Organizational structure of the Indonesian Ministry of Agriculture (simplified), showing locations of SMSs 3.3
Organization of agricultural extension in Indonesia
3.4 Booklet (brosur) produced by Medan Agricultural Information Center, North Sumatra 3.5 Liptan technical fact sheet, produced by Palembang Agricultural Information Center, South Sumatra 3.63.6 Buletin Informasi Pertanian, semiannual magazine produced by Lembang Agricultural Information Center, West Java 3.7
Sinar Tani, Indonesia's semiweekly agricultural newspaper
3.8
Trubus, a leading monthly horticultural magazine
4.1
Organizational structure of the Agency for Agricultural Research and Development
4.2 Simplified information flows from research to extension within the Ministry of Agriculture for a hypothetical food-crop pest-control technology 5.1 Information flows as characteristics of the relationship between a receiver and various sources 5.2 Information flows viewed as dependent on other relationships between a receiver and a source 5.3
Factors affecting information flow
8.1
Percentage of respondents obtaining information on selected technologies, by year
9.1
Regression of Information Flow against Familiarity, by source type
9.2
Regression of Information Flow against Proximity, by source type
9.3
Regression of Information Flow against Structure, by source type
9.4
Regression of Information Flow against Capacity, by source type
xix
9.5
Regression of Information Flow against Openness, by source type
9.6
Regression of Information Flow against Reward, by source type
9.7
Regression of Information Flow against Energy, by source type
9.8
Regression of Information Flow against Timeliness, by source type
xx
1
CHAPTER 1 INTRODUCTION Why study research-extension linkages? Agricultural technologies in the developing world must change continuously to keep pace with rising populations and rapidly changing social, economic, and environmental conditions. These technologies are generated by research institutes, universities, private companies, and farmers themselves. One of the tasks of government agricultural extension services is to communicate them among the farmer audience. Agricultural extension is expected to draw on public-sector research institutions for much of what it communicates to farmers. But poor linkages between research and extension are often blamed for the slow implementation of new agricultural technologies by farmers in the developing world (Kaimowitz 1990:xi). Despite increasing recognition of this problem, much of the evidence remains anecdotal. Relatively few quantitative studies have documented or sought reasons for it (Seegers and Kaimowitz 1989:i). Indonesia is a case in point. The country's approximately 2000 graduate extension subject-matter specialists (SMSs) have the task of bridging the gap between the Ministry of Agriculture's 30-plus research institutes and 29,000 field-level extensionists (Syam and Mundy, in press). But linkages between the SMSs and research are weak (Syam 1990:25, Padmanagara 1985:141). Most SMSs have few direct contacts with researchers (Hussein 1986:422), and readership of publications among extension specialists is low (Sophia 1988:52). If the specialists do not obtain information directly from research, where do they get it? And why do they use these sources? Very little research on these questions has been performed in Indonesia, or indeed elsewhere. Why Indonesia? I conducted this study in Indonesia for several reasons: • Indonesia's size, the diversity of its land and people, and its geographical fragmentation into thousands of islands, pose problems for both agricultural research and extension and provide challenges to smooth linkages between them. • The Indonesian research and extension systems are among the largest in the developing world. And many of their characteristics are typical of other Third-World countries. Findings from this study thus should have wide potential applicability within Indonesia and elsewhere.
2 • Despite their importance, the Indonesian research and extension systems remain under-researched. Few formal studies have been made of their activities, performance, and interaction. • Having previously worked for six years in the communication department of an Indonesian research institute, I knew that linkages were a problem. I hoped to be able to suggest pragmatic solutions to the problem. • This experience had also given me the language skills and institutional background necessary to conduct research there. Purpose of this study This study aimed to answer two main questions: •
Where do Indonesian extension specialists obtain technical information?
•
Why do they use these sources? Additional questions included:
• How important are research-extension linkages relative to other problems faced by extensionists in Indonesia? •
How quickly do research-derived technologies spread among SMSs?
•
What are SMSs' unmet information needs?
Answers to these questions should help policy makers design ways to improve research-extension linkages and thereby the application of research findings by Indonesia's farmers. Study outputs Expected outputs of the study included: • Data on the characteristics, communication behavior, and information needs of the primary audience of Indonesia's research institutions. •
Suggestions for improving research-extension linkages.
• A set of approaches for studying extensionists' information sources and researchextension linkages. •
A more enlightened perspective on agricultural knowledge systems as a whole.
3
Assumptions This study rests on some untested assumptions: • The Indonesian research system produces a flow of information that is potentially relevant and appropriate for extension and farmers. In many countries this is not the case: research is often criticized as irrelevant. I make this assumption because Indonesian farmers have rapidly adopted some research-derived technologies (especially for rice cultivation). I discuss the differences between rice and other commodities in Chapter 4. • Subject-matter specialists are motivated to seek and disseminate information, but face constraints in doing so. I base this assumption on the interviews I conducted as part of the study. • Within these constraints, subject-matter specialists are relatively free to choose among information sources available to them. While they may be mandated by their superiors to purvey certain types of information, they themselves can determine what other information to disseminate. This assumption is also based on the interviews and survey results. Contents of this volume This study reports the findings of two surveys of SMSs in Indonesia and numerous semi-structured interviews with SMSs and others involved in the research and extension system. Research-extension linkages form part of a broader knowledge system. Chapter 2 describes some approaches to studying agricultural knowledge systems and information flows within them. Chapters 3 and 4 provide the background for the study. Chapter 3 describes the agricultural extension system in Indonesia, while Chapter 4 discusses the research system and its linkages with extension. Chapter 5 defines the key concepts used in the study and proposes a model containing eight variables as predictors of information flow between a source (such as a research publication) and a receiver (such as an extensionist). Chapter 6 describes the methods used for the interviews and surveys, and outlines the analyses performed on the data. Chapters 7 to 10 give results of the research. Chapter 7 describes the characteristics of the SMSs who responded to the surveys. It goes on to analyze their extension activities and contacts with research. It concludes by addressing the question, "How important are research-extension linkages relative to other problems extensionist face?" This necessitated comparing the importance of information scarcity with other constraints that extensionists face: mobility, lack of teaching aids, and so forth.
4 Chapter 8 focuses on SMSs' information sources. It contains three parts, each corresponding to one of the questions listed above. • Where do SMSs obtain technical information? Answering this question required listing possible sources and asking SMSs to indicate how much they used each. I discuss which sources respondents used heavily and which they did not, and compare these with source they would like to use under ideal conditions. I also look at the readership and usefulness of several research and extension publications. • How quickly do technologies spread among SMSs? I report findings based on questions about six selected food crops and six livestock technologies. • What are SMSs' unmet information needs? The questionnaire listed more than 50 topics SMSs may need information on. Findings are reported here. Chapters 9 and 10 are devoted to answering the question, "Why do SMSs use certain sources?" They analyze the model proposed in Chapter 5, first looking at each variable individually (in Chapter 9), and then combining them into an overall model (in Chapter 10). Because of logistical and space limitations, it was possible to study only four of the many possible information sources. Chapter 11 summarizes the study findings and discusses their implications for improving research-extension linkages in Indonesia and elsewhere. The chapter concludes with suggestions for improving these ties. Appendices 1 to 4 contain copies of the questionnaires, together with their English translations. Appendices 5 and 6 present frequency tabulations for responses to questions in the two questionnaires.
5
CHAPTER 2 RESEARCH-EXTENSION LINKAGES : APPROACHES AND CONCEPTS Introduction The exchange of information is vital to the existence of societies, organizations, and other social groups. Given its importance, it is hardly surprising that this topic has received considerable research attention. It is of particular importance in agriculture, given the large numbers of people involved in this activity and its importance to society and national economies. The literature on information exchange is divided into several strands, each emphasizing a different set of theoretical relationships or practical problems. This chapter discusses these approaches, any of which could be used to study research-extension linkages in Indonesia. I draw from them in developing the approach used in this study. I review the literature on research-extension linkages in Indonesia in Chapter 4. Agricultural knowledge systems Agricultural knowledge systems are complex and diverse. They involve large numbers of people and organizations involved in generating, disseminating and using information, related to multifarious tasks in crop and livestock production, input supply, produce processing and marketing, consumption, and regulation. It should be noted in passing that agriculture is by no means unique in possessing a "knowledge system." Any field of human endeavor can be viewed in a similar way, though the agricultural arena has received the most attention. Two other areas that have attracted attention are education (e.g., Havelock 1986a:83) and industry (e.g., Silveira 1985, Rosenberg 1967). Another obvious candidate is health, though I am not familiar with the literature in this field. Information in the knowledge system The flow of information is vital to the smooth functioning of such systems. Without information about likely markets and prices, the producer cannot make decisions about what crops to grow and when to buy and sell. Without information about the location and size of a crop or the quality of produce, the processor cannot plan how much finished product to supply to consumers. In order to compete with each other and to maintain production in a sometimes hostile environment, producers need information about new technologies, most often developed by researchers at universities, research institutes, and private companies.
6 It is not possible, or even desirable, for individuals in one part of the knowledge system to maintain direct contact with all others in the system who may need information they can provide. There are many reasons for this (see Havelock's [1986b:213] example of technology dissemination). Here I focus on four that are particularly relevant to this study. They are arithmetic, distance, translation, and adaptation. Arithmetic Imagine a farmer who has found a way of controlling a crop pest, or a plant pathologist who has developed the same technique. Information about this technique might benefit large numbers of people (farmers whose crops are attacked by the pest). The inventor cannot possibly provide information to all the farmers personally. In many countries, farmers who could benefit from the technology number in the millions, while only a handful of pathologists are at work. Some kind of intermediary -- an extension service, or a mass medium -- is necessary to duplicate information about the pest control method and make it available to the audience. Distance Farmers and other users of information are scattered geographically. The inventor of a technology is located in one place. This means the information about the technology must be distributed (for instance, via a publication, the mass media, or training courses) so it reaches its audience. Translation In order for an audience member to understand the information, it must be translated into terms he or she is familiar with. This may involve rewriting a research report in non-technical language, combining it with other information to show how it is relevant and can be used, and presenting it in a readily understood format. Adaptation Audience members do not experience the same situations. It may be necessary to adapt the information to suit local social and agroecological conditions. It may even be necessary to tailor the information to suit individual farms. Usually this is done by the farmer him- or herself. But often the farmer lacks sufficient expertise to solve a problem, so an outsider's advice is needed. Given these constraints, it is clearly impossible for a technology inventor to perform the dissemination function except on the smallest scale. To do so would mean abandoning the invention of new technologies, since there would be no time left for anything except dissemination. The mass media can help overcome the barriers of arithmetic, distance and translation, but they are not suited to adapting information to suit individual needs. Hence the need for specialized organizations to handle these functions. In the developed world, a host of institutions has grown up to perform these functions. They include the sales forces of input and equipment suppliers, firms that specialize in market information, private crop consultants and veterinarians, agricultural magazines and broadcasting stations, and the information services of regulatory agencies. Similar organizations exist in the developing world, though these are typically much smaller and less well established than their developed country counterparts. In both the developed and developing worlds, formal and informal networks among farmers play a major role in the dissemination of information.
7
Agricultural extension Often the largest organization in an agricultural knowledge system is the extension service. In the developed world, extension is just one of many competing sources of information for farmers. In the developing world, however, the frequent lack of such suppliers gives the extension service a vital role in the dissemination of new technologies and in the solution of farmers' problems. Extension services field large numbers of personnel: their numbers are rivaled at the lowest levels only by village government officials and school teachers. In most of the developing world, the idea that agricultural extension must have close ties with formal research is relatively new (Kaimowitz 1991:101). Much early extension work was based on the idea that there was already a sufficiently large body of existing technology, both locally and abroad, that could be used to improve production, without having to ensure that the latest research findings reached the extension personnel. Colonialism meant that research activities focused on export crops and a few expatriate and elite indigenous farmers. Until well after the Second World War, research was largely confined to developing new crop varieties and to soils research, while extension was often coercive and included numerous other unrelated activities (Sims and Leonard 1990:44). More recently, extension began to be seen as a bridge between research and farmers -- a bridge to bear a one-way procession of newly developed agricultural technologies and policy directives on their way to be utilized by farmers. The task of extension personnel was to translate these technologies and directives into terms farmers could understand, then to persuade as many farmers as possible to employ them. The one-way, top-down view of extension has now been replaced by a more balanced view -- at least at academic institutions. But the dissemination of research-based technologies to farmers is still a major task of extension, and indeed should remain so if public agricultural research institutions continue to function. This does not mean that the dissemination of research findings is -- or should be -the only task of extension. Much of the advice an extensionist gives farmers draws on an already existing body of knowledge, based on both formal research and farmers' experience. A great deal of extension work consists of facilitating organization by and communication among farmers. Close ties with research are not necessary for such activities. In any case as Albrecht et al. (1989:185) point out, formal research has often failed to develop solutions to local problems, and other sources of innovations may also be useful.
8
Research-extension linkages Nevertheless, the importance of strong linkages between research and extension is now widely recognized (Baxter and Thalwitz 1985:42-48). But there have been far fewer studies of this link than of that between extension and users (Compton 1989:126). This is true in the developed world also (Feller 1986:283, Feller et al. 1984:45-47). In the United States and elsewhere, the topic of research-extension linkages has fallen between investigations of the research system, such as Busch and Lacy's (1983) study of U.S. agricultural scientists' research priorities, and assessments of extension services, such as that by Warner and Christenson (1984). Indeed, some recent texts on extension still pay little attention to research-extension ties (Albrecht et al. 1989:185; Van den Ban and Hawkins 1988:26-33) There has been a recent surge in interest in extension systems worldwide, and in research-extension linkages in particular. This has followed the pioneering work of Lionberger and Chang (1970) in Taiwan, and Nagel (1980) in India. It was given a major boost by a conference at Cornell University in 1980 (Compton 1992), and a series of dissertations at Cornell in the 1980s (Dhandhanin 1984, Hussein 1986, Lakoh 1986, Lupanga 1986, Malik 1988). It has been further fueled by the activities of INTERPAKS (the International Program for Agricultural Knowledge Systems at the University of Illinois) (INTERPAKS 1983-91), and a major review of research-extension linkages by ISNAR (the International Service for National Agricultural Research, in the Netherlands) (Kaimowitz 1990). In addition, a series of World Bank publications has looked at the Training-andVisit system the World Bank has promulgated in numerous countries, including the focus of this study -- Indonesia (e.g., Feder, Lau and Slade 1985; Feder and Slade 1983; Cernea, Coulter and Russell 1985). Table 2.1 lists some of the recent literature on extension and research-extension links worldwide.
9 Table 2.1 worldwide.a
References on agricultural extension and research-extension linkages
General
Organization-level studies
Albrecht et al. 1989, 1990
Agudelo and Kaimowitz 1989
Colombia
Arnon 1989
Busch and Lacy 1983
U.S.A.
Axinn and Thorat 1972
Cernea, Coulter and Russell (eds.) 1985
Bangladesh, India, Indonesia, Pakistan, Sri Lanka, Thailand
Hornik 1988
Ekpere and Idowu 1990
Nigeria
INTERPAKS 1983-92
El-Zoobi 1988
Pakistan
Kaimowitz (ed.) 1990
Engel 1990
Colombia
Lionberger and Gwin 1991
Eponou 1990
Ivory Coast
Rivera and Schram (eds.) 1987
Feller 1986
U.S.A.
Compton 1989
Rivera and Gustafson (eds.) 1991 FFTC 1981
E and SE Asia
Swanson (ed.) 1984
PR China
Fivawo 1987a
Van den Ban and Hawkins 1988 Fivawo 1987b
Fiji
Fivawo 1987c
Kenya
Individual-level studies
Hildreth 1965
U.S.A.
Anyawu 1982
U.S.A.
Kaimowitz 1989
Colombia
Curvo 1983
U.S.A.
Lionberger and Chang 1970, 1981
Taiwan
Gidley 1977
Australia
McCorkle and Esslinger 1992
Brazil, Indonesia, Kenya, Morocco, Peru
Hussein 1986
Indonesia
Ortiz et al. 1991
Guatemala
Lupanga 1986
Tanzania
Sigman and Swanson 1984
Worldwide
Mundy 1989
U.S.A.
Snyder 1987
West Africa
Nagel 1980
India
Warner and Christenson 1984
U.S.A.
Seegers and Kaimowitz 1990
(review)
Wyckoff 1965
U.S.A.
a
For references on research and extension in Indonesia, see Chapters 3 and 4.
10 In many countries until recently, it made little sense to talk of research-extension linkages, since both research and extension institutions were so weak that there was nothing to link. Growth in both sets of institutions, however, has given rise to a new concern -- how to ensure the smooth communication of information between them. The research-extension linkage is similar in many ways to the situation of the plant pathologist and the farm discussed above. The same four constraints apply: Extension personnel are numerous compared to researchers, they are geographically scattered, they require information in a different form from that provided by researchers, and they face different sets of environmental and social conditions. Just as extension services have begun to evolve as links between research and farmers, another set of job roles and institutions has developed to link research and extension. These include extension subject-matter specialists, research communication departments, and extension media production units. Subject-matter specialists In the United States, the functions of agricultural research, teaching and extension are combined in a single institution in each state: the land-grant university. The researchextension linkage is personified by extension specialists at these universities. While each state has its own organizational variant, these persons are generally based in the university's academic departments and have faculty rank. They may hold full-time extension appointments or combine this work with teaching or research. Their task is to collect, interpret, and translate information pertaining to their field, and to disseminate it to end-users -- directly, through county extension personnel, or via the mass media. They also provide feedback to their researcher peers on field problems, and may perform research themselves to help solve those problems (Mundy 1989:2). In most of the developing world, research, extension and teaching are the responsibilities of different institutions: research institutes, extension agencies, and universities. It is thus not possible to replicate the U.S. system (even if this were desirable) without massive and disruptive institutional change. Another model must be sought. In the Training-and-Visit extension system promoted by the World Bank in numerous developing countries, the role of linker is performed by extension subject-matter specialists (SMSs) (Benor and Baxter 1984:33-38). These are usually based not at research institutions but at provincial and district extension offices. Like their U.S. counterparts, it is their role to seek and translate information and to disseminate it to field extension agents. They are expected to divide their time roughly equally between three tasks: • Visiting farmers' fields to monitor problems and to check whether recommendations are appropriate and are being adopted by farmers. •
Training field agents in technology recommendations.
11 •
Maintaining contact with research. The SMSs are to maintain contact with research in four main ways:
•
Participating in monthly workshops that are also attended by researchers.
• Visiting research stations to meet researchers in their own specialty field, use library facilities, and observe experiments. •
Receiving formal training from researchers.
•
Conducting trials on farmers' fields, and analyzing them jointly with researchers.
SMSs are recognized as a vital but still weak link in the research-extension continuum (Hussein 1986:417; Padmanagara 1985:137; Pickering 1985:167; Wirasinghe, Weerasinghe and Fernando 1985:101). Problems include lack of sufficient specialists, poorly qualified personnel, irregular or inadequate training courses, distraction by administrative and other functions, and lack of interaction with research. Linkage institutions Two institutions aimed at linking research and extension are relevant to this study. These are research communication departments and extension media production units. The former are located within the research institutions. Examples are the experiment station information offices at U.S. land-grant universities, and the information units of research institutes in Indonesia (see Chapter 4). They typically publish research bulletins and annual reports, stage exhibitions, produce press releases, and engage in public relations work. Extension media production units are located within the extension institution. They perform similar activities to the information offices, but produce media materials to inform local extension personnel and for these personnel to use in serving their clients. In Indonesia, the provincial Agricultural Information Centers (AICs) perform this role.
12
Causes of poor linkages Along with a recognition of the importance of research-extension linkages has come an increasing awareness that they are a problem. Numerous causes of poor links have been proposed. A short list includes the administrative separation of the research and extension functions, complex institutional structures, status differences between scientists and extension personnel, the geographical distance between them, different time horizons, different motivations and personal orientations, different educational levels and other personal characteristics, lack of accountability to clients, lack of resources and infrastructure, lack of relevant research results, institutional rivalries, inadequate planning and coordination, lack of research continuity, researchers' unwillingness to take unconventional actions, failure to involve small farmers in research planning and implementation, researchers' ignorance of local knowledge, and their neglect of long term social and ecological effects (Albrecht 1989:186; Arnon 1989:786-787; Martínez Norgueira 1990; Kaimowitz, Snyder and Engel 1990; Padmanagara 1985). Most of the literature refers to poor linkages either as an organizational difficulty (to be dealt with by making the appropriate adjustments in organizational structures or operations), or as due to characteristics of the interrelationships among individuals in those organizations (to be dealt with through appropriate hiring, training and incentives). Table 2.1 lists references that use the two approaches. The organizational view (almost by definition) tends to ignore the constraints communication faces at the individual level. The individual view, on the other hand, has tended to push the blame for poor communication onto the shoulders of individual researchers and extensionists: they do not communicate because of mutual jealousies, differing motivations, etc. This view has also tended to ignore the constraints researchers and extensionists face in attempting to communicate. Communication between researchers and extension personnel can occur through two broad groups of channels: direct and indirect. By direct channels here, I mean those where the extensionist receives information directly from the researcher through interpersonal contact or through publications, correspondence, etc., authored by the researcher. Indirect channels have some intermediary who translates the research findings into another form before they are communicated to extensionists. Training courses given by professional trainers, ministry instructions, and magazine articles are examples of indirect channels. The intermediaries include research and extension administrators, subject-matter specialists, trainers, and journalists. Both direct and indirect channels are constrained by factors outside the control of the individual researcher or extensionist. Through no fault of their own, researchers and extension personnel in developing countries very rarely meet -- because of the large numbers of extensionists in comparison to researchers, the vast distances separating them, and the lack of funding for such meetings. Thus, high levels of interpersonal communication are not possible. Direct mediated communication is also difficult because of bureaucratic regulations (see Chapter 4), inadequate distribution of research publications, insufficient
13 funding, the lack of suitable publications as vehicles to carry findings, and the lack of incentive for researchers to communicate with extension personnel. Indirect contacts suffer from similar funding and organizational constraints. In such a situation, the information sources available to an individual extensionist are highly constrained. These constraints, as well as extensionist's personal characteristics, will help determine which sources the extensionist uses. This study is premised on the assumption that individual extension personnel are key actors in determining the flow of information from research to extension. Research supplies a range of information to various audiences, including extension. Extension personnel in turn select from a smorgasbord of information sources -- including research -- and pass this information on to their clients. Which sources the extensionists select depend on various factors that may or may not be under their control, including the constraints described above. This study aimed to find out which factors affect the choice of information sources used by Indonesian extension specialists. Approaches to studying information flows Uses and gratifications The uses-and-gratifications approach to studying the mass media views the audience as actively choosing and using media, rather than passively consuming and responding to them. It sees media use (or information flow) as a function of the receiver's needs and his or her perceptions of the ability of information sources to fulfill those needs (Katz, Blumler, and Gurevitch 1974:20). Thus, if you seek entertainment, you will watch those television programs you think will provide entertainment rather than (say) education or news. A problem with uses-and-gratifications research is that it mixes units of analysis. Your need for entertainment is a unitary characteristic dependent on your personality or some attribute pertaining to you. But your perceptions of a medium or program to provide entertainment are not characteristics of you alone. Rather, your perceptions describe the relationship between you and the medium or program. This mixing of units of analysis dilutes the usefulness of uses and gratifications in predicting information flows among a range of sources. The uses-and-gratifications approach fails to take into account other aspects of the source-receiver relationship. For instance, it assumes that audience members have easy access to the information source -- a reasonable assumption for television news in the United States, but not for agricultural information in Indonesia. Uses and gratifications generally assumes that the audience member has free choice in media selection and exposure, and is free to determine the gratifications sought. Neither of these assumptions is necessarily valid for organizational information flows. If you are watching television, you can choose whether to watch, what to watch, and why to watch. You have a wide range of possible gratifications, including surveillance, correlation, entertainment, and cultural transmission (Katz, Blumler, and Gurevitch 1974:23), relaxation,
14 to forget, companionship (Palmgreen and Rayburn 1979:163), keeping up with current events, finding out about government officials, and so forth (Palmgreen and Rayburn 1982:571). But extension specialists ought to seek information about agricultural technology in order to perform their jobs. Their possible sources and gratifications are constrained by this (McGuire 1974:168). While the uses-and-gratifications approach can include situations where a person is required to seek information, is likely to be of limited explanatory power for such a case. Finally, uses-and-gratifications research has been largely confined to studies of the mass media, though recently it has been applied also to interpersonal and other non-mass media information sources (e.g., O'Keefe and Sulanowski 1991:9). These problems do not mean that I could not have used the uses-and-gratifications approach in this study. But they did lead me to seek a more appropriate model. Information seeking Like uses and gratifications, the information seeking paradigm also assumes an active audience, one in which people seek information rather than are fed it. Unlike uses and gratifications, however, it includes constraints to information flows such as accessibility and choice. While much information seeking research focuses on unitary rather than relational variables, the approach does not require such a focus. And the information seeking approach is as applicable to non-mass media as to mass media. Much of the information seeking literature tries to identify characteristics of individuals or of their situations that affect their information seeking styles. This is not the focus of this study. Rather, I aimed to identify the reasons receivers use certain sources rather than others. The literature focusing on the costs and benefits of different sources (e.g., Atkin 1973) is thus more germane to this study. I discuss it in more detail in the section on source-receiver relationships in Chapter 5. Social interaction A large body of literature exists on information and innovation diffusion (Rogers 1983). Labeled the "social interaction" model by Havelock (1969, 1986a, Havelock and Lingwood 1973:274), this approach focuses on the attributes of innovations and the characteristics of adopters as determinants of the spread of an innovation (Rogers 1983). Innovations that possess certain characteristics (trialability, compatibility, observability, etc.) are expected to spread faster than those with the opposite features. And audience members who are better educated, have higher social status, are more intelligent, and who are more cosmopolite (and a host of other characteristics) are more likely to adopt than those without such features (Rogers 1983:260-261). Studies on communication channels within the diffusion paradigm have focused on channel use by different categories of adopters and at various stages in the adoptiondecision process. It has been generally found that the mass media and cosmopolite
15 channels are more important than interpersonal and localite channels among early adopters and in the first stages of the adoption process (Rogers 1983:197-202). Beyond the interpersonal-mass media distinction, however, this branch of diffusion theory is of little use in predicting which of several channels will be used. Three aspects of diffusion research are especially useful in the study reported here. • Diffusion studies have pointed to factors which affect the flow of information between individuals. People who are homophilous (similar to each other) communicate more frequently than those who are heterophilous (different) (Rogers 1983:274). And change agents who can empathize with their audiences are likely to be more effective than those without this ability (Rogers 1983:321). While I use neither homophily nor empathy in this study (for reasons outlined below), they are important because they show that communication between a source and a receiver is affected by the relationship between them rather than (or as well as) by factors pertaining to the source or receiver alone. Homophily and empathy alert us to other characteristics of this relationship that may determine information flows. • Diffusion research has indicated the importance of the strength of weak ties (Rogers 1983:295). These are infrequent contacts that can nevertheless be extremely important in providing someone with new information. Research-extension linkages may well be of this nature: contacts may occur seldom but be very useful when they do occur. • The diffusion paradigm has investigated the spread of innovations through a population. It focuses attention on the speed of diffusion and provides a vocabulary with which to describe this. Network analysis Growing out of early work on sociometry, network analysis has been used to extend the diffusion paradigm just described. This approach studies the flow of information among nodes (people or organizations) in a network (Rogers and Kincaid 1981, Rogers 1983:293). Information flow is measured between each node and all others (or with the most important partners only) to identify patterns of communication and key individuals. While it has great utility for studying information flows among individual farmers and extension personnel, network analysis has two drawbacks for the type of research I wanted to perform. First, for impersonal channels such as the mass media, it is difficult to distinguish the source from the channel (Rogers 1983:197). Network analysis requires that all nodes be true sources rather than source/channel combinations. And second, network analysis typically ignores the type of channel used. It groups together all channels from a single source as the measure of information flow from that source, and makes no predictions about which channel will be used.
16
Research, development and diffusion This approach is somewhat confusingly named in view of the label "diffusion" often attached to the social interaction perspective, described above. Havelock (1969:11/6) describes it as seeing innovations as developed through basic and applied research, passing through a development and testing stage, before being packaged and marketed to users. It focuses on the research, testing, and adaptation activities that precede an innovation's emergence rather than the process of spread within the audience, which is the main topic of the social interaction (viz. "diffusion") paradigm. This perspective is represented by the Coughenour-Lionberger and Meehan-Beal models (Lionberger 1986, Lionberger, Pope and Reddy 1979:13, Meehan and Beal 1977, Beal and Meehan 1986). In agriculture, it recognizes that researchers, extensionists and farmers must communicate with one another. But it does not predict which channels are used for such communication. Problem solving The problem-solving model focuses on the user's needs and actions to satisfy those needs (Havelock and Lingwood 1973:276, Havelock 1986a:89). It sees users as formulating needs, seeking and retrieving information that may help solve the needs, and then fabricating and applying solutions. The role of the change agent is to assist in and guide this process rather than to direct it. The problem-solving perspective focuses attention on users, their situations and needs, and on user participation in decision making. But it ignores the fact that users may lack the information or resources needed to initiate and carry through the change process. And it fails to provide a clear strategy for change agents to disseminate findings among a user population, especially to those unable to solve problems themselves (Havelock 1986a:91). Linkage Havelock (1986a:98 and 1986b) develops the concept of "linkage" to subsume aspects of the "social interaction," "research, development and diffusion," and "problem solving" perspectives described above. He see linkages as the two-way communication ties between information providers and users. Havelock's formulation is particularly useful for describing the processes that must take place if an item of technology developed in one place is to be applied in another. But it suffers from two shortcomings for the purposes of this study: • A "linkage" is conceptualized as being two-way -- though not necessarily using the same channels (Havelock 1986b:218). But previous research in Indonesia and elsewhere has shown that extension-to-research flows are very tenuous (Hussein 1986:435). I discuss this further in Chapter 5.
17 • Havelock does not develop, nor does the linkage concept readily lend itself to developing, a testable model to predict the level of information flows reaching a receiver from a source (such as a book, training course, or newspaper). The linkage concept is thus of limited use in predicting why a receiver should use one source rather than another. Somewhat confusingly, Havelock (1973:294) uses the term linkage to refer to another concept also: "the existence of person-to-person contacts where two-way communication was taking place." This is the meaning I adapt for use in this study and refer to as Familiarity (see below). Checklists of information flow determinants Several checklists of characteristics have been proposed to explain the amount of communication between individuals, group, or organizations (Glaser, Abelson and Garrison 1983). Perhaps the best known is the "A VICTORY" formulation (Davis 1971, Bedell et al. 1985). A VICTORY is an acronym standing for the seven elements in the list: Ability, Values, Idea, Circumstances, Timing, Obligation, Resistance, and Yield. This checklist has been widely used in evaluating the acceptance or rejection of innovations by organizations. However, the A VICTORY formulation is of limited use for this study for several reasons: •
It refers to organizations rather than individuals or publications.
• It presupposes that a known innovation (or group of innovations) is under study and tries to discover reasons for its (their) spread or failure to spread. • It does not attempt to explain the choice of sources by individuals for obtaining information on a range of topics over an extended period (Mundy 1989:20). • Negative attributes are grouped together as Resistance, even though they may be better seen as negative aspects of other elements (such as lack of Ability). Havelock and Lingwood (1973:294) provide a more useful checklist for this purpose. Labeled the "HELP SCORES" model, it is suggested as "a set of project or change variables as a schema for diagnosing problems in the communication of new knowledge or innovations from any source to any receiver" (Glaser, Abelson, and Garrison 1983:3334; emphasis added). The ten items in the list are Homophily, Empathy, Linkage, Proximity, Structure, Capacity, Openness, Reward, and Synergy. Note that Linkage as used here is different from the linkage concept described in the previous section. I discuss the individual factors in more detail in Chapter 5. However, we should note here that this model is particularly useful for this study because each factor can be seen as a characteristic of the dyadic link between two individuals -- and as potentially affecting the information flow between them. Each of these variables can be positive (leading to higher information flow) or negative (resulting in less flow).
18 Havelock (1969) developed seven of the factors in a review of 4000 studies on knowledge utilization, Havelock and Lingwood (1973) later added another three (Homophily, Empathy and Energy) as a result of Rogers and Shoemaker's (1971) work on innovation diffusion. The HELP SCORES model has not received much attention in the literature, and this is only the third empirical test of it I have been able to find. Havelock and Lingwood (1973:295) themselves used the HELP SCORES factors as a basis for coding oral interviews with staff members of the U.S. Department of Labor. Their coding sheet provided positive and negative instances of each factor. They did not develop a series of questions to measure them. The only attempt to develop a series of survey questions to measure the HELP SCORES factors was a study I conducted of communication between extension specialists and researchers at Iowa State University (Mundy 1989). This study represents a continuation of this line of research, expanded to include non-personal as well as personal sources. It is unclear why the HELP SCORES model has received so little attention in the literature. One possible reason is the influential volume by Glaser, Abelson and Garrison (1983). In an attempt to reconcile several lists of factors proposed by various authors, Glaser and colleagues (erroneously, I believe) equated the HELP SCORES concepts with those in the A VICTORY formulation. The A VICTORY list thus survived, while the HELP SCORES model sank from view. For the reasons outlined above, I chose to use the HELP SCORES model in this study. Summary It is possible to use a number of different approaches to study research-extension linkages. I choose to select concepts from several different approaches for this study. While I base the model I develop in Chapter 5 on Havelock and Lingwood's (1973) HELP SCORES list, I also draw on the information seeking and social interaction ("diffusion of innovations") approaches.
20
CHAPTER 3 AGRICULTURAL EXTENSION IN INDONESIA Introduction Indonesia's agricultural research and extension systems are large and complex. This chapter describes pertinent details of the extension subsystem. The next chapter discusses the research subsystem and research-extension linkages. The chapter begins with a brief description of Indonesia and its agriculture before turning to the administrative units of the Ministry of Agriculture relevant to this study. For convenience of exposition, I describe first the ministry directorates-general, Bimas, AAET, and the provincial and district units, before discussing the extension system in more detail. I focus especially on the roles of the subject-matter specialists (SMSs, the respondents in this study) and of the Agricultural Information Centers. The chapter concludes with a brief discussion of the mass media related to agriculture. Indonesia: The setting Indonesia is a land of superlatives. The world's largest archipelago, its 13,667 emerald-green islands arc along the equator from mainland Southeast Asia to Australia, dividing the Indian Ocean from the Pacific. More than 5000 km from east to west, Indonesia stretches further than from Lisbon to the Urals, from Hammerfest to Aswan, or from Los Angeles to Halifax. It contains five of the world's largest islands: Kalimantan (the Indonesian part of Borneo), Sumatra, Irian Jaya (the Indonesian half of New Guinea), Sulawesi and Java. It is home to numerous active volcanoes, vast swamps, and Asia's largest tropical forests. Indonesia's inhabitants are no less remarkable than its landscape. Its 188 million people make it the world's fourth most populous country (after China, India, and the United States). Over centuries, they have constructed intricate irrigation systems feeding beautiful rice terraces that climb mountain slopes toward the sky, developed a rich cultural heritage, and evolved unique forms of architecture, dance, textiles, and other art forms. They belong to numerous ethnic groups and speak more than 250 languages. While Indonesia has the world's largest population of Muslims (87% of Indonesians profess Islam), there are numerous adherents of Christianity (7%), Buddhism and Hinduism. This bewildering variety gives real meaning to Bhinneka tunggal ika "Unity in diversity," the national motto.
21
Figure 3.1
Provinces of Indonesia.
22 The island of Java, covering only 7% of Indonesia's land area, is home to 61% of its population (AARD 1988:3). With an average population density of 755 people/km2 (in 1985), Java rivals the valleys of the Ganges, Yangtze and Nile as the most densely populated area of the world. The rest of Indonesia, by contrast, has only 31 people/km2. Java has the most fertile soils and produces most of the country's food and manufactures. Indonesia's size, diversity, and geographic fragmentation pose unique challenges to government. Yet the country has been remarkably stable politically since 1966. Rapid economic growth has followed, particularly in the industrial sector, bolstered by exports of petroleum, natural gas and timber. In the 1980s Indonesia enjoyed high annual GNP growth rates, causing observers to name it along with Malaysia and Thailand as a potential newly industrializing country. Administratively, the country is divided into 27 provinces (Figure 3.1). At the second tier of local government are 242 districts (kabupaten) and 54 urban areas (kotamadya). Below this are subdistrict (kecamatan), village (kelurahan) and neighborhood (rukun warga and rukun tetangga) levels. Most government ministries are represented at the provincial and kabupaten/kotamadya level through specialized local government offices. Agriculture in Indonesia Despite growth in other sectors, agriculture continues to be of major importance in Indonesia's economy and society. In 1988 the agricultural sector contributed about 20% of the country's gross domestic product, and provided one-third of the nation's non-oil exports by value. Natural rubber, palm oil, coffee, tea, shrimp, and spices are major agricultural exports. Indonesia is also a major producer of rice and coconuts, though almost exclusively for local consumption. Agriculture employs about 35 million people, or about half the total labor force. About 71 million Indonesians (two-thirds of the total population) are directly dependent on agriculture. Croplands cover about 22 million of Indonesia's land area of 190 million ha. Of these, perennial crops cover 6 million ha, and 7.6 million ha are irrigated (FAO 1991). Rice is the major staple, accounting for more than half the area harvested to food crops. Much of the rice land is double- or even triple-cropped. Maize, cassava, soybeans, peanuts, and sweet potato are other major food crops. Tropical vegetables and fruits grown include chili, banana, mango, papaya and citrus. Temperate vegetables such as onions and cabbage are grown in the highlands. The major industrial crops are rubber, oilpalm, coffee, tea, sugarcane, pepper, coconut, cloves, and cacao. Livestock include chickens, ducks, sheep, goats, beef and dairy cattle, and water buffalo.
23
Table 3.1
Food crops and livestock production in Indonesia, 1966 and 1991a. Unitsb
1966 c
(est.)
% change 1966-91
188
+77
Population
000 000
Rice
000 t
13650
44321
+225
Maize
000 t
3717
6409
+72
Sweet potatoes
000 t
2308
1976
-14
Cassava
000 t
12100
16330
+35
Soybeans
000 t
353
1549
+339
Groundnuts
000 t
488
920
+89
Cattle
000 head
6700
10350
+54
Buffaloes
000 head
2790
3500
+25
Sheep
000 head
2340
5750
+146
Goats
000 head
11000
11300
+3
Chickens
000 000 head
64
590
+822
Ducks
000 000 head
22
30
+36
a
Source: FAO 1969, 1991.
b
Metric units used where applicable.
c
Estimate for 1965.
106
1991
Between 1966 and 1991, Indonesia tripled its rice production, changing Indonesia from the world's largest rice importer to self-sufficiency in this staple (AARD 1988:11) (Table 3.1). Production increases in soybeans and chickens have also been spectacular. While some of the fluctuations in output in other palawija (non-rice annual) crops and livestock species can be attributed to data reporting problems, it appears that progress has been less consistent for these commodities, especially when compared to the population growth over the same period. The Ministry of Agriculture The Ministry of Agriculture is one of the larger units within the Indonesian government, receiving a budget of Rp 1,994,200 million ($US 1,100 million) in 1989, second only to the Department of Transportation and Tourism (Syam and Mundy, in press).
24
Figure 3.2 Organizational structure of the Indonesian Ministry of Agriculture (simplified), showing locations of SMSs (based on Biro Humas Deptan 1991).
25 The organizational structure of the Ministry of Agriculture has undergone several revisions in response to changing circumstances. Further revisions were underway during 1991 while I was conducting the research for this study. As far as I could ascertain, the structure and organizational relationships described here were in effect at the time of the research. In 1991 the Ministry was composed of ten first-echelon units (Figure 3.2), five of which concern this study: the Directorates-General for Food Crops and for Livestock Services, the Agency for Mass Guidance (Bimas), and the Agencies for Agricultural Research and Development (AARD, discussed in the next chapter) and for Agricultural Education and Training (AAET). In addition, we must examine agricultural offices at the provincial and district levels. Directorate-General for Food Crops The Directorate-General for Food Crops has eight second-echelon units, all located in the capital, Jakarta: •
The Secretariat coordinates the activities of the other seven units.
• The Directorate of Food Crops Programming (Direktorat Bina Program Tanaman Pangan) sets production targets, plans agricultural projects, monitors and evaluates activities, and collects statistics. • The Directorate of Rice and Palawija Production (Direktorat Bina Produksi Padi dan Palawija) formulates recommendations, implements production programs, produces and certifies seed, and develops agricultural machinery for rice and palawija crops. Palawija crops include non-rice cereals, legumes, and root and tuber crops. • The Directorate of Horticulture Production (Direktorat Bina Produksi Hortikultura) performs similar functions for the horticultural commodities. • The Directorate of Food Crops Farming and Post-Harvest Processing (Direktorat Bina Usaha Tani dan Pengolahan Hasil) concentrates on the socio-economic and postharvest aspects of farming. • The Directorate of Plant Protection (Direktorat Bina Perlindungan Tanaman) monitors and attempts to predict attacks by pests and diseases in the field, tests pesticides, and assists in combatting pest outbreaks. • The Directorate for Land Rehabilitation and Extensification (Direktorat Bina Rehabilitasi dan Pengembahan Lahan) aims to expand the cropped area and rehabilitate areas subject to soil erosion and other problems. • Finally, the Directorate for Food Crops Extension (Direktorat Bina Penyuluhan Tanaman Pangan) coordinates extension activities relating to food crops. The Directorate-General for Food Crops operates a number of regional units
26 around Indonesia. A partial list is given below. The numbers of units may not be accurate since they are taken not from official ministry sources but from the publication mailing lists of an AARD research institute (Puslitbangtan 1991). • 14 Seed Control and Certification Centers (Balai Pengawasan dan Sertifikasi Benih). • 33 Rice Seed Centers (Balai Benih Induk Padi), which multiply high-yielding rice variety seed. •
25 Non-Rice Seed Centers (Balai Benih Induk Palawija).
•
10 Plant Protection Centers (Balai Perlindungan Tanaman).
Directorate-General for Livestock Services The Directorate-General for Livestock Services has seven units with equivalent functions to those in Food Crops: •
The Secretariat.
• The Directorate of Livestock Programming (Direktorat Bina Program Peternakan). •
The Directorate of Livestock Production (Direktorat Bina Produksi Peternakan).
• The Directorate of Livestock Farming and Post-Harvest Processing (Direktorat Bina Usaha Tani dan Pengolahan Hasil Peternakan). •
The Directorate of Animal Health (Direktorat Bina Kesehatan Hewan).
• The Directorate for Livestock Expansion and Development (Direktorat Bina Penyebaran dan Pengembangan Peternakan). • The Directorate for Livestock Extension (Direktorat Bina Penyuluhan Tanaman Pangan). Like the food crops equivalent, the Directorate-General for Livestock operates a number of regional units. These include the following. Again, the numbers are drawn from AARD institute mailing lists (Balitnak 1991, Balitvet 1991). •
7 Animal Disease Investigation Centers (Balai Penyelidikan Penyakit Hewan).
• 9 Animal Reproduction and Feed Centers (Balai Pembibitan Ternak dan Hijauan Makanan).
27
Agency for Mass Guidance The Agency for Mass Guidance (Badan Pengendali Bimbingan Massal, BP Bimas) coordinates intensification programs in rice and other commodities (Taslim 1991). Such programs provide information through the extension system; credit through local branches of the Bank Rakyat Indonesia; and inputs, post-harvest processing and marketing through village-level cooperatives and kiosks. Various patterns of group collaboration have also been promoted, resulting in a veritable soup of acronyms: Inmas, Insus, Supra Insus, Opsus, Inmum, amongst others. Hussein (1986) summarizes the history of extension in Indonesia. Bimas has units reaching down to the village level: • The provincial level Bimas unit is headed by the provincial governor. Day-to-day operations are handled by the head of the provincial office of the ministry (Kanwil). • At the district level, the district head (bupati) and the head of one of the district agricultural service offices (usually that of food crops, Dinas Pertanian Tanaman Pangan) perform these roles. • At the sub-district and village levels, the Bimas program is overseen by the subdistrict head (camat) and village head respectively. The existence of these local units and the participation in them of local government leaders at all levels contribute to the effectiveness of Bimas activities. The Bimas program has undergone a number of major changes since its creation in 1965. Hussein (1986:116-143) briefly describes its evolution from a small action research project by Bogor Agricultural University to its current form. At the time of this study, there was debate as to its future direction and even its existence. Agency for Agricultural Education and Training Until recently, this was the Agency for Agricultural Education, Training and Extension. However, the extension function, never fully unified under one body as has been the case for most research since 1974, was returned to the four directorates-general for food crops, livestock, estate crops, and fisheries. AAET coordinates and manages Indonesia's agricultural information and training institutions. These include (Biro Humas Deptan 1991:93-95): • 251 agricultural high schools: 30 of these belong to the ministry, 88 to local governments, and 133 to private institutions. • 32 Agricultural Staff Training Centers (Balai Latihan Pegawai Pertanian, some of which specialize on specific commodities, such as fisheries. •
28 provincial Agricultural Information Centers (AICs, Balai Informasi Pertanian,
28 BIP). The AICs are key to the flow of information on new agricultural technologies; they are described in more detail below. As mentioned above, responsibility for extension has never been brought into a single organization -- despite several changes in the allocation of extension duties within the Ministry. AAET is currently responsible for the education and training of extension personnel and for developing extension methods. Technical guidance of the personnel is the responsibility of the relevant directorates-general (SK Bersama 1991). This distinction is often unclear to those outside the agencies concerned, and can cause some confusion to extension personnel also -- since they are (or have been) responsible to several superiors in different branches of the Ministry. For instance, an extension specialist in food crops may be answerable to the Directorate-General for Food Crops for technical guidance, depend on Bimas for salary and operating funds, rely on research information generated by AARD, use extension materials produced by AAET's Agricultural Information Centers, and be answerable to the local district or provincial government head -- as well as serve the needs of local farmers. Provincial and district units The Ministry of Agriculture operates or coordinates an array of provincial and district technical units to oversee and implement different aspects of its work. These are: • At the provincial level, 27 provincial coordination offices (Kantor Wilayah, Kanwil), responsible directly to the Minister of Agriculture. • 108 provincial-level Agricultural Service (Dinas) offices: one for each of the four major commodity groupings (food crops, estate crops, livestock, and fisheries) in each of Indonesia's 27 provinces. These offices are responsible administratively to the provincial governor but are technically accountable to the relevant directorate-general at the national level. Dinas offices are divided into divisions corresponding to directorates at the national level. The Division of Agricultural Extension directs, monitors, and evaluates provincial extension programs (Hussein 1986:156). Also housed in each provincial Dinas office are several extension subject-matter specialists (SMS, Penyuluh Pertanian Spesialis, PPS) -the respondents used in this study. • At the district (kabupaten) level, approximately 1100 district-level Agricultural Service (Dinas) offices, one for each of the four commodity groups in each of Indonesia's 296 districts and municipalities. Some districts, such as urban areas (kotamadya) do not have a full complement of Dinas offices. Like their provincial counterparts, the district Dinas offices are administratively responsible to the district head (bupati) or city mayor (walikota); they are technically accountable to their corresponding provincial Dinas office. Also like their provincial counterparts, they are divided into units corresponding to the directorates at the national level. Each district Dinas office also houses a number of SMSs, many of whom were also respondents in this study.
29 • District offices representing the Mass Guidance (Bimas) program. These offices are frequently combined with the district's Food Crops Agricultural Service office (Dinas Pertanian Tanaman Pangan, Diperta). I was not able to determine their number. • Below the district level, 2217 local Rural Extension Centers (Balai Penyuluhan Pertanian, BPP). These house Indonesia's 29,400 field extension agents (Biro Humas Deptan 1991:94, 98). Extension The extension subsystem Linking policy makers and research institutes with field extension workers and farmers is a complicated network of institutions (Figure 3.3). The extension organization is extremely large. According to Bimas, in 1990 there were some 29,407 field extension workers and 1485 subject matter specialists (Biro Humas Deptan 1991:98). An organization of this size poses major challenges, especially in the absence of a sophisticated communication infrastructure. Coordination at the national level is performed by the National Agricultural Extension Commission (Komisi Penyuluhan Pertanian Nasional, KPPN), composed of the heads of the first-echelon agencies in the Ministry of Agriculture and chaired by the ministry's Secretary-General (Abbas, Tjitropranoto, and Yakub 1989, Abbas 1991, SK Mentan 1991). At the provincial and district levels, equivalent bodies are Agricultural Extension Coordination Forums (Forum Koordinasi Penyuluhan Pertanian, FKPP-I [at the provincial level] and FKPP-II [at the district level]). These are composed of local agricultural officials and extension specialists; scientists from local research institutes are also invited to attend (Suhardjo 1989:128).
30
Figure 3.3
Organization of agricultural extension in Indonesia (SK Bersama 1991).
31 With World Bank sponsorship, Indonesia introduced the "training and visit" system for extension in the late 1970s (Sukaryo 1983, Jahi 1991, Benor and Harrison 1977, Benor and Baxter 1984). Under this system, graduate extension subject-matter specialists (SMSs, penyuluh pertanian spesialis, PPS) train field extensionists in seasonally relevant material at regular fortnightly training sessions. Each field extension worker (FEW, penyuluh pertanian lapangan, PPL) is assigned to a number of villages, and visits each village once every two weeks. The field extensionist works with groups of contact farmers (kontak tani) in each village, discussing relevant topics for the time of year. These contact farmers in turn are expected to disseminate their knowledge to "follower farmers" in their village. According to the Ministry of Agriculture (Biro Humas Deptan 1991:96-97) there are some 250,000 farmer groups, with some 14 million members, throughout the country. However, ministry officials admit privately that this is an overestimate of the number of groups actually functioning. A number of village institutions are key to the success of this extension effort. These include the village cooperative (Kooperasi Unit Desa, KUD), which markets output to the national Food Logistics Board (Bulog); kiosks selling agricultural inputs; and the Village Unit Bank, a branch of the national Bank Rakyat Indonesia, which provides credit (Suhardjo 1989:134). All are coordinated through the Bimas program. This scheme is hoped to allow a relatively rapid transfer of technology from research institutes to the farmers. It is also expected to allow for feedback, since field extensionists can refer field problems back to the relevant subject-matter specialist, who can if necessary refer them back to researchers. The potential effectiveness of the extension set-up can be seen in the large increases in Indonesia's rice production experienced since 1966, and in the rapid adoption by farmers of modern wetland rice varieties. However, similar yield increases and adoption rates have not been evident with most other crops and commodities (Table 3.1). This may be related to other factors, though the emphasis given to rice in the extension system's efforts undoubtedly does play a role (see discussion in the next chapter).
32
Table 3.2
References on agricultural extension in Indonesia.
Extension * Abbas 1991
* Rusyana 1984
* Abbas et al. 1989
* Samsisaputra 1987
* Azis 1990
* SK Mentan 1984
* Dir. Penyuluhan RRL 1990 •
Ag Info Centers (AICs)
Harun 1987
• * Suharno 1986 Training and Visit extension, farmers' groups
* Hubeis 1987 •
Hussein 1986 Ludgate et al. 1990
• * Riyanto 1988 Röling et al. 1991
• * Bahraini 1984 •
Batoa 1985
• * Damayani 1988
* SK Mentan 1991
• * Holian 1990
* SK Bersama 1991
• * Nur 1986
* Slamet 1990
• * Subarma 1985
Sukaryo 1983 •
* Abdul Adjid 1990
•
Suradisastra and Soedjana 1990
Suhardjo 1989 Sukaryo 1983
• * Surialaga 1984 Syam and Mundy, in press
• * Suryono 1985
* Taslim 1991
• * Sutjipta 1982
* Wardojo 1990
• * Witjaksono 1990
•
Warsito 1989
•
Wiratmadja 1987
•
Widjono 1986
• Data-based study * In Indonesian.
Overview of the literature on extension
33 Much has been written on the Indonesian extension system in recent years: Table 3.2 gives a partial list of references. While much of the literature lacks empirical support, a body of data-based research does exist on the functioning of the "lower" levels of the extension system, particularly of the field agent-farmer interface. Most of this research seems to have been performed by sarjana and master's students in the development communication program at Bogor Agricultural University, mostly in West Java (apparently for logistical reasons). Examples of these are Holian (1990), Nur (1986), Riyanto (1988), Subarma (1985), Surialaga (1984), and Suryono (1985) in West Java; Bahraini (1984) in West Sumatra; Sutjipta (1982) in Bali; and Witjaksono (1990) in Yogyakarta. I was not able to visit other university libraries to seek references, and searches of the Indonesian Agricultural Index revealed no other studies. Most of the studies in English have been done by Indonesians studying overseas. They include Warsito's (1989) study of field agents' job performance in Yogyakarta; Wiratmadja's (1987) thesis on agents in West Java; Harun's (1987) study of extension publications in West Java; Widjono's (1986) research on upward communication within district Dinas offices; and Suhardjo's (1989) analysis of the T&V system in West Java. The most extensive and valuable study by a non-Indonesian is that by Hussein (1986). All these studies are dissertations or theses. I did not have access to any internal evaluations performed by the various extension agencies. To my knowledge, among these authors, only Widjono (1986) was an employee of AARD. Other AARD authors writing about extension have largely confined themselves to two topics: research-extension linkages, and the adoption of research findings by farmers (see the later discussion of research-extension linkages). Descriptions of the theoretical functioning of the upper echelons of the extension system abound (e.g., Taslim 1991; Abbas, Tjitropranoto and Yakub 1989), as do papers outlining extension policy and philosophy (e.g., Abbas 1991, Abdul Adjid 1990, Azis 1990, Dir. Penyuluhan RRL, Hubeis 1987, Slamet 1990, Wardojo 1990). However, I was unable to find any research on the functioning of extension agencies above the district level or of research-extension linkages above the level of individual researchers and extension personnel. Agricultural extension specialists Agricultural subject-matter specialists (penyuluh pertanian spesialis, PPS), the respondents in this study, are a key component of the extension system. They hold at least a sarjana (four years plus thesis) degree in an agricultural or social science. They are employed at various organizations: •
Provincial Dinas and Bimas offices (and some at Kanwils).
•
District Dinas and Bimas offices.
•
Agricultural Information Centers.
34
Their tasks are extremely diverse. They include (Sophia 1988:3-5): • Preparing local extension plans for discussion at Agricultural Extension Coordination Forum meetings. • Collecting information on new technologies from research institutes and universities, and providing feedback on field problems to these institutions. •
Maintaining links with other agricultural institutions at the national and local levels.
• Communicating with other SMSs about new technologies and government programs. •
Conducting surveys, analysis, and evaluations of extension activities.
•
Advising agricultural officials in their areas of specialization.
•
Writing, teaching, and participating in seminars.
•
Collecting and translating materials to help solve field problems.
•
Conducting field trials of research findings.
• Processing and analyzing survey and experimental data to develop extension plans and technology recommendations. •
Guiding field agents to use new technologies.
• Guiding and assisting field agents to prepare extension plans, improve their skills, and solve field problems. These tasks can be summarized as (1) obtaining information on new technologies and translating it into a form usable by field agents and farmers, (2) testing technologies for local applicability, (3) training field agents, (4) solving field problems, and (5) liaising with other actors in the extension and administration systems. SMSs face numerous constraints in obtaining information. Located at provincial and district capitals, they must travel to the nearest research institute or library obtain information that is not located in their offices. The budget to support such search is woefully inadequate: according to one interviewee in North Sumatra, an SMS receives an allowance (materi operasional penyuluhan) of only Rp 14,000 ($US 7) per month to pay for all work expenses (field visits, field trials, and training materials, as well as information search). Field extensionists get even less: Rp 11,500 ($US 6) per month. Unlike the AICs, provincial and district Dinas offices do not have libraries, and they are not routinely sent AARD publications (see next chapter). These constraints make it difficult for even highly motivated SMSs to obtain information. Most are therefore perforce likely to be dependent on information that arrives without any effort on their part. In Atkin's (1973:238) terms, they will thus engage in information receptivity rather than information search (see the section on Situation in Chapter 5).
35
Agricultural Information Centers There are 28 AICs: one in each of the 27 provinces, plus a center with nationwide responsibilities at Ciawi, near Bogor in West Java. They are coordinated by AAET. Twelve have been operational since 1978; the remaining 16 have been in existence since 1985. AIC roles The AICs have multiple roles (Harun 1987:5-7, Rusyana 1984, Samsisaputra 1987, Suharno 1986). The following list is based on the official ministerial decision establishing the centers (SK Mentan 1984): • Collect, select and process data and information from various sources, such as research institutes, universities, libraries, government institutions, farmers, and rural people. • Prepare information in a suitable form, such as publications, teaching aids, and audio-visuals. •
Disseminate information to extension institutions.
• Monitor and evaluate the efficiency and effectiveness of the information disseminated. Several tasks have since been added (Harun 1987:5-7), including participation in provincial Agricultural Extension Forums, preparing extension messages for the mass media, and acting as a provincial center for statistical information on agriculture. But the main task of the AICs remains providing extension personnel with information and teaching materials. AIC staff and facilities The AICs are staffed by subject-matter specialists, typically about six per center. Unlike the SMSs in province and district offices, those at AICs are typically responsible for a discipline area rather than a particular set of commodities. Some have a background in extension or communication rather than, or as well as, an agricultural topic. The Ciawi AIC is responsible for training staff of other AICs in communication methods. The AICs vary in the types of equipment they have available. At the four I visited (Ciawi, Lembang, Medan and Palembang), equipment included: a sheet-fed, single color, mini-offset printing press; a large-format camera for making halftones; a darkroom for processing black-and-white prints and slides; an audio studio; simple video recording and editing equipment; and a single computer (used mainly for word processing). Not all the AICs I visited possessed all these items. Each AIC also has a library, frequented mainly by students from local universities. Three of the libraries I visited had a good range of AARD publications on the shelves; the other had very few, possibly because the AIC had been established relatively recently. Media produced The AICs produce eight main types of media (described below). The number of copies produced seems to vary considerably among AICs and over time, apparently because of variations in funding. While I have no data on this, funding appears to have declined since the AICs were first established.
36 • Booklets (brosur, Figure 3.4), each about 30 pages long, printed in black and white with a 4-color cover and sometimes a 4-color insert. These cover topics ranging from the formulation of chicken feed to methods of forming farmers' groups. They are aimed at extension personnel. In 1984, the West Java AIC in Lembang printed between 10,000 and 20,000 copies of each issue (Rusyana 1984). According to an anonymous informant, the nationwide Ciawi AIC prints about 4000-4500 copies of its booklets, and aims them primarily at SMSs rather than field agents. The Palembang AIC produced four brochures in 1989/90. The Medan AIC prints 3000 copies each of the five booklets it produces in a year.
37
Front cover of a booklet on how to formulate rations for local chicken breeds, reproduced full size. The original contains 36 pages of text and two color photographs. The cover is in full color. Figure 3.1 Booklet (brosur) produced by Medan Agricultural Information Center, North Sumatra.
38 • Liptans, single-page technical fact sheets aimed at farmers (Figure 3.5). Liptan is an abbreviation for lembaran informasi pertanian, or "agricultural information fact sheet." They are printed in a single color, and a second color is used to code the issue by topic using a modification of the Agdex filing scheme: red for livestock, pink for fruits and vegetables, and so on (BPLPP/Pustaka 1983). Each Liptan covers a single topic, such as the control of vascular streak dieback on cacao, and growing mushrooms on sugarcane leaf substrate. Rusyana (1984) has no information on the number printed at the Lembang AIC. According to respondents, the Ciawi center prints about 2500 copies of each issue; the North Sumatra AIC prints 3000 copies of each of the approximately 20 Liptan titles it publishes in a year; AIC Palembang produced 20 Liptans in 1989/90, printing 2500 copies of each. • Posters, printed in full color on large sheets. In 1984, the Lembang AIC printed 8000 copies of each poster. None of the AIC personnel I interviewed mentioned producing any posters recently. • Folders, single sheets folded twice to form a brochure, aimed at contact farmers. In 1984, AIC Lembang printed 10,000-30,000 copies of each. In 1989/90, the Medan AIC printed 6000 copies of a single title. • Audiocassettes of recordings of songs and studio dramas, for extension agents' use and broadcast over local radio stations. The AICs at Ciawi and Palembang each produce about 10 titles a year. The number of copies is unknown; I estimate it to be about 250 (based on Suharno 1986:54 and BIP Ciawi 1990). • Slide sets and accompanying booklets with instructions for the presenter. These are distributed to Dinas offices. Because of funding limitations, AIC Ciawi now produces a master copy only of each slide set; other AICs and institutions can order copies at a cost of Rp 60,000 (about $US 30) each. AIC Medan produced three titles in 1991; because of funding limitations, only 10 copies of each were made, compared to the 60 copies normally produced. Respondents commented that slide sets were relatively expensive to produce: about Rp 1.5 - 1.8 million ($US 750) per title. • Videocassettes and 35 mm films. While at least some of the AICs have video recording and editing equipment, these are not of broadcast quality. Most district Dinas offices and Rural Extension Centers have no video playback units. AICs personnel also lack skills in video production. AIC Ciawi has made several videos and offers them to other AICs for Rp 15,000 ($US 7) per copy. The cost of producing videos is high: about Rp 12,000,000 ($US 6,000) per program. AIC Ciawi has cooperated with the national television corporation, TVRI, in producing television programs. The center produces about 10 35 mm movies a year and has made about 75 in all for distribution to other AICs. The Medan AIC produces videos occasionally on request.
39
Reproduced 65% of actual size. The original was printed on two sides in black, with the dark box at the top left in red. This Liptan is about cultivating King Grass; it is based on a publication by AARD's Central Research Institute for Animal Science. Figure 3.2 Liptan technical fact sheet, produced by Palembang Agricultural Information Center, South Sumatra.
40 • Buletin Informasi Pertanian is an agricultural information magazine aimed at extension personnel. Each AIC publishes its own version for distribution within its own province (Figure 3.6). There are typically two to four issues a year, with the Lembang institute producing 40,000 copies of each in 1984; AIC Ciawi currently prints 6000 copies; the North Sumatra AIC, 3000. • Other media, such as flip charts, models, and computer databases, are currently of minor importance. AIC Medan had by 1991 produced one flip chart with a print run of 200. AIC staff commented that the heavy paper and color printing required made flip charts expensive. AIC Ciawi cooperates with CALREC (see below) in reproducing and distributing computer diskettes containing CALREC's CDS/ISIS literature database (CDS/ISIS is a textbase program promoted by Unesco). However, other AICs lack computer facilities and skills to use these diskettes. Between 1984 and 1990, the 28 AICs produced a total of 4516 media titles in all formats except the Buletin (Table 3.3). Of these, 25% were on food crops, 22% on general topics such as extension methods and soil conservation, and 16-18% each were on livestock, fisheries, and estate crops.
Table 3.3 Numbers of media titles produced by the 28 Agricultural Information Centers, 1984-1990. Food crops
Livestock
Fisheries
Estate crops
General
Total
Booklets
159
135
126
131
178
729
Liptans
548
390
384
369
400
2091
Posters
70
57
41
50
83
301
Folders
92
49
55
47
42
285
197
143
122
84
228
774
Slide sets
88
48
47
51
47
281
Videos
10
8
1
7
29
55
1164
830
776
739
1007
4516
Audiocassettes
Total
Source: Tabulation from database of media titles compiled by BIP Ciawi (1990).
41
Front cover of a 1991 issue, reproduced 65% of actual size. The original has 28 pages of text with occasional black and white photographs and diagrams. The cover and centerfold are in high-quality full color. Other AICs produce similar magazines with the same name. Figure 3.3 Buletin Informasi Pertanian, semiannual magazine produced by Lembang Agricultural Information Center, West Java.
42 The total of 4516 items may seem a large figure. But this number was produced over a period of six years by institutions in 27 provinces; material produced in one province was distributed only in that province. (The AICs do exchange materials with each other, but the number of copies is small.) Each AIC thus on average produced about 4516 ÷ 27 ÷ 6 = 28 items per year. Dividing again by four to take into account the four commodity subsectors (food crops, livestock, fisheries, and estate crops), we obtain a figure of seven items that an extension specialist can expect to receive each year on his or her specialty. The number of items reaching a field agent on each commodity will be smaller because not all items are produced in sufficient numbers to be distributed to all agents. Contact farmers will fare worse still, and follower farmers probably can expect to obtain media extremely rarely from the AICs. Looking at the print runs of publications draws us to the same conclusion. The North Sumatra AIC produces about 3000 copies of each single-sheet Liptan. Yet there are 13,527 farmers' groups with 466,744 members in this one province (Biro Humas Deptan 1991:97-98). Three thousand copies of a publication are not even enough to provide each of the province's 1598 field extension workers with two copies -- one for themselves, and one to give away. Even the 30,000 copies of folders previously printed by the West Java AIC (Rusyana 1984) pale in comparison with the magnitude of the audience: 3337 field agents, and 37,177 farmer groups with 1,580,000 members (Biro Humas Deptan 1991:97-98). Two major problems thus face the AICs' media production programs: •
The number of titles produced each year is inadequate.
• Inadequate funding means that print runs are too small to serve the intended audience of extension personnel and farmers. The audience thus by default becomes extension personnel alone. Media development process Each year, the provincial Dinas offices determine the topics of forthcoming media materials to be produced by the local AIC. AIC SMSs then search for information on these topics. Respondents said they used Dinas information if this was available; they also sought information from the AIC library and nearby research institutes and universities. The SMSs prepare a draft in consultation with the Dinas personnel, and submit it for to the Dinas and their AIC colleagues for approval before going into production. In a scheme similar to that used for researcher civil servants (see next chapter), extension personnel receive credit points that can be used for promotion purposes if they author extension publications or teach courses. However, the number of credit points awarded per publication or course is minimal, and the limited number of titles produced or courses offered restricts opportunities for extensionists to amass credits. This has two effects:
43 • It restricts opportunities for advancement among extension personnel, markedly lowering morale. • Extension personnel and AICs have little motivation merely to reproduce research publications under an AIC imprint, since they would gain no credit points by doing so. Despite this, AICs do occasionally republish research publications directly. The Medan AIC, for instance, has adapted and shortened an AARD technical manual on sheep and goat raising and published it in its booklet series. Effectiveness of extension The largest, and probably the most critical, data-based study of the extension system is Hussein's (1986) dissertation on the Indonesian agricultural knowledge system. After interviewing farmers, extension personnel and researchers in West Java, Hussein concluded that an integrated knowledge system did not exist because of the weaknesses of linkages among its component parts. Other authors disagree on the effectiveness of the training and visit system in Indonesia. Some (e.g., Sukaryo 1983, Suryono 1985, Witjaksono 1990) see T&V as effective and a major improvement over the previous system, which was based at least in part on enforcing farmers' compliance with government programs. Others have criticized the system on various grounds. These include: • Field extension workers are inadequately trained (Suhardjo 1989, Subarma 1985, Sutjipta 1982, Nur 1986). • Farmers are little involved in planning meetings, and attendance at meetings with field agents is poor (Subarma 1985). • Meetings are poorly timed and held at inconvenient locations (Subarma 1985, Nur 1986). • Communication of extension information among farmers is poor (Hussein 1986, Surialaga 1984, Holian 1990). • Field agents have too many farmer groups to serve (typically 16 in a two-week period) (Nur 1986, Hubeis 1987). • T&V has had little impact on yields of 5 major non-rice crops (based on a cursory, macro-level, time-series analysis) (Hussein 1986:147-152). •
Farmer groups tend to include only wealthier farmers (Hussein 1986).
• Farmer groups often exist in theory rather than in practice (Hussein 1986, Hubeis 1987). Ministry officials will admit this in private. Despite the problems associated with the current system, there is little discussion in
44 the literature or among the people I interviewed in favor of radical change. The tone is more toward improving those aspects of the system that need fixing, such as revising the organizational structure of the ministry and adjusting the responsibilities of field agents. Both these have in fact been recently put into effect: • Responsibility for district extension staff and local Rural Extension Centers (BPPs) has been handed from the Ministry of Agriculture to local governments (which come under the jurisdiction of the Ministry of the Interior) (SK Bersama 1991). • Oversight of provincial and district extension specialists has been moved from Bimas to the Secretariat-General, in preparation for a further move to the extension arms of the relevant Directorates-General. • Instead of covering all topics, field agents will be responsible for extending information on a single commodity group, such as livestock of food crops. While the impact of these changes had not yet been felt at the field level at the time of this study, many of the ministry and local staff interviewed felt some uncertainty as to their future status. Further change may be on its way as the ministry evaluates results of a major attempt to train farmers in integrated pest management (Röling, Gallagher and van de Fliert 1991, Stone 1992). The mass media Four main types of mass media concern this study: broadcast media, general newspapers, the agricultural press, and commercially published books. Broadcast media Most radio and television stations are controlled by government corporations, though privately owned television stations have recently begun to operate. Both radio and television carry programming related to agriculture; these include the radio program Siaran Pedesaan ("Rural Broadcast") and television programs such as Dari Desa ke Desa ("From Village to Village"). AICs and AARD research institutes are occasionally featured on these programs: for instance, one 1992 TV program featured researchers at AARD's Lembang Research Institute for Horticulture responding to viewers' questions about vegetable growing. The field extension agents I interviewed said that these programs were not broadcast at convenient times for them to view. AICs prepare audiocassettes for radio broadcasts, and district Dinas offices have regular though infrequent opportunities (once every three months, according to an anonymous informant in a district office in North Sumatra) to collaborate in producing local radio broadcasts. Hussein (1986:212, 241-243, 450-454) concludes that radio has great potential for disseminating information on agriculture, with more than 80% of farmers in West Java
45 listening to agricultural radio programs at least once a week. I was unable to find any direct evaluation of the use of AIC or AARD materials by the broadcast media or of the effectiveness of these materials when broadcast. General newspapers Indonesia's general newspapers can be divided into two groups: large metropolitan dailies such as Kompas, and Surabaya Post, and small, non-metropolitan newspapers. Both carry agriculture-related information. I know of no analysis of the content or effect of this type of coverage in the Indonesian press. In the metropolitan dailies with which I am familiar, most agricultural coverage falls into two broad categories: • Hard news, such as the release of a new rice variety, the achievement of a record crop yield, or the inauguration of a new irrigation scheme. • Opinion, as expressed through guest articles by policy makers, academics, and essayists. The coverage of technical information in these newspapers is thus small. The Ministry of Information subsidizes the production and rural distribution of several dozen non-metropolitan newspapers through its Koran Masuk Desa ("Newspapers Enter the Village") program. An example of a newspaper supported in this way is Mitra Desa, a farmers' paper published in Bandung, West Java. Agricultural press The agricultural press includes the biweekly newspaper Sinar Tani and several magazines. Sinar Tani ("Farmer's Light," Figure 3.7) is an eight-page newspaper published twice a week by Duta Karya Swasta, a firm with links to the Ministry of Agriculture (my source on this is a member of the paper's editorial board). The editorial board includes ministry staff, one of whom is a senior AARD official responsible for coordinating the Agency's communication activities.
46
Front cover of a 1991 issue, reproduced 65% of actual size. The original has 28 pages of text with occasional black and white photographs and diagrams. The cover and centerfold are in high-quality full color. Other AICs produce similar magazines with the same name. Figure 3.4 Buletin Informasi Pertanian, semiannual magazine produced by Lembang Agricultural Information Center, West Java.
47 Sinar Tani has a circulation of 48,000. Of these, 23,000 copies are sent to extension personnel and farmers, 19,000 to estate crops parastatals, and the remaining 6,000 to other subscribers. Only a small number of senior officials receive complimentary copies; all others are paid subscriptions for Rp 3,000 ($US 1.50) per month. According to district extension personnel, subscriptions are typically drawn from funds allocated for extension activities at each institution. Trubus (Figure 3.8) is Indonesia's leading horticultural magazine. Published monthly, it focuses on ornamental plants, fruits, and pets (including ornamental fish) and occasionally covers field and plantation crops and livestock. It retails for Rp 3,000 per copy and is available at newsstands. According to a consultant to the magazine, Trubus readers comprise two main groups: business people and hobbyists. Many readers in university towns are students. The circulation is 51,000. This informant said that sources for articles in Trubus include academics, experienced horticulturists, AARD researchers, and other specialist media, including foreign horticultural magazines. Other magazines include Setia Kawan (a magazine aimed at farmers and published by the same group as Trubus), Poultry Indonesia, Tumpang Sari, Tani Mukti, Majalah Pertanian (Hussein 1986:215), and the AIC magazine Buletin Informasi Pertanian (Figure 3.6). Hussein found these to be of minor importance individually as information sources for farmers, though 10% of his farmer respondents read at least one of them. Hussein does not report readership of such magazines by extension personnel. Commercially published books Several firms publish agricultural books in the Indonesian language. The majority are technical guides on such topics as growing maize and keeping rabbits. The number of university-level textbooks in Indonesian is increasing, though few are yet available. Several textbooks aimed at the agricultural secondary school market are also available. A measure of the availability of agricultural books can be gained by inspecting the stock lists of bookstores. I obtained lists from the three largest private bookstores stocking agricultural titles in Bogor, West Java -- home of the country's largest agricultural university and center of the government's agricultural research activity. I also obtained lists from the ESCAP CGPRT Centre, a United Nations institute in Bogor that distributes titles from several international agricultural research centers. A summary of these lists is presented in Table 3.4.
48
Front cover of a 1991 issue, reproduced 65% of actual size. The original has 28 pages of text with occasional black and white photographs and diagrams. The cover and centerfold are in high-quality full color. Other AICs produce similar magazines with the same name. Figure 3.5 Buletin Informasi Pertanian, semiannual magazine produced by Lembang Agricultural Information Center, West Java.
49 Table 3.4 Numbers and prices of agricultural books stocked by bookstores in Bogor, West Java, 1991.a Bookstore
Indonesian Titles
Wisma Batik
b c
Gunung Agung P T Filia
d
English
Price (Rp)
Titles
Mean
Std Dev
618
5420
8149
301
4071
4225
No list available e
CGPRT Centre
0
-
-
Price (Rp) Mean
Std Dev
405
98,346
87,966
0
-
-
296
28,218
49,656
177
14,405
13,198
a
Sources: Stock lists provided by bookstores.
b
Caters mainly to university.
c
Agriculture section of general bookstore.
d
Mainly old stock.
e
United Nations institute. Stocks international research center titles only.
This table shows that a large number of reasonably cheap Indonesian-language books are available -- at least in metropolitan areas like Bogor. But the prices must be compared with the salary levels of Indonesian agricultural professionals: a recent graduate beginning work as an SMS or junior researcher earns about Rp 120,000 ($US 60) per month. This is not enough to maintain commonly accepted living standards for people with similar education levels. An SMS is thus likely to purchase even the moderately priced Indonesian books only rarely. Most English-language books are far too expensive for all but the wealthiest. Summary The Indonesian agricultural extension system is large and complex. It needs to cover a wide range of commodities in an extremely diverse and dynamic physical and social environment. Indonesia's farmers, whom this system must serve, are very numerous, diverse, and scattered across thousands of islands. Rapid population growth and the growing urban-industrial sector further add to the urgency of providing improved technology and information to this audience. The country's lack of strong private and university sectors that can do this puts a huge burden on the Ministry of Agriculture to provide these services. Research and extension are separated administratively within the ministry, as are the major commodity groupings from each other. This allows efficient administration of the various functions related to research and extension of each commodity. But it also presents challenges to smooth communication between research and extension. This is the subject of the next chapter.
52
CHAPTER 4 AGRICULTURAL RESEARCH IN INDONESIA Introduction This chapter discusses agricultural research in Indonesia and the various methods by which information flows from research to the extension organization. I first describe the research institutions, focusing on the Agency for Agricultural Research and Development (AARD) and its activities. I then briefly describe the process of developing an agricultural technology recommendation, as prescribed by ministerial decrees. This leads to a discussion of characteristics of the highly centralized recommendation-generation process, and a description of decentralized alternative procedures. Are research-extension linkages a problem in Indonesia? To answer this question, I next review the literature on linkages in Indonesia, weighing the evidence for and against poor linkages, and describing recent efforts to improve ties. I then describe the communication activities of AARD research institutes. I look first at the three types of communication departments within AARD. I then briefly discuss the three main audiences of these departments, before turning to their communication activities and the constraints they face. AARD and agricultural research Agricultural researchers are one of the largest groups of scientists in Indonesia. Like the extension system, the research system is rather complex, and it is necessary to simplify much in the brief description here. Table 4.1 lists sources of information on the research system, primarily on AARD, its largest component. This section first briefly describes other organizations involved in agricultural research. It then describes in greater detail AARD and the process of planning and performing research.
53
Table 4.1
References on the Indonesian agricultural research system.
AARD 1984a
•
AARD 1984b AARD 1985b
Cook and Wharton 1984 * Hadiwigeno 1991
•
Hussein 1986
AARD 1986
IARDJ 1986
AARD 1987
ISNAR 1981
AARD 1988
Nestel 1985
Abdurachman et al. 1991
Syam and Mundy, in press
* Badan Litbang Pertanian 1991 •
Data-based study
*
In Indonesian
Ward 1985
Other research organizations The bulk of formal agricultural research in Indonesia is conducted by AARD. This section will therefore concentrate on this agency. An incomplete list of other ministries and institutions involved in agricultural research and development work, and a brief summary of their relevant focus areas, are given below. • Research on estate crops (rubber, oilpalm, coconut, sugarcane, tea, coffee, and cacao) is performed by eleven research institutes under the auspices of the Indonesian Planters Association for Research and Development (IPARD, Asosiasi Penelitian dan Pengembangan Perkebunan Indonesia, AP3I). This is the research arm of a group of parastatal and private plantation corporations. The director-general of AARD is a member of the Board of Trustees of IPARD (Figure 4.1). • The Sugarcane Research Institute at Pasuruan also conducts research on sugar. It is under the direction of the Management Board for Sugarcane. • Universities combine a wide range of agricultural research with their teaching activities. Major universities with faculties of agriculture include Bogor Agricultural University (Institut Pertanian Bogor, IPB) in Bogor, West Java; Gadjah Mada University, in Yogyakarta; Pajajaran University, in Bandung, West Java; Brawijaya University, in Malang, East Java; Hasanuddin University, in Ujung Pandang, South Sulawesi; and Andalas University, in Padang, West Sumatra.
54 • The National Biology Institute of the Indonesian Institute of Sciences (Lembaga Biologi Nasional-Lembaga Ilmu Pengetahuan Indonesia, LBN-LIPI) concentrates on the description and conservation of plant genetic resources and minor crops not covered by AARD. It is located in Bogor. • The Ministry of Research and Technology conducts research on biotechnology and genetic engineering. • Research on post-harvest processing and crop use is performed by the Institute for Research and Development of Agro-Based Industries (IRDABI, Balai Besar Industri Hasil Pertanian, BBIHP) of the Ministry of Industry. IRDABI is located in Bogor. • The National Logistics Board (Bulog) conducts research on large-scale storage and post-harvest processing of food commodities through its Bureau for Research and Development of Logistic Systems. This is done mainly at the Food Technology Research and Training Center (FTRDC, Balai Penelitian Teknologi Pangan, BPTP) at Tambon, West Java. • The Regional Center for Tropical Biology (Biotrop) of the Southeast Asia Ministers of Education Organization (SEAMEO ) implements a program on the biology of tropical agricultural pests. This includes work on storage insects, pre- and post-harvest mycology of food crops, rodents, and weed ecology. Biotrop is sited near Bogor. • Multilocational field trials of agricultural technologies are performed by provincial Kanwil offices and district and provincial technical agricultural services (Dinas offices) (SK Mentan 1989). • Various private agro-chemical and seed companies develop and test products in Indonesia, often in cooperation with AARD research institutes. Agency for Agricultural Research and Development After Indonesia achieved independence in 1945, responsibility for research in agriculture was assigned to the various directorates-general of the Ministry of Agriculture: Food Crops, Estate Crops, Forestry, Fisheries, and Animal Husbandry. Recognition that research and development required stronger coordination led in 1974 to a presidential decree reorganizing these institutes under a new body, the Agency for Agricultural Research and Development (AARD, Badan Penelitian dan Pengembangan Pertanian, Litbang) (Baharsjah 1985). The Agency's organizational structure has undergone a number of changes since that date, most notably in the establishment of several central research institutes that coordinate the work of their subordinate research institutes in 1979; the creation of a separate Central Research Institute for Horticulture; the transfer of responsibility for forestry research to the newly created Ministry of Forestry in 1983; and the transfer of research on estate crops to the independent Indonesian Planters' Association for Research and Development (Schumacher et al. 1991:2, AARD 1985b:4). The agency now has ten Centers with coordinating, research, or technical functions,
55 16 research institutes, 43 research sub-institutes, and numerous experiment sites scattered around the country (Departemen Pertanian 1991) (Figure 4.1).
56
Figure 4.1 Organizational structure of the Indonesian Agency for International Research and Development.
57 The five central research institutes (for food crops, horticulture, industrial crops, livestock, and fisheries) conduct a limited amount of research themselves and manage multidisciplinary projects involving several lower-level institutes. Each coordinates a number of specialist research institutes working on their particular group of commodities. Each of these subordinate institutes has the national mandate for research on a particular commodity or agro-climatic type. In addition, the centers for soils and agro-meteorology and for agrosocio-economic research conduct research as well as coordinate studies at other AARD units in their topic areas. AARD has made major strides in developing its physical facilities and human resources. In 1975, shortly after its founding, the agency employed only 16 Ph.D. holders; by 1988, this figure had grown to 179 through an aggressive training program (Jahi 1991:7). The goal was to reach a total of 500 Ph.D.s by 1992 (but note that this total includes estate crops institutes no longer part of AARD) (AARD 1987:106). Of AARD's units, three concern this study: the food crops and livestock research branches, and the Center for Agricultural Library and Research Communication (CALREC). These are described below. Central Research Institute for Food Crops CRIFC (Pusat Penelitian dan Pengembangan Tanaman Pangan, Puslitbangtan) is the largest of the central institutes. It coordinates six Research Institutes for Food Crops (AARD 1985b, 1988:18; CRIFC 1986). These manage a total of 13 sub-institutes and 37 experiment farms. CRIFC's six research institutes are: • Sukarami Research Institute for Food Crops, in the highlands of West Sumatra, performs research on crops grown on dryland under a wet climate, and on high-elevation rice. • Maros Research Institute for Food Crops, in the coastal lowland of South Sulawesi, studies dryland crops under a dry climate. • Sukamandi Research Institute for Food Crops, in the northern coastal plain of West Java, focuses on wetland cropping, primarily of rice. • Malang Research Institute for Food Crops, in the uplands of East Java, researches non-rice food (palawija) crops. • Banjarbaru Research Institute for Food Crops, in the tidal swamps of South Kalimantan, studies cropping in the swamplands that cover much of southern Kalimantan and eastern Sumatra. • Bogor Research Institute for Food Crops, the largest of the six institutes, has research programs in many of these same areas. It supports the other units through pioneering and fundamental research and commodity analysis. Central Research Institute for Animal Sciences CRIAS (Pusat Penelitian dan Pengembangan Peternakan, Puslitbangnak) coordinates two research institutes, which in turn manage six sub-institutes (AARD 1985b, CRIAS undated):
58 • The Research Institute for Animal Production (RIAP, Balai Penelitian Ternak, Balitnak), located in Bogor, focuses on the production, reproduction, post-harvest handling, and farm system management of livestock and livestock feed. • The Research Institute for Animal Diseases (RIAD, Balai Penelitian Veteriner, Balitvet) is also sited in Bogor. It conducts research in livestock health. Center for Agricultural Library and Research Communication CALREC (Pusat Perpustakaan Pertanian dan Komunikasi Penelitian) houses the 80,000-volume national library for agricultural sciences, a major national asset. It also publishes agricultural journals in English and Indonesian and coordinates the external communication of other AARD units. CALREC is the lead institution in the Research-Extension Linkages project, a major effort to improve AARD's ties with its primary clientele. CALREC communication activities are discussed later in this chapter. Research planning in AARD The research process in AARD begins when a scientist proposes a research project to his or her research program coordinator. The coordinator revises the proposal, and passes it on to the institute director for prioritization, and so on up through AARD (AARD 1984a:40, AARD 1984b:25, Badan Litbang Pertanian 1991:45-46). Review teams have identified several weaknesses in this process (AARD 1984a:40, 1984b:26). These include the difficulty of coordinating research programs conducted at multiple sites and under the control of more than one institute director, and the danger that research topics reflect individual scientists' interests rather than national or AARD priorities. Since these comments were published in 1984, AARD has substantially strengthened its research programming activities, even creating a separate institute (Center for Agricultural Research Programming, CARP, Pusat Penyiapan Program Penelitian, Pusgram) to deal with it. Another positive step is that CALREC has begun soliciting ideas from extension personnel and local officials for research topics and information requirements; it passes these on to research institute administrators (Pustakom 1990). Nonetheless, the task of identifying research problems is a primary responsibility of AARD, in consultation with relevant directorates-general, the national Bimas secretariat, and AAET (Abbas, Tjitropranoto, and Yakub 1989:56). There is limited potential for input from extension agents and farmers in the setting of research priorities. Hussein (1986:409) found that extension workers and farmers were the least and second-least important sources of research ideas for scientists at three AARD institutes. The most important sources were research headquarters, informal discussions with colleagues, and the scientific literature. Researchers do receive requests from local farmers to solve problems, and those outside Bogor come under pressure from the provincial Kanwil to tackle immediate needs. Some researchers and administrators (including the members of an external review team [AARD 1984b:41, 43]) seem to regard such requests as distractions from AARD goals
59 rather than as vital links to its clientele. They saw many such problems as having limited applicability outside the immediate area, and as diverting scientists from addressing urgent national objectives. The research institutes face an inherent conflict between their nationally mandated research priorities and immediate local needs. There is clearly a need for balance between them if both local credibility and national impact are to be maintained. Research activities Because of the size and variety of AARD's research program, I will focus here on the two areas of relevance to this study: food crops and livestock. Readers interested in other areas are referred to AARD's quinquennial report series (AARD 1987). Food crops research AARD's food crops research activities are organized by commodities and problem areas: rice; palawija (non-rice food crops: maize, sorghum, soybean, peanut, mungbean, pigeonpea, cassava, and sweet potato); wheat; problem soils; farming systems; water management; and post-harvest technology and quality (CRIFC 1986). Research is conducted using CRIFC's own facilities, as well as on farmers' fields and special, often inter-institutional, projects such as farming systems research (e.g., SWAMPSII 1991). The output of one type of food crops research -- varietal breeding -- is easy to measure. In the period 1981-1986, about 30 new rice varieties were released in Indonesia, most developed by CRIFC breeders. In the same period, breeders released ten new varieties of maize, eight of soybean, five of peanut, and six of mungbean (AARD 1987:818). Several of these varieties have proved extremely popular and have been rapidly adopted by farmers. Livestock research Livestock research is organized into ten program areas: cattle and buffalo production; sheep and goat production; poultry, eggs and pig production; animal nutrition; breeding; farming systems; large ruminant diseases; small ruminant diseases; poultry diseases; and the diseases of other animals (CRIAS undated). A good example of the type of livestock research performed is the Small RuminantCollaborative Research Support Program (SR-CRSP). Funded by USAID and the Indonesian government, SR-CRSP researchers have studied the genetics and breeding of goats and sheep, their feeding and nutrition, labor needs and marketing. They developed methods for raising sheep under rubber trees, and established pilot projects to demonstrate the benefits of the improved methods to farmers, extension workers, and policy makers (SR-CRSP 1990).
60
Technology recommendations The process of turning a research finding into a recommendation and disseminating it to farmers is a complex one. The following discussion is based on the most recent relevant ministerial decree (SK Mentan 1989), an interpretation of this (Abbas, Tjitropranoto, and Yakub 1989), and discussions with numerous AARD researchers and administrators. The dashed lines in Figure 4.2 summarize the main communication channels prescribed by these sources.
61
Figure 4.2 Simplified information flows from research to extension within the Ministry of Agriculture for a hypothetical food-crop pest-control technology.
62 AARD's task is broadly defined as agricultural research. The agency is not allowed to make technology recommendations: this is the prerogative of the relevant directorategeneral at the national level and the Dinas offices at the local levels. Instead, researchers are to provide information that these other units can evaluate and base recommendations upon. Briefly, AARD researchers identify research problems, perform research, and inform the relevant directorate-general in Jakarta of the findings. Senior AARD administrators and directorate-general officials together divide research findings into three groups: 1.
Results that can be used in formulating technology recommendations.
2.
Findings that need to be subjected to local verification trials.
3.
Results that still need to be adapted so as to fit local agroecological conditions. From here on, AARD is in a merely supportive role.
Technologies in the first group are deemed suitable for immediate release. The directorate-general informs provincial Kanwils, Dinas offices, and other relevant local bodies of the technology. These bodies then include it in local recommendations where relevant. Technologies in the third group are returned to AARD for further adaptation and testing. The directorate-general chooses technologies from the second group to subject to local verification trials. The directorate-general, AARD, Bimas, and AAET cooperate in managing such tests. The tests themselves are performed by extension subject-matter specialists on government or farmers' land. The results are communicated via workshops held by the managing institution in cooperation with the provincial Bimas secretary and AIC. If necessary, local testing can be conducted at the provincial and district levels by the provincial Dinas office. The provincial AIC staff select findings from the AARD research and local trials for inclusion in a technology packet (see the discussion at the end of this chapter). The AIC then produces and distributes these packets. District Dinas offices select those packets suited to their areas, and run technology demonstrations and field days in conjunction with field extension agents and farmers. Farmers are then expected to disseminate the technology autonomously (curiously, Abbas, Tjitropranoto, and Yakub 1989 make no mention of how the training-and-visit system is integrated into this process). Ensuring feedback is the primary responsibility of provincial and district Dinas offices and the program extension agents.
63
Characteristics of the recommendation process While the above brief description obviously grossly oversimplifies reality, several things are clear. (I will discuss two exceptions allowed by the model later.) • Gatekeepers Most obviously, an item of information must pass through a large number of gatekeepers if it is to reach its destination. It is expected to rise from the individual researcher up through a series of research administrators to the AARD leadership. It must then cross over into another bureaucracy, and descend through another series of gatekeepers and across institutional boundaries to the end users. • Delay and distortion At each stage, the information must be processed to suit the needs of recipients in the next stage: summarized and translated from research language to policy language, converted into guides for verification trials, and translated and reproduced to form extension training materials. The possibilities for delay, loss and distortion are legion. • System size A large number of people and organizations, at various levels of government and scattered throughout the country, are involved throughout this process. Smooth communication, including personal consultations, regular distribution of publications, and systematized flows of messages are vital to the process. This communication is sometimes not as smooth as could be wished, given the lack of telecommunications and the formidable geographic and organizational barriers that must be surmounted. • Centralization The information flow is extraordinarily centralized. In theory at least, all relevant information must pass through the hands of a few key policy makers in Jakarta -- and they must discuss it at a meeting. It is unrealistic to expect them to understand all the nuances of the technology or appreciate all the possible local ramifications of its use. • Responsibility The direct responsibility of AARD ceases when technologies are identified for local testing. Even this is done by senior administrators rather than individual researchers. The individual researchers and senior administrators thus have little opportunity or incentive to ensure that technology practices are in fact tested adequately or disseminated to farmers. In fact, no one in the whole system appears to have any stake in the successful adoption of a technology. • Overlap There is considerable fear within the research and extension system of invading the "turf" of another institution, or having one's own turf invaded. Thus researchers are extremely wary about being seen to perform extension work, and many in extension agencies are sensitive to AARD attempting to take over such a role. Similarly, AARD scientists resent attempts by extension to perform research. This is despite the desirability of overlapping research and extension involvement in all aspects of technology generation and dissemination (e.g., Abbas 1991:13, Siwi and Mundy 1986).
64 • Feedback Despite the inclusion of "feedback" in the model, there is little opportunity for local involvement. There is little scope for local initiative in the process, and limited possibility for good, or bad, news to filter up the chain of institutions to the national level and then down again to AARD researchers (Widjono 1986:62). • Horizontal links There are few opportunities built into the model for horizontal communication among, say, farmers, extension personnel and researchers working in the same area. While such communication does of course occur in practice, especially in field research projects such as AARD's farming systems research activities, the majority of communication remains vertical. Alternative information routes The ministerial decree (SK Mentan 1989) allows two exceptions that give sufficient flexibility for it to be workable. These are as follows: • If a technology is locally applicable and is not in conflict with directorate-general guidelines, it may be adopted as a local recommendation while awaiting a decision by the directorate-general (SK Mentan 1989:§7.3). This allows local extensionists to tell farmers to do what works locally rather than what a bureaucrat in an office in Jakarta thinks ought to work. • Two-way information flows between research and extension are encouraged (SK Mentan 1989:§12). This opens up numerous possibilities for horizontal communication. For instance, it encourages researchers to visit farmers' fields to identify problems, allows research institutes to publish materials with an extension audience in mind, and permits them to send publications directly to extensionists instead of having to go through the bureaucracy. These two sections in the ministerial decree effectively create a fourth category of research finding: technology that is not important enough nationally to pass through the directorate-general. Technologies with a major potential national impact (for instance, a new rice variety or a vaccine against Newcastle disease) must pass through the directorategeneral. This is because they may greatly increase requirements for inputs such as seed, agro-chemicals, or pharmaceuticals, or may have a major effect on farmer incomes or marketing arrangements. Other technologies (for instance, changing the materials from which chicken nesting boxes are made) have a much smaller potential impact but still may be valuable. Such technologies are likely to be communicated through the horizontal channels. These are denoted by dotted lines in Figure 4.2.
65
The literature on linkages The linkages between research and extension in Indonesia have been subject to a wide range of comment. Table 4.2 lists publications that have dealt with this topic in the last decade. With few exceptions (most notably Wardojo 1990, see below), authors either note that research-extension linkages are poor, or accept this implicitly in seeking to improve them (e.g., Baharsjah 1985:30, Padmanagara 1985:137, Hadiwigeno 1991:22, Widjono et al. 1989).
66
Table 4.2
References on research-extension linkages in Indonesia.
Research communication, researchextension links
•
*
AARD 1985a
Siwi and Mundy 1986
* Abbas et al. 1989
•
Andyantoro et al. 1989
•
Baharsjah 1985
•
*
SK Mentan 1989
*
Sophia 1988 Sunarno 1983
Eriyatno 1988
*
Sunarno et al. 1989a
Ewell 1989
*
Sunarno et al. 1989b
Hussein 1986
*
Sunarno et al. 1989c
*
Syam and Widjono undated
* Jahi 1991 * Kanwil Deptan Sulsel 1990 •
Simandjuntak 1988
* Mansur 1989
SYGAP 1989 •
Tjitropranoto 1991
MORIF/AARP undated •
Mundy et al. 1991
•
Mundy et al. 1992
Tjitropranoto and Syam 1989 •
NIA and DGRD 1982
*
Wardojo 1990
*
Widjono et al. 1989
*
Widjono 1990
Padmanagara 1985 * Pustaka 1990a
Technology adoption
* Pustaka 1990b
•
IDRC 1986
* Pustakom 1991a
•
Ludgate and Priyanti 1988
* Pustakom 1991b
•
Thomas 1989
Sadikin 1982
•
Wahyuni et al. 1990
Schumacher et al. 1991 * Sejati 1991 • Data-based study * In Indonesian.
Evidence against poor linkages
*
Widjono 1989
67 The comments by Wardojo (1990:10-12), the current Minister of Agriculture, merit attention. At a seminar at Bogor Agricultural University in 1990, he stated that poor research-extension linkages are not a problem in Indonesia, "because if they were, we [Indonesia] would never have achieved the selfsufficiency in foodstuffs that we have maintained since 1984, and in the spread of new crop varieties, we would not be able to disseminate new varieties to every corner of Indonesia in a space of only three seasons" [my translation]. Wardojo admits that research-extension linkages were a problem before the First Five Year Plan (pre-1967), but that this had been overcome by increasing the number of SMSs to bridge the research-extension linkage, boosting the number of field agents, building Agricultural Information Centers in each province, using researchers as instructors in training courses, and encouraging private companies and farmers to multiply new varieties. Indonesia's attainment of rice self-sufficiency was indeed impressive, as are the speed with which new rice varieties are made available and farmers' willingness to adopt them. In wetland rice, at least, research-extension links seem to work. But production gains for most other commodities have been less spectacular (Table 3.1). Why this difference? One possible reason is the importance of rice to Indonesia and Indonesians. Rice has a high political profile, an enormous amount of resources have been poured into boosting its production, and the crop receives the lion's share of attention from research, extension agencies, and farmers. Other commodities enjoy much less attention, though soybeans and chickens have more recently attained a high political profile and have been given a large amount of extension resources through Bimas programs. A second reason relates to the physical environment. Wetland rice is grown under unique conditions: in a shallow, artificial swamp. The land is flat, physical soil characteristics are relatively unimportant as a determinant of yields because of the ameliorating effect of flooding (De Datta 1981:48, Grist 1986:24), the standing water suppresses weeds and precludes drought stress, and the rice is typically monocropped. Growing conditions are relatively uniform from one end of a rice field to the other, and even from one end of Indonesia to the other. This means that it is relatively easy to develop rice varieties and fertilizer and pest control recommendations that apply to large areas. Such conditions do not apply to other crops and livestock. (I am indebted to an anonymous interviewee for pointing this out.) They are grown on sloping land with a wide range of soil types and climatic regimes, and are subject to a wider range of weeds, pests, diseases, and environmental stresses at any stage of their growth cycle. They are typically grown together with numerous other species. Conditions may vary markedly from one end of a plot or barn to another, and from one day to the next. The job of researchers and extension personnel is far more difficult in such situations, and the requirements for communication among them are more complex. While possibly valid for wetland rice, Wardojo's explanation may thus not apply other commodities. Let us now turn to the evidence supporting the contention that linkages are poor.
68
Evidence for poor linkages Despite the attention given to the topic, most of the evidence for poor researchextension linkages remains anecdotal. There have been relatively few empirical evaluations; AARD institutes have conducted few such studies themselves, though this is being planned by CALREC (Schumacher et al. 1991:46). An exception is the study of a farming systems research project in West Java (Widjono et al. 1989), which looked at interpersonal communication among project researchers (mostly junior scientists) and extension personnel. It concluded that while the two groups worked and socialized together, the scientists rarely discussed research findings with their extension colleagues because they felt they had no right to do so. They saw this as the prerogative of their senior research colleagues, who were based outside the project area. AARD scientists have conducted several evaluations of technology adoption, mostly associated with farming systems research projects (e.g., IDRC 1986, Ludgate and Priyanto 1988, Wahyuni et al. 1990). The findings from these have been mixed; perhaps the main lesson to be drawn from them is the importance of government support in the form of credit and timely input supplies if farmers are to adopt new agricultural technologies. Two external reviews have been completed of AARD's communication activities (AARD 1985a, Schumacher et al. 1991), pointing out the need to strengthen links. External reviews of other AARD activities (e.g., of horticultural and non-rice food crops research) have also commented on this need (AARD 1984a, 1984b). Most of the published evaluations are theses; however, some have been of questionable quality, for instance testing uninteresting hypotheses, using poorly worded questionnaires, or applying inappropriate statistical tests. Mansur's (1989) survey of SMSs in South Sulawesi found no consistent relationships between the SMSs' personal characteristics and their use of three AARD publications. Simandjuntak's (1988) findings on the readership of a veterinary scientific journal by field extension agents must be treated with caution because of data and statistical validity problems: his questionnaire did not ask whether the respondents had ever received the journal (they are not sent it directly), and his analysis uses inappropriate statistical tests. Sunarno (1983) studied the reactions of SMSs in three provinces to various AARD publications: CALREC's well-illustrated, color publications, written in easily understood language; and technical reports and scientific journals published by the food crops research institutes. She found that the semi-popular publications were more readable and important for extension purposes than the scientific publications. While Hussein (1986:415-422) focussed mainly on the extension-farmer link, he also questioned researchers and SMSs about their communication behavior. He found that mutual visits were extremely infrequent: researchers made a mean of 0.5 visits to SMSs per year, while SMSs made 1.5 visits per year to researchers. More than half the researchers questioned never met with SMSs informally. Other types of interaction, such as producing written materials, conducting field days, making farm visits, and making and receiving visits
69 to or from other extension personnel, were also infrequent. Sophia's survey of food crops SMSs in West Java is probably the most useful study of this important group of AARD clients to date. Among her findings are (page references are to Sophia 1988): •
SMSs rarely read information on agricultural research findings (pp. 52, 117).
• Visits to information sources are infrequent: only one-third visited a research institute, university, AIC, or other institutional information source more than once every six weeks (p. 56). • An informal network among SMSs was their most important information source. Training, AICs, libraries, and research institutes were also relatively important; the mass media and universities were not (p. 58). • Major information needs were in extension methods, marketing, most aspects of legume cropping, and irrigation (p. 66). • SMSs wanted adaptive and applied research results rather than findings from basic studies (p. 72). • By an overwhelming majority, they preferred research information to be sent directly to them or to their home Dinas offices rather than to be made available only at AICs or research institutes (pp. 64, 73). Attempts to improve linkages Three notable recent developments are opening the way to improved linkages. One is farming systems research, in which AARD has been a world leader. Such projects allow researchers, extension personnel, and farmers to interact and solve problems together. They can be highly effective at communicating improved technologies to other farmers, extensionists and government officials (e.g., SYGAP 1989:12), though this is not always the case (Widjono et al. 1989). Some of these projects are managed jointly by AARD and extension agencies (Ewell 1989:22). Second, the Research-Extension Linkages Project, led by CALREC, is a major attempt to improve the two-way exchange of information among research and extension institutions (Tjitropranoto 1991). It includes workshops, publications, and research activities to identify and remove communication bottlenecks. It has prompted a spate of writing on linkages (e.g., Abbas 1991, Hadiwigeno 1991, Jahi 1991, Pustaka 1990a, Pustakom 1991a-b, Sejati 1991, Sunarno et al. 1989a-c, Widjono 1990). Much of this writing has resulted from a series of meetings to discuss linkages held in various locations around Indonesia (West Java, South Kalimantan, South and North Sulawesi, North and West Sumatra, and Bogor) attended by researchers, extension personnel, local government officials, and farmers. Third, and on a smaller scale, several individual research institutes have begun publishing materials aimed specifically at extension agents. These include:
70 • Illustrated manuals on sheep and goat management published by the Research Institute for Animal Production (Ludgate 1989). • Single-page research fact sheets called Sambung Litluh ("Research-Extension Link", Figure 4.3) published jointly by AARD institutes and projects and local agricultural authorities. Examples are those produced by Maros Research Institute for Food Crops in South Sulawesi and by the Upland Agriculture Conservation Project in Central Java, in cooperation with extension institutions and local governments (James H. French 1991, personal communication; Kanwil Deptan Sulsel 1990, MORIF/AARP undated).
71
Front side of a 1991 issue on using fodder troughs for livestock. Printed in black and white, with the masthead in red. Distributed by Ungaran Agricultural Information Center, Central Java. Figure 4.1 Sambung Litluh research fact sheet, produced by the Upland Agriculture Conservation Project, Central Java.
72 •
Technical guides on individual crops published by the food crops research institutes.
AARD research communication departments Each AARD central institute or research institute has a unit charged with the communication of research findings. The units' duties include managing the institute's library, providing in-house services such as producing slides for scientists' presentations, and managing communications with clients. They are coordinated by CALREC. CALREC Recognition of the importance of communication and the need to coordinate AARD's communication efforts (e.g., ISNAR 1981:5) led the National Library for Agricultural Sciences in 1990 to add the coordination of research communication to its mandate. It was renamed the Center for Agricultural Library and Research Communication (CALREC). CALREC produces a number of monographs and journals in both English and Indonesian, including the prestigious semiannual scientific journal Indonesian Journal of Crop Science. Serials intended for more general audiences include the quarterly Indonesian Agricultural Research and Development Journal in English, and its Indonesian equivalent, Jurnal Penelitian dan Pengembangan Pertanian. CALREC also publishes a bimonthly biological index and a semiannual list of abstracts of Indonesian agricultural research results. This publication effort is large and publication often lags several years behind schedule (Schumacher et al. 1991:12). Of particular note for this study is CALREC's bimonthly newsletter, Warta Penelitian dan Pengembangan Pertanian (often referred to as Warta Litbang) (Figure 4.4). This four-color, 12-page newsletter is the most widely distributed AARD publication. Previously 10,000, its print run is now 2500 to 5000, depending on the availability of funding. It is the only AARD publication regularly sent to Dinas offices at the district level. However, budget shortfalls sometimes mean that Warta Litbang is published late or not at all.
73
Front page of November 1988 issue, reproduced 65% of actual size. The original has high-quality, four-color printing. The lead story in this issue is on preserving soursop germplasm. Figure 4.2
Warta Litbang, AARD's bimonthly newsletter.
74 CALREC staff are relatively well qualified in communication fields. Several have obtained master's degrees through Bogor Agricultural University's graduate program in development communication. CALREC has held numerous training courses and workshops on research communication for staff of other AARD institutes and other parts of the Ministry of Agriculture. Central research institutes Until 1985, the only other operating communications units within AARD were at the central research institute level. The largest of these is at the Central Research Institute for Food Crops. From 1985 to 1990, this unit published 18 major book titles (mostly symposium proceedings), five issues of the occasional English-language periodical Contributions of CRIFC, and numerous research reports and brochures. Unlike most of its counterparts, the unit has its own mini-offset printing press, giving it considerable flexibility and the ability rapidly to produce single- or two-color publications. Again unlike most other units, several of the CRIFC unit staff have graduate degrees in communication; this has enabled them to conduct several training courses for communication personnel in other AARD institutes. The communication unit at the Central Research Institute for Animal Science has been less well supported. It lacks a printing press and adequate computing facilities. Nevertheless, it has been active in publishing symposium proceedings; between 1985 and 1991 it published 11 such volumes (Sejati 1991). It plans to renew publication of the review journal Wartazoa shortly. Research institutes Realization that research communication was being neglected led in the mid-1980s to the creation of Information Units in all AARD research institutes. These took over some of the functions of the existing communication departments at the central research institute level. For instance, the newly created information unit at Bogor Research Institute for Food Crops took over from CRIFC's communication department the publication of the prestigious Indonesian-language triannual scientific journal Penelitian Pertanian. The other food crops research institutes also began publishing their own journals: Pemberitaan Penelitian Sukarami by the Sukarami institute, Media Penelitian Sukamandi from Sukamandi, Penelitian Palawija from Malang, Agrikam from Maros, and Pemberitaan Penelitian Banjarbaru from Banjarbaru (Puslitbangtan 1990). The livestock research institutes publish the journals Ilmu dan Peternakan and Penyakit Hewan (Schumacher et al. 1991:28-29, Puslitbangtan 1990, Puslitbangnak undated). Nearly all AARD institutes now publish their own scientific journals (Tjitropranoto and Syam 1989:343).
75
AARD audiences Research institutes must serve three main clienteles: policy makers, scientists, and extensionists. They have so far concentrated mainly on reaching the policy makers and scientists. Policy makers Policy makers have been a priority because the communication units must respond to orders from above. The units have to provide the upper levels of the hierarchy with high-quality (and therefore costly) material, often on short notice. They also aim for visibility for their institute to ensure continued funding and attention. Scientists The communication units have also served scientists by publishing institute journals and symposium proceedings. There are a number of reasons for this. The agricultural literature in the Indonesian language is very small, and there has been a conscious attempt to expand and improve it. Scientists are familiar with the "scientific" style of writing, and find it easier to write a research paper than to summarize their findings for an extensionist's use. The communication units find publication of such journals easier for the same reasons. Researchers are keen to publish in such journals, as under Indonesian civil-service rules, their promotion depends largely on the number of scientific articles they write. Articles published in prestigious journals amass the author a larger number of credit points that can be used toward promotion. Scientific publications also contribute to the prestige of an institute, and so are likely to attract funds from administrators. Extension Providing information for extension has none of the benefits enjoyed by serving the policy maker and scientist clienteles. The prime task of AARD and its researchers ends once a research finding has been disseminated (Abbas, Tjitropranoto, and Yakub 1989:56). There is thus little commitment to (or funding for) publishing materials aimed at extension. Suitable publications for this audience are likely to be newsletters, brochures, technical and "how-to" guides, and summaries of a particular research area. None of these are "high-profile" or prestigious publication types, and scientists cannot achieve promotion by writing them. These factors combine to produce a preponderance of output in the form of scientific articles rather than formats more suited to extension and farmer audiences. AARD appears to serve policy makers and scientists relatively well. The extension clientele is only now beginning to receive serious attention. Nevertheless, the comments below about weaknesses in research-extension linkages should be seen in light of the multiple audiences
76 the research institutes must serve. AARD communication activities The most important channels for research communication are libraries, publications, scientific and consultative meetings, field days, on-farm research, training courses, and informal communication (Schumacher et al. 1991:22-44, Siwi and Mundy 1986, Tjitropranoto and Syam 1989:343-345). Libraries Each AARD institute has its own library, coordinated by CALREC. However, with certain notable exceptions, these libraries suffer from a lack of attention and funding from institute leaders, their facilities and services are under-utilized by researchers, and they fail to share scarce resources among themselves (Schumacher et al. 1991:24-24). Publications While there is considerable variation among institutes, most communication units seem to devote most of their energies to editing and producing publications. These include research journals, symposium proceedings, technical guides, annual reports, newsletters, and brochures. I discuss these in greater detail below. Scientific and consultative meetings Although they may be open for extensionists, these meetings are normally aimed at researchers and policy makers. They therefore often do not provide the type of information that extensionists can use in solving problems in the field. Moreover, many are ad-hoc in nature and lack coordinated follow up (Schumacher et al. 1991:37). Meetings are, however, an important means of exchanging information with policy makers, including the extension hierarchy. Field days Exhibitions and open days at research sites offer extensionists the opportunity to see the use of new technology in the field and to discuss directly with the researchers. They are most often held as part of farming systems research projects (Tjitropranoto and Syam 1989:344). However, they are held infrequently, only a limited number of topics can be covered at a time, and logistics and cost severely restrict the number of extensionists who can attend. The discussions are not published, and so there is no permanent record that can be referred to at a later date or by other extensionists.
77
On-farm research AARD has several on-farm research projects, particularly in a farming systems research context. However, information exchange among project researchers and extension personnel may not be as great as is sometimes supposed (Widjono et al. 1989, see above). Projects sometimes fail to take full advantage of their communication possibilities, and the number of people that can visit a site is limited by logistics and cost. Training courses AARD institutes occasionally hold training courses for extension personnel, and scientists are sometimes invited to make presentations to courses held at the ministry's inservice training centers. However, both are rather infrequent: researchers participate on average in one such course less than once every five years (Hussein 1986:422). Constraints to research communication Research communication by AARD institutes face a number of constraints (Schumacher et al. 1991:32-33). These are discussed below. Publication authorship and production Publications are produced irregularly or infrequently. This means that research results may be untimely when they reach their audience (Schumacher et al. 1991:32). The promotion system based on credit points militates against researchers' collaborating on joint publications, since the number of credits allocated per publication is divided up (unequally) among the authors. The same is true of collaborative articles between researchers and extension personnel (Schumacher et al. 1991:32). The format of publications is not standardized, adding confusion as to what is editorially acceptable. Some journals change their format frequently (Schumacher et al. 1991:2). Communication staff status and skills Senior staff in the institutes' information units are often drawn from among the institute scientists, so have technical backgrounds in science but limited skills in or commitment to communication. In some institutes, frequent changes in unit leadership lead to lack of continuity and momentum. In others, the information unit is even seen as a convenient backwater for placing personnel unwanted elsewhere (anonymous informant 1991). Many staff of institute communication departments have limited training in communication. There is a shortage of staff with skills in such areas as editing, publication design, desktop publishing, publication management, photography, and graphics.
78 Scientist authors may view editors as gatekeepers to be assuaged or bypassed rather than as allies in the struggle for audience understanding (Widjono 1990). Tensions can be heightened if the scientist regards the editor as having low status because of poor qualifications. Under such circumstances, it can be difficult for editors to require changes in a manuscript, enforce deadlines, and maintain compliance with manuscript submission requirements. Communication policy and funding There is no systematic policy for working with the mass media (Schumacher et al. 1991:34). Two agricultural periodicals, the newspaper Sinar Tani and the magazine Trubus, receive nationwide distribution (Figure 3.7 and Figure 3.8). In addition, the government subsidizes many rural newspapers through the Koran Masuk Desa program. But AARD has no policy for working with these and other media. They do not produce press releases, since this is seen as the prerogative of the Ministry of Agriculture's public relations department, and communication unit staff lack skills in this area. A number of well-known agricultural researchers (mainly at universities) do write for newspapers fairly regularly, but their articles tend to discuss policy questions rather than provide technical details. Nor do the communication units engage in public relations activities, except for handling visitors and holding occasional exhibitions of research findings. The publication budget is inadequate and often fails to cover the cost of distribution. Print runs and the speed of distribution depend on the budget available rather than the potential needs of the audience. Many copies remain undistributed because of insufficient funds (see discussion below). AARD publications must be given away free of charge to qualified individuals (such as extension personnel) and institutions (such as university libraries). There is no provision under government accounting rules for them to be self-financing. While some institutes do sell some publications from their own premises or through local bookstores, the number involved is very small and there is no attempt at marketing them to a wider audience. This is despite the potentially large audience for some publications, such as major symposium proceedings or textbooks.
79
Publication distribution AARD publication print runs are typically small. For instance, the Central Research Institute for Food Crops typically produces about 1000 copies of its publications, the number depending on the size of the publication and the amount of budget available. They are distributed to CRIFC researchers, plus a mailing list of 413 addresses, including ministry officials in Jakarta, university libraries, and provincial-level agricultural officials and institutions (heads of Kanwils, food crops Dinas offices, SMS provincial coordinators, AICs, Seed Centers and Seed Certification Centers, provincial planning offices, and agricultural training centers) (Puslitbangtan 1991). They are not sent to any institution at the district level, (e.g., to district Dinas or Bimas offices), to individual SMSs, or to agricultural high schools The situation is similar for animal science publications. Print runs of CRIAS publications range from 300 to 1650. CRIAS has a mailing list of about 550 addresses, including ministry officials in Jakarta, provincial institutions, and livestock Dinas offices in some (though not all) districts (Puslitbangnak 1991). The mailing list of the veterinary institute contains 181 addresses outside the institute itself; a total of 500 copies of the scientific journal Buletin Balai Penelitian Veteriner are printed (Balitvet 1991, Simandjuntak 1988:58). The animal husbandry institute's mailing list contains 161 names (Balitnak 1991). The only AARD publication with a print run large enough to be sent to all SMSs is the bimonthly newsletter Warta Litbang, published by CALREC (Figure 4.4). CALREC maintains a large database of addresses, but many of these are out of date. Mailing lists are maintained either on paper (with handwritten corrections) or in a word processing file. There has been an attempt by CALREC to train research institute personnel in database management (Haryani and Mulyati 1990), but such efforts have been limited. Limited publication dissemination budgets mean that some publications are not mailed out immediately (Schumacher et al. 1991:33). This reduces their timeliness and negates the purpose of producing the publication in the first place. AARD has recognized that the problem of publication distribution is severe (Schumacher et al. 1991:32, Tjitropranoto 1991:6). As described above, it has three causes: • Limited funding means print runs are too small to send copies of publications to everyone that needs one. •
Adequate mailing lists are not maintained.
• Limited budgets mean some publications remain in storage instead of being distributed to users. The first of these problems is also faced by AICs in their publication program (see the earlier discussion of this).
80
Summary Research-extension linkages in Indonesia are complex. The research and extension systems are separated into different administrative divisions within the Ministry of Agriculture, and the official process of developing technology recommendations is highly centralized. Direct contacts between researchers and extension personnel are limited. All AARD institutes have units responsible for fostering the flow of research-based information to extension and other audiences. These units use a variety of dissemination methods, but face numerous internal and external constraints. The result is that most AARD institutes have not actively served their extension audiences.
78
CHAPTER 5 INFORMATION FLOWS AND THEIR CAUSES Introduction Imagine you want to buy a new car. Think of the types of information you might want in order to decide what model to buy. You'd be interested in the number of passengers and amount of baggage space, fuel consumption, the engine size and acceleration, handling and comfort, safety features, and the cost of servicing. You'd probably look at the price and any financing options to make sure you get the best value for money. Now think where you'd get this information. You have many possible sources. You might browse through advertisements in magazines. You could visit car dealers and look at the vehicles for sale. You could ask the salesperson for information or pick up brochures on each model. You might talk to a friend who has just bought a new car. You could even ask your friend to let you test-drive it. There are so many possible sources that you cannot pay equal attention to all. You must be selective. But why choose some and not others? What determines where you get the information? Or in technical terms, what factors determine the level of information flows to a receiver from various potential information sources? You might read a magazine advertisement because it is attractive and easy to read. You might talk to a friend because you trust her judgement, or phone a salesperson because you think he has a lot of information. You visit the local car dealership rather than a larger one across town because it is closer and more convenient. It is clear that many factors -- ease of use, credibility, expertise, proximity, to name a few -- can influence your information sources. This chapter proposes a list of these factors. I've collapsed them into the ten "HELP SCORES" characteristics proposed by Havelock (1969) and Havelock and Lingwood (1973): Homophily, Empathy, Linkage, Proximity, Structure, Capacity, Openness, Reward, Energy and Synergy. I propose that these characteristics can be used to predict the level of information flows from each source to a receiver. Outline of this chapter After defining some terms I use throughout the study, this chapter focuses on the key concept of "information flow." I then briefly discuss various approaches to research on information flows, concentrating on the following independent variables: characteristics of the topic, message, situation, time, receiver, and source. Finally, I propose that with suitable adaptation, the HELP SCORES variables can be used to predict information flows from various sources to a receiver.
79
Concepts Information Somewhat surprisingly, information is something of a primitive term (Reynolds 1971:46) in communication science. It often remains undefined or is stated in terms of other concepts that are equally difficult to nail down, such as "knowledge" and "communication." Rogers (1983:6) defines information as "a difference in matter-energy that affects uncertainty in a situation where a choice exists among a set of alternatives." Rogers does not expand on what he means by the puzzling term, "a difference in matter-energy." And his focus on choices among a set of alternatives seems unnecessarily restrictive. We do not have to be in the situation of having a set of alternatives to obtain information; indeed, information may create alternatives where none existed before. The definitions of Atkin (1973:207) and Wiio (1980:18) come somewhat closer to the mark. Wiio conceptualizes information as something that reduces uncertainty, while Atkin defines it as "something that the receiver does not already know." But both these definitions are also limiting: they fail to admit as information "redundant" messages that reinforce already existing cognitions or remind a receiver of knowledge already held. And Wiio's definition ignores that information may also increase uncertainty. We can thus expand on Wiio by defining information as something that affects uncertainty about a topic. This subsumes both "new" messages that tell something that the receiver does not already know, and "old" messages that remind or reinforce. Information receivers An information receiver is a person who obtains, or who may potentially obtain, information on a given topic. Information sources Information sources are persons, institutions or channels identified by the information receiver (or an observer) as actually or potentially providing the receiver with information on a given topic. Note that I'm using the word source to mean some combination of source and channel, to use the terms of Berlo's (1960) SMCR (Source-Message-Channel-Receiver) model. Rogers (1983:197) uses channel to refer to the same concept. I do not distinguish between a true source (such as "your boss"), a channel (such as "the telephone"), or a source/channel combination (such as "phone conversations with your boss"). With some combinations, the true source may be unknown, unidentifiable or irrelevant. The true source may be identical for several different source/channel combinations, as when a researcher disseminates the same information through a publication, a seminar, and a radio program. The identity of the true source is further obscured because we often group individual source/channel combinations into categories or types. For instance, an extensionist may
80 draw on "research reports" for information. The term "research reports" is a category of many combinations of individual sources and channels -- numerous researchers (the true sources) writing in several different publication series published by different organizations. The extensionist may know which publication series she uses as a "source," but is unlikely to know, or care, who the original authors are. Indeed, in some instances the true source may not be important at all; rather, we may be interested in the channel, as in a comparison of information flows via newspapers, television, and radio. Combinations of sources and channels can be operationally defined as broadly or as narrowly as convenient. For instance, one study (such as this one) could compare among broad categories of source/channel combinations, weighing sets of publications and interpersonal communication against each other. Another might focus on information from different sources flowing via a single channel, such as radio. And another could focus on flows from the same source via different media (print, interpersonal, broadcast, etc.). Such comparisons are valid as long as the source/channel combinations used are mutually exclusive. Source/channel combinations can be personal (individuals or organizations) or nonpersonal (books, newspapers, television). Much of the communication literature refers to either one or other of these types. Most theories predicting the amount of information a person obtains from personal sources do not transfer readily to non-personal sources, and vice versa. In part this is a measurement problem: it is difficult to develop measures of information flow that apply equally well to both types (Chaffee and Mutz 1989). It is also due to differences in the nature of the sources: we can engage in two-way interaction with other people, but not with books or newspapers. Theories of interpersonal communication thus tend to be two-way in nature, while theories of media use tend to be one-way. We are thus faced with a problem if we want to include both personal and nonpersonal sources in a study. We must develop measures that are comparable across diverse source types. And we are forced to ignore the obvious two-way nature of interpersonal communication. The benefit of doing so is that we can compare directly between personal and non-personal sources. We do so at the cost of losing the explanatory power of theories that take two-way communication into account. Information flow Definition Information consists of stocks and flows. Each individual holds a stock of information in memory (in the form of knowledge). This stock is continually being added to by incoming information flows -- messages, perceptions, interpretations -- and reduced by memory loss. High levels of flow affect the receiver's uncertainty in a major way; low levels affect uncertainty less. Information flow is the quantity of information relevant to a given topic that a receiver obtains from a particular source. This is the variable we wish to predict for different information sources.
81
One-way vs. two-way flows The SMCR model and other communication theories have been subjected to severe criticism for assuming that the receiver is passive and that information flows only one way, from source to a receiver (Dissanayake 1986:64). Subsequent approaches, such as Havelock's linkage model (Havelock 1986a:98, Havelock and Lingwood 1973:277), credit the audience with a more active role, while the coorientation model (McLeod and Chaffee 1973) removes the distinction between source and receiver altogether. Readers hoping for a two-way approach in this study will be disappointed. An SMS's choice of sources seems unlikely to be related to the level of feedback to that source. There are several interrelated reasons for this: • What I refer to as a source is in fact a source/channel combination (Rogers 1983:198; see also the discussion of sources above). The true sources are often unknown or irrelevant, and may be identical for several source/channel combinations. For example, a researcher's finding may reach an extensionist via numerous source/channel combinations -direct contact, publications, training, etc. • It is difficult to relate any one research-to-extension channel with a single feedback mechanism. Hussein's research and my own experience suggest that much of the "upward" flow from farmers to researchers follows a different set of channels from the downward: the little that does take place is more likely to be direct researcher-farmer contacts rather than via the extension system (Hussein 1986:409). Similarly for research-extension relationships: most extension-to-research contacts appear to be interpersonal, while research-toextension flows tend to be through publications. • Experience and research in the developing world indicates that two-way communication between researchers and extension personnel is minimal (see for example, Seegers and Kaimowitz 1990, Kaimowitz 1990:xi, Benor and Baxter 1984:94, Coulter 1983:52, Cernea, Coulter, and Russell 1985:3). This is true also in Indonesia (Hussein 1986, Padmanagara 1985:137), despite the importance of farming systems research within AARD and the recent emphasis on Research-Extension Linkages Project (see Chapter 4). The Indonesian system is predominantly top-down in its orientation, with centrally determined recommendations transmitted "down" through the extension service to farmers. While attempts are being made to institutionalize bottom-up flows, such efforts have received relatively little attention. • Two-way flow models are most appropriate for interpersonal communications. But source/channel combinations include impersonal as well as personal contacts. A receiver may obtain information from a publication without ever providing feedback to the author or publisher. While direct researcher-extensionist contacts do occur, they are rare (Hussein 1986:422). On the other hand, we might expect print media to be especially important for Indonesian extension personnel, since the extensionists are geographically scattered, and research institutes have devoted most effort into disseminating their results via publications. It makes little sense in such circumstances to attempt to measure two-way information flows.
82 This does not mean that extension-to-research communication is unimportant; indeed, the strength-of-weak-ties hypothesis (Granovetter 1973) would indicate that such flows are of major importance. But such upward flows seem unlikely to affect directly which sources SMSs use. We can thus study top-down and bottom-up flows separately. This study looks at the former, thereby fitting into the "feed in" cell in Compton's (1984) matrix for guiding the analysis of agricultural knowledge systems. Nor do I ignore the desirability of two-way information flows or have eschewed the notion of an active audience. On the contrary, the model I propose is based on the premise that receivers actively choose their information sources based on their perceptions of the nature of the sources. This study thus falls into the information-seeking rather than the control paradigm (Krippendorff 1987:194). A top-down view of information flow normally assumes a passive audience. Rather, I see sources as providing an array of information, from which receivers can then choose (see Chapter 2). Such a view combines a one-way analysis with the notion of an active audience. One of the purposes of this study is to discover the reasons receivers choose one source rather than others. Treatment in the literature Information flow is treated in four ways in the literature: it is measured in descriptive studies, and used as an independent, intervening or dependent variable. 1. It is frequently studied descriptively (not related to theory) in research on information sources. For instance, Hussein (1986:421-422) lists researcher-extension contacts but does not relate these to any other variable. 2. It is used as an independent variable in studies of attitude formation, persuasion and learning. Much of this work has been done in social psychology. A typical design is to present subjects with an item of information and test its effect on attitudes and behaviors. Information flow has been studied in field settings by sociologists and development communicators; a typical design here is to distribute information to an audience (such as on a new agricultural technology) and then to measure knowledge gain and attitude and behavior change. 3. An example of information flow used as an intervening variable, is Rogers' (1969:102, 112-122) study of mass media exposure in five Colombian villages. He used villagers' exposure to the mass media, a measure of information flow, as an intervening variable between antecedent variables (literacy, education, cosmopoliteness, social status, and age), and consequent variables (empathy, innovativeness, political knowledge, achievement motivation, and aspirations. Information flow has also been studied in artificial small-group situations with different network patterns (e.g., wheel, circle, chain [Rogers and Agarwala-Rogers 1976:120). The independent variable in such studies is the pattern or the individual's
83 position in the network; the dependent variable is usually the efficiency of performing a task such as solving a puzzle, or the individual's satisfaction with her or his role. Information flow often is not measured in such studies. 4. As a dependent variable, information flow has been studied within the uses and gratifications, information seeking, diffusion, network analysis and knowledge utilization paradigms. These were briefly discussed in Chapter 2. In this study, I am interested in information flows as a dependent variable. In other words, I wish to find what influences the level of information flows to receivers from various sources. Unit of analysis Information flows can be seen as traits of a unit such as a person, organization, or society. They can also be seen as attributes of a relationship between two units, or dependent on a situation. Unitary, relational, and situational concepts can be used at any level, from macro to the individual. Whether information flow is a unitary, relational, or situational attribute determines what factors can be used to predict it. • If information flow is seen as a unitary characteristic of an individual, then other characteristics of the individual -- age, sex, and personality, for example -- must be used as predictors. • Viewing it as a situational attribute means we must explain it in terms relating to that situation -- for instance, problem recognition, constraint recognition, and level of involvement (Grunig 1983:10). • Viewing information flow as a relational attribute allows us to use other attributes of the relationship to explain it. Such attributes could include the source-to-receiver distance, the receiver's familiarity with the source, and the perceived credibility of the source. Information flow as a unitary attribute At the societal level, we can see the numbers of phone calls or cinema seats in a country as a unitary attribute of that country (e.g., De Sola Pool et al.'s 1984 communication censuses of the United States and Japan). Similarly, the number of programs broadcast by a radio station is an attribute of that station. At the micro level, we can view the total amount of information an individual receives on a topic as an attribute of that individual. Research at this level typically tries to discover which characteristics of the individual influence how much information the person obtains. Examples abound in the innovation diffusion and information seeking literature (Rogers 1983:241, Grunig 1983:8). The total amount of information a person obtains and his information seeking method may well be unitary attributes -- dependent on the characteristics of the individual in question. But it is difficult to justify regarding information flows from numerous different
84 sources all as attributes of an individual. If we seek to discover why a person uses one source (say, research journals) rather than another (say, symposium proceedings), it makes little theoretical sense to attribute causation to characteristics of the receiver. It makes more sense to examine the relationship between each source and each receiver. Hence the relational view of information flows. Information flow as a relational attribute The flow of information from a source to receiver is in some ways analogous to water flowing in an irrigation canal from a well to a field. Just as one field may be fed by many wells and canals, someone may get information from many sources -- books, other persons, the mass media, and so on. A single well can supply water to several fields; so too can each information source be a partner in many dyads with different receivers. The total amount of water pumped from a well is a unitary attribute of that well. Likewise, the total amount received by a field is a unitary attribute of that field. But the amount flowing from each well to each field is an attribute of the relationship between the well and the field, since a separate figure is needed to describe the quantity of water for each pair of wells and fields. Viewing information flows in an analogous way enables us to distinguish flows from each source to each receiver, rather than treating them as a whole. Instead of looking at the receiver or source alone, we can thus view them as a linked pair, or dyad. Information flow then becomes a characteristic of the dyad, or of the relationship between source and receiver (Figure 5.1).
85
Figure 5.1
Information flows as characteristics of the relationship between a receiver and various sources
Relational variables can be measured at various levels, from dyads of societies and organizations to pairs of individuals. At the macro level, studies of international news flows and information exchange among firms may use a relational view. At the interpersonal level, we regard each pair of persons as a unit. In this study, I was interested in the determinants of information flows to receivers from various sources. The sources could be individuals, classes of persons, or media types. A relational view of information flow, using individuals as receivers and categories of source/channel combinations as sources, would thus seem appropriate. Information flow as a situational attribute An alternative is to view information flow as an attribute of the situation. To return to our irrigation example, this is analogous to arguing that the flow of water depends not only on the well (or the receiving field, or the relationship between them), but also on the time of year (dry vs. wet season), the weather, and intermittent pilferage of water by landowners along the canal. Such factors can raise or lower the water flow in the canal. They relate to the situation of the particular irrigation system and may change over time.
86 I chose to ignore the situational aspect of information flows in this study. I explain more fully later in this chapter. Perspective Information flow in a dyad can be measured from three different perspectives: that of the source, the receiver, and an impartial observer. For instance, we could question a car salesperson (the source) about the amount of information she has given to a customer. Alternatively, we could question the customer on how much information he has obtained. Or we could observe a conversation between the two and rate it according to some objective scale, such as duration or number of topics discussed. The level of flow measured is likely to depend on whose perspective is used. The amount of information provided by the source may actually differ from the amount received by the receiver -- as in a radio broadcast that no-one is listening to (De Sola Pool 1984:56). In addition, differences may also result from subjective perceptions: the salesperson may overestimate the amount of information she has given, while the customer may rate the amount lower. Both estimates may be different from that of the observer. All three may be colored by the different perceptions and memory of the actors. Which is the most appropriate perspective for a study of research-extension linkages? Using each entails problems: Source In a situation where much of the information flows through impersonal channels such as publications, a source (such as an agricultural researcher) may be unable to identify receivers or provide adequate estimates of the amount of information flow. Impartial observer Using impartial measures such as numbers of publications received would seem to be ideal. But such measures suffer from two difficulties: • Some types of information flow are relatively easy to measure. For instance, the numbers of publications sent or received can be seen in administrative records. But other types are more difficult to measure: oral conversation is an example. It is thus impractical for an observer to measure all types, since this would entail accompanying a subject at all times, noting the duration and topics of conversations and the types of publications read. Such an approach also is likely to interfere with the validity of the data collected. • Someone can receive a publication but never read it, or can read it but not understand it. Directly observable measures cannot accurately reflect the amount of information actually reaching the receiver's brain. Receiver Asking a receiver how much information she receives entails advantages and disadvantages. Advantages include that the receiver can (probably) identify the source and estimate the amount of information obtained from it. The receiver can also provide estimates of the amount of information flow on a common scale for different sources. A disadvantage of questioning the receiver is that responses are biased by the
87 receiver's memory and perceptions. The receiver may have forgotten where she obtained an item of information, or how much she got from each source. And she may perceive the levels of flow from each source inaccurately. Nevertheless, judicious questions may counteract some of the bias inherent in every receiver's perceptions. For instance, we can ask the respondent to estimate objective criteria such as the frequency of obtaining information from a source. We can also pose several questions to tap the same variable, thereby ensuring the reliability, if not the validity, of responses. All in all, it seems that the receiver perspective is likely to yield more valid and usable data on information flows to extension personnel than would questioning sources or attempting to measure flows through an impartial observer. Accordingly, this is the approach I use in this study. Summary The level of information flow from a source to a receiver is a characteristic of the dyadic relationship between the source and receiver. Sources can be personal or nonpersonal. Information flows are most appropriately measured from the perspective of the receiver. Measuring information flows In order to measure information flows in a dyad, we must name the topic of the information flow, identify members of the dyad (the source and receiver), and gauge the quantity of information flowing between them. Identifying topics Information is specific to a topic. Knowing the price of a car does not tell you anything about whether it will rain tomorrow. The two topics, "car prices" and "weather," are mutually exclusive. Other topics, such as "car safety features" and "car prices," may be related. Still others are nested within each other: "car prices" is a subset of the general topic of "cars." Topics may be defined broadly or narrowly as a study requires. They must make sense to the respondent, and there must be a common understanding of what is meant by a topic among respondents to ensure that the data are valid. Studies normally specify the topic in one of two ways:
88 • Categories (e.g., national political news, rice cultivation, cattle feeding). For instance, "What is your main source of national news?" Using topic categories has the advantage of universality: most respondents are likely to have received some information in a category. But such broad categories may not be specific enough if we want to know when or how someone obtained an item of information. • Facts (e.g., the assassination of President Kennedy [Greenberg 1964]). Focusing on a fact allows us to discover when and how people get individual pieces of information. The fact must have the potential to reach a large proportion of respondents, and be salient enough for them to remember it. The Kennedy assassination is one such item. But highly salient facts may be communicated in a different way from more routine information. And it is difficult to identify salient facts of agricultural technology likely to be communicated or of use to all extensionists in Indonesia. In this study, I was interested in the flow of a broad category of information on agricultural technology to extension specialists, as well as subcategories of this (such as information on cattle breeding and legume agronomy). Most of the questions therefore focused on categories. I also included a set of questions on specific facts as a validity check. Identifying receivers Identifying sources and receivers is not necessary to study information flows per se; for instance, we can measure the volume of telephone traffic without knowing the origin or destination of calls. But identifying sources and receivers is necessary if we wish to measure or predict the level of flows in a particular dyadic link. For this study, the population of receivers -- Indonesian agricultural extension specialists -- was known (though obtaining their names and addresses was a challenge, as described later). Identifying sources A major problem lies in the wide variety of information sources. Different units are appropriate for each source. We cannot compare the number of publications a person receives with the number of hours he spends in conversation, since the two are measured on different scales. It is difficult to develop measures that can be used to compare information flows across sources (Chaffee and Mutz 1989). It is also difficult to conceptualize and explicate explanatory variables that apply to all types of sources. Measuring information flows from the receiver's perspective (see above) can help overcome the problem of measuring flows from different sources. This is because we can ask the receiver to estimate the amount of information obtained from each source using a common scale. While this approach is far from satisfactory -- it may yield ordinal rather than interval level data, and is subject to the receiver's perceptual bias -- it seems preferable
89 to the alternatives. The identity of sources can be seen as a "binary" measure (Monge and Contractor 1988:109): a particular receiver either has or does not have a link with each potential source for information on a particular topic. Note that this does not measure the strength of a linkage. Binary measures are often used to identify communication partners, and are then followed by questions of frequency of communication or the linkage strength (Rogers and Kincaid 1981:99). Various methods are used to identify the existence of such links: • The sources named by the respondent without prompting, as in, "Where do you get information on X?" or "With whom do you usually talk about [topic X]?" (Rogers and Kincaid 1981:97). The researcher may limit the number of sources to be identified, as in, "Please list the four faculty you use most as sources of extension information" (Mundy 1989:211), or "Who are the three other women in this village with whom you have discussed family planning?" (Rogers and Kincaid 1981:98). This method is used to identify the most-used sources of information on the topic. • A roster, in which the respondent is presented with a list of all other members of the system and is asked whether he or she talks with each of them (Rogers and Kincaid 1981:98-99). This method can help identify possible "weak ties" -- infrequent yet important links between closely-knit cliques. It is also used in questions such as, "Did you watch the national TV news last night?" A disadvantage is that it is confined to systems where the number of possible sources is manageable enough to fit on a roster, such as in small neighborhoods or organizational departments. • The sources of facts held by the respondent ("Where did you find out about X?"). This method too may identify weak ties, since people often find out new information from sources with whom they do not often talk (Granovetter 1973). Measuring flow levels Binary measures are often used to identify communication partners, and are then followed by questions of frequency of communication or the strength of linkages in order to measure information flow levels (Rogers and Kincaid 1981:99). The level of flows can be measured in a number of different ways:
90 • The ranking of information sources according to criteria such as importance, frequency of interaction, quantity of information obtained, etc. An example of such a question is, "Rank [the members of your system] in order of the frequency with which you talk to them" (Rogers and Kincaid 1981:102). Such measures are relatively easy for respondents to generate because they involve comparisons among sources. But they have two disadvantages: (a) they generate only ordinal-level data and (b) they assume that sources compete with (rather than supplement) each other. This is despite the frequent finding that people vary in the amounts of information they seek and obtain: e.g., an avid viewer of television news may also get a lot of information from newspapers or friends. Ranking obscures this. • The frequency of exposure or attention the respondent devotes to a source, such as the frequency of meetings or informal contacts or attendance to the mass media. Questions measuring frequency include, "How often do you talk with B?" (Monge and Contractor 1988:109) and "How many times, in an average 5-day, Monday-to-Friday week, do you watch the evening network television news?" (Palmgreen and Rayburn 1982:570). • The duration of exposure or attention the respondent devotes to a source. Questions include, "The last time you talked with this person, how long did your conversation last?" (Monge and Contractor 1988:109) and "How much time, in hours, did you spend with [each member of your system] this month?" (Rogers and Kincaid 1981:102). Other frequent questions measure the number of hours spent watching the television, reading newspapers, and so forth. • The perceived quantity of information on the topic derived from the source, as reported by the respondent. A typical question is, "How much information on X do you get from B?" The researcher provides the respondent with a suitable scale (e.g., 5 = very much, 1 = very little). A related question, "How important is B to you as a source of information on X?" could be interpreted as a measure of the relevance of information from B (therefore a determinant of information flow, see below) rather than an indicator of information flow itself. • The availability of information in the form of publications, telephones, television, etc. For instance, "Do you receive [publication X]?" and "How often do you receive it?" • The number of words. This measure was developed as the unit for a "communication census" of Japan and the United States (Takasaki and Ozawa 1983, De Sola Pool et al. 1984:3). The census compared a wide variety of mediated communication methods, including television, books, newspapers, classroom teaching, telephone conversations and data communication. For most media, it is a measure of the time spent receiving information multiplied by an estimate of the number of words received per time unit. The number of words has two major disadvantages as a measure of information flows: it is difficult to measure flows in interpersonal channels, and it takes no account of meaning. The second disadvantage is the more serious: using the number of words criterion, a thousand words of gibberish is equivalent to the same number of carefully weighed and well understood words.
91 • The respondent's ability to recall topics covered by the source. This is often used in studies of television news, as in, "Can you name any of the topics in yesterday's evening news?" The number of topics correctly named is used as a measure of information flow. • The respondent's learning of information provided by the source. Respondents are questioned about details of a topic communicated by a source; the number of correct responses is taken to measure information flow. Receivers vary in the total amount of information they obtain. In order to distinguish low from high information recipients, we should therefore develop absolute, rather than relative, measures of information flows. This means we should avoid rankings as an information flow indicator. An index composed of several absolute measures, such as availability, frequency, duration and perceived quantity of information from the source would seem to be the best indicator. A second problem is the need to identify "weak ties" (Granovetter 1973). Such ties are likely to be important for people in obtaining new information, but are difficult to detect using measures of frequency or volume. Using the roster method for identifying sources is one way of overcoming this difficulty. Another is to question respondents about where they obtained specific facts. This problem also implies a need for in-depth interviews of a small number of respondents in addition to a survey of a larger number. Interviews offer the flexibility needed to probe deeply into information gathering behavior -- a flexibility surveys do not provide. At the same time, a survey can provide a large quantity of uniform data in a form more tractable for statistical analysis. Summary In order to measure information flows, we must specify the topic and the source and receiver in the dyad. We can then gauge the level of information flow by questioning receivers. An interval scale should be used to ensure that total levels of information flow are comparable across respondents. Influences on information flows Many factors can affect information flows. As it is not practical to study all these factors at the same time, it is necessary to control for some by holding them constant (or assuming they vary randomly with respect to another sources of variation). Dependent and independent variables must be measured at the same unit of analysis to allow statistical tests of the relationships between them. If we view information flow as a relational attribute of the source-receiver dyad, we must seek influences on it that are relational also. We must explain it using other dyadic characteristics rather than with attributes of (say) the source or receiver alone (Figure 5.2).
92
Figure 5.2
Information flows viewed as dependent on other relationships between a receiver and a source.
This section discusses six influences on information flow. It describes how aspects of these influences can be regarded as attributes of the source-receiver dyad, and thus used as predictors of information flow. I then propose hypotheses to test the relationship between the predictors and information flow as the dependent variable. How may information flows vary? Naming ways in which a variable may vary can help us identify factors that may influence it. Information flows may vary in a wide variety of ways. They include: 1. Across topics. You obtain more information if you want to buy a car than if you're purchasing toothpaste. 2. Across messages. You pay attention to a glossy display ad for a car in a magazine, but not to a classified ad for the same model in the same magazine.
93 3. Across situations. If you are a first-time car buyer, you'll probably use different sources and get different amounts of information from the next time you buy. 4. Across time. As you near the decision to purchase the car, you probably seek larger amounts of information than when you first began to think about buying. 5. Across receivers. Even if both you and your friend are interested in buying a new car, you will probably not obtain the same amount of information or rely on the same sources. 6. Across sources. You do not obtain the same amount of information from a magazine advertisement as from a car salesperson. Each of these is discussed below. Topic Information flows vary according to the topic. We seek more information about a major purchase such as a car than about a minor one such as toothpaste. We may attend to local news rather than sports, scan the marriage announcements in the newspaper but not the births and deaths, and seek information on cars rather than computers. The topic also affects the source we use (Midgley 1983, Maguire and Kench 1984). For instance, we are not likely to learn anything about the weather from a magazine advertisement about cars. We use newspapers to get international news, conversation to inquire after the health of a neighbor's children, and memos from the boss to learn of administrative decisions. Midgley (1983) found that men shopping for a social product (a suit) were more likely to seek information through interpersonal than through objective or impersonal sources. Deacon and Firebaugh (1988) show that the type of product (car, jewelry, vacations, etc.) determines which family member or members seek information on the product and makes the decision to purchase. Similarly, Shaninger and Sciglimpaglia (1981) found that both product characteristics and personality influenced housewives' search for information on durable and nondurable goods. Extension specialists vary in their subject matter specialties: agronomy, pests and diseases, livestock husbandry, veterinary medicine, for example. Questioning them about their information sources on a single topic, such as rice agronomy, would be fruitless because most have no interest in this subject. We can control for topic by asking extensionist respondents about agricultural information in general, or about their own particular specialties. The latter controls for topic by providing data about each respondent's professional specialization.
94
Message Journalists and advertisers are well aware that the form and content of messages influence the level of information flow. Communication students are taught to write for different audiences, design publications for easy reading, and compose attractive advertisements. Sources become known for supplying messages with certain characteristics. For instance, a receiver may perceive a source as providing messages that are highly relevant, credible, and easy to use. Over time, these characteristics become associated with the source. Sources viewed as positive should therefore generate greater information flows to the receiver than other sources. Situation If you already own a car, you have some experience. You will obtain a different quantity of information and use different sources from when you bought your first car. If you've just written off your car in a crash, you'll probably be looking for a new one more urgently than otherwise. And you may be receptive to information you might ordinarily ignore -- such as a magazine ad touting a model's safety features. These are examples of how situations can influence information flows. You do not seek information on a topic or choose a particular source because of any characteristic of yourself or the source, but rather because of the situation you are in. Various types of situations have been shown to affect information seeking behavior: the need for political information (Woelfel 1977); the patient's (rather than the doctor's) description of the health situation (Dervin 1980) and the "situation movement state" (Atwood and Dervin 1982, Atwood et al. 1982). Situation often has been found to interact with receiver characteristics in predicting sources sought (Atwood and Dervin 1982). Receivers may actively seek information (information search), obtain it through routine scanning of messages (information receptivity), or acquire it because avoiding it is too much trouble (information yielding) (Atkin 1973:238). Grunig (1983:10-13) proposes that for people to become active seekers, they must recognize a problem in a situation that concerns them. Thus, if your car has fallen apart, you perceive a problem (the car is broken) and that you are connected with the problem (it's your car). You will thus seek information on alternatives, such as the characteristics and prices of new cars. In this study I make no distinction between information search, receptivity and yielding since I am interested not in how someone obtains information, but in the total quantity obtained. In general, however, we might expect extension specialists to be active seekers of information on their subject area, since they presumably recognize the problems local farmers face and accept that they are connected with the problems (since it is their job to help farmers). My own experience in Indonesia indicates that the extensionists' search is highly constrained by the availability of time, funding and transportation. The distinction
95 between search and receptivity thus becomes moot. Situations vary among receivers. For instance, a prospective car buyer in the United States faces a different set of situations from one in Indonesia. We can control for this variation to some extent by selecting respondents from a restricted group, such as people in similar occupations, thereby restricting the range of possible situations they may experience. Indonesian extension specialists perform generally similar tasks, have similar levels of education and experience, encounter similar problems, and have a similar set of possible information sources. While differences inevitably occur, these similarities mean that we can think of these extension specialists as facing broadly similar situations. Grunig (1983:8) argues that situational theories of communication have received more consistent support in the literature than have cross-situational theories (i.e., those that view information flow as unitary or relational). While his arguments have considerable merit and are in line with current trends in social psychological thinking, I choose here to use a cross-situational approach, viewing information flow as an attribute of the source-receiver relationship. This perhaps requires some justification. • Grunig's (1983:19) situational variables were (1) whether the respondent thought about the topic (problem recognition); (2) how involved the respondent felt to be in the topic (level of involvement); and (3) whether the respondent felt he/she could affect the topic (constraint recognition). My respondent group, extension specialists, should respond affirmatively to all these points, making them members of the "active public" (Grunig 1983:10-13). I would therefore expect little variance on the situational dimensions Grunig outlines (though I did not test this assumption). • When large numbers of respondents are randomly sampled, differences in their situations should be randomly distributed among subgroups. They should therefore balance out through the law of large numbers. Failure to take situation into account should increase the error term in statistical analysis, making any significant findings all the more credible. Time Situations may vary in relatively predictable patterns over time. For instance, Rubin (1979) found that the types of information sought in interpersonal relationships change as the relationships mature. And diffusion research indicates that people seek different amounts of information about innovations through different channels and from different sources as they progress through the stages of innovation adoption (Rogers 1983). However, information awareness may really depend on what channel has the information. As they move toward evaluation, people begin seeking information from any available source. In a study of the adoption of computers by farmers, both mass media and interpersonal source use peaked when respondents were in the evaluation/adoption stages (Abbott 1990). The important factors are thus source availability and the receiver's
96 adoption stage. The effect of time on information flows is best measured through a longitudinal study. Doing so is difficult in a single cross-sectional design. I choose to ignore the time element here, assuming that variations in information flows will tend to cancel out when questioning a large sample of respondents about a broad topic. Receiver Different people obtain different amounts of information, from different sources. Much of the research on information seeking has focused on the psychological and demographic characteristics of individual receivers. Because of their psychological makeup, social relations and experiences, people are seen as having different information seeking styles and obtaining different amounts of information. "Loners," for instance, obtain relatively little information about a topic before making a decision; "formal seekers" obtain information about a wide variety of options before committing themselves, while "risky seekers" obtain a lot of information about a narrow range of options (Donohew, Tipton, and Haney 1980). A whole slew of individual characteristics has been associated with information seeking behavior. A short list includes: prior knowledge, egotism and sensation seeking (Donohew, Tipton and Haney 1980), nationality, socio-economic status, parental socio-economic status, educational attainment, childhood magazine use, need for activation, race, age, tolerance for ambiguity, self esteem, cognitive style, trait anxiety, dogmatism, and home ownership (Atwood and Dervin 1982, Capon and Burke 1980, Donohew, Palmgreen and Duncan 1980, Flinn and Jamias 1984, Hirshchmann 1981, Lambert and Durand 1977, Schaninger and Sciglimpaglia 1981). But theories based on receiver characteristics have inconsistent results (Grunig 1983:8), and they fail to satisfactorily explain why people prefer one source over another. It is difficult to imagine why someone should use, say, television rather than books because of such personal characteristics as age, gender, or psychological makeup. It makes considerably more intuitive (and theoretical) sense to say she does so because she likes TV more than books, she thinks the TV is easier to understand, the TV is physically closer, and so forth. All these are relational characteristics because they describe the relationship a receiver has with a source, and they may vary among sources and receivers. In this study, I examine the effects on information flows of some receiver characteristics, such as age, gender, and industriousness. Other receiver characteristics, such as psychological makeup, are difficult to measure using a mailed survey, the only practical large-sample data collection method available. I assume that such characteristics vary randomly with respect to information flows and the explanatory variables. Other receiver characteristics can be subsumed under measures of the relationship between receiver and source.
97
Source Sources affect information flow. We get more news from the television than from newspapers. Farmers in the United States rely heavily on the agricultural press for information on new technologies (Fett and Mundy 1990:6). We prefer to talk to our friends rather than our enemies. Different sources supply larger or smaller amounts of information. A large publishing house produces more books than a small one; a 24-hour news television station puts more news on the air than an entertainment network. Such attributes do not vary from receiver to receiver, and can be measured relatively objectively -- as in number of minutes' news coverage aired, or the volume of publications produced. Other things being equal, we might therefore expect information flows to be higher from a prolific source than from a less prolific one. Because of their larger size and greater funding, Indonesia's food crops research institutes have been more prolific (as measured by numbers of research publications) than the livestock institutes. We would therefore expect that food crops extensionists to recognize a larger total information flow from research than would their livestock colleagues (Hypothesis 1). I expand further on this topic in the section below on Capacity. Source-receiver relationships Other characteristics that at first sight appear to be attributes of sources in fact vary from receiver to receiver. Such variation may be real or perceptual. For instance, receivers vary in their proximity to any one source. One receiver may view a source as exceptionally relevant, credible, and easy to use; another may think the opposite of the same source. And the specific sources that receivers refer to may be different: "your boss" refers to a different person for each receiver. These characteristics can be seen as features of the relationship between receiver and source -- a formulation that eases their use as predictors of information flows -- for two reasons. They may depend on the receiver's perception of the source, and thus be subjective in nature. The source's credibility is one such attribute. Or they can vary among receivers according to objective criteria. The source's proximity to the receiver is an example. The remainder of this chapter develops a model for predicting information flows
Hypothesis 1: Prolificacy A1:
Sources vary in the amount of information they provide.
A2: Ceteris paribus, the amount of information a source provides is proportional to the amount a receiver obtains. TF: Receivers will receive more information from a prolific source than from a less prolific one.
98 based on Havelock and Lingwood's (1973:294) HELP SCORES model. I propose eight characteristics of the source-receiver relationship that affect the level of information flows from a source to a receiver. The eight factors are: Familiarity, Proximity, Structure, Capacity, Openness, Reward, Energy and Synergy. I have omitted Havelock and Lingwood's first two concepts (Homophily and Empathy) because they are difficult to apply to non-personal as well as personal sources, and have changed the Linkage concept to Familiarity for reasons explained below. I adapt the remaining seven factors to fit both personal and non-personal sources. Adapting Havelock and Lingwood's original concepts is necessary because they relate to situations where the source is personal -- an individual or an organization. In such situations, both the source and receiver can be active in the communication process: twoway interaction can take place (Linkage), both source and receiver can be ready to give and receive information (Openness), and both can expend effort at exchanging information (Energy). Where non-personal sources (in reality, source/channel combinations) are involved, as in this study, it makes little sense to talk of two-way interactions (see the discussion of sources earlier in this chapter). A second problem associated with Havelock and Lingwood's model is that the concepts are not unidimensional or mutually exclusive. As Havelock himself points out (1969:11/20), they include: HELP SCORES
"a host of variables which are distinct and which could have been listed separately. At the same time, the headings chosen are not discrete; there is much overlap, and some categories may be seen as subcategories of others." Nevertheless, the HELP SCORES list is a useful starting point for a study of the determinants of information flows. Below I present a set of theoretical propositions that link the HELP SCORES concepts with information flow. For each of the characteristics, I present below theoretical propositions based on an axiom system. In doing so, I draw on the preceding discussion of influences on information flows. For clarity, I capitalize the concepts thus: Proximity. Homophily and Empathy Havelock and Lingwood envisaged the HELP SCORES variables as describing characteristics of the relationships between receivers and personal sources, where both receivers and sources are individuals or groups (such as organizations). Extending their model to non-personal sources entails difficulties. There is considerable evidence for the importance of Homophily and Empathy as influences on information flows (see for example, Rogers and Shoemaker 1971, Rogers 1983, Rogers and Kincaid 1981:133, Maguire and Kench 1984). Chapter 2 listed several aspects of both as problems in research-extension ties: the status differences between scientists and extension personnel, different motivations and personal orientations, different educational levels, and so forth. But these characteristics seem to apply only to linkages where the source is an
99
Hypothesis 2: Familiarity A1: There must be a connection between a source and a receiver for information to flow between them. A2:
People prefer the familiar to the unfamiliar.
A3: Familiarity with an information source means a person knows how to obtain information from it easily. TF:
Information flow will be greatest from sources familiar to the receiver.
identifiable person known to the receiver. I use the term "source" to mean a source/channel combination (see the discussion earlier in this chapter). But Homophily and Empathy are difficult to apply to non-personal "sources" such as books and newspapers: you cannot be homophilous or empathic with a book. With the book's author, perhaps -- but the author is often unknown both to the receiver and to an outside observer. This problem is compounded when individual sources are taken as a group (e.g., "books" in general, rather than an individual title with an identifiable author whose characteristics can be determined). This makes the measurement of Homophily and Empathy at best impractical, and at worst meaningless. Thus, while there is considerable evidence pointing to the importance of Homophily and Empathy in determining information flows, I am forced to drop these concepts from consideration here. Linkage/Familiarity Linkage is seen by Havelock (1969:11/21) as signifying "the degree of interpersonal or intergroup connection; the extent to which mutual communicative relations exist between two or more parties." Note that this use of the term is different from the linkage model he subsequently proposed (see Havelock 1969:11/21 and the discussion earlier in this chapter). In a study using interviews with managers in the U.S. Department of Labor, Havelock and Lingwood (1973:275) operationalized Linkage as any comment by the interviewee that indicated a "contact or relationship between persons or groups is sought or achieved." Linkages between individuals are often informal (e.g., Czepiel 1974, Eponou 1990, Mundy 1989:94, Rogers 1983:293). Havelock and Lingwood (1973:299, 307) maintain that Linkage is the most important of the ten HELP SCORES factors in determining information flows. My research in Iowa (Mundy 1989:182) found that it was a significant but not the major influence on flows between researchers and extension personnel. However, in this study I had interpreted Linkage as being social relationships (Mundy 1989:27).
100 Interpreting Linkage as "the degree of interpersonal or intergroup connection" and "mutual communicative relations," or as the existence of social relationships is clearly not valid for non-personal sources. As a substitute for the purposes of this study, I replace the Linkage concept with the receiver's Familiarity with a source. This requires some justification, especially since Havelock includes familiarity under the fourth HELP SCORES factor, Proximity. Havelock's concept of Linkage appears to measure the strength of the relationship between a source and a receiver. Frequent contacts indicate high Linkage; lack of contact reflects low Linkage. For non-personal sources, familiarity would appear a reasonable substitute for Linkage. Frequent or prolonged contact with books or newspapers makes us familiar with them. The same is true of personal sources such as individuals or organizations. I discuss the distinction between Proximity and familiarity in the next section. Some degree of Familiarity is a necessary but not sufficient condition for information flows to occur. It may refer to whether the receiver knows of a potential information source, such as a publication or a colleague. People rarely seek information from sources they do not already know of. For instance, they are more likely to ask a colleague or reach for a book in their personal library than they are to write a letter to an unknown researcher or perform a literature search for the required information (Kelly and Wolek, quoted in Busch and Lacey 1983:88). But this is not a binary, yes/no concept. Relationships vary in the strength, as reflected in the possible range of responses to the questions, "How well do you know [person B]?" and "How familiar are you with [source C]?" Answers to such questions can range from "not at all" (no Familiarity) to "very well" (high Familiarity). The mere exposure effect proposes that people prefer objects they come into contact with frequently (Worchel, Cooper, and Goethals 1988:351). It would also seem likely that they will know how to obtain information from a source they are familiar with. From this it follows that Familiar sources should provide the most information (Hypothesis 2). Proximity Proximity refers to the physical distance or accessibility of a source. Numerous studies have shown that propinquity breeds attraction and communication (Allen and Fustfeld 1975, Maguire and Kench 1984, Rogers 1983, Rogers and Kincaid 1981:133), though the precise mechanism for this remains under debate. Proximate sources are likely to be more convenient to use than are those farther away. While he does not classify sources in terms of distance, Rosenberg's (1967:122) findings suggest that professionals use nearby rather than distant sources. Other mechanisms include propinquity leading to rewards, expectations of future interactions, and familiarity or "mere exposure" (Worchel, Cooper and Goethals 1988:348). Whichever is correct, all explanations lead us to expect people to obtain more information from nearby or easily accessible sources than from those farther away
101 (Hypothesis 3). That Proximity is not necessarily important is shown by Gidley's (1977) and my own (Mundy 1987) findings that distance was not an important determinant of information flows between extension personnel and researchers. However, both of these studies were conducted in developed countries (Australia and the United States), where abundant telephones presumably dilute the effect of Proximity. Havelock's inclusion of familiarity as part of Proximity is based on analogy (Havelock 1969:11/27, emphasis added): "[U]sers who have close proximity to resources are more likely to use them. Anything which is 'handy,' i.e., easily accessible, is more likely to be used. This generalization applies to people and things but also, at least by analogy, to thinking processes (familiarity, recency, similarity)." However, physical distance and familiarity are conceptually distinct, and may be unrelated empirically -- though one may lead to another, as Havelock later points out. It seems better, therefore, to reserve Proximity for the concept of physical distance and accessibility, and to use Familiarity as a substitute for Linkage. Structure Structure is conceptualized by Havelock (1969:11/23) as the "degree of systematic organization and coordination of elements." Evidence of positive structuring is operationalized by Havelock and Lingwood (1973:297) as "any evidence or [sic] planning, ordering, systematic arranging, scheduling, mapping in a framework, quantitative analysis or evaluation of objectives, work or output," while lack of structuring is revealed by "confusion, disarray, ad hocracy, muddling through, lack of organization, irrationality." I take Structure to mean the organizational relationship between the source and the receiver. Such relationships can be determined by the organization's hierarchy or by it functions. An example of the former is where a superior seeks information from a clerk in her department. An example of the latter is where a hospital receptionist asks a patient for some information. Common to the two types of relationship is the idea of task: it is the superior's and clerk's job to talk to each other, just as it is the receptionist's duty to communicate with patients. Information flow is thus constrained by the roles individuals fulfill within the organization (Pfeffer 1982:98-102, 276; Rogers and Agarwala-Rogers
Hypothesis 3: Proximity A1:
Individuals attempt to minimize the costs of obtaining information.
A2: Information is less costly to obtain from sources that are close by than from more distant sources. TF:
Information flow will be greater from proximate than from distant sources.
102
Hypothesis 4: Structure A1: In order to attain its goals, an organization requires information to flow through certain channels. A2: An organization rewards its members for performing work that helps it achieve its goals. TF: Information flow will be greatest from those sources an individual perceives it his or her job to use.
Hypothesis 5: Capacity A1: Individuals attempt to minimize the costs and maximize the benefit of obtaining information. A2: Credibility is associated with the likelihood of information being correct and therefore of benefit to the receiver. A3: It is convenient (costs are low and benefits high) to obtain information from sources that provide large amounts of information on a topic. TF: Information flow will be greater from credible and complete sources than from those that are less credible or complete. 1976:77-107). Organizations have goals and reward their employees for helping achieve them (Morgan 1986:19-38). One of the goals of an extension institution is to take information from various sources and to pass it on to farmers. It is the task of extension subject-matter specialists to perform part of this task. It follows that specialists should tend to use information sources they see it is their job to use (Hypothesis 4). Structure proved an important influence on information flow in my Iowa study (Mundy 1989:185). Busch and Lacy (1983:88) found that U.S. agricultural scientists communicated more frequently with their departmental colleagues than with scientists in other departments or with extension staff. In many countries (including Indonesia), research and extension are separated administratively (Baharsjah 1985:30), and the interorganizational barriers are high (Coulter 1983:52, Benor and Baxter 1984:94). A frequent response to poor communications has been to adjust organizational structures to overcome these barriers (Kaimowitz 1989). Such revisions include changing the organization's hierarchical structure, shifting responsibilities among departments, combining organizations (such as research and extension) into a single body, forming joint
103 committees and working groups, and establishing new institutions (the AICs in Indonesia) or work roles (SMSs) to facilitate communication. Capacity Capacity is operationalized by Havelock and Lingwood (1973:297) as "any sign of affluence, experience, wisdom, intelligence, strength, or size..." This definition seems to include two dimensions: the source's ability to provide large quantities of information, and the source's authoritativeness, credibility and trustworthiness. Both meanings can be applied to both people and impersonal sources: both people and publications can provide large quantities of information. Both source types may also be deemed authoritative, credible and trustworthy -- or indeed their opposites, depending on the receiver's experience with messages originating from them. Capacity is important for reasons of efficiency of information seeking. We can view information seeking as an activity that incurs costs and provides benefits to the seeker (Atkin 1973). People prefer information from sources they regard as credible in order to reduce the risks of reduced benefits if the information turns out to be incorrect. They also should attempt to obtain information from the source with the greatest amount available so they can reduce their costs and the risk of having to go to other sources (Hypothesis 5). This is related to the idea of prolificacy (Hypothesis 1). In interpersonal linkages, it has been well established that followers seek information from individuals with higher socio-economic status and education (Rogers 1983:277). Rogers cites 17 studies that support this contention. Capacity is also the factor at work in the hierarchical diffusion of innovations described by Brown (1981). That Capacity may not always be important is shown by Rosenberg's (1967:124) finding that the amount of information respondents expected to obtain from a source did not affect the choice of source; however, ease of use was important (see below). Mundy (1989:187) also found Capacity to be irrelevant. Openness Havelock (1969:11/24) regards Openness as "the readiness to give and to receive new information." This refers to individuals or organizations only. For this study, I use Openness to mean the ease of obtaining information from a source -- a concept that can include non-personal as well as personal sources. A source may be willing or unwilling to divulge information, or may use scientific jargon or a foreign language. Similarly, publications may be written in a style difficult for a reader to understand. This is a charge often leveled at research reports, for instance. Havelock (1969:11/24) includes in Openness the readiness of the source to give and to receive information. While willingness to accept feedback may be empirically correlated with readiness to provide it, these two ideas would appear to be conceptually distinct. Further, it is difficult to envisage how impersonal sources such as books can be
104 seen as open to feedback. Their authors can be open or closed -- but they are often unknown (see the discussion on sources earlier in this chapter). While the feedback concept may be important in interpersonal flows, we are therefore forced to disregard it because we are trying to compare across widely differing source types. The Openness concept also is related to Atkin's (1973) idea of costs and benefits of information seeking. Sources that are Open are easier to use; since people try to minimize the expenditures they incur in seeking information, they will tend to choose the most Open sources (Hypothesis 6). This contention is supported by Rosenberg's (1967:124) finding that ease of use was a highly significant predictor of professionals' choice of source. He also found that ease of use (Openness) was more important than quantity of information obtainable from each source (Capacity). My earlier study of research-extension links at Iowa State University (Mundy 1989:188) found Openness to be important in determining choice of sources, but this relationship disappeared when other variables were controlled for. Reward One of the characteristics of an innovation likely to lead to its rapid adoption is relative advantage (Rogers 1983:213). Information flow about such innovations is likely to be high, and audiences will attend to sources of such innovations. The selective perception hypothesis proposes that people attend to messages (and source types) that fit with their pre-existing schemata (Worchel, Cooper, and Goethals 1980:72). Uses and gratifications theories posit that people attend to sources that consistently provide them with gratifications. People are seen as having a variety of needs, which they attempt to fulfill by attending to the media (McGuire 1974). They will attend to the media that best fulfill these needs. Common to these theories is the idea of reward. Receivers are likely to attend to information about topics which they deem relevant to their needs -- i.e., which provide them with some kind of reward. They will attend to information from a source when they estimate its reward value exceeds the expenditures incurred in seeking or avoiding it (Atkin 1973) (Hypothesis 7).
Hypothesis 6: Openness A1:
Individuals attempt to minimize the costs of obtaining information.
A2:
Difficulty in understanding or using information entails costs to the receiver.
TF: Information flow will be greater from sources that receivers regard as easier to understand and use.
105
Hypothesis 7: Reward A1:
Information flow may provide rewards for the receiver.
A2:
Individuals seek to maximize their rewards.
TF: Information flow will be highest from sources that provide receivers with the greatest rewards.
In my study in Iowa (Mundy 1989:188), Reward -- or the relevance of information held by a source -- was a major influence on the amount of information obtained from that source. Sources with little relevant information generated low information flows. Research institutions in the developing world are often accused of failing to produce information that is relevant to farmers' problems. In the absence of relevant information, extension services are left with nothing to extend. Energy Energy refers to the amount of effort the receiver devotes to obtaining information from a particular source. Obtaining information involves expenditures of time and energy (Atkin 1973). While people generally try to keep expenditures to a minimum, high expenditures generally result in more information gain than do low expenditures. For instance, we will obtain more information if we search longer or devote more resources to the search (Hypothesis 8). This proposition received some support in my Iowa study (Mundy 1989:190), though the reliability of the Energy index in this study was doubtful. Energy is another of the HELP SCORES variables that can apply also to the source -if the source is a person or organization. Sources such as change agents who devote a lot of Energy to communicating information are more successful than those who do not (Rogers 1983:317). However, it is difficult to conceptualize a non-personal source as expending Energy. We are therefore forced to ignore the possibility of personal sources as expending Energy in order to maintain comparability with the non-personal sources in the study.
Hypothesis 8: Energy A1:
Obtaining information involves costs of time, effort and money.
A2: Ceteris paribus, increasing expenditures result in increased levels of information gain for a receiver. TF: Information flow will be greatest from sources from whom the receiver devotes most effort to obtain information.
106
Hypothesis 9: Synergy A1: People often fail to accept or remember information the first time it is presented. A2: Information may become relevant to receivers only in certain situations or at certain times. A3: Repetition of information increases the likelihood that individuals will be receptive to it. TF: Information flow will be greatest from sources that repeat messages and time them favorably.
Synergy The "coming together of forces, orchestration, combining of diverse elements, synchronization..." is denoted by Havelock and Lingwood (1973) as "Synergy." As formulated by Havelock and Lingwood, Synergy does not appear to reflect a single concept; rather, it includes at least two elements: repetition and timing. The importance of repetition (redundancy) in learning is well known. Repetition may involve the simple repeating of the same message (as with advertisements on television), or providing the message through different channels (as when the same product is advertised through television and magazines). Timing is related to the time and situational factors discussed above; people are receptive to information on a topic only at certain times (e.g., at the evaluation stage of adoption) and in some situations (e.g., when they learn a power plant is to be built next door) (Hypothesis 9). Synergy is perhaps the least convincing of the HELP SCORES concepts. Conceptually, the ideas of timing and repetition are distinct, and they may not be related empirically. In my Iowa study (Mundy 1989:191), Synergy was reduced to the concept of timing because of low reliability in the original index. Even so, it failed to influence information flows. General model The changes in Havelock and Lingwood's list mean that the acronym HELP SCORES is no longer appropriate. Dropping Homophily and Empathy and switching Linkage to Familiarity gives the unfortunate abbreviation FP SCORES. Unfortunately these letters cannot be rearranged to form any new word, so I use this form to refer to the eight concepts below.
107
Figure 5.3
Factors affecting information flow.
We can derive the following general hypothesis from the above discussion: we would expect the level of information a receiver obtains from a source to be influenced by the following aspects of the dyadic linkages between the source and receiver: Familiarity, Proximity, Structure, Capacity, Openness, Reward, Energy and Synergy (Figure 5.3, Hypothesis 10). The model shown in Figure 5.3 is general enough to apply to a broad range of situations in which receivers can obtain information on a topic from a variety of potential sources. Such situations are frequent in communication within and between organizations and in the diffusion of innovations. Studying agricultural extension personnel's information sources offers the opportunity to test this model. Extensionists can obtain information on agricultural technologies from a wide variety of personal and impersonal sources: scientists at various universities and research organizations, other extensionists, farmers, books, scientific and trade journals, training courses, etc. Despite the frequent complaint of poor research-extension linkages in developing countries, it clear that extensionists there do
Hypothesis 10: FP SCORES Information Flow from a source to a receiver is influenced by the following aspects of the dyadic linkages between the source and the receiver: Familiarity, Proximity, Structure, Capacity, Openness, Reward. Energy, and Synergy.
108 obtain some types of information about technologies -- but it isn't clear where they get it and why. The model proposed above provides a framework for investigating this problem. The eight causal variables suggested also provide research and extension managers with guidelines for improvements, since many of the variables are amenable to manipulation by managers. Organizational structures can be changed, for instance, to bring researchers and extensionists closer together administratively. Research publications can be rewritten to ensure they are more understandable (more Open) and directly relevant (provide greater Reward) to extension personnel. And training courses, field days and other activities can be timed or repeated to ensure high Synergy. Model weaknesses The model proposed here has several weaknesses: 1. It does not predict which of the eight factors will dominate in determining information flows. The FP SCORES variables form a "laundry list": their relative importance must be determined empirically and may vary from case to case. For this reason I refer to the FP SCORES approach as a "model" rather than as a "theory." 2. Related to this, the model ignores the potential (indeed, the likely) relationships among the explanatory variables. Such relationships are suggested by Havelock (1969:11/31) and Havelock and Lingwood (1973:297). It is reasonable to expect, for instance, that one's position in an organizational structure is likely to affect one's physical proximity to potential sources: members of the same department tend to have offices in the same building, for instance. Similarly, the existence of strong (social?) links between a receiver and a source is likely to affect whether the receiver perceives the source as "open" or rewarding to work with. 3. The model fails to take into account variation in information flows over situations and time. Many information seeking activities are relatively short-term in nature, involve unique sources, and are unlikely to be repeated: you don't buy a new car very often. The theory seems better suited to longer-term or repeated information flows. We might expect such flows to vary over time as the situation adjusts, relationships between individuals change, and receivers become more adept at locating and using sources they find the most useful. A cross-sectional survey of the type described cannot measure such variations, but must assume they tend to cancel out in a large sample size and for a broad topic. 4. Because it is deliberately drawn broadly enough to include all possible sources of information, the model may omit characteristics unique to certain types of source. Two such characteristics were mentioned above: homophily and empathy, both of which seem less applicable to impersonal than to personal sources. 5. The model assumes information flow is one way, from source to receiver. I have deliberately excluded those elements of Havelock's model that are two-way in nature in order to be able to use it with non-personal sources. While the one-way assumption may be realistic for impersonal sources, it is less so if the source is a person.
109 6. A problem in all survey research is determining the direction of causality. While some of the independent variables are more likely to be causes rather than effects of communication (e.g., Structure and Proximity), others (e.g., Familiarity, Openness) can be (and in the literature often are) seen as consequences rather than antecedents of communication. 7.
A final problem is the need to control for situation and topic. This imposes restrictions on any empirical test of the theory that may effectively confine such tests to specialized communication activities, such as the extension personnel study reported here.
Model strengths Despite these weaknesses, I believe this model offers several major advantages over others in the literature: 1.
It takes a relational view of both information flows and their causal agents.
2.
It enables us to compare information flows and their causes across different types of sources. Previous theories have either focused on the total amount of information a receiver obtains (ignoring differences among sources altogether), or have compared very broad categories of sources (e.g., newspapers vs. TV). The latter have failed to include either a wider range of sources or a more detailed breakdown of these categories. The theory proposed here allows as fine a breakdown of source categories as the researcher desires. For instance, a receiver may be questioned about "researchers" in general as a source, a particular group of researchers (such as those at the local university) or individual researchers by name.
3.
It can be applied at different levels of analysis. Neither sources nor receivers need be individuals. Instead, the receiver could be conceptualized as a group or an organization, with little change in the explanatory or dependent variables used. Summary
Information flows are relational attributes of a source-receiver dyad. They are best measured from the receiver's perspective. They may vary across topics, messages, situations, time, receivers, and sources. It is necessary to control for each of these when studying information flows. I hypothesize that information flows are affected by eight factors: Familiarity, Proximity, Structure, Capacity, Openness, Reward, Energy, and Synergy. Each of these factors can also be regarded as relational attributes of the source-receiver dyad.
113
CHAPTER 6 M ETHODS Timing and location The study consisted of two related parts: a mail survey and a series of personal interviews conducted before and after the survey. This chapter deals first with the interviews before turning to the mail survey. The study was conducted in Indonesia between 9 March and 8 November 1991. During this time I was based at the Research Institute for Animal Production of the Central Research Institute for Animal Science, Bogor, West Java. I also made two extended field trips outside the Bogor/Jakarta area to pretest questionnaires and collect data: one around West Java, and the other to North and South Sumatra. Interviews I conducted personal interviews with two types of individuals: extension subjectmatter specialists (SMSs) (members of the population from whom the mail survey sample was drawn), and other officials concerned with research-extension links. All interviews were conducted in the Indonesian language. Interviews with extension specialists and local officials I interviewed a total of 66 local agricultural officials (37 SMSs and 29 others), using a semi-structured, open-ended questionnaire. The non-SMS officials were all directly involved in extension work; most were heads and extension section leaders of Dinas offices. These interviews consisted of two types: personal visits, and group interviews. I visited 42 livestock and food crops personnel employed in a total of 22 provincial and district offices and Agricultural Information Centers (AICs) in three provinces: West Java, North Sumatra, and South Sumatra. I also conducted a series of structured group interviews with 24 livestock officials, SMSs and Agricultural Staff Training Center lecturers attending a livestock training course in Ciawi, West Java. These officials came from eight of Indonesia's 27 provinces (West and South Kalimantan, North and South Sulawesi, Aceh, Bali, South Sumatra and Jambi). Before each interview, respondents were requested to complete a written questionnaire -- either the pretest (if the interview took place before the survey instrument had been finalized) or the final form of the questionnaire. The interview questions focused on the respondents' extension activities, information seeking behavior, and problems they faced in their work.
114
Other interviews I interviewed a total of 35 other persons connected with research-extension linkages, both as part of the research planning process and after the mail survey had been sent out. These included Ministry of Agriculture officials, Agency for Agricultural Research and Development (AARD) communication personnel, faculty at Bogor Agricultural University and Pajajaran University (Bandung), and expatriates working in agricultural research and administration. I interviewed several of these individuals more than once. Several interviews were with groups rather than individuals. I also talked with field extension agents and farmers where possible, though I made no attempt to do so on a systematic basis. Most of these interviews were conducted on an informal basis. Where possible, I made written notes during the interview; otherwise I made tape-recorded or written notes immediately afterwards. Analysis I have not made a formal analysis of the interview material for this study. However, the interviews were valuable for two reasons: • They provide the necessary background for the quantitative part of the study. This was especially valuable in designing the survey questionnaire and in interpreting its results. •
They provide a rich store of examples from which to illustrate the survey findings. Mail survey
The mail survey used two four-page questionnaires, both in the Indonesian language. The first, the Information Sources questionnaire, explored respondents' information sources and their extension activities (Appendices 1 and 2). I prepared four different versions of this instrument, each focusing on a different information source. The second, or Publications questionnaire, gathered data on problems the respondents faced, their use of AARD publications, and their information needs (Appendices 3 and 4). The two questionnaires were distributed to several different samples of extension specialists, as explained later in this chapter. For convenience, I have numbered references to questions from the Information Sources questionnaire as IS1, IS2, etc., and those from the Publications questionnaire as P1, P2, etc. Translations of the questionnaires are given in Appendices 2 and 4.
115
Questionnaire design and pretesting The instruments were designed after considerable discussion with communication professionals in AARD research institutes, Ministry of Agriculture extension agencies and training institutes, the Ministry's public relations division, Bogor Agricultural University, and representatives of the private agricultural press. These discussions served to identify problems in research-extension linkages and methods of evaluating them. The pretest questionnaire combined items of theoretical interest with questions of practical value for AARD and extension personnel. The instrument contained a series of identical closed-ended questions about four different source types: AIC publications, AARD publications, Directorate-General recommendations, and other SMSs. Other questions included both closed- and open-ended items. The instrument was edited for language and content by native Indonesian speakers. Forty SMSs and Ministry of Agriculture local officials involved in extension and training completed the pretest questionnaire during visits to six provincial and district agricultural offices in West Java; the AICs at Ciawi and Lembang, West Java; and the livestock training course mentioned above. Most of the 40 officials also were interviewed individually or in groups about their extension and information seeking activities. I analyzed the pretest using the Statistical Package for the Social Sciences (SPSS) PC+ version 4 (Norusis/SPSS 1990). This analysis included simple frequency tabulation to ensure adequate variation in responses, and reliability and factor analysis to test the usefulness of information flow and independent variable measures. In order to discover any reliability problems specific to any one of the four sources (AIC publications, etc.), I analyzed each separately. I dropped and added several items and made changes in question wording as a result of this reliability analysis. Because another question in the pretest indicated that the agricultural press was an important information source, I replaced the items dealing with Directorate-General recommendations with equivalents relating to the agricultural press. Several respondents complained that the eight-page pretest questionnaire was too long. I therefore dropped some items and split the questionnaire into two four-page instruments: one focusing on information sources, and the other on usage of publications. I reduced the instrument length by creating four separate versions of the information sources questionnaire -- one focusing on each of the four sources. This meant that it was not possible to compare the four different source types within subjects, as had been planned. Instead, comparisons must be made between groups of respondents responding to each of the four versions. This requires the assumption that the four groups are similar in all respects except for the questionnaire they were sent. I achieved this through random assignment of respondents to groups receiving each version and tested this assumption by comparing variables among the four groups. The four-fold versions also necessitated increasing the sample size to ensure a large enough n for significance in statistical tests within each version.
116 At the suggestion of AARD communication officials, several items were added to the publications questionnaire to provide data on specialists' information needs, their use of AARD publications, and their knowledge of AARD-developed innovations. Information Sources questionnaire The Questionnaire about Extension Specialists' Information Sources (Appendices 1 and 2) aimed to discover why extension specialists use certain information sources. Four versions of this instrument were distributed to random subsets of the respondent sample. The four versions were identical except for Questions IS7 and 8, on the third page of the questionnaire. I first discuss the questions common to all versions before describing questions IS7 and 8. Common questions Questions IS1 and 2 measured the respondent's personal characteristics, such as institutional affiliation, specialization, education, experience, gender, provenance (urban or rural), and outside work. Question IS2.9 attempted to measure the respondent's wealth by asking if he or she owned various items. Questions IS3, 4, and 9 asked about the respondent's extension activities: the allocation of time to various tasks, the frequency of seeking information, providing information to clients, and ties with research. Questions IS5 and 6 asked about actual and desired information flows from 24 sources. These sources included the mass media, scientific and extension publications, categories of individuals such as farmers and extension colleagues, training, and the respondent's own experience. The sources listed were those mentioned by extensionists and other experts during the planning phase of the study; the list was substantially revised in the light of the pretest results. Question IS5 measured how much information respondents obtained from each source. Question IS6 asked which they would like to use in an ideal situation. Questions IS9 to 11 measured the number of times the respondent had engaged in various extension and information-seeking activities (such as visits to farmers and seeking answers to clients' questions) in a specified period. Several items in this group measured specific linkages between research and extension, such as joint meetings, research seminars, training, and reading of publications. The time period specified ranged from three months to three years, depending on the expected frequency of the activity and respondent's ability to recall it. Question IS12 focused on the respondent's
117 attitudes toward centrally developed recommendations. This question was intended to distinguish specialists who viewed themselves as conduits of information generated by higher authorities from those who thought they had more autonomy in generating and adapting technology recommendations to suit local needs. Questions different in each version I was unable to discover any comparable studies that tried to measure the Information Flow and FP SCORES variables, so I had no standard set of questions that would give reasonably "clean" indices. I therefore based questions IS7 and 8 on Havelock and Lingwood's (1973) coding sheet for responses to open-ended questions in personal interviews. In addition, space limitations meant it was possible to include only a small number of questions to measure each concept. I attempted to overcome possible reliability problems by testing questions in the pretest and selecting those that had the best construct validity and reliability (as revealed by factor analysis) The wording of these questions differed among the four versions of the questionnaire. In each version, these questions explored the level of information flow from and the respondent's relationships with one of four sources: •
Publications from the Agricultural Information Centers (Version 1).
•
Agricultural newspapers and magazines (Version 2).
•
AARD publications (Version 3).
•
Information from other extension specialists (Version 4).
Each version posed the same questions about one of the sources, with nearidentical wording except for the name of the source type.
118 Questions IS5 and IS7.1 to 7.4 measured the amount of information flow from the Table 6.1 Questions in the Information Sources questionnaire used to measure Information Flow. Variable
Q#
Question wording (translated)a
Information Flow
5.
For extension purposes, how much information do you obtain from [source]?b
7.1
Frequency you read [source].c
7.2
Frequency you obtain information useful for extension from [source].c
7.3
Frequency you use information originating from [source] in your extension activities.c
7.4
Usefulness of [source] as a source of extension information for you.d
a
In each of the four versions of the instrument, "[source]" was replaced by one of four sources: AIC publications, agricultural press, AARD publications, and other extension specialists. Question wording differed minimally among the four versions. b
Information from AIC publications was measured by question IS5.4, the agricultural press by question IS5.1, AARD publications by question IS5.7, and other extension specialists by question IS5.13. Scale: 1 = very little, 2 = little, 3 = somewhat little, 4 = medium, 5 = somewhat much, 6 = much, 7 = very much. c
1 = never, 3 = once a year, 5 = once a month, 7 = once a week, 9 = every day.
d
1 = not useful, 3 = somewhat useful, 5 = rather useful, 7 = very useful.
source, while questions IS7.5 and IS8.1 to 8.15 measured the FP SCORES variables (Familiarity, Proximity, etc.). The order of items in questions IS8.1 to 8.15 was randomized to conceal the nature of the variables of interest, and several items were worded negatively to reduce the risk of monotonous responses. The relationships between the items in questions IS7.1 to 8.15 and the concepts they were intended to measure are shown in Table 6.1 and Table 6.2. I discuss the validity and reliability of these measures later in this chapter.
119
Table 6.2 Items in the Information Sources questionnaire and the FP SCORES variables they were intended to measure. Item wording (translated)a
Variable
Q#
Familiarity
8.11 You know the type of information [source] has. 8.14 You know [source] well. 7.5
Nearest place where [source] is usually located.b
8.2
[Source] is difficult to find.
Structure
8.3
Obtaining information from [source] is an important part of your job.
Capacity
8.4
[Source] is the most credible source.
Proximity
8.10 [Source] has more complete information than do other sources. Openness
8.5
Compared with other sources, [source] is easy to use for extension purposes.
8.9
[Source] provides information in a ready-to-use form.
8.15 Compared with other sources of information about agricultural technology, [source] is easy to understand. Reward
8.3
Much information from [source] is not relevant to problems faced by farmers in your area.
8.8
[Source] often discusses topics that are closely related to conditions in your area.
8.12 [Source] often provides information that is not new to you. Energy
8.7
You devote a lot of effort to obtain information from [source].
Synergy
8.1
Information from [source] often does not agree with information from other sources.
8.6
[Source] usually has information that is currently needed (timely).
a
See footnote a to Table 6.1. Unless otherwise indicated, responses were measured on a scale from 1 to 7 (1 = disagree, 3 = somewhat disagree, 5 = somewhat agree, 7 = agree). b
1 = in own office, 2 = in other room in same building, 3 = in other building, 4 = in same town, 5 = < 50 km away, 6 < 150 km away, 7 = >150 km away.
120 Publications questionnaire The Questionnaire about Publications Used by Extension Specialists (Appendix 2) evaluated extensionists' problems, their use of AARD publications, and their information needs. A single version of this instrument was produced. Questions P1 and 2 measured the respondent's personal characteristics: institutional affiliation, specialization, number of years experience, and gender. These questions were identical to items in the Information Sources questionnaire. Questions P3 and 4 sought to identify major problems extensionists faced in their work as a whole (question P3) and in obtaining information (question P4). The wording of question P3 was based on Sigman and Swanson's (1984) evaluation of problems facing extension systems worldwide. Changes from their questions included (1) simplified wording and translation into Indonesian, (2) division of their category of "linkage" into two: "obtaining information" and "feedback," (3) the addition of the category "reward," and (4) the use of a 7-point response scale instead of one containing 3 points. Items in question P4 sought to measure the relative importance of various problems mentioned by research and extension personnel during planning and by respondents during pretest interviews. Questions P5 and 6 compared the frequency respondents read and the usefulness of seven publication types. These included Indonesia's agricultural newspaper (Sinar Tani), three publication series produced by the Agricultural Information Centers (Liptan, Buletin Informasi Pertanian, and booklets), and three AARD publications types (Warta Litbang, scientific journals, and books). The pretest had included the farmers' magazine Setia Kawan, but so few of the pretest respondents had heard of this magazine that I excluded it from the full survey. Questions P7 and 8 sought to measure the actual information flow from AARD to respondents. Question P7 listed six items of food cropping technology recently developed by AARD. Respondents specializing in food crops were asked if they had obtained information about each item, and if so, from which source. Question P8 posed the same questions about six livestock technologies to livestock specialist respondents. Question P9 aimed to identify topics about which specialists most needed information. Sampling Sample size A sample size is a compromise between the desirable and the possible. The desired size depends on the amount of variation within the population and degree of accuracy and certainty required. The possible size depends on the resources available for the sample collection. In order to calculate the desired total sample size, it was necessary to determine the
121 size of subgroups that would be compared (McCall 1982:202). For the Information Sources questionnaire, the design called for a maximum of four subgroups (corresponding to the four versions of the questionnaire) to be compared within three institution types (province, district, and AIC). For the Publications questionnaire, a maximum of three subgroups (province, district, and AIC) were to be compared. I determined the desired sample sizes for each questionnaire were determined using a procedure given by Steel and Torrie (1980): r ≥ 2(Zα/2 + Zß)2 (σ/δ )2, where r=
The number of replications (= respondents per subgroup).
Z=
The value of the standard normal variable.
The population standard deviation. The pretest indicated that standard σ= deviations for Information Flow and the FP SCORES variables were in the range 0.95 to 1.77 on a scale from 1 to 7. I selected 1.77 as the value to use for σ. α = The desired confidence level (set to 0.05). This is divided by two for a two-tailed test. Zα/2 thus equals 1.96. ß= The assurance with which it is desired to detect the difference δ . (1 - ß = the power of the test). ß was set to 0.2; Zß thus equals 0.85. The minimum size of the difference to be detected. I set this arbitrarily to δ= equal σ. In other words, the smallest difference that could be detected with confidence would be equal to the standard deviation of the variable being tested. These calculations yielded an r of 15.79 (rounded up to 16) for the smallest subgroup. Information Sources questionnaire Based on the lists of SMSs collected (see below and Table 6.3), about 61% of SMSs were based at district offices, 28% at province offices, and the remaining 11% at AICs. The small numbers at AICs meant it was necessary to stratify the sample, over-selecting AIC specialists in order to obtain enough AIC respondents to respond to each version of the Information Sources questionnaire. It was neither possible nor necessary to sample province- and district-level specialists separately. It was not possible because of the address lists I had: for numerous SMSs I had an address "in care of" the provincial Kanwil or Bimas office only, without any indication of whether the specialist worked at the province or district level. It was not necessary because the numbers of province-level specialists (Table 6.3) were sufficient to allow pooling of province and district personnel and draw a common sample, being
122 reasonably confident of obtaining the minimum of 16 respondents in each subgroup. Multiplying by the minimum sample size of 16 by the proportion of district to province specialists (956/437) yielded 35, the number of district specialists desired in each subsample. The desired sample size for each questionnaire version was thus 16 + 35 = 51. Multiplying by four gave 202 as a total desired sample size for the Information Sources questionnaire. Multiplying this figure by 1.5 to compensate for an expected response rate of 67% yielded a total sample of 303 province and district specialists. For the AIC SMSs, I also aimed at a minimum subgroup size of 16. Multiplying this figure by three (only three of the four questionnaire versions were sent to AIC specialists) and by 1.5 to allow for non-responses yielded a total of 72 AIC SMSs to be sampled. Publications questionnaire Increasing the sample size above the minimum enables more confident estimation of smaller differences between means. The resources available allowed me to send the Publications questionnaire to 150 province and district specialists and 67 AIC specialists. After allowing for expected non-responses, this would yield a minimum sample size of 31 province SMSs, allowing differences of 0.7 σ to be detected with the confidence levels given above. Developing a list of specialists Respondents for the mailed survey were selected from a list of all Indonesian extension subject-matter specialists. Because no one organization at the national level is responsible for coordinating the specialists, no complete, up-to-date national list of them existed. It was therefore necessary to develop one. I obtained lists of specialists in each province from various branches of the Ministry of Agriculture: Bimas, the Secretariat-General and AAET in Jakarta, and the Agricultural Training Institute (Balai Penataran dan Latihan Pertanian) in Ciawi, West Java. These lists varied in their date and the types of information they contained: some included names and specializations but no addresses, while others showed names and addresses but no specializations. The oldest lists dated from 1988, the newest from 1991. By comparing these lists, I developed an aggregate list for each of Indonesia's 27 provinces, containing the specialists' names, work addresses and areas of specialization. These lists were mailed to the corresponding provincial offices of the Ministry of Agriculture (Kanwil, the body responsible for coordinating the SMSs' activities in that province), with a request to update and return it. Updates were received from 10 of the 27 provinces. Revisions for West Java were obtained by interviewing the specialists' coordinator in Bandung. A repeat request to the remaining 16 provinces failed to generate any response. A total of 1560 names and addresses were collected in this way. Because of the difficulty in compiling this master list and rapid turnover among specialists, doubt must be cast on the validity of the list. An indication of its accuracy can be obtained from the number of corrections received from the 11 provinces providing these: Of 575 SMSs in the original lists sent to the Kanwil offices in these provinces, 118 (20%)
123
Table 6.3
Numbers of specialists sent questionnaires. Provincea
Districta
AIC publications
22
53
0b
75
Agricultural press
24
53
28
105
AARD publications
25
53
28
106
Other specialists
31
45
27
103
102
204
83
389
44
108
67
219
Grand total
146
312
150
608
Estimated population
437
956
167
1560
Questionnaire/version
AIC
Total
Information sources questionnaire
Total Publications questionnaire
a
Province and district specialists were combined during the drawing of the samples. b
None sent because AIC specialists author AIC publications.
had changed address within the province, 29 (5%) were no longer extension specialists, and an additional 119 SMSs (20%) had been appointed. We might expect similar proportions to apply to the 985 specialists in the remaining 16 provinces which had not provided updated information. Taking the 11 provinces' corrections into account, this would mean that overall, some 184 of 1560 specialists (12%) in the master list had moved, 45 (3%) were no longer extension specialists, and another 185 newly appointed personnel (12%) are not represented. The total of 1700 specialists these calculations yield is an underestimate in light of President Soeharto's statement in August 1991 that the Ministry of Agriculture then employed 2247 specialists. This may mean that recently hired SMSs are under-represented in the sample. However, this causes little concern because recent hires are least likely to be familiar with local conditions or available information sources or to have developed firm information seeking habits. Sample selection The pretests and interviews suggested that specialists at province and district offices and at AICs obtained information from rather different sources. However, AIC specialists were much less numerous than those at the other two institutions. This meant it was necessary to oversample AIC specialists to obtain a sufficiently large n for statistical tests comparing them with personnel at the other institutions.
124 I therefore sampled the two groups separately, selecting nearly 90% of all AIC specialists to complete either one or the other of the instruments, but only one-third of all provincial and district specialists (Table 7.2). I drew the samples at random from the master list of 1560 names. Specialists who had been interviewed or participated in the pretests were excluded before the samples were drawn. All samples were mutually exclusive. The sampling scheme is summarized in Table 6.3. AIC specialists Of the 1560 specialists in the master list, 167 worked at AICs, and 1393 at provincial and district level agricultural offices. I randomly selected 67 of those working at AICs as respondents for the publications questionnaire. Of the remaining 100 AIC specialists, 83 received the Information Sources questionnaire (the other 17 had been interviewed or participated in the pretest). I randomly divided these 83 persons into three groups: 28 received version 2 of the questionnaire (asking about the agricultural press), 28 received version 3 (AARD publications), and 27 received version 4 (other specialists). I did not send version 1 (AIC publications) specialists at the AICs because these specialists are responsible for producing these publications. Province and district specialists I deleted names of those specialists known to specialize in topics other than food crops and livestock (e.g., in fisheries or estate crops) from the list of 1393 specialists working at provincial and district level offices. Some province lists classified specialists by discipline (e.g., economics, soil and water conservation) rather than by commodity. This made it difficult to determine whether the specialist was engaged in food crops or livestock activities. In such cases, included only those disciplines clearly related to food crops (e.g., agronomy) or to livestock. While this meant that some provinces (e.g., East Java) are under-represented in the sample, I felt this was better than risking the bias of including large numbers of respondents from outside the food crops and livestock areas. From the remaining exactly 1000 names, I randomly selected 152 to receive the publications questionnaire. I chose another 306 specialists at random from this same list to receive the information sources questionnaire. Of these, a random 75 were sent version 1 (focusing on AIC publications), 77 received version 2 (agricultural press), 78 got version 3 (AARD publications), while 76 were mailed version 4 (other specialists). Distribution logistics The questionnaires were mailed on 17-24 September 1991. Respondents at AICs received theirs via the directors of their institutes. In addition to the questionnaire, each mailing envelope contained: • A copy of a letter from the Directorate of Food Crops Extension (or Livestock Extension, depending on the recipient's area of specialization) indicating the importance of
125 the research • A letter from the director of the Central Research Institute for Animal Science, requesting respondents to reply to the questionnaire. •
Brief instructions on completing and returning the questionnaire.
• A stamped, addressed return envelope bearing the letterhead of the Central Research Institute for Animal Science. Each questionnaire bore a unique identification number. The anonymity of all responses was guaranteed in the instruction sheet. A small number of specialists completed questionnaires during personal interviews in North Sumatra and West Java. These have been added into the totals shown in Table 6.3. Data cleaning and additional variables I entered coded data on computer disk using the dBase III+ database program (Ashton-Tate 1986). I imported the data into SPSS PC+ version 4 (Norusis/SPSS 1990) and cleaned them using the SPSS Data Entry program. All analysis used SPSS PC+ and Lotus 1-2-3 (Lotus Development Corporation 1991). Missing values Respondents occasionally omitted to answer a question or gave more than one answer. I coded such instances as missing and excluded them from analysis involving that variable. A response of "Don't know" (for instance in question IS8) was coded as missing. One exception was necessary because of an error in the question wording in Indonesian: a response of "0" ("don't know") to question IS8.11 ("You know the type of information [source] has") was coded as "Disagree": someone who did not know the type of information could correctly circle either "don't know" or "disagree." It was necessary to recode 25 responses (9%) in this way. (This item was later dropped from the analysis anyway because of reliability problems.) Outliers Questions IS3 and 9 to 11 invited respondents to indicate the number of times they had performed a particular activity within a given period. A small number of specialists wrote unusually or unrealistically large numbers in response. I identified such cases using the SPSS EXAMINE procedure (Norusis/SPSS 1990:C -60) and treated them as missing for the variable involved.
126
Data transformations I reversed the coding of negatively worded items in the Information Sources questionnaire to ensure that high scores corresponded to high expected information flows. For uniformity, I converted responses to questions IS7.1 to 7.3 from a scale from 1 to 9 to one from 1 to 7 using a simple linear transformation: new score = ([old score - 1] × 6/8) + 1. Levels of measurement Several of the variables used in this study were measured on ordinal scales: questions IS5, 7, 8 and 12, and P3 to P6). Throughout, I assumed that these were measured at the interval rather than the ordinal level. This assumption is justified for four reasons: • The scales used approximate an interval scale, and their use as such is common in the social sciences. •
It enables the construction of indices composed of multiple variables.
• It allows the use of powerful parametric techniques such as multiple regression, correlation and reliability analysis • Such techniques as multiple regression, etc., are relatively robust with regard to minor violations of some assumptions. New variables I calculated or entered the following additional variables: Information Sources questionnaire •
Specialization, taken from question IS2.2.
• Wealth index, calculated as the total number of items the respondent reported owning in question 2.9. •
Province, taken from the respondent's address.
• Inner Island location, reflecting the respondent's location in the densely populated and relatively well developed "Inner Islands" of Java, Bali and Madura, as opposed to the sparsely populated Outer Islands. • Province category, reflecting beliefs commonly held in Indonesia about the level of agricultural development in each province: 2 (high) for Java and Bali, 1 for North and South Sumatra, Lampung, North and South Sulawesi; 0 (low) for all other provinces.
127 • Total information obtained, the mean of scores from all 24 information sources listed in question 5. •
Information flow and FP SCORES indices (see below).
I was interested in the effect distance has on the SMSs' use of sources. For each respondent, I therefore measured map distances from the respondent's office location to the nearest known source of each type. For simplicity, I used direct straight-line distances (i.e., not road or true air or sea distances). The maximum distance measured was 650 kilometers (from Merauke in Irian Jaya to Jayapura, the provincial capital). The distance to the nearest known location for each source depended on the source type the respondent was questioned on: • For AIC publications: the number of kilometers to nearest Agricultural Information Center (located in or near the provincial capital). • For the agricultural press: All respondents are assumed to receive Sinar Tani, so the distance was set to 0 km (but see below). • For AARD publications: the number of kilometers to the nearest AARD food crops (or livestock, depending on the respondent's specialty) research institute or subinstitute. The distance to the provincial AIC was used for respondents in other specialties, or if the AIC was closer than the AARD site. This was because the AICs are supposed to receive all AARD publications. • For other SMSs as sources: the number of kilometers to the nearest other SMS in the respondent's commodity group (e.g., food crops). The location and specialty of other SMSs was taken from the master list of 1560 addresses and plotted on the map to allow distances to be measured. Respondents with colleagues in the same institution were assigned a distance of 0 km; those with colleagues in a different institution in the same town were assigned a score of 5 km. I took the natural logarithm of the number of kilometers to approximate the nonlinear change in friction of distance at greater distances. Distances of zero (natural log undefined) were allocated a score of 1 km. The logarithm scores were linearly transformed and reversed to form a 7-point scale, so high scores corresponded to high proximity. Publications questionnaire •
Specialization, taken from question P2.2.
• Number of items on which respondents had information: the sum of answers to the first part of questions P7 and 8, asking whether the respondent had obtained information about six recent food crops (question P7) or livestock (question P8) technologies. Similar sums were calculated for each of the sources named in second parts of these questions.
128 • Information needs indices for each of the commodity groupings (rows) and disciplines (columns) in question P9. For instance, the index for rice information need was the number of boxes checked in the first row in this question; that for cultivation was number checked in the first column under the "food crops" rubric. I also calculated total information needs for food crops and for livestock information: the food crops index included the 24 food crops boxes, plus those for "estate and industrial crops," "machinery and equipment," and "irrigation." That for livestock included the 24 livestock boxes, plus "livestock fodder plants," "milk handling," and "livestock post-harvest." Initial analysis Significance level A level of p = 0.05 was selected as the criterion for significance. Frequencies and point estimates I calculated frequencies and appropriate descriptive statistics (e.g., mean, standard deviation, mode, median) for all variables. I ranked comparable variables (such as information flows from various sources, questions IS5.1 to 5.24) by their mean value for ease of analysis and presentation. The survey used a stratified sample containing two subgroups of specialists (those working at AICs and those at provincial and district offices), and different proportions of the two groups were sampled. Because of this, it is necessary to weight responses according to their proportions of the total population when making point estimates of the whole population of extension specialists (e.g., when estimating the proportion of all specialists who are women). The presentation of results distinguishes between the various samples as necessary. I weighted means when calculating point estimates for the population of specialists as a whole. Table 7.2 gives the weights used for this. Differences among groups I used two main statistical techniques to test for differences among groups of respondents. These were the chi-square for count data and one-way analysis of variance for ordinal (interval) data. If the analysis of variance identified significant variation, I used Student-Newman-Keul's and Scheffé's tests or contrasts using t-tests to test for differences among pairs of groups (Norusis/SPSS 1990:C-45, Steel and Torrie 1980:177-187). I used one-way analysis of variance to test for differences among respondents grouped by situation (institutional affiliation, institutional level, province, Inner Island location, province development category) and demographics (sex, education, rural background, specialization, engagement in outside work). I used regressions to gauge the effect of continuous variables such as wealth (index developed from question IS2.9) on specialists' responses.
129
Influences on information sources Uniformity of subsamples The analysis of factors affecting information flows rests on the assumption that the subsamples of respondents to the four versions of the questionnaire did not differ in important ways from each other. I tested this using chi-square (for categorical data) and oneway analysis of variance (for "interval" data) (see the earlier discussion of the assumption of interval-level data). I tested all major variables (51 in all) other than those in the Information Flow and FP SCORES indices (see below) for differences among the four subsamples. Subsets of respondents The survey design required that analysis take into account two categorical variables: • The respondents' institution (provincial vs district vs AIC). This is necessary because AIC specialists were oversampled and because pretests and interviews showed that these three groups have rather different characteristics. • The source type each respondent was questioned on (AIC publications vs agricultural press vs AARD publications vs other specialists. This must be taken into account because a missing cell (AIC respondents were not questioned about AIC publications) in the data may otherwise bias results. It is thus possible to compare subsets of responses: across source types or across institutions. While they are necessary for the reasons outlined above, comparisons among subsets of respondents are also valuable: • They can reveal differences among subsets. For instance, respondents at AICs may consistently score higher on a variable than their colleagues at district offices. Discovering this is a first step to finding out why. • They provide internal replications for tests of hypotheses. If a relationship holds over several subsets of respondents, it can be considered robust. If it fails to hold, doubt must be thrown on its wider applicability. Most studies fail to include such internal replications.
130
Scale construction Two problems were encountered in constructing multi-item indices for FP SCORES variables. First, items in some of the a priori scales (Table 6.1 and Table 6.2) were relatively poorly correlated with other items in the same scale. This raised questions about the validity of the indices, and meant that scales constructed solely on the basis of the a priori groupings would have low reliability. The second problem was that some items in an a priori scale were relatively highly correlated with items in other scales. This raised the danger of multicollinearity among independent variables, which would cause problems in interpreting regression results. These problems are discussed further in Chapter 9. I used five criteria to deal with these problems and ensure a consistent approach to index construction. I included an item in an index if: •
It contributed to the face validity of the index.
•
It was significantly correlated with all other items in the index.
•
It had an item-total correlation greater than r = 0.3.
•
Cronbach's α of the index would not be raised by deleting the variable.
•
All items in an index loaded onto the same factor in factor analysis. Details on these criteria are given below.
• Face validity was based on the groupings in Table 6.1 and Table 6.2. I did not consider an item for inclusion in any other scales than the one for which it is listed in these tables. If a choice had to be made between deleting one variable or another, I retained the item with the higher face correspondence with the concept in question. • Inter-item correlations I calculated simple Pearson correlation coefficients for pairs of items in each a priori scale. An item was retained in the index only if it was significantly (at p = 0.05) correlated with all other items in the scale. • Item-total correlation SPSS's RELIABILITY procedure (Norusis/SPSS 1990:C-77) was used to calculate item-total correlations. The item-total correlation measures the correlation between each item in the scale with the sum of all variables in the scale, excluding the item in question. I considered for possible deletion variables with such correlations of less than 0.3. • Cronbach's α RELIABILITY also calculates Cronbach's α , a measure of the overall mutual correlation among scale items. I removed items from an index if doing so would raise the value of α .
131 • Factor analysis I performed factor analysis on the responses to Questions IS7.5 and 8.1 to 8.15 (the hypothesized influences on information flows). The analysis used principal component extraction and the varimax rotation with pairwise deletion of missing variables. I considered for deletion any item from an a priori scale that loaded most heavily on a factor different from others in the scale. The five criteria were occasionally in conflict. For instance, two variables in the same a priori index might load onto the same factor but have a mutual correlation lower than the criterion level. Or an item loaded onto a different factor from others in its a priori scale, but deleting it would significantly decrease the Cronbach's α of the scale as a whole. Judgements in such instances were based primarily on the face validity and item-total correlation of the scale. They are described below in Chapter 9. I constructed the Information Flow and FP SCORES indices by summing responses to items remaining after evaluating these criteria and dividing by the number of items in the index. The result was a series of indices with a potential range from 1 (low) to 7 (high). Because of the small number of missing scores (a maximum of 13 cases, or 3.4%), I made no attempt to estimate any missing values. Differences among source types and institutions I used one-way analysis of variance to test for differences in Information Flow and each of the FP SCORES variables among respondents at the three institutions (province and district offices and AICs) and among the four source types (AIC publications, agricultural press, AARD publications, and other specialists). I also tested for differences between source types within each institution, and for differences between institutions within each source type. This was necessary because one source-type-by-institution combination was not surveyed: AIC specialists' use of AIC publications. This was because the AIC specialists themselves author these publications. This omission might bias any comparison that includes either AIC specialists or publications. Influences on information flow I ran several regression models with the information flow index as the dependent variable, using SPSS procedure REGRESSION (Norusis/SPSS 1990:C -53). Individual influences of FP SCORES variables If an SMS knows a source well, will he or she get a lot of information from it? If we assume that everything else is constant, we can answer this question by testing the relationship between Familiarity and Information Flow. I did this by running a simple regression using Familiarity as the predictor and Information Flow as the dependent variable. I did this for all respondents. I wished to know if specialists in the three institutions differed. For instance, is Familiarity with a source important for AIC specialists but not for their district counterparts? If specialists at the three institutions respond in similar ways, then the three groups can be
132 combined. If not, they must be treated separately. I tested this by running separate simple regressions for each institution for Familiarity as a predictor of Information Flow. Similarly for the four source types: Familiarity may be important for AIC publications but not for AARD publications, for instance. I tested this by running a similar set of simple regressions for each of the four sources. In order to gauge the overall effect of Familiarity on Information Flow after controlling for the source type, I ran a multiple regression using Familiarity and three dummy variables representing the source types as independent variables. To avoid the potential bias due to the missing source-type-by-institution combination (AIC specialists and AIC publications), I also ran a series of simple regressions using Familiarity to predict Information Flow for each of the 11 remaining combinations of sources and institutions. I used the z-test (Walker and Lev 1953:255) to discover differences among regression coefficients within each source type or institution. I repeated the above analyses using each of the other seven FP SCORES indices as a predictor of Information Flow. Combined effects of FP SCORES variables How do the FP SCORES variables affect Information Flow when all are considered together? To answer this question, I ran several multiple regression models. I first investigated the correlations among the FP SCORES variables to check for possible multicollinearity problems. I then ran a multiple regression model using the FP SCORES indices as predictors of Information Flow. First, FP SCORES variables were entered by stepwise forward regression with a minimum significance level of p = 0.05 for inclusion in the equation. Finally, all remaining FP SCORES variables were entered into the model. The same model was run a second time, controlling for the source type and institution. Dummy variables for these variables were entered into the equation before the stepwise inclusion of the FP SCORES indices. Entry of remaining FP SCORES variables was then forced. Comparison among sources and institutions To discover any relationships specific to individual sources or institutions, I also ran multiple regressions for seven subsets of cases: for each of the four sources and the three institution types surveyed. For each subset, I ran these regressions with and without controlling for source or institution. Violations of assumptions I inspected scatterplots and histograms of residuals for evidence of violations of regression assumptions such as non-normality of residuals and heteroscedasticity. I used variance inflation factors and other collinearity diagnostics to detect multicollinearity in the model.
135
CHAPTER 7 SPECIALISTS' EXTENSION ACTIVITIES Introduction This chapter summarizes the characteristics of the survey respondents (responses to questions IS1-4 and P1-2), their extension activities (questions IS9-12), and the problems they face (questions P3-4). The next chapter describes the information sources they use (IS5-6 and P5-8) and their information needs (P9). The following chapter attempts to explain why respondents use the sources they do (questions IS5-6). Frequencies of responses for both questionnaires are given in Appendices 5 and 6. Presentation of results I have used three methods of presenting numerical material in this and subsequent chapters: text, tables and graphs. The graphs are discussed in Chapter 9. I here describe the conventions used in the text and tables to indicate the statistical significance of findings. A single asterisk (*) indicates a statistically significant difference from zero at the p = 0.05 level. Two asterisks (**) denote significance at p = 0.01. These apply to chi-square values and regression slopes. Several tables compare values of interval-level variables among subgroups of respondents (e.g., Table 7.11). Table 7.1 (drawing on the data presented in Table 9.7) provides an example. The letters a, b and c denote vertical comparisons -- across rows within the same column. The letters x, y and z denote horizontal comparisons -- across columns within the same row.
136 Thus in the first column (dashed outlines), AIC publications (4.94 a), the agricultural Table 7.1 Example of presentation of statistically significant differences among subgroups of respondents as identified by analysis of variance.a Source type
Institution Province
District
Overall AIC
AIC publications
4.94 a
5.34 a
5.21 a
Agricultural press
5.50 a
5.39 a
6.18
5.63 a
AARD publications
5.75 a xy
5.26 a x
6.19 y
5.63 a
Other specialists
6.65 b
6.53 b
6.53
6.56 b
Overall
5.71 x
5.61 x
6.29 y
5.79
a
Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized. See the text for discussion of the outlined and shaded cells. press (5.50 a), and AARD publications (5.75 a) are all followed by the letter a; they are thus not significantly different from each other. All are, however, significantly different from other specialists (6.65 b) -- the only figure in the column followed by a b. The same pattern is true for the second column, (district specialists) and the fourth (Overall). The third column contains no a's or b's, so no significant differences exist among the three figures in this column. Comparisons across columns within the same row are denoted by the letters x and y. Thus for AARD publications (row outlined by dotted lines), district specialists (5.26 x) are not significantly different from province specialists (5.75 xy), since both figures are followed by an x. Similarly, AIC specialists (6.19 y) are not different from province SMSs (5.75 xy) but are different from those at district offices (5.26 x). Of the other four rows, only the final one (Overall) contained significant differences, as denoted by the letters x and y following figures in this row. Note that these comparisons are only valid within a single column or row. For instance, the table does not tell us whether the two shaded cells in the table are significantly different. Nor are comparisons between a subgroup (e.g., either of the shaded cells) and an "overall" figure (rightmost column or the bottom row) valid. To ease the rapid reading of some tables, I have used boldface to denote the highest value in a column, and italics to mark the smallest. Superscript letters a, b, etc., are used in tables to denote footnotes.
137
Table 7.2 Sampling and response rates from Information Sources and Publications questionnaires. Estimated population a
Weighting factor
Province
District
AIC
Total
437
956
167
1560
28%
61%
11%
100%
Information sources questionnaire Sample size Valid responses
103 68
203 66%
149
83 73%
63
389 76%
280
72%
Publications questionnaire Sample size
44
Valid responses
36
108 82%
73
67 68%
56
219 84%
165
75%
Overall Sample size
147
Valid responses
104
311 71%
222
150 71%
119
608 79%
445
73%
a
Figures used to weight data when calculating point estimates. Note that these weights assume that fisheries and estate crops SMSs (excluded from the non-AIC sample) have similar characteristics to the food crops and livestock specialists surveyed.
Response rate A total of 456 people returned the questionnaires before the cutoff date of 4 January 1992 (110 days after the first mailing). This represented a response rate of 75%. Of these, 11 persons indicated they were not extension subject-matter specialists, leaving 445 valid responses (73%). Table 7.2 presents sampling and response rates for the various groups sampled. Around 80% of Agricultural Information Center personnel returned questionnaires, while only 70% of provincial and district officials did so. Response rates varied among provinces also, with 100% of all specialists surveyed in Bengkulu and Central Sulawesi returning their questionnaire, but only 44% in Lampung, 53% in East Kalimantan, 54% in North Sumatra, and 56% in Aceh doing so. The reasons for the differences among provinces are unclear. Surprisingly, updated address lists did not seem to produce significantly higher response rates than the older lists. The overall response rate of 73% can be regarded as excellent, especially given the doubtful validity of the original address lists. For those 11 provinces where updated (1991)
138
Table 7.3
Respondents' institutional affiliationa. Province
District
AIC
Overall %b
n Kanwil
8
Dinas
46
Bimas AIC/other Total
% 7.7 44.2
n
% -
n
c
102
49
47.1
120
1
1.0
0
104
100
222
45.9 54.1 100
% -
c
2.2
-
c
40.5
-
c
46.3
119
100
11.0
119
100
100
a
Combined data from both questionnaires, questions IS2.1 and P2.1. n = 445.
b
Weighted for proportions of total specialist population in each group.
c
Does not apply.
lists were available, the updates indicated that we might expect some 15% of the total addresses sampled to be incorrect. Allowing for this, we can calculate a true overall response rate of 88% by current extension specialists sampled. In view of the adequate response, it was not considered necessary to send a follow-up mailing to non-respondents. Personal characteristics The two questionnaires contained several identical items about the respondents' personal characteristics: institutional affiliation, specialization, years as an extension specialist, and gender. These data are combined here for convenience of presentation. There were no significant differences in personal characteristics between the specialists responding to the two questionnaires except, inexplicably, a relative over-representation of women among Publications questionnaire respondents. Institutional affiliation Two-thirds (68.1%) of the non-AIC specialists were based at district offices, with slightly more at Bimas than at Dinas institutions (Table 7.3). Roughly equal numbers were based at provincial Bimas and Dinas offices, while few worked at the Kanwil. Weighting the responses for the different sampling percentages, the survey revealed that about 2% of all Indonesia's extension specialists work at Kanwils (about one person in each province), 12.4% are based at provincial Dinas and 13.2% at provincial Bimas offices, and 10.7% at Agricultural Information Centers. More than half the specialists work at district Dinas (28.2%) and Bimas offices (33.1%).
139 Differences in the duties and information seeking behavior of specialists based at the Bimas and Dinas offices were minor, and they are treated as a single group in subsequent analysis. Any distinction between extension personnel in these offices is likely to disappear with the current changes in the organizational structure of extension in Indonesia. Tests showed that provincial and district level staff differed significantly in their extension activities (see 7.14 and Table 8.1). Subsequent analysis therefore separated these two groups. One hundred and nineteen specialists worked at Indonesia's 28 Agricultural Information Centers (one national and 27 provincial). These, too, had rather different duties from the provincial and district staff, so are kept separate in analysis. Specialization Differences in policies among Indonesia's provinces and in respondents' reporting mean it is difficult to assign some respondents to certain categories. In some provinces, extensionists specialize in the broad commodity groups of "food crops" or "livestock." In others they are given more specific disciplines or sub-specialties, such as "soil conservation," "mechanization," or "economics," but continue to focus on a certain commodity group. In still other provinces, and in the AICs, some specialists are entirely discipline- rather than commodity-oriented. Some respondents reported their sub-specialty but not their commodity group (if indeed they had one). Table 7.4 reflects these difficulties in categorizing such respondents. This table shows that roughly one-quarter of the provincial and district respondents specialize in livestock, while two-thirds focus on food crops. A small number of respondents specializing in fisheries and other topics were inadvertently included in the
Table 7.4
Specialization of respondentsa. Province n
District %
n
AIC %
n
%
Livestock
23
23.0
57
27.1
26
22.4
Food cropsb
66
66.0
144
68.6
48
41.4
Otherc
11
11.0
9
4.3
42
36.2
Total
100
100
210
100
116
100
a
Combined data from both questionnaires, questions IS2.2 and P2.2. n = 426.
b
Includes 20 social science and economics specialists and 25 soils, mechanization and post-harvest specialists who probably focus primarily on food crops. c
Includes estate crops, fisheries, conservation and extension. Questionnaires were not sent to respondents at district and provincial offices known to be in these categories. Estimates for the whole specialist population therefore cannot be calculated.
140 sample. They are excluded from subsequent analysis that compares the two major groupings, food crops and livestock. Table 7.5
Education of respondents by institutional affiliationa. Province
District
AIC
Overall %b
n
%
n
%
n
%
Sarjana
48
75.0
137
100
51
86.4
91.5
Master's
16
25.0
0
0
8
13.6
8.5
Total
64
100
137
100
59
100
100
Chi square = 34.42**. a
Data from question IS2.3. n = 260.
b
Weighted for proportions of total specialist population in each group.
Education All specialists providing usable responses held at least a sarjana degree (four years plus thesis, roughly equivalent to a U.S. honors bachelor's degree). Several held a master's in addition (Table 7.5). None had a doctorate: presumably Ph.D.s are promoted out of the specialist role when they obtain their degree. A number stated that they had attended a university but did not indicate whether they had earned a degree. They are excluded from this analysis. None of the specialists at the district level had a master's degree, while one-quarter of those at the provincial level did. AIC respondents were intermediate. Overall, an estimated 8.5% of all specialists held a master's degree. It seems either that provincial level staff have greater opportunities for advanced study, or that those with advanced degrees are transferred to provincial offices as they gain their degrees. The basic salary for Indonesian government officials is low, and the opportunity for further study is often used as a reward for superior service. Work experience The median length of service as an extension specialist was six years, with responses ranging from less than a year to 20 years. More than 80% had 8 years or less experience in such positions, reflecting how recently most specialists' positions had been established and possibly the rapidity of turnover (including promotion to higher positions) among specialists.
141 Respondents at provincial offices had significantly more work experience (mean = 9.2 years) than those at the AICs (6.1 years) or the district level (6.0 years) (significant at p = 0.05 level by Student-Newman-Keul's test). This reinforces the suspicion that senior specialists transfer to provincial offices after working for a period at the district level. Gender Women were under-represented in all the institutions surveyed, and account for only 20% of all specialists (estimated by weighting for the different sampling rates). They are especially uncommon at the provincial and district offices, where less than one in five specialists was female (Table 7.6). At the Agricultural Information Centers, however, they accounted for one-third of respondents. The reasons for this cannot be ascertained from these data, but may be related to the nature of the work. The jobs of district specialists involve much travel, often over difficult terrain and with overnight stays (see Table 7.14). These specialists also meet with more farmers, and the Indonesian extension services usually deal with male rather than female farmers. Indonesian society traditionally deters (though does not prohibit) women from engaging in such activities, particularly alone. The AIC specialists (and province) typically travel less and meet fewer farmers. Qualified women may therefore eschew the district and provincial level specialist jobs. On the other hand, the unbalanced numbers of males and females in the provincial and district level jobs may reflect a lack of female graduates in agriculture or a hiring bias against women on the part of local authorities. Further investigation would be necessary to determine the causes and effects of this imbalance.
Table 7.6
Gender of respondents by institutional affiliation. Province n
District
%
n
Overall%b
AIC %
n
%
Male
83
79.8
184
82.9
78
66.7
80.3
Female
21
20.2
38
17.1
39
33.3
19.7
100
222
100
117
100
Total
104 a
100 b
Chi square = 11.95** Combined data from questions IS2.5 and P2.4. n = 443. Weighted for proportions of total specialist population in each group.
142
Urban/rural background of respondents by institutional affiliation.a
Table 7.7
Province n
District
%
n
Overall%b
AIC
%
n
%
Urban
34
50.0
69
46.6
40
63.5
49.4
Rural
34
50.0
79
53.4
23
36.5
50.6
Total
68
100
148
100
63
100
100
Chi square = 5.09ns (significant at p = 0.078).Chi square for (province + district) vs AIC = 4.87*a Data from question IS2.6. n = 279.b Weighted for proportions of total specialist population in each group.
Rural and farming backgrounds About half of the respondents at provincial and district offices were brought up in a village, while AIC personnel were rather more urban in background (Table 7.7). This difference was significant at p = 0.05 (chi-square = 4.87*). More AIC staff than local personnel stemmed from non-farm families (35% compared to 47%), though this difference was not significant (Table 7.8). Specialists brought up in the countryside tended to come from a farm family: of 135 respondents brought up in a village, 100 claimed a farm background, while only 22 of 139 specialists from urban areas did so (chi-square = 94.07**). The relatively large proportion of specialists with rural backgrounds could be seen as encouraging given the apparent urban bias among Indonesian government officials. But that nearly half of the country's extensionists do not have such a background is somewhat of concern given the need for extension specialists to understand and empathize with farm
Table 7.8
Farm background of respondents by institutional affiliation. Province n
%
District
Overall%b
AIC
n
%
n
%
Farm family
36
52.9
79
53.4
38
64.4
54.4
Not farm family
32
47.1
69
46.6
21
35.6
45.6
Total
68
100
148
100
59
100
100
Chi square = 2.34ns.a Data from question IS2.7. n = 275.b of total specialist population in each group.
Weighted for proportions
143 families. Other activities About 15% of respondents at provincial offices and AICs reported that they had outside income-generating activities, while significantly more district specialists (over one-third) did so (Table 7.9). Overall, about one-quarter of Indonesia's extension specialists are estimated to have outside work. (This figure does not include any extra income respondents may receive in the form of travel per diems or emoluments for service on committees.) The most common types of additional activities were teaching at high schools and universities (26 respondents) and additional duties in the specialist's office, such as coordinating the local Bimas program (20 respondents). Only 12 individuals (11 of whom were at district offices) were engaged in farming. There is considerable economic pressure on extensionists, as on other government employees, to supplement their incomes with other activities. It would seem that such pressure, or the opportunity for outside employment, is greater for extensionists working at the district rather than the provincial level. Wealth It would be inappropriate in Indonesian society to ask respondents how much they earned, and many respondents would be unable to provide an accurate response anyway. Respondents' wealth was therefore measured instead, by a series of substitute measures: by asking whether they owned a series of items, including land, housing, vehicles, and electrical equipment. The inclusion of electrical equipment is justified by the expectation that most specialists live and have their offices in towns, which unlike many villages, are typically supplied with electricity. The questions included the word "private" (e.g., "private motorbike") to exclude the respondents' use of government-owned items. The most common item was a color television, owned by 211 of 279 respondents Table 7.9
Other work of respondents by institutional affiliation. Province n
%
District n
Overall%b
AIC
%
n
%
Other work
10
14.9
51
34.5
10
16.4
27.1
No other work
57
85.1
97
65.5
51
83.6
72.9
Total
67
100
148
100
61
100
100
Chi square = 22.60**a Data from question IS2.8. n = 276.b of total specialist population in each group.
Weighted for proportions
144 overall (the questionnaire did not mention ownership of a black-and-white set). This was followed by a refrigerator, house, motorbike, stereo, and land (Table 7.10). Relatively few owned a video recorder or a car. The pretest also asked about a telephone, but this item was deleted because so few respondents reported having one and because owning a phone depends as much on the availability of service as on wealth. Summing scores on these questions created an index of wealth. Twenty-nine respondents reported having none of the items, while 5 owned all. The median response was 2 items, typically a television and fridge. Comparing the index across institutions shows that province specialists were significantly wealthier (mean of 3.44 items) than those in district offices (2.42 items) (significant at p = 0.05 by Scheffé's test). Specialists at AICs (2.87 items) were intermediate in wealth. Work time Table 7.10
Wealth, as reflected by number of respondents who own selected items a.
Item
n
%
Color television
211
75.6
Refrigerator
112
40.1
House
99
35.5
Motorbike
95
34.1
Stereo
92
33.0
Land
84
30.1
Video recorder
46
16.5
Car
34
12.2
a
Data from question IS2.9. n = 279.
The median respondent reported working a total of 42 hours a week. This is probably only a slight overestimate, given that the official Indonesian work week is 38 hours long and many specialists also visit the field outside office time. However, this figure conceals a very wide range, from only 7 to as many as 119 hours a week (or 17 hours a day, 7 days a week!). Further analysis on work time was conducted after removing such outliers (Table 7.11).
145
Table 7.12 Percentage of work time respondents spent solving field problems and providing information from "above"a. Province
District
AIC
Solving field problems
57.9 b
56.6 b
47.1 a
Providing information from "above"
42.1 a
43.4 a
52.9 b
a
Data from question IS4. n = 269. Figures in a row followed by the same letter are not significantly different at p = 0.05 by Scheffé's test. There was no difference between the total amount of time worked by respondents at the three different groups of institutions. Overall, respondents spent about one-quarter of their time seeking information, a little more than a quarter providing it to clients, about the same amount in administrative work, and the remainder in other activities (such as meetings and sport). However, AIC staff spent significantly more time seeking information and less in administrative duties than did their counterparts in other institutions.
Table 7.11 affiliation.a
Time respondents spent per week on various activities, by institutional Province
Seeking info
District
AIC
hours
%
hours
%
hours
%
9.9 a
23.6
9.0 a
21.4
13.8 b
32.6
11.9
28.1
Providing info
10.2
24.3
11.7
27.8
Administration
12.7 b
30.2
13.3 b
31.6
8.9 a
21.0
Other
9.2
21.9
8.1
19.2
7.7
18.2
Total
42.0
100
42.1
100
42.3
99.9
a
Data from questions IS3.1-3.5. n = 254. Figures in a row followed by the same letter are not significantly different at p = 0.05 by Scheffé's test. It should be remembered that it is difficult to obtain accurate information about time allocation. Many factors including genuine errors as well as willful misreporting, can bias a respondent's reporting to the socially acceptable. The figures in Table 7.11 should therefore be taken as indicating a general tendency rather than the actual numbers of hours spent.
146
Field vs. central orientation Time allocation Question IS4 asked respondents what percentage of their time they spent "solving field problems" as opposed to "promoting messages from above." Respondents reported spending a mean of 55% of their time doing the former and 45% doing the latter (Table 7.12). (The question requested that the two percentages sum to 100%.) Most respondents balanced roughly equally the time spent in each activity: only 18.6% of respondents spent more than 70% of their time performing either activity. Perhaps this marked central tendency is due to the social desirability of both activities: as extensionists, respondents are supposed to have farmers' interests at heart, while as civil servants they are to promote administration policies. Indeed, several respondents indicated on the questionnaire that the two coincided: i.e., that by disseminating government messages they are also solving field problems, and vice-versa. How much time specialists spent in each activity depended on where they worked. Provincial and district respondents said they spent 57 to 58% of their time solving field problems, significantly more (by Scheffé's test, p = 0.05) than the 47% that AIC specialists devoted to such work. It thus seems that the local personnel were rather more field-oriented than those in the AICs. Independence from central decision making Questions IS12.1 and 12.2 measured the respondents' independence from central decision making. Maximum independence is desirable from one point of view, but is undesirable from another. Specialists with the autonomy to adjust technical recommendations can provide farmers and field agents with advice suited to local conditions. But if such flexibility is to be fruitful, it must be based on adequate local technology testing, technical backup such as soil analyses and reliable sources of seed, and skilled local personnel (including extension specialists and field agents). Such preconditions do not exist in many parts of Indonesia. The overwhelming majority of respondents (95.7%) agreed or somewhat agreed that recommendations from above must be tested before being extended. And nearly three-quarters disagreed or somewhat disagreed with the statement that central recommendations could not be changed before being extended. While they were worded to tap the same concept -- autonomy from central decision making -- there was only a relatively weak correlation between responses to the two statements (r = -.16, p = .05) (the negative correlation was expected as question IS12.2 was negatively worded). For this reason the two questions are treated separately in the analysis below. Respondents at the three types of institutions differed somewhat in opinion (Table 7.13). Provincial level respondents appeared to favor local autonomy in adapting central
147 recommendations slightly (though not necessarily significantly) more than did their counterparts at other institutions. They agreed most strongly that central recommendations must be tested and could be changed before being extended. AIC specialists were the most cautious of the three groups, with somewhat less support for local testing and alterations. These differences were merely relative, however; AIC specialists still came down strongly in favor of local autonomy in testing and adapting recommendations. Women respondents were more cautious than men, but this was probably because they were over-represented in the AICs. Senior specialists (as measured by years of service) were more prone to favor local autonomy than were their younger colleagues. And wealthier respondents, as measured by the number of items they owned (question IS2.9) favored local testing and autonomy. There was no relationship between other demographic variables and attitudes toward local adaptation and testing. Personal interviews with specialists reinforced this finding of strong support for local autonomy. Specialists saw Ministry recommendations as flexible guidelines that could be adjusted to local conditions and needs rather than as strict, unalterable rules. They said they were allowed to, and frequently did, change recommendations that were inappropriate to their locality. But their source of information on whether and how to change recommendations is somewhat of a mystery. The survey shows that specialists rarely conduct field trials -- the median specialist had conducted just one trial in the previous year (Table 7.14). Even allowing for exchange of information on trials among specialists, the range and variety of soils and climates in each district or province means that the number of such trials is totally inadequate. The data suggest that instead of research or formal trials, specialists (especially the more experienced ones) base their technology recommendations on informal, interpersonal information sources. These sources -- farmers and field agents, other specialists, superiors and colleagues, and the specialists' own experience -- were the top six ranked information sources overall (see Table 8.1). Several interviewees suggested they also modified recommendations for non-technical reasons, especially economics and marketing, for
Table 7.13
Respondents' opinion about central recommendationsa.
Statement
Province
Recommendations from "above" must be tested before being extendedb
6.71
Recommendations from "above" may not be changed before being extendedb
2.88
a
c
District
AIC
6.44
6.26
3.18
3.29
Data from questions IS12.1 and 12.2. Scale = 1 (disagree with statement) to 7 (agree). n = 270 - 278.b Not significant at p = 0.05 by Scheffé's test. c Provincedistrict and province-AIC contrasts were significantly different at p = 0.05.
148 instance reducing recommended fertilizer dosages if farmers were unable to purchase the full amount. The evidence points to specialists' and other agricultural service staff's developing a flexible series of recommendations, often by adapting central Ministry advice to be sensitive to local economic and biophysical conditions. These recommendations are based primarily on observations as to what works on farms and what farmers can afford, rather than on formal trials. The advantages of this system are obvious. Recommendations are locally relevant and can be changed as necessary without recourse to higher administration. It is possible to take full advantage of indigenous knowledge and farmers' initiative, to feed such knowledge into the extension system, and spread it to nearby areas. And local decision making relieves the burden on higher administration and the research and development system. But there are also disadvantages. Links with research and the formal testing of technologies are extremely limited. Because of the lack of formal testing, there is an inflated risk of promoting ineffective practices -- especially given the low caliber of some extension personnel. For instance, some technology packages are ineffective unless applied in full: pest management practices are an example of this. There is also a possibility of confusion among conflicting recommendations, and little opportunity for message reinforcement through extension publications or the mass media. And the lack of feedback to research may cause scientists to pursue topics of little local relevance. Extension activities Questions IS9 to 11 asked how many times the respondent had performed certain extension and information seeking activities. Responses to these questions are summarized in Table 7.14.
Table 7.14 Number of times respondents engaged in extension and information seeking activities, by institutional affiliation. 149 Sourceb
Province
District
AIC
Overall Mean Median
In previous 3 months Visits to farmers
8.5 a
16.0 b
6.9 a
12.1
9
Visits to Rural Extension Centers
7.2 a
12.0 b
4.6 a
9.3
7
Sought information to answer agents' or farmers' questions
4.9
6.1
4.6
5.5
3
Read scientific journal articles
6.4
7.3
9.0
7.5
4
Read AARD books
3.7
3.5
5.1
3.9
2
In previous year Conducted field technology tests
1.5 b
1.9 b
0.6 a
1.5
1
Traveled to seek extension information
3.9 ab
3.9 a
6.0 b
4.3
3
Attended extension training
0.5 a
0.5 a
0.9 b
0.6
0
In previous 3 years b Attended AARD training
0.4
0.6
0.6
0.5
0
Collaborated in AARD research
0.2
0.7
0.2
0.4
0
Attended research exhibition or seminar
2.4
1.6
2.3
1.9
1
Attended technical meeting with AARD researchers
3.1
2.3
3.1
2.7
2
Visited (or received visit by) AARD researcher
2.2
1.7
2.3
1.9
1
Wrote letter to research institute
0.6
0.5
0.5
0.6
0
Provided information to researchers
2.2
1.4
1.7
1.6
1
a
Data from questions IS9.1 to 11.7. n = 261 to 272. Figures in a row followed by the same letter are not significantly different at p = 0.05 by Scheffé's test. b Adjusted for specialists who had worked less than 3 years. Outliers (respondents performing the activity more than 12 times in the previous 3 years) were dropped.
150 Four possible sources of error were encountered in analyzing responses to these questions. One was the question wording, which might be interpreted in different ways. A typical question was "How many times in the last three months have you visited farmers?," with the space for the response labeled "________ times." It is possible that some respondents misinterpreted this as "How many individual farmers have you visited?" and responded accordingly. Such confusion -- if it occurred -- would have two effects: • It would reduce the validity of the responses, diminishing our confidence in the means shown in Table 7.14, thereby possibly increasing the chances of spurious significant differences between the means. • It would inflate the error term, reducing the likelihood of obtaining statistically significant differences between subgroups. This makes any significance found all the more credible, and opens the possibility that non-significant differences between groups of respondents reflect genuine differences. The second problem involved outliers. Some specialists reported performing an activity with unusual frequency. Sometimes their claims strained credibility. For instance, one claimed to have read 285 scientific articles in the previous 3 months, or more than 3.5 a day -- a voraciousness matched by few university faculty with far easier access to literature. Another purportedly had collaborated in 20 AARD research projects in three years, more than three times as many as any other respondent. Cases with extreme outliers were deleted from analysis of the relevant variable. A maximum of six cases (of an n of 267) were thus dropped for any one variable. The third possible source of error in responses to questions IS9 to IS11 was that all relied on respondents' memory. The time periods (three months, one year, and three years) were selected to be appropriate to the type of activity measured: someone ought to be able to remember whether he or she had attended training in the previous three years, for instance. Nevertheless, it is possible that some errors occurred because of poor recall. It is difficult to predict which direction, positive or negative, this would tend to skew results. The fourth possible source of error was that of misreporting. All the activities in Table 7.14 are desirable for extension specialists to perform. Respondents may therefore have over-reported the frequency they had performed certain activities (the "halo effect"). While Table 7.14 must be interpreted with these caveats in mind, it should be added that the figures are comparable with those reported by Hussein (1986), and broadly agree with responses of the specialists I interviewed in person. In particular, there seems to be little halo effect given the low levels of many means and the willingness of respondents to admit they had not engaged in certain activities at all during the period in question. I therefore conclude that the figures given in the table are fairly accurate.
151
Field visits The median respondent had made nine visits to farmers and seven to Rural Extension Centers in the previous three months (Table 7.14). Five respondents (three of whom were at AICs) had visited no farmers, while three had visited no RECs. At the other end of the scale, one district specialist claimed to have made 87 farm visits (more than one every work day), while another had visited RECs 77 times. As might be expected, district level personnel made significantly more of these field visits than did those at the provincial offices or AICs. District personnel had made a mean of 16 visits to farmers and 12 to RECs in the previous 3 months, nearly twice the level of respondents in the other two institution types. If each visit occurs on a different day, this means that district personnel spend 16 of 78 workdays in 3 months, or one day in every five, visiting farmers. If they visit RECs on different days from their farm visits, they spend an additional 12 of 78 workdays, or one day in every 6.5, at RECs. Given these unlikely assumptions (since a specialist may make more than one farm visit a day, and interviewees said they often visited farmers and RECs on the same day), district specialists thus spend a maximum of about one-third of their work time visiting the field. When we compare this to the 28% of their time they said they spent disseminating information (in response to Questions IS3.1 and 3.5, Table 7.11), we see this estimate is not too far off the mark. Provincial and AIC specialists devoted far less time to field visits. Provincial respondents had visited farmers on 8.5 occasions (once every nine days), and RECs 7.2 times (once every 11 days) in the previous 3 months (Table 7.14). Corresponding figures for AIC specialists were once every 11 days to farmers and every 17 days to RECs. Province and AIC specialists also disseminate information by methods other than personal contacts and to audiences other than field agents and farmers. Comparing the number of field visits with the total time spent disseminating information (Table 7.11) can give us an idea of the effort devoted to these other methods and audiences. Using reasoning similar to that for district specialists above, we can conclude that province SMSs spend about 15-20% of their time in the field, and that this accounts for about two-thirds of their total information dissemination activities. AIC specialists, by contrast, spend about 12-15% of their time in the field, accounting for less than half of the time they spend disseminating information. Of course these are estimates only, but they do give some idea of the relative importance of field activities for SMSs in the three types of institutions. The respondents' education, sex, place brought up, and wealth (questions IS2.3, 2.5, 2.6 and 2.9) had no effect on the number of farmer visits they made. None of the demographic variables had any effect on the number of REC visits made. Livestock specialists (question IS2.2), junior specialists (question IS2.4) respondents from farm families (question IS2.7) and those who had outside activities in farming (IS2.8) reported visiting significantly more farmers than did their colleagues. Livestock specialists had made a mean of 16.5 visits in the previous 3 months, compared to
152 11.8 by food crops specialists (F1,172 = 4.54, p = .03). Junior specialists (those with fewer years of on-the-job experience) made more farm visits than did their senior colleagues (beta = -.13, p = .04), presumably because the latter are more engaged in administrative work. Respondents from farm families had made a mean of 15 visits in the three months, compared to only 10 for those from non-farm backgrounds (F1, 263 = 8.82, p = .003). And respondents with farming activities outside their SMS job made a mean of 25 visits, compared with 11 visits for those with no outside work (F4, 262 = 6.32, p = .0001). These relationships held even if the respondent's institution was controlled for. Information seeking Seeking answers to clients' questions Respondents stated that they had sought information to answer questions from field agents or farmers a median of three times (mean of 5.5) in the previous three months, or once every 26 work days (Table 7.14). Some 34 respondents indicated that they had not sought such information in this period. The modal response was twice. District respondents, who are in much more frequent contact with field agents and clients, had sought such information a mean of six times in this period, but this was not significantly more than specialists in the other institutions. None of the other demographic characteristics (questions IS2.1 to 2.9) had any effect on the frequency respondents sought information to answer clients' questions. Reading research publications The median respondent reported reading four articles in scientific journals in the previous three months, or one every 20 work days (Table 7.14). The mean was 7.5 articles, and the modal response was three. The median respondent had read two books published by AARD in the same period, or one every 39 work days (mean of four books and mode of one). Twenty-eight specialists had read no articles, and 38 had read no AARD books. Agricultural Information Center specialists appeared to read such literature somewhat more often than their colleagues at other institutions, though Scheffé's test revealed no significant differences. A t-test comparing AIC specialists' book reading (mean of 5.1 in 3 months) with that of district staff (mean of 3.5) yielded a significant difference, however. It is indeed somewhat surprising that district specialists report reading so many research publications, since most of these publications are distributed to AICs and provincial offices but not to the district level. Perhaps respondents understood something different from what was intended in questions IS9.4 and 9.5. For instance, they may have included AARD's quarterly newsletter, Warta Litbang, as a "book" or a "scientific journal," or misidentified non-AARD publications as published by the research agency. The personal interviews suggested that many specialists do not correctly identify the publishers of research and extension publications.
153 None of the other demographic measures (questions IS2.1 to 2.9) had any effect on the frequency of reading research publications. Technology testing The median respondent had conducted just one field technology test in the previous year (Table 7.14). There was a wide variation in the number of tests conducted, with one respondent claiming to have performed 24, and another 20. But 84% of respondents had conducted two or fewer tests, and more than a quarter of the respondents had performed none. District and provincial specialists conducted significantly tests more than AIC respondents. Livestock specialists had made significantly fewer field technology tests (mean of 1.2) in the past year than had food crops specialists (mean of 2.1) (F1, 175 = 12.52). Men respondents had conducted a mean of 1.6 field tests in the previous year, significantly more than the 1.1 tests women respondents had performed (F1, 268 = 4.34, p = .04). This was because there were more women at AICs than in the other institution types: when institution was controlled for, the relationship between sex and number of tests disappeared. None of the other demographic characteristics affected the number of tests performed. Traveling to seek information Another measure of information seeking was the number of out-of-town trips made to obtain information for extension purposes. The median specialist had made three such trips in the previous year, though the most traveled respondent reported making as many as 70 (one every 4.5 work days) (Table 7.14). This high number may reflect a broad interpretation of the terms "out-of-town" and "trips to seek information for extension purposes," which were not otherwise defined in the questionnaire. The modal value, reported by 18% of respondents, was zero trips. AIC specialists made significantly more information-seeking trips than other specialists, a mean of six a year, compared to under four by provincial and district respondents. The non-significant difference by Scheffé's test between AIC and provincial specialists reflects the conservative nature of this test; a t-test comparing these two groups bears a significance level of p = 0.017. AIC specialists' more frequent travel to seek information contrasts with their less frequent trips to disseminate information (see above). The only demographic measure to affect the number of out-of-town trips was where the respondent had been brought up. SMSs with an urban background reported making a mean of 5.1 such trips in the previous year, while those from rural areas had made only 3.6 (F1, 265 = 6.43, p = .01). This held even when the institutional level was controlled for. No obvious reason for this difference is apparent.
154 Training Few respondents had recently attended training conducted by either extension or research institutions. Less than half (43%) had attended one or more extension courses in the last year, with most of these (31% of the total) attending just one course (Table 7.14). The mean number of extension courses was 0.6 a year, suggesting that on average a specialist attends such a course once every 1.7 years. AIC specialists were more likely to attend such courses than were their colleagues, on average nearly once a year. Training at AARD institutes was rather more infrequent. Less than one-third (30%) of respondents had attended such training in the previous three years, reflecting the lack of resources devoted to such activities by the research and extension agencies. Again extrapolating from the mean frequency (0.5 in three years), we can calculate that the average respondent attends training at an AARD institute once every 5.8 years. The three groups of respondents did not differ significantly in the number of AARD training courses they had attended. Only one of the demographic measures influenced the frequency of extension training: after controlling for institution, female SMSs attended significantly fewer courses than did males (F1, 245 = 4.27, p = .04). Two measures affected training at AARD institutes. Respondents with activities in teaching and farming aside from their SMS jobs were less likely to attend AARD training (mean of .17 courses in the last 3 years), while those with activities other than these or additional Bimas or Dinas duties had attended significantly more (mean of 1.38 courses, twice as often as any other group) (F4, 261 = 3.11, p = .02). Respondents with no outside activities were intermediate (.50 courses a year). The reasons for these relationships are obscure. The second demographic measure influencing the frequency of training at AARD institutes was wealth (question IS2.9). Respondents who reported owning more items attended training more frequently than those with fewer items (beta = .19, t = 2.87**). This relationship held even when the respondents' institution was controlled for. Again, there is no obvious reason for this relationship. Other links with research Six other measures were used to reflect linkages with research institutions. These were collaboration with AARD research projects; attendance at exhibits, seminars, etc., at research institutes; attendance at technical meetings where researchers were also present; visits to or by researchers; correspondence with researchers; and feedback to research about local field problems. Respondents were asked how many times they had engaged in each activity in the previous three years. Scores on each of these measures were low (Table 7.14). The median respondent had attended only one exhibition or seminar, had participated in one researcher visit, and had met researchers twice at technical meetings. He or she had collaborated on no research projects, written no letters, and had offered feedback only once to researchers. By these measures, then, direct contacts between researchers and extensionists are dismally small. These figures generally confirm Hussein's (1986) finding of weak research-extension linkages.
155 These statistics conceal a few specialists who engaged in closer relations with researchers. One had collaborated in 20 research projects, and several had attended 15 or more seminars or exhibitions. Five had met more than 15 times with researchers at technical meetings, while four had received or made 15 or more researcher visits. One unusual specialist claimed to have written 15 letters to researchers in three years, while a group of 12 had provided feedback to researchers on eight or more occasions. We might expect such individuals to be located close to research institutions or to work for AICs or provincial offices. But neither appeared to be the case: there did not appear to be any pattern, geographical or institutional, in these unusual individuals. Nor did individuals with exceptionally high contacts on one measure appear to have similarly frequent contacts on another. Since fostering close contacts with research is a desirable goal for the extension agencies, investigating the circumstances of these individuals would seem to be fruitful. Even when the outliers described above are dropped from the analysis, there are few relationships between research contacts and institutional level (Table 7.14). None of the six measures yielded a significant difference among institutions by Scheffé's test. However, two (collaboration in research and technical meetings with researchers) did yield significant t-tests among pairs of institutions. District specialists reported collaborating in significantly more research projects (0.7 in three years, compared to 0.2 for other specialists). And provincial specialists attended more technical meetings than did their district equivalents. The only other demographic variable to affect research contacts was the respondents' family background: those not from a farm family had attended a mean of 2.4 seminars and exhibits in the previous year, while those with a farm background had attended only 1.4 (F1, 264 = 7.31, p = .007) There is no apparent reason for this relationship, which held even when the respondents' institution was controlled for. Summary The data reveal several general differences among respondents at the three types of institutions. District personnel made more field visits than did province and AIC specialists. Province and district specialists conducted more field tests than did AIC respondents, who appeared to seek information more from the literature than in the field. Livestock specialists made more farm visits but conducted fewer field technology tests than did their food crops counterparts. The reasons for this are easily understood. Field extension agents have fewer skills in livestock than in food crops, so more of the burden of farm visits falls to SMSs. And conducting field tests is more difficult and expensive for livestock than for food crops, limiting the number that can be performed. Senior specialists made fewer farm visits than did their junior colleagues (though not significantly fewer REC visits). It is likely that the more experienced SMSs have more administrative duties, preventing them from making as many field visits. SMSs with a farm background or an outside activity in farming made more farm
156 visits than those without these characteristics. The questionnaire did not measure whether these visits were associated with the respondents' extension work. It is possible that they took place as part of the respondents' other income-generating activities. There was no systematic or obvious relationship among other demographic characteristics on one hand and SMSs' extension behavior on the other. The respondent's education, sex and length of service affected none of the extension behaviors measured. Respondents from urban areas were more likely to travel to seek information, while those without farm backgrounds attended more exhibits and seminars. And wealthier SMSs and those with non-extension activities attended more AARD training. Reasons for these relationships are difficult to discern; some may merely represent the occasional "significant" result one should expect to occur due to random variations in a data set. Problems faced The Publications questionnaire contained two sets of questions on general extension problems (question P3) and problems specific to the flow of information from research to extension (question P4). Question P3 (Table 7.15) was adapted from Sigman and Swanson's (1984) survey of 50 extension directors worldwide.
157
Table 7.15
General problems faced by extension. a
Problem
Mean scoreb
Sigman & Swanson rank c
Mobility -- Adequate transport for extension personnel to visit farmers
5.75
1
Obtaining information -- Adequacy of information flows from researchers to extensionists
5.70
7d
Technology -- Availability of technology suitable for extension to farmers
5.49
8
Feedback -- Adequacy of information flows from extensionists to researchers
5.33
7d
Technical skills -- Extension personnel's practical skills in absorbing new technologies
5.24
5
Teaching aids -- Availability of teaching materials, printed materials, demonstration kits, etc.
5.21
6
Extension skills -- Extension personnel's skills in teaching and communicating
5.20
2
Rewards -- Payment (moral and material) received for performing extension duties
4.80
-d
Facilities -- Teaching and communication facilities for extension personnel (projectors, classrooms, telephones, etc.)
4.76
3
Organization -- Non-extension workload
3.28
4
a
Data from Publications questionnaire, question P3. Question wording was based on Sigman and Swanson (1984). Differences are due to translation into Indonesian and re-translation into English. b
7 = very important problem; 5 = important problem; 3 = somewhat important problem; 1 = not a problem. c
Ranking from Sigman and Swanson's (1984:8) study of 50 extension directors worldwide. d
Sigman and Swanson's study combined "Obtaining information" and "Feedback" as "Linkage" (7th) and did not include "Rewards."
Mobility
158 Respondents recognized mobility as the most important general problem facing extension (Table 7.15). This echoes Sigman and Swanson's (1984:8) findings. Many SMSs complained during the interviews of the large area they had to cover and the lack of transport facilities, especially of motorbikes. Only one-third of respondents owned their own motorbike, and 12% had their own car (Table 7.10). This meant that it was difficult for them to visit Rural Extension Centers, farmers, or other locations, especially at short notice. Several interviewees said they borrowed motorbikes from friends or took public transport. But the scarcity of public transport in remote areas and its high cost was a severe constraint to their reaching these areas, with the result that some Rural Extension Centers rarely received a visit from an SMS. That this problem is not confined to SMSs was confirmed by the few field agents I was able to interview. Two solutions to this problem are obvious: provide SMSs with a vehicle, or increase the amount of funds allocated for transportation costs. A third alternative is to improve the telecommunication facilities available in district offices and villages. This would enable SMSs to perform some activities, such as scheduling meetings, without the need to travel. Unfortunately, all three solutions would be rather expensive. Information After mobility, obtaining information from research was a close second as a problem for extension (Table 7.15). It is possible that this result is biased because respondents realized the nature of the questionnaire, which focused on this topic. To forestall this eventuality, the question on extension problems (P3) was placed before any other questions on research information (P4-9). And the difficulty of obtaining researchbased information was confirmed in the oral interviews. Sigman and Swanson's (1984) study combined my categories of "obtaining information" and "feedback" into a single category, "linkage." They found linkage ranked seventh of nine problems. The results of this study clearly disagree with theirs: Indonesian SMSs see both obtaining information (2nd overall) and feedback to research (4th) as major problems. I explore ways of improving information flows from research to extension in Chapter 11. Technology Respondents rated the availability of appropriate technology third among the ten problems listed. This contrasts with Sigman and Swanson's finding of technology as the least important of the eight problems they listed. It can be taken to mean three things: • Indonesian SMSs are not aware of the range of suitable research-based technologies that are available.
159 •
Such technologies do not exist.
• Such technologies exist, but sufficient adaptive research has not been carried out to test their applicability to local situations. The first is a communication problem: the technologies already exist, but have not been communicated adequately to the extension audience. This could be overcome relatively easily through more effective use of the available channels. The second implies a need for more (and a different type of) research, and a revision in the research planning process. For research to address field needs, input from the field is necessary in the form of field surveys and farmer and extension inputs. At present such inputs are rather limited: Hussein (1986:409) found that extension personnel and farmers were the two least important sources of research ideas for AARD scientists. The third possibility reflects the gap between the mandates of AARD and the local agricultural services (as coordinated by the directorates-general). The former has the mandate for technology generation; the latter have the responsibility for local verification trials, which are performed by the provincial and district SMSs (Abbas et al. 1989:55-57). But insufficient funding means local trials are rare: the median respondent had performed only one trial in the past year, and only 16% reported performing three or more (Table 7.14). And where local trials are conducted on a wide scale, researchers are sometimes critical of the validity of results. There is thus a need for greater funding for local trials, for more AARD involvement in their conduct, and probably for training of SMSs in appropriate research techniques. It is not possible from this study to discern which, if any, of the three possibilities outlined above is paramount. But it is likely that all three play a role, and that solving the problem of technology availability must involve them all. Feedback Poor feedback from extension to research ranked fourth among the problems facing extension in general (Table 7.15). That feedback is indeed poor was documented by Hussein (1986:416-417) and confirmed by this study: both direct contacts and correspondence were infrequent (Table 7.14): the median respondent had attended two meetings with researchers in the past three years, visited a researcher once, written no letters, and claimed to have provided feedback to researchers just once. It is easy to blame the SMSs for their inactivity in providing feedback to researchers. But research managers should seek ways of stimulating such feedback, such as inviting comments from extensionists and farmers on new technologies, involving them in research planning and technology testing, and including local suggestions in research protocols. Farming systems research projects are ideal vehicles for such interactions, but too often they are seen merely as AARD research that is performed in the field, rather than a
160 collaborative effort by all parties. And FSR projects are too scattered to allow a significant proportion of SMSs and field agents to participate in or even visit them. There is a need for a broader mechanism to generate feedback to research. Recent moves to discover technology needs in each province are encouraging in this regard (Tjitropranoto 1990). Technical skills Extension personnel's lack of technical skills ranked fifth among the ten problems listed (Table 7.15). (Sigman and Swanson's [1984:8] study also ranked this fifth.) This study did not differentiate between the SMSs' own skills and those of their field colleagues. Nevertheless, respondents felt their own skills needed upgrading, as reflected in the top ranking they gave "infrequent training" as a problem in question P4 (see below). Extension personnel can acquire technical skills in several different ways. These include training courses, media such as publications, collaboration with farmers, colleagues and researchers, and direct experience. They can also obtain such skills during their university careers before joining the extension service. Here I will comment only on training courses. SMSs typically receive two periods of training soon after they are hired: a onemonth course covering basic extension methodologies, and second course of the same length on approaches to identifying and solving farmers' problems. Both these courses are taught at a training center at Ciawi, West Java. The Ciawi center also teaches courses on specific technical skills for SMSs and other district agricultural officials; these typically last two weeks. SMSs, field agents, and other agricultural officials also attend courses at AAET's 31 other training centers (Figure 3.2). The centers have their own training staff, but AARD and university scientists may be invited to present sessions on specific topics. Training courses for SMSs are also held at AARD institutes and universities such as Bogor Agricultural University. Such courses are held only sporadically. Residential training has an added attraction for respondents because of Indonesian civil service rules about travel allowances. Personnel who travel on official business receive a per diem that varies according to their seniority, the distance traveled, and the location. The per diems normally exceed actual expenses incurred. They can often amount to more than the officials' (small) regular monthly salary. Residential training is thus particularly attractive because it provides the trainee with relatively large amounts of extra income. This consideration would tend to inflate the importance of training as a problem in the survey responses. Nevertheless, such opportunities are rare: the median respondent had attended no AAET training in the previous year, and none at AARD institutes in the previous three years (Table 7.14). Calculations using the mean number of courses attended indicate that a specialist can expect to participate in AAET training once every 1.7 years, and in AARD training once every 5.8 years. I must therefore conclude that lack of training for SMSs is a real concern, particularly at AARD institutions. Extension and research administrators should seek ways
161 of increasing the frequency of training, ensuring its practicability, and evaluating its effectiveness. Such training courses have the potential to benefit not only the SMS trainees, but also the host institution -- since trainees can provide valuable feedback to researchers on field problems and can build interpersonal linkages with researchers and each other that can prove invaluable in the future. Teaching aids The availability of teaching aids -- teaching materials, printed materials, demonstration kits, etc. -- ranked sixth on overall importance as a problem. Sigman and Swanson's study also placed them sixth. Nevertheless, their mean score of 5.21 (Table 7.15) meant that respondents still thought them to be an important constraint to extension. There does appear to be a lack of teaching aids available, both in terms of variety and number of copies. From 1984 to 1990, all the AICs produced a total of 301 posters, 774 audiocassettes, 281 slide sets, and 55 videos (Table 3.3). But divided among 27 provinces and spread over six years, this means that an individual extensionist might receive an average of 1.9 different posters, 4.8 cassettes, 1.7 slide sets, and 0.3 videos per year. In addition, the AICs occasionally produce flip charts and other teaching aids, but their numbers are insignificant. These must be regarded as maximum figures, since at least some of these materials were produced in quantities too small for distribution to all extension agents. Of course, publications such as Liptans, brochures and booklets might also be considered as teaching materials, but they too are produced in print runs too small for effective use with farmer groups. There does not appear to be any systematic attempt by AICs or the agricultural training institutes to train local extension personnel in preparing teaching materials, nor sufficient funding to enable the local staff to do so. Despite this, one AIC administrator said that local extension personnel were expected to overcome the lack of materials by reduplicating AIC publications and adapting them if necessary to local conditions. At least some district Dinas offices do indeed do this. One extension official in North Sumatra gave me a copy of the mimeographed materials used to train his district's field agents in integrated pest management. Another district official in West Java showed me a local magazine produced by the Dinas office and distributed to extension agents and village officials throughout the district. However, quality was low, illustrations were poor or absent, and the cost of producing such materials at district offices is undoubtedly larger than producing them at the provincial AIC would be. All the printed materials I saw were aimed at and distributed primarily to extension personnel. Very few seem to reach farmers. I am unable to comment on the effectiveness of teaching materials beyond some anecdotal evidence: •
Only one of the district or provincial Dinas offices I visited maintained an effective
162 library. In several cases, what purported to be a "library" was a locked cupboard containing unsorted AIC materials. • In some cases the materials appeared to have been used; in others, they were still in their mailing wrappers. • In one of the two Rural Extension Centers I visited, the AIC materials were openly displayed on a table and were readily accessible to staff and visitors. In the other, they were locked in a metal cupboard. • One cupboard in a district Dinas office contained 72 copies of an audiocassette jointly prepared in 1983 by the Ministries of Agriculture and Information. Local officials claimed these cassettes had never been used, and never been broadcast on the local radio station because the officials did not have the right to do so. This evidence is anecdotal and based on a small number of visits to non-randomly chosen sites. Nevertheless, it illustrates several points: • AIC publications appear to be disseminated to local Dinas offices and Rural Extension Centers. • At least some of these offices make good use of them. However, there is a need to evaluate the usage and effectiveness of AIC publications on a more systematic basis. • Non-print media such as slide sets and audio-cassettes are expensive to prepare and reproduce, yet little is known about their usage and effectiveness. There is an urgent need to evaluate these media. • In many local offices, publications are not accessible to visitors or even to office staff. These offices would benefit from having a formally established library to display materials and allow them to be checked out. Extension skills While extension skills ranked only seventh of ten problems, its mean score of 5.20 still indicated that respondents felt it to be an "important" problem (Table 7.15). Sigman and Swanson's survey ranked it much higher: second of eight problems. The question did not differentiate between respondents' view of their own skills and those of their extension colleagues. The low ranking of extension skills is surprising given the large number of respondents (50.3%) who indicated that they needed more information on extension (Table 8.10). I have briefly described the training SMSs receive in the section on technical skills above. I did not collect information on training for other extension personnel.
163
Rewards With a mean score of 4.80, inadequate rewards ranked eighth in terms of importance as a problem for extension. This means that respondents on average saw poor rewards as slightly less than "important." The rewards item (question P3.10) referred explicitly to both "moral and material" forms of payment for performing extension duties. Indonesian civil service salaries are low compared to amounts that can be obtained in the private sector. Nevertheless, once a person has obtained civil servant status (after a probationary period that may last several years), his or her job is secure. A major benefit is government health insurance. Many civil servants supplement their incomes through travel per diems, additional emoluments for serving on committees and task forces, and outside work. About one-quarter of all SMSs report having income-generating activities apart from their work as SMSs (Table 7.9). Low staff morale is often blamed for poor extension effectiveness (Pickering 1983:6). The low ranking of rewards compared to other problems does not support this view for Indonesia. This should not be taken as conclusive evidence of good extension morale, however, as the study did not focus on this question. Facilities The adequacy of teaching and communication facilities ranked ninth of the ten problems (Table 7.15). The mean score of 4.76 places this as slightly less than an "important" problem. Sigman and Swanson (1984:8), by contrast, found equipment ranked third of eight problems worldwide. Indonesian extension agents have few facilities to work with other than a building and furniture. Dinas offices and Rural Extension Centers typically have a classroom or meeting room but little equipment such as slide and overhead projectors. Most district Dinas offices have a telephone, but junior staff may not feel they have free access to it. Phones are rare in villages, meaning that such mundane messages as meeting schedules must be delivered to contact farmers in person. Nevertheless, SMSs felt that the lack of these facilities was less important than most of the other problems listed. Organization Ranking last of the ten problems, the amount of non-extension work extension personnel had to do was not considered a major constraint. The mean score of 3.28 places this item close to "somewhat important" as a problem. Sigman and Swanson (1984:8), by contrast, found this to be the fourth most important problem among extension programs worldwide. Reducing the amount of non-extension work that extension agents are required to perform is one of the major aims of the training-and-visit system (Benor and Baxter
164
Table 7.16
Problems respondents face in obtaining information.a
Problem
Mean scoreb
Training to raise SMSs' knowledge and skills is infrequent
5.84
Funding for performing extension duties is insufficient
5.49
Meetings between researchers and extensionists are infrequent
5.47
Publications are not received regularly
5.03
Publications are not relevant to field problems
4.68
SMSs lack skills in translating scientific information into extension language
3.89
Technology recommendations are difficult to derive from publications
3.85
SMSs have too little time to seek information
3.49
a
b
Data from Publications questionnaire, question P4. 7 = very important problem; 5 = important problem; 3 = somewhat important problem; 1 = not a problem.
1984:9). Nevertheless, SMSs report spending between 40% and 50% of their time in administrative work or "other activities" (Table 7.11). However, they do not seem to regard this percentage as excessive. Problems in obtaining information Question P4 asked respondents to rate the seriousness of several problems associated with obtaining information. Responses are shown in Table 7.16. The most serious was seen to be the lack of training opportunities for SMSs. This agrees with Sigman and Swanson's (1984) finding that extension training was a serious problem facing extension organizations worldwide. This was rather more important than the next two problems, inadequate funding for extension and the lack of meetings with researchers. The irregularity of publications (5th) irked respondents more than their possible lack of relevance to field problems (6th), though the latter was still seen as an important issue. Respondents did not seem to feel that they lacked skills in translating research findings into a usable form (7th), or that this was overly difficult to do from the available publications (8th). Most did not think time was a major constraint in obtaining information (9th), echoing the bottom-ranked placing of a similar itemreveal in question (see below).problems. Some are relatively easy and These rankings severalP3high-priority inexpensive to address; others are more complex and costly. For instance, improving extensionists' mobility by providing them all with motorbikes would be prohibitively expensive. Likewise, substantially increasing funding available for extension activities and technology testing would also be costly, given the large numbers of Indonesia's extension personnel. This does not mean that these actions should not be undertaken; on the
165 contrary, they are important if the country's agricultural development is to continue. Other actions are cheaper and easier to contemplate by middle-level decision makers. For instance, two-way information flows could be improved markedly by more frequent meetings between research and extension staff in seminars, workshops, field days, training courses, etc. This could be done on a provincial or regional base, with a significant proportion of a province's SMSs attending meetings with local researchers on an annual basis, if not more frequently. AARD is pioneering just such activities through the ResearchExtension Linkage Project. Other possibilities are still cheaper. One is to send all SMSs research publications regularly. Printing and distributing 2000 copies of a publication is far cheaper (albeit less effective) than bringing the same number of people together for meetings. Many AARD publications are suited for an SMS audience, but are not sent to them because of inadequate print runs or poorly maintained mailing lists. Both can be improved with a limited amount of funding and organization. The cheapest, and ironically possibly the most effective, method of reaching extension agents is through the mass media. The newspaper Sinar Tani stands out here: 79% of respondents read it at least once a week, yet it contains very little research information. Rural newspapers subsidized through the government's "Newspapers in the Village" (Koran Masuk Desa) program are another possible vehicle. It would be easy for communication personnel in AARD institutes to write news releases and distribute them to these newspapers, ensuring a wide and rapid coverage of research findings. The type of information disseminated through the agricultural and rural press would necessarily be constrained by the nature of the media. However, a large proportion of the subscribers to Sinar Tani are extension personnel (Chapter 3). And many of the extension personnel I questioned said they would welcome seeing research briefs in this newspaper.
169
CHAPTER 8 SPECIALISTS' INFORMATION SOURCES AND NEEDS Introduction This chapter describes the technical information sources of subject-matter specialists and their information needs. I first discuss responses to question IS5, which listed 24 possible sources of information for extension purposes, ranging from the mass media to farmers, and from research publications to training courses. I discuss the ranking of these sources and differences among the respondents at the three institutions surveyed (province and district offices, and Agricultural Information Centers). I then compare the total amount of information obtained by respondents at the three institutions. Question IS6 measured source desirability: it asked respondents to name the top five sources they would use in ideal conditions. I compare responses across institutions and then discuss the relationship between source desirability and the actual amounts of information respondents reported obtaining from each source. The Publications questionnaire contained several questions that paralleled those just described. I discuss these next. Questions P5 and P6 asked respondents how often they read several agricultural publications and how useful they found them. Questions P7 and P8 focussed on the major sources for six food crops and six livestock technologies released by AARD in the previous decade. I use responses to the latter questions to discover how fast and by what media AARD technologies diffuse through the SMS population. Finally, I turn to SMSs' information needs. Question P9 listed 62 topics and asked respondents to check those on which they felt they needed most information. Information sources Respondents obtained most information from field extension workers, followed by other extension specialists, farmers, and their superiors. This is shown in the last column of Table 8.1, which summarizes data from question IS5. The top six sources were all persons whom the specialists meet frequently or the respondents' own experience. Testing and research by respondents themselves ranked somewhat lower (11th), possibly because specialists seldom conduct such tests.
Table 8.1 Amount of information respondents received from 24 sources, by institutional affiliation.a b Source Province District AIC Weightedm eanc 170 1 Field extension workers 4.95 a 5.58 b 4.65 a 5.3 2 Extension specialists 5.18 4.86 5.05 5.0 Farmers 4.58 a 5.18 b 4.62 a 5.0 4 Superiors 4.85 4.89 4.95 4.9 5 Own experience/observation 4.86 4.86 4.48 4.8 Other colleagues 4.56 a 4.87 ab 5.10 b 4.8 7 Ag newspapers/magazines 4.52 4.43 4.47 4.5 8 AIC publications 4.00 a 4.23 a 5.21 b 4.3 9 Univ courses/notes 4.20 ab 4.31 b 3.49 a 4.2 Other mass media 4.03 4.20 4.18 4.2 11 Own testing and research 3.78 b 3.91 b 3.11 a 3.8 12 Dir Gen recommendations 3.91 ab 3.53 a 4.47 b 3.7 Training 4.06 b 3.42 a 4.03 ab 3.7 14 Warta/Jurnal Litbang 3.61 a 3.15 a 5.17 b 3.5 15 AARD technical guides 3.39 a 3.01 a 4.44 b 3.3 17 Seminars etc. at res. inst 3.45 ab 2.90 a 3.63 b 3.1 AARD books 3.31 a 2.79 a 4.46 b 3.1 Scientific journals 3.25 a 2.78 a 4.30 b 3.1 19 Field research projects 2.92 3.02 2.85 3.0 20 University textbooks 2.73 2.89 3.18 2.9 Researchers 3.09 ab 2.69 a 3.37 b 2.9 22 Private technical guides 2.12 2.40 2.70 2.4 23 Agribusinesses 2.31 2.00 2.13 2.1 24 Correspondence 2.13 ab 1.94 a 2.40 b 2.0 Total information obtainedd 3.69 ab 3.61 a 4.04 b Data from questions IS5.1-5.24. Scale: 1 = very little, 2 = little, 3 = somewhat little, 4 = medium, 5 = somewhat much, 6 = much, 7 = very much. n for each item = 262 to 278. Figures in a row followed by the same letter are not significantly different at p = 0.05 by Scheffé's test. Underlined sources are discussed in detail elsewhere in this study. Weighted for proportions of total specialist population in each group. Sources are listed in rank order of the weighted mean response. For specialists responding to all 24 items (n = 227).
171 Easily understood and accessible non-personal sources (agricultural press, extension publications, general mass media, and ministry recommendations) ranked between 7th and 12th. Training courses and notes ranked 9th (university courses) and 13th (training at extension institutions), the extension training being less important presumably because of its infrequency. The respondents said they obtained relatively little information from research sources, both printed and personal. The most-used of such sources, the serials Warta Litbang and Jurnal Litbang, ranked 14th, followed by AARD technical guides, books, scientific journals, and university textbooks. Personal contacts with researchers also fared badly, with seminars and other formal contacts (workshops, exhibitions, and training at research institutes) ranking 16th, field research projects 19th, contacts with individual researchers 21st, and correspondence with researchers a dismal 24th. Somewhat surprisingly, commercially published technical guides were rated low (22nd), even though these are relatively widely available locally through bookstores. Perhaps specialists are unable or reluctant to purchase copies. Agribusiness employees were also unimportant as a source. Comparison among institutions The overall ranking described above disguises important differences among specialists in the various institutions surveyed. Significant differences among the institutions are denoted by letters following the first three columns of figures in Table 8.1. In general, AIC specialists got more information than did their provincial and district colleagues from nearly all types of research and extension publications and from most forms of contact with scientists. But they got less information from their own research than did the others. Their two top sources were both publications: AIC publications and the AARD series Warta Litbang and Jurnal Litbang (Table 8.2).
172 District specialists' top two sources were field extension agents (by quite a wide Table 8.2
Top five information sources of respondents at provincial and district offices and Agricultural Information Centers.a
Sourceb
Province
District
AIC
score rank
score rank
score rank
Field extension workers
4.95
2
5.58
1
4.65
6
Extension specialists
5.18
1
4.86
5
5.05
4
Farmers
4.58
5
5.18
2
4.62
7
Superiors
4.85
4
4.89
3
4.95
5
Own experience & observations
4.86
3
4.86
6
4.48
8
Other colleagues
4.56
6
4.87
4
5.10
3
AIC publications
4.00
11
4.23
9
5.21
1
Warta/Jurnal Litbang
3.61
14
3.15
14
5.17
2
b
Data from questions IS5.1-5.24. n for each item = 275 to 278. Only those sources ranking in the top 5 of 24 sources for one of the groups are shown. Sources are listed in rank order of the weighted mean response. margin) and farmers. They receive significantly more information from these sources than do other specialists. They also rely significantly more on their university course work. Provincial specialists' most important sources were extension specialists and field agents. They also received more information through extension training courses. The three groups did not differ significantly in the amount of information they receive from the mass media, their superiors and peers, field research projects, textbooks and commercial technical guides, their own experience, or agribusiness employees. Differences seemed to be heavily influenced by specialists' institutional setting. District specialists are more field-oriented than those at the central institutions, so have more contact with and obtain more information from field agents and farmers. They are also more isolated, often being the only specialist in their subject area in their district, and are rarely able to travel to the provincial capital or local research institution. They must therefore rely on whatever materials are available, such as their university notes. That field agents and farmers are important sources is encouraging given the need for extension institutions to respond to local conditions. It may indicate that specialists are ready to take advantage of and promote locally developed technologies, or indigenous knowledge. Unfortunately the surveys did not include further questions about the role of extension clients as information sources. Agricultural Information Center personnel have less contact with the field but greater
173 access to the extension and scientific literature. All AICs have a library, and many are fairly well stocked, maintained and used. Indeed, the AIC libraries attract users from outside, especially students at local universities seeking information for term papers and theses. Provincial personnel are often intermediate between the district and AIC specialists. They visit the field less than do district staff, but have less easy access to agricultural publications since many of their institutions have no library. On the other hand, they work with and meet other extension specialists frequently, and probably travel more widely, if less frequently, than do district specialists. Notwithstanding the differences noted above, there was remarkable agreement among the specialists as to which were major sources and which were not. Interpersonal, easily accessed sources ranked high, while research sources and those associated with the private sector scored low. The mass media, extension publications and training were intermediate. The challenge for AARD and the extension agencies is to raise the flow of research-based information--through whatever channels--to supply the needs of extension personnel at all types of institution. Total information flows We can use the sum of scores on all 24 items in question IS5 (shown as "Total information obtained" in Table 8.1) to indicate the total amount of information a respondent obtains. On this crude measure, Agricultural Information Center specialists received significantly more information than their colleagues at district offices, and somewhat (but not significantly) more than those at the provincial level. I used one-way analysis of variance and simple regressions to test whether the total amount of information obtained was affected by any other criteria. The following characteristics (based on the respondents' address or the Information Sources questionnaire) were tested: Inner vs. Outer Island location, development stage of province, respondent's specialization (question IS2.2), education (2.3), number of years as an extension specialist (2.4), sex (2.5), rural background (2.6 and 2.7), engagement in outside work (2.8), wealth (2.9), hours spent seeking information (3.1), and time spent solving field problems (4.1). Of these characteristics, only wealth was significantly related to the total amount of information obtained (F = 7.18**). The reasons for this are not clear: province specialists were significantly wealthier than their district counterparts (Chapter 7), but did not obtain significantly more information (Table 8.1). AIC specialists obtained the most information, but were not significantly wealthier than their district colleagues. Hypothesis 1 proposes that food crops specialists will obtain a larger total amount of information than their livestock counterparts. This was not the case, as shown by the lack of a significant effect of specialization (question IS2.2) on either total information flow (the sum of responses to questions IS5.1 to 5.24) or flow from AARD publications (IS5.5 to 5.7 and 5.9). This may be because of the lack of a direct comparison in the questionnaires
174 between food crops and livestock. Because each specialist responded to questions about his or her own topic, he or she had no outside referent with which to make comparisons. Thus a livestock specialist might have answered "Very much" to an item in question IS5 ("How much information do you obtain from [source]?"), whereas a food crops specialist, faced with a larger total amount of information, might have compared the source in question with the larger total and answered only "Somewhat much." Measures of the amount of information flow may thus be consistent within a respondent, but less so among respondents. This would tend to reduce any difference between food crops and livestock respondents. A second reason for this finding may be that the amount of information available on food crops and livestock does not in fact differ a great deal. AARD's six food crops institutes have considerably more research staff than do the two livestock institutes (Figure 4.1) and the output of research publications reflects this, but the numbers of extension publications are not greatly different: 1164 food crops items in the period 1984-90, compared to 830 for livestock (Table 3.3). The lack of difference appears to be a result of a deliberate attempt by provincial authorities to maintain a balance among the various commodity groupings in terms of media materials the AICs produce.
Desirability of sources Respondents were asked to name the five sources or channels from the list of 24 that they would most like to use under ideal conditions. One point was given to a source if it appeared as one of a respondent's five choices. Giving choices different weights (5 points for a respondent's first choice, 4 for the second, and so on) yielded essentially the same ranking as the simpler method.
175 Table 8.3 Number of respondents naming sources among their five preferred information sources, by respondents' institutional affiliationa . Source Province District AIC Total n % n % n % n %b Warta/Jurnal Litbang 37 55.2 81 55.6 33 54.1 151 55.4 Dir Gen recommendations 43 64.2 64 44.0 31 50.8 138 50.4 AARD books 33 49.3 71 48.8 33 54.1 137 49.5 AARD technical guides 31 46.3 68 46.7 •c 43 70.5 142 49.1 AIC publications 25 37.3 55 37.8 18 29.5 98 36.8 Training 17 25.4 62 42.6 * d 15 24.6 94 35.8 Scientific journals 23 34.3 40 27.5 13 21.3 76 28.7 Seminars at res. insts. 16 23.9 39 26.8 21 34.4 76 26.8 Ag newspapers/magazines 18 26.9 36 24.7 16 26.2 70 25.5 Own testing & research 13 19.4 30 20.6 10 16.4 53 19.8 Field research projects 8 11.9 28 19.2 12 19.7 48 17.2 Farmers 11 16.4 23 15.8 14 23.0 48 16.7 Own experience/observ. 8 11.9 19 13.0 10 16.4 37 13.1 Researchers 11 16.4 14 9.6 9 14.8 34 12.1 Extension specialists 9 13.4 17 11.7 5 8.2 31 11.8 University textbooks 8 11.9 17 11.7 2 3.3 27 10.9 Field extension workers 5 7.5 14 9.6 2 3.3 21 8.3 Other mass media 5 7.5 9 6.2 6 9.8 20 6.9 Agribusinesses 3 4.5 10 6.9 1 1.6 14 5.6 Private tech. guides 3 4.5 6 4.1 7 11.5 16 5.0 Correspondence 3 4.5 7 4.8 1 1.6 11 4.4 Superiors 1 1.5 9 6.2 •e 0 0 10 4.2 Other colleagues 2 3.0 6 4.1 3 4.9 11 3.9 Univ. courses/notes 2 3.0 3 2.1 0 0 5 2.1 Total respondents f 67 145.6 61 273.6 Data from question IS6. Sources are ordered according to their weighted means of desirability (final column). "Other sources" (named by 6 respondents) are ignored. Weighted for proportions of total specialist population in each institution. Chi square = 5.23 (significant at p = 0.10). Chi squares calculated using the proportion of respondents in each group of the total (e.g., 67/273.6 for provincial specialists) multiplied by the total respondents naming that source (e.g., 151 for Warta/Jurnal Litbang) as the expected value. Chi square = 6.13 (significant at p = 0.05). Chi square = 5.63 (significant at p = 0.10). Number of specialists responding to questions. Calculated as the sum of the preceding column divided by 5 (because respondents named 5 sources). Some figures are non-integers because some respondents named less than five desired sources.
176 The last column of Table 8.3 shows the percentage of respondents (weighted according to their relative numbers in the general population of SMSs) who listed each source as one of the five they desired most. The most desired source was AARD's quarterly newsletter and semi- technical journal, Warta Litbang and Jurnal Litbang. These were named as the most desired source by 56 specialists (20% of respondents), and among the top five by 151 (55%). The AARD serials were followed by other technical publications: Directorate General recommendations, AARD books, and AARD technical guides were all among the top five most desired sources of about half the respondents. Two other types of publications (AIC publications and scientific journals) were named by fewer respondents but also ranked high (5th and 7th). All these sources share several features. They are obviously all publications, so can be stored and accessed at will and without disturbing another person. They also are likely to be invested with high credibility -- they are produced by government sources and contain official recommendations, or represent careful scientific research. This suggests that under ideal conditions (i.e., if publications are available at each extension office), credibility and ease of reference may be key in determining which sources are used. The remaining non-mass media publication types were less popular: university textbooks were mentioned by about 10% of respondents, and commercially published guides by only 5%. This is possibly because of the cost and non-official nature of these two publications. Specialists also favored training at extension institutions, making this the highest ranked non-publication source (6th overall). But this seemed to apply to training in the future rather than courses they had attended at university, which ranked last in desirability. They preferred the agricultural press (9th overall) over other mass media (18th). Research, and direct contact with researchers, scored in the middle range of desirability as sources. Seminars and other meetings was 8th in terms of desirability, field research projects 11th, and personal contacts with researchers 14th. Specialists appeared reluctant to rely on correspondence, though, which ranked 21st. The wish to rely on research included the specialists' own testing and research (ranking 10th), and possibly the specialists' own observations and experience (13th). Personal contacts scored low on desirability. The most desired personal contacts were farmers, ranking 12th in terms of desirability. These were listed by only 17% of respondents among their five most desired sources. Other extension specialists ranked 15th overall, followed by field agents at 17th. Agribusiness employees and projects ranked 19th. Superiors and other colleagues weighed in near the bottom.
177
Comparison among institutions Specialists at the three types of institutions had remarkably similar lists of desired sources (six leftmost columns of figures in Table 8.3). This can be seen from the similar percentages of respondents in each group that named a particular source as desirable. The three groups had nine of the ten most desired sources in common, with minor differences in the rankings among sources. Only three sources deviated significantly (or nearly so) from the common ordering. These were training, AARD technical guides, and superiors. Two-fifths of district specialists named training as a desired source, compared to only about one-quarter of provincial and AIC respondents. District specialists indeed received fewer training opportunities than AIC personnel, though not fewer than provincial specialists (see Table 7.14). Agricultural Information Center specialists' most desired source by a wide margin was AARD technical guides. This source was named as one of the five most desired sources by over 70% of AIC respondents, but by less than half of the specialists in the other two groups. Superiors were named as desirable sources by rather more district specialists and fewer AIC staff than might be expected, but the very small numbers involved induce us to disregard this result. Various other sources appeared to attract different groups. For instance, nearly two-thirds of provincial specialists named Directorate-General recommendations among their desired sources, while half or less of the other specialists did so. But chi-square tests revealed that this and other differences did not attain statistical significance. Relation between information flow and source desirability One might expect specialists to tend to use those sources that they see as most desirable -- those that they would use in ideal conditions. That this was not the case can be seen from Table 8.4, which lists the sources according to the difference in their ranks on information flow and desirability (final columns of Table 8.1 and Table 8.3). Information flow (as represented by its weighted mean) is correlated with neither desirability (r = -0.11ns) nor its natural logarithm (r = -0.07ns). Of the 11 most desired sources, only two ranked among the top ten actual sources.
178
Table 8.4
Comparison of desirability of 24 sources and actual information flows to respondents from those sources a . Source Desirability Info flow Differ-enced Weighted Rank Weighted Rank meanb meanc AARD books 49.5 3 3.1 17 +14 Warta/Jurnal Litbang 55.4 1 3.5 14 +13 AARD technical guides 49.1 4 3.3 15 +11 Scientific journals 28.7 7 3.1 18 +11 Dir Gen recommendations 50.4 2 3.7 12 +10 Seminars at res. insts. 26.8 8 3.1 16 +8 Field research projects 17.2 11 3.0 19 +8 Training 35.8 6 3.7 13 +7 Researchers 12.1 14 2.9 21 +7 University textbooks 10.9 16 2.9 20 +4 Agribusinesses 5.6 19 2.1 23 +4 AIC publications 36.8 5 4.3 8 +3 Correspondence 4.4 21 2.0 24 +3 Private tech guides 5.0 20 2.4 22 +2 Own testing & research 19.8 10 3.8 11 +1 Ag newspapers/magazines 25.5 9 4.5 7 -2 Own experience & observ 13.1 13 4.8 5 -8 Other mass media 6.9 18 4.2 10 -8 Farmers 16.7 12 5.0 3 -9 Extension specialists 11.8 15 5.0 2 -13 Univ courses/notes 2.1 24 4.2 9 -15 Field extension workers 8.3 17 5.3 1 -16 Other colleagues 3.9 23 4.8 6 -17 Superiors 4.2 22 4.9 4 -18 Data from questions IS5 and 6. n = 276. "Other sources" (named by 6 respondents) are ignored.b Percent of population of all specialists, estimated by weighting the scores of provincial, district and AIC respondents by their proportions in the total population.c Weighted mean information flow score for all specialists, estimated in the same way as in footnote 2.d A positive score indicates respondents wish to receive relatively more information from that source. A negative score denotes the opposite.
179 The final column in Table 8.4 gives the difference in ranks between desirability and actual information flow scores. As we are using ranks, all figures are relative: a negative score does not necessarily mean that respondents wish to get less total information from that source, but merely relatively less than from other sources. Sources in the top one-third of the table ("AARD books" to "Training") are ranked high in terms of desirability and low in actual flows. The reverse is true of the final eight sources ("Own experience" to "Superiors"). Sources listed in the middle of the table are of two types: those high in both desirability and actual flows (e.g., "AIC publications" and "Agricultural newspapers and magazines"), and those scoring low on both counts (e.g., "University textbooks," "Agribusinesses," "Correspondence," and "Private technical guides"). Table 8.4 reflects a remarkably clear distinction among types of sources in terms of desirability and actual information flows. Publications were much more desired as sources than their low scores for information flow might suggest. The five sources with highest differences between actual and desired flows were all technical publications. Indeed, all technical publications, including university textbooks and commercial guides, ranked higher on desired than on actual information flows. The only publications with lower desired than actual flows were mass media, both agricultural and general. Research and contacts with researchers were somewhat more desired as sources than we might expect from their information flow scores. For instance, seminars and other meetings were 8th in terms of desirability but 16th in information flow (difference in ranks of +8). This was mirrored by field research projects (difference also of +8), direct contacts with researchers (+7) and even correspondence (+3). Respondents also appeared to want more training at extension institutions (difference score of +7) than they at present receive. But they wanted to rely less on their university course work (difference of -15). We can speculate on the reasons for this: it may indicate that their course work did not prepare them adequately for their extension work, or that they hoped for improved opportunities to obtain information in the future rather than depending on past course work. Respondents wished to rely relatively less than at present on their own experience and observations (5th in information flow, compared to 13th in desirability, a difference of -8). They also seemed to set little store by research they conduct themselves (score of only +1), despite the very limited amount of research most specialists currently engage in. Perhaps this is because of a lack of familiarity with or skills in research, or a realistic view of funding limitations for it. The general mass media at present seem to be rather more important information sources than many respondents would like (difference score of -8). They currently used the specialist agricultural press about as much as they would wish (difference score of -2). They still saw the agricultural press as a desirable source, though, as reflected by its rank of 9th in desirability scores. With the exception of contacts with researchers, personal contacts were much less desirable than we might expect from their actual roles. The top four information sources
180 (field agents, other extension specialists, farmers and superiors) ranked 17th, 15th, 12th and 22nd in terms of desirability. Five personal sources, including these, were among the bottom six sources in terms of difference scores. Discussion The data suggest that the same channels could be used to reach specialists at all three institutions. Specialists currently rely primarily on personal contacts for information but obviously would prefer to use technical publications -- though they seldom do so at present. Research is prominent among their preferred sources. Other channels they would like to use more include seminars and other meetings with scientists, and training at both research and extension institutes. They also like to use the specialist agricultural press. Some of the differences between the desirability of sources and the actual amount of information obtained from them, as shown in Table 8.4, are consistent with the strength-ofweak-ties theory (Granovetter 1973, Rogers 1983). This predicts that the highest quality (and therefore most desirable) sources are often those with which we have limited contact. Thus, research sources ranked low on information flow because they are published infrequently, and SMSs seldom came into contact with them (Table 7.14). However, they ranked high on desirability. Workmates such as colleagues and superiors were the reverse. They are located close to SMSs and come into frequent contact with them, so ranked high on information flow, but low in terms of desirability as sources. The strength-of-weak-ties has been explained as being due to homophily/heterophily and communication proximity (Rogers 1989:296). I already know what my close friend knows because we are homophilous -- we are alike, so we tend to know the same things and exchange information frequently. We also know the same people and have the same set of sources -- in other words, we have high communication proximity, as defined as the degree to which two individuals have overlapping personal communication networks. On the other hand, I may gain valuable information from a slight acquaintance because we are heterophilous, have low communication proximity, and exchange information seldom. The concept of homophily cannot be applied to non-personal sources. For instance, a newspaper is not analogous to my friend, even though I see both every day. But the concept of communication proximity still applies: the newspaper draws on a wide range of sources -- sources my friend and I do not have. It therefore has the potential to provide me with new information every day. Frequent contact does not necessarily imply low desirability. For instance, if you were to ask me where I would most like to obtain news, I would answer "the newspaper." Thus, some sources both supply large amounts of information and are desirable. In Table 8.4 such sources are represented by AIC publications (5th on desirability and 8th on information flow) and agricultural newspapers and magazines (9th and 7th). On the other hand, infrequent contact need not imply high desirability. SMSs correspond with researchers rarely (Table 7.14) and get little information from
181
Table 8.5
Frequency respondents read selected agricultural publications, and ratings for publication usefulness.a
Publication type
Mean frequency readb
Rank
Mean usefulnessc
Rank
Sinar Tani (biweekly agricultural newspaper)
4.16
1
4.91
7
Liptan (AIC fact sheets)
3.01
2
5.42
6
AIC booklets
2.96
3
5.48
5
Buletin Informasi Pertanian (AIC magazine)
2.66
4
5.55
4
Warta Litbang (AARD newsletter)
2.19
5
5.77
3
AARD books
2.06
6
5.91
1
AARD scientific journals
2.04
7
5.87
2
Data from questions P5 and 6, Publications questionnaire. n = 160-164. Scale: 1 = not once, 2 = once in 3 months; 3 = once a month; 4 = once a week; 5 = every 3 days; 6 = every day Scale: 1 = not useful, 3 = somewhat useful, 5 = rather useful, 7 = very useful agribusinesses or privately published technical guides. And they indicated little willingness to use these sources (Table 8.4). Actual information flow from a source and that source's desirability appear to be affected by different things. Information flow is at least in part dependent on proximity and the frequency of contact (this is discussed further below and in the next chapter). The desirability of a source, on the other hand, is not dependent on these characteristics, but would seem more likely to be affected by such factors as the source's relevance, ease of use, and credibility. Publication readership and usefulness Readership Questions P5 and P6 asked respondents how often they read several agricultural publications, and how useful they found these same publications. Table 8.5 summarizes responses to these questions.
182 The most frequently read source by a wide margin was the agricultural newspaper Sinar Tani. Some 79% of respondents indicated that they read the newspaper at least once a week. Only seven (4%) read it less than once a month. AIC publications ranked second, third and fourth in terms of frequency read, both averaging about once a month. Thirty-one percent of respondents read a copy of Liptan at least once a week, 28% read an AIC booklet, and 18% read the AIC magazine Buletin Informasi Pertanian. Percentages reading these publications less than once a month ranged from about 40% for Liptans and AIC booklets to 54% for Buletin Informasi Pertanian. Readership of AARD publications was less frequent. Ten percent read the AARD newsletter Warta Litbang at least once a week. Eight percent read a scientific journal at least once a week, and seven percent an AARD book -- even though AARD mailing lists indicate most are not sent these publications. (This may be because of the "halo effect" -over-reporting of a desirable activity.) About 70% read all three publication types less than once a month. This ranking is at least in part due to the frequency the various publications are issued. Sinar Tani appears twice a week, while most of the AIC and AARD publications appear irregularly at longer intervals. Nevertheless, increasing the frequency of readership is a challenge facing AIC and AARD publication managers alike. Usefulness Respondents' opinions of the usefulness of the various publication types stand in marked contrast to the frequency they read them (Table 8.5). The relationship is almost perfectly inverse: the more frequently a publication is read, the less useful SMSs think it is. AARD publications were thus seen as the most useful, and Sinar Tani as the least useful, publication type. This echoes the findings in Table 8.4. This inverse relationship provides two opportunities for increasing information flows from research to extension. First, the research content of Sinar Tani could be increased. Since the majority of SMSs read the newspaper frequently, they would be exposed to any research-based information in it. Various models could be used: •
A regular column containing research information, possibly rotated among the various commodities or subsectors.
•
Articles contributed by AARD scientists or communication personnel.
•
Inserts for readers to remove and save, for instance integrated pest management techniques, the characteristics of newly released crop varieties, or a list of research sites and information sources. Sinar Tani has direct links with the Ministry of Agriculture, and ministry and AARD
183 communication personnel sit on the paper's editorial board. Adapting the newspaper's content to carry more research findings should thus not prove difficult. All the suggestions above could also be implemented with other newspapers, especially those in the "Newspapers Enter the Village" (Koran Masuk Desa) program. Studies of the readership of these newspapers would be necessary in order to target audiences effectively. Second, Table 8.5 indicates that SMSs regard AARD publications highly and would read them if they had access to them. The frequency and print runs of AARD publications should therefore be increased, and publications should be sent directly to SMSs at both provincial and district levels. This could also be done at relatively little expense and with some coordination -- for instance in the provision of updated, computerized mailing lists. Information diffusion rates Question P7 asked food crops SMSs whether they had obtained information on six specific food crops technologies recently developed by AARD, and if so, where they had obtained most information on those technologies. Question P8 posed an equivalent set of questions relating to livestock. Table 8.6 and Table 8.7 present the results of these two sets of questions. Some respondents indicated more than one main source per item; such multiple responses are reflected in the tables.
184
Table 8.6
Percentages of respondents specializing in food crops who obtained most information from various sources.a
Topic
Yearb
Totald
Main source AAR D publ
AIC publ
Mass media
Colleagues
Otherc
Stemborer enemies
-e
29
18
18
18
21
79
Ash in soybean storage
'89
21
16
15
13
20
71
Sesbania green manure
'89
20
9
27
10
13
67
P dosage on rice
'89
27
5
9
16
14
64
Rama maize variety
'89
12
4
8
4
4
31
Barumun rice variety
'91
10
3
7
5
3
26
Mean
'89
20
9
14
11
13
56
Data from question P7. n = 99-100. Approximate year technology was first reported by AARD. Includes "training," "other," and "don't know." Received from at least one source. Rows sum to more than "total" value because of multiple responses. Data unavailable.
Knowledge of technologies The percentages of respondents who claimed to have received information about each of the technologies ranged from 95% for conical nests and the use of Glyricidia as small ruminant feed, to 26% for the newly released rice variety Barumun (last columns in Table 8.6 and Table 8.7). A greater proportion of livestock specialists knew of the livestock technologies than did food crops SMSs for innovations in their field. This was merely because the livestock technologies were generally older (mean of 6 years) than were those relating to food crops
185 Table 8.7
Percentages of respondents specializing in livestock who obtained most information from various sources.a Yearb
Topic
Totald
Main source AARD publ
AIC publ
Mass media
Colleagues
Otherc
Conical nests
'85
28
30
18
10
24
95
Glyricidia as feed
'81
38
48
20
3
10
95
Ducks without ponds
'81
20
30
25
8
21
83
Urea molasses block
'88
31
15
15
21
23
82
Newcastle vaccine
'88
20
10
43
8
16
80
Prolific sheep
'90
13
5
13
8
16
48
Mean
'85
25
23
22
9
17
80
21
13
16
11
14
63
1.27
0.75
1.00
0.65
0.86
3.80
e
Overall mean
Mean items per respondent ef
Data from question P8. n = 39-40. Approximate year technology was first reported by AARD. Includes "training," "other," and "don't know." Received from at least one source. Rows sum to more than the "total" value because of multiple responses. Combined from food crops (Table 8.6) and livestock personnel (this table). Mean number of items that respondent had obtained information on. Maximum possible score = 6. (mean of 2 years). The greater length of time allowed information about the older technologies to have spread to more specialists than had the more recent innovations (see below).
186
Sources of information AARD publications with few exceptions provided the most information on each technology. These publications were the major source for between one-quarter and twofifths of respondents who had heard of an innovation. Overall, they were the main source for one-third of respondents who had obtained information on the twelve technologies (21 of 63%, second-from-last line in Table 8.7). The mass media were the second most named source overall, though not consistently so. They were the main information source on the Newcastle disease vaccine for more than half of the respondents who had heard of this innovation, but were the most important source for only one in seven respondents for phosphorus fertilization of rice. AIC publications were the main source for about one-fifth of respondents who had heard of the innovations. They were the main information supplier for the Glyricidia, conical nests, and ducks-without-ponds technologies, but were relatively unimportant for several other topics. The reasons for this are discussed below. Colleagues were relatively unimportant overall as a source. They were the main source for about one in six respondents who had heard of an innovation, and accounted for the most respondents for none of the twelve topics. About one in five respondents who had heard of the new technology had obtained most information through "other" sources. Slightly less than one-half of these named training courses as their prime source. About the same number named other sources (e.g., seminars). A small proportion indicated they did not know where they had obtained most information. At first glance, the percentage of specialists obtaining information through official sources (AARD and AIC publications, training and seminars) appears low. But questions P7 and P8 asked respondents to mark only their main source of information. Table 8.6 and Table 8.7 thus do not indicate how many obtained information from sources other than their main one. Number of topics The last line in Table 8.7 shows the mean numbers of topics respondents had obtained information on. Respondents had heard of a mean of 3.8 of the six technologies in their commodity area. Of these, AARD publications were the most important sources for a mean of 1.27 items, the mass media for one item, and the other source types for less than one item each. Diffusion rates Figure 8.1 regresses the percentage of respondents who had obtained information on the technologies listed in questions P7 and P8 against the number of years those
187 technologies had been released by AARD. This figure reveals several interesting features of the spread of AARD innovations. First, the curves shown in the figure conform remarkably well to the expected shape: the upper part of an ogive. This shape is typical of innovation or information diffusion: the curve is relatively steep at first, but eventually plateaus out at a ceiling level (for this reason a logarithmic regression was selected). It is encouraging to note that the ceiling level for agricultural technologies in Indonesia approaches 100% of the SMS population -- in other words, eventually, all SMSs in relevant subject areas will obtain information about AARD technologies. Second, and disturbingly, the figure reveals the slow rate of information spread among the SMS population. It takes about two years after a technology is released for half of the SMSs to hear about it (dashed line), and a total of six years before 80% of the SMSs have obtained the information (dotted line). Third, AARD publications are the major source of information on AARD technologies until about eight years after the technology is released, when AIC publications take over. It is impossible to draw a meaningful trend line for the other sources as their regressions were not significant; however, they were less important overall than AARD publications but more important than AIC publications. The relationship between the AARD and AIC lines reflects two things. It takes time for AARD technologies to become incorporated in AIC publications -- hence the lower height of this curve in the first few years. But once AIC publications begin to carry the technology, their wider availability means they grow more rapidly in importance than do AARD publications. Possible data shortcomings The data in Table 8.6, Table 8.7and Figure 8.1 suffer from several shortcomings that mean the results presented above should be treated as a coarse approximation. •
Figure 8.1 is based on a small number of technologies only (11). A larger n might reveal different results.
• Some of the technologies (e.g., the new crop varieties) passed through the centralized recommendation procedure at the relevant directorate-general (Figure 4.2). I do not know about the others. A comparison of technologies that undergo the recommendation process and those that are disseminated horizontally might reveal different patterns.
188
Curves for other sources (mass media, colleagues, and others) were lower than for AARD publications and not significant. Percentages based on responses from 140 respondents (100 food crops and 40 livestock SMSs).
Figure 8.1
Percentage of respondents obtaining information on selected technologies, by yeara.
189
•
Because of space limitations, the questions used to measure information on the twelve technologies were crude. For instance, they did not seek to measure flows from any sources other than the major one.
•
The release dates vary in accuracy. For some of the technologies -- such as crop varieties -- the date can be determined with some precision. For others, it is only approximate. For instance, some research findings are published as scientific articles well before they are officially accepted as a technology recommendation. Others, such as the use of conical bamboo nests for chickens, are derived from indigenous technologies, so SMSs in some areas may have heard of them well before the release date.
•
The technologies chosen may not be directly comparable. The figure assumes that information on a minor technology such as chicken nest types is equivalent to that on a major breakthrough such as an oral vaccine or new crop variety.
•
Because the livestock technologies were generally older than the food crops, there may be confounding between the commodity grouping and the year of release.
Discussion Nevertheless, the picture painted by the responses to questions P7 and P8 is remarkably clear. Information on AARD technologies spreads among SMSs through multiple channels, with AARD publications being the most important of these. And while information does eventually reach nearly all the SMS population, it does so slowly. AARD publications were most often named as a source for these technologies, yet were given low information flow scores in the Information Sources questionnaire (question IS5, Table 8.1). This seems to be because of a difference in question wording. Questions P7 and P8 ask respondents to identify a single main source for AARD-derived technologies. Questions IS5 and IS7 focus on total flows of information about all topics useful for extension purposes -- a much broader concept. The data thus indicate that AARD publications are important sources of information about AARD technologies (the "strength-of-weak-ties" phenomenon), but that these technologies form a minor part of all information useful in extension. The responses to questions P7 and P8 also reflect the slow overall flow of information from research to extension about AARD technologies. After an item of information reaches the SMSs, it must pass through at least three other hands before it can be widely known by rural people: field agents, contact farmers, and follower farmers (Figure 3.2). And this is still only awareness; it presumably takes longer still before a sizable number of farmers will adopt a technology.
190 Compared to farmers and the general public, SMSs are an unusually easy audience to reach: their work addresses and names are known, their numbers are limited, and it their official duty to obtain information from research. The time taken for information on a technology to spread could be shortened dramatically by using the mass media (especially Sinar Tani) and the AARD newsletter Warta Litbang to carry more research findings, and by increasing print runs and mailing publications directly to SMSs instead of relying on the extension hierarchy to transmit messages. AARD publications are relatively important conduits for information about AARD technologies. This is despite their relative inaccessibility for most SMSs. Direct mailings to SMSs would improve their importance further. Notwithstanding the importance of AARD publications, a variety of other channels are useful for carrying research information. One channel may be useful for some technologies, but not for others. In such circumstances, a multiple-channel approach would seem advisable, with AARD institutes using all available methods to communicate messages to their audiences. Information needs The final question in the Publications questionnaire (P9) listed 62 agricultural and socio-economic topics and asked respondents to indicate the five they most needed information on. Despite these instructions, so many respondents marked more than five that I included all responses in the analysis. The 62 topics were grouped into three categories: food crops, livestock, and "others" (some of which were possibly relevant to both food crops and livestock SMSs). The analysis below treats the food crops and livestock topics separately, since direct comparisons between them would merely reflect the larger numbers of food crops specialists in the sample. For food crops topics, responses from food crops SMSs only are therefore considered; likewise for livestock topics. For other topics, I consider all SMSs: food crops, livestock, and the small number (13) who fell into neither category.
191
Table 8.8
Percentages of respondents who needed information on food crops-related topics.a
Crop type
Cultivation
Seed & varieties
Pests & diseases
Postharvest
Mean
Rice
14
55
35
36
35
Maize, sorghum, wheat
25
38
39
46
37
Roots & tubers
31
49
39
52
43
Legumes
24
49
53
51
44
Vegetables
37
50
55
60
51
Fruits
49
47
49
62
52
Mean
30
48
45
51
Estate and industrial crops
30
Machinery and equipment
40
Irrigation
43
Overall
43
Food crops specialists only (n = 108). Numbers are the percentages of food crops respondents who checked the corresponding boxes in question P9.
Table 8.8, Table 8.9 and Table 8.10 present the percentages of respondents who felt they needed information on each topic. The most needed food-crops related topic was post-harvest handling of fruits: nearly two-thirds of food-crops respondents indicated they needed information on this topic (Table 8.8). Post-harvest handling of vegetables, root crops and legumes also scored high, as did seed and variety questions for rice and vegetables, and pest and disease problems on legumes and vegetables. The least needed topics were all in the area of cultivation: rice (named by only 14% of respondents), other cereals, legumes, and root crops. Relatively few respondents stated they needed information about estate and industrial crops; this is hardly surprising given the food crops emphasis of most of these SMSs. Overall, respondents indicated they most needed information on post-harvest handling (on average named by 51% of respondents) and on fruits and vegetables. They
192 Table 8.9
Percentages of respondents who needed information on livestock-related topics.a
Livestock type
Manage-ment
Reproduction/ breeding
Feed
Health
Mean
Dairy cattle
45
45
58
39
47
Beef cattle, buffaloes
37
74
63
47
55
Goats, sheep
34
61
66
47
52
Improved chickens
24
47
45
45
40
Local chickens
32
58
66
42
50
Ducks
50
47
61
45
51
Mean
37
55
60
44
Fodder plants
74
Milk handling
63
Livestock postharvest
95
Overall
52
Livestock specialists only (n = 38). Numbers are the percentages of food crops respondents who checked the corresponding boxes in question P9. needed least information about cultivation and about rice and other cereals. Among livestock specialists, the most desired individual topic was post-harvest handling of animals (slaughtering, meat handling and processing, hide tanning, etc.) (Table Table 8.10
Percentages of respondents who needed information on other topics.a
Topic
Percent
Topic
Percent
Fisheries
16
Economics
43
Marketing
54
Farm systems analysis
67
Extension
50
Rural sociology
50
Regional planning
76
Other
18
All specialists (n = 159). Numbers are the percentages of all respondents who checked the corresponding boxes in question P9.
193 8.9). Almost all (95%) livestock specialists indicated they wanted information on this subject. Other popular topics were fodder crops (wanted by 74%), breeding of beef cattle and small ruminants, feed for several different species, and milk handling. The least popular livestock topics were management of improved chickens (needed by only one-quarter of respondents), and management of other species. Overall, respondents wanted most information about feedstuffs and about beef cattle and buffaloes (though there was very little overall difference among the various livestock species). They needed least information about management and about improved chickens. Among general topics, the most needed were regional planning (named by threequarters of all respondents) and farm systems analysis (named by two-thirds) (Table 8.10). Marketing and rural sociology also scored high. Fisheries scored low, except among SMSs who specialized in this topic (most fisheries and estate crops specialists had been deliberately excluded from the sample). Livestock topics seemed to be in greater demand from livestock specialists than food crops topics were from SMSs in this commodity grouping. This is reflected in the generally higher scores in Table 8.9 than in Table 8.8. Overall, a mean of 52% of livestock SMSs indicated they needed information on a livestock topic, while only 43% of food crops specialists stated they desired information on a topic in their field. This may indicate a greater demand (or a smaller supply) of information on livestock than on food crops; or it might merely reflect a greater degree of specialization among food crops extensionists and hence their individual need for information on fewer topics. It is not possible from these data to distinguish between these two possibilities. Some livestock SMSs indicated they also needed information on food crops, as did some food crops specialists for livestock topics. Presumably this is because a livestock SMS may occasionally be called on to answer questions related to other commodities: some interviewees said they faced such situations, while others denied doing so. However, the numbers of respondents indicating such needs were small, so they have been excluded from the analysis above.
194
CHAPTER 9 FACTORS INFLUENCING INFORMATION FLOW Introduction What causes an extension specialist to use one source of information rather than another? This chapter examines influences on the amount of information specialists obtain from four sources: AIC publications, the agricultural press, AARD publications, and other specialists. Eight characteristics of the relationship between source and receiver (the FP SCORES variables) were hypothesized to affect such information flows: Familiarity, Proximity, Structure, Capacity, Openness, Reward, Energy and Synergy. Because the survey asked specialists at three institutions about four different sources, it is necessary also to investigate how source type and institution affected Information Flow and the FP SCORES variables, and to control for these when studying influences on Information Flow. This chapter begins by describing the construction of the indices representing Information Flow and the FP SCORES variables. It then focuses on each index in turn, starting with Information Flow. For each index, I detail the index construction and differences in index values among the four source types (AIC publications, the agricultural press, AARD publications, and other specialists) and across respondents in the three institutions (province and district offices and Agricultural Information Centers). For the FP SCORES variables, I then discuss how the variable is related to Information Flow, examining the overall relationship as well as the differences among source types and institutions. SCORES
This chapter looks at each variable in turn. The next chapter examines how the FP relate to each other and, as a group, to Information Flow. Index construction
Questions IS5 and 7.1 to 8.15 aimed to measure Information Flow and the FP variables thought to influence it. The a priori indices contained between one and three items (five for Information Flow) intended to tap each variable. Five criteria for index construction (face validity, correlations with other items in the index, item-total correlation, Cronbach's α , and factor loadings) were used to determine which items to include in the indices. The Information Flow index was relatively simple to construct; the FP SCORES indices proved somewhat less straightforward. This was because of the lack of a standard set of questions to measure these concepts and the small number of items (due to space limitations) used to measure each one, and occurred despite the screening of questions in the pretest. SCORES
195
Face validity The a priori indices were constructed on the basis of face validity. However, preliminary analysis of the responses exposed the weak face validity of several items, especially in the Reward and Synergy indices (see the sections on these indices below). And in the case of Reward there appears to be strong face validity but poor actual correlation among items in the same a priori index.
Correlations among variables in the FP SCORES indicesa.
Table 9.1
Know Know info well type
Question
Loca- Easy to tion find
Credible
Complete
Map
8.13
8.4
8.10
8.11 8.14
Know info typeb
8.11
1.00
Know well
8.14
.25 1.00
Locationc
7.5
.10
.14
1.00
Easy to findc
8.2
.08
.28
.31
1.00
Distance
Map
-.01
.11
.24
.35
1.00
Job to get
8.13
.10
.23
.01
-.11
-.09
1.00
Credible
8.4
.15
.13
-.10
-.19
-.18
.42
1.00
Complete
8.10
.24
.04
-.06
-.18
-.13
.29
.48
1.00
Easy to use
8.5
.14
.12
.04
-.04
-.07
.21
.51
.43
Ready to use
8.9
.20
.09
-.04
-.02
-.21
.14
.34
.43
Easy understand
8.15
.18
.20
.03
.05
-.06
.21
.35
.36
8.3
-.04
.13
-.09
.33
-.01
.04
.12
.08
Area related
8.8
.23
.13
-.07
-.03
-.11
.15
.34
.40
Newc
8.12
-.16
.10
-.05
.03
-.06
.14
.19
.24
Try to get
8.7
.05
.05
-.12
-.36
-.15
.26
.44
.28
Agreebc
8.1
-.09
.08
.09
.16
-.01
.20
.17
.07
Timely
8.6
.13
.12
-.04
-.05
-.03
.18
.44
.42
Relevant
c b
(Continued)
8.2
Job to get
no.
c
7.5
Dis tance
Familiarity Proximity Structure Capacity
196 Table 9.1 (continued). Easy to Ready use use
Question
no. 8.5
8.9
Easy Releunders vant tand
Area New rela-ted
Try to get
Agree
Timely
8.15
8.8
8.7
8.1
8.6
8.3
8.12
Easy to use
8.5
1.00
Ready to use
8.9
.52
1.00
Easy understand
8.15
.57
.43
1.00
Relevantc
8.3
.18
.14
.13
1.00
Area relatedb
8.8
.39
.54
.26
.21
1.00
Newc
8.12
.12
.01
.07
.31
.04
1.00
Try to get
8.7
.16
.12
.09
-.06
.31
-.03
1.00
Agreebc
8.1
.19
.08
.16
.42
.01
.29
-.00
1.00
Timely
8.6
.55
.42
.39
.21
.50
.18
.23
.13
Openness
Reward
Energy Synergy
1.00
a
Data from questions IS7.5 to 8.15. Pairwise deletion of missing values (minimum n = 253). Coefficients greater than ±.14 are significant at p < .05.b Deleted from revised version of index.c Coding has been reversed for negatively coded item.
Inter-item correlations All but two of the correlations between pairs of variables in the same original index (in the boxes along the diagonal of Table 9.1) were significant but generally low. The mean correlation among items along the diagonal was 0.32, compared to 0.53 for the revised form of the Information Flow index (Table 9.2). This leads us to suspect that the indices contain items that do not measure the same variable. The two non-significant correlations were r = 0.04 between two of the Reward measures and r = 0.13 for the Synergy items. In both cases, the reason for the poor correlations is evident from the question wordings. The off-diagonal cells of Table 9.1 show that some items were more closely related to components of other indices than to those in their own. For instance, questions IS8.6,
197
Table 9.2 Number of variables, mean correlations, and Cronbach's α of original and revised forms of indicesa. Item
Variables Orig Rev
Cronbach α b
Mean rb Orig Rev
Orig
Rev
Questions Retain
Drop
5, 7.1- 7.3
7.4
8.14
8.11
Info flow
5
4
.44
.53
.79
.81
Familiarity
2
1
.25
-
.39
-
Proximity
3
3
.30
.30
.56
.56
Structure
1
1
-
-
-
-
Capacity
2
2
.48
.48
.65
.65
8.4, 8.10
Openness
3
3
.50
.50
.75
.75
8.5, 8.9, 8.15
Reward
3
2
.19
.32
.42
.48
8.3, 8.12
Energy
1
1
-
-
-
-
8.7
Synergy
2
1
.13
-
.23
-
8.6
a
7.5, 8.2, Map 8.13
8.8 8.1
b
Data from questions IS5, 7 and 8. n = 262 to 271. Correlation coefficient and Cronbach's α cannot be calculated for indices containing a single variable. 8.8 and 8.9, all in different a priori indices, have a mean mutual correlation of 0.49. This alerts us to the danger of multicollinearity in the multiple regression analysis planned. It also lends support to the suspicion that the questions tap underlying concepts other than FP SCORES. I address the first danger in the section on correlations among the FP SCORES variables in the next chapter. I discuss the second below. Item-total correlations Low item-total correlations affected the Familiarity, Reward and Synergy scales. The two Familiarity items had item-total correlations of r = 0.25, while the equivalent figure for question IS8.8 (Reward) was 0.16 and for the two-item Synergy scale 0.13. All other item-total correlations exceeded the criterion level of 0.3. Cronbach's α Cronbach's α for the indices ranged from an acceptable 0.75 for Openness to a poor 0.23 for Synergy (Table 9.2). Familiarity and Reward also had low α scores, while Proximity and Capacity had slightly higher, though still somewhat depressed, values.
198 The problem of poor index reliability arose partly due to the small number of items in each index -- a number constrained by the need to keep the questionnaire short. Cronbach's α is sensitive to the number of items in a scale: a scale with many mutually correlated items will yield a higher α than one with a few items. Except for the Information Flow index, all scales in this study contained only two or three items, so α values were inevitably low. However, the low levels of α are a further indication of problems with the unidimensionality of the indices. Factor analysis Because of the relatively small correlations along the diagonal and large values among off-diagonal items in Table 9.1, several of the factors did not reveal loading patterns closely consistent with the original indices. Only five instead of the expected eight factors were identified, and several items that had been expected to tap a common underlying variable loaded heavily onto different factors. Table 9.3 presents the factor loadings of the 17 variables; Table 9.4 lists the questions in the a priori indices and those retained in their revised forms.
199
Table 9.3 Quest. no.
Factor loadings of variables intended to measure the FP SCORES concepts.a Concept intended to measure
Factor
1
2
3
4
5
8.5
Openness
.81
.12
.15
.13
-.07
8.9
Openness
.77
-.03
-.06
-.18
.13
8.6
Synergy
.72
.15
.13
-.05
.04
8.15
Openness
.69
.09
.11
.20
.04
8.8
Reward
.64
-.03
.03
-.31
.33
8.10
Capacity
.60
.05
.37
-.12
.03
8.3
Reward
.18
.73
-.26
-.19
.25
8.1
Synergy
.10
.72
.10
.14
-.04
8.12
Reward
.06
.69
.17
-.06
-.11
8.13
Structure
.10
.22
.72
.09
.20
8.7
Energy
.16
-.12
.65
-.27
.05
8.4
Capacity
.50
.17
.60
-.10
.05
8.2
Proximity
-.02
.31
-.49
.44
.45
7.5
Proximity
.03
-.06
-.01
.76
.08
Map
Proximity
-.09
-.01
-.16
.63
.08
8.14
Familiarity
.02
.17
.19
.20
.74
8.11
Familiarity
.24
-.32
.07
.01
.66
24.7
12.5
9.6
7.5
6.0
Percent of variance a
Data from questions IS7.5 to 8.15. Based on correlation matrix with pairwise deletion of missing values (minimum n = 253). Factor analysis used varimax rotation and principal components extraction. Total variance accounted for = 60.3%. b
Factor loadings greater than .40 are in boldface.
200 The first factor, accounting for 24.7% of variance, had all the Openness items loading heavily onto it, plus items intended to measure Capacity, Synergy, and Reward. While Openness seems to be central to this factor, it is difficult to label such a diverse range of concepts. Possibly "Applicability" would be appropriate.
201
Table 9.4
Loadings of items on factors identified through factor analysisa.
Question
Factor most heavily loaded
Original index
Revised index
8.5
S easy to use
1
Openness
Openness
8.9
S info ready to use
1
Openness
Openness
8.15
S easy to understand
1
Openness
Openness
8.6
S has timely info
1
Synergy
Synergy
8.8
S topics related to area
1
Reward
8.10
S has complete info
1
Capacity
8.1
S agrees with other sourcesc
2
Synergy
8.3
S information relevantc
2
Reward
Reward
8.12
S info newc
2
Reward
Reward
8.13
Part of job to get info from S
3
Structure
Structure
8.7
Devote effort to get info from S
3
Energy
Energy
8.4
S is most credible
3
Capacity
Capacity
8.2
S easy to findc
-3d
Proximity
Proximity
7.5
S's nearest usual locationc
4
Proximity
Proximity
Map
Distance to known source location
4
Proximity
Proximity
8.14
Know S well
5
Familiarity
Familiarity
8.11
Know type of information S has
5
Familiarity
a
Capacity
Data from questions IS7.5 to 8.15. Based on correlation matrix with pairwise deletion of missing values (minimum n = 253).b S = Source.c Coding has been reversed so high score corresponds with high expected information flows.d Negative loading on factor.
202 The remaining two Reward items loaded heavily on the second factor, which accounted for 12.5% of variance. Agreement with other sources (Synergy) also loaded on this factor. The source's relevance seems to be at the core of this factor. A peculiarity is that all items loading heavily on this factor were worded negatively: "Information from the source often disagrees with information from other sources," its "information is not relevant to problems farmers in your area face," and it often provides information "that isn't new." (Only one other item loading on another factor ["The source is difficult to obtain"] had similar negative wording.) Perhaps the wording somehow influenced responses similarly across all three items. The third factor accounted for 9.6% of variance. This was a hodge-podge of items from four original indices: Proximity (loading negatively on the factor), Structure, Capacity, and Energy. Again, a possible underlying dimension here is the sense of duty the respondent may have to seek the source. This idea is most closely reflected in the statement, "It is an important part of your job to obtain information from the source," but people agreeing with this would also tend to agree that they "devote much effort to obtain information from the source." At the same time, the nature of the four source types is such that the hardest to get (AARD publications) is the one many specialists might see as one they should use most if they did their jobs well -- and the easiest to get (agricultural press) has the lowest status on such a measure. A source that is hard to get would therefore correspond to one they should use while fulfilling their duties -- hence the negative loading on the factor. The fourth factor accounted for 7.5% of variance. It corresponded to the two remaining Proximity measures: "Nearest place source is usually located" (question IS7.5) and map distance to the nearest known source location. The third Proximity measure, "Source is difficult to find (question IS 8.2) had a high loading on this factor also but was more closely related to the third (see above). The fifth factor accounted for 6.0% of variance. It was composed of the two Familiarity items on the questionnaire. Nevertheless, the correlation between these two variables was relatively low (r = 0.25, Table 9.1). Causes of poor index reliability What caused the a priori indices to perform so poorly in the above correlation and factor analyses? Several possibilities are evident: • Poor conceptualization. The FP SCORES concepts the indices are designed to measure are not unidimensional. As mentioned in Chapter 5, this appears to be true of Synergy, which is composed of at least two separate concepts (timing and repetition). It may also be true of Capacity (completeness and credibility) and Proximity (physical distance and accessibility).
203 •
Poor operationalization. The items used did not adequately measure the FP SCORES concepts. Some of the items have poor face validity. For instance, some apply better to personal than non-personal sources. An example is question IS8.14, "You know [source] well," which can apply to non-personal sources (particularly with its wording in Indonesian), but better fits individuals. Question IS8.2, "[Source] is difficult to find," may not adequately measure Proximity. And IS8.12, "[Source] often provides information that is not new to you," has low face correspondence with the concept of Reward. • Inadequate pretesting. The above problems should have become evident during the questionnaire pretest. Several problems did emerge during pretesting, and I adjusted the questionnaire to deal with them. For instance, the Openness items in the pretest had a mean α (mean of the four source types tested) of .17. In response, I changed the wording of one item and added another in the final version of the questionnaire. This appears to have been successful: the Openness index in the full survey had an α of .75 (Table 9.2). The pretest Synergy items had a mean α of .20; I dropped the poorest one in the final version. This was not enough to improve reliability (full survey α = .23). However, other problems did not become evident until the full survey was complete. Values of α for Familiarity and Capacity were higher in the pretest than in the full survey: α for Familiarity was .84 in the pretest but only .39 in the full survey; equivalent figures for Capacity were .80 (pretest) and .65 (full survey). Inter-item correlations were also correspondingly higher. For space reasons, the pretest included only one item each to measure Proximity and Reward. It could therefore not predict the reliability problems encountered with these indices. From hindsight, the pretest thus included too few questions from which to select "clean" items for the final instrument. Whatever the cause, the poor reliability of the a priori indices presents a problem. Which set of variables should be used in subsequent analysis? Three possibilities present themselves: • Using the a priori indices, amended as far as possible by deleting variables that are poorly related to other items in the index. •
Using the groupings suggested by the factor analysis.
• Selecting the single most appropriate variable to represent each of the FP SCORES variables, thereby eliminating the problem of poor reliability. Each of these options entails disadvantages. The first raises questions of index validity and reliability. The second means abandoning an attempt to test the FP SCORES model, and difficulties in interpreting some of the factor analysis groupings. The third risks the possibility that the selected measures have poor validity. I chose the first option because of the validity problems associated with the third option and the wish to test the FP SCORES model. I included the factor analysis groupings
204 along with the other criteria rather than using them as the sole basis for developing indices. Final form of FP SCORES indices The five criteria for index construction were occasionally in conflict. For instance, an item loaded onto a different factor from others in its a priori scale, but deleting it would significantly decrease the Cronbach's α of the scale as a whole. Judgements in such cases were based primarily on the face validity and item-total correlation of the scale. Applying the criteria yielded the indices listed in the last column of Table 9.4. Table 9.2 gives characteristics of the original indices and their revised forms after deleting items that failed to satisfy the criteria. Deleting variables made it possible to raise the mean correlation among items and the Cronbach's α of one index (Reward). Four of the indices (Familiarity, Structure, Energy, and Synergy) were reduced to single variables. The mean correlation of items in the revised FP SCORES indices was 0.40, compared to 0.32 for the original indices and 0.53 for the Information Flow score. These are still low, but are the highest possible given the shortcomings outlined above. Further details on the makeup of each index are given below. Information Flow Constructing the index Five questions aimed to measure the level of Information Flow from source to respondent. Four of these were items in question IS7: IS7.1 Frequency you read [source]. IS7.2 Frequency you obtain information useful for extension from [source]. IS7.3 Frequency you use information originating from [source] in your extension activities. IS7.4 Usefulness of [source] as a source of extension information for you. The fifth component of the index was drawn from question IS5: IS5 [source]?
For extension purposes, how much information do you obtain from
This question asked about 24 sources. I used responses about one of the four sources germane to the questionnaire version each respondent was sent:
205 • For respondents asked about AIC publications (version 1 of the instrument), responses to item IS5.4 ("AIC publications such as booklets, Buletin Informasi Pertanian, Liptan") were used. • For version 2 (agricultural press), item 5.1 was selected ("Agricultural newspapers/magazines [e.g., Sinar Tani, Trubus]") • For version 3 (AARD publications), item 5.7 was used ("Books published by AARD [e.g., research summaries, symposium proceedings]") • For version 4 (other specialists), item 5.13 was used ("Discussions with other SMSs"). These five items tap different dimensions of Information Flow. Question IS5 measures the total quantity of information. IS7.1 measures the frequency of exposure, while IS7.2 and IS7.3 measure the frequency of obtaining and using useful information. Question IS7.4 measures the usefulness of the information supplied by the course. Including responses to all five questions in an index yielded a Cronbach's α of 0.79 (an acceptably high value), and a mean inter-item correlation of r = 0.44. But excluding question IS7.4 improved the scale reliability still further (to α = 0.81) and the correlation to r = 0.53 (Table 9.2). This was because of the relatively low correlations between responses to this question and the other items in the index (Table 9.5). Question IS7.4, reflecting the source's usefulness, also may differ conceptually from the quantity of information flow from that source. Furthermore, a reliable measure for the dependent variable is crucial to this study. I therefore deleted question IS7.4 from the index on both pragmatic and conceptual grounds, leaving a scale composed of four items. Differences among source types Both source type and institution significantly affected Information Flow (Table 9.6). The agricultural press provided significantly more information than did the other three sources, and did so for all three groups of specialists. Second, also consistently across institutions, was "other specialists," with AIC publications and AARD publications providing least information to all groups. One of the five component items in the Information Flow index was question IS5, which measured information obtained from 24 sources, including the four used in this part of the study. Despite this broader range of sources, Question IS5 yielded the same ranking as the Information Flow index (Table 8.1). There was one exception to this: the agricultural press had significantly higher Information Flow scores than did other specialists, while on question IS5 the reverse was the case. We can trace this to higher scores for the agricultural press on questions IS7.1 to 7.3 -- all items measuring frequencies rather than the "information obtained" measured by question IS5. Both AIC and AARD publications scored poorly on the Information Flow measure, despite the high score by the AIC materials on question IS5. The AIC publications failed to
206 score any higher than AARD publications on questions IS7.1 to 7.3, indicating some discrepancy among the measures used. The level of Information Flow is probably in part a function of how often respondents are exposed to the source -- for instance, how often a newspaper is published. While the density of useful information is probably lower in the agricultural press than in the other sources, the press's higher frequency (tapped by question IS7.1) seems to compensate for this. Differences among institutions Overall, AIC specialists reported receiving more information than both province and district specialists. This is consistent with the findings for all 24 sources measured in question IS5 (Table 8.1). (This table and Table 9.6 are based in part on the same data, since responses to question IS5 are one of four items in the Information Flow index.) While the interaction between source type and institution in the analysis of variance was not significant, two other features are of note. The highest Information Flow score of all was for the agricultural press among province specialists. And AIC specialists reported receiving much more information from AARD publications than did their province and district counterparts.
Table 9.5 Flow indexa.
Simple Pearson's correlation coefficients between items in the Information Ques- Qty of info Freq read/ tion obtained talk to
Quantity of info. obtained
Freq get Freq use Usefuluseful info info ness
5
1.00
Freq. read/talk
7.1
.39
1.00
Freq. get useful info.
7.2
.43
.71
1.00
Freq. use info.
7.3
.40
.59
.64
1.00
Usefulnessb
7.4
.34
.30
.25
.40
a
Data from questions IS5 and 7. n = 272.
b
Variable deleted from revised version of index.
1.00
207 Table 9.6
Information Flow scores by source type and institutiona
Source type
Institution Province
District
Overall AIC
AIC publications
3.56 a
3.98 b
-
3.84 a
Agricultural press
5.22 c
4.80 c
5.09 b
4.97 c
AARD publications
3.42 a x
3.41 a x
4.26 a y
3.65 a
Other specialists
4.60 b
4.23 b
4.63 ab
4.43 b
Overall
4.21 x
4.12 x
4.65 y
4.26
n = 273. Overall analysis of variance: F for Source: 26.55** Institution: 4.57*
Interaction: 1.97ns
a
Data from questions IS5 and 7.1 to 7.3. Score range = 1 (low Information Flow) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized.
Familiarity Constructing the index The Familiarity index aimed to reflect how familiar respondents were with the source. The questionnaire contained two items tapping this concept: IS8.11 You know the type of information [source] has. IS8.14 You know [source] well. These two items both appear to have good face validity. Both loaded on the same factor (Table 9.4), but their low mutual correlation (r = 0.25) gave the resulting index low reliability (α = 0.39, Table 9.2). The reasons for this are unclear. IS8.11 appears to relate better to publications, and IS8.14 to people (though this is at least in part due to the translation into English; the original Indonesian wordings relate to both). The poor index reliability led me to drop one of the items. I retained Question IS8.14 for two reasons: it had slightly better face validity than question 8.11, and respondents appeared to have difficulty answering question 8.11 -- 15 of them failed to provide any response and 25 answered "don't know" to this question (see Appendix 5).
208
Table 9.7
Familiarity scores by source type and institutiona
Source type
Institution Province
District
Overall AIC
AIC publications
4.94 a
5.34 a
5.21 a
Agricultural press
5.50 a
5.39 a
6.18
5.63 a
AARD publications
5.75 a xy
5.26 a x
6.19 y
5.63 a
Other specialists
6.65 b
6.53 b
6.53
6.56 b
Overall
5.71 x
5.61 x
6.29 y
5.79
n = 276.Overall analysis of variance: F for Source: 11.28**Institution: 3.89* Interaction: 1.35nsa Data from question IS8.14. Score range = 1 (low Familiarity) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized.
Differences among source type s and institutions Both source type and institution significantly influenced Familiarity scores (Table 9.7). Other specialists consistently scored highest on this measure. No significant differences occurred among the other three sources. The relatively high Familiarity scores for other specialists may result merely from a different interpretation of the question for individuals as opposed to publications (the other three source types). On the other hand, they may also reflect a real difference between interpersonal and mediated channels. The high Familiarity scores with other specialists indicates an opportunity for networking and information exchange among extension personnel -- though Familiarity guarantees neither that specialists see it as their job to seek information from each other, nor that they expend much effort to do so. And hence it may not result in information exchange among specialists (see the sections below on Structure and Energy and the results of the multiple regression analysis). The low Familiarity scores for AARD publications are understandable given that most specialists are not sent them. But the similarly low scores for the agricultural press and AIC publications are puzzling because the respondents supposedly receive both routinely -the agricultural press more often than the AIC publications. Responses to the Publications questionnaire (Table 8.5 and Appendix 3) show that 79% of specialists read the newspaper Sinar Tani at least once a week, compared to 19% for the AIC magazine Buletin Informasi Pertanian and 31% for the AIC Liptan fact sheets. By contrast, only 10% claimed to read the AARD newsletter Warta Litbang, 8% scientific journals, and 8%
209 AARD books at least once a week. Table 9.8 Standardized simple regression coefficients (beta) for Information Flow against Familiarity, by source type and institutiona Source type
Institution
Overall
Province
District
AIC
AIC publications
.40
.48 **
Agricultural press
.22
.25
AARD publications
.29
.44 **
Other specialists
.03
.22
.18
.17
Overall
.28 *
.33 **
.12
.31 **
.48 ** .23
.25 *
-.03
.39 **
a
Data from question IS8.14. For sample ns see Table 9.9. Significance of beta (one-tailed): * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test. A reason for the low Familiarity with the agricultural press despite the frequency specialists receive it may be the low opinion some specialists have of Sinar Tani. Several interviewees said that they pay little attention to the paper as they saw it as of low quality -large parts are written by field agents with little credibility for specialists, and attitudes about the paper may be colored its being seen as a government outlet rather than an independent organ. AIC publications' low Familiarity score is hard to explain, since respondents are supposedly among the main users of these publications. Admittedly, the score is reduced because AIC specialists were not questioned about AIC publications, which they themselves author; presumably these specialists are very familiar with them. But non-AIC specialists seem to lack familiarity with these publications: district specialists were only as familiar with AIC materials (Familiarity = 5.34) as with AARD publications (Familiarity = 5.26), and province specialists were somewhat (though not significantly) less so (Familiarity = 4.94 compared to 5.75). If substantiated, this lack of familiarity with AIC publications among extension specialists should be a cause for some concern for AIC administrators. Specialists who are unfamiliar with AIC publications may be unaware of technologies thought suitable for their province. Strategies to raise the level of Familiarity are suggested in Chapter 11 AIC specialists reported significantly higher overall Familiarity scores than did their counterparts, primarily because of their greater Familiarity with AARD publications. They also scored somewhat (though not significantly) higher than their colleagues on the agricultural press; indeed, their Familiarity with these two publication types approaches their interpersonal Familiarity scores with other specialists.
210
Presentation of FP SCORES -by-Information Flow relations Table 9.8 plots the relationship between Familiarity and Information Flow overall and for the four sources separately. It also shows mean scores of Familiarity (horizontally) and Information Flow (vertically), corresponding to the rightmost columns of Table 9.6and Table 9.7).
211
Figure 9.1
Regression of Information Flow against Familiarity for four source types.
212 Table 9.9 Maximum and minimum ns for regression coefficients for Information Flow FP SCORES against variables.a Source type
Institution
Overall
Province
District
AIC publications
16-17
34-35
Agricultural press
16-17
40-41
19-20
76-78
AARD publications
15-16
35-38
20-21
72-74
16
30-33
14-19
61-67
63-66
141-146
54-59
261-271
Other specialists Overall a
AIC 50-52
Regression coefficients are presented in Table 9.8 and similar tables in this chapter. Table 9.8 compares standardized simple regression coefficients for Familiarity as a predictor of Information Flow for the various combinations of source type and institution. The slopes in Figure 9.1 correspond to the rightmost column in Table 9.8. Each of the sections below on the individual FP SCORES variables contains similar plots and tables. To avoid cluttering these tables, I provide in Table 9.9 the maximum and minimum sample sizes used in calculating each of these regressions. The small numbers in the province and AIC columns mean that it is difficult to obtain significance for betas representing these source-institution combinations. The figures show the mean values for each of the FP SCORES indices overall and for the four source types individually. They also show the regression between each index and Information Flow. The height of the mean-point symbols thus corresponds to the value of Information Flow for that source (rightmost column in Table 9.6). The horizontal position of the mean-point symbol reflects the value of the FP SCORES variable in question (in the case of Familiarity, the rightmost column in Table 9.7). The slope of the regression lines reflects the relationship between the FP SCORES variable and Information Flow (rightmost column in Table 9.8). The length of the regression line gives an idea of the range along the horizontal axis: it equals two standard deviations either side of the mean of the FP SCORES variable measured. These data are summarized in the table below the figure.
213
Influence on Information Flow The respondents' Familiarity with a source was relatively strongly related to the amount of information obtained from the source (overall beta = 0.31**), accounting for 10% of variance in Information Flow. Table 9.8 shows that the relationship between Familiarity and Information Flow was closest for AIC publications (beta = 0.48**) and for AARD publications (beta = 0.39**), and less so, though still significantly, for the agricultural press (beta = 0.25*). For other specialists as sources, the relationship was not significant, possibly because of the limited range in Familiarity scores for this source: all respondents except one gave scores of at least five (on a seven-point scale) for their Familiarity with other specialists. The overall beta for Familiarity was .31** (see above). The strong relationship between Familiarity and Information Flow held even when controlling for source type using dummy variables in a multiple regression: its beta value was again .31**, with Familiarity accounting for 8% of Information Flow. Among the three institutions surveyed, district specialists had the closest relationship between Familiarity and Information Flows (beta = 0.33**). Province specialists had a virtually identical regression slope (b = 0.24) but a slightly lower standardized regression score (beta = 0.28*). The beta value for AIC specialists (0.12) was low, probably because of the same reason as for other specialists. Despite the apparent variations in beta values among subsamples in Table 9.8, there were no significant differences either between sources within an institution (among values within a column in the table), or between institutions within a source (across a row). This lack of significant differences among sources and institutions means it is legitimate to group all responses together and use the overall beta value of 0.31 as a measure of the effect of Familiarity on Information Flow. Familiarity appears to be most important as a determinant of Information Flow for district specialists in using AIC and AARD publications. These are the sources district specialists are the least likely to receive (see discussion of Proximity below) or be familiar with. But if these specialists are familiar with them, they tend to use them more heavily. Familiarity has the least effect on the use of other specialists as sources, and on the amount of information obtained by AIC specialists. This is probably because Familiarity scores for these two subsamples are already high (Table 9.7); increasing them further will therefore have little effect. It is plausible that Familiarity is the result of Information Flow rather than the other way round, or that causation is circular. In other words, when we use a source we become more familiar with it, and this leads us to use it more. Similar relationships between the other FP SCORES variables and Information Flow may also exist. I discuss this further in Chapter 11.
214
Proximity Constructing the index Proximity means the physical distance between the source and the receiver, or the availability of the source to the receiver. I used three measures to tap this concept. Two were questionnaire items, and the third was measured from a map: IS7.5 Nearest usual location of [source]. IS8.2 [Source] is difficult to find. Map
Distance to nearest known location of [source].
The map measure was generated not from the survey responses but by measuring distances on a map from the respondents' addresses to the nearest known likely location of each source (see Chapter 6). Such a measure inevitably entails some error. For instance, not all AICs regularly receive AARD publications (as was evident from personal observation during my field visits). And some individual specialists are sent some publications, as shown by the AARD mailing lists I was given (Balitnak 1991, Balitvet 1991). And counter to my assumption, some specialists undoubtedly do not receive Sinar Tani. However, time and the amount of space available in the questionnaire made it impossible to collect all of the information necessary to create a "cleaner" measure. While all these are weaknesses of the map measurement approach, this item does not necessarily contain more error than any of the others. The questionnaire items, for instance, could be criticized for over-reliance on perception and memory. The map measure was fairly well correlated with the two other measures in the Proximity index, even though it was taken from an independent source. I therefore retained it in the index. Question IS7.5 and the map measure tap the physical distance dimension, while question IS8.2 measures source availability. While these are perhaps conceptually different, correlations between IS8.2 and the other two variables (.31 and .35, Table 9.1) were higher than the mutual correlation of IS7.5 and the map measure (.24). Availability and physical distance thus seem to be highly related, so I have included both in the index. The Proximity index was an instance where the five criteria for index construction (chapter 6) did not agree. The three-item Proximity index had a somewhat low mean interitem correlation of r = 0.30 and a Cronbach's α of 0.56 (Table 9.2). In factor analysis, question IS8.2 loaded onto a different factor from the other two items (Table 9.4). However, its loading on this factor (3) was negative, and it also loaded heavily on factors 4 and 5. Deleting it would have lowered Cronbach's α from 0.56 to 0.50. In addition, all three items had item-total correlations of more than 0.35. I therefore retained all in the Proximity index. The Proximity index could also be criticized for being less relevant for human than
215 for non-personal sources. Humans, it could be argued, move around more often than publications, and they can use the telephone to overcome the friction of distance. However, Table 7.14 shows that specialists do not travel very much to seek information (only about once every 3 months), and few have regular access to a telephone, especially for longdistance calls. Any differences between the effect of Proximity between personal and nonpersonal sources should be revealed in the regression analysis reported below. An argument could also be made for the opposite: that the Proximity measures are more applicable to personal than non-personal sources. It is relatively easy to say where a person is usually located (question IS7.5) (though this leaves unanswered the question of which person, or persons -- some of whom may be close by and others distant). But multiple locations may be applicable to a publication, on the other hand: is its usual location at the publisher's, in the library (where the copy is normally kept), or my office (when I want to read it)? (The second alternative was the one I intended to measure.) This difficulty is further compounded when we ask about a class of publications, some of which may be close by and others distant. Respondents may have interpreted questions IS7.5 and 8.2 differently. This and the inevitable error in the map measure (see above) increase the amount of error in the Proximity index. In the absence of strong evidence to the contrary, I assume in the analysis below that this error is randomly distributed about the mean. Differences among source types and institutions Both institution and source type significantly influenced Proximity (Table 9.10). While the sources did not vary in Proximity for the AIC specialists, province and district specialists reported that the agricultural press and other specialists were significantly closer than the other two sources.
216
Table 9.10
Proximity scores by source type and institutiona
Source type
Institution
Overall
Province
District
AIC
AIC publications
5.82 ab y
4.12 ab x
Agricultural press
6.10 b
5.93 b
6.53
6.14 b
AARD publications
5.31 a y
3.92 a x
6.44 z
4.95 a
Other specialists
6.34 b y
5.53 b x
6.38 y
5.96 b
Overall
5.91 y
4.90 x
6.45 z
5.49
4.65 a
n = 271.Overall analysis of variance: F for Source: 26.21**Institution: 46.32** Interaction: 7.36**a Data from questions IS7.5 and 8.2 and map distances. Score range = 1 (low Proximity) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized.
This was in part because of the way one of the three components of Proximity was calculated: the distance to the source was set to 0 km for the agricultural press but for the other three sources depended on the number of kilometers measured on a map to the nearest known source. According to the address list of 1560 specialists, most respondents were in the same institution or town as another specialist in their own field. But both the agricultural press and other specialists also scored high on the other measures in the Proximity index (questions IS7.5 and IS8.2). The results from these two questions support information from question P5 (Table 8.5), personal interviews and observations that the agricultural newspaper Sinar Tani is distributed to and is received by extension offices throughout the country, and justifies the somewhat arbitrary calculation of the distance component in the index. The overall Proximity value for AIC publications (4.65, Table 9.10) would have been higher had AIC specialists been questioned about them. Its low score is thus misleading, especially when compared to AARD publications, which had lower Proximity for both district and province specialists. AIC specialists enjoyed the highest degree of Proximity to all sources and reported no significant differences among them. AIC libraries appear to fulfil their function as a repository of AARD publications and hold a reasonably wide range of agricultural periodicals. Personal interviews indicated that these libraries were frequently used by AIC staff. The AICs and province offices also are home to several extension specialists, resulting in high Proximity scores for other specialists. In contrast, district specialists were far more isolated. Most AARD publications are not sent to the district level -- this is clearly reflected in the low Proximity scores for this
217 source type. Of concern also is the low Proximity score for AIC publications at the district level, possibly indicating that district respondents receive few of these publications also (see the comments on Familiarity above). These low Proximity values are not merely an artifact of the distance component in the index: district specialists gave low scores to AIC and AARD publications on the other two components: 4.06 for AIC publications on question IS7.5 (compared to 5.19 for province specialists), and 3.31 to AARD publications on question IS8.2 (compared to 5.13 for province specialists) (see Appendix 5). Interviews and personal observations indicate that district offices lack libraries and formal methods of circulating incoming publications to relevant personnel. Both AARD and AIC publications may thus be difficult to find at district offices, even if copies are received. And while the newspaper Sinar Tani is distributed to most extension specialists nationwide, few agricultural magazines reach the district level. The interaction between source type and institution was significant, presumably because of the high Proximity between AIC specialists and AARD publications. Influence on Information Flow The amount of information a specialist obtained from a source was closely tied to the source's Proximity. These two variables had a simple correlation of r = 0.38, higher even than Familiarity, indicating that physical closeness or access was more important overall than how well the specialist knew the source. Proximity alone accounted for almost 15% of the variance in Information Flow. Figure 9.2 shows that there was considerable apparent variation among sources in the effect of Proximity on Information Flow. For other specialists and AARD publications, the slopes are relatively steep and beta values high (beta = 0.45** for other specialists, 0.33** for AARD publications). For the other two sources, however, the slopes are almost flat (beta = 0.07ns for AIC publications, 0.12ns for the agricultural press).
218
Figure 9.2
Regression of Information Flow against Proximity for four source types.
219 Figure 9.2 shows that the more Proximate sources also had higher Information Table 9.11 Standardized simple regression coefficients (beta) for Information Flow against Proximity, by source type and institutiona Source type
Institution
Overall
Province
District
AIC
AIC publications
.26
.20
Agricultural press
-.14
.04
.48 *
.12
AARD publications
.10
.16
.44 *
.33 **
Other specialists
.34
.40 *
.57 **
.45 **
Overall
.29 **
.37 **
.49 **
.38 **
.07
a
Data from question IS7.5 and 8.2 and map measurement. For sample ns see Table 9.9. Significance of beta (one-tailed test): * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test. Flow. This would tend to reinforce the apparent effect of Proximity in a simple regression of all respondents. Controlling for institution showed this indeed to be the case: the beta value fell from .38** to .23** when source-type dummy variables were included in the equation. Proximity accounted for about 4% of variance in Information Flow in this equation. Table 9.11 shows the beta values from regressions of Proximity against Information Flow for each combination of source type and institution. The rightmost column gives the overall beta values for each source type, represented by the regression lines in Figure 9.2. This table shows that while other specialists and AARD publications had slopes significantly greater than zero, they were not significantly different from the beta values of AIC publications and the agricultural press. There was less apparent variation among institutions (the bottom line of Table 9.11) than among source types. All three institutions show a strong relationship between Proximity and Information Flow, though this was highest for AIC specialists (beta = 0.49**) and lowest for province staff (beta = 0.29**). However, the differences among beta scores were not significant. The body of Table 9.11 shows that the relationship between Proximity and Information Flow was strong for AIC specialists using all sources, and less evident for specialists at the other institutions. Again, despite the apparent variation in beta values, there were no significant differences among betas either among institutions (across rows in the table) or among sources (across columns). We can thus use the overall beta value of 0.38 as an estimate of the effect of Proximity on Information Flow from sources overall.
220 Proximity would thus seem to be a major influence on Information Flow for all specialists, especially when using AARD publications and for other specialists as sources. The high beta values for AIC specialists is surprising given that they are closer to all sources than are province and district personnel (Table 9.10). We might thus expect Proximity to have less effect on them than on their counterparts. Examining Figure 9.2 gives a clue as to the cause of this. Proximity seems especially important for both respondents at AICs as receivers, and other specialists as a source. Both of these samples have high overall Proximity (Table 9.10), indicated by the position of the latter's mean values near the right side of Figure 9.2. This suggests that the higher a source's Proximity, the more important Proximity becomes -- a curvilinear relationship. In other words, when a source is close by, small distances may be critical in determining whether it is used. For instance, a specialist is far more likely to use a publication in his or her own room rather than one in the library in the building next door. As the distance between source and receiver increases, such differences become less critical: if one has to travel to reach a source, it makes little difference whether the source is 15 kilometers away or 50. Two of the three components of Proximity, question IS7.5 ("Nearest usual location of source") and the map distance are already curvilinear in nature. The map distance used a log scale, and the scale used in question IS7.5 is gives equal weight to the difference between "own office" and "other room in same building" as to the difference between "less than 50 km away" and "less than 150 km away." These two components thus already take into account this differential friction of distance. But this is not enough to account for the strength of the curvilinearity in the relationship. Structure Constructing the index Structure refers to the organizational relationship between the source and receiver. The questionnaire contained a single item to measure this: IS8.13 Obtaining information from [source] is an important part of your job. Structure and Energy (also measured by a single variable, see below) loaded onto the same factor (Table 9.4), but their mutual correlation was not large enough (r = 0.26) to justify grouping them into a single index. I therefore kept them separate.
221
Table 9.12
Structure scores by source type and institutiona
Source type
Institution
Overall
Province
District
AIC
AIC publications
4.24 a
4.94 a
Agricultural press
5.18 ab
4.76 a
5.23 ab
4.97 a
AARD publications
6.00 b
6.16 b
6.05 b
6.09 b
Other specialists
5.00 ab
4.44 a
4.65 a
4.63 a
Overall
5.09
5.09
5.31
4.71 a
5.14 ns
n = 277.Overall analysis of variance: F for Source: 11.85** Institution: 0.11 Interaction: 0.97nsa Data from question IS8.13.
Score range = 1 (low Structure) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized.
Differences among source types and institutions Only source type significantly affected the Structure scores (Table 9.12). Respondents at all three institutions gave the highest priority to obtaining information from AARD publications. They may have done this merely because they saw this as the desired response; however, such an effect may have been mitigated because each respondent was asked about only one source, so was presented with no explicit comparison. Province personnel gave low priority to AIC publications, though not significantly less than to the other two source types. While these publications scored somewhat better among district personnel, their relatively low score indicates that specialists do not generally see it as their job to seek information from them. This may be because of the range of materials included under "AIC publications," which range from magazines and booklets with four-color covers and inserts to cheaply produced, two-color Liptan fact sheets. Specialists may see themselves as an audience for the former but not the latter. Indeed, one interviewee remarked with disdain that Liptans looked like the used paper that "street vendors wrap peanuts in." AIC staff say that extension specialists are not the main target audience for AIC publications, which are aimed primarily at field agents and farmers. But AIC staff interviewed clearly saw specialists as a secondary audience for their materials. Perhaps AIC and AARD administrators should review the types of material produced in order to target specialists more specifically. Most respondents saw other specialists as a low priority source of information.
222 Nevertheless, their counterparts were major sources for many (Table 9.6). This suggests that Structure -- at least as measured here -- has little influence on actual Information Flows. Influence on Information Flow The plot of Information Flow against Structure (Figure 9.3) shows that the slope for all respondents was flat (beta = 0.02ns). Structure was thus not related to Information Flow: whether specialists saw it as their job to use a source had little influence on the amount of information they obtained from the source.
223
Figure 9.3
Regression of Information Flow against Structure for four source types.
224 Table 9.13 Standardized simple regression coefficients (beta) for Information Flow against Structure, by source type and institutiona Source type
Institution Province
District
AIC
AIC publications
-.02
Agricultural press
.07
.44 ** b
AARD publications
.35
.16
ab
-.26
.10
ab
Other specialists
.38
-.20
a
-.35
-.11
a
Overall
.11
.00
-.13
.02
a
.18
Overall
ab
.15 .29
ab
.35 ** b
Data from question IS8.13. For sample ns see Table 9.9. Significance of beta (two-tailed test): * p < 0.05; ** p < 0.01.
Common letters a-b in a column indicate no significant differences in beta at p = 0.05 by z-test. This was true for all source type and institution subsamples except one: the agricultural press. For this source the relationship was strongly positive (beta = 0.35**), with Structure accounting for 12% of variance in Information Flow. Table 9.13 shows that this was due to the strong relationship between Structure and Information Flow for this source among district specialists (beta = 0.44**). The beta for the press was significantly larger than that for other specialists both for district specialists and overall. Figure 9.3 shows that the mean for AARD publications is below and to the right of those for the other three source types. This would tend to dilute any positive effect of Structure on Information Flow in a simple regression using all respondents. Controlling for source type using dummy variables shows this is indeed the case: doing so raised the beta from .02ns to .13*. Structure accounted for a small amount (1.5%) of the overall variance in Information Flow when source type was controlled for. Two of the subsamples, other specialists as a source and AIC specialists, had negative slopes, indicating that respondents obtained less information from a source, the more important they thought using it to be in their job. However, neither of these slopes were significant; the negative relationship they reflect appears to be an aberration. Why the strong relationship between Structure and Information Flow for the agricultural press but not for the other sources? The mean respondent did not regard it as part of his or her job to use the press as a source (Table 9.12): indeed, there appears to be some stigma attached to using the newspaper Sinar Tani (though none for the more specialized magazines such as Poultry Indonesia or Trubus. Several interviewees admitted almost with embarrassment that they used Sinar Tani as a serious source. And during the study design an AARD communication official queried question wordings that implied that the press was a legitimate or desirable source for extension specialists.
225 Yet specialists clearly do use the agricultural press, Sinar Tani included, often because it is one of the few non-local sources available (Table 9.10). And if they see it as their job to do so, they tend to use it even more. Some specialists may see Sinar Tani as a feedback mechanism that informs them about farmers' problems and concerns. I did not collect data on this, and as far as I am aware, no content analysis of Sinar Tani or Trubus has been performed. Trubus and other magazines contain information about particular techniques or solutions to problems. Sinar Tani does contain articles about pest outbreaks and other problems, and stories by and about farmers as well as extension personnel. However, these seldom air problems or voice criticisms. Sinar Tani's ties with the Ministry of Agriculture would seem to make it even less likely than other, privately owned newspapers to perform the role of a local advocate. It seems more likely that Sinar Tani fulfills the important role providing extension personnel with news about government programs and farming successes outside their own areas. A second possible cause for the peculiar relationships displayed in Table 9.13 is suggested in the above discussion. It concerns the identity of the source. "Agricultural press" is shorthand for the words actually used in the questionnaire: translated as "agricultural newspapers/magazines (e.g., Sinar Tani, Trubus)." While Sinar Tani's reputation among extension specialists leaves something to be desired, that of Trubus and other agricultural magazines is high. Specialists may have given rather different responses if asked about Sinar Tani and agricultural magazines separately. In light of this, lumping these media together in the questionnaire may have been inappropriate. It may have caused two problems, one serious, and one less so. First, a respondent may have been thinking of different publications while completing each question. In such a case, responses to one question may not have any relationship to responses to another, and we are left with large random errors. Second, and less seriously, one respondent may have thought consistently about Sinar Tani in answering questions, while another may have based her responses on Trubus. This would have increased to variance within each source type, making significant differences among the four source types harder to obtain. But this is less serious a problem since the relationships due to the FP SCORES variables should still hold true for that individual. The same two problems are present for Information Flow and all the FP SCORES variables and all four source types. Both AARD and the AICs produce various publications, and respondents may view each differently. For instance, the AARD newsletter Warta Litbang would probably be seen as more Open and of lower Capacity, and possibly as providing more Information Flow, than say, a symposium proceedings on maize. And the AIC Liptan fact sheets would probably fare similarly when compared with the AIC magazine Buletin Informasi Pertanian. For other specialists, the problem lies in asking about the combined characteristics of several individuals, one of whom may be a close friend and office-mate, while another may be a casual acquaintance working in the provincial capital hundreds of kilometers away.
226 The same problem would have arisen even if I had asked specialists to respond to questions about a single publication, such as Liptan. Which issue? On what topic? Individual Liptans may vary as much along the FP SCORES dimensions as do any other source. And no single source serves all Indonesia's (or even a single province's) extension specialists. This problem could have been reduced by asking about the distinct kinds of publications separately, or avoided completely by using an experimental approach -- though this would have obviated the wish to measure the effect of all FP SCORES dimensions together. Capacity Constructing the index Capacity refers to the amount of information a source has and to its credibility. Two questions measured this: IS8.4 [Source] is the most credible source. IS8.10 [Source] has more complete information than do other sources. Based on their face validity, these two questions would seem to tap different concepts: credibility and quantity. In factor analysis these two items loaded onto separate factors: question IS8.10 onto factor 1 and IS8.4 onto factor 3 (Table 9.4). However, their mutual correlation was high (r = 0.48), making the Cronbach's α value 0.65, relatively high for an index containing only two items (Table 9.2). In addition, question IS8.4 loaded almost as heavily onto factor 1 (loading = .50, Table 9.3) as on factor 3 (loading = .60). I therefore retained both in the index. Differences among source types and institutions Both institution and source type also significantly affected Capacity scores, though the influence of institution was too diffuse to cause any significant multiple range tests (Table 9.14). AARD publications consistently enjoyed the highest Capacity, while other specialists had the lowest.
227
Table 9.14
Capacity scores by source type and institutiona
Source type
Institution
Overall
Province
District
AIC
AIC publications
3.15 a
3.78 a
Agricultural press
3.68 ab
3.61 a
3.39 ab
3.56 b
AARD publications
4.47 b
4.82 b
4.05 b
4.53 c
Other specialists
2.79 a
3.15 a
2.72 a
2.94 a
Overall
3.51
3.87
3.42
3.68
3.57 b
n = 271.Overall analysis of variance: F for Source: 17.90**Institution: 3.42* Interaction: 0.52nsa Data from question IS8.4 and 8.10. Score range = 1 (low Capacity) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized. AARD publications' high Capacity scores are not surprising, given the perceived role of science in providing complete and authoritative information on many topics. The low Capacity of the agricultural press and AIC publications may be related to the generally poor opinions many specialists appear to hold about these sources, discussed in the section on Structure above. Indeed, Structure and Capacity are highly correlated with each other (Table 10.1 below). The poor Capacity attributed to other specialists may be because publications are generally seen as having more information than do individuals. It may also reflect the possibility that specialists estimate their colleagues' skills (and by extension, their own abilities) as low. The surveys collected other data tangentially pertaining to this possibility. While extensionists' lack of technical and extension skills were not among the most important problems listed in question P4, specialists still thought these problems were on average more than "important" (Table 7.15). The lack of training opportunities was named as the most serious problem facing specialists in obtaining information (question P5, Table 7.16). And such opportunities are indeed rare (questions IS10.3 and 11.1, Table 7.14). Such problems compound the fact that half of Indonesia's extension specialists have no farm background (Table 7.7 and Table 7.8), most have less than six years of experience as a subject-matter specialist, and fewer than one in ten have an advanced degree (Table 7.5). While the overall analysis of variance presented in Table 9.14 indicates a significant effect of institution on Capacity, this was too weak to produce any significant differences among institutions. However, district specialists generally placed most credence in the sources, while AIC specialists had the least. The reasons for this are not clear. Being more familiar with a wide variety of
228 sources, AIC respondents may be more realistic about the completeness or credibility of any one source. On the other hand, we might expect district specialists to invest less credence in sources they see as having little relevance to local field problems. The data tend to support the former view. Having said that, other explanations are possible: unraveling the role of Capacity is particularly difficult because this index is significantly correlated with six of the other seven FP SCORES variables (Table 10.1). Influence on Information Flow In general, respondents' opinions of a source's Capacity did not significantly affect the amount of information they obtained from it (beta = 0.07ns). But as Figure 9.4 shows, the slope for all respondents was considerably flatter than that for each of the four sources considered individually. The slopes for AIC publications (beta = 0.40**) and the agricultural press (beta = 0.28**) were steep, while that for other specialists was only marginally non-significant (p = 0.051). The slope for AARD publications was weaker but still larger than the overall value (Table 9.15).
229
Figure 9.4
Regression of Information Flow against Capacity for four source types.
230 Table 9.15 Standardized simple regression coefficients (beta) for Information Flow against Capacity, by source type and institutiona. Source type
Institution Province
District
Overall AIC
AIC publications
.09
.44 **
Agricultural press
.14
.33 *
.33
.28 *
AARD publications
.25
.28 *
.18
.14
Other specialists
.24
.40 *
-.02
.21
Overall
.01
.14
.06
.07
a
.40 **
Data from question IS8.4 and 8.10. For sample ns see Table 9.9. Significance of beta: * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test.
Among the three institutions, only district specialists had a significant relationship between Capacity and Information Flow. As with the sample overall, this relationship was stronger for the individual sources than for district specialists as a whole, all four of which were significant. Despite the variations among the slopes of the various subsamples in the table, there were no significant differences in beta within sources or institutions. The lack of an overall relationship between Capacity and Information Flow despite the significant relationships within sources was because of differences between the four source types. The high Capacity AARD publications were associated with low Information Flow, while the other three sources, with their lower Capacity scores, generated higher flows. This appeared to be related to factors other than Capacity. Within each source, however, respondents who felt a source had high Capacity used it more than others who saw it as lower in Capacity. This can be seen from Figure 9.4: the higher Capacity sources generally have lower Information Flow scores than the lower Capacity sources. Controlling for source type by including dummy variables in the regression equation revealed a fairly strong underlying relationship between Capacity and Information Flow: its beta value was .23** (compared to .07ns without controlling for source type), and it accounted for 4% of variance in Information Flow. Thus, despite Capacity's overall apparent lack of significance, respondents do tend to use sources they think have higher Capacity. AARD administrators clearly do not have to worry about whether their publications are perceived as complete or credible: most specialists think they are both. And the flat regression for AARD publications suggests that improving the perceived Capacity will not have any effect on specialists' use of AARD publications as information sources. The story is different for the other sources, however. Their low mean Capacity
231 scores indicate that they have generally low credibility and are seen as providing less than sufficient amounts of information. At the same time, increasing the Capacity of these sources raises the amount of information specialists obtain from them. Openness Constructing the index Openness is the ease of understanding and using the source's information. Three questions measured this concept: IS8.5 Compared with other sources, [source] is easy to use for extension purposes. IS8.9 [Source] provides information in a ready-to-use form. IS8.15 Compared with other sources of information about agricultural technology, [source] is easy to understand. The three measures all loaded onto factor 1 (Table 9.4) and were highly mutually correlated. With an α of 0.75 and mean inter-item correlation of 0.5 (Table 9.1 and Table 9.2), Openness was the most reliable of all the FP SCORES indices. This is despite the conceptual distinction that could be drawn between the ease of understanding a source and the ease of using it. Differences among source types and institutions Source type and the source-institution interaction were both significant influences on Openness scores. Since AIC publications are designed for use by extension workers, it is not surprising that they were seen overall as the easiest sources to understand and use. The agricultural press, written in a journalistic style, was also seen as easy to understand. Somewhat surprisingly, other specialists were seen as the least Open source. One interviewee hinted that other SMSs were reluctant to share information. Another reason may be that relatively few specialists share offices with others in the same field: a livestock extensionist may indeed have little information of use to a crops specialist. In any case, the reported lack of Openness among specialists is disturbing given the importance of the oral transmission of information within the extension system. AARD publications also had relatively low Openness scores (Table 9.16). Most AARD publications are aimed at scientists rather than practitioners. They describe the results of research rather than contain technology recommendations, so are difficult for extension personnel to interpret and use. The Openness scores reflect this.
232
Table 9.16
Openness scores by source type and institutiona
Source type
Institution Province
District
Overall AIC
AIC publications
4.59
5.19 b
4.99 c
Agricultural press
4.81
4.57 ab
5.03 b
4.75 bc
AARD publications
4.40 y
4.78 b y
3.52 a x
4.35 ab
Other specialists
4.22
3.98 a
3.79 a
3.98 a
Overall
4.50 xy
4.65 y
4.14 x
4.50 ns
n = 270.Overall analysis of variance: F for Source: 6.69**Institution: 1.84 Interaction: 3.22**a Data from questions IS8.5, 8.9 and 8.15. Score range = 1 (low Openness) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student- Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized. Overall, district specialists thought that sources were more Open than did their AIC counterparts. This was particularly the case with AARD publications: unlike AIC specialists, district (and province) respondents saw them as being relatively easy to use. This must be tempered by the knowledge that AIC specialists were significantly both closer to AARD publications (Table 9.10) and more familiar with them (Table 9.7). The only AARD publication district specialists have regular access to is the newsletter Warta Litbang, which is written in a semi-popular rather than a scientific style. District specialists may have based their responses on this, while province and AIC respondents may have considered also AARD's more technical materials and hence given them lower Openness scores. Nevertheless, district and provincial specialists appear to view AARD publications as potential information sources -- if they can gain access to them. They may also have a higher tolerance for relatively hard-to-use sources. A district specialist with relatively few alternative sources of information may regard all as being Open, while a specialist at an AIC, with far greater information resources at hand, may be more critical of sources that are difficult to use. However, this speculation cannot be confirmed through the data available. Influence on Information Flow Openness was significantly related to Information Flow (beta = 0.24** overall), explaining about 6% of variance in the dependent variable. This means that if respondents saw a source as easy to understand and use in extension activities, they would draw on it for information.
233 This relationship held true for respondents at all three institutions (bottom line of Table 9.17) and all the source types except AARD publications (Figure 9.5). The strongest relationship among the source types was for AIC publications (beta = 0.46**), seen as the most Open source of the four (Table 9.16). High correlations between Openness and Information Flow were also evident for other specialists (beta = 0.37**) and the agricultural press (beta = 0.30**).
234
Figure 9.5
Regression of Information Flow against Openness for four source types.
235 AARD publications were the exception to this. Overall, specialists who found AARD publications easy to use and understand obtained no more information from them than did their colleagues who saw them as less Open (beta = 0.08ns). Looking along the AARD publications row in Table 9.17, we see that this was true for province and AIC respondents but not district personnel, for whom Openness was consistently an important determinant of Information Flow. The reasons for this inconsistency are unclear. Controlling for source type has no effect on the influence of Openness on Information Flow. Without such a control, the beta value was .24** and the amount of variance explained, 5%. When dummy variables for source type are included in the regression equation, both values remain unchanged. Beta values did not vary among themselves among either source types or institutions. The overall value of 0.24** can thus be used as an estimate of the effect of Openness on Information Flow. The ease of understanding and using a source thus appears to be a major influence on how much information specialists obtain from it. Reward Constructing the index Reward refers to the relevance of the source's information to the receiver. It was measured by three questions: IS8.3 Much information from [source] is not relevant to problems faced by farmers in your area. IS8.8 [Source] often discusses topics that are closely related to conditions in your area. IS8.12 [Source] often provides information that is not new to you.
Table 9.17 Standardized simple regression coefficients (beta) for Information Flow against Openness, by source type and institutiona Source type
Institution Province
District
Overall AIC
AIC publications
.50 *
.42 **
Agricultural press
.10
.30 *
.37
.30 **
AARD publications
.11
.41 **
.05
.08
Other specialists
.35
.38 *
.48 *
.37 **
Overall
.23 *
.25 **
.38 **
.24 **
a
.46 **
Data from question IS8.5, 8.9 and 8.15. For sample ns see Table 9.9. Significance of beta: * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test.
236
Table 9.18
Reward scores by source type and institutiona
Source type
Institution Province
District
Overall AIC
AIC publications
3.71 a
4.43 ab
4.19 a
Agricultural press
4.29 a
4.03 a
4.29
4.15 a
AARD publications
5.09 b y
4.33 abxy
4.33 x
4.50 a
Other specialists
5.29 b
5.13 b
4.87
5.11 b
Overall
4.59
4.45
4.46
4.49
n = 267.Overall analysis of variance: F for Source: 4.26**Institution: 0.74ns Interaction: 0.15nsa Data from questions IS8.3 and 8.12. Score range = 1 (low Reward) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student- Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized. Codes for responses to questions IS8.3 and IS8.12 were reversed because these items were worded negatively. The first two questions appear to tap the Reward concept better than the last. And their similar wordings lead us to expect a strong correlation between them, and weaker relationships between these and question IS8.12. This was not the case: question IS8.3 was more closely associated with question IS8.12 (r = .31) than with 8.8 (r = .21), and the correlation between questions IS8.8 and IS8.12 was negligible (r = .04) (Table 9.1). In addition, question IS8.8 loaded heavily onto factor 1 in factor analysis, while the other two Reward measures loaded onto factor 2 (Table 9.4). This may have been because both questions IS8.3 and IS8.12 were worded negatively. I therefore dropped question 8.8, leaving a Reward index composed of two items and raising the mean inter-item correlation from r = 0.19 to 0.32 and the Cronbach's α from 0.42 to 0.48 (Table 9.2). Differences among source types and institutions Source type significantly affected overall Reward scores, with other specialists scoring consistently higher than the remaining three sources (Table 9.18). Most "other specialists" work in the same locality as the survey respondents, so have information that is more locally relevant than do the nationally or provincially oriented publications. We might expect provincially produced AIC publications to score higher on Reward than do the national AARD materials and press. That they do not is because of their low score among province specialists -- a finding difficult to explain. Province
237 specialists gave AIC publications low scores on several other measures also: Familiarity, Structure, Capacity, Energy and Synergy. Low scores on each of these variables may be interrelated. The effects of institution and the source-institution interaction were not significant overall. However, AARD publications were significantly more Rewarding for province personnel than for their AIC colleagues, and the difference between province and district specialists falls just short of significance. It is easy to speculate why province specialists obtained more Reward from AARD publications than did their district colleagues. They are closer to and somewhat more familiar with the AARD publications than are district personnel (Table 9.7 and Table 9.10). Because of their larger territories, they also deal with a greater range of commodities and agro-ecosystems, making a greater proportion of AARD publications relevant to their needs. It is less easy to see why province specialists should get more Reward from AARD publications than do AIC staff. The two groups are similar in the frequency of their contacts with researchers (Table 7.14). Perhaps AIC staff are familiar with a wider range of AARD publications than are province personnel (Table 9.7) but find a smaller proportion relevant to their needs. Influence on Information Flow The Reward level of a source was significantly related to the amount of information specialists obtained from it (beta = 0.23**). This means that a source's relevance to local conditions helped determine how much specialists used that source. The relationship between Reward and Information Flow held for all the four source types (Figure 9.6). The strongest relationships were for AIC publications and the agricultural press. It held also when source type was controlled using dummy variables: in this regression equation, the beta value was .26** and the proportion of variance explained, 6%.
238
Figure 9.6
Regression of Information Flow against Reward for four source types.
239 It is possible that some respondents saw the agricultural press as providing Reward because it informed them of agricultural problems rather than how to solve them. I discuss this possibility above in the section on Structure. The relationship between Reward and Information Flow did not hold across respondents at all institutions, however: only district personnel had a significant beta score (0.30**, bottom line of Table 9.19). The body of the table shows that this was because Reward consistently affected Information Flow for district specialists across all source types, unlike the case for province and AIC personnel. The lack of significance for several of the source-institution combinations is undoubtedly due to the small ns involved (Table 9.9). But the small values and even negative sign of other combinations are difficult to explain, as are the small betas for province and AIC specialists. Despite these apparent variations, there were no significant differences in magnitude among beta scores either within columns or across rows. We can therefore use the overall figure of beta = 0.23** as representing the effect of Reward on Information Flow. Energy Constructing the index The amount of effort a specialist expended to obtain information from the source was reflected by the source's Energy. The questionnaire contained only one item to measure this: IS8.7 You devote a lot of effort to obtain information from [source]. In factor analysis, this question loaded most heavily on factor 3, along with a measure of credibility (Capacity) and ease with which the source could be found (Proximity) Table 9.19 Standardized simple regression coefficients (beta) for Information Flow against Reward, by source type and institutiona. Source type
Institution Province
District
Overall AIC
AIC publications
.35
.31 *
Agricultural press
.06
.35 *
.41 *
.32 **
-.04
.43 **
.22
.22
Other specialists
.22
.48 **
-.11
.23
Overall
.11
.30 **
.16
AARD publications
a
.35 **
.23 **
Data from question IS8.3 and 8.12. For sample ns see Table 9.9. Significance of beta: * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test.
240
Table 9.20
Energy scores by source type and institutiona
Source type
Institution Province
District
Overall AIC
AIC publications
2.56
3.53 a
3.22 a
Agricultural press
3.25
3.32 a
3.82 b
3.44 a
AARD publications
3.33
4.56 b
3.95 ab
4.15 b
Other specialists
2.76
3.00 a
2.78 a
2.88 a
Overall
2.97
3.62
3.56
3.45 ns
n = 273.Overall analysis of variance: F for Source: 5.90**Institution: 2.55 Interaction: 0.99nsa Data from question IS8.7. Score range = 1 (low Energy) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized. (Table 9.4). The relationship with credibility seems to reflect a causal relationship (we devote much effort to obtain a source we perceive as credible) rather than a unitary concept. And the relationship with ease of finding the source is negative: we must devote effort to obtaining a source that is difficult to find. Despite its common factor loading with the Energy item, the ease of finding the source seemed to fit better with the other Proximity measures (see the section on Proximity above). I therefore treated the Energy item as a separate index as originally planned. Differences among source types and institutions Respondents at all three institutions consistently said they devoted the most effort to obtaining information from AARD publications (Table 9.20). This was especially true for district specialists, possibly because these found such publications most difficult to find (Table 9.10). Of course, this may have been the socially acceptable response (see the discussion on Structure above), though the questionnaire item (question IS8.7) invited no specific comparison with other sources. Seeking information from other specialists was generally allocated the least effort. This supports the argument made above (see the sections on Familiarity and Structure) that while many specialists may have the opportunity to exchange information, they do not necessarily do so. While the differences are not significant, it seems that province specialists expend least effort to seek information. That they nevertheless report obtaining about the same amount as district specialists indicates that Energy may not be a major factor in determining Information Flows.
241
Influence on Information Flow The amount of Energy respondents put in to seeking information from a source was not significantly related to the amount of information they actually obtained. In fact, the overall relationship between Energy and Information Flow was marginally negative (beta = .06ns, Table 9.21).
242 This lack of relationship held true for all source types and institutions (Figure 9.7), Table 9.21 Standardized simple regression coefficients (beta) for Information Flow against Energy, by source type and institutiona. Source type
Institution Province
District
Overall AIC
AIC publications
-.12
.14
Agricultural press
.12
.02
.24
.10
AARD publications
.46 *
-.16
.24
-.02
Other specialists
.10
-.27
-.30
Overall
.15
-.16 *
a
.13
.08
-.22 * -.06
Data from question IS8.7. For sample ns see Table 9.9. Significance of beta: * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test.
and when source type was controlled for using dummy variables (beta = .00ns). Only one beta value was significantly positive: that for AARD publications among province specialists, indicating that the more effort these respondents put into using this source, the more information they obtained. This is what we would expect.
243
Figure 9.7
Regression of Information Flow against Energy for four source types.
244 However, two subsamples had significant negative relationships between Energy and Information Flow: other specialists as sources (beta = -0.22*), and specialists at district offices (beta = -0.16*). This strange result implies that putting more effort into using a source actually reduces the amount of information that source yields. A possible reason for this is the distinction between information search and receptivity (Atkin 1973:238; see also Chapter 5). Some information comes to us without out needing to look for it (information receptivity). In the context of this study, such information is likely to arrive on specialists' desks via the agricultural press and AIC publications without any effort on their part, and through conversations with colleagues. The total amount of information obtained from these sources is larger than from AARD publications, but the quality of a particular item is not as great. This phenomenon is akin to the strength-of-weak-ties theory (see Chapter 5). AARD publications, on the other hand, are not widely distributed because of their short print runs and limited funding for distribution (Chapter 4). They thus are not sent to most specialists, as reflected by their low Proximity scores relative to other sources (Table 9.10). Nevertheless, specialists think they are useful (Table 9.18 and Table 9.14) and expend effort to obtain information from them (Table 9.20). This is information search. Why, then, is Energy not significant for AARD publications? Surely, if these are hard to find, specialists should devote Energy to getting them, and this should pay off? This appears to indeed be the case for specialists at province offices (beta = .46*) and AICs (beta = .24ns). Small numbers of AARD publications are indeed sent to province Dinas offices and AICs, though not directly to SMSs at these institutions -- hence the Energy required to search for them. But this effort pays off through higher Information Flow. District specialists have a harder time finding AARD publications, however. Only the AARD newsletter Warta Litbang is sent regularly to district Dinas offices, and this has experienced lapses in publication due to funding shortages. The district specialists thus must try hard to find these publications, as shown by the high mean Energy score of 4.56 in Table 9.20. That the difficulties in doing so frustrate their efforts is shown by the negative regression slope (beta = -.16) in Table 9.21. Synergy/Timeliness Constructing the index Synergy refers to the timing and mutual reinforcement of information from different sources. I tried to measure this concept using two questions: IS8.1 Information from [source] often does not agree with information from other sources. IS8.6 [Source] usually has information that is currently needed (timely).
245
Table 9.22
Timeliness scores by source type and institutiona
Source type
Institution Province
District
Overall AIC
AIC publications
3.88
4.49
4.29
Agricultural press
4.88
4.78
4.86 b
4.83
AARD publications
4.56 xy
4.68 y
3.67 a x
4.37
Other specialists
4.59
4.13
3.95 ab
4.18
Overall
4.48
4.54
4.18
4.45
n = 274. Overall analysis of variance: F for Source: 2.87* Institution: 1.76ns
Interaction: 1.34ns
a
Data from question IS8.6. Score range = 1 (low Timeliness) to 7. Common letters a-c in a column and x-z in a row indicate no significant difference at p = 0.05 by Student-Newman-Keul's multiple range test. Highest scores in each column are in boldface; lowest scores are italicized. Synergy is perhaps conceptually the weakest of Havelock and Lingwood's original HELP SCORES variables. This is shown by the poor correlations between the two measures used (r = 0.13), the a priori index's low Cronbach's α of 0.23 (Table 9.2), and the items' loading onto different factors (Table 9.4). Synergy is thus clearly not a concept adequately measured by this study, if indeed it is a unitary concept at all. I chose to retain question IS8.6, measuring timeliness, as being the more likely to affect Information Flow. Since the term Synergy is not appropriate to refer to this measure, I refer to it as Timeliness in the following discussion. I dropped question IS8.1 from the analysis altogether. Differences among source types and institutions The agricultural press scored highest on Timeliness with respondents at all three institutions (Table 9.22). This was as expected as a biweekly newspaper like Sinar Tani or a monthly magazine like Trubus can provide more timely information than the other sources. The content of most AARD publications in particular depends more on research projects planned and executed up to several years beforehand, rather than on rapidly changing field conditions. And because they depend on the AICs' financial planning procedures and are based largely on other information sources, AIC publications also have long lead times. Other specialists have infrequent opportunities for visiting researchers and training, so may also lack up-to-date information.
246 Nevertheless, these differences among sources were significant only for AIC specialists. District respondents were rather more optimistic than their colleagues about the published sources, while province specialists saw AIC publications as least timely. Institution had no significant effect on Timeliness levels -- except for AARD publications, which district specialists saw as being timelier than did AIC respondents. This is difficult to understand, given that district personnel are not sent most AARD publications. Perhaps AIC specialists, more tied into information networks generally, see AARD publications as relatively less timely than other sources they are familiar with, whereas district personnel have few other sources to compare them with. Another possibility is that specialists have different views of what constitutes "timeliness." AIC specialists, and to a lesser extent those at province offices, tend to package information and pass it on to others. District specialists, on the other hand, tend to use information more directly. District personnel may thus see a publication as "timely" if it concerns problems they are currently trying to solve (as was indeed the intent of the question wording). AIC personnel may look more at how soon they received the publication after it was written (or when the research was performed, or when a symposium reported in the publication was held). Indeed, several interviewees mentioned that they had not recently received a copy of the AARD newsletter Warta Litbang (which had not been published because of funding shortages). Influence on Information Flow With an overall beta value of .25**, Timeliness was relatively closely related with Information Flow (Table 9.8). This means that the timeliness of a source is an important factor in determining whether SMSs use it.
247
Figure 9.8
Regression of Information Flow against Timeliness for four source types.
248 This relationship held generally for the various sources and institution types (Table Table 9.23 Standardized simple regression coefficients (beta) for Information Flow against Timeliness, by source type and institutiona Source type
Institution Province
Overall
District
AIC
AIC publications
.34
.36 *
Agricultural press
.22
.10
.52 **
.22 *
AARD publications
.20
.38 *
.17
.13
-.22
.36 *
.21
.21
.25 **
.39 **
.25 **
Other specialists Overall a
.24 *
.37 **
Data from question IS8.6. For sample ns see Table 9.9. Significance of beta: * p < 0.05; ** p < 0.01. No significant differences among beta values at p = 0.05 by z-test.
9.23). There was some variation in the magnitude of the betas among the subsets of respondents, with province specialists even returning a (non-significant) negative value for information flows from their colleagues. This appears to be an aberration due to a small n. The overall beta for AARD publications was non-significant but positive, while that for AARD publications among district respondents was reasonably strong. The relationship with Information Flow held also when using dummy variables to control for source type: the beta value for this regression was .21**, and the amount of variance explained, 4%.
Table 9.24 Standardized simple regression coefficients (beta) for amount of information obtained against Timeliness, by source type and institutiona Source type
Institution Province
District
Overall AIC
AIC publications
.45
.33 *
Agricultural press
.09
.04
.48 *
AARD publications
.02
.10
.07
.17
-.19
.29
-.18
.09
.15
.11
.11
.10
Other specialists Overall a
.36 **
Data from questions IS5 and IS8.6. For sample ns see Table 9.9. Significance of beta: * p < 0.05; ** p < 0.01.
-.09
249 That Timeliness is related to Information Flow may be merely because the two variables measure the same thing. The Information Flow index contains two variables that measure the frequency-of-receiving information (IS7.1 and IS7.2), plus another that is closely related (IS7.3). To test this, I regressed the remaining variable in the Information Flow index (IS5, measuring amount of information) alone against Timeliness (Table 9.24). Several of the beta values significant in Table 9.23 failed to attain significance in this table, indicating a weaker relationship between the stripped-down Information Flow indicator and Timeliness. Three betas retained their significance, however, confirming that the relationship between Timeliness and the Information Flow measure was not dependent only on the frequency information was received.
252
CHAPTER 10 COMBINED EFFECT OF FACTORS ON INFORMATION FLOWS
Introduction The previous chapter looked at each of the eight FP SCORES variables and their effects on Information Flow individually. This chapter investigates how they related to each other and to Information Flow when taken as a group. Because of problems with the a priori Synergy index, I used a single question from this index to measure Timeliness. The final S in FP SCORES thus becomes a T, making an infelicitous acronym even more so: FP SCORET. I first discuss the correlations among the FP SCORET variables. Ideally for multiple regression analysis, these correlations should be low. This is not likely given the probable relationships among the concepts I attempted to measure. I go on to summarize how the eight indices differed among the four source types and three institutions surveyed. The remainder of the chapter is devoted to an analysis of the effects of the eight variables, taken as a group, on Information Flow. Correlations among FP SCORET indices Despite attempts to reduce multicollinearity, several of the FP SCORET indices were significantly correlated with each other (Table 10.1). The overall mean correlation among the FP SCORET variables was r = 0.22.
253
Table 10.1
Correlations among FP SCORET indices. Familiarity Proximity
Famili-arity
Structure
Capacity
Openness
1.00
Proximity
.24
**
1.00
Structure
.23
**
-.09
Capacity
.11
Openness
.17
Reward
-.22
1.00 .43
**
1.00
.06
.22
**
.56
**
.14
.04
.10
.24
**
.16 *
Energy
.05
-.29
Timeliness
.16
-.06
*
Reward Reward
1.00
Energy
-.05
Timeliness
.24
**
**
Energy
1.00
.26
**
.43
**
.15 *
.18
*
.49
**
.55 **
Timeliness
1.00 **
.23
**
1.00
a
Data from questions IS5, 7 and 8. Minimum n = 260. Significance: * p = .01; ** p = .001 Coefficients greater than 0.4 are in boldface.
Familiarity Familiarity was significantly correlated with three other indices: Proximity (r = .24**), Structure (r = .23**), and Openness (r = .17*). These relationships are easily understood: we get to know well sources that are close by (Proximity), we have a duty to use (Structure), and that are easily understood (Openness). Proximity Proximity was significantly associated with Familiarity (see above), Capacity (r = -.22**) and Energy (r = -.29**). The negative correlations with the latter two variables are probably because of the nature of AARD publications: these have high Capacity (Table 9.14), and respondents devote a lot of Energy to obtain information from them (Table 9.20), but they are generally located far away (Table 9.10).
254
Structure Structure was correlated with Familiarity (see above), Capacity (r = .43**), Openness (r = .22**), Energy (r = .26**), and Timeliness (r = .18*). The strong correlations with Capacity and Energy show that respondents think that the sources they have a duty to use (i.e., AARD publications, Table 9.12) are also the most complete and credible (Table 9.14) and they devote most effort to using them (Table 9.20). Capacity Capacity was significantly correlated with all but one (Familiarity) of the other FP SCORET indices. The more complete and credible a source, the higher its Structure score (see above), Openness (r = .56**), Reward (r = .24**), Energy (r = .43**) and Timeliness (r = .49**). Proximity declined with higher Capacity (see above). The correlations of Capacity with Openness and Timeliness imply that the more easily understood and timely a source is, the complete and credible it is seen to be. And information that is complete (Capacity) is likely to be more ready to use (Openness) than information that is incomplete. The correlations among these three variables thus suggest circular causation among them. The correlation between Capacity and Energy implies that SMSs strive to obtain information from complete, credible sources. It is reflected by the high scores of AARD publications for both variables (Table 9.14 and Table 9.20). It is difficult to discern a cause for the moderate correlation between Capacity and Reward by comparing Table 9.18 and Table 9.14. Nevertheless, we might expect such a relationship intuitively: people may judge relevant sources as being credible, and vice-versa. Openness Openness was significantly correlated with six other variables: Familiarity, Structure and Capacity (see above), Reward (r = .16*), Energy (r = .15*), and Timeliness (r = .55**). The relationships with Reward and Energy are weak but are supported by common sense: we associate ease of understanding with relevance, and devote more effort to using sources we understand. The correlation with Timeliness may be because the most timely sources -- the agricultural press (Table 9.22) -- are written in a popular style. Comparing Table 9.16 and Table 9.22 seems to confirm this: the agricultural press scored highest on both Openness and Timeliness among both province and AIC personnel.
255
Reward Reward was related to Capacity and Openness (see above) and to Timeliness (r = .24**). The last relationship indicates that timely sources are seen as relevant to user needs. Energy Energy was associated with Structure, Capacity, Openness, and Reward (see above), and Timeliness (r = .23**). The last reflects that SMSs strive to obtain information from sources they see as timely. Timeliness Timeliness was significantly correlated with Structure, Capacity, Openness, Reward, and Energy. All these relationships have been discussed above. Summary With few exceptions, the correlations among the FP SCORET variables were intuitive. Most of the exceptions can be explained by inspecting the breakdowns of index scores by source type and institution. All correlations were in the expected direction. The highest correlations involved Openness, Capacity and Timeliness. Component items in these three indices fell into the same factor in the factor analysis (Table 9.4). Nevertheless, these relationships did not lead to serious multicollinearity problems (see below). Comparison among source types and institutions Comparison of the Information Flow and FP SCORET indices among the various source types and institutions is complicated because one cell is missing: AIC specialists were not questioned about publications their own institutions or they themselves produced. The absence of these data biases mean scores that would normally include this cell (such as the Information Flow score for AIC publications or for AIC specialists). For this reason it is necessary to break down each of the Information Flow and FP SCORET indices by both source type and institution to discover any differences due to these two variables. This is done in Table 9.6 and similar tables. Summaries by source type and institution separately are presented in Table 10.2 and Table 10.3.
256
Table 10.2
Information flow and FP SCORET index scores by source typea.
Index
AIC publications
Agri-cultural AARD Other press public-ations special-ists
Overall
Information flow
3.84 a
4.97 c
3.65 a
4.43 b
4.26
Familiarity
5.21 a
5.63 a
5.63 a
6.56 b
5.79
Proximity
4.65 a
6.14 b
4.95 a
5.96 b
5.49
Structure
4.71 a
4.97 a
6.09 b
4.63 a
5.14
Capacity
3.57 b
3.56 b
4.53 c
2.94 a
3.68
Openness
4.99 c
4.75 bc
4.35 ab
3.98 a
4.50
Reward
4.19 a
4.15 a
4.50 a
5.11 b
4.49
Energy
3.22 a
3.44 a
4.15 b
2.88 a
3.45
Timeliness
4.29
4.83
4.37
4.18
4.45
n rangeb
50-52
78-80
72-75
64-71
267-277
a
Data from questions IS5, 7 and 8. Scale range from 1 (low) to 7 (high). Figures in a row followed by the same letter are not significantly different at p = 0.05 by StudentNewman-Keul's test. Highest scores in a row are in boldface; lowest scores are italicized.b n differs among variables because of missing data.
Comparison among source types Table 10.2 repeats the overall Information Flow and FP SCORET measures by source type (i.e., it summarizes the rightmost columns in Table 9.6 to Table 9.22). Because of the lack of data on AIC specialists' views of AIC publications, the multiple range tests reported in this table must be treated with caution if AIC publications or specialists are being compared. It should also be remembered that two variables, Proximity and Openness, had significant interactions between source type and institution. Of the nine indices, all except Timeliness differed significantly among the four source types. The highest Information Flow score was for the agricultural press, while the lowest was for AARD publications. Among the FP SCORET variables, AIC publications scored highest on Openness, the agricultural press on Proximity and Timeliness, AARD publications on Structure, Capacity and Energy, and other specialists on Familiarity and Reward. Turning to the lowest scores, AIC publications scored worst on Familiarity and Proximity; the agricultural press had the lowest Reward score; and other specialists fared
257 poorest on the remaining indices. Examining the values in Table 10.2 for each source type gives us a clue as to which FP SCORET variables are likely to affect Information Flow. Despite high Openness, AIC publications had low Information Flow, probably because of their poor Familiarity, Proximity, Reward and Timeliness values. AARD publications show a similar pattern, with low Information Flow although specialists think it important to obtain information from these publications (high Structure), their claim to devote a large amount of effort to doing so (high Energy), and their opinion that AARD publications are being generally the most complete of the four source types (high Capacity). AARD publications' relatively low Proximity and Timeliness as well as their limited Familiarity and Reward seem to be related to their low Information Flow score. The low scores on Structure, Capacity, Openness, Energy and Timeliness for other specialists indicate that respondents see their colleagues as not particularly knowledgeable (low Capacity), as in possession of outdated information (low Timeliness), and unable or unwilling to provide information in an easy-to-use form (low Openness). Furthermore, they do not regard obtaining information from their colleagues as an important part of their job (low Structure), and they devote little Energy to doing so. But these low scores were offset by specialists' familiarity with the source (high Familiarity), its closeness (high Proximity), and local relevance (high Reward). The highest Information Flow score was for the agricultural press. This source has high Proximity and Timeliness, which appear to more than compensate for the low Reward and medium levels of other variables. This implies that specialists use the (poor) information that is immediately accessible to them rather than the better quality information that is more difficult to obtain. Overall, Table 10.2 leads us to suspect that Familiarity and Proximity will be the most important determinants of Information Flow. This will be tested further by regression analysis later in this chapter. Comparison among institutions Table 10.3 compares scores on the Information Flow and FP SCORET indices for respondents at the three institutions surveyed (i.e., it repeats values in the bottom lines of Table 9.6 and similar tables). Similar caveats to those above are applicable for the validity of multiple range tests comparing AIC publications or specialists and for the interactions for Proximity and Openness.
258
Table 10.3
Information flow and FP SCORET index scores by institution typea. Province
District
AIC
Overall
Information flow
4.21 x
4.12 x
4.65 y
4.26
Familiarity
5.71 x
5.61 x
6.29 y
5.79
Proximity
5.91 y
4.90 x
6.45 z
5.49
Structure
5.09
5.09
5.31
5.14
Capacity
3.51
3.87
3.42
3.68
Openness
4.50 xy
4.65 y
4.14 x
4.50
Reward
4.59
4.45
4.46
4.49
Energy
2.97
3.62
3.56
3.45
Timeliness
4.48
4.54
4.18
4.45
n rangeb
64-67
142-148
57-62
267-277
a
Data from questions IS5, 7 and 8. Scale range from 1 (low) to 7 (high). Figures in a row followed by the same letter are not significantly different at p = 0.05 by Scheffé's test. Highest scores in a row are in boldface; lowest scores are italicized.b n differs among variables because of missing data. The respondents' institution was generally less strongly related to the nine indices than was the source type. Only four indices were significantly affected by institution: Information Flow, Familiarity, Proximity, and Openness. On all these measures except Openness, AIC specialists scored highest and district specialists lowest, with provincial specialists in between. For Openness, the reverse was the case. Given the importance of Familiarity and Proximity revealed above in Table 10.2, we would expect AIC specialists to have the highest Information Flow score. Table 10.3 shows that this was indeed the case. Effect of FP SCORET on Information Flow Source type and institution account for significant differences in Information Flow (Table 10.2 and Table 10.3). Controlling for source and institution will therefore decrease the amount of variance in Information Flow explained by FP SCORET. In order to gauge their effect, I ran regressions with and without controlling for institution and source type. I concentrate here on the effect of source type, since this was more important than that of institution. and FP SCORET
I first describe the results of regression analysis of responses for all respondents, with and without using dummy variables to control for source type and institution. I then discuss the results of similar regression models using subsets of respondents classified
259 according to source type and institution. Both simple and multiple regressions are reported for each group. All respondents Effect of FP SCORET indices in isolation The first column in Table 10.4 gives simple standardized regression coefficients (betas, = rs) for source type, institution, and each of the FP SCORET indices as predictors of Information Flow. (The order of the FP SCORET variables has been changed to ease the construction of the table.) These reflect the effect of each index on Information Flow when considered in isolation of any other factors. Five of the eight indices were significant predictors in the simple regression models: Timeliness and Familiarity each accounted for about 10% of variance in Information Flow, while Reward, Openness and Proximity each accounted for 4% to 6% of variance. Capacity, Energy and Structure were not significant.
260 Table 10.4 Standardized regression coefficients (betas) for predictors of Information Flow in simple and multiple regressiona. Variable
Multiple regressionb
Simple regression No controls
Controlling for sourcec
No controls
With controls
Source type d AARD publs
-.31 **
-.31 **
-.15 *
AIC publs
-.20 **
-.20 **
-.02
Ag press
.22 **
.22 **
.31 ** d
Institution Prov specs
.03
AIC specs
.20 **
-.04 .07 FP SCORET
indices
Proximity
.38 **
.23 **
.35 **
.19 **
Familiarity
.31 **
.30 **
.18 **
.22 **
Timeliness
.25 **
.21 **
.22 **
.14 **
Reward
.23 **
.26 **
.13 *
.21 **
Openness
.24 **
.24 **
Capacity
.07
.23 **
R2
.28
.41
F
24.30 **
19.33 **
Energy Structure
a
-.06 .02
.00 .13 *
Data from questions IS5, 7 and 8. Minimum n = 260.b Multiple regression model used forced entry of dummy variables (in model with controls) and stepwise forward entry of FP SCORET variables. Coefficients are not given for variables not included in the models.c Regression model of each FP SCORET variable individually (similar to simple regression), but after controlling for source type using dummy variables.d Dummy variables were entered as a block into equations.
261 Effect of source type Taken as a group, the source type dummy variables alone accounted for 23% of variance, more than any of the FP SCORET indices. The institution dummy variables alone accounted for 4% of variance, indicating that while where the specialist worked was a significant influence on information obtained from the various sources, it was not nearly as important as the nature of the source itself. Effect of FP SCORET indices, controlling for source type The second column in Table 10.4 shows the beta values for the FP SCORET indices when source type was included in the equation using dummy variables. These figures represent the effect of each index, considered in isolation of other factors, but after the effect of the source types had been removed. Seven of the eight indices were significant, the exception being Energy. Capacity and Structure are significant in this column but not in the first. The differences between the first two columns in the table are due to the differences among the four sources. The plots of Structure and Capacity against Information Flow (Figure 9.3 and Figure 9.4) show that the four means for the source types were distributed from top left to bottom right: source scoring lower on Structure and Capacity had generally higher Information Flow scores, while those scoring higher on these indices had lower flows. This counteracts the positive slopes of the regression lines for individual sources, yielding an almost flat overall regression line. Controlling for source type overcomes this difficulty. Doing so gives us a better picture of the effects of Structure and Capacity on Information Flow -- positive in both cases, as hypothesized. Controlling for source type has no effect on Energy, however. All the regression lines in Figure 9.7 are flat, so removing the differences among them has no effect. We are forced to conclude that the amount of Energy a specialist expends on using a source has no (direct) effect on the amount of information gained. This does not mean that Energy is not important, however, as discussed in the next chapter. The second column in Table 10.4 indicates Familiarity (beta = .30) as the most important influence on Information Flow overall, accounting for about 8% of variance in the dependent variable (compared to 23% for the source types taken as a group). Second is Reward (beta = .26, accounting for 6% of variance), followed by Openness (beta = .23, 5% of variance), Proximity (beta = .23, 4% of variance) and Capacity (beta = .23, 4% of variance) and Timeliness (beta = .21, 4% of variance). Removing the variance due to source type had little effect on Familiarity, Reward, Openness, and Timeliness. It had a major effect on Proximity, however, reducing its beta to .23 from .38 when it was considered in isolation (first column in the table). Inspecting Figure 9.2 shows us why: the means of the source types trend along the line of the regressions, from bottom left to top right. Removing the effect of the sources will tend to thus diminish the influence of Proximity -- in contrast to the situation with Structure and Capacity discussed above, where it increased their influence.
262 Joint effect of FP SCORET indices without controls The third column in Table 10.4 show the results of a stepwise multiple regression with all the FP SCORET variables included but excluding the source and institution dummy variables. Four of the five FP SCORET indices significant in simple regression entered this model. The best predictor was Proximity (beta = 0.35), followed by Timeliness (beta = 0.22), Familiarity, and Reward. With a beta value slightly below the required significance level (beta = .12, t = 1.84ns), Openness did not enter the equation, probably because of its relatively high correlation with Timeliness (r = 0.55, Table 10.1). When the remaining FP SCORET variables were force-entered into the model, Openness did join the other four variables as a significant predictor of Information Flow (beta = 0.18, t = 2.08*) (results not shown). Joint effect of FP SCORET indices, controlling for source type and institution The final column in Table 10.4 shows the regression model that included the source and institution dummy variables as controls. The same four variables -- Proximity, Timeliness, Familiarity and Reward -- though in a different order, entered this model. The dummy variables alone accounted for 27% of variance in Information Flow. Familiarity was the best FP SCORET predictor, followed closely by Reward and then by Timeliness and Proximity. The final form of the equation accounted for 41% of variance in the dependent variable. Including all eight FP SCORET variables in the model failed to raise R2 above 42%, but caused the Timeliness beta value to drop below the significance level (results not shown). Despite the correlations among the FP SCORET indices (Table 10.1), there was little evidence of serious multicollinearity in the models. The mean variance inflation factor (VIF) for the model reported in the last column of Table 10.4 was 1.50, and the maximum value was 1.79 -- well below the critical value of 10 suggested by Neter et al. (1983:392). Even when all the FP SCORET variables were included in the model, the mean VIF was 1.77 and the maximum only 2.60. Other signs of multicollinearity, such as changed signs of betas when new variables were added to the model, were absent. Summary Both Proximity and Reward were identified earlier through the analysis of variance by source type and institution as being probable influences on Information Flow (Table 10.2). The regression models confirm this and show that Timeliness and Reward, likewise associated with high Information Flow levels in some of the earlier analyses, also influence Information Flow. There is some evidence that Openness may also be related to Information Flow. However, its correlation with other variables (Table 10.1) means that any effect is masked when these other variables are taken into consideration. Structure and Capacity were associated with Information Flow for individual sources, but differences among the sources masked their effect when all the sources were pooled. However, these indices failed to enter the final multiple regression equations,
263 presumably because they were correlated with other variables already in the model. Table 10.5 Regression of Information Flow on source type, institution, and significant a FP SCORET indices (by source types) . Variables
AIC publications r
Agricultural press
Beta
r
Beta
Institutionb Province
-.05
.15
AIC
.09 FP SCORET
indices
Proximity
.07
.12
Familiarity
.48 **
Timeliness
.37 **
Reward
.35 **
Openness
.46 **
.30 **
Capacity
.40 **
.28 *
Energy
.13
.10
Structure
.15
.35 **
.46 **
.25 * .22 *
.31 *
.32 **
n
49
76
R2
.34
.20
F
7.62 **
.23 *
.27 *
4.40 **
(Continued) Only Energy was not a significant predictor in any of the models tested. It is possible that other measures of this concept would reveal relationships between it and Information Flow; however, it is not possible to test this using the data available. It is possible that a different set of variables are important predictors for various subsets of the specialist population (for instance, for specialists at district offices) or for certain source types (such as AARD publications). I test this possibility next.
264
By source types Table 10.5 gives the simple correlations of the FP SCORET variables with Information Flow, broken down by the four source types tested. It also gives standardized multiple regression coefficients for the FP SCORET variables obtained in stepwise regression controlling for institution type. (These are equivalent to the first and fourth columns of Table 10.4.) Because correlation coefficients and beta scores are equivalent, these figures can be compared directly: the r value indicates the effect of the individual FP SCORET variables alone on Information Flow, while the beta value reflects its effect when all other significant FP SCORET variables have been taken into account.
265
Table 10.5 Regression of Information Flow on source type, institution and significant FP SCORET indices (by source types) (continued). Variables
AARD publications r
Other specialists
Beta
r
Beta
b
Institution Province
.36
AIC
**
.00
-.04 FP SCORET
.07 indices
Proximity
.33 **
Familiarity
.39 **
.31 **
.17
Timeliness
.13
.26 *
.21
Reward
.22
.23
Openness
.08
.37 **
Capacity
.14
.21
-.02
-.22
.10
-.11
Energy Structure
.45
**
70
61
R2
.30
.34
6.88 **
.39 ** -.26 *
n F
.30 *
5.73 **
a
Data from questions IS5, 7 and 8. Multiple regression model used forced entry of dummy variables and stepwise forward entry of FP SCORET variables. Coefficients are not given for variables not included in the models. Significance: ** p = .01, * p = .05.b Dummy variables included in model with controls. The importance of the FP SCORET variables appears to vary from source to source (Table 10.6). For AIC publications, five variables (Familiarity, Timeliness, Reward, Openness, and Capacity) all were related to Information Flow in simple correlations. For the agricultural press, Reward, Openness, Capacity, and Structure had significant simple correlations, while for AARD publications, Proximity and Familiarity were significantly correlated. For other specialists, Proximity and Openness were most closely correlated with Information Flow. This variation among sources is paralleled in the multiple regressions by sources. Beta values were significant for Proximity (other specialists), Familiarity (AIC and AARD publications), Timeliness (AARD publications), Reward (AIC publications and agricultural
266 press), Openness and Energy (other specialists) and Structure (agricultural press). Of the eight FP SCORET variables, only Energy failed to enter one of the four stepwise regression models. The first four FP SCORET variables listed in Table 10.6 (Proximity to Reward) were significant predictors of Information Flow in the overall regression (last column of Table 10.4). The mean of the simple correlations reported in Table 10.6 for these variables was 0.27, compared to 0.17 for the last four variables (Openness to Structure) (0.13 if Openness is excluded). In addition, six of the nine beta values in the table are for the first four variables. In general, therefore, the first four variables are better predictors than the last four.
267 Table 10.6 Regression of Information Flow on source type, institution, and significant FP a SCORET indices (by institution) . Vari-able
Province specialists r
District specialists
Beta
r
AICspecialists Beta
r
Beta
Source type b AIC publs Ag press AARD publs
-.28
*
.07
.35
**
.43
-.39
**
-.13
**
.11 -.18
FP SCORET indices
Prox-imity
.29
.37
**
Familiarity
.28
.33
**
Timeliness
.24
.25
**
Reward
.11
.30
**
Openness
.23
.25
**
Capacity
.01
.14
Energy
.15
-.16
Structure
.11
.00
n R F a
2
.22
*
.49 .31
**
**
.44
**
**
.24
*
.12 .39
.33
**
.16 .38
**
.06 *
.08 -.13
63
139
54
.51
.40
.40
15.21
**
7.53
**
8.25
**
Data from questions IS5, 7 and 8. Multiple regression model used forced entry of dummy variables and stepwise forward entry of FP SCORET variables. Coefficients are not given for variables not included in the models. Significance: ** p = .01, * p = .05.b Dummy variables included in model with controls.
268 While some variation is to be expected among the sources due to chance and the relatively small ns, the lack of consistency is surprising. While Proximity, Familiarity, Timeliness and Reward are generally more important than the other variables, it seems that individual sources may vary considerably from this pattern: • Specialists tended to use AIC publications they were familiar with (Familiarity), and that they found easy to use (Openness), complete and credible (Capacity), timely (Timeliness), and locally relevant (Reward). • Specialists got information from the agricultural press if they thought it their job to do so (Structure), and if they found the press stories relevant (Reward), easy to use (Openness), and complete and credible (Capacity). • They used AARD publications that they were familiar with (Familiarity), that were accessible (Proximity), and that were timely (Timeliness). • They obtained information from other specialists who were close by (Proximity) and who provided information in an easy to use form. These differences mean that it is not enough to attempt to increase the FP SCORET characteristics across the board for all sources in order to improve Information Flow. Rather, a more selective approach is needed, adjusting the characteristics of each type of source according to what is effective for that source. For instance, the data show it makes little sense to exhort SMSs to devote more effort to using the various sources (i.e., to increase Energy), since this will have negligible (or even a negative!) effect on Information Flow (r = -.02). It makes considerably more sense to focus on variables that are significant in Table 10.5 for these publications. The paragraph above assumes that the direction of causality is from the FP SCORET variables to Information Flow rather than the reverse -- in other words, that specialists use sources because they are Familiar with them, rather than they become Familiar with them because they use them. I discuss the problems of causality and possibilities for increasing Information Flow for each of the sources in the next chapter. By institutions Table 10.6 shows there was considerably less variation among the respondents at the different institutions than among the source types. In simple regressions, Proximity, Familiarity, Timeliness, Reward, and Openness were all significant predictors of Information Flow for the district specialists, who constituted the largest subset of respondents. These five indices accounted for all the significant predictors for province and AIC specialists, though not all were significant for these respondent subsets. The lack of significance may in part be due to the smaller sample n's for the province and AIC specialists, though the simple correlations are also somewhat lower for province specialists.
269 This similarity means that SMSs at different institutions appear to behave in similar ways toward the various sources. This has two implications: • The three subsets can be combined in subsequent discussion. Improving the Proximity of a source, for instance, should be about as effective in serving the needs of AIC specialists as of province or district specialists. • From a practical point of view, it is not necessary to treat SMSs at the different institutions as different audiences, to be served with different media. A source that effectively reaches one group has the potential to reach the others also. Summary This chapter has investigated factors affecting the flow of information from four sources to extension specialists at three types of institutions. The four sources were Agricultural Information Center publications, the agricultural press, Agency for Agricultural Research and Development publications, and other specialists. The three institutions were province- and district-level agricultural offices and provincial Agricultural Information Centers. Eight factors were tested for their influence on information flows. Overall, specialists obtained large amounts of information from a source if: •
The source was nearby (Proximity).
•
They knew the source well (Familiarity).
•
The source provided information in a timely manner (Timeliness).
•
The source provided information that was locally relevant (Reward). There was some evidence that they obtained much information if:
•
The source was easy to understand and use (Openness).
•
The source was seen as especially complete or credible (Capacity).
•
They saw it as their job to obtain information from the source (Structure). It made little or no difference overall whether:
• They devoted a large amount of effort to obtain information from the source (Energy).
270
There were few differences among the specialists at the three types of institutions. However, the influence of the various factors varied considerably among the four source types. This means that any strategy for improving information flows to extension personnel must consider the characteristics of the sources and channels and extensionists' likely usage of these. This is discussed further in the next chapter.
271
CHAPTER 11 TOWARD IMPROVED LINKS Introduction This chapter is divided into four sections. The first summarizes the major findings of the study as reported in Chapters 7 and 8. It covers the importance of research-extension links, contacts between researchers and extension specialists, extensionists' information sources, the rate of spread of research-based information among extension specialists, and specialists' unmet information needs. The second section reviews the findings on why subject-matter specialists use the sources they do, reported in Chapters 9 and 10. It discusses in turn each of the hypotheses proposed in Chapter 5 (and subsequently amended), relating to prolificacy, Familiarity, Proximity, Structure, Capacity, Openness, Reward, Energy, and Timeliness. For each predictor variable, it then suggests ways of boosting information flows by increasing levels of the determinant. Some of these suggestions would simultaneously increase the levels of two or more determinants -- for instance, improving publication distribution would raise both the Proximity of the publications to the SMSs, and their Familiarity with them. The suggestions apply mainly to research-extension links, and to AARD publications. Suggestions for other media and institutions are also provided. The third section focuses on specific methods that can be used to increase researchextension links. It discusses publications of the Agricultural Information Centers (AICs) and the Agency for Agricultural Research and Development (AARD), the agricultural press and mass media, and interpersonal communication. The fourth section discusses some implications and shortcomings of the theoretical model used in this research for studying information flows. Summary of findings Importance of research-extension linkages Lack of information was a major problem for Indonesia's agricultural extension specialists. It was not the most important problem -- which is poor mobility -- but it came a close second (Table 7.15). Respondents thought that lack of training opportunities, limited funding for extension activities, the infrequency of meetings with researchers, and the irregularity with which publications were received were the major problems in obtaining information (Table 7.16).
272
Contacts with research Direct contacts with research were infrequent among extension personnel at all three institutions surveyed (province and district offices and Agricultural Information Centers) (Table 7.14). Every three years, the median specialist attended one research exhibition or seminar, attended two technical meetings with researchers, exchanged one visit with researchers, and provided feedback to researchers once. He or she attended no training and collaborated in no research with AARD personnel during that period. Obviously there were extension personnel with much higher levels of interaction, but these were relatively rare. These findings generally support those of Hussein (1986:422). Indirect contacts with research were greater: the median specialist reported reading four scientific journal articles and two AARD books within the previous three months. Information sources Specialists' main information sources were field extension personnel, other extension specialists, farmers, the specialists' superiors and colleagues, and their own experience and observation. Agricultural newspapers and magazines ranked seventh, and AIC publications eighth. Research sources -- both personal and non-personal -- were unimportant relative to other sources. The most important research source was the AARD newsletter Warta Litbang, which ranked 14th of 24 sources. AARD books and journals ranked 17th (Table 8.1). Most specialists read the agricultural newspaper Sinar Tani frequently, but found it to be of relatively little use as a source of extension information. They read AIC publications less frequently, and AARD publications still less so, but found the latter to be more useful in their extension work (Table 8.5). Spread of research-derived technologies Information about AARD-developed technologies diffuses only slowly to extension specialists. It takes an average of about two years before half the specialists have heard of a newly released technology, and about six years before 80% of specialists have heard of it. These rates could be increased dramatically through improved communication strategies. AARD publications of various types are the major source of information about these technologies (Figure 8.1).
273
Unmet information needs Both livestock and food crops specialists felt that the topic they most needed information on was post-harvest processing and handling of their commodities. Food crops personnel felt they most required information on fruits and vegetables, and required least on rice and on crop cultivation (Table 8.8). Livestock personnel indicated that feed and fodder plants, and the breeding of beef cattle and buffaloes, were their top information needs. They least needed information on livestock management, especially of improved chickens (Table 8.9). Among general topics, the most needed were regional planning and farm systems analysis (Table 8.10). Determinants of information flow Prolificacy Hypothesis test
Hypothesis 1 Receivers will receive more information from a prolific source than from a less prolific one. Since Indonesia's food crops research institutes have been more prolific (in terms of numbers of publications), and government policies have focused on rice rather than other commodities, I expected that food crops extensionist would recognize a larger total information flow from research than would their livestock colleagues. The evidence for this is mixed. Food crops and livestock respondents' did not differ significantly in the amount of total information they reported obtaining, or in the amount of information they got from research sources (Chapter 8). This may be because of the lack of a direct comparison in the questionnaires between food crops and livestock. However, livestock specialists appear to need information more than do food crops personnel. Another finding lends support to the prolificacy hypothesis. The agricultural press is more prolific than the other sources: Sinar Tani is published twice a week, and Trubus and other magazines typically appear monthly. AARD publications appear less often (Warta Litbang is bimonthly); AIC publications, though published more frequently, are typically short. Although the density of relevant information is less in the press than in the other sources, it still provides more information overall. It was not possible in this study to test the effects of prolificacy jointly with the FP variables because my measure of Information Flow contains items on how frequently a person receives information from a source -- a measure closely related to prolificacy. Including the latter as an independent variable in the model would mean basing SCORET
274 both dependent and independent variables on the same measures. To summarize, the findings on prolificacy are equivocal. The measures I intended to test the hypothesis failed to show any effect of prolificacy on Information Flow, but other data tend to support the hypothesis. I am forced to conclude that the hypothesis has not been adequately tested in this study, and that no firm conclusion can be drawn. Familiarity Hypothesis test
Hypothesis 2: Familiarity Information flow will be greatest from sources familiar to the receiver. The evidence supporting the relationship between Familiarity and Information Flow is compelling. AIC respondents, the group most Familiar with the four sources, received more information than did their province and district counterparts (Table 10.3). Other specialists, far the most Familiar source overall, came second in the amount of information they provided (Table10.2). Familiarity was an almost uniformly significant predictor of Information Flow in simple regression equations (Figure 9.1, Table 9.8). It continued to be important even when other variables were taken into account: with a beta value of .22, it was the most significant predictor in the overall multiple regression (Table 10.4), and was significant in four of the seven multiple regressions of subsets of respondents (Table 10.5 and Table 10.6). A question remains, however, about the direction of causation of the relationship between Familiarity and Information Flow -- as pointed out toward the end of Chapter 5. Instead of using a source because we are familiar with it, the argument could be made that we become familiar with a source because we use it -- reversing the direction of causation hypothesized in this study. It is likely that causation is in fact circular, that Familiarity leads to use, which in turn leads to greater Familiarity. Reverse or circular causation may occur with other variables also: for instance, if we want to use a publication, we may keep it in our office (making it Proximate) -- rather than using it because it happens to be nearby. While such arguments have merit, it is not possible to test them using these data. Increasing Familiarity The evidence shows that improving SMSs' familiarity with print sources should increase the amount of information they obtain from those sources. Familiarity could be improved through a variety of means. These include: •
Improve publication distribution (see under Proximity below).
275 • Publicize publications more widely, for instance, by advertising books through Warta Litbang and the AIC magazines Buletin Informasi Pertanian. • Produce catalogs of current titles and distribute these to extension personnel and others. All AARD institutes should publish a catalog of their materials. CALREC could produce a master catalog containing titles from all institutes. •
Increase the use of the agricultural press to promote research findings.
•
Involve scientists more in extension training courses.
• Use research publications in extension training courses, and encourage collaboration between researchers and AAET trainers in developing course materials. • Publish indexes or computerized databases to improve access to information already available at AIC libraries. • Increase the interpersonal interactions between researchers and extension personnel through meetings, exhibitions, seminars, training courses, joint projects, field surveys, and the involvement of extension personnel in field research. • Explore ways of bringing extension personnel to work at research institutes on a longer term basis. Options include the secondment or exchange of staff between AARD and the extension agencies, joint appointments, AARD's hiring of people with degrees in extension and development communication, or simply providing a desk and office space to extension personnel within the research institution. Proximity Hypothesis test
Hypothesis 3 Information flow will be greater from proximate than from distant sources. Physical closeness and access seem to be important determinants of the amount of information SMSs receive from sources. The most Proximate source, the agricultural press, was the one providing the most information (Table10.2). AIC specialists, the most Proximate of the three subsets of respondents, obtained the most information; the least Proximate, district specialists, obtained least (Table 10.3). In simple regressions, Proximity was the best single predictor of Information Flow overall, and was significant for respondents at all three institutions. Part of this correlation may be spurious, due to the reinforcing effect of source type on the regression. The closest source (the press) also provided the most information (Figure 9.2), though it is not clear whether this was because it was close or because it provided more information (prolificacy) on a more regular basis than did the other sources.
276 Controlling for source type revealed that the effect of Proximity was still strong, though somewhat diminished (Table 10.4). Among the four source types, Proximity was significant for AARD publications and other specialists, though not for AIC publications or the agricultural press (Table 9.11, Figure 9.2). In other words, specialists use AARD publications if they are accessible, and they obtain information from those colleagues who are close by. It is unclear why Proximity had no effect for the agricultural press or AIC publications. Proximity was important overall even when controlling for other variables, yielding a beta value of .19, third after Familiarity and Reward (Table 10.4). In multiple regression for subsets of respondents, Proximity was significant only for other specialists as sources and among AIC specialists (Table 10.5 and Table 10.6), though closer inspection reveals that this seems to be because of correlations with Familiarity and the differences in Proximity between district and province personnel. There is some indication of a curvilinear relationship between Proximity and Information Flow, with Proximity becoming more important at closer distances. This emphasizes the importance of improved distribution of research findings if they are to reach extension specialists and through them, field agents and farmers. It is not enough to ensure copies of publications are distributed to province-level offices; rather, they must be sent to those individuals that need them at both province and district levels -- at least to their offices, and ideally to their very desks. Increasing Proximity Low Proximity is perhaps the most serious problem facing AARD's communication effort. Specialists cannot obtain information from publications they never see. Extension personnel do not get information because they are never sent it. This is not (as is sometimes claimed) a matter of poor postal services; rather, it is because mailing lists do not contain the addresses of extension personnel urgently in need of research-based information. It is unrealistic to expect the 200 copies of a publication many AARD institutes currently distribute to meet the needs of 2000 extension specialists and 29,000 field agents scattered throughout a country the size of Indonesia. It is unrealistic to expect a district SMS or a field agent to travel to the provincial capital (which may be more than a day's journey away) to consult a research publication which may (or may not) be in an institute library. If the publication is not in the specialist's own office building, or in one next door, the data show that the specialist is unlikely to use it. It is also unrealistic to expect AICs, with their limited budget and facilities, to reproduce AARD publications for immediate distribution in their provinces. AIC personnel have little incentive to do so because of the credit-point system that rewards original authorship. Figure 8.1 shows that the delay in publication is so great that AIC publications exceed AARD sources in importance only in the eighth year after a technology is released. This is too slow. And it is unrealistic to expect the few province-level staff who do currently receive
277 AARD publications to transmit the information contained therein to district personnel. The information is too voluminous, technical and complex to be transmitted orally, specialists travel and meet too infrequently to make this practicable, and users may need to refer back to the material frequently. There is thus an urgent need to improve the distribution of AARD publications to ensure that they reach the people that can use them. Measures could include: • Explore the use of alternative media to disseminate research findings (see the section on using the mass media below). •
Print larger print runs of existing publications.
• Increase publication distribution budgets to ensure the publications are sent out promptly. • Distribute publications more widely. Priority should be given to ensuring at least one copy of each relevant publication is sent to each province and district Dinas office and the AIC libraries. Multiple copies should be sent where possible, especially to libraries where they are likely to receive heavy use. Publications should be sent to office (rather than home) addresses and to job titles (e.g., "Extension Section Head") rather than individual names. This will ease address list management. • Improve the management of address lists. This should be computer-based, and coordinated by the Central Research Institutes or CALREC. An easy-to-use address list management program should be acquired and its use made standard throughout AARD. The necessary hardware and supplies (computers, printers, labels, envelopes) should be earmarked specifically for publication distribution. Distribution staff should be given training in how to maintain lists, and a system of regular updates and list cleaning should be initiated. Updated lists should be obtained annually from Ministry offices as well as from internal sources. Institutes should share lists to avoid duplication of effort. CALREC should have a key role in coordinating these activities and providing training. • Target publication distribution to potential audiences. For instance, a publication on tidal swamp rice should be sent to areas with such an environment; only a limited number of copies need be distributed elsewhere, e.g., to libraries and universities. While this seems obvious, AARD institutes at present have no mechanism of ensuring this is done. The use of a computerized address database should ease this task. Eventually this could be in conjunction with a geographic information system containing information about land types and production systems in each area. • Print as many copies as are needed. The print run of publications should be commensurate with the likely audience. This will depend on the topic of the publication, its geographical and commodity coverage, and the type of publication (e.g., technical guide, symposium proceedings, scientific journal). • Make AARD publications available for sale through commercial bookstores throughout Indonesia. There appears to be a sizeable demand for some AARD titles,
278 especially among students and teachers. Proceeds from such sales should be placed in a revolving fund to support future publication activities. •
Allow individuals outside AARD to subscribe to serial publications such as journals.
• Make AARD publications available at a variety of locations: from the publishing institute, a central facility (such as a bookstore within CALREC or the separate Central Research Institutes), and the AARD headquarters and Ministry of Agriculture offices in Jakarta. It may be necessary to establish a system for transferring funds among institutions to enable such transactions. • Clarify policies regarding who is qualified to receive AARD publications free of charge, and who is required to pay. • Establish libraries in all Dinas offices to serve all personnel in that office. Once established, the library (rather than individuals) should be sent copies of AARD publications. • Improve the cataloguing and shelving of materials in AIC and AARD libraries, and train users in library use. Many materials are currently inaccessible because of poor catalogs, unorganized storage, and poor librarian and user skills. • Conduct occasional research on who receives AARD publications and where they get them. This would elucidate problems in the distribution system and ways they might be overcome. Such evaluations could also measure other aspects, such as preferred content, aspects that should be covered, the amount of scientific and technical detail required, and so forth. Structure Hypothesis test
Hypothesis 4: Structure Information flow will be greatest from those sources an individual perceives it his or her job to use. The relationship between Structure and Information was weak. Among the four sources tested, respondents most felt it their job to obtain information from AARD publications, yet this was their least important source. Their lowest priority as a source was other specialists, yet these ranked second in importance (Table 10.2). The relationship was stronger when the three institutions were compared: AIC specialists had the highest scores in both Structure and Information Flow (Table 10.3). The overall simple regression was negligible, and simple regressions were non-significant for all institutions and for all sources except the agricultural press (Figure 9.3). When other variables were taken into account in multiple regression, Structure was
279 significant in only one: for the agricultural press. This indicates that if SMSs see it as their job to use the press as the source, they will do so. For other sources and for all three subsets of respondents, however, Structure was not a significant predictor of Information Flow. Hypothesis 4 is thus rejected for all sources except the press. Increasing Structure The data indicate that specialists already regard it as their job to use AARD publications, and that exhorting them to do so is unlikely to have much effect on the amount of information they obtain. Structure was also non-significant for both AIC publications and other specialists. Among the four sources studied, the opportunities for using Structure to improve research-extension linkages are thus confined to the agricultural press. Interaction with the press should not conflict with government rules or interfere with ministry policies on the premature release of new technologies, especially those that must pass through the centralized recommendation process described in Chapter 4. Nevertheless, numerous topics could legitimately be discussed in the press under current rules, including already released technologies, research in progress, and current farming practices. If specialists think it is part of their job to use the press as a source, they will do so. They could be encouraged to do so by the following: • Include research briefs in Sinar Tani on a regular basis, preferably in an easily identified column. • Cooperate more closely with the private agricultural press, including Trubus and other specialized magazines. Such cooperation requires a greater understanding among AARD staff of the needs and limitations of the press. • Produce press releases aimed at national and local newspapers and magazines. Such releases should include instructional stories with technical information aimed at extension and farmers, as well as public relations and policy pieces. AARD communication staff would need training in how to prepare such articles. At present, SMSs have little incentive to stay current and knowledgeable in their specialties or actively to seek new information. (In fact, such search is likely to cost them money because of the smallness of their monthly working budget). They can gain promotion credits by attending training courses, to be sure, but access to these is not under the individual's control. Ways should be sought to provide such incentives -- such as increasing the opportunities for authoring articles based on current findings, stimulating peer recognition as a reward, and providing funding for attendance at research seminars and for conducting field research.
280
Capacity Hypothesis test
Hypothesis 5 Capacity Information flow will be greater from credible and complete sources than from those that are less credible or complete. Capacity did not prove to be a major predictor of Information Flow overall, though it was significant for individual sources. The highest Capacity source, AARD publications, provided the lowest Information Flow (Table 10.2). The group of respondents with the lowest opinions as to the Capacity of various sources, AIC specialists, scored highest on Information Flow (Table 10.3). The overall simple regression of Information Flow against Capacity was non-significant, despite all the simple regression coefficients for the various source types being positive, and two (AIC publications and the agricultural press) being significant (Figure 9.4). This apparent paradox was because higher Information Flow sources (the press and other specialists) had low Capacity, while the high-Capacity AARD publications yielded low flows, as revealed by the regression model that controlled for source types (Table 10.4). Multiple regression failed to reveal any relationship between Capacity and Information Flow for any of the sub-samples (Table 10.4, Table 10.5, and Table 10.6), presumably because it was masked by the effects of other variables. Hypothesis 5 is thus only partially supported. Increasing Capacity Overall, Capacity was an important influence on Information Flow only for the agricultural press and for AIC publications, though for district specialists it was significant for all sources. Many of the suggestions under Structure above for the agricultural press would also increase the Capacity of this source. Other suggestions include: • Produce publications that cover topics in sufficient detail so as to be self-contained. For instance, Warta Litbang often reports that a research project has met with success but does not give enough information for readers to be able to apply the findings themselves. The same is true of some Sambung Litluh fact sheets. Because such publications are often the only source of information a reader has on a topic, they should include all necessary details to allow readers to use the information.
281 • Books should cover single topics or commodities (e.g., maize) rather than a broad range of unrelated topics (as in some symposium or seminar proceedings). • Ensure that widely distributed publications such as Warta Litbang contain enough information to be useful. Warta Litbang should be expanded in size to accommodate this, and its content should be adapted to give it a more practical orientation. • Ensure that authoritative experts (such as AARD scientists) are quoted in press articles and AIC publications. • Include more research-based information in the agricultural press. This can be achieved by writing press releases and by closer cooperation with the press. • Increase the research-based content of AIC publications through closer collaboration between the AICs and AARD institutes -- for instance, by joint publications, collaborative authorships, mutual editing of materials, and ensuring that all AICs have a full range of AARD publications in their libraries. •
Provide more and better training to boost the Capacity of individual specialists.
• Adapt a publication's format and design to make it appear more authoritative (e.g., improve the appearance of Liptans). Provide illustrations where applicable. Openness Hypothesis test
Hypothesis 6 Openness Information flow will be greater from sources that receivers regard as easier to understand and use. There was some evidence supporting Openness as an influence on Information Flow. The agricultural press, with the second-highest Openness score among the four sources, yielded the highest Information Flow (Table 10.2). However, district specialists felt sources were most Open but had the lowest Information Flow; AIC specialists felt sources were least Open but had the highest flows (Table 10.3). Against this, the simple regression revealed that Openness was a significant predictor of Information Flow overall and for most of the sub-samples (Table 9.17, Figure 9.5). This situation is akin to that of Capacity, where the highest Information Flows come from the subsets scoring lowest in the predictive variable. Possibly because of this, or because of its high correlation with Timeliness (Table 10.1), Openness failed to enter the final equation in the overall multiple regression (Table 10.4). It was significant for only one of the sub-samples, other specialists. The effect of Openness evident in the simple regressions thus seems to be masked by other variables when all are considered together. Hypothesis 6 is thus only partially supported.
282
Increasing Openness There is some evidence that improving Openness can increase Information Flow. Below are some suggestions for increasing the Openness levels of various sources: • Design publications with extension needs in mind; they should contain enough information, written in a suitable form, to permit ministry officials and local extension staff to develop technology recommendations. The language used should facilitate their use in generating recommendations and their translation into materials suited for farmers. Technical guides are more suited to extension needs than are scientific journals (Sophia 1988, Sunarno 1983). • Ensure that the practical implications of a research study are clearly stated. Many research publications currently fall short of providing clear guidelines for the practical uses of research findings. Examples are: expanding agronomic studies to include economic analyses, providing illustrations of equipment or techniques, including maps of areas in which a technology is thought to be suitable (e.g., if it is confined to a particular soil type), providing concrete examples of the use of a technology, and ensuring that each publication or article contains a separate section (labeled "Practical Implications" or a similar title) that summarizes the practical import of a set of findings. • Increase the accessibility of scientists to answer questions and respond to field problems through more field visits in collaboration with extension personnel. • Improve the handling of visitors to AARD institutes by establishing standard procedures -- such as showing slide-tape or video programs, tours of exhibit plots, and the provision of informational materials. Encourage extension personnel and farmers to meet with scientists. Such open access should be monitored so it does not interfere with research activities. • Establish a rapid-response service to provide inquirers with references or copies of publications. This function could be performed by institute libraries under CALREC coordination. Such services should be advertised through Warta Litbang and AIC magazines. Reward Hypothesis test
Hypothesis 7 Reward Information flow will be highest from sources that provide receivers with the greatest rewards.
283 There is considerable evidence indicating Reward is a major influence on Information Flow. Among the four source types, other specialists had the highest Reward scores and the second highest Information Flow (Table 10.2). There was very little difference in perceived Reward levels among respondents at the three institutions (Table 10.3). Reward was a strong predictor in the simple regressions (Table 9.19 and Figure 9.6). It was the second strongest predictor of Information Flow in the overall multiple regression (Table 10.4), and entered three of the seven equations for the sub-samples (Table 10.5 and Table 10.6). Hypothesis 7 is thus supported. Increasing Reward Publications can be made more Rewarding for extensionists by the following measures: •
Target publications to specific agroclimatic regions or commodity types.
• Ensure that research addresses field problems as identified by farmers and extension personnel. Where possible, research reports should be couched in terms of a field problem rather than a scientific or theoretical one. • Ensure that research reports spell out the locations and conditions where findings can be expected to hold. For instance, the soil type, climate, season, and other key criteria should be restated in the conclusions (or "Practical Implications" section [see above]) to ease interpretation by extensionists. • Outline in publications the practical implications of research findings in specific situations. • Regularly survey extension personnel and farmers about problems they face, and develop publications and other media (such as training courses) that summarize research findings on these problems. Energy Hypothesis test
Hypothesis 8 Energy Information flow will be greatest from sources from whom the receiver devotes most effort to obtain information.
284 There was little evidence to support the amount of effort they devote to using a source as a determinant of the amount of information SMSs get from that source. The relationship is even negative in some instances. Respondents devote most effort to obtaining information from AARD publications, yet they actually obtained least from these sources (Table 10.2). Among the three groups of respondents, those at district offices expended the most effort to obtain information, but they obtained least (Table 10.3). Most simple regression slopes were flat or even negative (Table 9.21 and Figure 9.7). Energy failed to enter any of the multiple regression equations, except one, where it had a significant negative effect on Information Flow (Table 10.4, Table 10.5, and Table 10.6). Hypothesis 8 is thus not supported. This does not mean that Energy is unimportant, however. Rather, the findings indicate that respondents recognize and seek to overcome the constraints to obtaining information. They expend the most effort trying to obtain information from AARD publications, but these are too inaccessible for them to meet with much success. Increasing Energy The data indicate that in general, the amount of effort SMSs devote to obtaining information from a source does not affect the amount actually obtained. But this appears to be because of the costs of obtaining high-value sources such as AARD publications. Ways must be found to reduce the effort required to obtain these high-value sources, particularly by improving the publication distribution system (see under Proximity above). Timeliness Hypothesis test
Hypothesis 9 Timeliness Information flow will be greatest from sources that time them favorably. Hypothesis 9 aimed to test the concept of Synergy. However, this was not a single concept, at least as measured in this study. I therefore replaced it with one of its components, Timeliness. Hypothesis 9 is thus revised to refer to Timeliness only. There was considerable support for Timeliness as a determinant of Information Flow. The most Timely source, the agricultural press, also provided the most information (Table 10.2). Respondents at the three institutions did not differ significantly in their opinions as to the Timeliness of sources (Table 10.3). Simple regression revealed mostly significant positive effects of Timeliness on Information Flow (Table 9.23 and Figure 9.8). When considered together with other predictors in the overall multiple regression model, Timeliness was the fourth and last to enter the equation (Table 10.4), and it was
285 significant in two of the seven multiple regressions performed on sub-samples (for AARD publications and for district specialists) (Table 10.5 and Table 10.6). Hypothesis 9, as amended, is thus accepted. Improving Timeliness A number of measures can be taken to improve the Timeliness of sources. These include: • Speed the production and distribution of publications. Such measures as strengthening the editorial staff of AARD communication units and computerizing the generation of mailing labels should help contribute to this goal. • Produce and distribute certain publication types rapidly in response to field problems as they occur. Examples could include responses to pest and disease outbreaks and newly released crop varieties. • Improve the regularity of existing serial publications such as Warta Litbang by allocating them higher priority in budgets. Where it is necessary to save money, it seems better to reduce the number of pages or print quality of a publication rather than to eliminate issues. • Increase the use of the private mass media and especially Sinar Tani, through the means described below. • Produce press releases and foster closer relationships with private-sector media reporters. • Develop a regular column in Sinar Tani containing research information (possibly rotated among the various commodities or sub-sectors). •
Invite scientists or AARD communication staff to write articles for the media.
• Develop inserts in Sinar Tani and other publications for readers to remove and save. Topics could include integrated pest management techniques, the characteristics of newly released crop varieties, and a list of research sites and information sources. Overall
Hypothesis 10: FP SCORET Information Flow from a source to a receiver is influenced by the following aspects of the dyadic linkages between the source and the receiver: Familiarity, Proximity, Structure, Capacity, Openness, Reward. Energy, and Timeliness.
286 There is considerable support for four of the eight FP SCORET variables as determinants of Information Flow. Specialists use a source if it is close (Proximity) and well-known (Familiarity), and provides relevant, timely information (Reward and Timeliness). There is some evidence that specialists will use a source if it is easy to use and understand (Openness), see it as complete and credible (Capacity), and view it as part of their job to use it (Structure). There was little evidence that the amount of effort the specialist devotes to using a source (Energy) affects the amount of information a specialist obtains from it. It seems, however, that the importance of the FP SCORET variables varied from source to source. There was less variation among the three institutions surveyed. This means that each source must be treated separately in planning a strategy to communicate with extension personnel (and presumably, with any other audience also). For instance, increasing a source's Proximity will improve Information from one source but not from another. A number of strategies for each source are suggested by the regression analyses. These are discussed in the next section. Source-specific strategies Strategies for increasing Information Flows from research to extension should consider the characteristics of each source and audience group. What is effective for one source will not necessarily be useful for another. This section discusses the approaches that could be used for several source types. Three basic strategies can be pursued to increase Information Flows: • Attempt to increase the levels of important determinants of Information Flow for each source. This is equivalent to moving to the right along the regression lines shown in the figures in Chapter 9. • Adapt source types so they are similar to other, more successful types -- for instance, adapting AARD publications so they are more like AIC publications. This is equivalent to shifting the AARD publications regression line to coincide with that of AIC publications. • Carry research messages on other channels. In terms of the regressions in Chapter 9, this is equivalent to jumping from one line to another. AIC publications For AIC publications, multiple regression revealed that Familiarity and Reward were the most important determinants of Information Flow. Openness, Capacity, and Timeliness were important in simple regressions, but did not enter the multiple regression equation when all factors were taken into account (Table 10.5). In other words, extension specialists tended to use AIC publications if they were
287 familiar with them and if they thought they were relevant. Publications that were easy to use, that were complete and credible, and that addressed timely issues, also received high use. Efforts should be made to increase the levels of each of these determinants, using some of the strategies listed under each determinant in the preceding section. When compared with AARD publications, AIC materials provide SMSs with somewhat (though not significantly) more information. Specialists see them as easier to understand and use than the AARD publications. Efforts should thus be made to improve the Openness of AARD publications, for instance by writing in practical terms rather than scientific jargon, providing concrete examples, and spelling out the practical implications of research findings so as to ease the task of ministry and local staff in translating the findings into extension recommendations. AARD should not rely on the AICs to be the sole source of research-based information for subject-matter specialists. Doing so places an unnecessary extra hurdle in the way of smooth research-extension linkages. The AICs were established to develop media materials aimed at farmers. In fact, their primary audience is field extension agents. They lack the resources to serve even this audience adequately, and their current publication print runs can serve only a handful of farmers (I was unable to evaluate the impact of their materials intended for broadcast). Figure 8.1 shows that the AICs are a minor source of information on AARD technologies until about eight years after the technologies are released. And the SMS population is small enough for AARD to serve directly. A promising option is to develop joint working groups and editorial boards composed of AARD and AIC staff to develop materials aimed at extension personnel. In such teams, AARD would provide the technical expertise, while the AICs would ensure that the materials produced would fit the needs of extension personnel. Funding could be drawn from both institutions. A start has been made in this direction in the form of Sambung Litluh, the fact sheets jointly produced by the AIC and researchers in South Sulawesi and Central Java. More such collaboration is required -- but care must be taken to avoid creating yet another institution or rigid set of rules in an already complex bureaucracy. Other methods of increasing collaboration between extension and research personnel have been outlined above in the section on Familiarity. The agricultural press and mass media Structure was the most important determinant of Information Flow for the agricultural press, indicating the extension personnel will use this source if they feel it their job to do so. The relevance of the press (Reward) was also important, as were its ease of use and the amount and credibility of information it contained. Some ways of using the agricultural press and the mass media have been outlined in the section on Timeliness above. Research findings and other extension information could be disseminated more rapidly through channels the extensionists come into contact with most often: i.e., the agricultural press and other mass media.
288 But is putting research information into the mass media really a good idea? The decision as to which sources to use depends on several factors: • Nature of the medium Some media are more suited to conveying technical information than others. Television, for example, can show motion and color well, but is not well suited to giving detailed instructions. Newspapers typically carry relatively short and non-technical material, and the quality of graphics is not high. • Audiences Media have different audiences. Television has a mass audience of millions of both rural and urban dwellers. General newspapers are read mainly by the urban middle class. Magazines circulate among the elite. • Costs Preparing and disseminating information can be costly and difficult. Producing a video is expensive. Printing and distributing publications can also be costly. And there may be institutional barriers -- for instance, in persuading a television or radio station to produce and broadcast a program, or getting a newspaper or magazine to print a story. Not all types of agricultural technology are suited for widespread dissemination through the press and other mass media: • The suitability of a new crop variety, for instance, may depend on local conditions, such as soil types and weather patterns. Skilled technicians may be needed to advise farmers on whether to use the new seeds. • Some technologies may be suited only to limited areas, making the use of more widespread mass media inappropriate. • Some technologies are too complex to allow dissemination via the media. Extension personnel and farmers may have to undergo intensive training in order to use them correctly (as in integrated pest management, Stone 1992). • It may be undesirable to generate premature demand for a technology before the necessary inputs (e.g., seed, fertilizer, chemicals, vaccines, credit) are available. • The media can carry only a brief summary of many types of technology. Their primary use may thus be to raise awareness of a technology's existence rather than to teach how to use it. • The media should supplement rather than replace other channels. In order to apply technologies correctly in a wide range of situations, extension personnel will require considerably more in-depth and technical information than can be carried in a newspaper article. This information should continue to be supplied through training, technical publications, field testing, and other methods.
289 Nevertheless, many techniques could be disseminated more actively through the mass media. Examples are the use of conical woven bamboo containers as nests for chickens, the provision of water and salt to small ruminants under confined management, and post-harvest processing methods for food crops. These generally fall into the category of technology that is not important enough nationally to pass through the directorate-general, but is locally applicable and is not in conflict with official guidelines (Chapter 4). AARD has created or tested a large supply of such technologies, but they reach extension personnel only slowly. Improved use of the agricultural and other mass media could greatly improve the dissemination of these technologies. AARD publications The expense of travel and training courses, coupled with the large numbers and scattered distribution of extension personnel in Indonesia, mean that publications must continue to be the major conduit for research information reaching extension personnel. SMSs use AARD publications if they are familiar with them, if they are timely, and if they are close by. Efforts to increase use of these publications should therefore focus on these three characteristics. Several suggestions for increasing the amount of information SMSs obtain from AARD publications are given above. Let me repeat: perhaps the most important problem to be solved is the poor distribution of these publications. Print runs must be increased, copies distributed to district Dinas offices, and mailing list management improved. This does not mean that all AARD publications should be sent to district extension personnel, however. The more practically oriented publications such as Warta Litbang and technical guides should be sent to all district offices. Scientific journals are less useful to extension personnel because they are less easy to translate into extension recommendations. Whether books such as symposium proceedings are distributed so widely depends on the topic: a symposium on a basic research topic is less useful to extension personnel than is one on an applied topic. Much of AARD's publication output is in the form of scientific articles of limited use to extension or policy makers (but providing large numbers of promotion-supporting credit points to their authors). AARD institutes should review their publication policies to ensure that the extension audience is receiving due attention. It may be necessary to develop new publication types and to devote extra resources to institute communication units to enable them to serve the extension audience better.
290
Interpersonal communication This study has shown that interpersonal communication is an important means by which SMSs obtain information. In Chapters 9 and 10, this was represented by the "other specialists" source category. SMSs obtained information from their peers if the information these provided was easy to understand and use (Openness), and if the peers were close by (Proximity). Extension personnel should be encouraged to exchange information with each other and with others in the agricultural knowledge system -- particularly farmers, field agents, and researchers. While social relationships are important to put people at their ease, structured opportunities for interchange should be provided. These include meetings, exhibitions, training courses, joint research, and field tests. All these activities are currently being conducted by AARD and the extension agencies. Other possible methods not at present used are the secondment or exchange of staff, the provision of office space to extensionists, and other measures listed in the Familiarity section above. Implications of the FP SCORET model I undertook this study with a practical aim in mind: to understand researchextension linkages in Indonesia in order to propose ways of improving them. I chose Havelock's HELP SCORES model as a theoretical base since it promised to elucidate the reasons extension personnel use certain information sources rather than others. The model has indeed proved valuable in doing so. Nonetheless, a study of this nature is fraught with peril. Conceptual, theoretical, and operational pitfalls abound. Some can be foreseen and avoided; others will be stumbled into. Below are some of the conceptual and theoretical issues encountered in this study. Operationalization of concepts Several weaknesses are evident in the operationalization of the FP SCORET variables. Major among these are the poor face validity and low reliability of several of the indices used. In part this was due to my inadequate operationalization of Havelock's original HELP SCORES concepts. In part it may also be due to the multidimensionality of some of those concepts (Havelock 1969:11/20). It is clear that they must be more carefully conceptualized and operationalized if this approach to studying information flows is to prove fruitful.
291
Direction of causality A major flaw in the FP SCORET model is in the uncertain direction of causality between the FP SCORET variables and Information Flow. For some, the direction is reasonably clear: for instance, the source's Proximity to the receiver is more likely to affect the level of Information Flow than the other way round. For other variables, however, the case is less clear. Do I use a source because I am Familiar with it, or do I become Familiar with it as I use it? Causation appears at least circular here, and may be stronger from Information Flow to Familiarity than vice-versa. Even for the apparently clear case of Proximity above, it is possible to imagine a specialist keeping a book in her office or home because she gets information from it, rather than obtaining information from it because it happens to be close by. Libraries would not function if causality were all from Proximity to Information Flow, since no one would ever travel to a (relatively distant) library to borrow a book to take home to read. The practical implication of the direction of causality is in determining which variables should be manipulated in order to increase information flows. The suggestions given above assume that causality is in the hypothesized direction, from the FP SCORET variables to Information Flow, or is at least circular. This study cannot rule out the alternative possibility, however. Nevertheless, the hypothesized direction of causation is plausible, leading us to believe that attempts to influence Information Flow by manipulating the FP SCORET variables should meet with at least some success. Manipulating variables Some of the FP SCORET variables lend themselves to manipulation more easily than others. It is difficult and expensive to change the location of an experiment station (to increase its Proximity to its clients), change the structure of an organization (to increase Structure), or train 29,000 extension agents (to improve their Energy and Capacity). And such large changes are likely to cause as many problems as they solve. It is easier to change the format of an existing training course, or assign personnel to new roles within an existing organization. While sometimes major changes are called for, they should not be resorted to unless there is a clear and urgent need. If the system is already performing reasonably well -- as the research and extension systems in Indonesia seem to be doing (witness the gains in rice, soybean and chicken production), it is probably better to fine-tune them than to wield a hatchet.
292
Interrelationships among variables Another flaw is that the model fails to take into account the obvious interrelationships among the independent variables. Throughout, I have assumed that the FP SCORET variables influence Information Flow independently of one another. There is ample evidence that this is not the case: significant correlations among items making up different variables and among the indices themselves, and the failure of several indices to enter the multiple regression model despite significant simple correlations with Information Flow. Several theoretical reasons can be cited for these interrelationships. The "mere exposure" theory links Proximity to Familiarity. Role theory could link Structure and Energy. Havelock himself (1969:11/31, Havelock and Lingwood 1973:297) provides some brief ideas on how the variables might be causally related. With suitably conceptualized and operationalized variables, a path analysis could be used to test the proposed interrelationships among them. Such an analysis might also throw light on the direction of causality problem just described. A practical implication of the existence of causal relationships among the FP SCORET variables is that it may be possible to affect Information Flow indirectly by adjusting one or other of the variables. For instance, increasing the research-based content of a publication (Capacity) may encourage SMSs to view it as part of their job to use that publication (Structure). And increasing the Timeliness of a publication or training program may also raise its credibility (Capacity) in the minds of the extensionists. The general lack of strong negative correlations among the indices indicate that it may be possible to improve scores on several factors simultaneously without jeopardizing the levels of others. Information search and receptivity The respondents in this study appear to exert very little effort to obtain any information. This is not necessarily through any fault of their own, given the limited opportunity and funds they have for seeking information from outside. But the FP SCORET model lacks any measure of the ease with which information reaches the receiver. If such a variable were included, it might well prove to be a major influence on the amount of information obtained. The lack of such a measure also restricts the present FP SCORET to situations where information search is highly constrained. In Atkin's (1973) terms, it may apply better to information receptivity rather than information search, or indeed to his other categories of information yielding, rejection, and so forth. Nevertheless, highly constrained sources are typical of the situation of extension personnel in the developing world. The importance of Proximity and Familiarity in this study shows that it is imperative to ensure that information arrives on an extension specialist's desk if that information is to be used. This lesson should not be forgotten by research and extension organizations, many of which seem to rely on an easily broken oral chain to transmit messages up and down hierarchies and across vast distances.
293
Information flow: One-way or two-way? Throughout this study, I have treated Information Flow as if it were a one-way, source-to-receiver, movement. In reality, this is not so. Even in highly rigid, top-down systems, some feedback does filter back to the sources of information. In an agricultural knowledge system, such two-way flows are highly desirable. Can the FP SCORET model be extended to form a two-way model? Certainly. Many of the variables in Havelock's original model are inherently two-way in nature: Linkage, Openness, and Reward. This task is facilitated because the FP SCORET model uses characteristics of the relationship between source and receiver, not features of the source and receiver ass individuals. The difficulty comes in trying to operationalize such two-way concepts, particularly when dealing with non-personal sources. Missing variables As Havelock points out, the HELP SCORES formulation omits many important factors that may affect information exchange. Some of these are primacy ("being first"), status, and values. This study identified prolificacy and ease of acquisition as possible additions to the list. Atkin's notions of the costs and benefits of obtaining information from sources may be particularly useful. I omitted or changed other factors (Homophily, Empathy, Linkage) because they were not compatible with the need to compare personal and non-personal sources. Like other "laundry lists" or factors, (A VICTORY, HELP SCORES), FP SCORET is thus by no means exhaustive. Rather, it should be seen as a useful checklist that can produce practical suggestions for change. Personal and non-personal sources It is difficult to compare personal and non-personal sources directly. This study is no exception: including both in the study forced me to drop some concepts and alter others, and constrained the choice of question wordings. Using the receiver perspective helped overcome these difficulties: receivers are in the best position to judge the amount of information they acquire from various sources, and are able to compare widely different sources on the same scale -- an impossible task for more objective measures. The personal/non-personal dichotomy is only an extreme version of the dilemma facing those who wish to compare information flows among source types. Similar problems face us when we try to compare the amount of information obtained from two non-personal media, such as television and newspapers, or even TV and radio. Each source type has characteristics that make it unique, and that may affect the amount of information we obtain from it. We have yet to identify suitable measures. Meanwhile, an index approach based on the receiver's opinions and memory seems the best route.
294
Information flow as a relational characteristic I have conceptualized information flow and its antecedents as features of the relationship between a source and a receiver. As stated above, this approach could be extended to embrace two-way communication. The important aspect of this view is that it is relational, viewing each source-receiver pair as a dyad. This makes it compatible with other relational approaches, such as network analysis and coorientation. Links with these approaches should be explored further. Conclusion Poor agricultural research-extension links are a problem in Indonesia. This study has outlined some of the problems involved and suggested some solutions to these. The impression one has is that research-extension links are poor because there has not been any systematic attempt to improve them. The AICs have filled a major gap in the agricultural knowledge system by developing materials for field agents. But these institutions are too under-funded to be able to fulfill even this role adequately. They thus have not aggressively sought out new research findings from AARD and disseminated these to their audiences. Commitment from AARD has also been lacking. AARD institutes' communication activities have been skewed toward serving the needs of scientists and policy makers rather than extension. While many useful publications have been published, these have been produced in too small numbers and have received too restricted a distribution to have major impact. This study did not investigate the relevance of AARD's research findings to Indonesia's farmers. The existence of a pool of relevant information was assumed at the outset. Rather, I sought to discover whether the information that has been generated is reaching extension personnel. The study shows that research information does reach extension personnel, but it does so very slowly. Indonesia's extensionists are eager to obtain research-based information. A large body of research information exists, but it is not written in a form useful for extensionists, and it is not distributed effectively to them. Nevertheless, AARD institutes can take several simple measures to boost the amount of research information flowing to extension via publications. While this will require increased funding, the amounts involved are modest, and the potential payoff is large. This study does not propose radical restructuring of the Indonesian research and extension system. I believe the changes I suggest will help these institutions serve farmers better and are pragmatic in that they build on existing institutions and activities rather than replacing them with new ones. Indonesian authorities increasingly recognize that improved research-extension linkages are key to agricultural development. Recent moves within AARD to improve
295 research-extension links should have a significant impact on the spread of agricultural technologies in Indonesia. Finally, with 41% of variance in Information Flow explained, the FP SCORET model has proved useful in predicting information flows. Further research will demonstrate whether it can be developed into a theoretical as well as a diagnostic tool. While this study has focused on the information sources of agricultural extension specialists in Indonesia, the FP SCORET model could be applied to other situations also. It may prove useful in explaining why people in general use certain sources for any type of information.
298
REFERENCES AARD. 1984a. An evaluation of the palawija crop research program of AARD. Volume 1 - Main report. International Service for National Agricultural Research, The Hague/Agency for Agricultural Research and Development, Indonesia. AARD. 1984b. An evaluation of the fruit and vegetable crop research program of AARD. Volume 1 - Main report. International Service for National Agricultural Research, The Hague/Agency for Agricultural Research and Development, Indonesia AARD. 1985a. An evaluation of the library, information and publication activities of AARD. Agency for Agricultural Research and Development, Jakarta, Indonesia. AARD. 1985b. This is AARD. Agency for Agricultural Research and Development, Jakarta, Indonesia. AARD. 1986. Indonesian farming systems research and development: The food crop subsystem. Central Research Institute for Food Crops, Bogor, Indonesia. AARD. 1987. Five years of agricultural research (1981-1986): Its contribution to agricultural development in Indonesia. Agency for Agricultural Research and Development, Jakarta, Indonesia. AARD. 1988. Indonesia and its food crops agriculture. Agency for Agricultural Research and Development, Indonesia. Abbas, Syamsuddin. 1991. Mekanisme dan kelembagaan penyuluhan pertanian. Paper presented at the Temu Teknis Keterkaitan Penelitian-Penyuluhan, 13-21 February 1991, Pusat Perpustakaan dan Komunikasi Penelitian, Bogor, Indonesia. Abbas, Syamsuddin, Prabowo Tjitropranoto, and Yusuf Yakub. 1989. Hubungan antara penyuluhan dan penelitian dalam menunjang pelestarian swasembada pangan. Pp. 45-74 in: Syam, Mahyuddin, M. Ismunadji, and Adi Widjono (eds.). Risalah simposium II penelitian tanaman pangan. Vol. 1. Ciloto, 21-23 March 1988. Pusat Penelitian dan Pengembangan Tanaman Pangan, Bogor, Indonesia. Abbott, Eric A. 1990. Personal communication. Department of Journalism and Mass Communication, Iowa State University, Ames, IA 50011. Abdul Adjid, Dudung. 1990. Peranan kelompok dalam menggerakkan partisipasi masyarakat dalam pembangunan. Paper presented at the Seminar Sehari Peranan Penyuluhan Pembangunan Menyongsong Era Tinggal Landas. 25 September 1990, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Abdurachman, Adi, Soleh Sukmana, and James H. French. 1991. Compilation of a technology inventory: A framework for organizing what we know about the performance of upland conservation and production systems. Paper presented at
299 the International Workshop on Conservation Policies for Sustainable Hillslope Farming, 11-15 March, Solo, Indonesia. Upland Agriculture and Conservation Project, Agency for Agricultural Research and Development, Indonesia. Agudelo, Luis Alfonso, and David Kaimowitz. 1989. Institutional linkages for different types of agricultural technologies: Rice in the eastern plains of Colombia. Linkages Discussion Paper 1, International Service for National Agricultural Research (ISNAR), The Hague, Netherlands. Albrecht, Hartmut, Herbert Bergmann, Georg Diederich, Eberhard Großer, Volker Hoffmann, Peter Keller, Gerhard Payr, and Rolf Sülzer. 1989. Agricultural extension. Volume 1: Basic concepts and methods. Rural Development Series, Deutsche Gesellschaft für Technische Zusammenarbeit, Eschborn, Germany. Albrecht, Hartmut, Herbert Bergmann, Georg Diederich, Eberhard Großer, Volker Hoffmann, Peter Keller, Gerhard Payr, and Rolf Sülzer. 1990. Agricultural extension. Volume 2: Examples and background material. Rural Development Series, Deutsche Gesellschaft für Technische Zusammenarbeit, Eschborn, Germany. Allen, T. J. and A. R. Fustfeld. 1975. Research laboratory architecture and the structuring of communications. R and D Management 5(2):153-164. Andyantoro, S., A. Hasanuddin, and J. H. French. 1989. Hasil penelitian dan penyebarannya: Balai Penelitian Tanaman Pangan Maros, 1983-88. Paper presented at the Temus Tugas Peneliti-Penyuluh, Balai Penelitian Budidaya Pantai, Maros, 6-9 February 1989. Anyawu, Alphonsus Chijioke. 1982. Communication behavior of professional disseminators of scientific agricultural knowledge: A study of the Iowa Cooperative Extension Service. Ph.D. dissertation, Iowa State University, Ames, IA. Arnon, I. 1989. Agricultural research and technology transfer. Elsevier Applied Science, London. Ashton-Tate. 1986. dBase III+, version 1.1. Computer database program. Atkin, Charles. 1973. Instrumental utilities and information seeking. Pp. 205-242 in: Clarke, Peter (ed.). New models for mass communication research. Sage, Beverly Hills. Atwood, R. et al. 1982. Children's realities in television viewing: Exploring situational information seeking. Pp. 605-626 in: Burgoon, M. (ed.). Communication Yearbook 6. Sage, Beverly Hills, CA. Atwood, R. and B. Dervin. 1982. Challenges to sociocultural predictors of information seeking: A test of race versus situation movement state. Pp. 549-569 in: Burgoon, M. (ed.). Communication Yearbook 5. Transaction Books, New Brunswick, NJ. (abstract).
300 Axinn, George H., and Sudhakar Thorat. 1972. Modernizing world agriculture: A comparative study of agricultural education systems. Praeger, New York. Azis, Amin. 1990. Peranan masyarakat sebagai sistem sosial dan pelaku pembangunan. Paper presented at the Seminar Sehari Peranan Penyuluhan Pembangunan Menyongsong Era Tinggal Landas. 25 September 1990, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Badan Litbang Pertanian. 1991. Lokakarya pemantapan rencana penelitian lingkup Badan Penelitian dan Pengembangan Pertanian. Pp 45-49, Rapim Departemen Pertanian, August 1991. Baharsjah, Sjarifuddin. 1985. Improving research-extension linkages: The Indonesian experience. Pp. 28-33 in: Cernea, Michael M., John K. Coulter, and John F. A. Russell (eds.). Research-extension-farmer: A two-way continuum for agricultural development. World Bank, Washington, DC. Bahraini. 1984. Hubungan dinamika kelompok tani hamparan dan penerapan teknologi padi sawah pada peserta dan non peserta intensifikasi khusus. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Balitnak. 1991. Alamat pengiriman hasil-hasil penelitian. Unpublished list of addresses for mailing research findings of the Research Institute for Animal Production. Balai Penelitian Ternak, Ciawi, Bogor, Indonesia. Balitvet. 1991. Daftar pengiriman publikasi Balitvet. Unpublished list of addresses for mailing publications of the Research Institute for Animal Diseases. Balai Penelitian Veteriner, Bogor, Indonesia. Batoa, Mahmud Hamundu. 1985. Perceptions, attitudes and levels of participation of farmer-members of Koperasi Unit Desa in Southeast Sulawesi, Indonesia. M.S. thesis, University of the Philippines at Los Baños, Philippines. Baxter, Michael, and Wilfried Thalwitz. 1985. National policies and institutional constraints to linking research and extension in Asia. Pp. 42-48 in: Cernea, Michael M., John K. Coulter and John F. A. Russell (eds.). 1985. Research-extension-farmer: A World Bank and UNDP Symposium. Based on an international workshop convened by the World Bank and the United Nations Development Programme in Denpasar, Indonesia, in March 1984. The World Bank, Washington, DC. Beal, George M., and Peter Meehan. 1986. Communication in knowledge production, dissemination, and utilization. Pp. 135-150 in: Beal, George M., Wimal Dissanayake, and Sumiye Konoshima (eds.). Knowledge generation, exchange, and utilization. Westview, Boulder, CO. Bedell, Jeffrey R., John C. Ward, Jr., Robert P. Archer, and M. Kirk Stokes. 1985. An empirical evaluation of a model of knowledge utilization. Evaluation Review 9(2):109-126. Benor, Daniel, and Michael Baxter. 1984. Training and visit extension. World Bank, Washington, DC.
301 Benor, Daniel, and James Q. Harrison. 1977. Agricultural extension: The training and visit system. World Bank, Washington, DC. Berlo, David K. 1960. The process of communication: An introduction to theory and practice. Holt, Rinehart and Winston, New York. BIP Ciawi. 1990. Daftar judul bahan informasi pertanian BIP/PIP seluruh Indonesia tahun 1984/1985-1990/1991. Unpublished printout of database containing titles of media produced by Agricultural Information Centers. Balai Informasi Pertanian Ciawi, Bogor, Indonesia. Biro Humas Deptan. 1991. Buku kerja Deptan (Ministry of Agriculture diary). Biro Hubungan Masyarakat, Ministry of Agriculture, Jakarta. BPLPP/Pustaka. 1983. Pedoman penggunaan sistem Agdex. Badan Pendidikan, Latihan dan Penyuluhan Pertanian and Pusat Perpustakaan Biologi dan Pertanian, Bogor, Indonesia. Brown, Lawrence A. 1981. Innovation diffusion: A new perspective. Methuen, London. Busch, Lawrence, and William B. Lacey. 1983. Science, agriculture, and the politics of research. Westview, Boulder. Capon, N. and M. Burke. 1980. Individual product class, and task-related factors in consumer information processing. Journal of Consumer Research 7(3):314-326. (abstract) Cernea, Michael M., John K. Coulter and John F. A. Russell (eds.). 1985. Research-extension-farmer: A World Bank and UNDP Symposium. Based on an international workshop convened by the World Bank and the United Nations Development Programme in Denpasar, Indonesia, in March 1984. The World Bank, Washington, DC. Chaffee, Steven and Diana Mutz. 1989. Comparing mediated and interpersonal communication data. Pp. 19-43 in: Robert P. Hawkins, John M. Wiemann, and Suzanne Pingree (eds.). Advancing communication science: Merging mass and interpersonal processes. Sage Annual Reviews of Communication Research 16. Sage, Newbury Park, CA. Compton, J. Lin. 1984. Linking scientist and farmer: Re-thinking extension's role. Pp. 79-84 in: Drosdoff, M. (ed.). World food issues. 2nd. ed. Program in International Agriculture, Cornell University, Ithaca, NY. Compton, J. Lin. 1989. The integration of research and extension. Pp. 113-136 in: Compton, J. Lin (ed.). The transformation of international agricultural research and development. Lynne Rienner, Boulder, CO. Compton, J. Lin. 1992. Personal communication. Department of Continuing and Vocational Education, University of Wisconsin-Madison. Cook, L. J., and R. H. Wharton. 1984(?). Resources for research in animal health and
302 production. P3T - A case study. Pp. 241-257 in: Animal health and nutrition in the tropics. Research for development seminar one. James Cook University of North Queensland, Townsville/Australian Development Assistance Bureau. Coulter, John K. 1983. The interdependence of research and extension: A comment. Pp. 49-52 in: Cernea, Michael M., John K. Coulter, and John F. A. Russell (eds.). Agricultural extension by training and visit: The Asian experience. The World Bank, Washington, DC. CRIAS. Undated. Central Research Institute for Animal Science. Brochure. Central Research Institute for Animal Science, Bogor, Indonesia. CRIFC. 1986. Central Research Institute for Food Crops. Brochure. Central Research Institute for Food Crops, Bogor, Indonesia. Curvo, Placido Flaviano. 1983. Analysis of the informal communication patterns among professionals in agricultural research and extension service activities. Ph.D. dissertation, Case Western Reserve University. (abstract). Czepiel, John A. 1974. Word-of-mouth processes in the diffusion of a major technological innovation. Journal of Marketing Research 1(2):172-180. Damayani, Maya. 1988. Penyelenggaraan penyuluhan pertanian dengan sistem kerja latihan dan kunjungan dalam upaya pembinaan kelompok tani Supra Insus (Studi kasus pada kelompok tani Mulus dan Harapan Jaya, desa Karangjaya, kecamatan Cikampek, kabupaten Karawang). Sarjana thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Davis, H. R. 1971. A checklist for change. In: A manual for research utilization. Government Printing Office, Washington, DC. De Datta, Surajit K. 1981. Principles and practices of rice production. Wiley, New York. De Sola Pool, Ithiel, Hiroshi Inose, Nozumo Takasaki and Roger Hurwitz. 1984. Communication flows: A census in the United States and Japan. NorthHolland, Amsterdam. Deacon, Ruth and Francille Firebaugh. 1988. Family resource management. Allyn and Bacon, Boston, MA. Dervin, B. et al. 1980. The human side of information: An exploration in a health communication context. Pp. 591-608 in: Nimmo, D. (ed.). Communication Yearbook 4. Transaction Books, New Brunswick, NJ. (abstract) Dhandhanin. M. 1984. An analysis of organizational structure, role perceptions, and inter-role relationships affecting the functioning and effectiveness of the agricultural extension education system in Thailand. Ph.D. dissertation, Cornell University. Direktorat Penyuluhan RRL. 1990. Peranan penyuluhan pembangunan dalam pelestarian sumberdaya alam. Paper presented at the Seminar Sehari Peranan Penyuluhan
303 Pembangunan Menyongsong Era Tinggal Landas. 25 September 1990, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Dissanayake, Wimal. 1986. Communication models and knowledge generation, dissemination, and utilization activities: A historical survey. Pp. 61-75 in: Beal, George M., Wimal Dissanayake, and Sumiye Konoshima. Knowledge generation, exchange, and utilization. Westview, Boulder, CO. Donohew, Lewis, Leonard Tipton and Roger Haney. 1980. Analysis of information-seeking strategies. Ch. 21 in: Wilhoit, G. and H. deBock (eds.). Mass Communication Review Yearbook I. Sage, Beverly Hills. Donohew, L., P. Palmgreen and J. Duncan. 1980. An activation model of information exposure. Communication Monographs 47(4):295-303. (abstract). Ekpere, Johnson, and Isiaka Idowu. 1990. Managing the links between research and technology transfer: The case of the Agricultural Extension-Research Liaison Services in Nigeria. Linkages Discussion Paper 6, International Service for National Agricultural Research (ISNAR), The Hague, Netherlands. El-Zoobi, A. 1988. Linkages between agricultural research and extension: The situation in Pakistan. Pp. 112-121 in: Food and Agriculture Organization of the United Nations; United Nations Educational, Scientific and Cultural Organization; and International Labor Organisation. 1986-87 training for agriculture and rural development. FAO, Rome. Engel, Paul. 1990. The impact of improved institutional coordination of agricultural performance: The case of the Nariño Highlands in Colombia. Linkages Discussion Paper 4, International Service for National Agricultural Research (ISNAR), The Hague, Netherlands. Eponou, Thomas. 1990. Informal linkage mechanisms and technology transfer: The PACO Project in Côte d'Ivoire. Linkages Discussion Paper 5e, International Service for National Agricultural Research (ISNAR), The Hague, Netherlands. Eriyatno. 1988. Information systems analysis and design for extension workers. Paper presented at the Farming Systems Research/Extension Symposium, University of Arkansas, Fayetteville, October 1988. Ewell, Peter T. 1989. Linkages between on-farm research and extension in nine countries. OFCOR Comparative Study 4, International Service for National Agricultural Research, The Hague. Also published as Links between on-farm research and extension in nine countries, Pp. 151-196 in: Kaimowitz, David (ed.). 1990. Making the link: Agricultural research and technology transfer in developing countries. Westview, Boulder, CO. FAO. 1969. FAO production yearbook 23. Food and Agriculture Organization of the United Nations, Rome. FAO. 1991. FAO production yearbook 45. Food and Agriculture Organization of the United Nations, Rome.
304 Feder, Gershon, Lawrence J. Lau, and Roger H. Slade. 1985. The impact of agricultural extension: A case study of the training and visit system in Haryana, India. World Bank Staff Working Papers 756, World Bank, Washington, DC. Feder, Gershon, and Roger H. Slade. 1983. Experiences with the monitoring and evaluation of training and visit extension in India. World Bank Staff Working Papers 595, World Bank, Washington, DC. Feller, Irwin. 1986. Research and technology transfer linkages in American agriculture. Chapter 21 in: Busch, Lawrence, and William B. Lacey. 1983. Science, agriculture, and the politics of research. Westview, Boulder. Feller, Irwin, Lynne Kaltreider, Patrick Madden, Dan Moore, and Laura Sims. 1984. The agricultural technology delivery system: A study of the transfer of agricultural and food-related technologies. Volume 3: Review of previous case studies. Institute for Policy Research and Evaluation, The Pennsylvania State University, University Park, PA. Fett, John H., and Paul Mundy. 1990. Disseminating annual crop practice recommendations via supplements in weekly agricultural newspapers. Paper presented at the annual convention of the Agricultural Communicators in Education, Minneapolis, MN, 15-18 July 1990. FFTC. 1981. Support communication for rural development programs. Food and Fertilizer Technology Center, Taipei, Taiwan. Fivawo, Anna Wuyts. 1987a. Agricultural research-technology transfer system: The People's Republic of China. International Service for National Agricultural Research, The Hague, Netherlands (draft). Fivawo, Anna Wuyts. 1987b. Agricultural research-technology transfer system: Fiji. International Service for National Agricultural Research, The Hague, Netherlands (draft). Fivawo, Anna Wuyts. 1987c. Agricultural research-technology transfer system: Kenya. International Service for National Agricultural Research, The Hague, Netherlands (draft). Flinn, A. J., and J. F. Jamias. 1984. Socio-cultural factors influencing professional communication between UPLB and IRRI researchers. Devcom Quarterly 2(1):110. Gidley, V. N. 1977. Towards better liaison: A survey of communication between agricultural research and extension. Journal of the Australian Institute of Agricultural Science, Sept.-Dec. 1977:110-116. Glaser, Edward M., Harold H. Abelson and Kathalee N. Garrison. 1983. Putting knowledge to use: Facilitating the diffusion of knowledge and the implementation of planned change. Jossey-Bass, San Francisco. Granovetter, M. S. 1973. The strength of weak ties. American Journal of Sociology
305 73:1361-1380. Greenberg, Bradley S. 1964. Diffusion of news of the Kennedy assassination. Public Opinion Quarterly 28(2): 225-232. Grist, D. H. 1986. Rice. 6th edition. Longman, London. Grunig, James E. 1983. Communication behaviors and attitudes of environmental publics: Two studies. Journalism Monographs 81. Hadiwigeno, Soetatwo. 1991. Program penelitian dan pengembangan pertanian: Tantangan dan peluang untuk memperkuat keterkaitan penelitian-penyuluhan-petani. Paper presented at the Temu Teknis Keterkaitan Penelitian-Penyuluhan, 13-21 February 1991, Pusat Perpustakaan dan Komunikasi Penelitian, Bogor, Indonesia. Haryani, Mimi, and Rusmini Mulyati. 1990. Pembuatan database "mailing list." Pp. 91-96 in: Pustaka. Prosiding latihan komunikasi ilmiah bidang pertanian. Training course materials, Bogor, 8-21 Feb. 1990. Pusat Perpustakaan Pertanian dan Biologi, Bogor, Indonesia. Harun, Rochajat. 1987. Extension workers' perception of information media published by the Agricultural Information Center (AIC) Lembang, West Java, Indonesia. Ph.D. dissertation, University of the Philippines at Los Baños, Philippines. Havelock, Ronald G. 1969. Planning for innovation through dissemination and utilization of knowledge. Center for Research on Utilization of Scientific Knowledge. University of Michigan, Ann Arbor, MI. Havelock, Ronald G. 1986a. Modeling the knowledge system. Pp. 77-104 in: Beal, George M., Wimal Dissanayake, and Sumiye Konoshima. Knowledge generation, exchange, and utilization. Westview, Boulder, CO. Havelock, Ronald G. 1986b. Linkage: key to understanding the knowledge system. Pp. 211-243 in: Beal, George M., Wimal Dissanayake, and Sumiye Konoshima. Knowledge generation, exchange, and utilization. Westview, Boulder, CO. Havelock, Ronald G. and David A. Lingwood. 1973. R&D utilization strategies and functions: An analytical comparison of four systems. Center for Research on Utilization of Scientific Knowledge, Institute for Social Research, University of Michigan, Ann Arbor, MI. Hildreth, R. J. 1965. Tensions between research and extension workers: Three hypotheses. Journal of Farm Economics 47(3):838-840. Hirshchmann, E. C. 1981. Social and cognitive influences on information exposure: A path analysis. Journal of Communication 31(1):76-87. (abstract) Hoffman, Eric and Paul M. Roman. 1984. Information diffusion in the implementation of innovation process. Communication Research 11(1): 117-140. (abstract) Holian, Basruddin. 1990. Peranan kontak tani dalam pelaksanaan penyuluhan sistem
306 kerja latihan dan kunjungan (Laku): Suatu kasus di kecamatan Balaraja, kabupaten Tangerang, Jawa Barat. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Hornik, Robert C. 1988. Development communication: Information, agriculture, and nutrition in the Third World. Longman, New York. Hubeis, Aida Vitayala Sjafri. 1987. Memantapkan penyuluhan pertanian di Indonesia. Pp. 27-32 in: Prosiding kongres penyuluhan pertanian 1987. Subang, 4-6 July 1987. Perhiptani. Hussein, Shakir. 1986. An analysis of the agricultural knowledge system of Indonesia. Ph.D. dissertation, Cornell University, Ithaca, NY. IARDJ. 1986. Development in Indonesian agriculture. Indonesian Agricultural Research and Development Journal 8(1):26-31. IDRC. 1986. Cropping systems research in Indonesia. Pp. 158-172 in: International Development Research Centre. With our own hands. Research for Third World development: Canada's contribution through the International Development Research Centre, 1970-1985. IDRC, Ottawa. INTERPAKS. 1983-1991. INTERPAKS interchange. University of Illinois at UrbanaChampaign. Newsletter. ISNAR. 1981. The Agency for Agricultural Research and Development of Indonesia, reviewed by ISNAR. International Service for National Agricultural Research, The Hague. Jahi, Amri. 1991. Strategi komunikasi untuk meningkatkan keterkaitan penelitian dan penyuluhan. Paper presented at the Temu Teknis Keterkaitan PenelitianPenyuluhan, 13-21 February 1991, Pusat Perpustakaan dan Komunikasi Penelitian, Bogor, Indonesia. Kaimowitz, David. 1989. Placing agricultural research and technology transfer in one organization: Two experiences from Colombia. Linkages Discussion Paper 3, International Service for National Agricultural Research (ISNAR), The Hague, Netherlands. Kaimowitz, David (ed.). 1990. Making the link: Agricultural research and technology transfer in developing countries. Westview, Boulder, CO. Kaimowitz, David. 1991. The evolution of links between extension and research in developing countries. Pp. 101-112 in: Rivera, William M., and Daniel J. Gustafson (eds.). Agricultural extension: Worldwide institutional evolution and forces for change. Elsevier, Amsterdam. Kaimowitz, David, Monteze Snyder, and Paul Engel. 1990. A conceptual framework for studying the links between agricultural research and technology transfer in developing countries. Pp. 227-269 in: Kaimowitz, David (ed.). Making the link: Agricultural research and technology transfer in developing countries.
307 Westview, Boulder, CO. Kanwil Deptan Sulsel. 1990. Surat Keputusan Kepala Kantor Wilayah Departemen Pertanian Propinsi Sulawesi Selatan no. OT 22/33/SK/29/XII/90 tentang pembentukan tim kerja Sambung Litluh. Kantor Wilayah Departemen Pertanian Propinsi Sulawesi Selatan, Ujung Pandang. Unpublished administrative decision. Katz, Elihu, Jay G. Blumler, and Michael Gurevitch. 1974. Utilization of mass communication by the individual. Pp. 19-32 in: Blumler, J. G., and E. Katz (eds.). The uses of mass communication: Current perspectives. Sage Annual Reviews of Communication Research, Vol. 3. Sage, Beverly Hills, CA. Klauss, Rudi and Bernard M. Bass. 1982. Interpersonal communication in organizations. Academic Press, New York. Krippendorff, Klaus. 1987. Paradigms for communication and development with emphasis on autopoiesis. Pp. 189-208 in: Kincaid, D. Lawrence (ed.). Communication theory: Eastern and Western perspectives. Academic Press, San Diego, CA. Lakoh, A. K. 1986. Organizational behavior and program effectiveness: Analysis of the Northern Agricultural Development Project Extension Delivery. Ph.D. dissertation, Cornell University. Lambert, Zarrel V., and Richard M. Durand. 1977. Purchase information acquisition and cognitive style. Journal of Psychology: The General Field of Psychology 97(5):3-13. Lionberger, Herbert F. 1986. Toward an idealized systems model for generating and utilizing information in modernizing societies. Pp. 105-134 in: Beal, George M., Wimal Dissanayake, and Sumiye Konoshima (eds.). Knowledge generation, exchange, and utilization. Westview, Boulder, CO. Lionberger, Herbert F., and H. C. Chang. 1970. Farm information for modernizing agriculture: The Taiwan system. Praeger, New York. Lionberger, Herbert F., and H. C. Chang. 1981. Development and delivery of scientific farm information: The Taiwan system as an alternative to land grant universities-U.S. style. Chapter 12 in: Crouch, Bruce R., and Shankariah Chamala (eds.). Extension education and rural development. Vol. 1: International experience in communication and innovation. Wiley, New York. Lionberger, Herbert F., and Paul H. Gwin. 1991. Technology transfer from researchers to users: A textbook of successful research extension strategies used to develop agriculture. University Extension publication MX381, University of Missouri, Columbia, MO. Lionberger, Herbert F., LaVern A. Pope, and B. Anji Reddy. 1979. Information development and flow: A study of the communication behavior of social scientists in a U.S. Land Grant and two Taiwan universities. Research Bulletin 1033, University of Missouri-Columbia, Agricultural Experiment Station, Columbia, MO.
308 Lotus Development Corporation. 1991. Lotus 1-2-3, release 2.3. Computer spreadsheet program. Ludgate, Patrick J. (ed.). 1989. Kumpulan peragaan dalam rangka penelitian ternak kambing dan domba. 2nd ed. Balai Penelitian Ternak/Small RuminantCollaborative Research Support Program, Bogor, Indonesia. Ludgate, Patrick, Atien Priyanti, and Tjeppy D. Soedjana. 1990. Women's participation in agricultural extension services: A Southeast Asian perspective. Pp. 260-65 in: Resource utilization for livestock production in Malaysia. Proceedings of the 13th annual conference of the Malaysia Society of Animal Production, Malacca, Malaysia, 6-8 March 1990. Malaysian Society of Animal Production, Serdang/Malaysian Agricultural Research and Development Institute, Kuala Lumpur. (Also published as Small Ruminant-Collaborative Research Support Program Working Paper 108, Balai Penelitian Ternak, Bogor. Winrock International/University of Missouri, 1989.) Ludgate, Patrick J., and Atien Priyanti. 1988. Small ruminant farmers participating in farming systems research and extension in West Java, Indonesia. Paper presented at the Farming Systems Research/Extension Symposium, University of Arkansas, Fayetteville, October 1988. Lupanga, Ildefons Joseph. 1986. Linking research and extension in Tanzania: Some communication behaviors of researchers and extension workers. Ph.D. dissertation, Cornell University, Ithaca, NY. (abstract). Maguire, C. and R. Kench. 1984. The introduction and diffusion of technological innovation in industry: An information search perspective. Pp. 163-204 in: B. Dervin and M. J. Voigt (eds.). Progress in communication sciences 4. Ablex, Norwood, NJ. (abstract). Malik, W. H. 1988. An analysis of the agricultural knowledge system in Pakistan. Ph.D. dissertation, Cornell University. Mansur, Hana. 1989. Pemanfaatan publikasi Badan Litbang Pertanian oleh penyuluh pertanian spesialis (PPS) tanaman pangan di Sulawesi Selatan. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. McCall, Chester H., Jr. 1982. Sampling and statistics handbook for research. Iowa State University Press, Ames, IA. McCorkle, Constance M., and Donald L. Esslinger. 1992. Communicating project results: A model from an international R&D programme. Journal of Development Communication 1(3):53-68. McGuire, William J. 1974. Psychological motives and communication gratification. Pp. 167-196 in: Blumler, J., and E. Katz (eds.). The uses of mass communication: Current perspectives on gratifications research. Sage, Beverly Hills, CA. McLeod, Jack M., and Steven Chaffee. 1973. Interpersonal approaches to communication research. American Behavioral Scientist 16(4):465-500.
309 Meehan, Peter M., and George M. Beal. 1977. Knowledge production and utilization: A general model, third approximation. Sociology Report 138, Dept. of Sociology and Anthropology, Iowa State University, Ames, IA. Midgley, D. F. 1983. Patterns of interpersonal information seeking for the purchase of a symbolic product. Journal of Marketing Research 20(1):74-83. (abstract). Monge, Peter R. and Noshir S. Contractor. 1988. Communication networks: Measurement techniques. Pp. 107-134 in: Tardy, Charles H. (ed.). A handbook for the study of human communication: Methods and instruments for observing, measuring, and assessing communication processes. Ablex, Norwood, NJ. Morgan, Gareth. Images of organization. Sage, Newbury Park, CA. MORIF/AARP. Undated (1991?). Sambung Litluh: A summary of experience gained in preparation of research fact sheets. Maros Research Institute for Food Crops/Applied Agricultural Research Project - Phase II. Brochure. Mundy, Paul G. W. 1989. Communication linkages between researchers and extension specialists at Iowa State University. M.S. thesis, Iowa State University, Ames, IA. Mundy, Paul, Hadi Budiman, and Marwan Rangkuti. 1991. Livestock research-extension links faces challenges. Indonesian Small Ruminant Network Newsletter 2(2):23. Mundy, Paul, Mahyuddin Syam, Marwan Rangkuti, and Hadi Budiman. 1992. Information sources of agricultural extension specialists in Indonesia. Paper presented at the 1992 annual conference of the Association for International Agricultural and Extension Education, Columbus, OH, 12-16 May 1992. Nagel, Uwe Jens. 1980. Institutionalization of knowledge flows: An analysis of the extension role of two agricultural universities in India. Zeitschrift für ausländische Landwirtschaft: Materialsammlung Heft No. 30. DLG-Verlag, Franfurt (Main), Germany. Nestel, Barry. 1985. Indonesia and the CGIAR centers. CGIAR study paper 10. World Bank, Washington, DC. Neter, John, William Wasserman, and Michael H. Kutner. 1983. Applied linear regression models. Irwin, Homewood IL. Martínez Norgueira, Roberto. 1990. The effect of changes in state policy and organization on agricultural research and extension links: A Latin American perspective. Pp. 75-108 in: Kaimowitz, David (ed.). 1990. Making the link: Agricultural research and technology transfer in developing countries. Westview, Boulder, CO. NIA and DGRD (National Institute of Administration and Directorate General of Rural Development). 1982. Delivery system in support of small farmers in the context of
310 rural development: Indonesia. CIRDAP study series 20. Center on Integrated Rural Development for Asia and the Pacific, Kotbari, Comilla, Bangladesh. Norusis, Marija J./SPSS Inc. 1990. SPSS/PC+ 4.0 for the IBM PC/XT/AT and PS/2. Computer statistical data analysis program. Nur, Andri Priawijaya. 1986. Penyelenggaraan kegiatan penyuluhan pertanian melalui sistem kerja Laku: Studi kasus tiga BPP di wilayah Dinas Pertanian Tanaman Pangan kabupaten daerah tingkat II Bogor, Jawa Barat. Sarjana thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. O'Keefe, Garrett, and Barbara Sulanowski. 1991. How we use the telephone: Informational, instrumental and social perspectives. Paper presented to the Midwest Association for Public Opinion Research Annual Conference, November 1991, Chicago. Ortiz, Ramiro, Sergio Ruano, Horacio Juárez, Francisco Olivet, and Adlai Meneses. 1991. A new model for technology transfer in Guatemala: Closing the gap between research and extension. OFCOR Discussion Paper 2, International Service for National Agricultural Research (ISNAR), The Hague, Netherlands. Padmanagara, Salmon. 1985. The joint formulation of extension messages by research and extension staff in Indonesia. Pp. 136-143 in: Cernea, Michael M., John K. Coulter, and John F. A. Russell (eds.). Research-extension-farmer: A two-way continuum for agricultural development. World Bank, Washington, DC. Palmgreen Philip, and J. D. Rayburn II. 1979. Uses and gratifications and exposure to public television: A discrepancy approach. Communication Research 6(2):155180. Palmgreen, Philip and J. D. Rayburn II. 1982. Gratifications sought and media exposure: An expectancy value model. Communication Research 9(4):561-580. Pfeffer, Jeffrey. 1982. Organizations and organization theory. Ballinger, Cambridge, MA. Pickering, Donald C. 1983. Agricultural extension: A tool for rural development. Pp 313 in: Cernea, Michael M., John K. Coulter, and John F. A. Russell (eds.). Agricultural extension by training and visit: The Asian experience. The World Bank, Washington, DC. Puslitbangnak. 1991. Unpublished list of addresses for mailing publications of the Central Research Institute for Animal Sciences, Bogor, Indonesia. Puslitbangnak. Undated. Pusat Penelitian dan Pengembangan Peternakan. Brochure. Central Research Institute for Animal Sciences, Bogor, Indonesia. Puslitbangtan. 1990. Publikasi Pusat Penelitian dan Pengembangan Tanaman Pangan. Publications list of the Central Research Institute for Food Crops, Bogor, Indonesia. Puslitbangtan. 1991. Unpublished list of addresses for mailing publications of the Central
311 Research Institute for Food Crops, Bogor, Indonesia. Pustaka. 1990a. Prosiding temu tugas penelitian-penyuluhan pertanian. Pusat Perpustakaan Pertanian dan Biologi, Bogor, Indonesia. Pustaka. 1990b. Prosiding latihan komunikasi ilmiah bidang pertanian. Training course materials, Bogor, 8-21 Feb. 1990. Pusat Perpustakaan Pertanian dan Biologi, Bogor, Indonesia. Pustakom. 1990. Teknologi/informasi yang diperlukan oleh beberapa wilayah (Unpublished list of technologies and information topics required by several provinces in Indonesia). Pusat Perpustakaan dan Komunikasi Penelitian, Bogor, Indonesia. Pustakom. 1991a. Prosiding lokakarya komunikasi penelitian pertanian untuk memperkuat keterkaitan penelitian-penyuluhan. Pusat Perpustakaan Pertanian dan Komunikasi Penelitian, Bogor, Indonesia. Pustakom. 1991b. Hasil diskusi kelompok, Temu Teknis Keterkaitan PenelitianPenyuluhan. 13-21 February 1991. Pusat Perpustakaan dan Komunikasi Penelitian, Bogor, Indonesia. Unpublished manuscript. Rivera, William M., and Daniel J. Gustafson (eds.). 1991. Agricultural extension: Worldwide institutional evolution and forces for change. Elsevier, Amsterdam. Rivera, William M., and Susan G. Schram (eds.). 1987. Agricultural extension worldwide: Issues, practices and emerging priorities. Croom Helm, London. Riyanto, Sutisna. 1988. Hubungan beberapa karakteristik terpilih penyuluh lapangan di daerah aliran sungai (DAS) Citanduy dan kebutuhan komunikasi mereka. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Rogers, Everett M. 1969. Modernization among peasants: The impact of communication. Holt, Rinehart and Winston, New York. Rogers, Everett M. 1983. Diffusion of innovations. 3rd ed. Free Press, New York. Rogers, Everett, and Rekha Agarwala-Rogers. 1976. Communication in organizations. Free Press, New York. Rogers, Everett M. and D. Lawrence Kincaid. 1981. Communication networks: Toward a new paradigm for research. Free Press, New York. Rogers, Everett M. and F. Floyd Shoemaker. 1971. Communication of innovations: A cross-cultural approach. 2nd ed. Free Press, New York. Röling, Niels. 1989. The agricultural research-technology transfer interface: A knowledge systems perspective. Linkages Theme Paper 6, International Service for National Agricultural Research, The Hague, Netherlands. Röling, Niels, Kevin D. Gallagher, and Elske van de Fliert. 1991. Beyond the Green Revolution: Capturing efficiencies from integrated agriculture. Indonesia's IPM
312 Programme provides a generic case. Informal report, National Programme for Development and Training of Integrated Pest Management in Rice-Based Cropping Systems. Yogyakarta, May 1991. Rosenberg, Victor. 1967. Factors affecting the preferences of industrial personnel for information gathering methods. Information Storage and Retrieval 3:119-127. Rubin, R. B. 1979. The effect of context on information seeking across the span of initial interaction. Communication Quarterly 27(3):13-20. (abstract) Rusyana, Yana. 1984. Pengadaan alat bantu penyuluhan media tercetak di Balai Informasi Pertanian Kayuambon, Lembang, Jawa Barat. Sarjana thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Sadikin S.W. 1982. Overcoming technology gaps. Indonesian Agricultural Research and Development Journal 4(1):1-7. Samsisaputra, Waluya. 1987. Penyiapan penyuluhan pertanian di Jawa Barat. Pp. 145148 in: Prosiding kongres penyuluhan pertanian 1987. Subang, 4-6 July 1987. Perhiptani. Schaninger, Charles M. and Donald Sciglimpaglia. 1981. The influence of cognitive personality traits and demographics on consumer information acquisition. Journal of Consumer Research 8(2):208-216. (abstract) Schumacher, Henry A., Rahim Md. Sail, Aida Vitayala Sjafri Hubeis, and Marwah Daud Ibrahim. 1991. Review of research communication of Agency for Agricultural Research and Development. Final report. Agricultural Research Management Project, Agency for Agricultural Research and Development, Indonesia. Seegers, Stephan and David Kaimowitz. 1990. Relations between agricultural researchers and extension workers: The survey evidence. Journal for Farming Systems Research-Extension 1(2):29-45. Sejati, Wahyuning Kusuma. 1991. Komunikasi hasil penelitian dan pengembangan peternakan. Paper presented at the Temu Teknis Keterkaitan Penelitian Penyuluhan, Pusat Perpustakaan dan Komunikasi Penelitian, Bogor, Indonesia, 1321 February 1991. Sigman, Vickie A., and Burton E. Swanson. 1984. Problems facing national agricultural extension in developing countries. Interpaks Series 3, University of Illinois at Urbana-Champaign. Silveira, Mary Pat Williams (ed.). 1985. Research and development: Linkages to production in developing countries. United Nations Science and Technology for Development Series. Westview Press, Boulder, CO. Simandjuntak, Spener Jafar. 1988. Persepsi aparat peternakan terhadap Bulletin Balai Penelitian Veteriner (kasus: Jawa Barat). M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Sims, Holly, and David Leonard. 1990. The political economy of the development and
313 transfer of agricultural technologies. Pp. 43-73 in: Kaimowitz, David (ed.). Making the link: Agricultural research and technology transfer in developing countries. Westview, Boulder, CO. Siwi, B. H., and Paul Mundy. 1986. Management of agricultural research in relation to extension in Indonesia. Paper presented at the consultation meeting of agricultural research and extension experts, SEARCA, Philippines, 14-19 April 1986. SK Bersama. 1991. Pedoman penyelenggaraan penyuluhan pertanian. Surat keputusan bersama (joint ministerial decree) no. 65 1991/no. 539/Kpts/LP.120/7/91. Ministries of the Interior and Agriculture, Indonesia. SK Mentan. 1984. Organisasi dan tata kerja Balai Informasi Pertanian. Surat keputusan (ministerial decree) no. OT.210/308/Kpts/5/1984. Ministry of Agriculture, Indonesia. SK Mentan. 1989. Tata hubungan fungsi penelitian dan penyuluhan pertanian. Surat keputusan (ministerial decree) no. 439/Kpts/0...[illegible]..0/6/1989. Ministry of Agriculture, Indonesia. SK Mentan. 1991. Koordinasi penyuluhan pertanian. Surat keputusan (ministerial decree) no. 58/Kpts/LP.110/2/91. Ministry of Agriculture, Indonesia. Slamet, R. Margono. 1990. Perspektif ilmu penyuluhan pembangunan menyongson era tinggal landas. Paper presented at the Seminar Sehari Peranan Penyuluhan Pembangunan Menyongsong Era Tinggal Landas. 25 September 1990, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Snyder, Monteze Mundy. 1987. A framework for analysis of agricultural research organizations and extension linkages in West Africa. Ph.D. dissertation, George Washington University. Sophia, Sulastuti. 1988. Informasi hasil penelitian yang diperlukan oleh penyuluh pertanian spesialis (PPS) Jawa Barat: Hubungannya dengan karakteristik PPS sebagai pengguna informasi. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. SR-CRSP. 1990. Partners in productivity: Farmers and scientists collaborating in the research and development process. Small Ruminant-Collaborative Research Support Program, Research Institute for Animal Production, Bogor, Indonesia. Steel, Robert G. D., and James H. Torrie. 1980. Principles and procedures of statistics: A biometrical approach. 2nd. ed. McGraw-Hill, New York. Stone, Richard. 1992. Researchers score victory over pesticides -- and pests -- in Asia. Science 256:1272 (29 May 1992). Subarma, Bambang. 1985. Evaluasi pelaksanaan sistim kerja latihan dan kunjungan (Laku): Studi kasus di wilayah kerja Balai Penyuluhan Pertanian Situraja, kabupaten Sumedang, Jawa Barat. Sarjana thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia.
314 Suhardjo, Djodjo. 1989. Operationalizing the training and visit system in Bandung district of West Java, Indonesia. Ph.D. dissertation, University of the Philippines at Los Baños, Philippines. Suharno, Iski. 1986. Telaah kegiatan kerja Balai Informasi Pertanian Kayuambon Lembang, Jawa Barat dan penggunaan kaset sebagai alat bantu penyuluhan terhadap perubahan perilaku petani (Studi kasus di BIP Kayuambon dan desa Cibodas, Lembang). Sarjana thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Sukaryo, Dady Ganda. 1983. Farmer participation in the Training and Visit system and the role of the village extension worker: Experience in Indonesia. Pp. 17-25 in: Cernea, Michael M., John K. Coulter, and John F. A. Russell (eds.). Agricultural extension by training and visit: The Asian experience. World Bank, Washington, DC. Sunarno, Sylvia A. 1983. Evaluation of AARD publications by subject matter specialists in Indonesia. M.S. thesis, University of the Philippines at Los Baños, Philippines. Sunarno, Sylvia, Sumardi Dahlan, Endang Setyorini, and Penny I. Iskak (eds.). 1989a. Prosiding temu tugas penelitian-penyuluhan pertanian No. 1. Pusat Perpustakaan Pertanian dan Biologi, Badan Penelitian dan Pengembangan Pertanian, Bogor, Indonesia. Sunarno, Sylvia, Sumardi Dahlan, Endang Setyorini, and Penny I. Iskak. 1989b. Prosiding temu tugas penelitian-penyuluhan pertanian No. 2. Pusat Perpustakaan Pertanian dan Biologi, Badan Penelitian dan Pengembangan Pertanian, Bogor, Indonesia. Sunarno, Sylvia, Sumardi Dahlan, Endang Setyorini, Penny I. Iskak, and Maksum. 1989c. Prosiding temu tugas penelitian-penyuluhan pertanian No. 3. Pusat Perpustakaan Pertanian dan Biologi, Badan Penelitian dan Pengembangan Pertanian, Bogor, Indonesia. Suradisastra, Kedi, and Tjeppy D. Soedjana. 1990. A comparative study on socioeconomic aspects of small ruminant production in West Java. Working Paper 114, Small Ruminant-Collaborative Research Support Program, Balai Penelitian Ternak, Bogor. Surialaga, Tjetjep S. 1984. Peranan penyuluhan dalam mengembangkan kemampuan kelompok tani melaksanakan intensifikasi padi di tujuh desa lingkar kampus IPB, kabupaten Bogor. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Suryono, Dwi Windu. 1985. Efektivitas penerapan penyuluhan pertanian dengan sistem kerja latihan dan kunjungan (Laku) dalam program intensifikasi khusus (Insus): Suatu studi kasus di BPP Rawamerta, kabupaten D.T. II Karawang. Sarjana thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia.
315 Sutjipta, Nyoman. 1982. Hubungan pelaksanaan penyuluhan sistem Laku dan keberhasilan PPL melaksanakan tugasnya di Bali. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. SWAMPS-II. 1991. Farming systems in Indonesia's tidal swamps: Research highlights, 1987-90. Agricultural Research and Development in the Tidal Swamps Project, SWAMPS-II, Agency for Agricultural Research and Development, Indonesia. Swanson, Burton E. 1984. Agricultural extension: A reference manual. 2nd. ed. Food and Agriculture Organization of the United Nations, Rome. Syam, Mahyuddin. 1990. Komunikasi penelitian Badan Penelitian dan Pengembangan Pertanian. Pp. 24-28 in: Badan Litbang Pertanian. Prosiding lokakarya komunikasi penelitian pertanian untuk memperkuat keterkaitan penelitian-penyuluhan. Pusat Perpustakaan Pertanian dan Komunikasi Penelitian, Bogor, Indonesia. Syam, Mahyuddin, and Paul Mundy. In press (1992). Agricultural research, extension and education in Indonesia. In: Miller, Freda A. Agriculture in Indonesia. Centre for Agricultural Strategy, Reading University/Oxford University Press, Oxford. Syam, Mahyuddin, and Adi Widjono. undated (1991?). Keterkaitan antara penelitian dan penyuluhan: Persepsi, institusi, dan tata hubungan kerja. Pusat Penelitian dan Pengembangan Tanaman Pangan, Bogor, Indonesia. Manuscript. SYGAP. 1989. Annual report, September 1988-August 1989. Volume I: Agricultural research and extension organization in Indonesia and Thailand. Soybean yield gap analysis project, phase II. Agency for Agricultural Research and Development, Indonesia/Ministry of Agriculture and Co-operatives, Thailand. University of Missouri, Columbia, MO. Takasaki, N., and T. Ozawa. 1983. Analysis of information flow in Japan. Information Economics and Policy 1(2):177-193. Taslim, Haeruddin. 1991. Sistem peningkatan produksi pangan di Indonesia. Seminar paper presented at Balai Penelitian Tanaman Pangan Sukamandi, 1 March 1991. Thomas, Soedarsono. 1989. Adoption of recommended potato production technology in Pangalengan, West Java, Indonesia. Ph.D. dissertation, University of the Philippines at Los Baños, Philippines. Tjitropranoto, Prabowo. 1990. Teknologi/informasi yang diperlukan oleh beberapa wilayah. Unpublished list of information requests collected by provincial Technology Application Workshops. Tjitropranoto, Prabowo. 1991. Research communication and research-extensionfarmers linkage. Paper presented at the Evaluation Meeting of the Agricultural Research Management Project, Ciloto, June 1991. Agency for Agricultural Research and Development, Jakarta.
316 Tjitropranoto, Prabowo, and Mahyuddin Syam. 1989. Communication between research and extension in Indonesia. Pp. 339-346 in: Sukmana, Soleh, Pervaiz Amir, and Djojo M. Mulyadi (eds.). Developments in procedures for farming systems research. Proceedings of an international workshop, 13-17 March 1989, Puncak, Bogor, Indonesia. Agency for Agricultural Research and Development, Jakarta. Van den Ban, A. W., and H. S. Hawkins. 1988. Agricultural extension. Longman, Harlow, U.K. Wahyuni, Sri, Patrick J. Ludgate, and Michael F. Nolan. 1990. Dissemination of small ruminant production information in outreach pilot project (OPP) villages in the Bogor District of West Java. Pp. 250-255 in: Resource utilization for livestock production in Malaysia. Proceedings of the 13th annual conference of the Malaysia Society of Animal Production, Malacca, Malaysia, 6-8 March 1990. Malaysian Society of Animal Production, Serdang/Malaysian Agricultural Research and Development Institute, Kuala Lumpur. Walker, Helen M., and Joseph Lev. 1953. Statistical inference. Holt, Rinehart and Winston, New York. Ward, William B. 1985. Science and rice in Indonesia. Oelgeschlager, Gunn and Hain, Boston, MA. Wardojo. 1990. Pendekatan penyuluhan pertanian untuk meningkatkan partisipasi masyarakat. Paper presented at the Seminar Sehari Peranan Penyuluhan Pembangunan Menyongsong Era Tinggal Landas. 25 September 1990, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Warner, Paul D., and James A. Christenson. 1984. The Cooperative Extension Service: A national assessment. Westview, Boulder, CO. Warsito. 1989. Perception of the Pancasila and job performance of agricultural extension workers of Daerah Istimewa Yogyakarta, Indonesia. M.S. thesis, University of the Philippines at Los Baños, Philippines. Widjono, Adi., 1986. Perception of superiors' openness to receive upward information in an Indonesian food crops agricultural development organization. M.S. thesis, University of Wisconsin-Madison, Madison, WI. Widjono, Adi. 1989. Beberapa pertimbangan dalam pengamatan efektifitas penelitian sistem usahatani. Pp. 53-60 in: Sistem usahatani tanaman-ternak di lahan kering. Risalah seminar hasil penelitian Proyek Penelitian Sistem Usahatani Tanaman-Ternak (Crop-Animal Systems Research Project), Bogor, 19-21 September 1989. Badan Penelitian dan Pengembangan Pertanian and International Development Research Centre, Bogor, Indonesia. Widjono, Adi. 1990. Beberapa masalah penerbitan publikasi ilmiah di lingkup Badan Litbang Pertanian. Paper presented at the Lokakarya Komunikasi Penelitian Pertanian, 5-11 December 1990, Bogor. Pusat Penelitian dan Pengembangan Tanaman Pangan, Bogor, Indonesia.
317 Widjono, Adi, Mahyuddin Syam, Sutoro, Husni Malian, and M. Nainy A. Kirom. 1989. Komunikasi hasil penelitian pertanian dalam pengembangan DAS Citanduy. Pp 245-261 in: Risalah lokakarya penelitian dan pengembangan sistim usahatani konservasi di DAS Citanduy. Proyek Penelitian Penyelamatan Hutan, Tanah dan Air, Badan Penelitian dan Pengembangan Pertanian, Jakarta. Wiio, Osmo A. 1980. Information and communication: A conceptual analysis. Report, Department of Communication, University of Helsinki, Finland. (summarized in Rogers 1983). Wirasinghe, Suranimala, S. P. R. Weerasinghe, and M. H. J. P. Fernando. 1985. On-farm validation of technology in heterogeneous Sri Lankan environments. Pp. 93-109 in: Cernea, Michael M., John K. Coulter, and John F. A. Russell (eds.). Researchextension-farmer: A two-way continuum for agricultural development. World Bank, Washington, DC. Wiratmadja, Ruyat. 1987. Empathic ability of agricultural extension workers in West Java, Indonesia. Ph.D. dissertation, University of the Philippines at Los Baños, Philippines. Witjaksono, Roso. 1990. Hubungan perilaku komunikasi dan tingkat pemahaman informasi anggota kelompok tani tentang paket teknologi Supra Insus di WKBPP Sanden, kabupaten Bantul, D.I. Yogyakarta. M.S. thesis, Institut Pertanian Bogor (Bogor Agricultural University), Bogor, Indonesia. Woelfel, John C. 1977. Changes in interpersonal communication patterns as a consequence of need for information. Communication Research 4(3):235-256. (abstract) Worchel, Stephen, Joel Cooper, and George R. Goethals. 1988. Understanding social psychology. The Dorsey Press, Chicago. Wyckoff, J. B. 1965. Closer cooperation between research and extension. Journal of Farm Economics 47(3):834-837.
320
APPENDIX 1 INFORMATION SOURCES QUESTIONNAIRE Four versions of this questionnaire were distributed, identical except for the source type respondents were asked about in questions 7 and 8. Version 1 (reproduced here, slightly reduced) focused on publications of the Agricultural Information Centers (publikasi BIP). The other three versions substituted the following text for this (text is followed by the English translations):
Rubric to Question 7
Text in Qs. 7 and 8
Indonesian
English
Indonesian
English
Mengenai publikasi terbitan BIP (misal Buletin Penyuluhan Pertanian, Liptan, brosur)
About AIC publications (e.g., Buletin Penyuluhan Pertanian, Liptan, booklets).
publikasi BIP
AIC publications
Mengenai koran/majalah pertanian (misal Sinar Tani, Trubus)
About agricultural newspapers and magazines (e.g., Sinar tani, Trubus)
koran/ majalah pertanian
agricultural newspapers/ magazines
Mengenai buku terbitan Badan Litbang Pertanian serta lembaga penelitian di bawahnya (misal ringkasan hasil penelitian, risalah simposium)
About publications produced by AARD and its subsidiary research institutes (e.g., research summaries, symposium proceedings)
buku Badan Litbang Pertanian
AARD books
Mengenai informasi dari PPS lain (di instansi Anda atau di instansi lain)
About other SMSs (in your institution or other institutions)
PPS lain
other SMSs
321
322
323
324
325
APPENDIX 2 ENGLISH TRANSLATION OF INFORMATION SOURCES QUESTIONNAIRE
326 Questionnaire about SMSs' Information Sources Please circle the most appropriate answer or write your response in the place provided. Example: If you live about 30 km away from your office: How far is your house from your office? 1: 50 m 2: 200 m 3: 500 m 4: 2 km 5: 5 km 6: 20 km 7: 50 km About you 1
Are you an extension subject-matter specialist? (circle one) No
Yes
If no, please return this questionnaire immediately in the stamped envelope provided. If yes, please answer the following questions. 2
Please write your response:
1
Your place of work (circle one): 1
Provincial office of the Ministry of Agriculture (Kanwil)
2
Agricultural Information Center (BIP)
3
Provincial Agricultural Service office (Dinas)
4
District Agricultural Service office (Dinas)
5
Provincial Mass Guidance office (Bimas)
6
District Mass Guidance office (Bimas)
7
Other (specify) _____________________
2
Your specialization as an SMS:
____________
3
Your educational level
____________
4
How long have you worked as an SMS?
______ years
5
Sex (circle one):
Male
6
Place you were brought up (circle one):
Town Village
Female
327 7
Are you from a farm family? (circle one):
8
Do you have any work/employment apart from your job as an SMS? (circle one):
9
No
No
Yes
Yes (specify) ________
Do you have the following? (write an "X" next to those you own) Private house
Private car
Refrigerator
Land
Private motorbike
Color TV
Video recorder
Stereo
3
Approximate length of time you spend on the following activities:
1
Seeking or translating information: ________hours/week
2
Disseminating information:
3
Administrative work:
________hours/week
4
Other activities:
________hours/week
________hours/week
-----------------5
Total number of hours' work
________hours/week
4
Percentage of your work time you spend:
1
Trying to overcome problems that arise in the field: ________%
2
Passing on central extension messages:
________% Total: 100%
328 Information sources you use 5 For extension purposes, how much information do you obtain from the following media/sources? Please circle one number for each medium/source: Amount of information you obtain -------------------------------Very little 1
Somewhat Little
little
Medium
2
Somewhat
Very
5
7
much Much much
3
4
6
1
Agricultural magazines/newspapers (e.g., Sinar Tani, Trubus)
2
Other mass media (general newspapers/magazines, TV, radio)
3
Written recommendations from the Directorate General/extension program
4 Agricultural Information Center publications (e.g., booklets, Buletin Informasi Pertanian, Liptan) 5 Warta Penelitian dan Pengembangan Pertanian, and Jurnal Penelitian dan Pengembangan Pertanian 6
Scientific journals
7 Books published by the Agency for Agricultural Research and Development (e.g., research summaries, symposium proceedings) 8
University textbooks
9
Technical guides published by the Agency for Agricultural Research and Development
10 Technical guides published by private publishers 11 Discussions with farmers 12 Discussions with field extension agents 13 Discussions with other SMSs 14 Discussions with your superiors 15 Discussions with other agricultural officials (e.g., other colleagues) 16 Discussions with researchers/professors 17 Correspondence with researchers, professors, or libraries 18 Courses/notes from your university courses 19 Training at Agricultural Training Institutes 20 Seminars, workshops, exhibitions, or training at research institutes or universities
329 21 Field research projects 22 Agribusinesses (e.g., staff or projects of private firms) 23 Your own observations and experience 24 Your own tests and research 25 Other (specify) ________________ 6 Of all the media/sources listed above, which would you like to use (in ideal conditions) as information sources? Please choose five and put them in your order of preference: Most wanted: No:_____
3rd:
No:_____
2nd-most:
4th:
No:_____
No:_____
5th: No:_____
330 7 About agricultural newspapers/magazines (e.g., Sinar Tani, Trubus): please circle the most appropriate number on each line: Never 1 1
Once 2
Once
Once
a year a month
a week day
3
6
4
5
7
Every 8
9
Frequency you read agricultural newspapers/magazines
2 Frequency you obtain information useful for extension from agricultural newspapers/magazines 3
Frequency you use information from agricultural newspapers/magazines
Don't know
Not
this source
useful useful useful useful
1
3
2
Somewhat 4
5
6
Rather Very 7
8
9
4 Usefulness of agricultural newspapers/magazines as a source of extension information for you Don't Your own
Other Other Same
Distance
know office room in buil- <50 km <150 km
Distance
Distance
>150 km
building ding 5
Nearest place agricultural newspapers/magazines normally are located
8 Your opinion about the following questions (please circle the one most appropriate number on each line): Don't Disagree
Somewhat
Somewhat
know
disagree
agree
3
5
0
1
2
4
6
Agree 7
1 Information from agricultural newspapers/magazines often does not agree with information from other sources
331 2
Agricultural newspapers/magazines are difficult to find
3 Much information from agricultural newspapers/magazines are not relevant to problems faced by farmers in your area 4
Agricultural newspapers/magazines are the most credible sources
5 Compared with other sources, agricultural newspapers/magazines are easy to use for extension purposes 6 Agricultural newspapers/magazines usually contain information that is currently needed (timely) 7 You devote a lot of energy to obtain information from agricultural newspapers/magazines 8 Agricultural newspapers/magazines often discuss topics that are closely related to the situation in your area 9
Agricultural newspapers/magazines provide information in a ready-to-use form
10 Agricultural newspapers/magazines contain the most complete information compared to other sources 11 You know the type of information contained by agricultural newspapers/magazines 12 Agricultural newspapers/magazines often provide information that is not new to you 13 Obtaining information from agricultural newspapers/magazines is an important part of your job 14 You know agricultural newspapers/magazines well 15 Compared with other sources of information on agricultural technology, agricultural newspapers/magazines are easy to understand
332 About your extension activities 9
How many times in the last 3 months have done the following?
1
Visited farmers
________times
2
Visited Rural Extension Centers
________times
3
Sought information to answer questions from field agents or farmers ________times
4
Read articles in scientific journals
5
Read books published by the Agency for Agricultural Research and Development ________times
________times
10 How many times in the last year have you done the following? 1
Tested technology in the field
________tests
2
Traveled out of town to seek information for extension purposes ________times
3 Attended training held by the Agency for Agricultural Training and Education (e.g., at an Agricultural Training Institute) ________times 11 How many times in the last three years have you done the following? 1 Attended training courses organized by Agency for Agricultural Research and Development ________courses 2
Cooperated with AARD research projects (e.g., farming systems research, field trials) ________projects
3
Attended exhibits, seminars, etc. at research institutes ________times
4 Attended technical meetings (e.g., Extension Coordination Forums, etc.) also attended by AARD researchers ________times 5
Visited or received visits from AARD researchers ________times
6
Written a letter to a research institute
7
Given information to researchers about problems in your area ________times
________times
12 You opinion about the following statements (please circle the most appropriate number on each line):
333 Disagree 1
2
Somewhat
Somewhat
disagree
agree
3
4
5
Agree 6
7
1
Recommendations from "above" must be tested first before being extended
2
Recommendations from "above" may not be changed before being extended
Thank you for completing this questionnaire. Please return this questionnaire in the stamped envelope provided to: Evaluation of SMSs' Information Sources Research Communication Department Central Research Institute for Animal Science PO Box 210, Bogor
334
APPENDIX 3 PUBLICATIONS QUESTIONNAIRE
335
336
337
338
339
APPENDIX 4 ENGLISH TRANSLATION OF PUBLICATIONS QUESTIONNAIRE
340 Questionnaire about publications used by SMSs Please circle the most appropriate answer or write your response in the place provided. Example: If you live about 30 km away from your office: How far is your house from your office? 1: 50 m 2: 200 m 3: 500 m 4: 2 km 5: 5 km 6: 20 km 7: 50 km About you 1
Are you an extension subject-matter specialist? (circle one) No
Yes
If no, please return this questionnaire immediately in the stamped envelope provided. If yes, please answer the following questions. 2
Please write your response:
1
Your place of work (circle one): 1
Provincial office of the Ministry of Agriculture (Kanwil)
2
Agricultural Information Center (BIP)
3
Provincial Agricultural Service office (Dinas)
4
District Agricultural Service office (Dinas)
5
Provincial Mass Guidance office (Bimas)
6
District Mass Guidance office (Bimas)
7
Other (specify) _____________________
2
Your specialization as an SMS:
3
How long have you worked as an SMS? ________ years
4
Sex (circle one):
Problems you face
________________ Male
Female
341 3 Following are ten items that may be extension problems. Please circle the one number you think is most appropriate for each item: 1: not a problem
3: Somewhat of a problem
5: Important problem 7: Very important problem 1
Technology -- Availability of technology suitable for extension to farmers
2 Obtaining information -- Adequacy of information flows from researchers to extensionists 3
Feedback -- Adequacy of information flows from extensionists to researchers
4 Technical skills -- Extension personnel's practical skills in absorbing new technologies 5
Extension skills -- Extension personnel's skills in teaching and communicating
6
Mobility -- Adequate transport for extension personnel to visit farmers
7 Facilities -- Teaching and communication facilities for extension personnel (projectors, classrooms, telephones, etc.) 8 Teaching aids -- Availability of teaching materials, printed materials, demonstration kits, etc. 9
Organization -- Non-extension duties
10
Rewards -- Payment (moral and material) received for performing extension duties
11
Other problems (specify) ________________________
342 4 Following are several problems that you and other SMSs may face in obtaining information for extension purposes. Please circle the one number you think is most appropriate for each item: Scale: Same as for question 3 1
Publications are not received regularly
2
Publications are not relevant to field problems
3
Technology recommendations are difficult to derive from publications
4
Funding for performing extension duties is insufficient
5
Training to raise SMSs' knowledge and skills is infrequent
6
SMSs have too little time to seek information
7
SMSs lack skills in translating scientific information into extension language
8
Meetings between researchers and extensionists are infrequent
9
Other (specify) ________________________
Publications you use 5 How many times in the last 3 months have you read the following publications? (circle the most appropriate number on each line): 1: Never
2: Once in 3 months
3: Once a month
4: Once a week
5: Every 3 days
6: Every day
1
Sinar Tani (biweekly agricultural newspaper)
2
Buletin Informasi Pertanian (AIC magazine)
3
Liptan (AIC factsheets)
4 Warta Penelitian dan Pengembangan Pertanian (Warta Litbang, AARD newsletter) 5
Scientific journals published by AARD
6 Books published by AARD and its research insititutions (inclduing proceedings volumes) 7
Booklets published by AICs
343
6 How useful are the following publications to you as sources of information for extension? (cirlce the most appropriate number on each line): 0: Don't know this publication 1: Not useful
3: Somewhat useful
5: Rather useful
7: Very useful
1
Sinar Tani (biweekly agricultural newspaper)
2
Buletin Informasi Pertanian (AIC magazine)
3
Liptan (AIC factsheets)
4 Warta Penelitian dan Pengembangan Pertanian (Warta Litbang, AARD newsletter) 5
Scientific journals published by AARD
6 Books published by AARD and its research insititutions (inclduing proceedings volumes) 7
Booklets published by AICs
344 7 (Please answer this question only if your specialty is food crops): Have you obtained information about the following agricultural technologies? If yes, from which source/medium did you obtain the most information? 0: No 1: Yes If Yes, Main source of information (circle one): 0: Don't know
1: AARD publications
2: AIC publications
3: Mass media
4: Colleagues
5: Training
6: Other (specify) ________________ 1
Characteristics of the new wetland rice variety Barumun
2
Characteristics of the new maize variety Rama
3
Application rates of P fertilizer can be reduced in wetland rice intensification areas
4
The use of Sesbania as green manure
5
Natural enemies of whitebacked planthopper on rice
6
The use of rice-hull ash in stroing soybean seeds
8 (Please answer this question only if your specialty is livestock): Have you obtained information about the following agricultural technologies? If yes, from which source/medium did you obtain the most information? Scale: Same as for question 7 1
The use of urea molasses block (appetizer for ruminants)
2
Use of conical bamboo nests for hatching local chickens
3
Raising ducks in confinement (without pond)
4
The use of Gliricidia for sheep and goat feed
5
Vaccination of local chickens against Newcastle disease
345 6
Prolific thin-tailed sheep (produce twins)
346 Types of information you need 9 Do you still need information about the following topics? (Please write an "X" in the five boxes that you feel you most need information on. Food crops Culti- Seed and
Pests &
Post
vation varieties
diseases
harvest
Rice Maize, sorghum, wheat Roots & tubers Legumes Vegetables Fruits Livestock Manage-
Reproduction/
ment
breeding
Feed
Dairy cattle Beef cattle, buffaloes Sheep, goats Improved chickens Local (village) chickens Ducks (layers and meat) Other Livestock fodder plants
Estate and industrial crops
Milk handling
Machinery and equipment
Livestock post-harvest
Irrigation
Fisheries
Economics
Marketing
Faming system analysis
Extension
Rural sociology
Health
347 Regional planning
Other (specify)___________
Thank you for completing this questionnaire. Please return this questionnaire in the stamped envelope provided to: Evaluation of SMSs' Information Sources Research Communication Department Central Research Institute for Animal Science PO Box 210, Bogor
348
APPENDIX 5 RESPONSE FREQUENCIES FROM INFORMATION SOURCES QUESTIONNAIRE Questions are listed in the same order as on the questionnaire (Appendices 1 and 2). The figures in the variable name refer to the question number. Thus, INST21 refers to question 2.1. Frequency tabulations are provided for all variables except questions IS7 and IS8, which are broken down by source type.
349 INST21
Institutional affiliation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7
7 63 31 64 30 84 1 ------280
2.5 22.5 11.1 22.9 10.7 30.0 .4 ------100.0
2.5 22.5 11.1 22.9 10.7 30.0 .4 ------100.0
2.5 25.0 36.1 58.9 69.6 99.6 100.0
Kanwil BIP Dinas Tk 1 Dinas Tk 2 SPH Bimas Tk 1 SPH Bimas Tk 2 Other
Total Mean Std dev
4.082 1.605
Valid cases
280
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
6.000 7.000
0
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SPEC22
Specialization
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9
17 113 51 7 20 25 9 12 11 15 ------280
6.1 40.4 18.2 2.5 7.1 8.9 3.2 4.3 3.9 5.4 ------100.0
6.4 42.6 19.2 2.6 7.5 9.4 3.4 4.5 4.2 Missing ------100.0
6.4 49.1 68.3 70.9 78.5 87.9 91.3 95.8 100.0
Vet med Food crops Livestock Estate crops Social/ econ Soil, mech, postharv Extension Fish Conservation, other
Total Mean Std dev
2.517 2.187
Valid cases
265
Median Minimum
2.000 .000
Missing cases
Mode Maximum
1.000 8.000
15
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - DIK23
Education
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 9
14 236 24 6 ------280
5.0 84.3 8.6 2.1 ------100.0
5.1 86.1 8.8 Missing ------100.0
5.1 91.2 100.0
Attended university Ir > Ir
Total Mean Std dev Valid cases
2.036 .371 274
Median Minimum Missing cases
2.000 1.000
Mode Maximum 6
2.000 3.000
350 LAMA24
Years as PPS
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 99
15 9 13 22 50 60 23 22 12 14 7 9 3 6 2 2 1 3 4 2 1 ------280
5.4 3.2 4.6 7.9 17.9 21.4 8.2 7.9 4.3 5.0 2.5 3.2 1.1 2.1 .7 .7 .4 1.1 1.4 .7 .4 ------100.0
5.4 3.2 4.7 7.9 17.9 21.5 8.2 7.9 4.3 5.0 2.5 3.2 1.1 2.2 .7 .7 .4 1.1 1.4 .7 Missing ------100.0
5.4 8.6 13.3 21.1 39.1 60.6 68.8 76.7 81.0 86.0 88.5 91.8 92.8 95.0 95.7 96.4 96.8 97.8 99.3 100.0
Total Mean Std dev
6.853 3.806
Valid cases
279
Median Minimum
6.000 1.000
Missing cases
Mode Maximum
6.000 20.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SEX25
Sex
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
227 52 1 ------280
81.1 18.6 .4 ------100.0
81.4 18.6 Missing ------100.0
81.4 100.0
Male Female
Total Mean Std dev
.186 .390
Valid cases
279
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BESAR26
Place brought up
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
143 136 1 ------280
51.1 48.6 .4 ------100.0
51.3 48.7 Missing ------100.0
51.3 100.0
Town Village
Total Mean Std dev Valid cases
.487 .501 279
Median Minimum Missing cases
.000 .000
Mode Maximum 1
.000 1.000
351 TANI27
From farm family
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
153 122 5 ------280
54.6 43.6 1.8 ------100.0
55.6 44.4 Missing ------100.0
55.6 100.0
No Yes
Total Mean Std dev
.444 .498
Valid cases
275
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
5
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - USAHA28
Outside work
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 9
205 20 26 12 13 4 ------280
73.2 7.1 9.3 4.3 4.6 1.4 ------100.0
74.3 7.2 9.4 4.3 4.7 Missing ------100.0
74.3 81.5 90.9 95.3 100.0
Dinas, Bimas Teaching Farming Other
Total Mean Std dev
.580 1.124
Valid cases
276
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 4.000
4
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RUMAH29
Own house
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
180 99 1 ------280
64.3 35.4 .4 ------100.0
64.5 35.5 Missing ------100.0
64.5 100.0
No Yes
Total Mean Std dev
.355 .479
Valid cases
279
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LAHAN29
Own land
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
195 84 1 ------280
69.6 30.0 .4 ------100.0
69.9 30.1 Missing ------100.0
69.9 100.0
No Yes
Total Mean Std dev Valid cases
.301 .460 279
Median Minimum Missing cases
.000 .000
Mode Maximum 1
.000 1.000
352 VIDEO29
Own video
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
233 46 1 ------280
83.2 16.4 .4 ------100.0
83.5 16.5 Missing ------100.0
83.5 100.0
No Yes
Total Mean Std dev
.165 .372
Valid cases
279
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MOBIL29
Own car
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
245 34 1 ------280
87.5 12.1 .4 ------100.0
87.8 12.2 Missing ------100.0
87.8 100.0
No Yes
Total Mean Std dev
.122 .328
Valid cases
279
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SEPED29
Own motorbike
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
184 95 1 ------280
65.7 33.9 .4 ------100.0
65.9 34.1 Missing ------100.0
65.9 100.0
No Yes
Total Mean Std dev
.341 .475
Valid cases
279
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LEMAR29
Own fridge
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
167 112 1 ------280
59.6 40.0 .4 ------100.0
59.9 40.1 Missing ------100.0
59.9 100.0
No Yes
Total Mean Std dev Valid cases
.401 .491 279
Median Minimum Missing cases
.000 .000
Mode Maximum 1
.000 1.000
353 TV29
Own TV
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
68 211 1 ------280
24.3 75.4 .4 ------100.0
24.4 75.6 Missing ------100.0
24.4 100.0
No Yes
Total Mean Std dev
.756 .430
Valid cases
279
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - STER29
Own stereo
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
187 92 1 ------280
66.8 32.9 .4 ------100.0
67.0 33.0 Missing ------100.0
67.0 100.0
No Yes
Total Mean Std dev
.330 .471
Valid cases
279
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - WEALTH29
Wealth index
Value Label 0 1 2 3 4 5 6 7 8
items item items items items items items items items
Value
Frequency
Percent
Valid Percent
Cum Percent
.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 .
29 55 57 44 41 26 16 6 5 1 ------280
10.4 19.6 20.4 15.7 14.6 9.3 5.7 2.1 1.8 .4 ------100.0
10.4 19.7 20.4 15.8 14.7 9.3 5.7 2.2 1.8 Missing ------100.0
10.4 30.1 50.5 66.3 81.0 90.3 96.1 98.2 100.0
Total Mean Std dev Valid cases
2.771 1.928 279
Median Minimum Missing cases
2.000 .000
Mode Maximum 1
2.000 8.000
354 CARI31
Seeking info (hours/week)
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19 20 21 24 25 28 30 32 40 48 99
4 7 13 9 13 11 24 9 26 4 36 33 1 21 13 3 2 12 1 11 5 3 3 2 3 1 1 2 7 ------280
1.4 2.5 4.6 3.2 4.6 3.9 8.6 3.2 9.3 1.4 12.9 11.8 .4 7.5 4.6 1.1 .7 4.3 .4 3.9 1.8 1.1 1.1 .7 1.1 .4 .4 .7 2.5 ------100.0
1.5 2.6 4.8 3.3 4.8 4.0 8.8 3.3 9.5 1.5 13.2 12.1 .4 7.7 4.8 1.1 .7 4.4 .4 4.0 1.8 1.1 1.1 .7 1.1 .4 .4 .7 Missing ------100.0
1.5 4.0 8.8 12.1 16.8 20.9 29.7 33.0 42.5 44.0 57.1 69.2 69.6 77.3 82.1 83.2 83.9 88.3 88.6 92.7 94.5 95.6 96.7 97.4 98.5 98.9 99.3 100.0
Total Mean Std dev Valid cases
10.912 7.244 273
Median Minimum Missing cases
10.000 .000
Mode Maximum 7
10.000 48.000
355 SAMP32
Providing info (hours/week)
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 19 20 21 22 24 25 26 28 30 32 35 36 99
2 12 8 22 5 23 7 16 5 35 28 3 17 21 14 2 14 1 9 5 1 7 4 2 3 2 2 1 2 7 ------280
.7 4.3 2.9 7.9 1.8 8.2 2.5 5.7 1.8 12.5 10.0 1.1 6.1 7.5 5.0 .7 5.0 .4 3.2 1.8 .4 2.5 1.4 .7 1.1 .7 .7 .4 .7 2.5 ------100.0
.7 4.4 2.9 8.1 1.8 8.4 2.6 5.9 1.8 12.8 10.3 1.1 6.2 7.7 5.1 .7 5.1 .4 3.3 1.8 .4 2.6 1.5 .7 1.1 .7 .7 .4 .7 Missing ------100.0
.7 5.1 8.1 16.1 17.9 26.4 28.9 34.8 36.6 49.5 59.7 60.8 67.0 74.7 79.9 80.6 85.7 86.1 89.4 91.2 91.6 94.1 95.6 96.3 97.4 98.2 98.9 99.3 100.0
Total Mean Std dev Valid cases
11.978 7.012 273
Median Minimum Missing cases
12.000 .000
Mode Maximum 7
10.000 36.000
356 ADMIN33
Admin work (hours/week)
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 13 14 15 16 18 19 20 21 22 23 24 25 26 27 28 30 34 35 36 38 40 42 47 99
6 1 5 7 13 16 28 13 17 5 33 28 1 11 8 9 12 2 12 7 2 1 5 2 2 2 7 7 1 3 2 1 2 1 1 7 ------280
2.1 .4 1.8 2.5 4.6 5.7 10.0 4.6 6.1 1.8 11.8 10.0 .4 3.9 2.9 3.2 4.3 .7 4.3 2.5 .7 .4 1.8 .7 .7 .7 2.5 2.5 .4 1.1 .7 .4 .7 .4 .4 2.5 ------100.0
2.2 .4 1.8 2.6 4.8 5.9 10.3 4.8 6.2 1.8 12.1 10.3 .4 4.0 2.9 3.3 4.4 .7 4.4 2.6 .7 .4 1.8 .7 .7 .7 2.6 2.6 .4 1.1 .7 .4 .7 .4 .4 Missing ------100.0
2.2 2.6 4.4 7.0 11.7 17.6 27.8 32.6 38.8 40.7 52.7 63.0 63.4 67.4 70.3 73.6 78.0 78.8 83.2 85.7 86.4 86.8 88.6 89.4 90.1 90.8 93.4 96.0 96.3 97.4 98.2 98.5 99.3 99.6 100.0
Total Mean Std dev Valid cases
12.853 8.839 273
Median Minimum Missing cases
10.000 .000
Mode Maximum 7
10.000 47.000
357 LAIN34
Other work (hours/week)
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 24 28 30 31 32 35 40 41 42 50 84 99
30 5 15 17 26 15 20 14 23 10 20 2 14 2 12 5 7 1 8 1 4 1 8 2 3 1 1 1 1 1 1 1 1 7 ------280
10.7 1.8 5.4 6.1 9.3 5.4 7.1 5.0 8.2 3.6 7.1 .7 5.0 .7 4.3 1.8 2.5 .4 2.9 .4 1.4 .4 2.9 .7 1.1 .4 .4 .4 .4 .4 .4 .4 .4 2.5 ------100.0
11.0 1.8 5.5 6.2 9.5 5.5 7.3 5.1 8.4 3.7 7.3 .7 5.1 .7 4.4 1.8 2.6 .4 2.9 .4 1.5 .4 2.9 .7 1.1 .4 .4 .4 .4 .4 .4 .4 .4 Missing ------100.0
11.0 12.8 18.3 24.5 34.1 39.6 46.9 52.0 60.4 64.1 71.4 72.2 77.3 78.0 82.4 84.2 86.8 87.2 90.1 90.5 91.9 92.3 95.2 96.0 97.1 97.4 97.8 98.2 98.5 98.9 99.3 99.6 100.0
Total Mean Std dev Valid cases
9.165 9.315 273
Median Minimum Missing cases
7.000 .000
Mode Maximum 7
.000 84.000
358 TOT35 Value
Total work (hours/week) Cum Freq Pct Pct Value
7 8 11 12 13 17 22 23 24 25 26 29 30 31 32 33 34 35 36 37 38
1 1 1 1 2 2 2 1 1 2 1 2 4 1 5 6 7 4 20 11 23
Value
Freq
999
8
Mean Std dev
0 0 0 0 1 1 1 0 0 1 0 1 1 0 2 2 3 1 7 4 8
44.647 14.605
Valid cases
272
0 1 1 1 2 3 4 4 4 5 6 6 8 8 10 12 15 16 24 28 36
Cum Freq Pct Pct
39 12 4 40 11 4 41 2 1 42 30 11 43 3 1 44 14 5 45 9 3 46 6 2 47 5 2 48 16 6 49 3 1 50 5 2 51 1 0 52 4 1 53 1 0 54 5 2 55 1 0 56 7 3 58 2 1 60 7 3 61 1 0 M I S S I N G D A T Value Freq
Median Minimum
40 44 45 56 57 63 66 68 70 76 77 79 79 81 81 83 83 86 86 89 89 A
42.000 7.000
Missing cases
Value
Cum Freq Pct Pct
63 64 66 69 70 72 73 74 75 76 77 78 79 80 81 82 84 85 90 119
2 1 1 2 2 2 2 1 1 1 2 1 1 2 1 1 1 3 1 1
Value
Freq
Mode Maximum
1 90 0 90 0 91 1 92 1 92 1 93 1 94 0 94 0 94 0 95 1 96 0 96 0 96 1 97 0 97 0 98 0 98 1 99 0 100 0 100
42.000 119.000
8
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MASAL41
Solving field problems (percent)
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
5 10 15 20 25 30 35 40 42 45 50 52 55 60 65 70 72 74 75 80 85 90 99
1 3 3 2 6 15 3 30 1 4 56 1 4 64 7 34 1 1 16 10 2 5 11 ------280
.4 1.1 1.1 .7 2.1 5.4 1.1 10.7 .4 1.4 20.0 .4 1.4 22.9 2.5 12.1 .4 .4 5.7 3.6 .7 1.8 3.9 ------100.0
.4 1.1 1.1 .7 2.2 5.6 1.1 11.2 .4 1.5 20.8 .4 1.5 23.8 2.6 12.6 .4 .4 5.9 3.7 .7 1.9 Missing ------100.0
.4 1.5 2.6 3.3 5.6 11.2 12.3 23.4 23.8 25.3 46.1 46.5 48.0 71.7 74.3 87.0 87.4 87.7 93.7 97.4 98.1 100.0
Total Mean Std dev Valid cases
54.870 16.356 269
Median Minimum Missing cases
60.000 5.000 11
Mode Maximum
60.000 90.000
359 PROGR42
Providing info from above (percent)
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
10 15 20 25 26 28 30 35 40 45 48 50 55 58 60 65 70 75 80 85 90 95 99
5 2 10 16 1 1 34 7 64 4 1 56 4 1 30 3 15 6 2 3 3 1 11 ------280
1.8 .7 3.6 5.7 .4 .4 12.1 2.5 22.9 1.4 .4 20.0 1.4 .4 10.7 1.1 5.4 2.1 .7 1.1 1.1 .4 3.9 ------100.0
1.9 .7 3.7 5.9 .4 .4 12.6 2.6 23.8 1.5 .4 20.8 1.5 .4 11.2 1.1 5.6 2.2 .7 1.1 1.1 .4 Missing ------100.0
1.9 2.6 6.3 12.3 12.6 13.0 25.7 28.3 52.0 53.5 53.9 74.7 76.2 76.6 87.7 88.8 94.4 96.7 97.4 98.5 99.6 100.0
Total Mean Std dev
45.130 16.356
Valid cases
269
Median Minimum
40.000 10.000
Missing cases
Mode Maximum
40.000 95.000
11
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KORAN51
Ag newspapers/magazines
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
4 24 26 99 52 55 17 3 ------280
1.4 8.6 9.3 35.4 18.6 19.6 6.1 1.1 ------100.0
1.4 8.7 9.4 35.7 18.8 19.9 6.1 Missing ------100.0
1.4 10.1 19.5 55.2 74.0 93.9 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
4.458 1.379 277
Median Minimum Missing cases
4.000 1.000
Mode Maximum 3
4.000 7.000
360 MEDIA52
Other mass media
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
11 31 40 86 41 58 7 6 ------280
3.9 11.1 14.3 30.7 14.6 20.7 2.5 2.1 ------100.0
4.0 11.3 14.6 31.4 15.0 21.2 2.6 Missing ------100.0
4.0 15.3 29.9 61.3 76.3 97.4 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
4.157 1.473
Valid cases
274
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
4.000 7.000
6
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - DITJEN53
Dir Gen recommendations
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
21 44 43 72 50 42 4 4 ------280
7.5 15.7 15.4 25.7 17.9 15.0 1.4 1.4 ------100.0
7.6 15.9 15.6 26.1 18.1 15.2 1.4 Missing ------100.0
7.6 23.6 39.1 65.2 83.3 98.6 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
3.826 1.542
Valid cases
276
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
4.000 7.000
4
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BIP54
BIP publications
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
9 29 31 75 61 55 18 2 ------280
3.2 10.4 11.1 26.8 21.8 19.6 6.4 .7 ------100.0
3.2 10.4 11.2 27.0 21.9 19.8 6.5 Missing ------100.0
3.2 13.7 24.8 51.8 73.7 93.5 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
4.392 1.506 278
Median Minimum Missing cases
4.000 1.000
Mode Maximum 2
4.000 7.000
361 WARTA55
Warta/Jurnal Litbang
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
37 44 38 59 53 32 14 3 ------280
13.2 15.7 13.6 21.1 18.9 11.4 5.0 1.1 ------100.0
13.4 15.9 13.7 21.3 19.1 11.6 5.1 Missing ------100.0
13.4 29.2 43.0 64.3 83.4 94.9 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
3.718 1.736
Valid cases
277
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
4.000 7.000
3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MAJ56
Scientific journals
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
52 57 42 60 31 27 5 6 ------280
18.6 20.4 15.0 21.4 11.1 9.6 1.8 2.1 ------100.0
19.0 20.8 15.3 21.9 11.3 9.9 1.8 Missing ------100.0
19.0 39.8 55.1 77.0 88.3 98.2 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
3.226 1.666
Valid cases
274
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
4.000 7.000
6
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LITB57
AARD books
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
46 55 49 63 38 18 9 2 ------280
16.4 19.6 17.5 22.5 13.6 6.4 3.2 .7 ------100.0
16.5 19.8 17.6 22.7 13.7 6.5 3.2 Missing ------100.0
16.5 36.3 54.0 76.6 90.3 96.8 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
3.295 1.630 278
Median Minimum Missing cases
3.000 1.000
Mode Maximum 2
4.000 7.000
362 TEKS58
University textbooks
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
69 67 35 51 27 23 2 6 ------280
24.6 23.9 12.5 18.2 9.6 8.2 .7 2.1 ------100.0
25.2 24.5 12.8 18.6 9.9 8.4 .7 Missing ------100.0
25.2 49.6 62.4 81.0 90.9 99.3 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
2.916 1.645
Valid cases
274
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
1.000 7.000
6
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PEDLIT59
AARD technical guides
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
37 56 47 64 35 28 7 6 ------280
13.2 20.0 16.8 22.9 12.5 10.0 2.5 2.1 ------100.0
13.5 20.4 17.2 23.4 12.8 10.2 2.6 Missing ------100.0
13.5 33.9 51.1 74.5 87.2 97.4 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
3.423 1.627
Valid cases
274
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
4.000 7.000
6
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PEDSW510
Private technical guides
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
105 57 43 37 20 8 1 9 ------280
37.5 20.4 15.4 13.2 7.1 2.9 .4 3.2 ------100.0
38.7 21.0 15.9 13.7 7.4 3.0 .4 Missing ------100.0
38.7 59.8 75.6 89.3 96.7 99.6 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
2.402 1.472 271
Median Minimum Missing cases
2.000 1.000
Mode Maximum 9
1.000 7.000
363 PET511
Farmers
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
2 10 20 83 60 69 33 3 ------280
.7 3.6 7.1 29.6 21.4 24.6 11.8 1.1 ------100.0
.7 3.6 7.2 30.0 21.7 24.9 11.9 Missing ------100.0
.7 4.3 11.6 41.5 63.2 88.1 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
4.906 1.324
Valid cases
277
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
4.000 7.000
3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PPL512
Field extension workers
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
4 7 13 46 74 95 37 4 ------280
1.4 2.5 4.6 16.4 26.4 33.9 13.2 1.4 ------100.0
1.4 2.5 4.7 16.7 26.8 34.4 13.4 Missing ------100.0
1.4 4.0 8.7 25.4 52.2 86.6 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
5.217 1.286
Valid cases
276
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
6.000 7.000
4
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PPS513
Extension subject-matter specialists
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 9 16 61 82 83 22 4 ------280
1.1 3.2 5.7 21.8 29.3 29.6 7.9 1.4 ------100.0
1.1 3.3 5.8 22.1 29.7 30.1 8.0 Missing ------100.0
1.1 4.3 10.1 32.2 62.0 92.0 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
4.982 1.243 276
Median Minimum Missing cases
5.000 1.000
Mode Maximum 4
6.000 7.000
364 BOSS514
The boss
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 16 15 72 66 77 28 3 ------280
1.1 5.7 5.4 25.7 23.6 27.5 10.0 1.1 ------100.0
1.1 5.8 5.4 26.0 23.8 27.8 10.1 Missing ------100.0
1.1 6.9 12.3 38.3 62.1 89.9 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
4.895 1.359
Valid cases
277
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
6.000 7.000
3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - REKAN515
Other colleagues
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
2 3 4 5 6 7 9
11 27 65 84 68 21 4 ------280
3.9 9.6 23.2 30.0 24.3 7.5 1.4 ------100.0
4.0 9.8 23.6 30.4 24.6 7.6 Missing ------100.0
4.0 13.8 37.3 67.8 92.4 100.0
Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
4.848 1.232
Valid cases
276
Median Minimum
5.000 2.000
Missing cases
Mode Maximum
5.000 7.000
4
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PENEL516
Researchers
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
53 68 53 60 25 13 2 6 ------280
18.9 24.3 18.9 21.4 8.9 4.6 .7 2.1 ------100.0
19.3 24.8 19.3 21.9 9.1 4.7 .7 Missing ------100.0
19.3 44.2 63.5 85.4 94.5 99.3 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
2.938 1.468 274
Median Minimum Missing cases
3.000 1.000
Mode Maximum 6
2.000 7.000
365 SURAT517
Correspondence
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 9
116 68 52 29 2 6 7 ------280
41.4 24.3 18.6 10.4 .7 2.1 2.5 ------100.0
42.5 24.9 19.0 10.6 .7 2.2 Missing ------100.0
42.5 67.4 86.4 97.1 97.8 100.0
Very little Little Somewhat little Medium Quite a lot A lot
Total Mean Std dev
2.088 1.206
Valid cases
273
Median Minimum
2.000 1.000
Missing cases
Mode Maximum
1.000 6.000
7
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UNIV518
University courses/notes
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
29 25 27 84 49 50 14 2 ------280
10.4 8.9 9.6 30.0 17.5 17.9 5.0 .7 ------100.0
10.4 9.0 9.7 30.2 17.6 18.0 5.0 Missing ------100.0
10.4 19.4 29.1 59.4 77.0 95.0 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
4.097 1.657
Valid cases
278
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
4.000 7.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LATIH519
Training
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
39 37 35 74 40 36 12 7 ------280
13.9 13.2 12.5 26.4 14.3 12.9 4.3 2.5 ------100.0
14.3 13.6 12.8 27.1 14.7 13.2 4.4 Missing ------100.0
14.3 27.8 40.7 67.8 82.4 95.6 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
3.714 1.719 273
Median Minimum Missing cases
4.000 1.000
Mode Maximum 7
4.000 7.000
366 SEMIN520
Seminars etc. at res. insts
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
57 47 49 62 37 17 7 4 ------280
20.4 16.8 17.5 22.1 13.2 6.1 2.5 1.4 ------100.0
20.7 17.0 17.8 22.5 13.4 6.2 2.5 Missing ------100.0
20.7 37.7 55.4 77.9 91.3 97.5 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
3.196 1.640
Valid cases
276
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
4.000 7.000
4
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PROY521
Field research projects
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
64 61 47 46 24 24 3 11 ------280
22.9 21.8 16.8 16.4 8.6 8.6 1.1 3.9 ------100.0
23.8 22.7 17.5 17.1 8.9 8.9 1.1 Missing ------100.0
23.8 46.5 63.9 81.0 90.0 98.9 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
2.959 1.642
Valid cases
269
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
1.000 7.000
11
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BIS522
Agribusinesses
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 9
126 57 35 23 13 8 18 ------280
45.0 20.4 12.5 8.2 4.6 2.9 6.4 ------100.0
48.1 21.8 13.4 8.8 5.0 3.1 Missing ------100.0
48.1 69.8 83.2 92.0 96.9 100.0
Very little Little Somewhat little Medium Quite a lot A lot
Total Mean Std dev Valid cases
2.099 1.378 262
Median Minimum Missing cases
2.000 1.000 18
Mode Maximum
1.000 6.000
367 PENG523
Own experience and observations
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
2 13 20 88 60 71 21 5 ------280
.7 4.6 7.1 31.4 21.4 25.4 7.5 1.8 ------100.0
.7 4.7 7.3 32.0 21.8 25.8 7.6 Missing ------100.0
.7 5.5 12.7 44.7 66.5 92.4 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
4.775 1.293
Valid cases
275
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
4.000 7.000
5
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UJI524
Own testing and research
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
26 40 42 86 46 24 8 8 ------280
9.3 14.3 15.0 30.7 16.4 8.6 2.9 2.9 ------100.0
9.6 14.7 15.4 31.6 16.9 8.8 2.9 Missing ------100.0
9.6 24.3 39.7 71.3 88.2 97.1 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev
3.699 1.519
Valid cases
272
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
4.000 7.000
8
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LAIN525
Other sources
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
7 7 7 20 15 5 8 211 ------280
2.5 2.5 2.5 7.1 5.4 1.8 2.9 75.4 ------100.0
10.1 10.1 10.1 29.0 21.7 7.2 11.6 Missing ------100.0
10.1 20.3 30.4 59.4 81.2 88.4 100.0
Very little Little Somewhat little Medium Quite a lot A lot A great deal
Total Mean Std dev Valid cases
4.101 1.733 69
Median Minimum Missing cases
4.000 1.000 211
Mode Maximum
4.000 7.000
368 INGIN61
1st wanted source
Value Label Ag newspapers/magazi Other mass media Dir Gen recommendati BIP publications Warta/Jurnal Litbang Scientific journals AARD books University textbooks AARD technical guide Private technical gu Farmers Field extension work The boss Researchers University courses/n Training Seminars etc. at res Field research proje Own experience and o Own testing and rese Other sources
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 14 16 18 19 20 21 23 24 25 99
18 3 39 16 56 6 31 3 34 1 7 2 2 3 1 23 9 11 5 5 1 4 ------280
6.4 1.1 13.9 5.7 20.0 2.1 11.1 1.1 12.1 .4 2.5 .7 .7 1.1 .4 8.2 3.2 3.9 1.8 1.8 .4 1.4 ------100.0
6.5 1.1 14.1 5.8 20.3 2.2 11.2 1.1 12.3 .4 2.5 .7 .7 1.1 .4 8.3 3.3 4.0 1.8 1.8 .4 Missing ------100.0
6.5 7.6 21.7 27.5 47.8 50.0 61.2 62.3 74.6 75.0 77.5 78.3 79.0 80.1 80.4 88.8 92.0 96.0 97.8 99.6 100.0
Total Mean Std dev
8.627 6.551
Valid cases
276
Median Minimum
6.500 1.000
Missing cases
Mode Maximum
5.000 25.000
4
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - INGIN62
2nd wanted source
Value Label Ag newspapers/magazi Other mass media Dir Gen recommendati BIP publications Warta/Jurnal Litbang Scientific journals AARD books University textbooks AARD technical guide Private technical gu Farmers Field extension work Extension subject-ma The boss Other colleagues Researchers Correspondents Training Seminars etc. at res Field research proje Agribusinesses Own experience and o Own testing and rese
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 20 21 22 23 24 99
14 3 38 27 32 20 30 5 30 2 4 2 5 1 1 3 1 12 16 6 2 7 15 4 ------280
5.0 1.1 13.6 9.6 11.4 7.1 10.7 1.8 10.7 .7 1.4 .7 1.8 .4 .4 1.1 .4 4.3 5.7 2.1 .7 2.5 5.4 1.4 ------100.0
5.1 1.1 13.8 9.8 11.6 7.2 10.9 1.8 10.9 .7 1.4 .7 1.8 .4 .4 1.1 .4 4.3 5.8 2.2 .7 2.5 5.4 Missing ------100.0
5.1 6.2 19.9 29.7 41.3 48.6 59.4 61.2 72.1 72.8 74.3 75.0 76.8 77.2 77.5 78.6 79.0 83.3 89.1 91.3 92.0 94.6 100.0
Total Mean Std dev Valid cases
9.159 6.972 276
Median Minimum Missing cases
7.000 1.000
Mode Maximum 4
3.000 24.000
369 INGIN63
3rd wanted source
Value Label Ag newspapers/magazi Other mass media Dir Gen recommendati BIP publications Warta/Jurnal Litbang Scientific journals AARD books University textbooks AARD technical guide Private technical gu Farmers Field extension work Extension subject-ma The boss Other colleagues Researchers Correspondents University courses/n Training Seminars etc. at res Field research proje Agribusinesses Own experience and o Own testing and rese Other sources
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 99
12 5 31 19 28 21 32 8 35 4 7 1 7 1 3 6 1 2 24 9 7 2 6 4 1 4 ------280
4.3 1.8 11.1 6.8 10.0 7.5 11.4 2.9 12.5 1.4 2.5 .4 2.5 .4 1.1 2.1 .4 .7 8.6 3.2 2.5 .7 2.1 1.4 .4 1.4 ------100.0
4.3 1.8 11.2 6.9 10.1 7.6 11.6 2.9 12.7 1.4 2.5 .4 2.5 .4 1.1 2.2 .4 .7 8.7 3.3 2.5 .7 2.2 1.4 .4 Missing ------100.0
4.3 6.2 17.4 24.3 34.4 42.0 53.6 56.5 69.2 70.7 73.2 73.6 76.1 76.4 77.5 79.7 80.1 80.8 89.5 92.8 95.3 96.0 98.2 99.6 100.0
Total Mean Std dev Valid cases
9.326 6.416 276
Median Minimum Missing cases
7.000 1.000
Mode Maximum 4
9.000 25.000
370 INGIN64
4th wanted source
Value Label Ag newspapers/magazi Other mass media Dir Gen recommendati BIP publications Warta/Jurnal Litbang Scientific journals AARD books University textbooks AARD technical guide Private technical gu Farmers Field extension work Extension subject-ma The boss Other colleagues Researchers Correspondents University courses/n Training Seminars etc. at res Field research proje Agribusinesses Own experience and o Own testing and rese
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 99
12 2 13 23 21 15 28 5 21 5 17 11 10 4 6 11 5 1 16 14 11 5 4 16 4 ------280
4.3 .7 4.6 8.2 7.5 5.4 10.0 1.8 7.5 1.8 6.1 3.9 3.6 1.4 2.1 3.9 1.8 .4 5.7 5.0 3.9 1.8 1.4 5.7 1.4 ------100.0
4.3 .7 4.7 8.3 7.6 5.4 10.1 1.8 7.6 1.8 6.2 4.0 3.6 1.4 2.2 4.0 1.8 .4 5.8 5.1 4.0 1.8 1.4 5.8 Missing ------100.0
4.3 5.1 9.8 18.1 25.7 31.2 41.3 43.1 50.7 52.5 58.7 62.7 66.3 67.8 69.9 73.9 75.7 76.1 81.9 87.0 90.9 92.8 94.2 100.0
Total Mean Std dev Valid cases
11.203 6.836 276
Median Minimum Missing cases
9.000 1.000
Mode Maximum 4
7.000 24.000
371 INGIN65
5th wanted source
Value Label Ag newspapers/magazi Other mass media Dir Gen recommendati BIP publications Warta/Jurnal Litbang Scientific journals AARD books University textbooks AARD technical guide Private technical gu Farmers Field extension work Extension subject-ma The boss Other colleagues Researchers Correspondents University courses/n Training Seminars etc. at res Field research proje Agribusinesses Own experience and o Own testing and rese Other sources
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 99
14 7 17 13 14 14 16 6 22 4 13 5 9 2 1 11 4 1 19 28 13 5 15 13 4 10 ------280
5.0 2.5 6.1 4.6 5.0 5.0 5.7 2.1 7.9 1.4 4.6 1.8 3.2 .7 .4 3.9 1.4 .4 6.8 10.0 4.6 1.8 5.4 4.6 1.4 3.6 ------100.0
5.2 2.6 6.3 4.8 5.2 5.2 5.9 2.2 8.1 1.5 4.8 1.9 3.3 .7 .4 4.1 1.5 .4 7.0 10.4 4.8 1.9 5.6 4.8 1.5 Missing ------100.0
5.2 7.8 14.1 18.9 24.1 29.3 35.2 37.4 45.6 47.0 51.9 53.7 57.0 57.8 58.1 62.2 63.7 64.1 71.1 81.5 86.3 88.1 93.7 98.5 100.0
Total Mean Std dev Valid cases
12.478 7.574 270
Median Minimum Missing cases
11.000 1.000 10
Mode Maximum
20.000 25.000
372 BACA71
Frequency read/speak to source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total BACA71 --------+--------+--------+--------+--------+ 1 | | | 2 | | 2 Never | | | | | .7 +--------+--------+--------+--------+ 2 | 3 | | 2 | | 5 | | | | | 1.8 +--------+--------+--------+--------+ 3 | 6 | | 10 | 7 | 23 Once a year | | | | | 8.3 +--------+--------+--------+--------+ 3 | 6 | 1 | 7 | 3 | 17 | | | | | 6.1 +--------+--------+--------+--------+ 4 | 16 | 8 | 25 | 24 | 73 Once a month | | | | | 26.3 +--------+--------+--------+--------+ 5 | 6 | 5 | 9 | 8 | 28 | | | | | 10.1 +--------+--------+--------+--------+ 6 | 12 | 28 | 12 | 19 | 71 Once a week | | | | | 25.5 +--------+--------+--------+--------+ 6 | 2 | 22 | 4 | 5 | 33 | | | | | 11.9 +--------+--------+--------+--------+ 7 | 1 | 16 | 4 | 5 | 26 Every day | | | | | 9.4 +--------+--------+--------+--------+ Column 52 80 75 71 278 Total 18.7 28.8 27.0 25.5 100.0 Number of Missing Observations: 2 Count
373 OLEH72
Frequency get info from source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total OLEH72 --------+--------+--------+--------+--------+ 1 | | | 2 | | 2 Never | | | | | .7 +--------+--------+--------+--------+ 2 | 6 | | 8 | 3 | 17 | | | | | 6.1 +--------+--------+--------+--------+ 3 | 9 | 4 | 21 | 8 | 42 Once a year | | | | | 15.2 +--------+--------+--------+--------+ 3 | 12 | 5 | 15 | 7 | 39 | | | | | 14.1 +--------+--------+--------+--------+ 4 | 20 | 18 | 22 | 27 | 87 Once a month | | | | | 31.4 +--------+--------+--------+--------+ 5 | 3 | 5 | 3 | 9 | 20 | | | | | 7.2 +--------+--------+--------+--------+ 6 | 1 | 27 | 4 | 13 | 45 Once a week | | | | | 16.2 +--------+--------+--------+--------+ 6 | 1 | 13 | | 3 | 17 | | | | | 6.1 +--------+--------+--------+--------+ 7 | | 7 | | 1 | 8 Every day | | | | | 2.9 +--------+--------+--------+--------+ Column 52 79 75 71 277 Total 18.8 28.5 27.1 25.6 100.0 Number of Missing Observations: 3 Count
374 GUNA73
Frequency use info from source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total GUNA73 --------+--------+--------+--------+--------+ 1 | 2 | 1 | 1 | 2 | 6 Never | | | | | 2.2 +--------+--------+--------+--------+ 2 | 4 | 2 | 5 | 3 | 14 | | | | | 5.1 +--------+--------+--------+--------+ 3 | 6 | 1 | 12 | 6 | 25 Once a year | | | | | 9.1 +--------+--------+--------+--------+ 3 | 12 | 8 | 14 | 9 | 43 | | | | | 15.6 +--------+--------+--------+--------+ 4 | 14 | 24 | 21 | 30 | 89 Once a month | | | | | 32.4 +--------+--------+--------+--------+ 5 | 5 | 18 | 8 | 5 | 36 | | | | | 13.1 +--------+--------+--------+--------+ 6 | 8 | 10 | 10 | 12 | 40 Once a week | | | | | 14.5 +--------+--------+--------+--------+ 6 | 1 | 10 | 3 | 1 | 15 | | | | | 5.5 +--------+--------+--------+--------+ 7 | | 5 | 1 | 1 | 7 Every day | | | | | 2.5 +--------+--------+--------+--------+ Column 52 79 75 69 275 Total 18.9 28.7 27.3 25.1 100.0 Number of Missing Observations: 5 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
MANF74
Usefulness of source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total MANF74 --------+--------+--------+--------+--------+ 1 | 1 | | 1 | | 2 Not useful | | | | | .7 +--------+--------+--------+--------+ 2 | 3 | | 1 | 2 | 6 | | | | | 2.2 +--------+--------+--------+--------+ 3 | 5 | 10 | 5 | 8 | 28 Somewhat useful | | | | | 10.1 +--------+--------+--------+--------+ 4 | 6 | 5 | 3 | 4 | 18 | | | | | 6.5 +--------+--------+--------+--------+ 5 | 23 | 30 | 26 | 35 | 114 Rather useful | | | | | 41.2 +--------+--------+--------+--------+ 6 | 4 | 12 | 7 | 10 | 33 | | | | | 11.9 +--------+--------+--------+--------+ 7 | 10 | 23 | 31 | 12 | 76 Very useful | | | | | 27.4 +--------+--------+--------+--------+ Column 52 80 74 71 277 Total 18.8 28.9 26.7 25.6 100.0 Number of Missing Observations: 3 Count
375 TEMP75
Usual location of source (R)
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total TEMP75 --------+--------+--------+--------+--------+ 0 | | | 1 | | 1 | | | | | .4 +--------+--------+--------+--------+ 1 | 8 | 6 | 8 | | 22 >150 km | | | | | 8.0 +--------+--------+--------+--------+ 2 | 3 | 1 | 2 | 1 | 7 <150 km | | | | | 2.5 +--------+--------+--------+--------+ 3 | 11 | 1 | 6 | 4 | 22 <50 km | | | | | 8.0 +--------+--------+--------+--------+ 4 | 4 | 13 | 6 | 25 | 48 Same town | | | | | 17.4 +--------+--------+--------+--------+ 5 | 5 | 7 | 6 | 4 | 22 Other building | | | | | 8.0 +--------+--------+--------+--------+ 6 | 6 | 7 | 14 | 14 | 41 Same building | | | | | 14.9 +--------+--------+--------+--------+ 7 | 14 | 45 | 31 | 23 | 113 Own office | | | | | 40.9 +--------+--------+--------+--------+ Column 51 80 74 71 276 Total 18.5 29.0 26.8 25.7 100.0 Number of Missing Observations: 4 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
SES81
Source AGREES with other sources (R)
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total SES81 --------+--------+--------+--------+--------+ 0 | 1 | | 4 | 1 | 6 Dont know | | | | | 2.2 +--------+--------+--------+--------+ 1 | 1 | 2 | 1 | 1 | 5 Disagree | | | | | 1.8 +--------+--------+--------+--------+ 2 | 2 | 1 | | 1 | 4 | | | | | 1.5 +--------+--------+--------+--------+ 3 | 6 | 11 | 7 | 4 | 28 Somewhat disagre | | | | | 10.3 +--------+--------+--------+--------+ 4 | 1 | 3 | 3 | 1 | 8 | | | | | 2.9 +--------+--------+--------+--------+ 5 | 21 | 34 | 26 | 27 | 108 Somewhat agree | | | | | 39.7 +--------+--------+--------+--------+ 6 | 5 | 6 | 7 | 6 | 24 | | | | | 8.8 +--------+--------+--------+--------+ 7 | 13 | 19 | 28 | 29 | 89 Agree | | | | | 32.7 +--------+--------+--------+--------+ Column 50 76 76 70 272 Total 18.4 27.9 27.9 25.7 100.0 Number of Missing Observations: 8 Count
376 TEMU82
Source is EASY to find (R)
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total TEMU82 --------+--------+--------+--------+--------+ 0 | | 1 | | | 1 Dont know | | | | | .4 +--------+--------+--------+--------+ 1 | 2 | 5 | 9 | 2 | 18 Disagree | | | | | 6.5 +--------+--------+--------+--------+ 2 | 2 | 3 | 6 | | 11 | | | | | 4.0 +--------+--------+--------+--------+ 3 | 4 | 3 | 13 | 1 | 21 Somewhat disagre | | | | | 7.6 +--------+--------+--------+--------+ 4 | | 2 | 7 | 2 | 11 | | | | | 4.0 +--------+--------+--------+--------+ 5 | 23 | 16 | 17 | 12 | 68 Somewhat agree | | | | | 24.6 +--------+--------+--------+--------+ 6 | 3 | 4 | 4 | 5 | 16 | | | | | 5.8 +--------+--------+--------+--------+ 7 | 18 | 45 | 19 | 48 | 130 Agree | | | | | 47.1 +--------+--------+--------+--------+ Column 52 79 75 70 276 Total 18.8 28.6 27.2 25.4 100.0 Number of Missing Observations: 4 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
REL83
Source is locally RELEVANT (R)
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total REL83 --------+--------+--------+--------+--------+ 0 | | | 1 | 1 | 2 Dont know | | | | | .7 +--------+--------+--------+--------+ 1 | 3 | 8 | 5 | 2 | 18 Disagree | | | | | 6.5 +--------+--------+--------+--------+ 2 | 1 | 7 | 2 | 1 | 11 | | | | | 4.0 +--------+--------+--------+--------+ 3 | 12 | 20 | 22 | 3 | 57 Somewhat disagre | | | | | 20.6 +--------+--------+--------+--------+ 4 | 5 | 1 | 7 | 2 | 15 | | | | | 5.4 +--------+--------+--------+--------+ 5 | 22 | 27 | 25 | 25 | 99 Somewhat agree | | | | | 35.7 +--------+--------+--------+--------+ 6 | 3 | 3 | 3 | 9 | 18 | | | | | 6.5 +--------+--------+--------+--------+ 7 | 6 | 13 | 10 | 28 | 57 Agree | | | | | 20.6 +--------+--------+--------+--------+ Column 52 79 75 71 277 Total 18.8 28.5 27.1 25.6 100.0 Number of Missing Observations: 3 Count
377 CAYA84
Source most credible
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total CAYA84 --------+--------+--------+--------+--------+ 0 | | | 1 | 1 | 2 Dont know | | | | | .7 +--------+--------+--------+--------+ 1 | 3 | 10 | 2 | 11 | 26 Disagree | | | | | 9.4 +--------+--------+--------+--------+ 2 | 3 | 2 | 1 | 4 | 10 | | | | | 3.6 +--------+--------+--------+--------+ 3 | 15 | 23 | 12 | 29 | 79 Somewhat disagre | | | | | 28.4 +--------+--------+--------+--------+ 4 | 7 | 6 | 2 | 3 | 18 | | | | | 6.5 +--------+--------+--------+--------+ 5 | 12 | 21 | 26 | 16 | 75 Somewhat agree | | | | | 27.0 +--------+--------+--------+--------+ 6 | 5 | 8 | 8 | 3 | 24 | | | | | 8.6 +--------+--------+--------+--------+ 7 | 7 | 9 | 24 | 4 | 44 Agree | | | | | 15.8 +--------+--------+--------+--------+ Column 52 79 76 71 278 Total 18.7 28.4 27.3 25.5 100.0 Number of Missing Observations: 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
MUDAH85
Source easy to use in extension
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total MUDAH85 --------+--------+--------+--------+--------+ 0 | | | 1 | | 1 Dont know | | | | | .4 +--------+--------+--------+--------+ 1 | 1 | 1 | 5 | 5 | 12 Disagree | | | | | 4.3 +--------+--------+--------+--------+ 2 | 2 | 1 | | 2 | 5 | | | | | 1.8 +--------+--------+--------+--------+ 3 | 9 | 16 | 16 | 18 | 59 Somewhat disagre | | | | | 21.1 +--------+--------+--------+--------+ 4 | 5 | 13 | 8 | 10 | 36 | | | | | 12.9 +--------+--------+--------+--------+ 5 | 16 | 21 | 30 | 25 | 92 Somewhat agree | | | | | 33.0 +--------+--------+--------+--------+ 6 | 7 | 9 | 7 | 8 | 31 | | | | | 11.1 +--------+--------+--------+--------+ 7 | 12 | 19 | 9 | 3 | 43 Agree | | | | | 15.4 +--------+--------+--------+--------+ Column 52 80 76 71 279 Total 18.6 28.7 27.2 25.4 100.0 Number of Missing Observations: 1 Count
378 TEPAT86
Source timely
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total TEPAT86 --------+--------+--------+--------+--------+ 0 | | | 1 | 2 | 3 Dont know | | | | | 1.1 +--------+--------+--------+--------+ 1 | 1 | 3 | 4 | 5 | 13 Disagree | | | | | 4.7 +--------+--------+--------+--------+ 2 | 2 | 2 | 1 | 3 | 8 | | | | | 2.9 +--------+--------+--------+--------+ 3 | 16 | 12 | 17 | 17 | 62 Somewhat disagre | | | | | 22.4 +--------+--------+--------+--------+ 4 | 11 | 12 | 12 | 6 | 41 | | | | | 14.8 +--------+--------+--------+--------+ 5 | 12 | 26 | 28 | 24 | 90 Somewhat agree | | | | | 32.5 +--------+--------+--------+--------+ 6 | 4 | 10 | 8 | 9 | 31 | | | | | 11.2 +--------+--------+--------+--------+ 7 | 6 | 15 | 5 | 3 | 29 Agree | | | | | 10.5 +--------+--------+--------+--------+ Column 52 80 76 69 277 Total 18.8 28.9 27.4 24.9 100.0 Number of Missing Observations: 3 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
TEN87
Devote effort to use source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total TEN87 --------+--------+--------+--------+--------+ 0 | | 1 | | | 1 Dont know | | | | | .4 +--------+--------+--------+--------+ 1 | 11 | 12 | 11 | 22 | 56 Disagree | | | | | 20.4 +--------+--------+--------+--------+ 2 | 7 | 10 | 4 | 5 | 26 | | | | | 9.5 +--------+--------+--------+--------+ 3 | 16 | 28 | 14 | 24 | 82 Somewhat disagre | | | | | 29.9 +--------+--------+--------+--------+ 4 | 3 | 6 | 10 | 6 | 25 | | | | | 9.1 +--------+--------+--------+--------+ 5 | 6 | 11 | 18 | 6 | 41 Somewhat agree | | | | | 15.0 +--------+--------+--------+--------+ 6 | 3 | 7 | 6 | 1 | 17 | | | | | 6.2 +--------+--------+--------+--------+ 7 | 4 | 5 | 12 | 5 | 26 Agree | | | | | 9.5 +--------+--------+--------+--------+ Column 50 80 75 69 274 Total 18.2 29.2 27.4 25.2 100.0 Number of Missing Observations: 6 Count
379 ERAT88
Source relevant to local conditions
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total ERAT88 --------+--------+--------+--------+--------+ 0 | | 1 | 2 | 1 | 4 Dont know | | | | | 1.4 +--------+--------+--------+--------+ 1 | 1 | 6 | 11 | 16 | 34 Disagree | | | | | 12.3 +--------+--------+--------+--------+ 2 | 4 | 6 | 1 | 9 | 20 | | | | | 7.2 +--------+--------+--------+--------+ 3 | 10 | 29 | 30 | 13 | 82 Somewhat disagre | | | | | 29.6 +--------+--------+--------+--------+ 4 | 9 | 10 | 4 | 8 | 31 | | | | | 11.2 +--------+--------+--------+--------+ 5 | 17 | 17 | 13 | 15 | 62 Somewhat agree | | | | | 22.4 +--------+--------+--------+--------+ 6 | 6 | 7 | 9 | 5 | 27 | | | | | 9.7 +--------+--------+--------+--------+ 7 | 5 | 4 | 5 | 3 | 17 Agree | | | | | 6.1 +--------+--------+--------+--------+ Column 52 80 75 70 277 Total 18.8 28.9 27.1 25.3 100.0 Number of Missing Observations: 3 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
SIAP89
Source info ready to use
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total SIAP89 --------+--------+--------+--------+--------+ 0 | | | 1 | 2 | 3 Dont know | | | | | 1.1 +--------+--------+--------+--------+ 1 | | 4 | 10 | 18 | 32 Disagree | | | | | 11.7 +--------+--------+--------+--------+ 2 | 3 | 9 | 6 | 4 | 22 | | | | | 8.0 +--------+--------+--------+--------+ 3 | 4 | 21 | 24 | 16 | 65 Somewhat disagre | | | | | 23.7 +--------+--------+--------+--------+ 4 | 10 | 9 | 6 | 8 | 33 | | | | | 12.0 +--------+--------+--------+--------+ 5 | 19 | 23 | 16 | 17 | 75 Somewhat agree | | | | | 27.4 +--------+--------+--------+--------+ 6 | 6 | 5 | 6 | 1 | 18 | | | | | 6.6 +--------+--------+--------+--------+ 7 | 9 | 8 | 7 | 2 | 26 Agree | | | | | 9.5 +--------+--------+--------+--------+ Column 51 79 76 68 274 Total 18.6 28.8 27.7 24.8 100.0 Number of Missing Observations: 6 Count
380 LENG810
Source info complete
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total LENG810 --------+--------+--------+--------+--------+ 0 | 1 | | 1 | 4 | 6 Dont know | | | | | 2.2 +--------+--------+--------+--------+ 1 | 11 | 16 | 8 | 24 | 59 Disagree | | | | | 21.1 +--------+--------+--------+--------+ 2 | 3 | 6 | 4 | 8 | 21 | | | | | 7.5 +--------+--------+--------+--------+ 3 | 25 | 38 | 25 | 28 | 116 Somewhat disagre | | | | | 41.6 +--------+--------+--------+--------+ 4 | 5 | 7 | 13 | 5 | 30 | | | | | 10.8 +--------+--------+--------+--------+ 5 | 5 | 7 | 13 | 1 | 26 Somewhat agree | | | | | 9.3 +--------+--------+--------+--------+ 6 | 1 | 4 | 7 | 1 | 13 | | | | | 4.7 +--------+--------+--------+--------+ 7 | 1 | 2 | 5 | | 8 Agree | | | | | 2.9 +--------+--------+--------+--------+ Column 52 80 76 71 279 Total 18.6 28.7 27.2 25.4 100.0 Number of Missing Observations: 1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
TAHU811
Know info held by source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total TAHU811 --------+--------+--------+--------+--------+ 0 | 6 | 2 | 3 | 14 | 25 Dont know | | | | | 9.4 +--------+--------+--------+--------+ 1 | 1 | | 4 | 4 | 9 Disagree | | | | | 3.4 +--------+--------+--------+--------+ 2 | 1 | 2 | 3 | 3 | 9 | | | | | 3.4 +--------+--------+--------+--------+ 3 | 10 | 9 | 8 | 11 | 38 Somewhat disagre | | | | | 14.3 +--------+--------+--------+--------+ 4 | 3 | 6 | 6 | 9 | 24 | | | | | 9.1 +--------+--------+--------+--------+ 5 | 15 | 24 | 26 | 21 | 86 Somewhat agree | | | | | 32.5 +--------+--------+--------+--------+ 6 | 7 | 11 | 11 | 5 | 34 | | | | | 12.8 +--------+--------+--------+--------+ 7 | 8 | 20 | 10 | 2 | 40 Agree | | | | | 15.1 +--------+--------+--------+--------+ Column 51 74 71 69 265 Total 19.2 27.9 26.8 26.0 100.0 Number of Missing Observations: 15 Count
381 LAMA812
Source info IS NEW (R)
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total LAMA812 --------+--------+--------+--------+--------+ 0 | | | 1 | 3 | 4 Dont know | | | | | 1.5 +--------+--------+--------+--------+ 1 | 5 | 7 | 2 | 5 | 19 Disagree | | | | | 6.9 +--------+--------+--------+--------+ 2 | 6 | 7 | 5 | 2 | 20 | | | | | 7.3 +--------+--------+--------+--------+ 3 | 13 | 20 | 11 | 11 | 55 Somewhat disagre | | | | | 20.1 +--------+--------+--------+--------+ 4 | 4 | 8 | 8 | 2 | 22 | | | | | 8.0 +--------+--------+--------+--------+ 5 | 17 | 23 | 28 | 30 | 98 Somewhat agree | | | | | 35.8 +--------+--------+--------+--------+ 6 | 1 | 4 | 6 | 5 | 16 | | | | | 5.8 +--------+--------+--------+--------+ 7 | 6 | 10 | 14 | 10 | 40 Agree | | | | | 14.6 +--------+--------+--------+--------+ Column 52 79 75 68 274 Total 19.0 28.8 27.4 24.8 100.0 Number of Missing Observations: 6 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
TUGAS813
Part of job to get info from source
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total TUGAS813 --------+--------+--------+--------+--------+ 0 | | | 1 | | 1 Dont know | | | | | .4 +--------+--------+--------+--------+ 1 | 4 | 3 | 1 | 3 | 11 Disagree | | | | | 4.0 +--------+--------+--------+--------+ 2 | 2 | 1 | 1 | 3 | 7 | | | | | 2.5 +--------+--------+--------+--------+ 3 | 9 | 17 | 3 | 12 | 41 Somewhat disagre | | | | | 14.7 +--------+--------+--------+--------+ 4 | 5 | 10 | 1 | 13 | 29 | | | | | 10.4 +--------+--------+--------+--------+ 5 | 15 | 15 | 16 | 22 | 68 Somewhat agree | | | | | 24.5 +--------+--------+--------+--------+ 6 | 4 | 11 | 9 | 4 | 28 | | | | | 10.1 +--------+--------+--------+--------+ 7 | 13 | 23 | 43 | 14 | 93 Agree | | | | | 33.5 +--------+--------+--------+--------+ Column 52 80 75 71 278 Total 18.7 28.8 27.0 25.5 100.0 Number of Missing Observations: 2 Count
382 KENAL814
Know source well
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total KENAL814 --------+--------+--------+--------+--------+ 0 | | | 1 | 1 | 2 Dont know | | | | | .7 +--------+--------+--------+--------+ 1 | | 2 | 1 | | 3 Disagree | | | | | 1.1 +--------+--------+--------+--------+ 2 | 2 | 1 | 1 | | 4 | | | | | 1.4 +--------+--------+--------+--------+ 3 | 4 | 4 | 5 | | 13 Somewhat disagre | | | | | 4.7 +--------+--------+--------+--------+ 4 | 7 | 6 | 3 | 1 | 17 | | | | | 6.1 +--------+--------+--------+--------+ 5 | 19 | 23 | 25 | 10 | 77 Somewhat agree | | | | | 27.7 +--------+--------+--------+--------+ 6 | 8 | 11 | 13 | 8 | 40 | | | | | 14.4 +--------+--------+--------+--------+ 7 | 12 | 32 | 27 | 51 | 122 Agree | | | | | 43.9 +--------+--------+--------+--------+ Column 52 79 76 71 278 Total 18.7 28.4 27.3 25.5 100.0 Number of Missing Observations: 2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Count
NGERT815
Source easy to understand
by
SOURCE
Source type
SOURCE | |BIP publ Ag press AARD pub Other SM |ications lication Ss Row | 1.00| 2.00| 3.00| 4.00| Total NGERT815 --------+--------+--------+--------+--------+ 0 | | 1 | 1 | | 2 Dont know | | | | | .7 +--------+--------+--------+--------+ 1 | | 1 | 3 | 2 | 6 Disagree | | | | | 2.2 +--------+--------+--------+--------+ 2 | | 1 | | 4 | 5 | | | | | 1.8 +--------+--------+--------+--------+ 3 | 8 | 12 | 17 | 16 | 53 Somewhat disagre | | | | | 19.0 +--------+--------+--------+--------+ 4 | 6 | 4 | 8 | 6 | 24 | | | | | 8.6 +--------+--------+--------+--------+ 5 | 19 | 25 | 24 | 26 | 94 Somewhat agree | | | | | 33.7 +--------+--------+--------+--------+ 6 | 10 | 15 | 12 | 6 | 43 | | | | | 15.4 +--------+--------+--------+--------+ 7 | 9 | 21 | 11 | 11 | 52 Agree | | | | | 18.6 +--------+--------+--------+--------+ Column 52 80 76 71 279 Total 18.6 28.7 27.2 25.4 100.0 Number of Missing Observations: 1 Count
383 PETANI91
Visits to farmers in 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19 20 21 22 24 25 29 30 32 36 40 45 48 50 60 67 87 99
5 11 10 29 16 12 35 3 12 9 14 5 32 3 19 8 1 2 1 4 2 1 10 1 1 4 1 6 1 1 4 2 3 1 1 10 ------280
1.8 3.9 3.6 10.4 5.7 4.3 12.5 1.1 4.3 3.2 5.0 1.8 11.4 1.1 6.8 2.9 .4 .7 .4 1.4 .7 .4 3.6 .4 .4 1.4 .4 2.1 .4 .4 1.4 .7 1.1 .4 .4 3.6 ------100.0
1.9 4.1 3.7 10.7 5.9 4.4 13.0 1.1 4.4 3.3 5.2 1.9 11.9 1.1 7.0 3.0 .4 .7 .4 1.5 .7 .4 3.7 .4 .4 1.5 .4 2.2 .4 .4 1.5 .7 1.1 .4 .4 Missing ------100.0
1.9 5.9 9.6 20.4 26.3 30.7 43.7 44.8 49.3 52.6 57.8 59.6 71.5 72.6 79.6 82.6 83.0 83.7 84.1 85.6 86.3 86.7 90.4 90.7 91.1 92.6 93.0 95.2 95.6 95.9 97.4 98.1 99.3 99.6 100.0
Total Mean Std dev Valid cases
12.122 12.526 270
Median Minimum Missing cases
9.000 .000 10
Mode Maximum
6.000 87.000
384 BPP92
Visits to BPP in 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 11 12 14 15 16 18 20 21 22 24 25 28 30 36 77 99
3 11 15 30 20 14 42 9 13 10 16 2 37 2 11 6 4 6 3 2 6 2 1 2 4 1 8 ------280
1.1 3.9 5.4 10.7 7.1 5.0 15.0 3.2 4.6 3.6 5.7 .7 13.2 .7 3.9 2.1 1.4 2.1 1.1 .7 2.1 .7 .4 .7 1.4 .4 2.9 ------100.0
1.1 4.0 5.5 11.0 7.4 5.1 15.4 3.3 4.8 3.7 5.9 .7 13.6 .7 4.0 2.2 1.5 2.2 1.1 .7 2.2 .7 .4 .7 1.5 .4 Missing ------100.0
1.1 5.1 10.7 21.7 29.0 34.2 49.6 52.9 57.7 61.4 67.3 68.0 81.6 82.4 86.4 88.6 90.1 92.3 93.4 94.1 96.3 97.1 97.4 98.2 99.6 100.0
Total Mean Std dev
9.217 8.027
Valid cases
272
Median Minimum
7.000 .000
Missing cases
Mode Maximum
6.000 77.000
8
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - CARI93
Seek answers to questions in 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 11 12 15 16 18 20 24 26 30 32 36 40 48 60 99
34 21 41 36 21 15 23 4 8 6 14 2 13 9 2 1 3 5 1 1 1 1 1 1 1 15 ------280
12.1 7.5 14.6 12.9 7.5 5.4 8.2 1.4 2.9 2.1 5.0 .7 4.6 3.2 .7 .4 1.1 1.8 .4 .4 .4 .4 .4 .4 .4 5.4 ------100.0
12.8 7.9 15.5 13.6 7.9 5.7 8.7 1.5 3.0 2.3 5.3 .8 4.9 3.4 .8 .4 1.1 1.9 .4 .4 .4 .4 .4 .4 .4 Missing ------100.0
12.8 20.8 36.2 49.8 57.7 63.4 72.1 73.6 76.6 78.9 84.2 84.9 89.8 93.2 94.0 94.3 95.5 97.4 97.7 98.1 98.5 98.9 99.2 99.6 100.0
Total
385 Mean Std dev Valid cases
6.072 7.620 265
Median Minimum Missing cases
4.000 .000 15
Mode Maximum
2.000 60.000
386 MAJ94
Read sci journals in 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 13 14 15 18 19 20 24 25 26 30 32 35 36 38 50 68 70 200 285 999
28 20 28 43 17 21 23 3 8 4 20 13 1 2 8 3 1 6 2 3 1 4 1 1 1 1 2 1 1 1 1 11 ------280
10.0 7.1 10.0 15.4 6.1 7.5 8.2 1.1 2.9 1.4 7.1 4.6 .4 .7 2.9 1.1 .4 2.1 .7 1.1 .4 1.4 .4 .4 .4 .4 .7 .4 .4 .4 .4 3.9 ------100.0
10.4 7.4 10.4 16.0 6.3 7.8 8.6 1.1 3.0 1.5 7.4 4.8 .4 .7 3.0 1.1 .4 2.2 .7 1.1 .4 1.5 .4 .4 .4 .4 .7 .4 .4 .4 .4 Missing ------100.0
10.4 17.8 28.3 44.2 50.6 58.4 66.9 68.0 71.0 72.5 79.9 84.8 85.1 85.9 88.8 90.0 90.3 92.6 93.3 94.4 94.8 96.3 96.7 97.0 97.4 97.8 98.5 98.9 99.3 99.6 100.0
Total Mean Std dev Valid cases
9.219 22.714 269
Median Minimum Missing cases
4.000 .000 11
Mode Maximum
3.000 285.000
387 BUKU95
Read AARD books in 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 13 14 15 16 20 21 25 30 33 45 99
38 55 47 40 18 16 13 2 3 3 12 5 1 3 4 1 3 1 1 2 1 1 10 ------280
13.6 19.6 16.8 14.3 6.4 5.7 4.6 .7 1.1 1.1 4.3 1.8 .4 1.1 1.4 .4 1.1 .4 .4 .7 .4 .4 3.6 ------100.0
14.1 20.4 17.4 14.8 6.7 5.9 4.8 .7 1.1 1.1 4.4 1.9 .4 1.1 1.5 .4 1.1 .4 .4 .7 .4 .4 Missing ------100.0
14.1 34.4 51.9 66.7 73.3 79.3 84.1 84.8 85.9 87.0 91.5 93.3 93.7 94.8 96.3 96.7 97.8 98.1 98.5 99.3 99.6 100.0
Total Mean Std dev
4.144 5.676
Valid cases
270
Median Minimum
2.000 .000
Missing cases
Mode Maximum
1.000 45.000
10
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UJI101
Field technology tests in last year
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 8 12 16 20 24 99
76 75 80 13 13 4 8 1 1 2 1 1 5 ------280
27.1 26.8 28.6 4.6 4.6 1.4 2.9 .4 .4 .7 .4 .4 1.8 ------100.0
27.6 27.3 29.1 4.7 4.7 1.5 2.9 .4 .4 .7 .4 .4 Missing ------100.0
27.6 54.9 84.0 88.7 93.5 94.9 97.8 98.2 98.5 99.3 99.6 100.0
Total Mean Std dev Valid cases
1.782 2.655 275
Median Minimum Missing cases
1.000 .000
Mode Maximum 5
2.000 24.000
388 LUAR102
Trips to seek info in last year
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 13 15 16 17 18 20 21 24 26 36 46 70 99
49 41 41 27 22 12 17 5 6 2 13 17 1 6 2 1 1 3 1 1 1 1 1 1 8 ------280
17.5 14.6 14.6 9.6 7.9 4.3 6.1 1.8 2.1 .7 4.6 6.1 .4 2.1 .7 .4 .4 1.1 .4 .4 .4 .4 .4 .4 2.9 ------100.0
18.0 15.1 15.1 9.9 8.1 4.4 6.3 1.8 2.2 .7 4.8 6.3 .4 2.2 .7 .4 .4 1.1 .4 .4 .4 .4 .4 .4 Missing ------100.0
18.0 33.1 48.2 58.1 66.2 70.6 76.8 78.7 80.9 81.6 86.4 92.6 93.0 95.2 96.0 96.3 96.7 97.8 98.2 98.5 98.9 99.3 99.6 100.0
Total Mean Std dev
4.934 7.052
Valid cases
272
Median Minimum
3.000 .000
Missing cases
Mode Maximum
.000 70.000
8
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LATIH103
BPLP training courses in last year
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 99
155 89 22 7 1 6 ------280
55.4 31.8 7.9 2.5 .4 2.1 ------100.0
56.6 32.5 8.0 2.6 .4 Missing ------100.0
56.6 89.1 97.1 99.6 100.0
Total Mean Std dev
.577 .777
Median Minimum
.000 .000
Mode Maximum
.000 4.000
Valid cases 274 Missing cases 6 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LATIH111
AARD training courses in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 99
188 48 19 9 3 3 10 ------280
67.1 17.1 6.8 3.2 1.1 1.1 3.6 ------100.0
69.6 17.8 7.0 3.3 1.1 1.1 Missing ------100.0
69.6 87.4 94.4 97.8 98.9 100.0
Total Mean
.519
Median
.000
Mode
.000
389 Std dev Valid cases
.974 270
Minimum Missing cases
.000 10
Maximum
5.000
390 PROY112
Collaborate in AARD research in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 6 20 99
209 37 5 8 5 2 1 13 ------280
74.6 13.2 1.8 2.9 1.8 .7 .4 4.6 ------100.0
78.3 13.9 1.9 3.0 1.9 .7 .4 Missing ------100.0
78.3 92.1 94.0 97.0 98.9 99.6 100.0
Total Mean Std dev
.461 1.532
Valid cases
267
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 20.000
13
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PAMER113
Exhibits/seminars attended in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 8 9 10 12 15 20 99
106 62 38 28 10 5 10 3 1 5 1 1 1 9 ------280
37.9 22.1 13.6 10.0 3.6 1.8 3.6 1.1 .4 1.8 .4 .4 .4 3.2 ------100.0
39.1 22.9 14.0 10.3 3.7 1.8 3.7 1.1 .4 1.8 .4 .4 .4 Missing ------100.0
39.1 62.0 76.0 86.3 90.0 91.9 95.6 96.7 97.0 98.9 99.3 99.6 100.0
Total Mean Std dev Valid cases
1.760 2.592 271
Median Minimum Missing cases
1.000 .000
Mode Maximum 9
.000 20.000
391 RAPAT114
Tech meets with researchers in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 15 20 21 27 30 99
48 54 50 52 24 8 19 2 1 3 3 3 1 1 1 1 2 7 ------280
17.1 19.3 17.9 18.6 8.6 2.9 6.8 .7 .4 1.1 1.1 1.1 .4 .4 .4 .4 .7 2.5 ------100.0
17.6 19.8 18.3 19.0 8.8 2.9 7.0 .7 .4 1.1 1.1 1.1 .4 .4 .4 .4 .7 Missing ------100.0
17.6 37.4 55.7 74.7 83.5 86.4 93.4 94.1 94.5 95.6 96.7 97.8 98.2 98.5 98.9 99.3 100.0
Total Mean Std dev
2.996 3.963
Valid cases
273
Median Minimum
2.000 .000
Missing cases
Mode Maximum
1.000 30.000
7
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KUNJ115
Visits to or by researchers in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 8 9 10 12 15 20 30 99
80 70 44 23 17 8 9 2 1 2 4 2 2 1 1 14 ------280
28.6 25.0 15.7 8.2 6.1 2.9 3.2 .7 .4 .7 1.4 .7 .7 .4 .4 5.0 ------100.0
30.1 26.3 16.5 8.6 6.4 3.0 3.4 .8 .4 .8 1.5 .8 .8 .4 .4 Missing ------100.0
30.1 56.4 72.9 81.6 88.0 91.0 94.4 95.1 95.5 96.2 97.7 98.5 99.2 99.6 100.0
Total Mean Std dev Valid cases
2.154 3.237 266
Median Minimum Missing cases
1.000 .000 14
Mode Maximum
.000 30.000
392 SURAT116
Letters to researchers in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 15 99
195 35 25 4 5 2 3 1 10 ------280
69.6 12.5 8.9 1.4 1.8 .7 1.1 .4 3.6 ------100.0
72.2 13.0 9.3 1.5 1.9 .7 1.1 .4 Missing ------100.0
72.2 85.2 94.4 95.9 97.8 98.5 99.6 100.0
Total Mean Std dev
.593 1.416
Valid cases
270
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 15.000
10
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UMPAN117
Feedback given to researchers in 3 years
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 8 9 10 11 12 99
131 38 43 20 11 6 10 2 4 2 1 3 9 ------280
46.8 13.6 15.4 7.1 3.9 2.1 3.6 .7 1.4 .7 .4 1.1 3.2 ------100.0
48.3 14.0 15.9 7.4 4.1 2.2 3.7 .7 1.5 .7 .4 1.1 Missing ------100.0
48.3 62.4 78.2 85.6 89.7 91.9 95.6 96.3 97.8 98.5 98.9 100.0
Total Mean Std dev
1.613 2.416
Valid cases
271
Median Minimum
1.000 .000
Missing cases
Mode Maximum
.000 12.000
9
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UJI121
Recommendations from above must be teste
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 3 4 5 6 7 9
4 6 2 31 32 203 2 ------280
1.4 2.1 .7 11.1 11.4 72.5 .7 ------100.0
1.4 2.2 .7 11.2 11.5 73.0 Missing ------100.0
1.4 3.6 4.3 15.5 27.0 100.0
Disagree Somewhat disagree Somewhat agree Agree
Total Mean Std dev Valid cases
6.468 1.100 278
Median Minimum Missing cases
7.000 1.000
Mode Maximum 2
7.000 7.000
393 UBAH122
Recommendations from above cannot be cha
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
77 13 107 12 22 10 29 10 ------280
27.5 4.6 38.2 4.3 7.9 3.6 10.4 3.6 ------100.0
28.5 4.8 39.6 4.4 8.1 3.7 10.7 Missing ------100.0
28.5 33.3 73.0 77.4 85.6 89.3 100.0
Disagree Somewhat disagree Somewhat agree Agree
Total Mean Std dev Valid cases
3.130 1.899 270
Median Minimum Missing cases
3.000 1.000 10
Mode Maximum
3.000 7.000
394
395
APPENDIX 6 RESPONSE FREQUENCIES FROM PUBLICATIONS QUESTIONNAIRE
396 LEVEL21
Administrative level
Value Label Province BIP District
Value
Frequency
Percent
Valid Percent
Cum Percent
1.00 2.00 3.00
36 56 73 ------165
21.8 33.9 44.2 ------100.0
21.8 33.9 44.2 ------100.0
21.8 55.8 100.0
Total Mean Std dev
2.224 .784
Valid cases
165
Median Minimum
2.000 1.000
Missing cases
Mode Maximum
3.000 3.000
0
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - INST21
Institutional affiliation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7
1 56 15 38 19 35 1 ------165
.6 33.9 9.1 23.0 11.5 21.2 .6 ------100.0
.6 33.9 9.1 23.0 11.5 21.2 .6 ------100.0
.6 34.5 43.6 66.7 78.2 99.4 100.0
Kanwil BIP Dinas Tk 1 Dinas Tk 2 SPH Bimas Tk 1 SPH Bimas Tk 2 Other
Total Mean Std dev
3.770 1.576
Valid cases
165
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
2.000 7.000
0
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SPEC22
Specialization
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 7 8 9
3 100 35 10 3 8 2 4 ------165
1.8 60.6 21.2 6.1 1.8 4.8 1.2 2.4 ------100.0
1.9 62.1 21.7 6.2 1.9 5.0 1.2 Missing ------100.0
1.9 64.0 85.7 91.9 93.8 98.8 100.0
Vet med Food crops Livestock Estate crops Social/ econ Fish Conservation, other
Total Mean Std dev Valid cases
1.764 1.575 161
Median Minimum Missing cases
1.000 .000
Mode Maximum 4
1.000 8.000
397 LAMA23
Years as PPS
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 99
3 9 7 23 33 23 19 19 5 5 1 3 2 1 1 1 2 3 2 1 2 ------165
1.8 5.5 4.2 13.9 20.0 13.9 11.5 11.5 3.0 3.0 .6 1.8 1.2 .6 .6 .6 1.2 1.8 1.2 .6 1.2 ------100.0
1.8 5.5 4.3 14.1 20.2 14.1 11.7 11.7 3.1 3.1 .6 1.8 1.2 .6 .6 .6 1.2 1.8 1.2 .6 Missing ------100.0
1.8 7.4 11.7 25.8 46.0 60.1 71.8 83.4 86.5 89.6 90.2 92.0 93.3 93.9 94.5 95.1 96.3 98.2 99.4 100.0
Total Mean Std dev
6.650 3.788
Valid cases
163
Median Minimum
6.000 1.000
Missing cases
Mode Maximum
5.000 21.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SEX24
Sex
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
118 46 1 ------165
71.5 27.9 .6 ------100.0
72.0 28.0 Missing ------100.0
72.0 100.0
Male Female
Total Mean Std dev
.280 .451
Valid cases
164
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SEDTEK31
Technology availability
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 6 12 9 51 19 62 3 ------165
1.8 3.6 7.3 5.5 30.9 11.5 37.6 1.8 ------100.0
1.9 3.7 7.4 5.6 31.5 11.7 38.3 Missing ------100.0
1.9 5.6 13.0 18.5 50.0 61.7 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
5.494 1.549 162
Median Minimum Missing cases
5.500 1.000
Mode Maximum 3
7.000 7.000
398 OLEH32
Information availability
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 1 11 7 51 24 67 1 ------165
1.8 .6 6.7 4.2 30.9 14.5 40.6 .6 ------100.0
1.8 .6 6.7 4.3 31.1 14.6 40.9 Missing ------100.0
1.8 2.4 9.1 13.4 44.5 59.1 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.695 1.403
Valid cases
164
Median Minimum
6.000 1.000
Missing cases
Mode Maximum
7.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UMPAN33
Feedback to research
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
4 1 13 15 63 20 47 2 ------165
2.4 .6 7.9 9.1 38.2 12.1 28.5 1.2 ------100.0
2.5 .6 8.0 9.2 38.7 12.3 28.8 Missing ------100.0
2.5 3.1 11.0 20.2 58.9 71.2 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.331 1.427
Valid cases
163
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
5.000 7.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - TERTEK34
Technical skills
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
4 11 15 9 49 21 53 3 ------165
2.4 6.7 9.1 5.5 29.7 12.7 32.1 1.8 ------100.0
2.5 6.8 9.3 5.6 30.2 13.0 32.7 Missing ------100.0
2.5 9.3 18.5 24.1 54.3 67.3 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
5.241 1.682 162
Median Minimum Missing cases
5.000 1.000
Mode Maximum 3
7.000 7.000
399 TERSUL35
Extension skills
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
15 7 13 2 47 18 62 1 ------165
9.1 4.2 7.9 1.2 28.5 10.9 37.6 .6 ------100.0
9.1 4.3 7.9 1.2 28.7 11.0 37.8 Missing ------100.0
9.1 13.4 21.3 22.6 51.2 62.2 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.201 1.945
Valid cases
164
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
7.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MOB36
Mobility
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 6 8 3 45 26 73 1 ------165
1.8 3.6 4.8 1.8 27.3 15.8 44.2 .6 ------100.0
1.8 3.7 4.9 1.8 27.4 15.9 44.5 Missing ------100.0
1.8 5.5 10.4 12.2 39.6 55.5 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.750 1.492
Valid cases
164
Median Minimum
6.000 1.000
Missing cases
Mode Maximum
7.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SAR37
Equipment
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
6 8 27 16 53 30 24 1 ------165
3.6 4.8 16.4 9.7 32.1 18.2 14.5 .6 ------100.0
3.7 4.9 16.5 9.8 32.3 18.3 14.6 Missing ------100.0
3.7 8.5 25.0 34.8 67.1 85.4 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
4.756 1.583 164
Median Minimum Missing cases
5.000 1.000
Mode Maximum 1
5.000 7.000
400 PERAG38
Teaching aids
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 4 18 10 62 23 40 5 ------165
1.8 2.4 10.9 6.1 37.6 13.9 24.2 3.0 ------100.0
1.9 2.5 11.3 6.3 38.8 14.4 25.0 Missing ------100.0
1.9 4.4 15.6 21.9 60.6 75.0 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.206 1.467
Valid cases
160
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
5.000 7.000
5
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ORG39
Organization
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
43 21 28 23 27 9 12 2 ------165
26.1 12.7 17.0 13.9 16.4 5.5 7.3 1.2 ------100.0
26.4 12.9 17.2 14.1 16.6 5.5 7.4 Missing ------100.0
26.4 39.3 56.4 70.6 87.1 92.6 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
3.276 1.900
Valid cases
163
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
1.000 7.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - HARG310
Rewards
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
12 12 22 10 40 30 38 1 ------165
7.3 7.3 13.3 6.1 24.2 18.2 23.0 .6 ------100.0
7.3 7.3 13.4 6.1 24.4 18.3 23.2 Missing ------100.0
7.3 14.6 28.0 34.1 58.5 76.8 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
4.805 1.876 164
Median Minimum Missing cases
5.000 1.000
Mode Maximum 1
5.000 7.000
401 LAIN311
Other problems
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
3 5 6 7 9
1 11 10 24 119 ------165
.6 6.7 6.1 14.5 72.1 ------100.0
2.2 23.9 21.7 52.2 Missing ------100.0
2.2 26.1 47.8 100.0
Somewhat important Important problem Very important
Total Mean Std dev
6.217 .964
Valid cases
46
Median Minimum
7.000 3.000
Missing cases
Mode Maximum
7.000 7.000
119
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - REG41
Publications irregular
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
4 5 27 12 53 22 40 2 ------165
2.4 3.0 16.4 7.3 32.1 13.3 24.2 1.2 ------100.0
2.5 3.1 16.6 7.4 32.5 13.5 24.5 Missing ------100.0
2.5 5.5 22.1 29.4 62.0 75.5 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.031 1.593
Valid cases
163
Median Minimum
5.000 1.000
Missing cases
Mode Maximum
5.000 7.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - REL42
Publications relevance
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
9 8 26 15 57 23 26 1 ------165
5.5 4.8 15.8 9.1 34.5 13.9 15.8 .6 ------100.0
5.5 4.9 15.9 9.1 34.8 14.0 15.9 Missing ------100.0
5.5 10.4 26.2 35.4 70.1 84.1 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
4.683 1.653 164
Median Minimum Missing cases
5.000 1.000
Mode Maximum 1
5.000 7.000
402 CERNA43
Publications difficult to translate
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
23 19 33 18 38 15 16 3 ------165
13.9 11.5 20.0 10.9 23.0 9.1 9.7 1.8 ------100.0
14.2 11.7 20.4 11.1 23.5 9.3 9.9 Missing ------100.0
14.2 25.9 46.3 57.4 80.9 90.1 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
3.852 1.856
Valid cases
162
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
5.000 7.000
3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ANGG44
Funding for extension
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
5 5 12 8 45 31 58 1 ------165
3.0 3.0 7.3 4.8 27.3 18.8 35.2 .6 ------100.0
3.0 3.0 7.3 4.9 27.4 18.9 35.4 Missing ------100.0
3.0 6.1 13.4 18.3 45.7 64.6 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
5.488 1.576
Valid cases
164
Median Minimum
6.000 1.000
Missing cases
Mode Maximum
7.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LATIH45
Training infrequent
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
3 3 12 8 27 32 79 1 ------165
1.8 1.8 7.3 4.8 16.4 19.4 47.9 .6 ------100.0
1.8 1.8 7.3 4.9 16.5 19.5 48.2 Missing ------100.0
1.8 3.7 11.0 15.9 32.3 51.8 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
5.835 1.496 164
Median Minimum Missing cases
6.000 1.000
Mode Maximum 1
7.000 7.000
403 WAKTU46
Time insufficient
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
32 26 26 17 43 12 8 1 ------165
19.4 15.8 15.8 10.3 26.1 7.3 4.8 .6 ------100.0
19.5 15.9 15.9 10.4 26.2 7.3 4.9 Missing ------100.0
19.5 35.4 51.2 61.6 87.8 95.1 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
3.494 1.818
Valid cases
164
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
5.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - TERAMP47
Translation skills poor
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
28 17 25 15 48 16 15 1 ------165
17.0 10.3 15.2 9.1 29.1 9.7 9.1 .6 ------100.0
17.1 10.4 15.2 9.1 29.3 9.8 9.1 Missing ------100.0
17.1 27.4 42.7 51.8 81.1 90.9 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev
3.890 1.903
Valid cases
164
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
5.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - TEMU48
R-E meetings infrequent
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
2 5 17 7 44 32 54 4 ------165
1.2 3.0 10.3 4.2 26.7 19.4 32.7 2.4 ------100.0
1.2 3.1 10.6 4.3 27.3 19.9 33.5 Missing ------100.0
1.2 4.3 14.9 19.3 46.6 66.5 100.0
Not a problem Somewhat important Important problem Very important
Total Mean Std dev Valid cases
5.472 1.509 161
Median Minimum Missing cases
6.000 1.000
Mode Maximum 4
7.000 7.000
404 LAIN49
Other problems
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
3 4 5 6 7 9
2 2 10 8 21 122 ------165
1.2 1.2 6.1 4.8 12.7 73.9 ------100.0
4.7 4.7 23.3 18.6 48.8 Missing ------100.0
4.7 9.3 32.6 51.2 100.0
Somewhat important Important problem Very important
Total Mean Std dev
6.023 1.165
Valid cases
43
Median Minimum
6.000 3.000
Missing cases
Mode Maximum
7.000 7.000
122
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SINAR51
Read Sinar Tani in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 9
2 5 27 66 58 6 1 ------165
1.2 3.0 16.4 40.0 35.2 3.6 .6 ------100.0
1.2 3.0 16.5 40.2 35.4 3.7 Missing ------100.0
1.2 4.3 20.7 61.0 96.3 100.0
Not once Once in 3 months Once/month Once/week Every 3 days Every day
Total Mean Std dev
4.165 .935
Valid cases
164
Median Minimum
4.000 1.000
Missing cases
Mode Maximum
4.000 6.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BULIP52
Read Bul Info Pert in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 9
13 75 45 20 5 5 2 ------165
7.9 45.5 27.3 12.1 3.0 3.0 1.2 ------100.0
8.0 46.0 27.6 12.3 3.1 3.1 Missing ------100.0
8.0 54.0 81.6 93.9 96.9 100.0
Not once Once in 3 months Once/month Once/week Every 3 days Every day
Total Mean Std dev Valid cases
2.656 1.091 163
Median Minimum Missing cases
2.000 1.000
Mode Maximum 2
2.000 6.000
405 LIPTAN53
Read Liptan in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 9
10 55 47 33 7 10 3 ------165
6.1 33.3 28.5 20.0 4.2 6.1 1.8 ------100.0
6.2 34.0 29.0 20.4 4.3 6.2 Missing ------100.0
6.2 40.1 69.1 89.5 93.8 100.0
Not once Once in 3 months Once/month Once/week Every 3 days Every day
Total Mean Std dev
3.012 1.236
Valid cases
162
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
2.000 6.000
3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - WARTA54
Read Warta Litbang in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 9
39 72 33 12 4 5 ------165
23.6 43.6 20.0 7.3 2.4 3.0 ------100.0
24.4 45.0 20.6 7.5 2.5 Missing ------100.0
24.4 69.4 90.0 97.5 100.0
Not once Once in 3 months Once/month Once/week Every 3 days
Total Mean Std dev
2.188 .972
Valid cases
160
Median Minimum
2.000 1.000
Missing cases
Mode Maximum
2.000 5.000
5
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MAJ55
Read sci journals in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 9
53 65 30 10 3 4 ------165
32.1 39.4 18.2 6.1 1.8 2.4 ------100.0
32.9 40.4 18.6 6.2 1.9 Missing ------100.0
32.9 73.3 91.9 98.1 100.0
Not once Once in 3 months Once/month Once/week Every 3 days
Total Mean Std dev Valid cases
2.037 .968 161
Median Minimum Missing cases
2.000 1.000
Mode Maximum 4
2.000 5.000
406 BUKU56
Read AARD books in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 9
50 67 35 9 2 2 ------165
30.3 40.6 21.2 5.5 1.2 1.2 ------100.0
30.7 41.1 21.5 5.5 1.2 Missing ------100.0
30.7 71.8 93.3 98.8 100.0
Not once Once in 3 months Once/month Once/week Every 3 days
Total Mean Std dev
2.055 .925
Valid cases
163
Median Minimum
2.000 1.000
Missing cases
Mode Maximum
2.000 5.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BROS57
Read BIP brochures in last 3 months
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 9
6 58 52 36 5 6 2 ------165
3.6 35.2 31.5 21.8 3.0 3.6 1.2 ------100.0
3.7 35.6 31.9 22.1 3.1 3.7 Missing ------100.0
3.7 39.3 71.2 93.3 96.3 100.0
Not once Once in 3 months Once/month Once/week Every 3 days Every day
Total Mean Std dev
2.963 1.088
Valid cases
163
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
2.000 6.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SINAR61
Usefulness of Sinar Tani
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
2 3 28 15 72 15 29 1 ------165
1.2 1.8 17.0 9.1 43.6 9.1 17.6 .6 ------100.0
1.2 1.8 17.1 9.1 43.9 9.1 17.7 Missing ------100.0
1.2 3.0 20.1 29.3 73.2 82.3 100.0
Not useful Somewhat useful Rather useful Very useful
Total Mean Std dev Valid cases
4.909 1.392 164
Median Minimum Missing cases
5.000 1.000
Mode Maximum 1
5.000 7.000
407 BULIP62
Usefulness of Bul Info Pert
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
2 3 4 5 6 7 9
1 6 8 75 34 40 1 ------165
.6 3.6 4.8 45.5 20.6 24.2 .6 ------100.0
.6 3.7 4.9 45.7 20.7 24.4 Missing ------100.0
.6 4.3 9.1 54.9 75.6 100.0
Somewhat useful Rather useful Very useful
Total Mean Std dev
5.555 1.064
Valid cases
164
Median Minimum
5.000 2.000
Missing cases
Mode Maximum
5.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LIPTAN63
Usefulness of Liptan
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 2 3 4 5 6 7 9
2 3 10 13 64 23 49 1 ------165
1.2 1.8 6.1 7.9 38.8 13.9 29.7 .6 ------100.0
1.2 1.8 6.1 7.9 39.0 14.0 29.9 Missing ------100.0
1.2 3.0 9.1 17.1 56.1 70.1 100.0
Dont know publicatio Somewhat useful Rather useful Very useful
Total Mean Std dev
5.421 1.401
Valid cases
164
Median Minimum
5.000 .000
Missing cases
Mode Maximum
5.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - WARTA64
Usefulness of Warta Litbang
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 3 4 5 6 7 9
3 6 12 44 32 67 1 ------165
1.8 3.6 7.3 26.7 19.4 40.6 .6 ------100.0
1.8 3.7 7.3 26.8 19.5 40.9 Missing ------100.0
1.8 5.5 12.8 39.6 59.1 100.0
Dont know publicatio Somewhat useful Rather useful Very useful
Total Mean Std dev Valid cases
5.774 1.385 164
Median Minimum Missing cases
6.000 .000
Mode Maximum 1
7.000 7.000
408 MAJ65
Usefulness of sci journals
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 3 4 5 6 7 9
5 5 9 34 36 75 1 ------165
3.0 3.0 5.5 20.6 21.8 45.5 .6 ------100.0
3.0 3.0 5.5 20.7 22.0 45.7 Missing ------100.0
3.0 6.1 11.6 32.3 54.3 100.0
Dont know publicatio Somewhat useful Rather useful Very useful
Total Mean Std dev
5.866 1.501
Valid cases
164
Median Minimum
6.000 .000
Missing cases
Mode Maximum
7.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BUKU66
Usefulness of AARD books
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
2 1 2 2 11 33 41 72 1 ------165
1.2 .6 1.2 1.2 6.7 20.0 24.8 43.6 .6 ------100.0
1.2 .6 1.2 1.2 6.7 20.1 25.0 43.9 Missing ------100.0
1.2 1.8 3.0 4.3 11.0 31.1 56.1 100.0
Dont know publicatio Not useful Somewhat useful Rather useful Very useful
Total Mean Std dev
5.915 1.345
Valid cases
164
Median Minimum
6.000 .000
Missing cases
Mode Maximum
7.000 7.000
1
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BROS67
Usefulness of BIP brochures
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 3 4 5 6 7 9
3 8 14 60 36 42 2 ------165
1.8 4.8 8.5 36.4 21.8 25.5 1.2 ------100.0
1.8 4.9 8.6 36.8 22.1 25.8 Missing ------100.0
1.8 6.7 15.3 52.1 74.2 100.0
Not useful Somewhat useful Rather useful Very useful
Total Mean Std dev Valid cases
5.479 1.269 163
Median Minimum Missing cases
5.000 1.000
Mode Maximum 2
5.000 7.000
409 BAR71A
Got info on Barumun rice variety
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
74 26 65 ------165
44.8 15.8 39.4 ------100.0
74.0 26.0 Missing ------100.0
74.0 100.0
No Yes
Total Mean Std dev
.260 .441
Valid cases
100
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
65
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RAMA72A
Got info on Rama maize variety
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
68 31 66 ------165
41.2 18.8 40.0 ------100.0
68.7 31.3 Missing ------100.0
68.7 100.0
No Yes
Total Mean Std dev
.313 .466
Valid cases
99
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
66
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PUPUK73A
Got info on P dosage in wetland rice
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
36 64 65 ------165
21.8 38.8 39.4 ------100.0
36.0 64.0 Missing ------100.0
36.0 100.0
No Yes
Total Mean Std dev
.640 .482
Valid cases
100
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
65
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SESB74A
Got info on Sesbania as green manure
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
33 67 65 ------165
20.0 40.6 39.4 ------100.0
33.0 67.0 Missing ------100.0
33.0 100.0
No Yes
Total Mean Std dev Valid cases
.670 .473 100
Median Minimum Missing cases
1.000 .000 65
Mode Maximum
1.000 1.000
410 MUSUH75A
Got info on enemies of stemborer
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
20 79 66 ------165
12.1 47.9 40.0 ------100.0
20.2 79.8 Missing ------100.0
20.2 100.0
No Yes
Total Mean Std dev
.798 .404
Valid cases
99
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
66
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ABU76A
Got info on ash in soybean seed storage
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
30 70 65 ------165
18.2 42.4 39.4 ------100.0
30.0 70.0 Missing ------100.0
30.0 100.0
No Yes
Total Mean Std dev
.700 .461
Valid cases
100
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
65
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BAR71B
Info source on Barumun rice variety
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 6 7 9
1 9 2 6 5 2 1 139 ------165
.6 5.5 1.2 3.6 3.0 1.2 .6 84.2 ------100.0
3.8 34.6 7.7 23.1 19.2 7.7 3.8 Missing ------100.0
3.8 38.5 46.2 69.2 88.5 96.2 100.0
Dont know AARD publications BIP publications Mass media Colleagues Other
Total Mean Std dev
2.692 1.828
Valid cases
26
Median Minimum
3.000 .000
Missing cases
Mode Maximum
1.000 7.000
139
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RAMA72B
Info source on Rama maize variety
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 6 7 9
2 11 4 8 4 1 1 134 ------165
1.2 6.7 2.4 4.8 2.4 .6 .6 81.2 ------100.0
6.5 35.5 12.9 25.8 12.9 3.2 3.2 Missing ------100.0
6.5 41.9 54.8 80.6 93.5 96.8 100.0
Dont know AARD publications BIP publications Mass media Colleagues Other
Total Mean
2.323
Median
2.000
Mode
1.000
411 Std dev Valid cases
1.641 31
Minimum Missing cases
.000 134
Maximum
7.000
412 PUPUK73B
Info source on P dosage in wetland rice
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
1 23 4 7 14 4 5 6 101 ------165
.6 13.9 2.4 4.2 8.5 2.4 3.0 3.6 61.2 ------100.0
1.6 35.9 6.3 10.9 21.9 6.3 7.8 9.4 Missing ------100.0
1.6 37.5 43.8 54.7 76.6 82.8 90.6 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.125 2.089
Valid cases
64
Median Minimum
3.000 .000
Missing cases
Mode Maximum
1.000 7.000
101
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SESB74B
Info source on Sesbania as green manure
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
5 14 4 22 9 1 5 7 98 ------165
3.0 8.5 2.4 13.3 5.5 .6 3.0 4.2 59.4 ------100.0
7.5 20.9 6.0 32.8 13.4 1.5 7.5 10.4 Missing ------100.0
7.5 28.4 34.3 67.2 80.6 82.1 89.6 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.104 2.039
Valid cases
67
Median Minimum
3.000 .000
Missing cases
Mode Maximum
3.000 7.000
98
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MUSUH75B
Info source on enemies of stemborer
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
1 23 11 10 11 8 6 8 87 ------165
.6 13.9 6.7 6.1 6.7 4.8 3.6 4.8 52.7 ------100.0
1.3 29.5 14.1 12.8 14.1 10.3 7.7 10.3 Missing ------100.0
1.3 30.8 44.9 57.7 71.8 82.1 89.7 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev Valid cases
3.218 2.081 78
Median Minimum Missing cases
3.000 .000 87
Mode Maximum
1.000 7.000
413 ABU76B
Info source on ash in soybean seed stora
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
2 13 12 11 9 7 7 10 94 ------165
1.2 7.9 7.3 6.7 5.5 4.2 4.2 6.1 57.0 ------100.0
2.8 18.3 16.9 15.5 12.7 9.9 9.9 14.1 Missing ------100.0
2.8 21.1 38.0 53.5 66.2 76.1 85.9 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.563 2.136
Valid cases
71
Median Minimum
3.000 .000
Missing cases
Mode Maximum
1.000 7.000
94
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UMB81A
Got info on urea molasses block
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
7 32 126 ------165
4.2 19.4 76.4 ------100.0
17.9 82.1 Missing ------100.0
17.9 100.0
No Yes
Total Mean Std dev
.821 .389
Valid cases
39
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
126
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SANG82A
Got info on conical nest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
2 38 125 ------165
1.2 23.0 75.8 ------100.0
5.0 95.0 Missing ------100.0
5.0 100.0
No Yes
Total Mean Std dev
.950 .221
Valid cases
40
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
125
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ITIK83A
Got info on ducks without pond
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
6 34 125 ------165
3.6 20.6 75.8 ------100.0
15.0 85.0 Missing ------100.0
15.0 100.0
No Yes
Total Mean Std dev Valid cases
.850 .362 40
Median Minimum Missing cases
1.000 .000 125
Mode Maximum
1.000 1.000
414 GLIR84A
Got info on Gliricidia as feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
2 38 125 ------165
1.2 23.0 75.8 ------100.0
5.0 95.0 Missing ------100.0
5.0 100.0
No Yes
Total Mean Std dev
.950 .221
Valid cases
40
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
125
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - VAKS85A
Got info on vaccination against ND
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
8 32 125 ------165
4.8 19.4 75.8 ------100.0
20.0 80.0 Missing ------100.0
20.0 100.0
No Yes
Total Mean Std dev
.800 .405
Valid cases
40
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
125
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - DOMBA86A
Got info on prolific sheep
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
21 19 125 ------165
12.7 11.5 75.8 ------100.0
52.5 47.5 Missing ------100.0
52.5 100.0
No Yes
Total Mean Std dev
.475 .506
Valid cases
40
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
125
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UMB81B
Info source on urea molasses block
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
1 7 2 3 6 4 2 7 133 ------165
.6 4.2 1.2 1.8 3.6 2.4 1.2 4.2 80.6 ------100.0
3.1 21.9 6.3 9.4 18.8 12.5 6.3 21.9 Missing ------100.0
3.1 25.0 31.3 40.6 59.4 71.9 78.1 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.906 2.305
* Multiple modes exist.
Median Minimum
4.000 .000
Mode Maximum
The smallest value is shown.
1.000 7.000
415 Valid cases
32
Missing cases
133
416 SANG82B
Info source on conical nest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
1 10 10 4 2 5 3 3 127 ------165
.6 6.1 6.1 2.4 1.2 3.0 1.8 1.8 77.0 ------100.0
2.6 26.3 26.3 10.5 5.3 13.2 7.9 7.9 Missing ------100.0
2.6 28.9 55.3 65.8 71.1 84.2 92.1 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.000 2.053
Median Minimum
2.000 .000
Mode Maximum
* Multiple modes exist.
The smallest value is shown.
Valid cases
Missing cases
38
1.000 7.000
127
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ITIK83B
Info source on ducks without pond
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
4 6 7 2 6 1 7 132 ------165
2.4 3.6 4.2 1.2 3.6 .6 4.2 80.0 ------100.0
12.1 18.2 21.2 6.1 18.2 3.0 21.2 Missing ------100.0
12.1 30.3 51.5 57.6 75.8 78.8 100.0
AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.939 2.091
Median Minimum
3.000 1.000
Mode Maximum
* Multiple modes exist.
The smallest value is shown.
Valid cases
Missing cases
33
3.000 7.000
132
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - GLIR84B
Info source on Gliricidia as feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 5 6 7 9
9 15 2 2 2 8 127 ------165
5.5 9.1 1.2 1.2 1.2 4.8 77.0 ------100.0
23.7 39.5 5.3 5.3 5.3 21.1 Missing ------100.0
23.7 63.2 68.4 73.7 78.9 100.0
AARD publications BIP publications Mass media Training Other
Total Mean Std dev Valid cases
3.237 2.342 38
Median Minimum Missing cases
2.000 1.000 127
Mode Maximum
2.000 7.000
417 VAKS85B
Info source on vaccination against ND
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
1 2 3 4 5 6 7 9
6 2 14 2 2 2 4 133 ------165
3.6 1.2 8.5 1.2 1.2 1.2 2.4 80.6 ------100.0
18.8 6.3 43.8 6.3 6.3 6.3 12.5 Missing ------100.0
18.8 25.0 68.8 75.0 81.3 87.5 100.0
AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.438 1.900
Valid cases
32
Median Minimum
3.000 1.000
Missing cases
Mode Maximum
3.000 7.000
133
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - DOMBA86B
Info source on prolific sheep
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 2 3 4 5 6 7 9
2 2 1 4 2 3 2 3 146 ------165
1.2 1.2 .6 2.4 1.2 1.8 1.2 1.8 88.5 ------100.0
10.5 10.5 5.3 21.1 10.5 15.8 10.5 15.8 Missing ------100.0
10.5 21.1 26.3 47.4 57.9 73.7 84.2 100.0
Dont know AARD publications BIP publications Mass media Colleagues Training Other
Total Mean Std dev
3.789 2.299
Valid cases
19
Median Minimum
4.000 .000
Missing cases
Mode Maximum
3.000 7.000
146
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PADBUD9
Rice cultivation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
146 17 2 ------165
88.5 10.3 1.2 ------100.0
89.6 10.4 Missing ------100.0
89.6 100.0
Dont need Need
Total Mean Std dev
.104 .307
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - JAGBUD9
Maize cultivation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
130 33 2 ------165
78.8 20.0 1.2 ------100.0
79.8 20.2 Missing ------100.0
79.8 100.0
Dont need Need
Total Mean
.202
Median
.000
Mode
.000
418 Std dev Valid cases
.403 163
Minimum Missing cases
.000
Maximum 2
1.000
419 UBIBUD9
Roots/tubers cultivation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
125 38 2 ------165
75.8 23.0 1.2 ------100.0
76.7 23.3 Missing ------100.0
76.7 100.0
Dont need Need
Total Mean Std dev
.233 .424
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KACBUD9
Legumes cultivation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
135 28 2 ------165
81.8 17.0 1.2 ------100.0
82.8 17.2 Missing ------100.0
82.8 100.0
Dont need Need
Total Mean Std dev
.172 .378
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SAYBUD9
Vegetables cultivation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
121 42 2 ------165
73.3 25.5 1.2 ------100.0
74.2 25.8 Missing ------100.0
74.2 100.0
Dont need Need
Total Mean Std dev
.258 .439
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BUABUD9
Fruits cultivation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
106 57 2 ------165
64.2 34.5 1.2 ------100.0
65.0 35.0 Missing ------100.0
65.0 100.0
Dont need Need
Total Mean Std dev Valid cases
.350 .478 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
420 PADBEN9
Rice seed & varieties
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
101 62 2 ------165
61.2 37.6 1.2 ------100.0
62.0 38.0 Missing ------100.0
62.0 100.0
Dont need Need
Total Mean Std dev
.380 .487
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - JAGBEN9
Maize seed & varieties
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
115 48 2 ------165
69.7 29.1 1.2 ------100.0
70.6 29.4 Missing ------100.0
70.6 100.0
Dont need Need
Total Mean Std dev
.294 .457
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UBIBEN9
Roots/tubers seed & varieties
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
104 59 2 ------165
63.0 35.8 1.2 ------100.0
63.8 36.2 Missing ------100.0
63.8 100.0
Dont need Need
Total Mean Std dev
.362 .482
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KACBEN9
Legumes seed & varieties
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
105 58 2 ------165
63.6 35.2 1.2 ------100.0
64.4 35.6 Missing ------100.0
64.4 100.0
Dont need Need
Total Mean Std dev Valid cases
.356 .480 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
421 SAYBEN9
Vegetables seed & varieties
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
104 59 2 ------165
63.0 35.8 1.2 ------100.0
63.8 36.2 Missing ------100.0
63.8 100.0
Dont need Need
Total Mean Std dev
.362 .482
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BUABEN9
Fruits seed & varieties
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
108 55 2 ------165
65.5 33.3 1.2 ------100.0
66.3 33.7 Missing ------100.0
66.3 100.0
Dont need Need
Total Mean Std dev
.337 .474
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PADHAM9
Rice pests & diseases
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
119 44 2 ------165
72.1 26.7 1.2 ------100.0
73.0 27.0 Missing ------100.0
73.0 100.0
Dont need Need
Total Mean Std dev
.270 .445
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - JAGHAM9
Maize pests & diseases
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
116 47 2 ------165
70.3 28.5 1.2 ------100.0
71.2 28.8 Missing ------100.0
71.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.288 .454 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
422 UBIHAM9
Roots/tubers pests & diseases
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
116 47 2 ------165
70.3 28.5 1.2 ------100.0
71.2 28.8 Missing ------100.0
71.2 100.0
Dont need Need
Total Mean Std dev
.288 .454
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KACHAM9
Legumes pests & diseases
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
100 63 2 ------165
60.6 38.2 1.2 ------100.0
61.3 38.7 Missing ------100.0
61.3 100.0
Dont need Need
Total Mean Std dev
.387 .488
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SAYHAM9
Vegetables pests & diseases
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
99 64 2 ------165
60.0 38.8 1.2 ------100.0
60.7 39.3 Missing ------100.0
60.7 100.0
Dont need Need
Total Mean Std dev
.393 .490
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BUAHAM9
Fruits pests & diseases
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
103 60 2 ------165
62.4 36.4 1.2 ------100.0
63.2 36.8 Missing ------100.0
63.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.368 .484 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
423 PADPAS9
Rice post harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
119 44 2 ------165
72.1 26.7 1.2 ------100.0
73.0 27.0 Missing ------100.0
73.0 100.0
Dont need Need
Total Mean Std dev
.270 .445
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - JAGPAS9
Maize post harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
105 58 2 ------165
63.6 35.2 1.2 ------100.0
64.4 35.6 Missing ------100.0
64.4 100.0
Dont need Need
Total Mean Std dev
.356 .480
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - UBIPAS9
Roots/tubers post harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
99 64 2 ------165
60.0 38.8 1.2 ------100.0
60.7 39.3 Missing ------100.0
60.7 100.0
Dont need Need
Total Mean Std dev
.393 .490
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KACPAS9
Legumes post harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
103 60 2 ------165
62.4 36.4 1.2 ------100.0
63.2 36.8 Missing ------100.0
63.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.368 .484 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
424 SAYPAS9
Vegetables post harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
90 73 2 ------165
54.5 44.2 1.2 ------100.0
55.2 44.8 Missing ------100.0
55.2 100.0
Dont need Need
Total Mean Std dev
.448 .499
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BUAPAS9
Fruits post harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
88 75 2 ------165
53.3 45.5 1.2 ------100.0
54.0 46.0 Missing ------100.0
54.0 100.0
Dont need Need
Total Mean Std dev
.460 .500
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PERMIL9
Dairy cattle management
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
134 29 2 ------165
81.2 17.6 1.2 ------100.0
82.2 17.8 Missing ------100.0
82.2 100.0
Dont need Need
Total Mean Std dev
.178 .384
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - POTMIL9
Beef/buffalo management
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
134 29 2 ------165
81.2 17.6 1.2 ------100.0
82.2 17.8 Missing ------100.0
82.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.178 .384 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
425 KAMMIL9
Sheep/goats management
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
135 28 2 ------165
81.8 17.0 1.2 ------100.0
82.8 17.2 Missing ------100.0
82.8 100.0
Dont need Need
Total Mean Std dev
.172 .378
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RASMIL9
Improved chickens management
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
138 25 2 ------165
83.6 15.2 1.2 ------100.0
84.7 15.3 Missing ------100.0
84.7 100.0
Dont need Need
Total Mean Std dev
.153 .361
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BURMIL9
Local chickens management
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
130 33 2 ------165
78.8 20.0 1.2 ------100.0
79.8 20.2 Missing ------100.0
79.8 100.0
Dont need Need
Total Mean Std dev
.202 .403
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ITIMIL9
Ducks management
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
123 40 2 ------165
74.5 24.2 1.2 ------100.0
75.5 24.5 Missing ------100.0
75.5 100.0
Dont need Need
Total Mean Std dev Valid cases
.245 .432 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
426 PERREP9
Dairy cattle reproduction
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
132 31 2 ------165
80.0 18.8 1.2 ------100.0
81.0 19.0 Missing ------100.0
81.0 100.0
Dont need Need
Total Mean Std dev
.190 .394
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - POTREP9
Beef/buffalo reproduction
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
113 50 2 ------165
68.5 30.3 1.2 ------100.0
69.3 30.7 Missing ------100.0
69.3 100.0
Dont need Need
Total Mean Std dev
.307 .463
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KAMREP9
Sheep/goats reproduction
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
123 40 2 ------165
74.5 24.2 1.2 ------100.0
75.5 24.5 Missing ------100.0
75.5 100.0
Dont need Need
Total Mean Std dev
.245 .432
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RASREP9
Improved chickens reproduction
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
135 28 2 ------165
81.8 17.0 1.2 ------100.0
82.8 17.2 Missing ------100.0
82.8 100.0
Dont need Need
Total Mean Std dev Valid cases
.172 .378 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
427 BURREP9
Local chickens reproduction
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
123 40 2 ------165
74.5 24.2 1.2 ------100.0
75.5 24.5 Missing ------100.0
75.5 100.0
Dont need Need
Total Mean Std dev
.245 .432
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ITIREP9
Ducks reproduction
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
127 36 2 ------165
77.0 21.8 1.2 ------100.0
77.9 22.1 Missing ------100.0
77.9 100.0
Dont need Need
Total Mean Std dev
.221 .416
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PERPAK9
Dairy cattle feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
129 34 2 ------165
78.2 20.6 1.2 ------100.0
79.1 20.9 Missing ------100.0
79.1 100.0
Dont need Need
Total Mean Std dev
.209 .408
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - POTPAK9
Beef/buffalo feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
122 41 2 ------165
73.9 24.8 1.2 ------100.0
74.8 25.2 Missing ------100.0
74.8 100.0
Dont need Need
Total Mean Std dev Valid cases
.252 .435 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
428 KAMPAK9
Sheep/goats feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
127 36 2 ------165
77.0 21.8 1.2 ------100.0
77.9 22.1 Missing ------100.0
77.9 100.0
Dont need Need
Total Mean Std dev
.221 .416
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RASPAK9
Improved chickens feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
134 29 2 ------165
81.2 17.6 1.2 ------100.0
82.2 17.8 Missing ------100.0
82.2 100.0
Dont need Need
Total Mean Std dev
.178 .384
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - BURPAK9
Local chickens feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
118 45 2 ------165
71.5 27.3 1.2 ------100.0
72.4 27.6 Missing ------100.0
72.4 100.0
Dont need Need
Total Mean Std dev
.276 .448
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ITIPAK9
Ducks feed
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
121 42 2 ------165
73.3 25.5 1.2 ------100.0
74.2 25.8 Missing ------100.0
74.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.258 .439 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
429 PERKES9
Dairy cattle health
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
133 30 2 ------165
80.6 18.2 1.2 ------100.0
81.6 18.4 Missing ------100.0
81.6 100.0
Dont need Need
Total Mean Std dev
.184 .389
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - POTKES9
Beef/buffalo health
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
129 34 2 ------165
78.2 20.6 1.2 ------100.0
79.1 20.9 Missing ------100.0
79.1 100.0
Dont need Need
Total Mean Std dev
.209 .408
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KAMKES9
Sheep/goats health
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
127 36 2 ------165
77.0 21.8 1.2 ------100.0
77.9 22.1 Missing ------100.0
77.9 100.0
Dont need Need
Total Mean Std dev
.221 .416
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - RASKES9
Improved chickens health
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
129 34 2 ------165
78.2 20.6 1.2 ------100.0
79.1 20.9 Missing ------100.0
79.1 100.0
Dont need Need
Total Mean Std dev Valid cases
.209 .408 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
430 BURKES9
Local chickens health
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
122 41 2 ------165
73.9 24.8 1.2 ------100.0
74.8 25.2 Missing ------100.0
74.8 100.0
Dont need Need
Total Mean Std dev
.252 .435
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ITIKES9
Ducks health
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
125 38 2 ------165
75.8 23.0 1.2 ------100.0
76.7 23.3 Missing ------100.0
76.7 100.0
Dont need Need
Total Mean Std dev
.233 .424
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - HIJAU9
Fodder crops
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
114 49 2 ------165
69.1 29.7 1.2 ------100.0
69.9 30.1 Missing ------100.0
69.9 100.0
Dont need Need
Total Mean Std dev
.301 .460
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SUSU9
Milk handling
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
121 42 2 ------165
73.3 25.5 1.2 ------100.0
74.2 25.8 Missing ------100.0
74.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.258 .439 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
431 PASNAK9
Livestock post-harvest
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
108 55 2 ------165
65.5 33.3 1.2 ------100.0
66.3 33.7 Missing ------100.0
66.3 100.0
Dont need Need
Total Mean Std dev
.337 .474
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - IKAN9
Fisheries
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
136 27 2 ------165
82.4 16.4 1.2 ------100.0
83.4 16.6 Missing ------100.0
83.4 100.0
Dont need Need
Total Mean Std dev
.166 .373
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - PASAR9
Marketing
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
76 87 2 ------165
46.1 52.7 1.2 ------100.0
46.6 53.4 Missing ------100.0
46.6 100.0
Dont need Need
Total Mean Std dev
.534 .500
Valid cases
163
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SULUH9
Extension
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
81 82 2 ------165
49.1 49.7 1.2 ------100.0
49.7 50.3 Missing ------100.0
49.7 100.0
Dont need Need
Total Mean Std dev Valid cases
.503 .502 163
Median Minimum Missing cases
1.000 .000
Mode Maximum 2
1.000 1.000
432 REKWIL9
Regional planning
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
39 124 2 ------165
23.6 75.2 1.2 ------100.0
23.9 76.1 Missing ------100.0
23.9 100.0
Dont need Need
Total Mean Std dev
.761 .428
Valid cases
163
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - KEBUN9
Estate & industrial crops
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
121 42 2 ------165
73.3 25.5 1.2 ------100.0
74.2 25.8 Missing ------100.0
74.2 100.0
Dont need Need
Total Mean Std dev
.258 .439
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - MESIN9
Machinery
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
101 62 2 ------165
61.2 37.6 1.2 ------100.0
62.0 38.0 Missing ------100.0
62.0 100.0
Dont need Need
Total Mean Std dev
.380 .487
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - IRIG9
Irrigation
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
108 55 2 ------165
65.5 33.3 1.2 ------100.0
66.3 33.7 Missing ------100.0
66.3 100.0
Dont need Need
Total Mean Std dev Valid cases
.337 .474 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
433 EKON9
Economics
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
94 69 2 ------165
57.0 41.8 1.2 ------100.0
57.7 42.3 Missing ------100.0
57.7 100.0
Dont need Need
Total Mean Std dev
.423 .496
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ANAL9
Farming systems analysis
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
52 111 2 ------165
31.5 67.3 1.2 ------100.0
31.9 68.1 Missing ------100.0
31.9 100.0
Dont need Need
Total Mean Std dev
.681 .468
Valid cases
163
Median Minimum
1.000 .000
Missing cases
Mode Maximum
1.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - SOS9
Rural sociology
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
83 80 2 ------165
50.3 48.5 1.2 ------100.0
50.9 49.1 Missing ------100.0
50.9 100.0
Dont need Need
Total Mean Std dev
.491 .501
Valid cases
163
Median Minimum
.000 .000
Missing cases
Mode Maximum
.000 1.000
2
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - LAIN9
Other topics
Value Label
Value
Frequency
Percent
Valid Percent
Cum Percent
0 1 9
134 29 2 ------165
81.2 17.6 1.2 ------100.0
82.2 17.8 Missing ------100.0
82.2 100.0
Dont need Need
Total Mean Std dev Valid cases
.178 .384 163
Median Minimum Missing cases
.000 .000
Mode Maximum 2
.000 1.000
