— F-vf TYPOLOGTES OF CHANGE AGENTS BASED ON THEIR ACCURACY IN ESTIMATENG CDMMUNTW ADUPTION LEVEL OF INNOVATION Thesis for the Degree of M. S. MICHRGAN STATE UNIVERSITY VICENTE DE PAULA VITOR 1968 ...... "“3 LIBRARY Michigan State University 1“ Imam; av ”‘TL’.’ 1:3; ”ME & WW .:1 Mommas.- T“ umITTTTTTTTTTTTETTLTLTTTTTTHII 01310Q ABSTRACT TYPOLOGIES OF CHANGE AGENTS BASED ON THEIR ACCURACY IN ESTIMATING COMMUNITY ADOPTION LEVEL OF INNOVATIONS BY Vicente de Paula Vitor This study was concerned with correlates of change agent accuracy in estimating community adOption level. As part of a larger study, 20 Brazilian Extension agents were asked to estimate the level of adoption of innovations in their community, and their estimates were checked against responses from all farmers in that community. There were agents who under- and over—estimated, and agents who were quite accurate in judging their clients' adoption level. This thesis was carried out to see if important differences exist between accurate and inaccurate agents. The 20 agents were also asked about the adOption level of six innovations introduced by their offices in the last three years. They also were asked about their level of education, farming experience, years working as an extensionist, years working in the community, leadership style with the farmers, communications methods most used in their teaching Vicente de Paula Vitor and about their relationship with their client farmers. These variables were grouped in five main tOpics: 1. Agent education\/ 2. Agent background.J 3. Agent client interaction x/ 4. Agent style of communication.¢ 5. Agent at his job.\] Fifteen hypotheses using these variables were tried, using two statistical tests: "Kruskal-Wallis One-Way Analysis of Variance by Ranks" and "Spearman Rank Correlation Coefficient." Only three of the hypotheses reached a statis- tical level of significance: The agents with farming eXperience were more accurate than the agents who lacked this kind of experience. The agents with less time living in the communities were more accurate than those living there a longer time. The agents who considered individual discussion with farmers as the most effective method to take ideas to them were more accurate than the agents who chose group meetings as the most effective method. The implication of these findings is that the extension agency could have more confidence in the reports of the agents who present the above characteristics, and also considering the importance of this study for the extension service, another research could be carried out with a larger sample of agents. TYPOLOGIES OF CHANGE AGENTS BASED ON THEIR ACCURACY IN ESTIMATING COMMUNITY ADOPTION LEVEL OF INNOVATION BY Vicente de Paula Vitor A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Institute for Extension Personnel Development 1968 557/52 5:? - ‘/-/ ”.2” “,9 .~ . f .. ACKNOWLEDGMENTS The author wants to express his sincere gratitude to all people whose work made it possible to write and to finish this study. Special mention is owed to: Dr. Mason E. Miller, his academic adviser, for his Optimism, patience in reading and helping to reword the drafts, and also for his advising and orientation in the program at Michigan State University. Dr. Everett M. Rogers for carrying out the Diffusion Research Project in Brazil, India, and Nigeria; the data for the thesis came from part of this Project. Dr. Gordon C. Whiting, Director of the Brazilian Diffusion Project in 1968; his dedication and efforts in analyzing the data and preparing the first major report of this Project inspired the realization of this thesis. Agency for International Development (AID) and IRI Research Institute Inc. for sponsoring and coordinating his program in the United States. Associacao de Credito e Assistencia Rural (ACAR) directors and its hard working extensionists, whose cooperr ation and patience in answering the questionnaire made the field data available. ii His wife for her encouragement and arduous work of typing the first draft. Mrs. Margaret V. Hudson and Mrs. Nancy Stumhofer for their clerical work and great help in preparing the last part of this thesis. And also his appreciation and many thanks due to the committee members: Dr. Mason E. Miller, Director of the Institute for Extension Personnel DevelOpment; Dr. Kirk Lawton, Director of Institute of International Agriculture; Mr. Tom Carroll, Department of Communication faculty member. iii Chapter I II III IV TABLE OF CONTENTS INTRODUCTION. . . . . Origin of the Research Findings: A Three Nation Comparison Project. The Brazilian Research Project . Phase I - Sampling Procedure . Description of ACAR. The Dependent Variable: The Measurement of ACAR's Success (A Summary from the Phase I Report). IMPORTANCE OF ACCURACY IN ESTIMATING COMMUNITY ADOPTION LEVEL FOR THE LOCAL ACAR AGENT . . . . . THE LOCAL ACAR AGENT. Local ACAR Agent Characteristics The Local ACAR Agent Described . SOME THOUGHTS ON ACCURACY IN ESTIMATING COMMUNITY ADOPTION LEVEL AND CHANGE AGENT EFFECTIVENESS . . . . How Would the Extensionist Accomplish His Role as Change Agent if He Knows the ACAR's Success. Community Adoption Level with Accuracy?. Accuracy-ran Index of Effectiveness of the Change Agent . iv Page 16 20 20 31 32 35 38 Chapter Page Effectiveness of Change Agents in Other Studies: A Review of Literature. . . 39 V HYPOTHESES AND FINDINGS TO BUILD A TYPOLOGY OF CHANGE AGENTS BASED ON THEIR ACCURACY IN ESTIMATING COMMUNITY ADOPTION LEVEL. . . . . . 48 sampling. 0 O O O O O O O O O O O O O O O O 48 Data. 0 O O I O O O O I O O O O O O O O O O 49 Hypotheses O O O O O O O O C O O O O O O O O 50 Statistical Analysis. . . . . . . . . . . . 52 Presentation and Analysis of the Findings I O O O O O O O O O O O O O I O 54 VI SUMMARY AND CONCLUSIONS. . . . . . . . . . . . 87 Needed Research . . . . . . . . . . . . . . 96 Suggestions to Improve Research . . . . . . 97 BIBLIOGRAPHY. . . . . . . . . . . . . . . . . . . . . 98 APPENDIX. 0 O O O O O O O O O O O O O O O O O O O O O 102 Table 1. ll. 12. l3. 14. LIST OF TABLES Factor loadings for six estimates of practice adOption given by local ACAR agents (adjusted for the number of farmers to whom the practice is applicable. . . . . . . . . . . . . . . . . Correlations of selected Phase I estimates of success with Phase II adoption levels in 20 communities . . . . . Rankings of the adoption level of the same communitiés according to farmer answers and agent estimations . . . . . . . Number of ACAR technicians in each section and the number males and females. . Age, in years, ACAR local agents, 1964. . . Education level of the ACAR's agents, 1964. Time spent by ACAR agents in receiving technological training... .-. . . . . . . .‘ Time spent by ACAR agents in training about extension methods . . . . . . . . . . The of experience as an ACAR change agent . Time that each ACAR agent had been in his most recent ACAR field office . . . . . . . Help from Extension specialists . . . . . . COOperation from the supervisors. . . . . . Policy of the organization. . . . . . . . . Factors influencing decisions of ACAR technicians to leave the organization each ye ar 0 C O O O O O O O O O O O O O O 0 vi Page 11 13 15 22 22 23 24 25 26 27 28 28 29 30 Table Page 15. Difference between farmer reports and agent estimates of community adOption level. I O O O O O O O O O O O O O O O O I O 33 16. Rank order of local agents according to their accuracy of estimating community adOption level . . . . . . . . . . . . . . . 34 17. Rank order of accuracy of the agent with college and high school education. . . . . . 57 18. Rank order of accuracy in estimating the community adoption level for agents born in the country and in the city . . . . . . . 59 19. Agents' years of farming, ranking of farm experience and rank order of accuracy of the agents . . . . . . . . . . . . . . . . . 61 20. Years agents have worked for ACAR, rank order of work time, and rank of accuracy . . 63 21. Time in months of agent-client tenure, its ranking, and the accuracy ranking order. 65 22. Ranks on accuracy of the agents preferring authoritarian and agents preferring per- missive styles of dealing with peOple with little intelligence. . . . . . . . . . . . . 69 23. Agent's use of the extension methods chosen to represent interpersonal communication . . 72 24. Agent's use of the extension methods chosen to represent use of mass media in communi- cation with farmers. . . . . . . . . . . . . 74 25. Accuracy rankings of the agents who chose individual discussion as the most effect- ive method and those who chose group meetings as the most effective method. . . . 76 26. Hours a week spent on the extension job, rank order of this time, and accuracy ranking. 0 O O O O I. O O I O O O O O O 0 O O 79 vii Table 27. 28. 29. 30. 31. 32. Page Percentage of agent time spent in field work, its ranking order and accuracy ranks O O O O O O O O O O O O O O O O O O O O 80 Percentage of time spent in the office by the agents, its ranking and accuracy rank order O O O O O O O O O O O O O O O O O 81 Rank order of accuracy of the agents who are more field-oriented and the agents who are more office-oriented . . . . . . . . 84 Rankings of accuracy of the agents who said yes or no to the question about supervisor listening to the agent's ideas. . 86 Summary of the Kruskal-Wallis Analysis . . . 94 Spearman rank correlational analysis . . . . 9S viii Figure 1. LIST OF FIGURES Page Basic paradigm of Phase I core variables (Independent Predictor variableS) O O O O O O O O O O O O O O O O O 4 Organization Chart of ACAR. . . . . . . . . 8 Total social, psychological, situational, and cultural context for source and receivers O O O O O O O O O O O O O O O O O 37 ix CHAPTER I INTRODUCTION The objective of this study was to relate change agent accuracy in estimating community adOption level with several agent characteristics such as education, farm background, preferential work, etc., and by using these relationships to build typologies of agents. The data came from a very broad diffusion research project which was carried out in three nations. Origin of the Research Findings: A Three Nation Comparison Pro-ect Considerable empirical research has been conducted in developed countries, attempting to explain diffusion and adOption of new ideas by farmers. The Diffusion Document Center at Michigan State University is filled with the reports of these researches. Based on these reports, some hypotheses have been supported and generalizations drawn to give prac- tical direction to agencies and organizations working in the area of adOption of innovations. However, in the so-called underdeveloped world, there is a great scarcity of this kind of research, which makes somewhat tentative any cross appli- ance of generalizations based strictly on research in develOped countries. In order to test some of these previously mentioned generalizations in underdevelOped countries, a three-nation research project was organized under the direction of Professor Everett Miuflmdl Rogers and sponsored by the Agency for International Deve10pment. This project began in 1964, when sites for the research were chosen using the criterion of representativeness of the world's develOping agriculture. These places were India, Brazil, and Nigeria. The general research plan for the three nations was to use agricultural modernization programs as innovations and (1) community characteristics, (2) individual farmer characteristics and (3) strategy of communication used by these programs of modernization, as explanations for the success or lack of success of these modernization programs. The Brazilian Research Project This part of the world project was set up in 1965 and was carried out by two members of the Department of Communi— cations, Michigan State University, and three Brazilian co- workersl. The research was divided into three phases: 1These three Brazilians were hired by the MSU Diffusion Research Project. Among them was the author of this report. Besides these three research assistants, the Brazilian country was represented in the Diffusion Project by: Federal Univer- sity of Minas Gerais, State Agricultural University of Minas Gerais, Association for Credit and Rural Assistance (ACAR) and Brazilian Association for Credit and Rural Assistance (ABCAR). Phase I - Study of community characteristics as an explanation for success or lack of success of the programs of agricultural modernization introduced by ACAR (Association for Credit and Rural Assistance), a government organization which tries to increase agricultural production by introducing new agricultural practices in rural communities. The adOption or nonadoption of practicesyadvocated by this agency is used as the dependent variable in all research phases. Phase II - Study of individual farmer characteristics as an explanation for acceptance of agricultural modernization programs in Brazilian rural communities in the last three years. Phase III — A field experiment to introduce change (agricultural modernization) by varying communication strategy and using communities and individual typologies drawn from Phases I and II. The Phase I major report lays out the basic paradigm of Phase I core community variables. This paradigm is: Fig. l - Basic paradigm of Phase I core variables Independent Predictor Variables (44:6). I - COMMUNITY SOCIAL STRUCTURE (1) Social differentiation (2) Opinion leadership concentration (3) Social status concentration (4) Consensus on village problems (DePEAdent (5) Norms ~wwVarwuhlsLM II - COMMUNITY MODERNIZATION LEVEL f RELATIVE SUCCESS 0F’\( ; CHANGE AGENCY IN (1) Institutional develOpment 3 TERMS OF THE LEVEL (2) External contact \ OF ADOPTION OF ITS (3) Modernization of leaders \\ PROGRAMS IN THE (4) Economic and social development ‘xS?MMUNITY III - CHANGE AGENCY VARIABLESb ‘\\\ ‘‘‘‘‘‘ ”we“/ (1) Integration of change agencies J (2) Recognized needs J (3) Social-cultural valuesJ (4) Feedback data ' T (5) Rapport with villagers (6) Communication channel apprOpriateness q (7) Perception of change agent v (8) Availability of credit aour result of trying to measure this dependent variable was the securing of data on the accuracy of agents in estimating community adOption level. bWithin "CHANGE AGENCY VARIABLES", several agent personal characteristics such as education, farm background, leadership style, etc., were measured. These findings will be used to build the typology of change agents on the basis of their accuracy in estimating community adoption level. Phase I - Sampling Procedure: A stratified random sample of 40 local agents working for ACAR was chosen. This sample was meant to represent two different regions of Minas Gerais State, in terms of geography and economy, and also an area where ACAR had been working for at least three years. At the time of sampling, ACAR had 131 local offices. However, only 78 had existed for at least three years. So the 78 local offices constituted the uni- verse or pOpulation for the study. From this pOpulation of local offices, the 40 agents were selected. These chosen agents were to identify the communities to be analyzed. They were asked to name two communities in their counties: one in which they had had the most success and another in which their program had had a minimum of acceptance among the farmers. These 80 communities became the community sample for Phase I. Their characteristics were measured, as well as the acceptance of the change program introduced by ACAR. Description of ACAR ACAR--Association for Credit and Rural Assistance--was created in 1953 through a cooperative program between the state of Minas Gerais and the American International Association. Its function wfis to develOp and to increase agricultural production. (24:163) This organization today is an institution With more than 230 technicians. It is part of the total Brazilian Extension System. The Brazilian Extension Service is very similar to the- United States Extension Service. Its philosophy of work is Extension education which tries to help the rural people to help themselves. Its technicians are mostly agriculturally oriented. However, there are some homemaking and youth educatars. In order to accomplish its task, ACAR uses a.kind of credit-oriented program, differing in this particular from American Extension education. Loans are made through the state bank. These loans help farmers handle the costs of the several innovations introduced by the Extension Service. In spite of some controversy around the use of credit as a means to get the Extension program accepted by the farmers, this device has helped ACAR to bring about changes of traditional agriculture throughout the State. As an institution, ACAR is divided into four levels: 1. A Board of Directors, located in the central office at the state capital. ,2. A group of Extension Specialists working on program. planning and orientation of the local Extension workers for their task of taking agricultural innovations to farmers. 3. Regional supervisors, a more administratively oriented group. Each regional superviser oversees an average of ten local offices. 4. Local offices--made up of two agents (a male and a female), a clerical worker and a jeep. These agents are in direct contact with farmers--teaching, orienting, and taking agricultural innovations to them. This section of ACAR will be the principal focus of the present study. The Dependent Variable: ACAR's Success~ Recalling what was said earlier, the dependent variable in this broad research project would be the acceptance of agricultural modernization programs. The over-all objective of this study was to find a tentative explanation for this acceptance by manipulating community characteristics, indi- vidual farmer characteristics, and the use of different methods to bring about this modernization. Since ACAR is an organization which works in this area of adOption, their efforts should be very representative of this dependent variable. So by using the adoption level of ACAR-promoted programs, the researchers Operationalized the dependent variable acceptance g£.agricultural modernization program . The way acceptance of ACAR's program was measured, and how the findings on change agent accuracy in estimating the community adoption level (which is the objective of this writing) were arrived at, is described in detail in the Phase I_report done by Whiting and others. (44: 17-26) This is one of the first major reports from the "Diffusion Research Project on Rural Societies." Others, reporting Phases II and III, are being prepared. m¢0¢ mo unmau coaumecmmuo I ... 1 TJ «Leena ma_..._z«... u momma: I a - .300; 0253.3: 2633 on .1 .- 7539; 5025.... 333.23 on .o .n 0:33:01! .4 .- 3219. 3.2. 824: o: .> :Euxih 32.: an «x: 38...: 030.. :a do .n .o .6: 3305 up: .028... :03. loo 3:. «I: a -295 .r 3.2.2 .35. 55:. 095.. 0001 $81.5 8.523 .22.. 31.0505 .1 «5 up: 332?. aze u.- oEUo-SL . 2:: «:35: .28» an .5: «9:5» 3: .3 52.3 .5qu 5.5 «53985 5:23.] 0:35.- .n :5: a... 8:1 -.urzo.:uo: .n 1.5.3133 n3.- -<3oz34 #m H _ m. <00... _ _ mo_m.o._._mumu I u 1 __ cau>eau _ A 738 25.34 a _ if h M (TL _ _ _ _ a w .4 m a J T85 3.1.81; ~56: 3:31— mezi “32$; 75:: we .éfl :32: «2 _ TEE we NEIL fl (0.5:: I L H ~ . _ _ _ _ L fl qfuzwemmai D I QEEEEZES 1:230 m.¢uw 36.1928 32:; 3855;: m 35.36 on 338034 :23. «.0555: u 3.3.33 ”3 (53.3% 32.68? V Jl <3 43419 on ouz:wmoou pom mopoom pom uxmucoo amusufiso paw .Hmcowumoufim .Hmoflonosomwo .HMwoom HmDOH .m .mHm u pomOQEoo coflumsuflm wchHmmH op ou manw mm .5 xcHLH .m Hmwm .N BocM .H "umrz unuow>mzmm HmcHEuoH m.uo>wooom pom m_oopoom mmcHuuom pofl>msmb paw mopomwao uow>mcon mo op ou Emuwoum oflwwommm oabm mm .8 How mu0fl>mnon VETS. .m muuco pmufiovou ’ muoch Hmom .N LDHB mocmflpsm HoCGMLOImemmoE Bocx .H Eoum muo>woomm "umc3 mosmfips< HmDOH uau0w>msmm kuus 38 agent may be very important at the beginning of the intro- duction of an innovation but may be unnecessary after this diffusion reaches a certain point within the community where other forces now maintain its spread. The agent can use this knowledge as strategy in his work of diffusing several inno- vations among the farmers. He is able to make a better distribution of his efforts and to be more effective as a change agent. Another aspect of looking at accuracy as a factor in the effectiveness of local ACAR agent is concerned with the validity of his report to ACAR. ACAR is interested in intro- ducing and getting adopted as many new ideas to farmers as it can. If the agent is able to make a valid report of the ACAR acceptance program from his office, this agency can concentrate on those innovations that could be more adOpted, instead of losing time with others less accepted in the region. By providing this accurate report this agent will be more effective as a member of ACAR and as an extension worker. But if he cannot make this report with validity he certainly will lose much of his effectiveness. ACAR could also put more resources into an area or innovation that isn't going we11--help the agent, and investigate causes, etc. Accuracy--an Index of Effectiveness of the Change Agent A number of points, then, support the relevance of the local ACAR agent knowing with accuracy the community adoption 39 level: (1) accuracy in perception of receiver's effect for revision of the communication process, (2) sureness in deter- mining the terminal behavior level in teaching-learning situations, (3) accuracy in estimating the need of more or less effort in diffusing innovations by the agent, and (4) validity of report for the agency judgment of his program acceptance. It would seem that this accuracy in estimating the community adoption level could be considered an index of effectiveness of this agent in performing his job as an extension worker. So in this study this accuracy index will operationalize the effectiveness of the change agent in his job performance. The more accurate the agent, the more effective he will be as a change agent in his community. Effectiveness of Change Agents in Other Studies: A Review-of Literature V/ Quite a few researches have been conducted to try to explain variation in effectiveness of changelagents, by r————*--——~ll__,__._lflfl—r-~“‘ l_ull‘““m,_w,ll__ relating effectiveness to several other independent variables {V a such as characteristics ofhchange agents, workmpattern%;’time Ml spent in the field and in the office, etc. Some of these * \\\ _ 11 \_ ‘Hb—M studies will be discussed in order to clarify the approach that will be used to attempt to explain accuracy of the change agent as related with some of the agent characteristics that this study will consider later on. 40 Nye, studying the relationship of "Certain Factors to County Agents Success,‘ states that there are three methods of measuring effectiveness of the county agents: 1. Judgment of qualified peOple 2. Possession or non-possession of characteristics or methods which are believed to be associated with good job performance. 3. Actual results achieved by the individual. This author took a sample of figflworkers of the Missouri Extension Service and classified them:_using effectiveness judged by qualified peOple, into two groups: high and low effectiveness. He analyzed background and training, voca- tional interests, attitudes and personality of the agents. Nye found 63% of effectiveness variation explained by these factors. In an evaluation of the Extension Service in India, sponsored by the Ford Foundation and carried out by the Allahabad Agriculture Institute, the performance and charac- teristics of 428 village workers were analyzed in order to explain thelevel of performance of these 'village workers. In the report of this evaluation, the authors present the following conclusions: {/1. Differences in the educational level and type of _1,,,— training have a bearing on the performance level m of the village workers. 41 2. Among the single workers the graudates turned in a higher level of performance than other categories of workers. 3. Within each category of workers there were sub- stantial differences with regard to the level of performance among the individual workers. 4. Among all categories of workers, an over-all improvement in efficiency with the passing of time was indicated. 5. About two units and graduates got off to a better start and maintained the lead throughout the period of the experiment, except for intermediates who although slow to start, surpassed the graduates in their performance in the fourth season. ~ 6. The largest number of practice changes was effected where the_felE_needs_app£oachflwas_followed. Among other researches done in the United States with agent effectiveness the more relevant ones are: Curry, who found that high ranked agents organized more groups, and attracted more peOple due to their efficiency plus their in- tensity of interaction and positive knowledge of farm practices. They were also oriented toward the needs and desire of the local peOple. This characteristic, according to Curry, is positively related to success in the Extension Service and can be seen as a desire to serve others. This desire will increase the interaction between farmers and the extension worker. 42 Benn (4), 1952, found effectiveness of the extension worker related to certain attitudes and values. For in— stance, the effective agents: 1. Considered teaching people the value of an organized approach to the solution_9f_their problems and develOping efficiency of group action as highly important. 2. Considered increase in extent and effectiveness of group action and number of improved practices adopted by the extension audience as significant measures of teaching effectiveness. 3. Were vocationally better adjusted, liked their jobs better, and preferred it to other jobs. 4. Had more advanced college training since gradu- ation from college. One very singular analysis of the more and less effect- ive agent was done by a group of top level peOple, related with the Cooperative Extension Service in the United States. The title of this study is: Progress Report, Summary of the Research Study "The Differential Characteristics of More Successful Versus Less Successful Informal Extension Teachers." The committee which carried out this analysis was headed by Stone (John T. Stone), specialist in extension training, Department of Extension, Michigan State University. By a judgment method they differentiated the more and the less effective agent, who showed some difference in the following 43 characteristics: (1) Work Pattern - reflected in the county agent's annual report, Stone found that the more effective agent had more phone calls and more individual contacts. However, they did not differentiate on farm visits and num- ber of meetings held. Interpreting this, Stone said that this could indicate the more effective agent is more sought by the clients than the less effective agent. But as work initiators both had the same rank. (2) Division of time between office and field - this difference did not show up very distinctly in this study. And Stone's explanation is that it depends on the type of county and number of farmers worked with. The implication was that the larger the clients' number, the more time will be spent in planning and office work. (3) Use of time of the less and more successful agents: In reading, sorting mail and planning, there was no signif- icant difference between the two groups. But the higher agent was found to plan more and set up more goals for his work than the lower agents. They also discussed subject matter more with the specialists and supervisors. (4LIWork- ing with the farmers - the highly effective agent discussed more agricultural problems than the less effective agent. (5) Role behavior and interaction patterns - the more success- ful agent was found as: 5.1 - playing a moremactivewrolelin public programs administration, going frequentlylheyond the expectancy of his‘fOlerasflan extension worker. .nh" . aka...— v: *1, r __4 p.-- _...r...<, . A. h, _. 44 - working more with his colleagues from other agencies. - having a great ability to transfer work to the county leaders. - introducing a large variety of programs and trying to solve great numbers of county problems. - showing a great ability to recognize the needs in his county. His programs are planned in such a way that they can meet the needs of the county and clients. Swan (John Curtis Swan), in his MS thesis, "A Study 3" of Value as a Differential Characteristic of More~Effective and Less Effective County_§xtension Agents," presented a summary of predictors 6”“— factors studied in several other researches as of change_agent effectiveness. Reporting the W» ~~CW . conclusions of these research studies Swan states: do 2. «3 4. 5. /\ Age. The younger agents were equally as effect— ive as the older agents. Tenure in extension was not a factor in one study. In another it differentiated to some extent. Agents who had had general psychology courses were more effective. The more effective agents had more advanced college training since graduation from college. The vocational interests of the more effective agents were more like those of personnel directors and social science teachers. The vocational in- terests of the less effective agents were more like those of farmers and carpenters. The more effective agents liked their extension job better, preferred itmto othér jobs, and showed bétférrvocational adjustment. 45 7. The more effective agents assumed positive leadership in county program planning, having a more widespread formal planning group with membership from all segments of the population. 8. The more effective agents made a greater effort to reach rural people personally. y/ 9. The rural peOple made more effort to seek information from the more effective agents than they did from the less effective agents. 10. As a supervisor and organizer of events, the more effective agents used local leaders to per- form this role more often than did the less effective agents. 11. As an organizer of groups, the more effective agents organized and worked with more groups than did the less effective agents. 12. As a "salesman" of information and ideas, the more effective agents spent more time in the performance of this role and showed evidence of more initiative and originality in convincing persons that they should use the service. 13. The more effective agents had closer working relationships with their own staff members and with other agency representatives. \ a The more effective agents showed grfigaint‘en- sitywof interaction--greaterfldepthlandliréggency ofmcofifacfiwgth peOpleJ A very similar study of change agents' accuracy of estimating their clients' behavior was found, despite the lack of research on this particular subject. This was re- search done by Lutz, (Arlene E. Lutz) in Nebraska. In this research, a group of 16 change agents6 estimated the. 6Lutz defines change agent as: a professional who attempts to change his clients in a direction that he feels is desirable. 46 innovativeness7 of their clients, 92 farmers. This esti- mation was compared with the farmers' rating themselves and also with rating done by the interviewers in this research. The findings showed that: 31.9% of the agents agreed with the farmers' ratings, 44% missed by one category, and 25.1% were widely inaccurate. This variance in accuracy of estimating the clients' innovativeness was later tested with some agents' character- istics and the results were: as age and experience of change agent increased, accuracy decreased, at the 5% sig- nificance level. As score on the Quick word test increased, accuracy decreased, 10% level of significance. As years lived in the county increased, accuracy decreased, 10% of significance. (21:74) In Spite of the fact that none of these studies dealt with the accuracy of change agents in estimating farmer adOption, they researched the agent's effectiveness as re- lated to some characteristics which this study will consider; for instance, education, farm background, etc. On the other hand, considering the importance of perceiving the client's behavior with accuracy as an index of change agent effective- ness, this literature helps to build the hypotheses in this thesis. For example, farm background, agent-client inter- action, hard working agents, agent-client interpersonal *7 7 o I o o I Innovativeness refers relative time in Wthh a farmer adOpts an innovation in the community. 47 contact and so forth were found as positive factors in change effectiveness in these various studies. These variables will be hypothesized to be in the same relation- ship with agent's accuracy of estimating community adOption level. CHAPTER V HYPOTHESES AND FINDINGS TO BUILD A TYPOLOGY OF CHANGE AGENTS BASED ON THEIR ACCURACY IN ESTIMATING COMMUNITY ADOPTION LEVEL This study carries out Whiting's suggestion in the Phase I report (44:22). It attempts to build a typology of ACAR agents in terms of their accuracy in estimating the level of change. The limitation of this analysis is the sample size: N equals 20. To reach a level of significance with so small an N will be rather difficult. But the import- ance of the study for ACAR will overcome the lack of very conclusive results. Sampling The sample for this analysis was 20 agents out of the 38 selected for Phase I research. They are the same agents selected for Phase II, whose community sampling con— stituted a purposeful one to represent the dichotomy of the communities in terms of the most and least success of ACAR programs. Remembering the discussion of Phase I sampling of the communities, it was said that each agent was asked to name two communities under his jurisdiction--one in which his office had most success and the other in which it had the least success in changing programs. The representativeness 48 49 of this community dichotomy was the criterion for Phase II choosing communities on Phase II; ten most successful and ten least successful. Data The procedure for building this typology of agents was to test the relationship of some agent characteristics to agent accuracy in estimating community adoption level. This testing will tell if the highly accurate and low accurate agents come from different populations, classified by each characteristic. The findings on accuracy of the agents came from Phases I and II in which the estimations made by the agents were correlated with answers given by the farmers, both around community adOption level. The farmers' answers were considered the "real" community adoption level, and the deviance given by the estimation of the agents measured the accuracy of them in doing this estimation. This deviance was ranked, and the ranks varied from 1 for the mOSt accurate agent, to 19.5 for the least accurate agent based on 20 estimations. The characteristics of the agents were measured in Phase I, when the researchers analyzed change agent character- istics as predictors of farmer adoption of new ideas. Each agent was interviewed by a member of the Research Diffusion 50 Project team.8 The interview guide contained questions measuring: (1) education, (2) agent communication style, (3) work adjustment, etc. These 15 characteristics will be tested against accuracy of the agent in estimating com- munity adOption level. Hypotheses:9 In a hypothetical schema the relationships to be studied would be stated as: I. The agent at his job: A. Time spent to perform the job: the agent who spends more time in his job will be more accurate. B. Time in the different sections of the job. 1. The agent who spends more time in the field will be more accurate. 2. The agent who Spends more time in the office will be less accurate. C. Agent interaction with his supervisors: Agents who feel their supervisors will listen to them will be more accurate 0 8The interviews were carried out in January and Feb- ruary of 1966. The author was the supervisor of the field collectors data. His title in the Diffusion Research Project was Research Assistant. The interviews took about one hour with the local ACAR agent, at his office located in the town. The author did some of the interviews and also he worked in all the offices in Phase II data gathering. A c0py of the interview guide will be placed at the end of this report. 9The hypotheses will be discussed with the statistical analysis and presentation of the findings. II. III. 51 D. Work preferential: The agents whose work prefer- ential is more community oriented will be more accurate than the agents who prefer office work. Agent style of communication A. As user of different communication methods: The agents who use more personal kinds of communication in taking information to farmers will be more accurate than the other agents more mass oriented in their communication methods. B. Preference for certain kind of communication methods: Agents who feel that individual contact is more effective in communicating than group discussion will be more accurate than agents preferring group discussion over individual contact. Agent-client interaction A. Sought by clients 1. The agents who feel the community asks for their help without their alerting the community will be more accurate than agents who feel they need to alert the community to the kind of help they can give. B. Leadership style: The agents who prefer more author- itarian style of leadership will be less accurate than the agents preferring a more permissive style. -52 C. Disposition to listen to ignorant people: Those agents with more disposition to listen to ignorant peOple will be more accurate than the agents who dislike listening to ignorant people. IV. Agent background A. Farming experience 1. Agents born on farms will be more accurate. 2. Agents with farming experience will be more accurate. B. Experience with ACAR service: The agents who worked for ACAR for a longer period of time will be more accurate. C. Agent-client tenure:r The agent who has worked more time in the same area will be more accurate. V. Agent education: College-degree agents will be less accurate . Statistical Analysis Two statistics were used to test the null hypotheses in this study of change agent accuracy in estimating the community adoption level as related to some agent character- istics: "The Kruskal-Wallis One-Way Analysis of Variance by Ranks" and "The Spearman Rank Correlation Coefficient: rs." This duality of statistical approach is due to different Operationalization of the independent variables in the study. 53 For instance in the case of "education," the data presented two groups of agents or two independent samples of agents--one group with college education and another group with high school training. As such this variable calls for a test of "K independent samples," (37:174) and the test chosen was "the Kruskal-Wallis." On the other hand, in "time the agent spends working in the field" was measured by asking each agent the percent- age of his time Spent in field work, which gave different percentages in each case. These numbers were ranked, the highest received rank 1 and the lowest received 20. With this agent ranking and the accuracy measurement, it was possible to use a correlational process to test the relation- ship between the two variables. The correlation coefficient would tell if there was a correlation, its magnitude, and also its direction. The correlational method used was "The Spearman Rank Correlation." "The Kruskal-Wallis One-Way Analysis of Variance bngank-ETr This very useful nonparametric statistic does not require many assumptions for its use. The basic assumptions are: the variable must have an underlying continuous distri- bution and present at least ordinal measurement. (37:185) The present data satisfy these assumptions. Presenting this test, Hays states: "in comparison with the analysis of variance, the Kruskal Wallis test Shows up extremely well." 54 (13:639) Another very well known statistical book, "Non- parametric Statistics for the Behavioral Sciences," Siegel presents this test as "an extremely useful test for deciding whether k independent samples are from different pOpulations." (37:184) "The Spearman Rank Correlation Coefficient: rs" This is a very old nonparametric correlational tech- nique and according to Siegel (37:202) it is still the best method known to test correlation of measurement in ranking form. The only requirement is that the scores must be in rank order measurement. Referring to this process of com- puting correlation, Downie and Heath, (N. M. Downie and R. W. Heath, 1965) say that: "This is the most widely used of the rank correlational processes. It is particularly well suited to situations where the number of cases are 25 to 30 or less." (11:207) Also Borg, in his book, refers to this test as very useful, time saving, and almost as precise as the "Product Moment Correlation: r." Borg says that the difference caused by ranking the scores does not affect very Significantly the size of the real correlation if it exists between the variables. (6:150) Presentation and Analysis of the Findings The agent characteristic findings collected in Phase I of the Diffusion Research Project were used in building a 55 typology of agents according to their accuracy of estimating the level of adOption of their clients. So the dependent variable in this study is accuracy of the agents in estimating the communityyadoption level. This was measured in part by asking the agents to estimate the percentages of farmer adOption on six ACAR promoted new agriculture practices, in each community under the agent's jurisdiction. The Independent Variables By using findings from Phase I, it is possible to operationalize the following variables: 1. Agent's education 2. Agent background 3. Agent-client interaction 4. Agent preferred style of communication 5. Agent at his job. A series of hypotheses manipulating these variables will test if a group of agents is more accurate and also if some characteristics can explain to some extent the variation in accuracy in estimating the community adoption level. 1. Agents' education This characteristic of the agents was defined pre- viously as being the period in which the agents attended school before going to work for ACAR. The agent's education was di- vided into: college (14.5 years of schooling); high school 56 (10.5 years); and, partial high school education (5.5 years attending school). Those with college education were mostly agronomists and veterinarians, while high and partial high school education included agents primarily with some years of agriculture and home economics training. Due to scarcity of high school and partial high school agents, these groups were combined and considered as high school education. Thus, the agent's education was divided into college and high school. These data are presented in Table 17 with agents ranked according to their accuracy in estimating community adOption level. The hypothesis relating education with accuracy was that the less educated agents will be more accurate than the more educated agents. This was the alternative hypothesis and the null hypothesis for the statistical testing was: no difference between the groups, high school and college education agents. The probability that the difference, given by "Kruskal Wallis" test will be, by chance, 10 percent, is quite a high probability within research analysis. So the null hypothesis of no difference stands and the alternative is rejected. It is possible to say that in this sample of 20 agents there is no difference in accuracy between the more and less educated. The hypothesis relating these two variables was based on the amount of heterophily10 placed by high education on 10HeterOphily is defined as the cultural distance between source and receiver in communication process. (35) "M "u"z'u‘fiv .nn.. m: “I.“ . c .- "ap 57 Table 17. Rank order of accuracy of the agent with college and high school education Accuracy in Estimating Community Adoption Level Education College High School 9.5 7.0 5.5 12.0 15.5 4.0 2.5 12.0 15.5 1.0 19.5 12.0 17.0 8.0 19.5 5.5 14.0 9.5 12.0 2.5 Since these two groups constituted "k independent samples," "The Kruskall Wallis Analysis of Variance" was used to tell if the group with high school education and the group with college came from different populations with respect to accuracy. 58 the interaction between the agents and farmers (35) When the amount of heterophily is high as in case of college degree agents and the farmers, the interaction tends to decrease between them. By interacting less the agents probably would lose some of the client knowledge and conse- quently would be less accurate also. However, this hypothesis was not confirmed by the present findings. Lutz, studying characteristics of the agents as pre— dictors of change agent ability to estimate farmer innovative— ness, did not find any significance between agent accuracy to estimate the farmers' innovativeness and years of schooling attended by the agents. (21:86) 2. Agent Background» The Phase I data collection measured this variable by asking the agents: a. If they were born in the country (rural area) or in the city. b. How long they had worked in farming before going to ACAR. c. How long they had worked for ACAR. d. How many months they had worked in the area for which they estimated farmers' adoption level. a. Agent Place of Birth: Agents born in the country were supposed to have more experience in farming business and also would share more contact with 59 farmers than the agents born in the city. So the hypothesis in this case was that: the agents born in the country would be more accurate than the agents of city birthplace. Table 18 presents the rank of accuracy of these two groups of agents. Table 18. Rank order of accuracy in estimating the community adOption level for agents born in the country and in the city. Birthplace gipy Country 5.5 9.5 15.5 12.0 2.5 17.0 12.0 19.5 15.5 9.5 7.0 2.5 1.0 19.5 18.0 4.0 8.0 5.5 14.0 12.0 The "Kruskal-Wallis" test gave H equals 3.5, signif- icant at 10% level. of Siegel calls for a H of 3.84. To reach a significance of 5%, "Table C" So this alternative hypothesis 60 does not stand and the null hypothesis of no difference‘ between groups will stay. Adding some comments more to it, one can notice that the findings showed slightly the Opposite trend from the predicted hypothesis. Agents born in the city tended to be a little more accurate than their partners born in the country. b. Agent with Farming Experience: The same logical thought given to the first hypothesis can be applied to this second. Agents who worked in farming would have more eXperience in this field, would interact more with their clients and also would have a better understanding of farmers' way of life. Table 19 shows the data concerning this variable. However, a different approach toward measurement was used in which the years of agent working in farming are replaced by ranks. The agent with the most experience received 1 and the agent with least experience received number 14. With two variables in rank order measurement it is possible to apply the "Spearman Rank Correlation Coefficient," whose formula corrected for ties was taken from Siegel. (37: 207) The significance of rs is tested by using a t test. This t test value of significance is brought by "table B" of Siegel, which tells the probability of the rs being dif- ferent from zero. (36:210) 61 Table 19. Agents' years of farming, ranking of farm experience and rank order of accuracy of the agents Rank order Years in farming Rank order of accuracy 1 7.0 9.5 0 14.0 5.5 0 14.0 15.5 3 5.5 2.5 0 14.0 12.0 3 5.5 15.5 6 3.0 7.0 8 1.0 1.0 O 14.0 19.5 0 14.0 18.0 0 14.0 12.0 0 14.0 4.0 O 14.0 17.0 6 3.0 8.0 0 14.0 19.5 0 14.0 5.5 0 14.0 14.0 0 14.0 9.5 0 14.0 12.0 6 3.0 2.5 The "Spearman rank order correlation" formula applied in Table 19 of this study gave a coefficient of .517. By using a t test of Significance, it showed that this coefficient has a probability of being different from zero of 99, which is quite high and is an accepted research standard level of 62 significance. This correlation coefficient thus supports the alternative hypothesis that the agents with farming background are more accurate in estimating community adop- tion level. Wondering about the large number of ties, the author divided these agents into groups low and high in farming experience. He considered those with zero or one year as low, and those with 3, 6, and 8 as high farming experience. Applying "Kruskal-Wallis" to these and the accuracy ranks gave an H of 8.16. This H is significant at .01 level, showing that the agents did differ as far as accuracy is concerned. Therefore, the hypothesis of the agent with farming experience tending to be more accurate than the agent who missed this kind of experience is confirmed by the present analysis. .b. Experience in ACAR Service: The average time working for ACAR in this group of 20 agents was 3.5 years. Thus, it is possible to assume that some of them have an acceptable amount of experience in ACAR service working as extensionists. As such they should know quite well how to identify those clients using the ACAR-promoted innovations. With this assumption in mind it is possible to say that those agents who are older in ACAR service also will be more accurate. The years working for ACAR were transformed into ranks. They are presented in Table 20. The oldest received rank 1 and the youngest received 19. 63 Table 20. Years agents have worked for ACAR, rank order of work time, and rank of accuracy. Years Agents worked for ACAR Rank order Accuracy 5 4.5 9.5 4 7.0 5.5 2 14.0 15.5 4 7.0 2.5 1 19.0 12.0 10 2.5 15.5 13 1.0 7.0 5 4.5 1.0 2 14.0 19.5 2 14.0 18.0 2 14.0 12.0 3 9.5 4.0 10 2.5 17.0 2 14.0 8.0 3 9.5 19.5 2 14.0 5.0 1 19.0 14.0 2 14.0 9.5 1 19.0 12.0 4 7.0 2.5 64 The "Spearman rank correlation" was equal to .29, which is not significantly different from zero. The findings do not support the alternative hypothesis that the more ex- perience an agent has in ACAR service the more accurate he will be in estimating the client's level of adOption. d. Time Working in the Area: This characteristic tells about the agent-client tenure or the amount of time the agent had worked in the area for which he estimated the client adoption level. The average, in months, on the 20 agents was 26.17. This is not a long time to get acquainted with the community situation, but there were some agents in this group with considerable time in the area, for instance those with 36, 48, 72, and 94 months. These agents supposedly should know their clients much better than those who did not have such long tenure. Tenure was ranked and compared with the agent accuracy ranking. The agent with the most time was ranked 1. (Table 21). Applying "Spearman rank correlation coefficient" the results showed a rs of -.49, a negative cor- relation between agent-client tenure and accuracy of these agents in estimating the client adoption level. Checking "t" significant level of this correlation, it presents 95% of being different from zero. Its magnitude explains almost 25 of the variance on accuracy measurement. (20:205) Therefore, it can be said that in this sample the agent with less time in the community or with a shorter period of client 65 Table 21. Time in months of agent-client tenure, its ranking, and the accuracy ranking order. Rank order Accuracy Months of this time ranking 6 19.0 9.5 20 13.0 5.5 24 7.0 15.5 12 16.0 2.5 10 17.0 12.0 22 11.5 15.5 72 2.0 7.0 5 20.0 1.0 24 7.0 19.5 24 7.0 18.0 19 14.0 12.0 24 7.0 4.0 94 1.0 17.0 24 7.0 8.0 36 4.0 19.5 22 11.5 5.5 18 15.0 14.0 23 10.0 9.5 8 18.0 12.0 48 3.0 2.5 tenure was more accurate in judging the clients' level of adOption. The alternative hypothesis was the contrary, say- ing the agents with more client tenure would be more accurate. This hypothesis is rejected in favor of the evidence opposite, showed by the coefficient of correlation: -.49. 66 Berlo gives some possible explanation for the above results. He points out that after a certain time the com— munication between source and receiver becomes a kind of routine and no longer provokes the necessary reaction in both elements. (5:14) Rogers, in explaining the effective- ness of the change agent in the community, said that: this effectiveness will grow, will have a flat period, and after time will drOp. (35) The author explains by saying that the change agent in long-time client contact will absorb the client culture and cease to be a change agent in the com- munity. These generalizations might shed some light on the present findings of agent accuracy and time worked in the community. However, it is necessary to be cautious before making any decision based on these results. The small sample size and the amount of explanation given by the correlation coefficient would seem to make a very precise conclusion difficult. In Table 21 the data Showed that the agent with the longest tenure was highly inaccurate-—he ranked 17 out of 20. The most accurate agent had the Shortest agent-client tenure-~only 5 months in the area. 3. Agent-client interaction This variable--very important in this analysis of local ACAR agent accuracy-«was measured also in Phase I data collecting. The researchers asked the agents: 67 a. Which kind of help their clients mostly sought. b. If the agents would prefer an authoritarian leadership with their clients or would rather be permissive in dealing with the farmers. c. If the agents were disposed to listen to ignorant people. a. Kind of help mostly sought by the clients: A five- I item scale measuring this subject was given to the agent. F He was asked to choose the one mostly fitted to his own situations: ij (1) "People ask help exactly on those problem areas a they are ready to confront." (2) "Recognize their problems and accept help when offered." (3) "Do not know or no answer." (4) "NecesSary to point out existing problems the ACAR agent can solve." (5) "The majority of them refuse to recognize that they have problem the ACAR agent can help with." The 20 agents used in this report chose (3) and (4) mostly. Only one chose (2), and this was also the most accurate agent in the group. Unfortunately the lack of vari- ation made it impossible to check the difference in accuracy of the two groups of agents since at least four more choosing (l) and (2) were needed to apply the statistical analysis. 68 b. Agent leadership style: This was measured by ask- ing the agents which method of working with less educated peOple was the best: (1) "be authoritarian and tell the- people what to do", or (2) "encourage them to make their own decisions." People in general do not appreciate authoritarian persons and they tend to avoid this kind of individual. This applies even more to farmers who are very proud of their "knowledge" about agriculture and dislike being given orders. So agents who preferred an authoritarian style of leadership would tend to rate lower with clients. This very fact would cause some misjudgment as to clients' adoption level. The reverse could be applied to permissive agents. So the hypothesis in this case was that the agents who preferred a more permissive style of leadership would be more accurate than their partners who preferred an author- itarian way of dealing with farmers. In Table 22 the accuracy ranks of these two groups are reported. The two group averages did differ in the hypothesized direction: the permissive agents tend to be more accurate than the authoritarians. "Kruskal-Wallis" was applied to data-in Table 22 in order to check the significance of the difference between the two groups. The result was an H equal to 3.4. "Table C" of Siegel's book gives this value of H as significant at .10. At this level, the alternative hypo- thesis is rejected and the permissive and authoritarian agents A L {hIJ'RI .{fi-‘khll ‘- -h .. . In. 4' 69 are considered as having the same accuracy in estimating their client level of adOption, as far as these 20 agents Showed. Table 22. Ranks on accuracy of the agents preferring authoritarian and agents preferring per- missive styles of dealing with pe0ple with little intelligence. Leadership Style Authoritarian Permissive 15.5 9.5 15.5 5.5 14.0 2.5 2.5 12.0 7.0 1.0 19.5 18.0 12.0 4.0 17.0 8.0 19.5 5.5 9.5 12.0 utfl‘ipl‘i‘j. ‘_.'E' .‘ firm :.n1 (3m T. y . . 70 c. Disposition to listen to ignorant peOple: This was measured by asking the agent if he "appreciated their talk- ing about their problems" or "he does not want to waste time with this." The 20 agents did not Show any difference in answering this question. All of them answered that they had the disposition to listen to ignorant peOple. Thus, the hypothesis was not borne out. 4. Agent Style of Communication Two devices were used to measure this characteristic: a. An average monthly use10 of several communication techniques was supplied by the central office as being the general extension communication methods used by its agents: (1) conversations; (2) personal demonstrations: (3) demonstrations of results; (4) small group meetings; (5) demonstrations for groups; (6) excursions; (7) talks; (8) movies; (9) printed materials, and (10) radio. b. By asking the agent what he thought was the most effective way to take ideas to the farmers; (1) individual discussion or (2) group meetings. Using the above measures, inter-personal contact be— tween the agent and his clients was operationalized this way: 10The agent was asked: how many times per month do you use (1) "conversation" with the farmers; (2) "personal demonstration"; (3)"demonstration of results," etc. 71 ad. A group of communication techniques was chosen to represent interpersonal communication: conversations, personal demonstrations, demonstration of results, and meetings. The agents high in rank using these techniques were considered high in personal contact with clients. These communication methods were weighted: conversation received a weight of 4, personal demonstration 3, demon- stration of results 2, and small group meetings 1. This weight was multiplied by the agentJS monthly use score on each of these techniques, the four weighted uses of the interpersonal communication methods were added, and this score represented the index of interpersonal communication use of each agent. These indices were ranked; the largest received number one and the smallest received number 20. Table 23 presents the total weighted use score of each agent in all four interpersonal communication techniques, its ranking and the rank order of accuracy of the agents. a.2. Another group of communication techniques was chosen to represent mass media communication use. The agent frequently using these communication methods would be expected to be low in interpersonal contact with his clients. The methods chosen were: demonstration for group, excursions, talks, movies, printed material, and radio. These techniques were also weighted: radio received weight 6, printed material 5, movies 4, talks 3, excursions 2, and demonstrations for groups 1. This weight was multiplied by each agent's score Faun __i ‘8’\m.‘ gotta-m”: ana‘ 72 Table 23. Agent's use of the extension methods chosen to represent interpersonal communication Total weighted score of each agent in the four Ranking of Accuracy interpersonal communi- these ranking cation techniques scores order 27 7.5 2.5 23 17.0 5.5 27 7.5 15.5 28 3.5 2.5 26 11.0 12.0 26 11.0 15.5 22 18.0 7.0 25 14.0 1.0 26 11.0 19.5 28 3.5 18.0 25 14.0 12.0 28 3.5 4.0 24 16.0 17.0 27 7.5 8.0 29 1.0 19.5 18 20.0 5.5 20 19.0 14.0 25 14.0 9.5 27 7.5 12.0 28 3.5 2.5 aThe number of times per month he used the method was multiplied by its weight, for instance: in case of "conversation" it was multiplied by 4 and the products on the 4 chosen methods were added to represent the agent use of interpersonal communication. 73 in using these techniques per month, and these scores were added, giving the agent index of mass media communication approach used in the community. These indices were ranked; the smallest received 1 and the largest received 20. Table 24 presents these indices, and their ranks and the accuracy ranking. Applying "Spearman rank correlation" to data from these two tables, the results showed that: The correlation coefficient between the agent highly using interpersonal contact communication approach and the accuracy of this agent is equal to .07. This extremely low coefficient has a very high probability of being not dif- ferent from zero, so this sample of 20 agents did not show relationship between use of interpersonal communication and accuracy in estimating community adOption level. This makes it possible to reject the alternative hypothesis that the agents who ranked high in using interpersonal communication techniques also would be more accurate. This same procedure was used with the mass media use, and coefficient correlation found between limited use of mass media by the agents and accuracy was equal to .35. Applying a test of significance on this coefficient (t), it was found Significant at a level of 10%. So, the alternative hypothesis of the agents who are lesser users of 74 Table 24. Agent's use of the extension methods chosen to represent gse of mass media in communication with farmers. Score weighted and added on six mass Accuracy communication Its rank techniques ranking order 49 14.5 9.5 34 2.0 5.5 g 55 19.5 15.5 f 45 9.5 2.5 41 5.5 12.0 49 41.5 15.5 25 1.0 7.0 37 3.5 1.0 50 18.0 19.5 41 '5.5 18.0 45 9.5 12.0 49 14.5 4.0 49 14.5 17.0 37 3.5 8.0 44 8.0 19.5 55 19.5 5.5 49 14.5 14.0 46 11.0 9.5 49 14.5 12.0 48 7.0 2.0 aThe number of times per month he used the method was multiplied by its weight, for instance in the case of use of the radio, it was multiplied by 6 and the products on the 6 chosen methods were added to represent agent use of mass media. 75 of mass media in communicating with the farmers also being more accurate was-rejected.ll b. Group meetings vs. individual discussion. This com- munication agent style was also measured by another question in the interview guide. Agents were asked to choose between group meetings and individual discussion as to which they considered the most effective to take ideas to farmers in their communities. The answer given by the agents permitted division into two groups: (1) Those choosing individual discussion as the most effective method; (2) those choosing group meetings as the most effective. By using this, it was tried again to Operationalize personal contact between agent and clients. The agents who chose individual discussion as the most effective method could be considered in closer personal contact than the group choosing group meetings. 11This analysis was tried again using a different approach; the methods were weighted in total: conversation received weight 10, personal demonstration 9, personal demonstration of results 8, small group meetings 7, demon— strations for group 6, excursions 5, talks 4, movies 3, printed material 2, and radio 1. These weighted scores of the agent use per month of the methods were added and ranked and these rankingscorrelated with accuracy and the results showed a coefficient correlation of .25, which is not sig- nificant. 75 Applying the "Kruskal-Wallis" test to these data, H=5.89, Table C of Siegel's book gives this value of H as significant at the 2% level. With this level of signifi- cance it is possible to say that these groups do differ as far as accuracy in estimating community adOption level is concerned. Looking at Table 25, one can see the group who chose individual discussion as the most effective method has lower accuracy rank on the average. Therefore, the alternative hypothesis that the agents who preferred indi- vidual discussion would be more accurate is confirmed at 2% level of significance. And the null hypothesis of no difference between groups was rejected. Table 25. Accuracy rankings of the agents who chose individual discussion as the most effective method and those who chose group meetings as the most effective method. Group Individual meeting discussion l—' PH PH mowmwwmm . O O I O C O O omomommm H \l O 77 The agent who has high interpersonal contact with his clients tends to be more accurate in estimating the adOption level of these clients. 5. Agent at His Job This was another aspect of the local ACAR agent anal- yzed by Phase I research. The data pertinent to this sub- ject were used to try to explain variation in accuracy of the agents. The interviewers asked each agent: a. How many hours a week he Spent doing his job. b. How much of this time, in percentage, he spent at the office. c. What percent was used in field work. d. The aspect of the job he preferred. e. About feedback in the organization. a. Time spent on the job. The local ACAR agent is a full-time employee. As such, he is supposed to work from 8:00 to 11:00 a.m. and from 12:00 to 5:00 p.m., with half hour break for lunch. However, this time schedule frequently is even longer due to field work. The agent stays visiting or in meetings with farmers more than he expected and this will put him with different hours of working a week. So if agents work more time a week than the regular hours they are supposedly harder workers, like their job more, and so tend to be more accurate in knowing the results of their performance. Therefore, the hypothesis on the time spent 78 on the job was: agents who Spend more time doing their job will be more accurate than agents who spend less time a week in the Extension Service. Table 26 reports the agent hours per week on the job, the rank order of this time and also the rank order of the accuracy of the agents. The agent with the largest number of hours per week received rank one and the smallest number of hours received rank 19.5. These ranks were correlated with the ranks of the agent accuracy. The result was rs equals .085, not significant. Thus, accuracy of agents in estimating com— munity adOption level is not correlated with the hours spent on the job. b. Time Spent in the office and time spent in the field. Working as an Extensionist the agent works mainly in the field in which he conducts meetings with farmers, visits the rural families, and supervises the demonstration of results or organizes clubs and group demonstrations. How- ever, part of his time is spent in the office, when he makes plans, contacts peOple related to his job that live in the urban section of his area of working and also answers farmers calls. The agent who spends more time in the field in direct contact with the farmers naturally Should have more evidence of his work. Also he will have more contact with his clients which enables him to distinguish more clearly the adOpters ‘31....- 1'. 79 Table 26. Hours a week spent on the extension job, rank order of this time, and accuracy ranking. Rank order Accuracy Hours a week of this time ranking 60 5.0 9.5 44 16.5 5.5 60 5.0 15.5 70 1.5 2.5 a 47 13.0 12.0 5‘1} 70 1.5 15.5 } 44 16.5 7.0 L 60 5.0 1.0 L. 50 10.5 19.5 * 60 5.0 18.0 54 9.0 12.0 50 10.5 4.0 48 12.0 17.0 45 14.5 8.0 38 19.5 19.5 55 8.0 5.5 38 19.5 14.0 60 5.0 9.5 44 16.5 12.0 40 18.0 2.5 of new ideas from nonadopters. Hence, this agent should be more accurate in knowing the community adoption level. On the other hand the agent who Spends more time in the office will have less contact with the rural community and so will be less accurate in knowing the level of adoption in this community. 80 Tables 27 and 28 bring together the time of agents working in field and office and their accuracy measurements. Table 27. Percentage of agent time spent in field work, its ranking order and accuracy ranks Percent time working in Rank Accuracy the field order ranking 75 9.5 9.5 79 14.5 5.5 80 4.0 15.5 68 16.0 2.5 73 13.0 12.0 90 1.0 15.5 81 2.0 7.0 65 17.0 1.0 75 9.5 19.0 75 9.5 18.0 40 20.0 12.0 70 14.5 4.0 75 9.5 17.0 80 4.0 8.0 55 18.0 19.5 75 9.5 5.5 55 18.0 14.0 76 6.0 9.5 80 4.0 12.0 75 9.5 2.5 81 Table 28. Percentage of time Spent in the office by the agents, its ranking and accuracy rank order. Percent time working in Rank Accuracy the office order ranking- 20 12.5 9.5 20 12.5 5.5 20 12.5 15.5 30 17.5 2.5 25 17.5 12.0 10 1.0 15.5 19 10.0 7.0 15 4.5 1.0 15 4.5 19.5 15 4.5 18.0 30 17.5 12.0 25 19.5 4.0 20 12.5 17.0 15 4.5 8.0 40 20.0 19.5 15 4.5 5.5 35 19.0 14.0 16 9.0 9.5 15 4.5 12.0 15 4.5 2.5 The"Spearman rank correlation" was applied to data in Tables 27 and 28. The results were: For Table 27 in which the agent with most time in the field received 1 and the agent with least time received rank 20, the correlation between these ranks and accuracy was .03. 82 For Table 28, where the agent with least time in the office received rank 1 and the agent with most office time received 19.5, the correlation between these ranks and accuracy was -.14. Both coefficient correlations have a high probability of being no different from zero, according to a "t" test of significance. Therefore these results do not support the alternative hypothesis that the agents with more time spent in field would be more accurate in judging their clients' level of adoption. d. Agent Preferential Kind of Work: Agents were given 7 items to choose among concerning that aSpect of their job they liked most. The items were furnished by central ACAR office. They were supposed to cover all the different aspects of the extensionist job in ACAR service. They were: 1. Office work: administrative and bureaucratic details, reports, investigations, etc. 2. Introduction of new technology: work with specific agricultural products: milk, cows, corn, etc. 3. Demonstrations. 4. Other answers, no answers, don't know.12 5. Work with credit. 12This item was added to facilitate later tabulation of the data. 83 6. Visits in the field: work in the communities, rural contact, teaching, develOpment of rural life. 7. Work with clubs and groups: 4-H, adult groups, etc . The 20 agents used in this study chose in terms of l, 2, 4, 6, and 7. By using these preferences for certain kinds of work it is possible to Operationalize the amount of con- tact of the agent with his clients. For instance the agent who likes to work with clubs and groups would be expected to have a higher amount of contact with his clients than the agent who prefers office work. Only one agent chose 4 and he was arbitrarily considered an office-oriented agent. The agents who preferred 5, 6, and 7 were considered field work-oriented agents and as such these agents, by having more opportunity to contact their clients in the field, would be expected to be more accurate than the agents who chose any of the above items (4,3,2,l). So, the hypothesis in this case was: the agents who are more field-oriented are more accurate than office-oriented agents. Table 29 presents the rank order of accuracy of the two groups. The "Kruskal-Wallis" was called to judge this differ- ence in accuracy between the groups. The result was H=3.82. To reach a significance of 5% level, it was necessary to reach a value for H of 3.84 (37:210). So, in this case, the prob- ability is that the existing difference is a matter of chance. '84 Table 29. Rank order of accuracy of the agents who are more field-oriented and the agents who are more office-oriented. Agent's preferential work Field-oriented Office-oriented 9.5 5.5 12.0 15.5 7.0 2.5 18.0 1.0 12.0 19.5 4.0 19.5 17.0 5.5 8.0 2.5 14.5 9.5 9.5 12.0 So, the null hypothesis of no difference between groups is supported and the alternative saying that the field-oriented agents would be more accurate was rejected. However, it is worthwhile to mention that the difference between groups did not come up as it was predicted. Looking at Table 29, one can see that in average the office—oriented agents were slightly more accurate than their partners in judging community adoppion level. 85 f. Feedback in the organization (ACAR). It was measured by asking the agent if: 1. "He thought his regional supervisor would listen to him." 2. "He does not know." 3. "His supervisor does not pay attention to his ideas" (agent's). If the agent thinks his supervisor listens to him, he would be expected to have a high interaction with him and he would try to tell the supervisor with accuracy the results of his work. Besides administration work, the regional super- visor is charged with orientation and help to the local agents, advising them how to do a better job. Therefore, the hypo- thesis in organization feedback was that the agent who feels his ideas are listened to by his immediate supervisor would be more accurate than the agent who does not feel this way. Table 30 compares the accuracy of these two groups. The "Kruskal-Wallis" test was applied to these two groups and it was found that H=3.75. The probability of this difference to be by chance is 10%. Hence, the alternative hypothesis of the agent who felt his supervisor would listen to his ideas being more accurate is not supported. Unfortunately, the lack of statistical significance rejected moSt of the hypotheses in this analysis. It is only possible to say that: 86 Table 30. Rankings of accuracy of the agents who said yes or no to the question about supervisor listening to the agent's ideas. Agent's Answers Yes NO 5.5 9.5 15.5 1.0 2.5 18.0 2.5 12.0 12.5 19.5 15.5 9.5 7.0 19.5 4.0 17.0 8.0 5.5 14.0 12.0 2.5 l. The agents with farming experience were more accurate than the agents who lacked this experience. 2. The agents with less client-tenure showed more accuracy in judging the level of adOption. 3. Agents who considered individual discussion the most effective method of teaching farmers were more accurate than the agents who considered group meetings as the most effective approach to take ideas to their clients. Probably the small sample size made impossible further distinctions between accurate and inaccurate agents. CHAPTER VI SUMMARY AND CONCLUSIONS The objective of this study was to build a typology of change agents based on their accuracy in estimating com- munity adoption level. Change agent was defined as a pro- fessional individual who attempts to change his clients in a direction he feels is desirable. (32:254) To represent these agents, 20 extensionists of a Brazilian state extension service agency were chosen. These extensionists attempt to increase agriculture production and farmers' income by intro- ducing and getting adOpted new agricultural techniques. The 20 agents were asked to estimate the percentage of their clients using six new agriculture practices promoted by their offices in the last three years. These estimates permitted ranking the 20 local offices from highest to lowest (20 to l) on the acceptance of the extension service promotion of farming innovations. Later on, this program acceptance was measured on the farmer level; in each community an exhaustive interviewing of all farmers measured this program acceptance. By using their answers it was possible to rank- order the 20 communities from the highest to lowest (20 to l) as far as acceptance of the ACAR promotion program is concerned. This constituted the "real” community adOption level of pro- grams of this agency extension service. 87 88 The farmer measurement was compared with the agents' estimation. The results showed that: 1. Eight of the 20 agents underestimated the clients' level of adOption, with a mean error of 8 ranks. 2. Six over—estimated, with a mean error of 11 1/2 ranks. 3. Six were quite accurate in their judgment of com- munity adOption level. For some agents, the discrepancies were quite large. With others, they were not so. Farmer answers were considered the "real" community adOption level. The difference between that and the agent ratings measured agent accuracy in esti- mating client adOption level. The discrepancies were ranked: the smallest received 1 and the largest 19.5. This also represented the most and least accurate agent, respectively. These accuracy ranks were tested against several agent characteristics in order to see if there was any important difference between the most and least accurate agent. Two statistics were used to test the hypothesis: "The Spearman Rank Order Correlation Coefficient, "rs" and "The Kruskal- Wallis One Way Analysis of Variance." By using data gathered about the 20 agents, it was possible to operationalize the following subjects which characterize the local ACAR agents: 1. Agent Education 2. Agent Background 89 3. Agent-Client Interaction 4. Agent Style of Communication 5. Agent at His Job 1. Agent Education The findings showed that the college educated agents (14.5 mean years of schooling) were as accurate as the high school agents (10.5 mean years). 2. Agent Background This factor was measured by: a. Agent birthplace: country or city. b. Agent experience working on farm. c. How many years they have worked for ACAR. d. How long they have worked in the area in which they estimated the adoption level. The only items in the above list that correlated with agent accuracy were agent experience in farming work and time in the area in which they estimated the adoption level. The experience in.farming with accuracy of the agents correlated positively, with a correlation coefficient of .57, which is significantly different from zero at the .01 level. In this case the previous hypothesis was confirmed that said: phg agents with more farming eXperience would be more accurate. Rahudkar, studying characteristics which differentiate the more and less effective agents, found that the most 90 effective agents had rural background; some of them were born on farms, had worked or lived for some time in rural areas. (30:120) Similar findings were reported by Maunder. (22:75) The second item in the list of agent background was time worked in the area in which the adOption level was estimated. This variable correlated negatively with accuracy and its correlation coefficient was -.49, significant at .05 level. However, the agent Should not be too recent, nor be too old in his area of working. The most accurate agent had 5 months in his community and the oldest agent (with 94 months), was among the most inaccurate agents, ranking 17 out of 20, in which the most accurate agent ranked 1. 3. Agent-Client Interaction This subject was measured by: a. Kind of help the clients mostly felt. b. Agent leadership style: if the agent preferred to be authoritarian or permissive in dealing with less educated people. 0. Agent disposition to listen to ignorant peOple. Lack of variance made the statistical analysis of a and g impossible. Item 2 was found Significant at .10 level, test- ing the hypothesis that: the more permissive agent would be more accurate in estimating client adOption level. This hypothesis was invalidated because of the low level of significance. 91 4. Agent Communication Style The attempt was to operationalize agent-client inter- personal contact by using his communication approach. So the analysis tested: a. Agent high use of interpersonal communication methods. b. Agent low use of mass media communication methods. c. Agent preference for certain kind of communication method: individual discussion versus group meeting. The agent low in using mass media to communicate with clients, correlated .35 with accuracy. This coefficient was not significant. The coefficient found between high use of interpersonal communication and accuracy was -.08, not significant. The agent preferential communication approach did Show some difference on accuracy ranking. The level of significance was 2%. The agents who preferred individual discussion were more accurate than their partners who preferred group meeting as the most effective method to take ideas to farmers. 5. The Agent at His Job: This factor was measured by considering: a. Time spent in the job by the agents during a week (hours). b. Time in percentage spent in field work. c. Time Spent in office work. d. Aspect preferential of the job. e. Feedback in the organization. 92 None of the above variables showed significant cor- relations with agent accuracy in estimating community adop- tion level. However, g and 3 almost reached 5% level of significance. d. Aspect preferred on the job—-this was used to Operationalize the agent contact with the clients in the field. The agents who preferred more field work such as work in the rural communities, work with clubs or groups of clients, etc., were considered in more contact with the clients and were also expected to be more accurate in judging the clients' level of adoption. Those agents who preferred more office- oriented work would be less accurate, accordingly. The dif- ference almost reached 5% level of significance: A 3.82 value was found with a value of 3.84 being necessary in order to be Significant at 5% level. However, this difference was in Opposite direction from the previously hypothesized: the agents more Office-oriented showed slightly more accurate than the field-oriented work agents. e. Feedback in the organization— This was measured by asking the agents if they felt their ideas were listened to by their supervisors. Those who felt so were more accurate than those who did not. But the significance only reached 10% level. 93 Tables 31 and 32 bring a final summary of the total results, their level of significance and also the direction of accuracy: more and less accurate groups. Conclusions: 1. It was said previously that the small size of the sample (N=20) would make it difficult to draw very definite conclusions from the statistical analysis; this statement will guide the conclusion of the present study. With this thought in mind, it is possible to say: the agents with earlier farming expgrience and also the agents with more client.interpersonal contact are more accurate in estimating community adOption level. 2. The time of the agent working in the area also shows a significant negative correlation with accuracy: the agent with least time in the area was the most accurate. How- 13 of these findings can be questioned; ever, the reliability it seems that other factors such as age of the agent, his willingness to do a good job, pseudo knowledge of extension performance given by a long period of experience in service 13Reliability can be defined as degree of reproduction of similar findings by a measurement instrument. (36:166) Table 31. Summary of the 94 Kruskal-Wallis Analysis. Agent Accuracy Variables Characteristics More Less college degree - Education high school + country - BirthplaceC city + authoritarian - Leadership stylec permissive + individual discussion + Preference for comm.a methods group discussion - office oriented + Preferential workb field oriented - yes + Feedback in the organization no _ aSignificant at .02 level bSignificant at .05 level CSignificant at .10 level 95 Table 32. Spearman rank correlational analysis Correlation Variables ‘ with accuracy 1. Years working in farming .51a 2. Years working for ACAR .30 3. Months working in the community -.49b 4. Less use of mass media in communicating with farmers .35 5. Use of interpersonal communication with farmers .07 6. Total weighted use of communication methods, in which interpersonal communications received larger weights .26 7. Time spent in doing the job .08 8. Time Spent in field working -.14 9. Time Spent in office working .04 aSignificant at .01. bSignificant at .05. and so forth may have some influence in this relationship between agent's tenure and accuracy in estimating the com- munity adOption level. 3. The other findings despite giving some clues between agent characteristics and accuracy in estimating the client level of adOption were obscured by the small sample size and the low level of significance which made it impossible to draw further conclusions. 96 Needed Research From the results, it seems reasonable to recommend a repetition of the same study with a larger sample of the agents. The great concern would be to measure the adoption level of all farmers in the community. This would demand very extensive work. However, findings from Phase I and II of the Diffusion Research Project demonstrated that Opinion leaders14 in the community can estimate with high accuracy the farmers' adOption level. (44:20) The leaders' estimation could be checked against those from the agents. The validity of accuracy as an index of agent effective- ness should also be researched. Other methods to measure change agent effectiveness could be: "judgment of qualified peOple", and also other analysis of agent characteristics such as adjustment on the job, preferential work, client orientation, etc., can be used to identify the less and more effective change agents. (27) And furthermore, a research to determine the knowledge of the agents on evaluation approach and the amount of time they are spending in this subject should be carried out. The findings on ACAR agents' accuracy in estimating the community adoption level to some extent 14Opinion leaders were identified by the "reputational process" of leader identification. 97 | suggested this recommendation. The analysis Showed that agents need to do some evaluation on their clients' terminal behavior, which is adOption of ACAR promoted programs. Suggestions to Improve Research The great problem in this research was to translate the questions from English to Portuguese—-a major difficulty in any cross-cultural inquiry. The words, when literally trans- lated, seldom have the same meaning. Even in English the questions could be improved by just rewording or cutting some items in the scales. This was done by Whiting who made some changes in the interview guide used with the change agent. For example, the questions about agent communications style were reworded (verbal information instead of conversation, etc.). The agent preferential work item was removed and only a question concerning which part of the job the agent prefers was retained. The Appendix Shows part of this inter- view guide which was improved by Whiting. (44) The ranking order procedure seems to work very well; however, a larger sample is necessary in order to show the existent differences. 10. ll. 12. BIBLIOGRAPHY ACAR Publication. Programa 1963-64. Belo Horizonte, Minas Gerais, 1963. American Management Association. Effectivegommuni- cation on the Job. New York, 1515 Broadway Times Square, 1962. Allahabad Agricultural Institute. Extension Evaluation. Community Project in India, Allahabad, India, 1952. Benn, Harold W. Identification of Attitudes and Values Associated with the Teaching Effectiveness of New York County Extension Agents. (Unpublished Ph.D. Thesis, Cornell, 1952). Berlo, David K. The Process of Communication. New York: Holt Rinehart and Winston, Inc., 1963. Borg, Walter. Educational Research An Introduction. New York: David McKay Company, Inc., 1965. Byrn, Darcie gt 31. Evaluation in Extension. Division of Extension Research and Training, Federal Extension Service. U. S. Department of Agriculture, 1959. Campbell, William Giles. Form and Style in Thesis Writing. Boston: Houghton Mifflin Company, 1967. Cartano, David G. and Everett M. Rogers. The Role of the Change Agent in Diffusing New Ideas. Journal of the Pakistan Academy for Rural Development, Comilla, Pakistan, 1963. Curry, Donald Glenn. A Comparative Study of the Way in Which Selected County Agricultural Agents Perform Their Roles (Unpublished MS Thesis, Michigan State College, 1951). Downie, M. N. and Heath, R. W. Basic Statistical Methods. New York: Harper and Row Publishers, 1965. Eallading, Harold. The Contribution of Sociology to Effective Agricultural Extension. Sydney, Australia: University of Sydney, Department of Agricultural Economics, 1960. 98 13. 14. 15. l6. 17. 18. 19. 20. 21. 22. 23. 24. 99 Hays, William L. Statistics for Psychologists. New York: Holt, Rinehart and Winston, Inc., July, 1960. Herzog, William A. Jr. Diffusion of Innovations to Peasants in Brazil, India, and Nigeria. East Lansing, Michigan: Department of CommunicatiOn, Michigan State University, 1968. Herzog, William A., gt gt. Patterns of Diffusion in Rural Brazil, Correlations.of.Innovativeness and Opinion Leadership; in 20 Minas Gerais Communities. East Lansing, Michigan: Department OfCCommunication, Michigan State University, 1968. ' Hoffer, Charles R. Selected Social Factors Affecting Participation of Farmers in Agricultural Extension Work. East Lansing, Michigan: Agricultural Experiment Station, Special Bulletin, 1944. Hook, Lucyle and Mary Virginia. The Research Paper. New Jersey: Prentice-Hall, Inc., 1962. Howard, Thelma. An Exploratory Study of a Comparison of Expressed Acceptance of Self and Others Between 4-H Members Involved in Two Types of Project Evaluation. (Unpublished MS Thesis, Michigan State University, 1966). Gomes, Rita gt gt. Estudo Analitico de Certas Caracter- isticas of Pessoal Tecnico da ACAR. Belo Horizonte, Minas Gerais: ACAR's Publications, 1964. Kerlinger, Fred N. Foundations of Behavioral Research. New York: Holt, Rinehart and Winston, Inc., 1964. Lutz, Arlen E. Change Agent as Predictors of Rate of Farm Practices Adoption. (Unpublished Ph.D. Thesis, University of Nebraska, 1966.) Maunder, A. H. Improvement of Agricultural Extension Service in Eurgpean Countries. Rome, Italy: Fao Development Paper 41, 1954. Miller, Mason E. Teaching-Learning Process and Methods for Extension Workers. Institute for Extension Personnel Development, Cooperative Extension Service, Michigan State University, 1965. (Mimeographed) Mosher, Arthur T. Techpical Co-gperation in Latin American Agriculture. Chicago, Illinois: University of Chicago Press, 1957. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 100 Nye, Ivan. The Relationship of Certain Factors to County Agent Success. Research Bulletin 498, University of Missouri, College of Agriculture, 1952. Pierson, Rowland Ray. Vocational Interest of Agricul- tural Extension Workers as Related to Selected Aspects of Work Adjustment. Report for Research Project, Michigan State College, 1951. Posz, A. Conrad. The Academic Background of Agricultural Extension Workers as Related to Selected Aspect of Work Adjustment. (Unpublished Ph. D. Thesis, Michigan State College, 1952). Preiss, Jack J. The Functions of Relevant Power of County Agent Performance. (Unpublished Ph. D. Thesis, Michigan State University, 1954). Progress Report, Summary of the Research Study, The Differential Characteristics_gt More versus Less-_— Sucgessffil’lnformai Extension Teachers. Michigan State College, February, 1952. Rahudkar, Wasudeo B. The Relationship of Certain Factors to the Success of Village Level Workers, Rural Sociology, Vol. 27, No. 4, 1962. Robinson, A. V. An Evaluation of the Cooroy Extension Grou . Brisbane, Australia: Queensland, Department of Primary Industry, 1963. Rogers, Everett M. Diffusion of Innovations. New York: The Free Press of Glencoe, 1962. .Bibliography on the Diffusion of Innovations. Report No. 6, Department of Communication, Michigan State University, July, 1967. . . Elementos Del Cambio Social En America Latina. Bogota, Colombia: Ediciones Tercer Mundo, 1966. , with Floyd Shoemaker. Communication of Innovations a Cross Cultural Approachf New York: Free Press Of GIencoe (in press), 1968. Selltiz, Claire gt a1. Research Methods in Social Relations.. New YorK?’ Halt, Rinehart and WinSton, Inc., 1961. 37. 38. 39. 40. 41. 42. 43. 44. 45. 101 Siegel, Sidney. Nonparametric Statistics for the tghaviora1-Sciences.. New York: McGraw-Hill Book 1 Company, Inc., 1956. Straus, Murray A. Cultural Factors in the Function- ing of Agricultural Extension in Ceylon. Rural Sociology No. 18, Wisconcin, 1953. Swan, John C. A Study of Values as a Differential Characteristic of More and Less Effective County Agent. (Unpublished MS thesis, Michigan State University, 1960). Vivian, Charles H. and-M. Bernetta Jackson. En lish Composition. New York: Barnes and Nobles, Inc., 1960. Webster's Seventh New Collegiate Dictionary. Spring- Field, Massachusetts: McGraw—Hill Book Company, Inc., 1966. Willson, E. Bright, Jr. An Introdtction to Scientific Research. New York: McGraw-Hill Book Company, Inc., Whiting, Gordon C. and John Winterton. Methodological Background Of the Phase I_MSU/AID Brazil Diffusion Research Project. East Lansing, Michigan: Department of Communication, Michigan State University, 1968. Whiting, Gordon C. gt gt. Innovations in Brazil. East Lansing, Michigan: Department of Communication, Michigan State University, 1968. Wilkening, Eugene A. Roles of Communicating Agents in Technological Change in Agriculture, Social Forces 34, 1950. 192 APPENDIX ACAR Local Agent Schedule DIFFUSION OF AGRICULTURAL PRACTICES: PHASE I Interview with Change Agents Summer, 1966 County (Municipio, administrative unit) Final Edition 1/24/66 ' .Interviewee's name: 5,7 8,9 Name of his organization ' Interviewer's name: Supervisor's name: Date of interview: ‘ ‘ Begun at o'clock, finished at .o—— Total time ' hours. 10 ,11 __ Number of y o 103 INTERVIEW WITH CHANGE AGENTS AT "hUNICIPIO" (COUNTY) LEVEL ~‘- Have you been brought Up in a city or in the country? 0‘12. 0. Town, City 2. Country 13 How many years have you worked on a farm? ' 14,15 0. Never Number of years Were you born in Minas Gerais or in another state of Brazil? .0. Minas _ .2. Other 16 For how long have you been working for ACAR? e“r 3 17,18 a In how many other organizations which aim at bringing about change have you worked, and for how long? . Number of organizations _ ' . . 19 Total period of work (in—acnths) 20,21,22 For how many years have you been working in.this area (or office)? . months _ 23,24,25 What do yOu like best about your work, and what do you like the least? . 3. Likes best ° b. Likes least 26 ' 27 Do-you thin that your "regional" pays considerable attention to your ideas, or not? 0. Never listens to me 1. He listens to me 3. ‘ Other . 28 If a farmer from outside of your working area needed your professional counseling urgently, would you make use of ACAR's vehicle without your "regional's" authorization? 2. Yes ’ 1 '. No . Other ' ' - 29 16.. 10a. 10b. ‘11. 12. 13. 'In the better community: Number 104 Of these four situations, which one best describes your_ work in ( . )? A (BETTER COMMUNITY) 1. PeOple ask for help concerning exactly those problems they are prepared to face, or . 2. People recog :ize their problems and sec pt your help when offeied, or 3. You have. to ma