SYSTEM VARIABLES AND AGRICULTURAL
INNOVATIVENESS III EASTERN NIGERIA

Thesis for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY
BURL EDWARD DAVIS

- 1968

 

   

  
     

LIBRARY
Michigan State
University

IIIIIIIIIIIIIIIQIIIIIIIIIIIIIISIIIQIIIIIIIIIIIIIII

This is to certifg that the
thesis entitled
SYST I3". VI‘IRIA BL ES \ND .‘IGR I CULT URN
INEIICVAT IV EN EL) 5 III EA ST 13115 N 13 ERIA

presented bg

BURL SJCJI‘IRJ L)?‘."JI';}

has been accepted towards fulfillment
of the requirements for

1 I1 . 3 degree in Communication

 

0-169

 

 

 

   

 

ABSTRACT

The present study was designed to explore the relations between
individual modernizing characteristics of Nigerian farmers, the cor—
responding system characteristics, (the mean level of the individual
characteristics in each of the villages), and the individuals' innovao
tiveness. Of primary concern was the interaction between the two
.levels of modernizing characteristics in explaining greater amounts of
variance in innovativeness.

A portion of the data gathered in the three-nation Diffusion
Project conducted by the Department of Communication, Michigan State
University, and Sponsored by the United States Agency for International
Development, was analyzed.

Eighteen villages were chosen as the basis for the study. The
sample was comprised of 1,1u2 respondents, who were in each case a
male, the head of his household, at least 20 years old, and farming
some amount of land.

Fourteen agricultural innovations were selected for the study
including: fertilizer, NS-l maize, Aldrin dust, poultry, oil palm
rehabilitation, community plantations, citrus, rice, improved cassava,
vegetable seeds, cocoa planting schemes, rubber planting scheme, cashew,

and PAID agricultural credit program.

Burl Edward Davis

Seventeen characteristics of individuals were utilized as inde-
pendent variables, including cosmopoliteness, education, familiem,
educational aspiration, empathy, knowledge of agricultural workers,
economic aspiration, literacy, newspaper exposure, radio exposure,
film exposure, agricultural media exposure, credit orientation, level
of living, correct agricultural knowledge, formal social participation,
and achievement motivation. Each of these variables was aggregated
through a measure of central tendency (mean) and utililzed as an inde-
pendent variable at the system level.

Three guiding hypotheses were selected as the focus of the
analysis: (1) that the individual level of innovativeness for respon-
dents is positively related to the individual variables, (2) that the
individual level of innovativeness is positively related to the system
variables, and (3) that the individual level of innovativeness is positively
related to the joint influence of the individual and system variables.

Three statistical methods were utilized in the analysis of the
data: (1) zero-order correlation, (2) first-order and higher-order
partial correlations, and (3) multiple correlation.

First, zero-order correlations were computed between innova-
tiveness and each independent variable. Then, higher-order partial
correlations between innovativeness and each of the independent variables
were computed, holding constant all other independent variables. To
determine the contribution of the system variables, partial correlations
were computed for each of the system variables, with the individual level

variables held constant.

Burl Edward Davis

After submitting the independent variables to both zero-order
and higher-order partial correlation analysis, the best predictors
of innovativeness (variables most highly correlated with innovativeness)
from the system variables and the individual variables were included
in the multiple correlation analysis, using the multiple regression
least squares delete computer program.

When zero-order correlations were computed between the inde-
pendent variables and innovativeness, 15 of the 17 individual level
variables were significantly correlated with innovativeness: cosmo-
politeness, education, familism, educational aspiration, literacy,
newspaper exposure, level of living, radio exposure, film exposure,
agricultural media exposure, correct agricultural knowledge, and formal
social participation. System level variables significantly correlated
with innovativeness were education, familism, educational aspiration,
achievement motivation, empathy, knowledge of agricultural workers,
economic aspiration, literacy, newspaper exposure, radio exposure,
film exposure, agricultural media exposure, level of living, correct
agricultural knowledge and formal social participation. In general,
individual level variables were more highly correlated with innova-
tiveness than system level variables.

First-order partial correlations were computed separately
for individual level and system level variables with innovativeness.
First, individual variables were correlated with innovativeness with
system variables held constant. All but three of the individual level
variables were significantly correlated with innovativeness. Next,

system level variables were correlated with innovativeness with

 

 

Burl Edward Davis

individual level variables held constant. "Only 7 of_the 17 system
variables reached statistical significance in this analysis.

Highest-order partial correlations were then computed, to
select variables from both individual and system levels most highly
correlated with innovativeness, to be included in a multiple cor-
relation analysis. Of the 34 variables submitted to the least
squares delete computer program, ll were deleted when they did not
reach the minimum statistical significance criterion. Thus, a total
of 23 variables were retained in the analysis as statistically signi-
ficant after the highest-order partial correlation analysis, and
these were combined in a multiple correlation analysis. The multiple
correlation coefficient for these 23 variables was .68, with a
coefficient of determination of .H7.

When the individual level variables only were submitted to
highest order partial correlation analysis, only six were retained
as statistically significant. When entered into a multiple correla-
tion analysis, these six individual level variables produced a
multiple correlation coefficient of .62, and a coefficient of deter—
mination of .38. The difference between the multiple correlation
coefficient for individual variables only, and the multiple cor-
relation coefficient including both individual and system variables

was statistically significant ( p<:.05 ).

Accepted by the faculty of the Department of Communication,
College of Communication Arts, Michigan State University, in partial

fulfillment of the requirements for the Doctor of Philosophy degree.

 

Guidance Committee: ,Zflfldm, K /%(/c//é€
27 at. 3 Item--
xii/Me—

SYSTEM VARIABLES AND AGRICULTURAL INNOVATIVENESS
IN

EASTERN NIGERIA

by

Burl Edward Davis

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Communication

1968

Kym?
(fizz/<07

PREFACE

The research on which this dissertation was based was
conducted in Eastern Nigeria during 1966-67. The Regional
Government of Eastern Nigeria seceeded from the Federation of
Nigeria in May, 196T and became the Republic of Biafra. As
this dissertation goes to press, civil strife is still raging
in the area and neither the present nor future status of the
territory is not clear. Thus, the name "Eastern Nigeria" is
used throughout this report. A tragic aftermath of the civil
strife in this nation is that a large proportion of the res—
pondents included in the present study are now no longer living,

in all probability.

ii

ACKNOWLEDGEMENTS

The author wishes to express his sincere appreciation to the
chairman of his guidance committee, Dr. Everett M. Rogers, for the
assistance and counsel tendered, not only during the writing of
the author's dissertation, but during his entire graduate program.
Gratitude is also extended to the other membbrs of the author‘s
committee, Dr. Hideya Kumata, Dr. Mason Miller, and Dr. Bradley S.
Greenberg.

The author is also indebted to the United States Agency for
International Development and the Department of Communication,
Michigan State University, for permission to make use of data
collected in Eastern Nigeria for the project, "Diffusion of Innova-
tions in Rural Societies."

Thanks are also expressed to the numerous co-workers and
fellow graduate students in the Department of Communication for the
insights and theoretical assistance gained through many hours of
discussion and consultation. Special thanks are due to Anant Saxena,
who was of great assistance during the analysis of data for this
report, and Niels Roling, Graham Kerr, and Joe Ascroft, who furnished
invaluable background knowledge and helpful suggestions based on
intimate knowledge of the conduct of the project in Nigeria.

The author expresses his sincere appreciation to Miss Judy
MacGregor, for her many hours of work spent in typing and collating

the manuscript.

iii

 

TABLE OF CONTENTS

Chapter

I. INTRODUCTION AND DESCRIPTION OF THE PROBLEM

Diffusion Research . . .

Growth of Cross-Cultural Research
Focus of the Present Study

The Role of Communication

Value of the Present Study

Setting for the Study .

Objectives of the Present Study
General Outline of Following Chapters

II. THEORETIC RATIONALE e . . . .

Review of Literature . .
System Analysis . . . .
Theoretic Rationale . .

How to Determine System Effects

Some Possible Fallacies
Dependent Variable . . .

Description of Innovations

Independent Variables .

III. RESEARCH METHODOLOGY . . . .
The Diffusion Project .
Conduct of the Study . .

Phase One . . . . .
Phase Two . . . . .

Characteristics of the Sample
Knowledge of Agricultural Innovations
Adoption of Agricultural Innovations

Analysis of the Data . .

Operationalization of the Variables

O

O

Dependent Variable . . . .

Independent Variables

Construction of Variables

IV. RESEARCH FINDINGS . . . . . .

Correlates of Innovativeness .

Individual Characteristics

System Characteristics . .
Comparison of Zero—Order and

Variables . . .

0

O

0

Partial

Confirmation of Hypotheses . . . . .

Analysis for System Effects

iv

0

0

Correlations
Joint Influence of Individual and System

O

O

Page

F'F‘
ooooooucwI-o

|-‘
M

12
18
22
27
30
33
3W
37

II?

47
49
49
51
53
55
59
62
63
63
65
69

71

71
71
76
79

84
89
94

Chapter

Implications for Change Agencies
Implications for Future Research

.V. SUMMARY AND CONCLUSIONS

I Summary I: e e b e

Conclusions .. ..
Limitations of Present Study

BIBLIOGRAPHY . . . . . . . .

e

Page

98

98
103
106
109
110

116

LIST OF TABLES
Table Page

1. List of Innovations Studied in Phases I and II
in Eastern Nigeria . . . . . . . . . . . . . . . . . 38

2. Zero-Order Intercorrelations of Newspaper Exposure,
Radio Exposure, and Film Exposure . . . . . . . . . 43

3. List of Independent Variables and Their Hypothesized
Relationship to Innovativeness . . . . . . . . . . . 46
u. Village Populations, Sample Sizes, and Sampling Rates
forNigeria,PhaseII..........e....51+
5. Personal Characteristics of the Respondents . . . . . . . 56
6. Number of Innovations Respondents Knew Correctly . . . . 57
7. Respondents' Correct Information About 1n Agricultural

Innovations O O O O ‘ O O O O O O O O O O O O I O O 0 58

8. Percentage of Respondents Adopting Agricultural
Innovations . . . . . . . . . . . . . . . . . . . . 60

9. Percentage of Respondents Adopting Each of the la
Agricultural Innovations . . . . . . . . . . . . . . 61

10. Zero-Order Correlations of Individual Level Variables
and System Level Variables with Innovativeness . . . 73

ll. First—Order Partial Correlations of Individual Level
Variables, System Level Variables with Innova-
tiveness . . . . . . . . . . . . . . . . . . . . . . 75

I2. Sixteenth-Order Partial Correlations of Six Individual
Level Variables with Innovativeness, After
Deletion of ll Variables Not Reaching Minimum
Significance . . . . . . . . . . . . . . . . . . . . 77

13. Thirty-Third Order Partial Correlations of 3M Inde-
pendent Variables with Innovativeness . . . . . . . 80

la. Comparison of Zero-Order and First-Order Partial Cor-

relations of Both Individual and System Variables
with Innovativeness . . . . . . . . . . . . . . . . 82

vi

Table Page

15. First-Order Multiple Correlations of Each Variable
with Innovativeness, Including Botn Individual
andSYStemLCVQISeeeeeeeeeeeeeeeee86

16. Twenty-Second Order Partial Correlations of 23 Inde-
pendent Variables with Innovativeness, After
Deletion of 11 Variables Not Reaching Minimum
Significance . . . . . . . . . . . . . . . . . . . 90

17. Multiple Correlation Coefficients and Coefficients
of Determination of Individual Level Variables,
System Level Variables, and Individual and System
Level Variables Considered Simultaneously . . . . o 91

18. Comparison of Zero-Order Correlations of 3% Individual
and System Level Variables with Three Indices of
Innovativeness ( N = 1,1u2 ) . . . . . . . . . . . 126

19. Intercorrelations Between lu-Item, 6-Item, and 3-Item
Indices of Innovativeness ( N = 1,1u2 ) . . . . . . 128

20. Comparison of Zero-Order Product-Moment Correlation
(r) and Coefficient of Nonlinear Relationship
(eta) for an Individual and System Level Inde-
pendent Variables and Innovativeness (N = 1,1u2) . 131

21. Mean Values and Standard Deviations for 17 Individual
Level, 17 System Level Variables (N = 1,142) . . . 133

vii

 

LIST OF FIGURES

Figure Page

1. Comparison of Conventional Direct Effects Model,
and Hypothesized Direct-Plus—Mediational
Effects Model of Innovativeness . . . . . . . . . . . 26

viii

 

APPENDIX A:
APPENDIX B:
APPENDIX C:

APPENDIX D:

LIST OF APPENDICES

Correction for Inflated Multiple Correlation . . .

Alternative Indices of Innovativeness. . . . . . .

AnaIYSis Of cuPVilinearity e e e e e e e e e e e e

Mean Values and Standard Deviations of Independent

Variables .

ix

0

O

O

Page
123

125

129

133

 

CHAPTER I
INTRODUCTION AND DESCRIPTION OF PROBLEM

The introduction of technological innovations in traditional agricul-
ture is one of the basic elements in rural development programs in less
developed countries. The study described in the present thesis represents
an attempt to understand one such development program in Eastern Nigeria.
Consideration is given to social, psychological, cultural, and overt behav-
ioral factors in attempting to describe and explain the processes of diffusion
and adoption of fourteen agricultural innovations by individuals in eighteen
Nigerian villages. Primary attention is given to the influence of system

variables in increasing the amount of variance in innovativeness explained.

Diffusion Research

For a number of years, research workers in the field of communication
and other social sciences have been interested in the spread and acceptance
of new ideas. The diffusion of innovations in social systems and among
individuals involves the communication of information about the practices
from their sources of origin, and their acceptance by the end-users or adop-
ters. Generally, the analysis of the diffusion of innovations involves
four crucial elements: (1) the innovation, (2) its communication from one
individual to another, (3) in a social system, (A) over time (Rogers, 1962,
p. 12).

Since its initiation in the 1920's, diffusion research has produced

an imposing body of research findings, both in the United States and in

1

 

other nations. Typically, the findings have pointed out the individual,
group, and situational elements which influence the innovation-decision
process. The elements involved in the investigation of diffusion pro-
cesses are manifold; however, a few have received major emphases, for
various reasons.

For instance, diffusion researchers generally agree that adop-
ters can be categorized according to their individual tendencies to be
relatively early or late to adopt a new practice. Operationally,
receivers are divided into categories according to their relative innova-
tiveness. For any given innovation, the period of time since each
ad0pted the innovation is recorded and standardized across the unit of
analysis (e.g., the village, region, or state). Assuming a normal dis-
tribution, and utilizing standard deviations, the earliest 2.5 percent
are classified as innovators; the next 13.5 percent, early adopters; the
next 3M percent, early majority; another 3n percent, late majority; and
the final 16 percent as laggards (Rogers, 1962, pp. 160-171).

Another major area of interest in the diffusion research tradition
centers around the explication of the innovation decision-making process.
The North Central Regional Rural Sociology Subcommittee for the Study of
Diffusion of Farm Practices (1961) divided the adoption process into five
stages: awareness, interest, evaluation, trial, and adoption. Dissatis-
faction with a five-stage conceptualization led others to propose other
classifications. Rogers with Shoemaker (in press) speak of four "functions"
(rather than "stages"): knowledge, persuasion, decision, and confirmation.

Other major generalizations from the diffusion research tradition

deal with the S-shaped curve of adoption, the role of the various

communication channels at the different stages of the adoption process,
and the characteristics of the innovation itself that enhance (or retard)

its rate of adoption.

Growth of Cross-Cultural Research

One striking development in the field of diffusion research is the
marked increase in the number of diffusion studies being conducted in
countries outside the United States, especially since about 1960. The
value of this trend is illustrated in the testing of the cross-cultural
applicability of communication principles and generalizations that emerged
originally from intra-national research. To achieve maximum utility,
researchers need to find hypotheses about the diffusion of innovations
that are generally true, regardless of the geographic and cultural locality
of the study.

This trend toward internationalization of the field of inquiry has
been prompted both by the overseas migration of North American researchers,
and by a growing number of non-0.8. scientists conducting diffusion re-
search. A recent analysis of the publications in the Diffusion Documents
Center at Michigan State University revealed that less than 10 percent
of the studies listed had been conducted outside the United States in
1960, but by 1965 this percentage had risen to 26.8 percent, or a net
increase from 38 to 190 studies conducted in the developing countries
over a five-year period (Rogers and Stanfield, 1968, p. 10). A cursory
examination of a recent issue of the Bibliography on the Diffusion of
Innovations (Rogers, 1967) indicates that well over one-half of the
studies listed were authored or co-authored by nationals from developing

countries.

 

The study described in the present dissertation represents a
further contribution to the growing number of investigations of the

diffusion of innovations in less developed countries.

Focus of-the Present Study
When one considers the various units that may form the bases of
analysis in the diffusion of innovations, several possibilities appear.

The following simplified typological classification illustrates this

 

 

point:
Independent Variable
Individual System
Individual A I .B
Dependent +__
Variable-, System C J D

 

 

 

In many of the diffusion studies conducted to date, much emphasis

has been placed upon the individual characteristics of the receiver, and

 

their effect upon his individual decision to adopt or not adopt the

 

innovation (category A in the paradigm). For instance, the relatively
earlier adopters in a social system in the United States have generally
been found to be younger in age, have higher social status, a more
favorable financial position, more specialized operations, and a different
type of mental ability than later adopters (Rogers, 1962, pp. 171-178).
Socio—economic,demographic, and socio-psychological characteristics of
individuals were utilized widely as independent variables in explaining

innovativeness in various individuals.

However, individuals engaged in the innovation decisionrmaking
process clearly do not exist in an isolated state. They are surrounded‘fl
by a myriad of influences, ranging from prevailing social system norms _$
and the level of education, wealth, and age of others in their system,
to contact with cosmopolite friends, distance from urban centers of
population, and the availability of communications media.’ Thus, the
present study focuses on the additional amount of variance in innova-
tiveness explained by taking into account the "system variables" in the
social system under consideration, as well as the "individual variables"
(categories A and B in the paradigm). System variables are defined as
those structural, compositional, or social-psychological characteristics
whose distribution in the system influences the behavior of individuals
in that system.1

The central problem considered in the present study is represented
by the following questions:

(1) Do the properties of a social system influence the behavior
of its members? If so, what is the nature and the direction of such
influence?

(2) How much additional variance in the behavior of the indivi-
duals in a social system can be explained by the simultaneous considera-

tion of both individual and system variables?

 

1A more detailed definition is given in Chapter II.

 

The Role of Communication

Members of a number of research traditions have been interested
in answers to these two questions, especially sociologists. However,
the present study is designed to focus on the communication aspects of
the problem, from three viewpoints:

(l) the introduction of innovations to villagers in a less
developed country is essentially a communication process. When viewed
from the standpoint of the source or instigator of the change, the
introduction of innovations may be classified as either immanent change,
which occurs when members of a social system with little or no external
influence, create and develop a new idea, or contact change, which occurs
when the source of the new ideas is outside the social system (Rogers
with Shoemaker, forthcoming, p. 17). Contact change is a change phenom-
enon occurring between systems. The present study is an example of
centact change, inasmuch as the agricultural innovations included in the
study were introduced into each local social system by an outside source
(agents from the Eastern Nigerian Ministry of Agriculture), through
their communication with the villagers in the social systems.

Obviously, other types of consideration are inevitably involved
in the process of diffusion of innovations (such as economic, political,
cultural, etc.), but communication from the change agency to receivers
in the receiving social system is basic to the process of contact change.

(2) Once introduced into a social system, the innovation is
diffused within the social system by a process of communication. Both

interpersonal and mass media channels are utilized in the within-system

diffusion process. If communication is viewed as the process by which
messages are transferred from a source to a receiver, it may be illus-
trated by the SMCR model of communication (Berlo, 1960, p. 72), in
which a source (8) sends a message (M) via certain’channels (C) to the
receivers (R). Communication factors are vitally involved in the various
aspects of the diffusion process within the social system, and are an
important element pervading the social change process. A number of social,
psychological, and cultural attributes of individuals are relevant as
antecedent correlates of diffusion and a number of social, psychological,
and cultural attributes of individuals will be changed as a result of
the diffusion process, but the central area of concern integral to the
entire realm of activities is communication, from one member of the
receiving social system to another member of the receiving social system.
The diffusion process is conceptualized, therefore, as a multi-faceted
communication process involving the spread of new ideas from a source to
an audience of receivers via a series of sequential transmissions, as well
as by direct contact with the change agency itself. Thus, communication
is vital to within-system diffusion of innovations.

(3) A number of specific communication variables are considered
as independent variables in our analysis of innovativeness within the
social system. These include exposure to agricultural communications
media, exposure to newspapers, exposure to radio, and exposure to films
(see Table 3). In addition, a number of modernization characteristics
such as empathy, education, achievement motivation, cosmOpoliteness,
etc., which are expected to exert influence upon the receivers' perception

and reception of the modernization messages, are included.

 

Value of the Present Study

The investigation of innovativeness described in this dissertation
may be justified by the following considerations:

(1) It is a further contribution to the evaluation and testing of
theories of diffusion through studies conducted in crossocultural
situations, specifically in less developed countries where studies are
few. Many of the generalizations derived from diffusion research have
been based on investigations conducted in the United States. The locale
of the present study is one of the less developed countries of the world
(Nigeria), so that application of existing principles of diffusion in
this situation represents an extreme test of their validity and reliability.

(2) The present study represents an extension of the analysis of
innovativeness to include elements of the village system which affect the
rate of adoption, as well as personal characteristics of each individual
in the village. The analysis of both levels of variables is widely
discussed and utilized in other social sciences, but very few studies
use this type of analysis in diffusion research.

(3) The inclusion of both system variables and individual variables
in the analysis helps prevent the commission of the "ecological fallacy"
(Robinson, 1950), the "psychologistic fallacy" and the "sociologistic
fallacy" (Riley, 1963, pp. 703-709).1

(A) The present study represents a merging or convergence of the
individual level analysis typically employed by psychologists, social
psychologists, sociologists, etc., and the aggregative, molar type of

analysis typically employed by economists, ecologists, geographers, etc.

 

1Definitions of these fallacies are discussed in Chapter III.

 

The Setting for the Study
The present study is based on an analysis of data representing
one portion of a much larger, more comprehensive Diffusion Project
conducted by the Department of Communication, Michigan State University.
This study was sponsored by the United States Agency for International
Development, and was conducted concurrently in three nations in three

widely separated and differing areas of the world, Brazil, India and

 

Nigeria. The present analysis derives from the research conducted in
Nigeria, where the project was affiliated with the Economic Develop-
ment Institute, University of Nigeria, Enugu, and the Nigerian
Ministry of Agriculture.

The Nigerian diffusion project was designed to investigate the
diffusion of health and agricultural information and the acceptance of
health and agricultural innovations. Data were gathered on numerous
social, psychological, cultural, economic and geographical factors to
describe and explain the diffusion and adoption processes. More com-
plete information regarding the collection of data, sampling procedures,
etc., are given in Chapter IV.

The Nigerian study was conducted in three phases. In Phase One,
the unit of analysis was the village; the purpose was to analyze villages
in terms of their success in programs of planned change. In Phase Two,
the unit of analysis was the individual head of a farm household, and
the purpose was to determine how the individual's social characteristics,
family structure, farm business, communication behavior and various
social psychological characteristics were related to his innovativeness,

attitude toward change, opinion leadership, etc.

 

10

The present study utilized data from both Phase One and Phase
Two.1 Phase One data were used to derive measures of the social system's
characteristics, and form the basis for the determination of which
villages to include or exclude in Phase Two; Phase Two data were used

to derive correlates of innovativeness for the individuals.

Objectives of the Present Study

Our objectives for the present study were threefold:

 

(1) To determine whether various attitudinal, cognitive and overt
behavioral characteristics found related to innovativeness in other
studies were relevant in diffusion research in a less developed nation.

(2) To determine whether the distribution of modernizing
characteristics within a social system influence individuals toward
greater innovativeness, in addition to the individual's own personal
characteristics.

(3) To derive generalizations and form conclusions which may be
helpful in the formulation of a strategy for change by various change

agencies.

General Outline of Following Chapters
A brief summary and introduction to the discussion outline to be
followed in the following chapters may be helpful at this point.
In Chapter II, a theoretic foundation for the analysis of system
effects is presented, including a review of relevant literature, a
definitional discussion of system analysis, a description of the meth-

odology utilized to determine system effects, some theoretic and

 

1Phase Three of the Diffusion Project is described in Chapter III.

 

 

"' .

11

methodological fallacies that must be avoided, and a listing of both
independent and dependent variables used in the present study.
'Chapter III discusses the research methodology utilized in
the investigation, including a description of how villagers, respon-
dents, and interviewers were selected, how data were collected, and
how the research instrument was formulated and revised. .A brief

description of methods of correlational analysis utilized is included,

 

along with a description of how both dependent and independent
variables are operationalized.

In Chapter IV, the findings of the investigation are presented,
in both verbal and tabular form, and in Chapter V, the findings are
summerized, conclusions are stated, and implications for future research

are given.

CHAPTER II
THEORETIC RATIONALE

Review of Literature
Although scarcely mentioned by diffusion researchers, the concept
of "system effects" is not new. Durkheim (IDS?) showed not only that
suicide rates vary considerably among different religions, but also that
suicide rates for members of a given religion are much lower when these

individuals are a minority in the society. He maintained that research

 

methods dealing with social groups rest wholly on the basic principle that
social facts must be studied as "things"; that is, as realities external
to the individual. Although Selvin (1958) subsequently accused Durkheim
of ignoring the individual, treating the group as an entity, thus com-
mitting the error of reification,l Durkheim apparently was one of the
first to recognize the influence of system effects in addition to indi-
vidual characteristics in personal behavior.

Groves and Ogburn (1928) investigated the marriage rates for men
and women, and found that they varied inversely with the sex ratios of
the communities in which the subjects lived. Paris and Dunham (1939)
obtained similar results in studies of rates of psychosis. Stouffer
(19u9) investigated attitudes toward promotion among soldiers in a
military police unit and an air force unit, and found that the indivi-

dual's attitude toward promotion was influenced markedly by the general

 

lReification is defined as the conversion of an abstraction or
mental construct into a supposed real thing (Webster's New Collegiate
Dictionar , 1958. Springfield, Massachusetts: G. 8 C. Merriam Co.)

12

 

13

rate of promotion in the unit of which he was a member. Generally, pro-
motion was evaluated more highly in units where group promotion rates were
low, than in units where group promotion rates were high.

Blau (1960) reported that data from an analysis of a public assis-
tance agency showed that the prevailing values in a work group had
structural (system) effects. In some cases, the group values and the

individual‘s orientation had similar, but independent, effects on the con-

 

duct of the individual; in other cases, they had opposite effects; and
in still others, the effects of the individual‘s orientation were contingent
on the prevalence of this orientation in the group, a pattern which
identifies characteristics associated with deviancy. Blau further
commented that most social surveys tend to make individuals the focus of
the analysis, and ecological studies typically examine social units with-
out separating social conditions from the individual‘s own characteristics.
Blau thus called for the simultaneous use of indices of social structure fit
(system effects) and of individual behavior, such as the present study
contemplates.

Other sociological studies dealing with system variables include
Berelson 3: 2i! (1959),who showed the effects of community composition
in terms of party affiliation on voting behavior; Lipset 23 ii“ (1956),
who found system effects in their study of a labor union; and Davis g£_§£3
(1961), who encountered system effects in their study of the Great Books
reading groups.

A number of research investigations examined the influence of

system effects in formal organizations. Becker and Stafford (1967) conducted

 

14

a mail survey of luo savings and loan associations in Illinois, to

discover the effect upon organizational efficiency of five independent
variables, including the size of the organization (measured in assets),

the growth rate of the surrounding community, the adoption of innova- ,K;
tions, the size of the administrative component, and the management's
leadership style. These five variables explained 40 percent of the
Variance in the organizational efficiency of the institutions.

Sapolsky (1967) studied nine retail organizations in six depart-

 

ment stores, using personal interviews, and found that three major
innovations suggested by store executives were not implemented because
of the nature of the stores' organization and reward systems. Similarly,
in a study of factors associated with the success or failure of various
staff proposals (innovations), Evans and Black (1967) found that the
structure of the organizations studied, the attributes of the staff-line
relationships that affected the innovation process, and the attributes
of the innovative proposal were positively associated with the adoption
of innovations.

Shepard (1967) classified various organizations as either innova-
tion-resisting or innovation-producing, based on an analysis of numerous
system variables which influenced the attitude of the members toward
innovations.

Within the rural sociology diffusion research tradition, a number
of studies gave further evidence of the influence of system variables on
the adoption of innovations. Marsh and Coleman (1954), for instance,
studied 393 farm operators in one Kentucky County, and found that socio-

economic characteristics and the neighborhood of residence were both

 

15

positively related to the adoption of recommended farm practices. Even
when socio-economic characteristics were held constant, it was found
that the extent to which farm operators adopted farm innovations was, in
part, a function of the neighborhood of residence.

Flinn (1963) investigated the influence of community norms in ‘fl?
predicting agricultural innovativeness. Defining norms as a pattern for

behavior in a social system with both objective and subjective qualities

which can be inferred from overt actions and verbal responses, he found

 

that community norms relative to innovativeness, as inferred from overt
action, account for more variation in farmer innovativeness than any
other variable studied. Five variables taken together explained 69.1
percent of the variance in innovativeness, with community norms alone
accounting for 20 percent of the variance explained.

In a study of 393 farmers in 12 neighborhoods in Kentucky, Young
and Coleman (1959) found that farmers in high adoption neighborhoods
ascribed scientific farming attitudes to their neighbors, and said they
were frequently guided by the influence and opinions of their neighbors.

Van den Ban (1960) studied #7 Wisconsin townships to test the
hypothesis that the social organization and culture of locality groups
influence adoption more than individual socio—economic factors (such as
education, farm size, net worth, etc.). After dividing the townships
into four groups according to their innovativeness scores, van den Ban
found significant differences in innovativeness among the groups, be-
cause of differences in social structure.

Coughenour'(l96u) analyzed data on the diffusion of five farm

innovations in 12 Kentucky localities for speed of diffusion and factors

 

16

relating to it, and found positive relationships with socio-economic
and attitudinal resources of each locality, along with the nature of
social relationships with information sources and media contacts.

Duncan and Kreitlow (195“) matched and compared 19 pairs of
rural neighborhoods on the adoption of 30 school innovations, using
an index of 25 farming practices and four elements of organizational
participation. The neighborhood was the unit of analysis, and the mean
score of 10 respondents in each neighborhood was set as the acceptance
value for the entire neighborhood. Heterogeneous neighborhoods were
found consistently more favorable toward a majority of the innovations,
indicating the influence of both social structure and norms.

By including a system variable called "norms on innovativeness,"
Rogers (1961) was able to improve predictions regarding the innovative-
néss of truck farmers, when earlier studies had been characterized by
low prediction levels.

Qadir's (1966) analysis of "compositional"1 effects and their
influence on the adoption of technological innovations by some 600
villagers in 26 Philippine neighborhoods indicated that system variables
(e.g., mean media exposure, mean education level, etc.) were effective
as predictors of individual innovativeness. He concluded, "The in-
vestigation in the rural area of the Philippines confirms that the barrio
as a locality group has an effect on the individual‘s behavior toward

adoption of technological change."

 

QadirHSusage of the term "compositional effects," is practically
synonymous with the usage of the term, "system effects" in this dis-
sertation.

 

 

17

Regarding this review of literature, several generalizations may
be drawn:

1. Although a number of studies mentioned system effects as part
of the overall research design, almost none focused on system effects as
the central problem of the research. In fact, many of the studies
considered system variables only after other variables were unsuccessful
in predicting innovativeness; thus, system effects were considered only
secondarily and in a residual capacity.

2. Many studies investigated the effect of the personal character-
istics of a large number of individuals upon the dependent variables.
Typically, little attention was given to the socio-cultural and communi-
cation environment surrounding these individuals. These dimensions are
essentially system variables.

3. When the influence of the social system was considered in
the analysis, often only one system (e.g., one village, one community,
etc.) was analyzed for its influence upon the dependent variable. Al-
most no earlier studies included a sufficiently large number of com-
munities (or other systems) to permit testing of hypotheses about
community characteristics and adoption of innovations with any degree
of rigor.

In the study herein described, we go beyond the analysis of the
characteristics of individuals, and include characteristics of the
system (i.e., the village) in which the individuals (whose behavior

we seek to explain) reside.

I."

.-

._
l...
_
u
I
7

..._4
HT

 

18

System Analysis

Because of its emphasis upon the complex functional inter-relations
between members of a group, organization, culture, or social system,
system analysis seems to be peculiarly appropriate to the study of the
introduction of innovations in less developed countries. What are the
distinctive characteristics of system analysis? What is the precise
meaning of the various terms and concepts utilized in this type of
investigation?

Schramm (1965) offers the following definitional discussion of
a system:

When we refer to a system, we mean a boundary-main-
taining set of interdependent particles. By inter-
dependence, we mean a relationship [among] parts
such that anything happening to one component of a
system affects, no matter how slightly, the balance
and relationship of the whole system. By boundary-
maintaining we mean a state in which the components
are so related that it is possible to tell where the
system ends and its environment begins.

From the standpoint of diffusion research, we can define a develop-
ment system as a complex of interrelated parts where innovation-receiving
units are linked to each other and to a central innovation-introducing
unit or change agency through channels of communication. The linkage of
the two subsystems-~the change agency (or the source) and the adopter
population (or the receivers)--is a basic requisite for innovation dif-
fusion and development. For the purposes of the present study, the major

components are: (1) the source or the change agent responsible for in-

troduction of (2) innovations (3) in a receiver (or adoEter) pOpulation

 

19

of individuals, groups, or organizations (u) who are linked to each

other and to the sources by channels of communication.

 

An individual variable is defined as any of a group of seledted
social, psychological, cultural, or communication characteristics of
individuals within the social system, which affect their individual
innovativeness in adopting a given innovation.

System variables are defined as the mean level of any of a
group of selected social, psychological, cultural, or communication
characteristics within the social system, which affect the innovative-
ness of individuals within the social system.

A system effect is defined as any observable or measurable
change in the attitude or behavior of the individuals within the social
system, that can be attributed to the influence of the system variables.

In certain research traditions, analyses are sometimes made
which explore the complete network of interrelationships between the
social, cultural, psychological, and communication variables in a
given social system. However, such an investigation (often called a
"social system analysis") lies beyond the realm of the present study, for
several reasons.

First,many attempts to conduct such investigations have been
primarily descriptive in nature. The present investigation purposes to
find correlates of innovativeness which can be useful in predicting the
adoption or rejection of new technologies.

Second, few reliable and appropriate research methods have been

devised to conduct such a study. Of course, for many purposes, a full

 

 

20

analysis of the entire social system with all the complexities and
interrelationships of its components is not necessary. Most studies
have been somewhat selective in their fecus.

The present study also is designed as only a partial analysis.
It is not designed to present a comprehensive description of every
facet of the social systems selected for study, but rather to explore
certain rather general hypotheses regarding various socio-cultural
and communication variables, at both the individual and system level,
and their relation to a single dependent variable (innovativeness).

This study, therefore, deviates from the "system analysis"
often utilized in physics, engineering, or the natural sciences. It
is a "system analysis" only to the extent that it utilizes selected
system variables to explain variance in the dependent variable.

The essential feature of system analysis pertinent to this
study is that data for selected characteristics are gathered from
various levels in a total system (i.e., from the sub-systems compos-
ing the total system), and analysis is made of their function, both
independently and jointly, within the system under study. Thus, one
is concerned in the system approach not only with the major or primary
effects of the independent variables, but with their interaction
effects as well.

Various terms have been used by different authors in describ-
ing the various elements involved in systems analysis. Blau (1957)

gave the following definition of structural effects:

 

 

21

The general principle is that if ego's X affects not
only ego's Y but also alter's Y, a structural effect
will be observed, which means that the distribution
of X in a group is related to Y, even though the in-
dividual's X is held constant. Such a finding indi-
cates that the network of relations in the group with
respect to X influences Y. It isolates the effects
of X on Y that are entirely due to or transmitted by
the process of social interaction.

Contextual analysis has been described by both Sills (1961) and

 

Riley (1963, pp. 700-703) as an approach by which the effects of the
properties of groups or collectives on individuals have been studied.
Davis Efnéif (1961, pp. 215-225) use the term compositional effects to
describe the analysis of the influence of variables formed from the
properties of members of the collectivities. Another term used

(Valkonen, 1966) is ecological analysis, which may take either of two

 

forms: (1) taking areal units as wholes in their own right, for some-
what global analysis, or (2) taking properties obtained by aggregating
properties of smaller units (e.g., of individuals).

Selvin and Hagstrom (1963) classify the properties of a system

into (1) aggregate properties, based on characteristics of smaller units

 

within the system being described, and (2) integral properties, which

 

are not based on smaller units, but on the system as a whole.

Lazarsfeld and Menzel (1961) disagree with these terms, and speak instead
of analytic and global properties of a system. To illustrate these
concepts, an aggregate or analytic property of a system might be the mean
level of exposure to the mass media of communication, whereas an integral

or global property might be the presence or absence of a mass media facility

such as a radio station, in the system under consideration.

 

22

Cattell posits three categories of system variables (1951):
(1) sygtality‘variables, which describe the behavior of the social
system acting as a whole, (2) structure variables, which describe
internal characteristics, and interaction among members, and (3) 22227
lation variables, which describe the distribution of status, persona-
lity, and attitude-interest variables among the members of the system.

Undoubtedly these semantic disagreements have been occasioned
by efforts to speak more precisely and accurately about the variables
being considered by various researchers. In the present study, our
primary concern is not to end this debate but to ascertain whether
variables of this general description help explain additional amounts
of variance in the innovativeness of the respondents in the study. We
employ two broad categories of variables in this respect: (1) 229$?
vidual variables, which pertain to individual characteristics of the
respondents, and (2) systems variables, described elsewhere as con-
textual, compositional, Structural, aggregate, analytic, or population

variables.

Theoretic Rationale
In considering the effects of system variables within the Nigerian
villages studied, we ask: How do they influence the innovativeness of
the individuals in these villages? In what ways do the system variables
affect the individual's decision to adopt or not adopt the innovations?
Several possibilities exist.
l° There is, first of all, the possibility that the differences

between the villages included in the study are random. This would mean

 

23

that the variance in the system variables from village to village
simply is an expression of the normal distribution of such variables,
and has no systematic relationship to the dependent variable. Thus,
one would expect to find differences regarding the system variables as
one moves from village to village, but these differences would not be
significantly correlated; i.e., they would not covary concomitantly
beyond chance expectations with differences in innovativeness in the
various villages. Such a situation would be expressed by the null
hypothesis:
HO = The various elements of the total system in
the villages are not related to the innova-
tiveness of the residents of these villages.
2. A second possibility is a situation in which the differences

in the adoption of innovations by the villages are associated with

variance in a single independent variable among the villages. Take,

 

for instance, a situation where innovativeness varies beyond chance
expectations from village to village. If the presence of a single
variable (e.g., educational level) also varies from village to village,
one might hypothesize that different levels of education among the
villages explain the different adoption rates. Such a situation may
be expressed by the following mathematical equation:

Y;j = a + 1313?]. + bzxij
In this expression, Yij is the expected value of the dependent variable
of individual i living in the social system j and having the value xij

of the independent variable; E5 is the aggregated (e.g., the mean) level

of the independent variable in the social system j, and a, bl’ and b2

 

24

are constants. For example, if Y is the innovativeness of individual
farmers in Nigerian villages, the individual level variance of it in
the whole population would be explained by the education (for example)
of the individuals (assuming b2 # 0), and additionally, by the mean
level of education inthe social system (assuming bl # 0). If b1 and
b2 have the same sign, both individual level education and mean level
of education have an effect in the same direction. If b1 and b2 have
different signs, the variables on the two levels have opposite effects.
Further, if bl = O and h2 = 0, there is no non-random effect at all;
if bl = 0 while b2 # 0, there is only an individual level effect; if
bl # O and b2 = 0, there is only a systems effect. In the last case,
education would affect the behavior of individuals only through the
mean level of education in the given social system.

In most cases, however, because of the complexity of the total
situation confronted by the villages, it is probably not realistic to
assume that the variance in the dependent variable (such as innovative-
ness) can be explained by one independent variable, such as level of
education. We therefore must look further.

I 3. A third possibility, which occupies the position of central

concern in the present investigation, is one in which a number of in-

dividual variables and a number of systems variables operate to

 

explain the variance in the dependent variable. Individual A, living
in social system Z, would have some score on the dependent variable Y,
produced both by his individual score for independent variable X,and

by its interaction with the mean level for independent variable X,

25

throughout his social system. Similarly, individual A would have indi-
vidual scores for every independent variable considered (x1, x2, x3...xn),
and the social system would have a mean level for every independent

variable considered (E1, :5, I,...§£). Thus, the individual variables

and the system variables (aggregated individual variables) act independently
and jointly to influence the dependent variable for individual A in the
social system. The entire system's position on the dependent variable

is represented by some measure of central tendency (SUCh as the mean) for
every individual included in the sample.

In summary, the present study is designed to examine the effect
on the individual's innovativeness of individual independent variables
plus system independent variables. We hypothesize that the two levels
of the independent variables (i.e., individual and system) operate
linearly and interactively to explain the variance of the dependent
variable within the individuals studied.

We propose to examine whether the level of the independent
variables within the system operates as an intervening variable, to
augment the effect of the level of the independent variable within the
individual. Figures 1a and 2a illustrate this point.

Many studies have been designed to explicate direct relation-
ships Qa in the paradigm, the effect of individual characteristics ”g¥f
upon innovativeness). We propose to investigate whether these indivi—
dual characteristics are mediated in some way (i.e., either increased
or decreased impact) by the characteristics of the system in which the

individuals are found. We therefore are proposing to compare 1a, the

 

26

 

Individual Characteristics Individual Innovativeness
of Nigerian Farmers }.of Nigerian Farmers
(Independent Variables) (Dependent Variable)

Figure la: Conventional Direct Effects Model
of Individual Innovativeness.

 

Individual Characteristics Individual Innovativeness
of Nigerian Farmers ) of Nigerian Farmers
(Independe t Variables) (Dependent Variable)

ystem Characteristics
of Nigerian Villages
(Independent Variables)

Figure lb: Hypothesized Direct—Plus-Mediational
Model of Individual Innovativeness

Figures la and lb: Comparison of Direct Effects Model
and Direct-Plus-Mediational Effects
Models of Individual Innovativeness

 

27

direct effects, with B, the individual or direct effects as mediated by
the system effects, and are hypothesizing thatJUDwill be greater.

The guiding hypotheses on which this analysis is predicated may
be stated formally in the following manner:

H1 The individual level of innovativeness for Nigerian
villagers is positively related to the individual
level of selected moderrization characteristics.

H The individual level of innovativeness for Nigerian
villagers is positively related to the system level
of selected modernization characteristics within
each village.

H3 The individual level of innovativeness for Nigerian
villagers is positively related to the joint in-
fluence of the individual level and the system level
of selected modernization characteristics within each
village.

Naturally, if H3 is supported, H1 and H2 must also be supported.

How to Determine System Effects
As indicated earlier, Durkheim (1938, 1951) was one of the first
social scientists to give attention to the study of systems effects. In
calling for the recognition of influences external to the individual, ‘fifi
Durkheim (1951, pp. 37-38) observed:

But if no reality exists outside of individual con-
sciousness, it [i.e., sociological method] wholly
lacks any material of its own...0n the pretext of
giving the science a more solid foundation by estab-
lishing it upon the psychological constitution of
the individual, it is thus robbed of the only object
proper to it. It is not realized that there can be
no sociology unless societies exist, and that soci-
eties cannot exist if there are only individuals.

 

28

Thereupon Durkheim described a method for isolating these social
influences (roughly comparable to system effects) by finding the relation-
ship between the distribution of a given characteristic in-a social
system and an effect criterion, while holding this characteristic con-
stant for individuals. Since individual differences have been thus
controlled, any significant covariation between the variables and the
effect criterion is due to the characteristics of the system, or system
effects.

Blau (1960) also argued for the simultaneous consideration of
individual and system characteristics, so as to isolate the effect of
the system's characteristics. He pointed out:

...by treating individuals or even subgroups as
independent units of analysis that can be classi-
fied and re-classified according to any one of

their characteristics, this procedure (i.e., the
observation of regularities among individual
members or subgroups) necessarily ignores the unique
constellation of relationships between groups and
individuals in the organization-—its Gestalt.

Blau therefore proposed a type of analysis similar to that described
by Durkheim. In speaking of the influence of values, he stated: "To
isolate the external constraints of social values from the influence of
the individual's internalized values, it must be demonstrated that the 1T}
prevalence of a value in a group is associated with social conduct when
this value is held constant for individuals" (Blau, 1960, pp. 178-179).

He outlined the following 3-step strategy:

(1) An empirical measure Z is obtained on some characteristic of

individual members of a system.

 

29

(2) These individual scores are aggregated into one index for
each system, ng.

(3) Determine the relationship between the system attribute (ng)
and some dependent variable (W), while holding the individual variable
(Z) constant. Thus, the effect of ng on W will be system effect.l

Tannenbaum and Bachman (196%) criticize this strategy of Blau’s
on the grounds that Blau dichotomizes continuous variables, both for
individual (rows) and system (column) characteristics. Two serious weak-
nesses are pointed out: (1) the problem of contaminating individual
differences with system effects, and (2) the problem of contaminating
system effects with individual effects. Tannenbaum and Bachman say these
weaknesses may be remedied in several ways.

One method is to utilize more precise matching of the individual
variable (Z). The larger the number of categories of the individual
variable used in matching, the greater the accuracy. Instead of a 2 X 2
table, this would produce a N X 2 table, N being the number of categories
used in matching on the individual variables. This method could also be
utilized to expand the range of group variables (ng).

Another method described by Tannenbaum and Bachman utilizes cor-
relational analysis. By the use of partial correlation, a system effect
can be isolated by correlating between ng and W with Z partialled out.
An individual level effect can be determined by correlating Z and W,

with ng partialled out.

1 ._
Blau's ng corresponds to blxj, z to b2xi., and W to Yi' as used
in the present dissertation. ] 3

 

30

'If linearity between the variables can be assumed, the individual
and group variables may be used as independent variables for multiple
regression analysis.

In the present study, we utilize correlational analysis, includ-
ing Pearson product-moment zero-order correlation, partial correlation,
and multiple correlation techniques. For further details of the plan

of analysis, see Chapter III.

Some Possible Fallacies

In analyzing the data gathered in a study of individual and
system variables, and in subsequent generalizations from the analysis,
several fallacies are possible.

Riley (1963) describes two general sets of fallacies that may
be committed: (l):fallacies arising because analytical methods fail to
fit the model, and (2) fallacies arising because methods fail to fit the
facts. If the analyst's model refers to individuals in roles, but the
analysis is based on systems (large or small collectivities or aggregates),

there is a possible aggregative fallacy. On the other hand, if his

 

model refers to the system, but his analysis is based on individuals,

there is a possible atomistic fallayy.

 

Perhaps the best known error commonly committed in the form of
an "aggregative fallacy" was pointed out by Robinson (1950): the

ecological fallacy. This is the error of assuming that relationships

 

found between characteristics of groups of people, taken as a group,
hold true for the individuals within those groups. In Robinson's

example, census data for groups of individuals indicated a negative

 

31

correlation between foreign birth and illiteracy. However, it cannot
be assumed, on the basis of this data, that a foreign-born individual
will necessarily be literate. Thus, a group analysis is inappropriate
if the hypothesis refers to the individual.

Conversely, a conclusion about groups of individuals in census
areas based only upon individual data is subject to a possible atomistic
fallacy. If the hypothesis refers to the group, then the analysis must
be based on groups.

The second set of fallacies described by Riley, in which the
method fails to discover the relevant facts, also subsumes two easily

committed errors.

 

A psychologistic fallacy is committed if the researcher. when
explaining individual behavior, looks only at the characteristics of

each individual, disregarding or overlooking the significance of - n

ML! 5

factors such as the character of the village or community. Rogers
(forthcoming) remarks that, in much previous diffusion research,

"Because the data were gathered from individuals as the units of

response, our focus has been upon individuals' intra-personal variables,
largely to the exclusion of social structural and interpersonal
variables." By including an analysis of the characteristics of the

social system, in addition to individual intra-personal variables, the \f
present study avoids this error.

The second error, the sociologistic fallacy, is committed when

 

the individual level processes are disregarded when interpreting rela-

tionships between variables describing collectivities. For example,

 

32

a group analysis might show a positive correlation between transiency
rates and suicide rates. If suicide rates are explained in terms of a
social process, and individual level variables are omitted, a sociologistic
fallacy may have been committed.1
Selvin (1958) reanalyzed the methodological reasoning in Durkheim's
Suicide (1951) and concluded that Durkheim paid too little attention to
the individual. He accused the early sociologist of treating the group
as an entity, and committing the error of reification.2
Kendall and Lazarsfeld (1950) contend that the principle of ana-

lyzing group composition can lead to positive gains if the data allow

for simultaneous group and individual level analysis.

 

 

There is no reason why unit data cannot be used to
characterize individuals in the unit. A man who does
not have malaria in a unit where the incidence of
malaria is very low probably feels differently about
his state of health than does the man who has no malaria
but serves in a unit with high incidence.... In
terms of actual analysis, the matter can be restated
in the following terms: Just as we can classify
people by demographic variables or by their atti-
tudes, we can also classify them by the kind of en-
vironment in which they live. The appropriate vari-
ables for such a classification are likely to be unit
data. A survey analysis would then cover both per-
sonal and unit data simultaneously.

In the investigation described herein, we have attempted to avoid

the psychologistic and sociologistic fallacies by including both system

 

lRiley's psychologistic and sociologistic fallacies are similar
td Scheuch's (1965) description of the individualistic fallacy, which
is the negation of the usefulness of an explanation that treats the
collectivity as collectivity, and the group fallacy, which is the con-
verse of the above.

 

 

Reification is defined in a footnote on page .12.

 

33

and individual variables in our analysis, and avoid the aggregative and
atomistic fallacies by basing the\analysis on the appropriate level (i.e.,
system or individual). The partial correlation technique also aids in
preventing confounding system variables with individual verifibles, and

‘—a—-

vice versa.

Dependent Variable

The dependent variable utilized throughout thelresent study is
innovativeness, defined as "the degree to which an individual is relatively
earlier in adopting new ideas than other members of his social system"
(Rogers, 1962, p. 20).

Over the ten years preceding this study, the Eastern Nigerian
Ministry of Agriculture introduced and actively promoted a number of
agricultural innovations, some 1n of which were chosen for inclusion in
the second phase of the Nigerian study (see Table 1). Although these
14 innovations vary widely in type, in amount of skill required for
successful cultivation, in adaptability to various climatic and soil con-
ditions, initial investment required, etc., they are available and
economically feasible for a large number of Eastern Nigerian farmers.
Many of these innovations are utilized simultaneously, and others are
complementary. It is therefore held to be theoretically logical to use
adoption or non-adoption of these 1n innovations as a criterion of inno-

vativeness.l

 

1See Appendix Bfor analyses using two additional indices of
innovativeness.

 

34

Description of Innovations

Aldrin dust is an insecticide manufactured by the Shell Oil
Company. Although it kills all kinds of insects, it is especially
used against the yam beetles-~a pest that flies in from its breeding
grounds near a river at or near planting time, sustains itself tem-
porarily on the yam seed tuber, and later attacks young tubers. Ap-
plying Aldrin dust to the seed tuber before planting kills the yam
beetle. The insecticide is distributed by extension agents.

GCH-7 Cassava is a new cassava variety that was introduced be-

 

cause of its high-yielding characteristics. From the standpoint of
quantities grown and consumed, cassava is the most important food crop
in Eastern Nigeria. Stem cuttings, about one foot long, are inter-
planted on the sides of the yam heaps.

Three innovations—~oil palm rehabilitation, rubberplanting,

 

 

and cocoa plantingf-are highly similar. These are major tree crops,

 

and are an important part of the six-year development plan of the
Eastern Region. In order to adopt one of these innovations, a farmer

is required to furnish a relatively large amount of land, a minimum of
five acres. He signs a formal contract with the Ministry of Agriculture,
which obligates him to clear, space, mulch, and maintain his land ac-
cording to specific procedures. The Ministry supplies seedlings,
fertilizer, and instructions without charge for a period of five years.

Fertilizer isea chemical compound designed to supplement the

 

natural fertility of the soil. An aggressive promotion program has been

conducted to encourage its purchase and use. This program has a

 

3S

specially trained staff, its own sales campaigns, and distributors, and
special demonstrations are conducted to show farmers how to apply
fertilizers on cassava, yam, and maize.

The Fund for Agricultural and Industrial Development (FAID) is

 

an organization established to encourage the use of government agri-
cultural credit. At the time of the study, however, the program was
suffering from administrative difficulties, and loans were difficult
to obtain.

Community plantations are organized by representatives of the
Ministry of Rural Development to encourage cooperative use of village
land for agricultural development. Generally, the plantation is de-
voted to one of the major tree crOps being promoted by the Ministry of
Agriculture. The rural development officers organize village cOOpera-
tives, formally register the association with the government, survey
and measure land, instruct villagers as to the availability and
requirements of Ministry of Agriculture programs and other development
projects, and assist them in getting SUpplies and services from the
Ministry of Agriculture.

Improved livestock is an innovation designed to promote the
breeding and raising of higher quality farm animals. Generally, the
adoption of this innovation required economic means for purchasing the
necessary foodstuffs, veterinary services, and pens for the animals.
Only a few farmers have adopted the program, and most of the beef and

pork consumed was imported from other parts of the country.

 

36

NS-l Maize is a hybrid corn variety that has high-yielding
qualities. The new variety was developed Specifically to supply food
fer the poultry program, but many people use it for food also. The
Ministry of Agriculture supplies the seed without charge and it sup-
posed to purchase the farmer's harvest later. However, the latter
feature of the program was suffering from administrative difficulties
at the time of the study.

Improved poultry breeding consisted mainly of teaching farmers

 

proper management, supplying them with chicks and foodstuffs, providing
them with disease preventives, and making markets available for the
eggs. Problems such as high cost of feed, high incidence of disease
and death among chicks, and low price of eggs were besetting the pro-
gram at the time of the study.

Improved rice is an innovation that requires a whole complex of

 

new technologies, and is characterized by considerable expense. Rice
improvement programs include improved seed, new cultural practices, seed
and nursery preparation, weeding, use to fertilizers, and of dams,
irrigation, and drainage. Because of its complexity and expense

rice improvement is sometimes incorporated into village plantations.

The Stork oil press is a hydraulic hand press for obtaining oil

 

from palm nuts. Although it is a considerable improvement over the
conventional screw presses and other traditional methods, the new press
costs a great deal of money (about $612.00) in addition to relatively
high labor costs. Detailed, technical operating instructions are also
required. These disadvantages probably prevented wide scale purchase

and utilization of the innovation.

 

37

Eggetable seeds are sold in small packets at minimal cost, and

 

include onions, tomatoes, and other European vegetables, as well as
local varieties. At the time of the study, vegetable planting was
receiving increasing attention, with some 80,000 packets being sold in
1966.

From the standpoint of their use and characteristics, these lu
innovations can be classified according to the following pattern:

1. Food Crops-—Cassava, maize, rice, vdgetables, and cocoa.

 

2. Land=Commitmen€L InnovationsL-Oil palm, cocoa, rubber,
community plantations, and cassava.

3. Divisible InnovationsL-Aldrin dust, fertilizers, NS-l maize,
and citrus crops.

4. Elite innovationss-PAID agricultural credit, livestock,

 

rice planting, and the Stork oil press.

Independent Variables
Three criteria were utilized in the determination of the variables

to be included in the present analysis:

 

lLand commitment: innovations are innovations which require the
setting aside of a considerable portion of land in order to adopt.

2Divisible innovations are innovations which may be tried on a
small scale prior to complete adoption. -

3Elite innovations are innovations which, because of high
capital requirements, and agricultural skill demanded, can be adopted
6nly by farmers in an advantageous economic situation.

 

38

 

Table 1: List of Innovations Studied in
Phases I and II in Eastern Nigeria

 

 

 

Name of Innovation Percent of Farmers Adopting:___
Phase I Phase II
Fertilizer 22 39
Oil Palm (rehabilitation scheme) 20 17
NS-l Maize (hybrid variety) 16 32
Aldrin dust (insecticide) / 12 33
Poultry (improved breeds) 6 19
Rice production 5 8
Rubber production 5 u
Vegetable production 5 7
Cocoa production a 5
Community plantations u 13
Cassava (improved variety) 2 7
Livestock l NSc
Cashew l 3
Pineapple -b NS
Citrus crops - 10
PAID - 2
Farm settlements - NS
Stork oil press - NS
Adopted no innovations 53 37

 

aPercentages are based on 947 respondents in Phase I and 1,3u7
respondents in Phase II.

bDashes indicate that the percentage adOpting is less than one percent.

CNS means this innovation was not studied in Phase II.

39

(1) Previous research, both in the United States and in cross-
cultural settings, with special attention to research in less-developed
nations.

(2) Earlier analyses of similar data from the Nigerian Diffusion
Project by coworkers on the staff.

(3) Intuitive and theoretical reasoning. Very little research
has been conducted earlier in diffusion studies which deal with the
influence of system variables on innovativeness, so that background
and previous investigations afford limited empirical assistance in the
final choice of variables to be included.

0n the basis of the above criteria, the following independent
variables were selected and included in the data analysis:

(1) Cosmopoliteness. Cosmopoliteness is "the degree to which

 

an individual's orientation is external to a particular social system"
(Rogers, 1962, p. 183). It is hypothesized that earlier adopters are
more cosmopolite than later adopters. In the United States, Ryan and
Gross (1993) found that hybrid corn innovators traveled more often to
urban centers such as Des Moines than did average farmers. Menzel

and Katz (1955) found that innovative medical doctors made more trips
to out-of—town professional meetings than non-innovators. Goldsen and
Ralis (1957, pp. 25-28) found that Thailand farm innovators were more
likely to visit Bangkok. Out of 73 publications in the Diffusion
Documents Center at Michigan State University dealing with the relation-
ship between cosmopoliteness and innovativeness, almost 81 percent

reported a positive relationship (Rogers and Stanfield, 1966, p. 26).

 

#0

(2) Education. As a social characteristic that can enable potential
adopters to perceive the relative advantages of innovations more readily,
and can assist in implementing the break with traditionalism, education
is expected to be positively related to innovativeness, at both the indi-
vidual and system levels. Rahim (1961) found education positively
correlated with innovativeness in Pakistan, and Rogers and Stanfield
(1966, p. 22) reported that 79.6 percent of 193 publications in the
Michigan State University Diffusion Documents Center dealing with educa-
tion's relation to innovativeness indicated a positive relationship.

(3) Familism. To the extent that an individual farmer's orien—
tation toward his own primary family group is superordinate to his
orientation toward other groups within the social system, he will
be disinterested or resistant to change, especially if change is per-
ceived as threatening maintenance of the family. Familism is there-
fore expected to be negatively related to innovativeness at both the
individual and system level.

(9) Educational Aspiration. Educational aspirations are defined

 

as the level of education individuals would desire if they could be
young again. Since a high level of educational aspiration is believed
to indicate a more modern outlook, this variable is expected to be
positively related to innovativeness at both the individual and systems

level.

 

(5) Achievement Motivation. A precise and consensually accep-
table definition of achievement motivation is yet lacking in diffusion

literature. McClelland defined achievement motivation as "the desire

 

.+1

to do well, not so much for the sake of social recognition or prestige,
but to attain an inner feeling of personal accomplishment" (1961, p. 76).
Neill and-Rogers stated that achievement motivation is "that value,
instilled in the individual through the socialization process, in which
the individual feels a need or desire to excel in reaching certain goals
only for the satisfaCtion of reaching the goals and not for the rewards
of the goals or ends involved" (1963, p. 2). Generally, achievement
motivation is taken to indicate a desire for excellence.

McClelland reported that this variable was positively related to
levels of entrepreneurial activity, economic growth, and rate of
national economic growth. Neill and Rogers (1963, p. 12) found occupa-
tional achievement motivation among Ohio farmers positively related to
productive man work units, man days of labor on the farm, size of farm
and innovativeness. There is, therefore, considerable theoretical
reasoning to indicate that villagers with high levels of achievement
motivation are more innovative in adOpting new ideas and practices.

(6) Empathy. Empathy was described by Lerner (1958) in various
ways: "the capacity for identification with new aspects of [the res-
pondent's] environment" (p. 99), "the capacity to see oneself in the
other fellow's situation" (p. 50), "the capacity for rearranging the
self-system on short notice" (p. 51), "mobile sensibility" (p. 99),
and "psychic mobility" (p. 51). Throughout all these descriptive
phrases, the central theme is "the individual's ability to put himself
in another's role." Although findings regarding the role of empathy

in the modernization process have been mixed, it is generally believed

 

E2

that empathy is positively related to innovativeness, at both the indi-
vidual and system level. '

(7) Knowledge of Agricultural Change Agents. Contact with change
agencies has been found to be positively related to innovativeness in a
great many studies reported in diffusion research literature. Rogers
and Stanfield (1966, p. 26) report that, out of 136 studies dealing with
the relationship between change agency contact and innovativeness, 91.9
percent reported a positive relationship. In Nigeria, agricultural workers
from the Ministry of Agriculture were the change agents seeking to intro-
duce change into the rural areas. It is hypothesized that acquaintance
with these agricultural workers and familiarity with their work is
positively related to innovativeness, at both the individual and systems
level.

(8) Economic Aspiration. As measured in the Nigerian study,

 

economic aspiration is more oriented toward a desire for economic develoP-
ment of the entire village, than toward a desire for personal aggrandize-
ment. Thus, a high degree of economic aspiration_should indicate a
high level of desire for modernization, economic advancement, and
improvement of the village conditions. It is therefore hypothesized
that economic aspiration is positively related to innovativeness, at
both the individual and system level.

(9) Literacy. In many research investigations conducted, literacy
(along with education) seems to emerge as one of the general correlates
of innovativeness. The innovative person is likely to be literate, as

well as have more education than others in his social system. The

 

43

weight of evidence seems to overwhelmingly indicate a positive
relationship between these variables and innovativeness. Rogers and
Stanfield (1966, p. 22) found that 19 out of 27 publications dealing
with the relationship between innovativeness and literacy reported a
positive relationship. 1

(10) Newspaper, Radio and Film Exposure. Taken together, these

 

three variables form a mass media exposure index. They are somewhat
interrelated, as Table 2 indicates. However, because there are marked
differences in the mean exposure to these three communications media,
they will be included in this analysis as separate variables. When
ranked according to mean amount of exposure to these three media, the

order is: (1) radio, (2) newspaper, and (3) film (see Appendix D ).

 

Table 2: Intercorrelations Among Newspaper Exposure, Radio
Exposure, and Film Exposure (N = 1,1H2).

 

 

 

Variable ‘Variable
2 3
1. Newspaper Exposure .47 .28
2. Radio Exposure -—- .40
3. Film Exposure _--

 

Exposure to the mass media of communication is probably a crucial
factor in promoting innovativeness and economic development in less-

developed nations. Lerner (1956), Deutschmann and Fals Borda (1962),

   

an

and Frey (1966) all found a strong positive relationship between mass
media exposure and innovativeness. Rogers (1966) found that mass media
exposure was highly correlated with innovativeness in Bolivia, explain-
ing 67 percent of the variance in innovativeness in one city, and 10
percent of the variance in another. Newspaper, film, and radio expo-
sure in Nigeria is therefore expected to be positively correlated with
innovativeness at both the individual and the systems level.

(11) Agricultural Media Exposure. From one viewpoint, this

 

variable overlaps the previous three variables considerably, and is in
fact highly correlated with newspaper exposure, radio exposure, and film
exposure (r = .33, .43, .42, respectively). Its specific function is to
measure the exposure to agricultural messages being disseminated by the
mass media. It has the additional function, however, of being an
indirect measure of contact with extension agents or agricultural workers.
Agricultural media exposure is therefore expected to be positively re-
lated to agricultural innovativeness at both the individual and the
systems level.

(12) Credit Orientation. A favorable attitude toward the use of
agricultural credit to finance further investment in agricultural pro—
duction is a characteristic generally associated with more innovative
farmers. In a traditional or subsistence system, decisions for agri-
cultural alternatives are based not on monetary gains but rather on the
protection of one's livelihood. A subsistence-level farmer cannot

jeopardize his supply of food and his income without some knowledge of

   

 

us

what is involved in such a venture. Thus, innovativeness is expected‘
to be positively related to credit orientation, at,both the individual
and systems level.

(13) Level of Living. As an indicant of socio-economic status,

 

level of living could be viewed as either a product or an antecedent

of innovativeness. Since many of the innovations included in this

study require some degree of economic prosperity, it is hypothesized
that level of living is positively related with innovativeness. Further,
since certain of the 14 innovations are oriented toward community

action, level of living at the system level should also be positively
related to innovativeness.

(l4) Correct Knowledge of Agricultural Innovations. A necessary

 

prerequisite to the successful adoption of agricultural programs, cor-
rect knowledge of the innovations is expected to be positively related
to innovativeness. As a cognitive correlate of innovativeness,

Rogers and Stanfield (1966, p. 24) found that knowledgeability was re-
ported to have a positive relationship in 53 out of 66 research studies
dealing with the variable.

(15) Formal Social Participation. Individuals who more actively

 

participate in the activities of their social system are more likely
to be innovative. As one index of social participation, group member-
ship is expected to be positively related to innovativeness. Rogers
and Stanfield (1966, p.26) found that 123 out of 156 research studies
dealing with group participation reported a positive relationship to

innovativeness.

 

46

Table 3 lists the independent variables utilized in this analysis,

with the hypothesized relationship to innovativeness. '

 

Table 3: List of Independent Variables with their
Hypothesized Relationship to Innovativeness.

 

 

Variable‘ Relationshipa
Individual System
1. Cosmopoliteness + +
2. Education + +
3. Familism' - -
4. Educational Aspiration + +
5. Achievement Motivation + +
6. Empathy + +
7. Knowledge of Agricultural Workers + +
8. Ebonomic Aspiration + +
9. Literacy + +
10. Newspaper Exposure + +
11. Radio Exposure + +
12. Film Exposure + +
13. Agricultural Media Exposure + +
14. Credit Orientation + +
15. Level of Living + +
16. Correct Knowledge of Innovations + +
17. Formal Social Participation + +

 

aSince the relationship of these variables is expected to be interactive
for both the individual level and system level, the same direction of relationship
is gypggbggiggdéforoeagh_yariable at both the individual and system level.

 

 

CHAPTER III

RESEARCH METHODOLOGY

The Diffusion Project

The present study is based on an analysis of a portion of the
data gathered in the three-nation Diffusion Project conducted by
the Department of Communication, Michigan State University, and
sponsored by the United States Agency for International Development.
This project was initiated in December, 1964, and terminated in
December, 1968. Broadly speaking, the study concentrated on pro-
blems associated with the introduction of modern technology among
peasants in the nations of Brazil, India, and Nigeria. Only the
data gathered from the Nigerian portion of the study were used in
the present investigation.

Among the objectives of the Diffusion Project was the specific
goal of gathering information useful to change agents in their efforts
to introduce innovations. This information included the identification
of village innovators and opinion leaders, and their distinctive social
and economic characteristics, communication behavior, attitudes, and
values; the role and influence of various communication channels such
as the mass media, opinion leaders, interpersonal communication, and

demonstrations; the response to various communication and economic

47

 

48

incentives which were introduced into the country;.gpd the influence of
price incentives, land tenure, credit, marketing practices, the shift

from Subsistence to commercial agricultural production, and other economic
factors.

The Diffusion Project was designed to consist of three data-
gathering phases in each of the three nations. The first phase was an
analysis of the relative success or failure of various programs of
change in agricultural production in some 80 villages in each country.
The unit of analysis was the village, and the data were secured from
secondary sources and through interviews with village leaders and change
agents.

The second phase was an analysis of data obtained mainly through
structured personal interviews with villagers living in some 20 vil-
lages in each of the three countries. The unit of analysis was the
farm family, and respondents were male household heads. The purpose
was to trace the diffusion of farm innovations within the villages,
and especially to study the role of innovators and opinion leaders in
the diffusion process.

In the third phase, certain communication techniques and incen-
tives for the ad0ption of innovations were introduced into each of the
nations. The effectiveness of these incentives and techniques were
evaluated through observation and follow-up interviews as part of a
controlled field experiment to last over several years.

However, political disturbances and civil strife in Nigeria

forced curtailment of the project in this nation} so that Phase

 

See Preface.

1+9

Three had to be cancelled in Nigeria after its initiation. Thus, only
the first two phases of the Nigerian Diffusion Project were utilized

in the present study.

Conduct of the Study
Phase One

Selection of Villages: A random sample of 30 countries was

 

selected from among the 76 counties in the Ibo- and Ibibio-speaking
areas of the Eastern Region of Nigeria. Within each county, at

least two villages were chosen for the study. The villages were
chosen in pairs: half of them were "success" villages, and half of
them were "failure" villages, in terms of their acceptance of Ministry
of Agriculture extension programs.

Several criteria were used in selecting the villages. The
change agent must have worked in the village for at least nine months.
The selection was also based on ratings by the agricultural super-
visors and the county supervisors regarding the "most successful"
and the "least successful" villages in the counties, as related to
innovations sponsored by the Ministry of Agriculture, and on examina-
tion of extension records concerning the number of innovations in
operation in a given village, the number of farmers participating,
and the amount of supplies distributed during the previous years.

After some additions and substitutions necessitated by defi-
ciency in pairings, the list of villages included 34 "success" vil-
lages, and 37 "failure" villages. Of the 71 villages, 52 were Ibo-

speaking and 19 were Ibibio-speaking.

 

50

Two weaknesses in village selection procedures should be
mentioned. Extension agents and county supervisors were generally
reluctant to designate a village as a "failure" village. Thus the
reliability of this designation may be questionable. In addition,
the 30 counties chosen in the first stampling step were not ranked
on relative success, so that a "success" village in one county may
well be a "failure" village in another.

Selection of Respondents: Data were obtained in Phase One

 

from village leaders, change agents, and innovators.

Leaders were chosen through sociometric nominations. Six
typologies of leadership were utilized: (1) civic, (2) religious-
traditional, (3) religious-modern, (4) administrative-traditional,
(5) administrative-modern, and (6) educational. In addition, two
informal leadership typologies were utilized: (1) village affairs
opinion leadership, and (2) village farming opinion leadership.
Leadership nominations were taken first from the village school head-
master and village chiefs, then from those nominated by the headmaster
and chiefs, then throughout the village until some ten informants had
been interviewed. From 13 to 16 leaders and innovators were selected
and interviewed in each village.

Instrument Construction: The interview schedules were translated

 

into Ibo and Ibibio, and reviewed and revised by the research staff.
These revised schedules were then back-translated into English, to check
for ambiguities. A two-week pretest revealed further weaknesses in the

instrument, leading to deletion or revision of certain words or questions.

 

51

a

 

Interviewer Selection and Training: Criteria used in selecting
interviewers included education, cultural origin, age, and experience
with rural people. All interviewers were school teachers (or the
equivalent) holding the Grade II teaching certificate. All interviewers
were at least 25 years old, and spoke the dialect of the village where
they worked.

Interviewer training included one week's orientation at research
headquarters, Enugu, two week's pretest in the field, and another week
at headquarters in revising procedures and discussing problems that
were encountered.

Data Collection: Phase One interviewing was conducted between
May 15 and August 27, 1966. Altogether, a staff of 11 Nigerian inter-
viewers contacted 947 rural people in 71 villages throughout the Eastern
Region. Interviewers averaged three completed interviews per day, with
each interview consuming an average of one to one and one-half hours.
Each interviewer worked in each village for two weeks, and was visited
at least once each week by a supervisor who corrected mistakes and re-

viewed the interviewer‘s conduct in the village.

~Phase Two

Village Selection: From the total of 71 villages studied in
Phase One, the 34 "success" villages were chosen as the basis for the
sample population in Phase Two. "Failure" villages were dropped from
consideration, since it was desired to study change where it had occurred

and "failure" villages were characterized by very low levels of adoption.

 

52

After ranking on access to the outside world and institutional
development, 18 villages were chosen as the basis for the Phase Two
study, with 9 villages used in the pre4test.

Respondent Selection: In Phase Two, the unit of analysis was

 

the individual. The respondents were in each case a male, the head
of his household, at least 20 years old, and farming some amount of
land. As indicated in Table 4, the general sample was comprised of
1,142 re3pondents, with the modal seleCtion from each village ap-
proximating 65, and the unweighted sampling rate approximating 57
percent of the respondents in each village.

To insure an adequate number of innovators and early adopters
in each village for incorporation into the field experimental activi-
ties in Phase Three, some 205 innovators and ex-servicemen were added
purposively to the sample. In certain descriptive statistics, these
1,142 general sample respondents and 205 innovators and ex-servicemen
are considered together, but for all correlational analyses conducted
in the present investigation, the general sample of 1,142 respondents
was utilized. The population on which the correlational analyses are
based are indicated in each table of findings in Chapter IV.

Instrument Construction: The initial interview schedule was

 

constructed, based upon two pilot studies, one in Western Nigeria and
one in the Midwest Region. A five-week pretest was conducted in nine
villages, and appropriate revisions and deletions were made in the
interview schedule. The schedule included structured response items,

opennended response items, and interviewer ratings.

 

53

Interviewer Selection and Training: From among the Phase One
interviewers, nine "senior" interviewers were selected fer Phase Two.
In addition, nine "junior" interviewers (with less education and age)
were selected. Each interviewing team was comprised of one senior
and one junior interviewer.

A total of seven weeks was spent both in the field and at
headquarters, learning interviewing techniques, rapport-building,field
behavior, etc.

Data Collection: Phase Two interviewing was conducted between
November 2, 1966 and February 15, 1967. Altogether, a staff of 18
Nigerian interviewers talked with 1,347 farmers in 18 villages in the
Eastern Region. Each team of two interviewers worked in each village
for 6 to 7 weeks, and was visited at least once per week by a super-
visor. In addition to personal interviews, interviewers made personal
observations of the village and respondents' property. Each interview
averaged one and one-half hours in length, and two or three interviews

were completed each day.

Characteristics of the Sample
All respondents in the study were males. Nearly four-fifths
(78 percent) said that farming was their major occupation; i.e., the
job at which they spent the most time and/or earned the most money.
Of the remaining 22 percent of the sample, all but three respondents
said farming was a secondary occupation.
Slightly over half (52 percent) of the respondents were between

30 and 49 years of age; well over half (56 percent) had never attended

 

54

 

Table 4: Village Populations, Sample Sizes, and
Sampling Rates for Nigeria, Phase Two.

 

 

General Innovators, Total
Village Sample or Bx-Service- Total Eligible Sampling
Census men Sample Resp's. Rate

Omasi-Agu 65 14 79 162 49%
Obolo-Eke 61 l 62 62 100
Eka Uruk Eshiet 65 16 81 287 28
Nung Ikot-Ikot 63 16 79 250 32

Udo Offiong
Utit Uruan 52 l 53 61 85
Nsukara 66 10 76 152 50
Itu Ezinihite/ 66 16 82 156 53

Amaova
Oduma/Amankanu 65 16 81 300 27
Umuduru 80 O 80 83 96
Umuoke/Uga 65 9 74 116 64
Okwudo/Umuoseke 64 16 80 147 54
Ulli/Umuaku 64 16 80 ---a ---a
Obigbo 67 15 82 250 33
Umubiakwe 69 12 81 209 39
Uwana 64 16 80 197 41
Owutu 65 17 82 149 55
Ihiala/Umuezeawala 66 14 80 162 49
Obgoji 36 o 36 36 100
Totals 1142 205 13a? b b

 

aThe population totals for this village were not available, so no
sampling rate can be computed.

bSince only incomplete data are available, accurate totals cannot be
computed. However, of the villages where data are given, the mean population
is approximately 164 people, and the unweighted sampling rate is approximately

 

55

school, and another third had not completed their primary education.
Thus, the typical respondent in the study was a male farmer, the
head of his household, who was between 30 and 49 years of age, and who

had little or no formal education (Table 5).

Knowledge of Agricultural Innovations

 

Respondents were read a list of the 14 agricultural innovations
studied in Phase Two of the project, and asked what information they had
concerning each innovation. The interviewers then scored respondents
as to whether the information was correct. All interviewers were care—
fully trained and tested in the technical aspects of each innovation,
and specific points of information about each innovation were printed
on the interview schedules to assist the interviewer in his rating.

Only 3 percent of the respondents had no correct information about
any of the 14 innovations. At the other extreme, 9 percent of the res—
pondents had correct information about 11 to 14 of the innovations.

Over one-third (35 percent) knew something about four innovations, and
another third (33 percent) knew about seven innovations. Table 6
shows the number of innovations about which respondents knew at least
one correct item of information.

From 70 to 80 percent of the respondents had correct information
about the four best—known agricultural innovations: poultry (79 percent),
fertilizer (79 percent), oil palm rehabilitation (74 percent), and NS—l
maize (70 percent). However, between these four best-known innovations
and the fifth best-known innovation, a sharp decrease in information

appeared. Only 40 percent of the respondents had correct information

 

56

 

Table 5: Personal Characteristics of the Respondents (N = 1,347).

 

 

Characteristics Percentagesa
Occupation:
Farmer 78

Non-farmer 22

Age Group:

20 to 29 years 15
30 to 39 years 24
40 to 49 years 28
50 to 59 years 18
60 to 69 years 10
70 and older 5

Formal Education:

Never attended school 56
Primary, incomplete 33
Primary, complete 8
Secondary, incomplete 1
Secondary, complete 2
University, incomplete 0

University, complete --

 

aIn order to include every respondent interviewed in Phase Two,
these percentages are based on N = 1,347, which includes 205 innovators
and ex-servicemen purposively selected for the sample.

bLess than 1 percent.

 

 

57

 

Table 6: Number of Innovations Respondents Knew Correctly.

 

Innovations Percentagea
None 3
One 7
Two 6
Three 10
Four ' 12
Five 12
Six 11
Seven 10
Eight 7
Nine 6
Ten 7
Eleven to fourteen 9
100

aThese percentages are based on an N of 1,347 (see footnote,
Table 5).

 

58

 

Table 7: Respondents' Correct Information
About Agricultural Programs.

 

 

 

Innovations Known Correctly Percentagea
Poultry 79
Fertilizer 79

Oil Palm Rehabilitation 74

NS-l Maize 7O
Citrus 4O
Aldrin Dust 38 i
Improved Cassava 35
Rubber Planting Scheme 33
Community Plantations 33

Cocoa Planting Scheme 32

Rice DevelOpment Scheme 25
Vegetable Seeds 1?

FAID Credit Program 12
Cashew ll

 

aPercentages, which are based on N = 1,347, sum to more than
100 percent because many respondents knew about more than one innova-
tion (see footnote, Table 5).

 

59

regarding the citrus innovation. At the lowest level of correct infor- w
mation, 12 percent of the respondents had correct information about the

PAID credit program, and 11 percent knew about the cashew innovation.

Adoption of Agricultural Innovations

 

Respondents were asked if they had ever (planted, joined, used,

or bought) each of the 14 innovations studied. Somewhat less than two-

 

fifths (37 percent) of the respondents had never used any of the inno-
vations, less than one-fifth (19 percent) had tried one, and smaller
percentages had at some time used two (14 percent) or three (10 percent)
of the innovations. Only 1 percent had used nine or more of the innova~
tions. Eighty percent had used three or less (including none) of the
innovations (see Table 8).

Considerable differences were found in the percentage of respon-
dents adopting each of the innovations. As indicated in Table 9, almost
two-fifths of the respondents had adopted fertilizer (39 percent),
almost a third had adopted NS—l maize (32 percent), almost a fourth had
adopted Aldrin dust (23 percent), and almost a fifth had adopted poultry
(19 percent). Below these first four innovations, the percentage of
adoption declined rapidly. Less than 10 percent of the respondents had
adopted either rice (8 percent), improved cassava (7 percent),
vegetable seeds (7 percent), cocoa (5 percent), rubber (4 percent), or
cashew (3 percent). Only two percent of the respondents had adopted
the use of FAID credit.

Thus, the adoption rate was quite low for most of the innovations,

with certain innovations being adopted by hardly anyone. In general,

 

 

 

60

the highly divisible innovationsl (fertilizer, NS-l maize, Aldrin dust,
and citrus) had the highest rate of adoption, with these innovations
requiring a substantial commitment of land (such as cocoa, cassava,

oil palm, rubber, community plantations) being adopted by fewer res—

, pondents. Almost two-fifths (37 percent) of all respondents had adopted
no innovations at all (see Table 8). As indicated in Table 1, similar

percentages of respondents adopting the innovations were found in

 

both Phase One and Phase Two of the study, with one or two notable ex—

ceptions.

 

Table 8: Percentage of Respondents Adopting
Agricultural Innovations.

 

a
Number of Innovations Adopted Percentage of Respondents

 

 

0 37
l 19
2 l4
3 10
4 7
5
6
7
8
9

l-‘MMQU’I

or more

 

100

aPercentages based on N = 1,347 (see footnote, Table 5).

 

1See Chapter II for the definition of divisible innovations.

   

61

 

Table 9: Percentage of Respondents Adopting Each of the
14 Agricultural Innovations (Phase Two).

 

Innovation Percentage Adoptinga
Fertilizer 39
NS-l Maize 32
Aldrin Dust 23
Poultry 19
Oil Palm Rehabilitation 17
Community Plantations l3
Citrus 10
Rice 8
Improved Cassava 7
Vegetable Seeds 7
Cocoa Planting Scheme 5
Rubber Planting Scheme 4
Cashew 3
FAID Credit Program 2

aPercentages, which are based on N = 1,347, sum to more than
100 percent, since many respondents adopted more than one innovation
(see footnote, Table 5)°

 

 

62

Analysis of Data

Three statistical methods were utilized in the analysis of data
in this study:

(1) Zero-order correlation.

(2) Higher-order partial correlation.

(3) Multiple correlation.

First, zero-order correlations were computed between‘innovative-
ness and all other independent variables (both individual and system).
Then, higher-order partial correlations between innovativeness and each
of the independent variables were computed, holding constant all other
independent variables. To specifically determine the contribution of
the system variables, partial correlations were computed for the system
level independent variables, with the individual level effects held
constant.

After submitting the system variables outlined previously to both
zero-order and higher-order partial correlational analysis, we selected
the best predictors of innovativeness for inclusion in a multiple cor-
relation analysis. Similarly, the best predictors from the zero-order
and partial correlation analysis of the individual level variables were
included in this further analysis. Thus, the best predictors from the
system variables analysis, and the best predictors from the individual
level analysis, were submitted to multiple correlation analysis, using
the multiple regression least squares delete computer program, to for—
mulate a paradigm of variables that predict innovativeness and possess

significant explanatory power.

 

63

Operationalization of Variables

Dependent Variables

In recent years, attempts have been made to devise indices of
innovativeness which will accurately measure respondents' behavior
concerning the adoption of recommended innovations.l

In operationalizing the index of innovativeness utilized in
the present study, a measure was chosen which indicated the total
number of innovations<Iout of 14) adopted by the respondents, weighted
by the year it was first used. Each respondent was asked: "When
did you first (plant) (join) (use) (buy) the oil palm rehabilitation
scheme...the agricultural cocoa scheme...the agricultural rubber scheme...

etc." The answers were originally coded as:

 

lDasgupta (1968) compared and evaluated five indices of adoption
of recommended farm practices in terms of their ability to predict adOp—
tion behavior of farmers. The five indices represent a continuum from
a relatively simple index to a methodologically sophisticated one. They
included:

Percent of Applicable Practices Adopted. The practices adopted
by the farmer were first added and then divided by the number of practices
which were applicable on his farm. The figure so obtained was then
multiplied by 100. The farmers were then classified into five categories
according to their adoption scores.

Years of Use of AdOpted Applicable Practices. An adoption score
was derived for each farmer by adding the number of years he had used
each of the applicable practices. The figure so obtained was then
divided by the number of practices applicable to his farm. The farmers
were then classified into groups according to their adOption scores
(Bose and Dasgupta, 1962, p. 14).

Innovativeness Scale. The mean and the standard deviation of the
distribution of adoption over time are used to divide the distributions
into five areas, labelled as innovators, early adopters, early majority,
late majority, and laggards (Rogers, 1958).

Adoption Scale (Trace Line Analysis). Through the examination
of trace lines on a two-dimensional line graph, attempts were made to
select a number of recommended farm practices which were consistently
related to the latent variable continuum of adoption behavior for the

 

 

 

 

 

64

00 = never used

nn = last two digits of year first used

99 = not applicable, has not heard of innovations.

In order to give higher ratings to earlier adopters, these answers

were subsequently receded, according to the following pattern:

1967 = 00
66 = 01
65 = 02
64 = 03
63 = 04
62 = 05
61 = 06
60 = 07

The new codes therefore may be interpreted as follows:

00 = never uSed any innovation.
01 = one adopted in 1966.
91 : all 14 adopted in 1960 or before.

 

purpose of constructing an adoption scale. Farmers were given scores
on the basis of the number of these six practices each of them had adopted.

Adoption Scale (Guttman Scalogram Analysis). The seven prac-
tices examined by trace line analysis were also submitted to Guttman
scalogram analysis. After eliminating tw0 practices which did not scale
well, a minimally satisfactory scale was produced, with a coefficient
of reproducibility of .90, the errors being randomly distributed.

Of the five indices examined by Dasgupta, none were found to be
far superior to others, although two--years of use of applicable prac-
tices, and the adoption scale constructed with the aid of trace line
analysis and subsequently tested by Guttman scalogram analysis-—were
of more value in predictive ability.

 

 

65

Independent Variables

 

The following operationalizations were utilized to measure the
17 independent variables included in this study:

(1) Cosmopoliteness was operationalized as a total score ranging
from 0 to 40 for number of visits outside the village. These visits
included towns within the Eastern Region, cities within the Eastern
Region, cities outside the Eastern Region, other towns outside Eastern
Region, and places outside Nigeria. Each category was weighted from 1
to 4, based on distance from the respondents' village, to yield a total
possible score of 40.

(2) Education was operationalized as the re3pondents' score on
a scale running from O to 6, indicating he had never attended school,
attended primary (incomplete), attended primary (complete), attended
secondary (incomplete), attended secondary (complete), attended university
(incomplete), or attended university (complete).

(3) Familism was operationalized as the respondents' score on a
scale running from 0 to 4, indicating loyalty and orientation toward
members of his family. Respondents were asked two questions: "If-you
had enough money to start a trade, but a land dispute arose between your
family and another, what would you do with the money?" and, "If you
had saved enough money to build a zinc house but a brilliant relative
needed money for schooling, what would you do?" Respondent's answers
were coded so as to give higher scores for the latter alternative

in each question, to indicate a higher degree of familism.

 

66

(4) Educational aspiration was operationalized as respondents'

 

answer to a question regarding what level of education they would want
to reach, if they could be young again. Possible answers included none,
primary, secondary, or university, coded 0, l, 2, or 3, respectively,

so that a higher score indicated higher aspiration.

(5) Achievement motivation was operationalized as respondents'

 

score for a series of three questions designed to measure the degree to

 

which occupational and economic successwerevalued. The questions were:
"Do you think that work should come first, even if it implies lack of
rest?" "Do you think the best way to judge a man is by his success in
his occupation?" and,"Do you think the most important qualities of a
real man are determination and driving ambition?" Scores ranged from

0 to 6, with a high score indicating high achievement motivation.

(6) Empathy was operationalized as the respondents' score on a
scale ranging from 0 to 9, indicating answers to the following three
questions: "How could the newspaper help farmers in this village to
progress?" "If electricity were brought to this village, how would your
life change?" and, "How could the cinema be used to make this village
more civilized?" Specific scores for each question were assigned,
ranging from O to 3, depending on whether respénse was (0) don't know,
or no response, (1) vague or unrealistic, (2) specific relationship
but no direct improvement, and (3) specific relationship plus direct
improvement.

(7) Knowledge of Agricultural Change Agents was operationalized

 

as respondent's score on a question indicating whether he knew any

 

67

agricultural workers who come to his village. Possible responses were:
No, or incorrect, and Yes, coded O or 1, respectively.

(8) Economic Aspiration was operationalized as respondent's score
on a scale ranging from O to 9, indicating various responses to the
question: "If you had all the money you needed, what would you do with’
it?" Answers were coded in the 9 categories according to the degree of
desire for economic improvement exemplified, with higher scores indica-
ting higher degrees of economic aspiration.

(9) Literacy was operationalized as respondent's score on a
functional reading test. Respondent was asked to read a card containing
35 English words: "He who cannot read is like a blind person who is
being led lest he goes astray. He is dependent upon others. The book
which he cannot read mocks him as a slave to ignorance." The number of
incorrect words was recorded, and respondent was scored from 00 to 35,
according to the number of words known. A higher score indicated a
higher degree of literacy.

(10) Newspaper EXposurewaSoperationalized as respondent's score
on an index ranging from 00 to nn, indicating the actual number of
papers read, or read to him, during the previous four weeks.

(11) Radio Exposure was operationalized as respondent's score on
an index ranging from 00 to 14, indicating the number of days he had
listened to the radio during the previous two weeks. The score represen-
ted the actual number of days, with 14 as the maximum.

(12) Film Exposure was operationalized as respondent's score on

an index ranging from 0 to 9, indicating the number of films he had seen

 

68

during 1966. Answers were coded as follows: 0 = none; 1 to 8 = actual
number; and 9 = nine or more films seen. Thus a higher score indicated
higher film exposure.

(13) Agricultural Media Exposure was operationalized as respondents'

 

score, based on a total number of exposures to various types of agri-
cultural media messages during 1966. Possible scores ranged from 00 to
12, with answers coded as follows: 0 = No, l = don't know, 2 = Yes. Six
types of agricultural messages included: Read agriculture newsletter

or pamphlet, saw an agricultural demonstration, heardan agricultural radio
program, saw an agricultural film, attended an agricultural lecture,
talked with an agricultural officer. Thus, a higher score indicated
higher exposure to agricultural media.

(lu) Credit Orientation was operationalized as respondent's score
on an index ranging from 0 to 9 indicating the source of credit he would
contact to raise money to improve the farm. Answers were coded so as
to give higher scores for individuals who indicated greater orientation
to sources external to the family and close friends.

(15) Level of Living was operationalized as respondents' score on
an index, based on the total number of selected personal possessions the
respondents owned. Answers were coded O for No, and l for Yes, indica-
ting if they owned a radio, a wristwatch, cushioned chairs, laced shoes,
a cupboard, an iron bed, a clock, a bicycle, framed photographs, and a
1966 calendar. Scores ranged from O to 9, indicating the actual number

of these possessions.

 

69

(16) Correct Agricultural Knowledge was operationalized as respon-

 

dents' score on an index ranging from 00 to 28, indicating whether
respondents had correct information about each of the 14 innovations.
Respondents were asked, "Do you know anything about..." and interviewers
judged the correctness of the response by comparing against a checklist
carried to the interview. Respondents were scored as either 0 for no
correct information, or 2 for correct information. A previous index had
recorded the simple fact of knowledge or lack of knowledge about the
programs.

(17) Formal Social Participation was operationalized as respon-

 

dent's score on an index indicating the number of groups in the village
in which the respondent held memberships. A list of eight categories

of village groups, plus "others" was used as the basis for the clas-
sification, and scores ranged from 00 to nn, indicating the actual number
of groups belonged to.

In each of the foregoing questions, and all others asked during
the interview, confidentiality and a friendly and courteous manner were
used by the interviewer. In addition, the interviewers were instructed
to seek specific answers for each question, to indicate by appropriate
coding whether the respondent was qualified to answer the question, and

to record every answer given.

Construction of Variables

 

As outlined earlier, two levels of modernization characteristics

were included in the present analysis as independent variables:

 

70

(1) Individual level variables, constructed from each respondent's

 

score on each of the 17 variables on which data were obtained.

(2) System level variables, constructed from the mean value of

 

each of the 17 variables within the villages included in the study. Each
of the 18 villages in the study thus had 17 different scores, represen-
ting the mean value for each of the 17 independent variables. Each
respondent in each village was assigned a score based on the mean level
of each of the 17 independent variables in his village, for inclusion in
the correlational analyses as a system variable. For example, each indi-
vidual's innovativeness was correlated with his own individual level

of achievement motivation, as well as with the mean level of achievement

motivation in his village.

 

CHAPTER IV
RESEARCH FINDINGS
Correlates of Innovativeness

The research design of the present study calls for the selection
of certain modernization characteristics of individuals, to be submitted
to zero-order, partial, and multiple correlation analyses, with a view
to developing a paradigm of variables that best predict innovativeness.

As outlined in the guiding hypotheses (Chapter II), three basic
propositions are examined:

(1) That selected individual modernization characteristics of
the respondents are positively related to innovativeness.

(2) That the distribution of these individual characteristics
throughout the social system (herein called system level variables)
are positively related to innovativeness.

(3) That the interaction of these two levels of independent
variables is positively related to innovativeness, and the consideration
of both levels of variables simultaneously explains more variance in
farmers' innovativeness than do the individual level variables alone.

Let us now examine each of these propositions in the light of

various data analyses conducted in order to examine them.

Individual Characteristics and Innovativeness

 

Zero-Order Correlations. As indicated in Table 10, 15 of the

 

17 individual level variables included in the analysis had zero-order

7l

 

72

correlations which were statistically significant from zero at the .05
level of probability. Of the 15 statistically significant correlations,
correct agricultural knowledge had the highest zero-order correlations
with innovativeness (.53), followed closely by agricultural media expo—
sure (.52). The individual respondent's education (.35), level of
living (.38), social participation (.35), knowledge of agricultural
workers (.3”), literacy (.3”), and newspaper and radio exposure (.32
and .31, respectively) were all highly correlated with innovativeness.
The individual level of cosmopoliteness (.26), educational aspi—
ration (.23), empathy (.29), economic aspiration (.12), and film expo-
sure (.2H) were also significantly correlated with innovativeness.
Familism was negatively related to innovativeness, as expected, but
the correlation barely reached significance at the .05 level of pro-
bability (-.O6). Neither achievement motivation nor credit orientation
were significantly related to innovativeness at the individual level.

Partial Correlations. Partial correlational analysis is a

 

technique whereby the correlation between a dependent variable and
an independent variable can be determined, while holding constant one
or more other independent variables. Such a technique statistically
eliminates the intercorrelation between the independent variables, and
results in a "cleaner" measure of the relationship between the dependent
and the independent variables.

Two types of partial correlation were computed in the present

analysis:

 

73

 

Table 10: Zero-Order Correlations between Individual Level
Variables, System Level Variables, and Innovative-

 

 

 

 

 

ness (N = 1,142).
Zero-Order Correlation Correlation of
with Innovativeness Individual Vari-
IndiVidual Level System Level able with
of Independent of Indepen- System Variable
Variable Variable dent Variable
Cosmopoliteness .26a .03b .33
Education .35a .luab .31
Familism -.06a .iaab .39
Educational Aspiration .23a .Ogab .29
Achievement Motivation .03 .06a .22
b

Empathy .29a .lGa .31
Knowledge of Agricultural .3ua .lla;b .43

Workers
Economic Aspiration .12a .11a .1u
Literacy .3ua .13ab .23
Newspaper Exposure .32a .llab .21
Radio Exposure .Sla .OBab .20
Film Exposure .2ua .17ab .27
Agricultural Media .52a .2zab .33

Exposure
Credit Orientation —.O2 -.05 .31
Level of Living .38a .lBab .33
Correct Agricultural .53a .28ab .52

Knowledge
Formal Social Participation .35a .18ab .58

 

a
Significantly different from zero at the .05 level of probability.
bSignificantly different from individual level variables at .05 level
of probability.

 

 

74

(l) First-order partial correlation, in which the correlation be-
tween innovativeness and individual characteristics was computed, while
holding constant (partialling out) the system characteristics. This was
computed for each variable.

(2) Sixteenth-order partial correlations, in which the correlations
between innovativeness and a given independent variable was computed,
while holding constant (partialling out) the other independent variables
included in the analysis. This also was computed for each variable.

Let us look first at the first—order partial correlations. As
indicated in Table 11, all but three of the 17 individual variables were
significant at the .05 level of probability, when system variables were
partialled out. The three non-significant correlations were for familism
(.01), achievement motivation (.02), and credit orientation (-.01).

Among the significant correlations, the correlations for agricultural media
exposure (.ug) and correct agricultural knowledge (.u7) were markedly
higher than all others. Level of living (.36), education (.33), knowledge
of agricultural workers (.33), literacy (.32), and social participation
(.31) had correlations with innovativeness above .30. Cosmopoliteness
(.27), empathy (.26), educational aspiration (.22), film exposure (.20)

had correlations with innovativeness above .20, and economic aspiration

was correlated with innovativeness at .11.

Sixteenth-order partial correlations with innovativeness were
also computed for the individual level variables, and the results are
shown in Table 12. In this analysis, the correlation between each of

the individual level variables and innovativeness was computed, while

 

75

 

Table ll: First Order Partial Correlations between Individual
Level Variables, System Level Variables, and Inno-

vativeness (N = 1,1u2).

 

Individual Variable
and Innovativeness,
Holding System

System Variable and
Innovativeness, Hold-
ing Individual
Variables Constant

 

Variable variables Constant
Cosmopoliteness .27a
Education .33a
Familism .01
Educational Aspiration .22a
Achievement Motivation .02
Empathy .26a
Knowledge of Agricultural .33a
Workers
Economic Aspiration .lla
Literacy .32a
Newspaper Exposure .31a
Radio Exposure .30a
Film Exposure .20a
Agricultural Media .uea
Exposure
Credit Orientation -.01
Level of Living .36a
Correct Agricultural .473
Knowledge
Formal Social Participation .31a

 

-.06a
.04
-.17a
.03
.05

.08a

“.04

.09a
.063
.05
.02
.11al

. 06a

-.05
.Ol

.01

—.O3

aSignificantly different from zero at the .05 level of probability.

 

 

 

76

holding constant all other individual level variables. As indicated in
Table 12, only six of the individual level variables were significantly
correlated with innovativeness when their intercorrelations with all
other individual variables was mathematically removed. As was true
with first-order partial correlation analysis (see Table ll), correct
agricultural knowledge (.26) and agricultural media exposure (.22) were
most highly correlated with innovativeness. Formal social participa-
tion (.17) and newspaper exposure (.1u) were both rather highly cor-
related with innovativeness, while education_(.06) barely reached
significance at the .05 level of probability. Contrary to expectations,
credit orientation (-.10) was negatively correlated with innovativeness.

Multiple Correlation. Multiple correlational analysis is a

 

technique whereby the contribution to the prediction of the dependent
variable by a number of independent variables when considered toget-
her is computed. In this particular case, we computed the contribution
made by all individual level variables to innovativeness. After
deleting 11 variables which did not reach the minimum significance
criterion, six variables taken together produced a multiple correla-
tion coefficient of .62 (Table 12), accounting for 38 percent of the

variance in innovativeness.

System Characteristics and Innovativeness

 

As discussed in Chapter III, the mean level of each of the in-
dependent variables was taken as an indicator of the distribution of

the selected modernization characteristics throughout the social

 

77

 

Table 12: Sixteenth-Order Partial Correlations of Six Individual
Level Variables with Innovativeness, after Deletion of
11 Variables Not Reaching Minimum Significancea (N = 1,1u2).

 

 

Partial
Variable Correlation Coefficients'
(lBth-Order)
Education .06
Newspaper Exposure .14
Agricultural Media Exposure .22
Credit Orientation -.lO
Correct Agricultural Knowledge .26
Formal Social Participation .17
Multiple Correlation _
. . - .62
Coefficient
Coefficient f
° = .38

Determination (R2)

 

aSignificance criterion established at .05 level of probability.

 

 

78

system. Let us now examine the relationship of these system variables
to innovativeness.)

Zero—order Correlations. As indicated in Table 10, all of the
zero—order correlations for the system variables and innovativeness were
statistically significant, except two. Cosmopoliteness (.03) was not
significantly correlated with innovativeness at the system level, nor
was credit orientation (—.05).

Most highly correlated with innovativeness were correct agricul-
tural knowledge (.28) agricultural media exposure (.22). Familism, as
expected, was negatively and significantly correlated with innovative-
ness (-.18), and social participation (.18), film exposure (.17),
empathy (.16), education (.14), literacy (.13), and level of living
(.13) were all positively and significantly correlated with innovative-
ness at the system level. Knowledge of agricultural workers, economic
aspiration, and newspaper exposure were all significantly correlated
(.11) with innovativeness at the system level. Educational aspiration
(.09), radio exposure (.08), and achievement motivation (.06) were
also correlated positively with innovativeness.

Partial Correlations. Both first—order partial correlations and

 

higher-order partial correlations were computed for the system level
variables in the analysis. Let us consider first-order partial cor-
relations first in our discussion.

As indicated in Table 11, when system variables were correlated
with innovativeness, holding individual level variables constant, seven

of the 17 variables were statistically significant at the .05 level.

 

79

Familism (-.l7), empathy (.08), economic aspiration (.09), literacy
(.06), film exposure (.11), and agricultural media exposure (.06) were
statistically significant at the .05 level of probability, and in the
expected direction. Cosmopoliteness (-.06), however, was negatively
correlated with innovativeness, contrary to expectation, and was
statistically significant.

Although film exposure and agricultural media exposure were both
significantly correlated with innovativness at the system level, news—
paper exposure (.05), radio exposure (.02), and knowledge of agricultural
workers (-.04) were not significantly correlated with innovativeness,
contrary to expectations.

With the exception of familism (-.17) and film exposure (.11),
all the first-order partial correlation coefficients for the system
level variables were quite low.

As indicated in Table 13, 33rd-order partial correlations were
computed for the system variables, and the individual variables. TWenty
of these higher-order partial correlations were significantly correlated
with innovativeness at the .05 level, but seven were negatively cor-
related with innovativeness. Because of the large number of independent
variables (34) utilized in the analysis, and the high degree of inter-
correlation between them, these negative higher order partial correlation
coefficients are difficult to interpret with confidence.

Comparison of Individual and System Variables

 

When one examines the zero-order correlations of both the indivi-

dual and system variables (see Table 10), it is seen that 15 of the 17

 

80

 

Table 13: Thirty—Third—Order Partial Correlations of 34 Inde-
pendent Variables with Innovativeness (N = 1,142).

 

Partial Correlation Coefficient

 

 

Variable Individual Level System Level
Cosmopoliteness .00 -.lSa
Education ’ .05 -.19a
Familism . oo . 183
Educational Aspiration .03 —.03
Achievement Motivation -.04 .09a
Empathy -.08a -.19a
Knowledge of Agricultural Workers .06a .03
Economic Aspiration .04 —.20a
Literacy .04 .18a
Newspaper Exposure .11a -.04
Radio Exposure -.02 .20a
Film Exposure -.O2 .123
Agricultural Media Exposure .22a -.02
Credit Orientation -. 06a ., 21a
Level of Living .02 .20a
Correct Agricultural Knowledge .20a .17a
Formal Social Participation .17a -.19a

aSignificantly different from zero at the .05 level of probability.

 

 

81

variables at each level were significantly correlated with innovativeness.
However, they were not the same variables in both cases.

Cosmopoliteness' correlation with innovativeness was .26 at the
individual level, and .03 at the system level, and was not statistically
significant at the system level. On the other hand, the correlation of
achievement motivation with innovativeness increased from .03 at the
individual level to .06 at the system level, and was barely significant
at the system level. Credit orientation, which was not significant at
the individual level,was also not significant at the system level. With

four exceptions, all zero-order correlations with innovativeness were

 

significantly lower at the system level than at the individual level.

 

Sixteen of the zero-order correlations between the individual and
system variables and innovativeness were significantly higher than the
partial correlations with the individual and system level effects
partialled out (see Table 13). This was to be expected, if in fact the
system level variables exerted any influence upon the individual's in-
novativeness, as hypothesized. For example, the zero—order correlation
between the individual level of film exposure and innovativeness was .24,
whereas the partial correlation between film exposure and innovativeness,
with the system level of literacy held constant, was .20, indicating per-
haps that the system level of this variable was contributing somewhat to
the zero-order correlation at the individual level. A similar result was
observed for formal social participation (.35 for zero-order correlation
at the individual level versus .31 with the system level partialled out),

and correct agricultural knowledge (.53 versus .47).

 

 

82

 

Table 14: Comparison of Zero-Order and First-Order Partial Cor-
relations for Both Individual and System Variables and
Innovativeness ( N = 1,142 ).

 

 

 

 

Zero-Order Correlation With First-Order Partial Correla-
Variable _fi Innovativeness tion With Innovativeness
Individual Level System Level Individual Level System Level

Cosmopoliteness .26 .03 . .27 -.OSa

Education .35 .14 .33 .04a

Familism -.06 -.18 .Ol -.17

Educational As- .23 .09 .22 .03a
piration

Achievement Mo- .03 .06 .02 .05
tivation

Empathy .29 .16 .26 .08a

Knowledge of Ag- .34 .11 .33 --.04a
ricultural Workers

Economic Aspira- .12 .ll .11 .09
tion

Literacy .34 .13 .32 .06a

Newspaper Exposure .32 .ll .31 .05a

Radio Exposure .31 .17 .30 .02a

Film Exposure .24 .17 .20a .11a

Agricultural Media .52 022 .49 .06a
Exposure

Credit Orientation -.02 -.05 -.01 -.05

Level of Living .38 .13 .36 .01a

Correct Agri- .53 .28 .47a .01
cultural Knowledge

Formal Social .35 .18 .31a -.03a
Participation

 

aSignificantly different from zero-order correlations at .05 level of
probability, as indicated by t test for significance of difference between
two correlation coefficients, with 1139 degrees of freedom. For this test
of significance, the first-order partial correlation coefficients were
treated as zero order correlation coefficients. An alternative method of
comparison is to randomly divide the sample into two groups, and compute
the zero-order correlations from one group and the first order partial
correlation from the other. For large samples, however, (N greater than
300), the difference between the two methods is negligible.

 

 

83

Although system level film exppsure and system level agricul- V
tural media exposure were both significantly correlated with innova-
tiveness when individual level variables were partialled out (Table 11),
system level newspaper exposure and system level radio exposure were
not significantly correlated with innovativeness. All four of these
variables were highly related to innovativeness at the individual level,
however, suggesting that the primary relation of these variables

with innovativeness is at the individual level, not at the system level.

 

Variables which were statistically significant at the individual
level, but not at the systmnlevel,when first order partial correlations
were computed (see Table 11), were education (.33 at the individual
level versus .04 at the system level), educational aspiration (.22 versus
.03), newspaper exposure (.31 versus .05), radio exposure (.32 versus
.02), knowledge of agricultural workers (.33 versus -.04), level of
living (.36 versus .01), correct agricultural knowledge (.47 versus .01),
and social participation (.31 versus -.03). Fourteen variables were
significantly correlated with innovativeness at the individual level
when system level variables were held constant, but only seven vari-
ables were Significantly correlated with innovativeness at the system
level when individual level variables were held constant. In eneral,
then, system level variables were less highly correlated with inngvative—

ness than were individual level variables.

 

 

84

Joint Influences of Individual and System Variables

In order to determine the joint influence of individual and system
variables to innovativeness, two types of analysis were performed:

(1) A comparison of zero-order and first-order partial correla-
tions for both individual and system variables and innovativeness.

(2) First-order multiple correlations of individual level and
system level of each variable, and higher-order multiple correlations

of all individual and system variables which met the minimum signifi-

 

cance criterion of .05 level of probability.

Let us consider the first type of analysis. As indicated in
Table 14, the first-order partial correlations of each variable with
innovativeness were all lower than the zero-order correlations of each
variable with innovativeness. The only exception to this generalization
was the correlation of cosmopoliteness with innovativeness, which was
higher in correlation with innovativeness when first-order partial cor-
relation was computed than when zero-order correlation was computed.

As indicated in Table 11, cosmopoliteness was negatively correlated
with innovativeness at the system level when the individual level of
cosm0politeness was held constant.

If indeed the individual and system level of each variable inter-
act to increase the correlation of that variable with innovativeness,
the partialling out of either the individual level or the system level
of that variable should decrease the correlation with innovativeness.

With the exception of cosmopoliteness, this was the case for each of

 

85

the 17 variables. However, the zero-order correlations of the indivi-
dual variables with innovativeness was decreased less by partialling
out the system variables, than the zero—order correlations of the
system variables were decreased by partialling out the individual
variables, indicating that the system variables explained less of the
variance of innovativeness than did the individual characteristics of
the respondents.

Let us now consider the first-order multiple correlations. As
indicated in Table 15, multiple correlational analysis was performed for
each variable, taking both the individual level and system level of
that variable into account. Only seven of the 17 variables had higher
first-order multiple correlation coefficients than zero-order correla-
tion coefficients at the individual level; moreover, the increase was
quite small in almost every case. The first-order multiple correla-
tions, therefore, did not clearly indicate an interaction effect between
the individual level and system level variables.

The results of higher—order multiple correlation analysis were,
however, somewhat more marked and easier to interpret. In describing
how the higher-order multiple correlation coefficient was determined,
consideration must be given to the role of highest-order partial correla-
tion coefficients, and the least squares delete computer program.

Highest-order partial correlation is a special type of partial
correlation in which the correlation between the dependent variable and
the independent variable under consideration are determined, while hold—

ing constant all other variables in the analysis. In the present

 

86

 

Table 15: First-Order Multiple Correlations for Each Variable,
When Both Individual Level and System Level for Each
Variable Are Taken into Account, in Predicting
Innovativeness (N = 1,142).

 

 

 

Zero-Order Correlation Multiple
with Iggovativeness Correlations
Individual System Level with
Level of of Independent Innovativeness
Independent Variable
Variable Variable
Cosmopoliteness .26 .03 .27a
Education .35 .14 .35
Familism -.os -.18 .13al
Educational .23 .09 .23
Aspiration
Achievement .03 .06 .06a
Motivation
Empathy .29 .16 .30a
Knowledge of .34 .11 .34
Agricultural Workers
Economic .12 .11 .15a
Aspiration
Literacy .34 .13 .34
Newspaper .32 .ll .32
Exposure
Radio Exposure .31 .08 .31
Film Exposure .24 .17 .26a
Agricultural Media .52 .22 .52

Exposure

 

 

87

Table 15—-Continued.

Zero-Order Correlation
with Innovativeness

 

Individual System Level

 

Level of of Independent
Independent Variable
Variable Variable
Credit —.02 -.05
Orientation
Level of .38 .13
Living
Correct Agri— .53 .28
cultural Know-
ledge
Formal Social .35 .18
Participation

Multiple
Correlations
with
Innovativeness

. 05a

.38

.53

.35

aVariables in which multiple correlation for both the individual
level and system level shows some increase over the zero-order individual

level.

 

 

88

analysis, a total of 34 variables were used in the analysis, so that
the partial correlation was of the 33rd order.

Two types of highest-order partial correlation were utilized in
the analysis of the data:

(1) Highest-order partial correlations for all 34 variables.
Such correlations include both significant and non-significant variables,
and are not of primary interest in the present study. Results of

this analysis are shown in Table 13.

 

(2) Highest-order partial correlations for the 23 variables re-
maining in the analysis, after 11 variables had been deleted because
they did not reach the minimum statistical significance criterion.
Results of this analysis are shown in Table 16.

Deletion of the non-significant variables from the analysis
occurred in the following manner. The data were submitted to the
computer for analysis by the "least squares delete" computer program,
with the stipulation that all variables that did not reach a minimum
significance level of .05 should be deleted from the analysis. Of
the 34 variables submitted to computer analysis (including both indi-
vidual and system level variables), 11 were deleted. The 11 variables
in the order deleted included:

Cosmopoliteness

Familism

System level agricultural media exposure
Radio exposure

Level of living

Film exposure

Literacy

System level educational aspiration
Educational aspiration

Achievement motivation
Economic aspiration.

O O O

0 0 O

Lomxlmm-thI-J
e

O

F‘F‘
F‘O
0 O

 

89

Thus, a total of 23 variables were retained in the analysis as
statistically significant after the highest-order partial correlation
analysis, and were combined into a multiple correlational analysis.
The multiple correlation coefficient for these 23 variables was .68,
with a coefficient of determination of .47 (see Table 17).

In order to determine how much additional variance could be
explained by taking both individual level and system level variables
into account, a multiple correlation coefficient was computed for
individual level variables only. After deleting the individual level
variables which did not meet the minimum significance criterion of
.05, six variables (see Table 12) were retained in the analysis, pro-
ducing a multiple correlation coefficient of .62, and a coefficient
of determination of .38.

When the multiple correlation coefficient for the individual
level plus system level variables is compared with the multiple cor-
relation coefficient for the individual level only, it is seen that
the former was somewhat higher (.68 as compared with .62). Thus,
taking both individual level variables and system level variables
into account in the analysis produced a multiple correlation coef-
ficient which accounts for 47 percent of the variance in innovative—
ness, whereas the coefficient for the individual level variables only

explained 38 percent of the variance in innovativeness (Table 17).

Confirmation of Hypotheses

 

Although this analysis was not designed to rigorously test

hypotheses, three rather general guiding hypotheses were set forth

 

90

 

Table 16: Twenty-Second Order Partial Correlations of 23 Inde-
pendent Variables with Innovativeness, After Deletion
of 11 Variables by Least Squares Delete Program.a
(N = 1,142).

 

Variable Partial Correlation Coefficients

 

I. Individual Level)

Education .09
Empathv -.07
Knowledge of Agricultural Workers .07
Newspaper Exposure .12
Agricultural Media Exposure .24
Credit Orientation -.07
Correct Agricultural Knowledge .20
Formal Social Participation .17

II. System Level

Cosmopoliteness -.24
Education -.21
Familism .22
Achievement Motivation .22
Empathy -.22
Knowledge of Agricultural Workers .13
Economic Aspiration -.23
Literacy .19
Newspaper Exposure -.09
Radio Exposure .23
Film Exposure .19
Credit Orientation .21
Level of Living .24
Correct Agricultural Knowledge .23
Formal Social Participation -.23

aSignificance criterion established at .05 level of probability.
Thus, all of these variables are statistically significant.

 

 

 

91

 

 

Table 17: Comparison of Multiple Correlation Coefficients and
Coefficients of Determination of Tndividual Level
Variables Only, System Level Variables and Individual
Levelwand System Level Variables Considered Simultaneously
(N a 1,142).
Level of Multiple Coefficient
Variable Correlation of
Considered Coefficient Determination
Individual Level Only .62 .38
System Level Only .40 .16
Individual and System Level .68a .47

Considered Simultaneously

aSignificantly different from either individual level only or system
level only at .05 level of probability, as indicated by t test for
significance of difference between two correlation coefficients, with
1139 degrees of freedom.

 

92

in Chapter II. Perhaps a consideration of these hypotheses in relation
to the analytical results can bring the goals and conclusions of this
study into sharper focus.
H1 The individual level of innovativeness for
Nigerian Villagers ls pOSitively related
to the individual level of selected moderni-
zation characteristics.
As described earlier in this chapter, three analytical techniques
were utilized to test this hypothesis:
(1) Zero-order correlations
(2) First-order and highest—order partial correlations
(3) Multiple correlation
With the exception of two variables (achievement motivation and
credit orientation), the zero-order correlations of the selected indi-
vidual level variables to innovativeness were positive and significant.
All but three of the individual level variables (achievement motivation,
credit orientation, and familism) were positively and significantly
related to innovativeness when first-order partial correlations were com-
puted, but only six variables were significant when highest-order partial
correlations were computed, probably because of the high intercorrelations
between the independent variables. The multiple correlation coefficient
for the six significant individual level variables was .62.
With the minor exceptions here noted, the first guidigg_hypothesis
was generally confirmed.
H2 The individual level of innovativeness for
Nigerian villagers is positively related to

the system level of selected modernization
characteristics within each village.

 

93

As described earlier in this chapter, two analytical techniques
were utilized to test this hypothesis:

(1) Zero-order correlations

(2) First-order and highest-order partial correlations.

All the system level variables were somewhat lower in zero-order
correlation with innovativeness, than the individual level variables.
Fifteen of the system level variables were statistically significantly
correlated with innovativeness, however, and all but one
were in the hypothesized direction. When first—order partial correla-
tions were Icomputed, seven of the 17 system variables were significantly
correlated with innovativeness. When higher—order partial correlations
were computed for the system variables, seven were statistically
significant but in a negative direction. The high level of intercorrela-
tions among the independent variables, and the large numer included in
the present analysis, make the results of partial correlations analysis
somewhat difficult to interpret.

We conclude, therefore, that the second guiding hypothesis was

only partially confirmed.

 

H3 The individual level of innovativeness for
Nigerian villagers is positively related to
the joint influence of the individual level
and the system level of selected moderniza—
tion characteristics within each village.
The primary analytic technique utilized in testing this hypothesis
was multiple correlation. First-order multiple correlations were computed,

utilizing individual and system levels for each variable, but the results

were not clear-cut. Including the 23 individual and system level variables

 

94

which were statistically significant in highest-order partial analysis

into a multiple correlation, however, increased the correlation coefficient
to .68, accounting for 47 percent of the variance in innovativeness.

This is compared with a multiple correlation for individual variables of
.62, accounting for 38 percent of the variance in innovativeness.

We conclude, therefore, that the thirdgguiding hypothesis is

 

generally confirmed, although the additional amount of variance in

 

innovativeness explained by taking system level variables into account

as well as individual level variables was not very great.

 

Analysis for Isolating System Effects

 

As set forth in Chapter II, the present study was designed to
make the following determinations for system effects:

(1) To investigate whether system level variables contribute
significantly to the amount of variance explained in the innovativeness
of Nigerian farmers.

(2) To investigate whether the individual level variables desc-
ribing individual characteristics of Nigerian farmers are mediated by
the system level variables in their influence on innovativeness.

Both of these research questions were answered in the affirmative,
with certain reservations and exceptions. Generally, the following two
statistical techniques gave strongest support for these two conclusions:

(1) Partial correlations, in which the correlation between system
level variables and innovativeness was computed, while holding constant

the individual level variables. Of the 17 pairs of variables for which

 

95

this partial correlation was computed, seven were significantly correlated
with innovativeness at the .05 level. Of these seven significant variables,

six were correlated in the hypothesized direction, as follows:

Familism Negative

Empathy Positive

Economic Aspiration Positive

Literacy Positive

Film Exposure Positive

Agricultural Media Positive
Exposure

Cosmopoliteness, however, with a correlation of 4.06, was signifi-

 

cantly but negatively related to innovativeness at the system level.

(2) Multiple correlations, in which the amount of variance explained
by utilizing both individual and system level variables was compared with
the amount of variance explained by utilizing individual level variables
only. The consideration of both levels of variables simultaneously in-
creased the amount of variance explained in innovativeness by 9 percent
(from 38% to 47%).

The system level variables seem to exert mediational influence

 

upon the individual level variables in three different ways:
(1) Augmenting their correlation with innovativeness, as
indicated by:

(a) The increase over zero-order correlation obtained by
computing a multiple correlation coefficient incorporating both indi-
vidual and system variables, for each variable.

(b) The increase in the amount of variance explained in

innovativeness by incorporating both individual and system level

   

96

variables, for all variables in the analysis, in a multiple correla-
tion coefficient, when compared with the amount of variance‘explained
by incorporating only individual level variables in a multiple cor-
relation coefficient.

(c) The decrease in correlation between innovativeness and
individual level variables, when system level variables were partialled
out.

(2) Decreasing their correlation with innovativeness, as indicated
by the increase in correlation of individual level cosmopoliteness with
innovativeness, when system level cosmopoliteness was partialled out.

As indicated earlier, the partial correlation (first-order) for system
level cosmopoliteness and innovativeness, with individual level cosmo—
politeness held constant, was -.06. To interpret this negative
correlation verbally, the higher the mean level of cosmopoliteness in

an individual's village, the lower will his individual innovativeness be.
Thus, system level cosmopoliteness exerted a depressing influence upon
the individual farmer's innovativeness.

(3) Serving as an alternative to individual correlates of innova—
tiveness. In the case of familism, for example, the correlation of
individual level familism when system level familism was held constant,
was practically nil (.01). The system level correlation of familism,
however, with individual familism held constant, was highly significant
(-.17). Further, the multiple correlation coefficient obtained by

taking both individual and system level familism into account was .18,

 

97

as contrasted with -.06 for individual level familism only. This

would seem to indicate that an individual farmer's familism matters
little in predicting innovativeness, so long as his village is low
in familism. Thus, system level familism serves as an alternative
to individual level familism, in predicting innovativeness, in some

instances.

 

 

 

CHAPTER V
SUMMARY AND CONCLUSIONS

Summary

The present study was designed to explore the relations between
individual modernizing characteristics of Nigerian farmers, the corres-
ponding system characteristics, computed as the average level of those
characteristics in each of the villages where the respondents lived,
and the individuals' innovativeness. Of primary concern was the inter-
action between the two levels of modernizing characteristics in ex-
plaining greater amounts of variance in innovativeness.

A portion of the data gathered in the three-nation Diffusion
Project conducted by the Department of Communication, Michigan State
University, and sponsored by the United States Agency for International
Development, were analyzed. The Diffusion Project was conducted in
three phases: Phase One was an analysis of the relative success or
failure of various programs of change in agricultural production in
some 80 villages in each country. The unit of analysis was the
village, and data were secured primarily from interviews with village
leaders and change agents. Phase Two was based on data obtained
through personal interviews with individuals living in some 20 vil-

lages in each of the three countries. Respondents were male household

98

 

99

heads. Phase Three was designed as a field experiment using selected

communication techniques and economic incentives. Hewever, Phase

Three was curtailed in Eastern Nigeria after its initiation because of

civil strife and political disturbances in the country. Thus, the first

two phases of the Nigerian Diffusion Project were utilized in the

present study, with the major portion of the data coming from Phase Two.
From the 71 villages studied in Phase One, 18 villages were

chosen as the basis for the Phase Two study. The sample was comprised

 

of 1,142 respondents, who were in each case a male, the head of his
household, at least 20 years old, and farming some amount of land.
The sampling rate within each village was approximately 57 percent.
Fourteen agricultural innovations were selected for inclusion as
innovations inthe study. They included: fertilizer, NS—l maize, Aldrin
dust, poultry, oil palm rehabilitation, community plantations, citrus,
rice, improved cassava, vegetable seeds, cocoa planting schemes,
rubber planting scheme, cashew, and FAID agricultural credit. Both
knowledge and adoption of these innovations were generally quite low,
with 50 percent of the respondents knowing five innovations or less,
and 37 percent of the respondents adopting no innovations at all.
Seventeen cognitive, attitudinal, and overt behavioral charac-
teristics of individuals were utilized as independent variables, in-.
cluding cosmopoliteness, education, familism, educational aspiration,
empathy, knowledge of agricultural workers, economic aspiration,
literacy, newspaper exposure, radio exposure, film exposure, agricul-

tural media exposure, credit orientation, level of living, correct

 

 

100

agricultural knowledge, formal social participation, and achievement
motivation. Each of these 17 variables was aggregated through a
measure of central tendency (mean) and utilized as an independent
variable at the system level. I

Three guiding hypotheses were selected as the focus of the
analysis: (1) that the individual level of innovativeness for the
respondents is positively related to the individual level of selected

modernization characteristics, (2) that the individual level of in-

 

novativeness is positively related to the system level of selected
modernization characteristics, and (3) that the individual level of
innovativeness is positively related to the joint influence of the
individual and system levels of selected modernization characteris—
tics.

Three Statistical methods were utilized in the analysis of
the data{(l) zero-order correlation, (2) first-order and higher-
order partial correlations, and (3) multiple correlation.

First, zero-order correlations were computed between innova-
tiveness and each independent variable, both individual and system.
Then, higher-order partial correlations between innovativeness and
each of the independent variables were computed, holding constant
all other independent variables. To specifically determine the
contribution of the system variables, partial correlations were
computed for each of the system variables, with the individual level

variables held constant.

 

101

After submitting the independent variables to both zero-order
and higher-order partial correlation analysis, the best predictors
of innovativeness (the variables most highly correlated with innova-
tiveness) were submitted toes multiple correlation analysis. Both
individual level and system level variables were included in the
multiple correlation analysis. Thus, the best predictors from the
system variables and the best predictors from the individual
variables were included in the multiple correlation analysis, using
the multiple regression least Squares delete computer program, to
formulate a paradigm of variables that predict innovativeness, and
have significant explanatory power.

When zero-order correlations were computed between the inde-
pendent variables and innovativeness, 15 of the 17 individual level
variables were significantly correlated with innovativeness. These
15 variables were cosmopoliteness, education, familism, educational
aspiration, empathy, knowledge of agricultural workers, economic
aspiration, literacy, newspaper exposure, radio exposure, film
exposure, agricultural media exposure, level of living, correct
agricultural knowledge, and formal social participation. Also, 15
of the 17 (not the same 15) system level variables were significantly
correlated with innovativeness when zero-order correlations were
computed. These 15 variables were education, familism, educational
aspiration, achievement motivation, empathy, knowledge of agricultural
workers, economic aspiration, literacy, newspaper exposure, level of

living, correct agricultural knowledge, and formal social participation.

‘r

   

102

Most highly correlated with innovativeness at the individual level were
correct agricultural knowledge (.53) and agricultural media exposure
(.52). Most highly correlated with innovativeness among the system
level variables were correct agricultural knowledge (.28) and
agricultural media exposure (.22). In general, individual level
variables were more highly correlated with innovativeness than system
level variables.

First-order partial correlations were computed separately for
individual level and system level variables with innovativeness.
First, individual variables were correlated with innovativeness, with
system variables held constant. All but three of the individual level
variables were significantly correlated with innovativeness, with cor-
relations ranging from .49 for agricultural media exposure to .11 for
economic aspiration (the lowest significant correlation). Next,
system level variables were correlated with innovativeness, with
individual level variables held constant. Only 7 of the 17 system
variables reached statistical significance in this analysis. Thus,
fewer system level variables were significantly correlated with in-
novativeness, when individual level variables were held constant,
and correlations were generally considerably lower for system level
variables than for individual level variables.

Highest-order partial correlations were then computed, to
select variables from both individual and system levels most highly
correlated with innovativeness, to be included in a multiple cor—

relation analysis. Of the 34 independent variables submitted to the

   

103

least squares delete computer program, 11 were deleted when they did
not reach the minimum statistical significance criterion of .05 level
of probability. Thus, a total of 23 variables were retained in the
analysis as statistically significant after the highest-order partial
correlation analysis, and these were combined in a multiple correla—
tion analysis. The multiple correlation coefficient for these 23
variables was .68, with a coefficient of determination of .47.

When the individual level variables only were submitted to
highest order partial correlation analysis, only six were retained

as statistically significant° When entered into a multiple

cQEESIQFiPD .analysis, these six individual level_ variables '

—~_

produced a multiple correlation coefficient of .62, and a coef-

ficient of determination of .38.

Conclusions

 

On the basis of the present analysis, several conclusions
regarding the relation to innovativeness of individual modernizing
characteristics, system modernizing characteristics, and their
interaction, seem to be possible.

1. The individual modernization characteristics selected

 

for inclusion as independent variables in this study were generally

highly correlated with the individual Nigerian farmer's innovative-

 

ness. As suggested by the paradigm (Figure l) in Chapter II, indi-

 

vidual characteristics of respondents have been frequently investigated

for relationships to innovativeness, and to a considerable extent

 

 

104

the findings reported herein agree with many previous studies utiliz—
ing many of these same individual level variables.

Worthy of note, however, are three variables included in the
present study which were not significantly correlated with innova-
tiveness at the individual level: familism, achievement motivation,
and credit orientation. Possibly these modernizing characteristics
were not relevant individual variables in the Nigerian context, or,

as discussed later in this chapter, the failure of these variables

 

to be significantly correlated with innovativeness in this study
may have been due to measurement error.

2. The system level modernizing characteristics selected for

 

inclusion as independent variables in this study were related to

 

some degree to the innovativeness of the individual Nigerian farmers.

In general, however, they were not nearly so strongly related to

 

innovativeness as the individual level variables.

 

Familism, which was not significantly correlated with innova-
tiveness at the individual level, was significantly correlated with
innovativeness at the system level, indicating that an individual
farmer may be highly innovative if the general level of familism in
his social system is low, even if his individual level of familism
is quite low.

System level cosmopoliteness, which was not significantly
correlated with innovativeness in the zero-order correlation analysis,
and negatively and significantly correlated with innovativeness in

the partial correlation analyses, is of considerable interest. Since

 

105

this characteristic is strongly related to innovativeness at the indi-
vidual level, and either non-significant or negatively correlated at
the system level, one might conclude that an individual farmer is
likely to be innevative when he is individually highly cosmopolite,
but not likely to be highly innovative when the social system in which
he lives is highly cosmopolite. Perhaps the relation of cosmopoliteness
to innovativeness is curvilinear, so that after some undetermined
point of inflection on the curve, further increments of cosmopoliteness
act as a suppressant to individual innovativeness.l

Credit orientation, which was not significantly correlated
with innovativeness at the individual level, was not significantly
correlated with innovativeness at the system level either. Apparently,
this characteristic was a poor predictor of innovativeness in the
Nigerian context.

3. Taking both individual level variables and system level

 

variableg into account in a multiple correlational analysis in-

cre§§es the amgunt of vgrignce egplaingg in inngyatiyengss, so that

some interaction effects occurr between individggl level epd system

 

level independent variables. However, the amount of variance ex-
plained by individual level variables only was far greater than the
amount of variance explained by system level variables only. The
increase in amount of variance explained by taking both levels of

independent variables into account amounts only to 9 percent.

See Appendix A for analysis of curvilinearity.

 

106

Therefore, taking both levels of independent variables into account
can increase the ability to predict individual innovativeness, but

not greatly.

Limitations of the Study

Although several comments regarding weaknesses in sampling
procedure, measurement error, etcg, appear in earlier chapters of
this dissertation, several limitations of the present investigation
need specific mention°

l° The usual problems of field research (sampling, inter-
viewer selection and training, biased responses, lack of information
regarding conditions preceding introduction of innovations, problems
of causality inference, etco) are considerably magnified when the
site of the study is in a less developed country such as Eastern
Nigeriao Language difficulties, low homophily between researchers
and respondents, danger of bias because of economic, philosophic, or
political considerations, etco, were all present during the conduct
of this study. Consequently, even though special efforts were made
to overcome or counteract these difficulties in conducting the study,
less than perfect confidence exists regarding the degree of success
achieved in this effort.

2. Caution should be exercised in generalizing the results
of this investigation to populations other than that included in
the study° Although Eastern Nigeria farmers share many problems

and characteristics with farmers in other nations of the world,

 

107

several factors make the subjects of this investigation somewhat
unique, including the extremely low state of modernization exist-
ing generally in the country, the peculiar clique and kinship
ties existing within the social systems, and the tension and anéiéiy
characterizing the nation because of extremely unsettled civil and
political conditions at the time of the investigation°

3. A most important limitation of the present study concerns
the rather narrow focus on a very small portion of the total social
system and its environment, In looking for correlates of innovative—
ness within the individual farmer himself and his environment, we
chose only a very few of the many social, psychological, and overt
behavioral characteristics that could be examined in relationship to
innovativeness. A myriad of influences--political, economic,
philosophical, etc°-—surround the individual farmer, even in an
extremely underdeveloped nation such as Nigeria, which relate either
negatively or positively to his innovativenesso As explained in
Chapter II, this present analysis was not designed as an investiga-
tion of the total social system, in all its hundreds of facets, but
we chose instead to focus specifically on a relatively small number
of attitudinal, cognitive, and behavioral characteristics which have
been found to be related to innovativeness at the individual level
in previous investigationso We attempted to determine the relation of
these variables to individual innovativeness at both the individual

and system level, and to determine if their interaction explains any

 

108

more variance in innovativeness than either taken alone. Such an
approach is highly specific, and the results should not be generalized
beyond the implications of these research findings.

u. A clear weakness of the present study is that no direct
measurements were taken of the interpersonal communication processes
within the villages, although several of the independent variables
included in the analyses clearly exert their influence through the
interpersonal communication network. Intuitively, one would expect
the mean level of the independent variables within the villages
(herein called system variables) to affect the individual farmers
through the interaction among the individuals who live within the
village. Yet, no direct measurements were taken of these inter‘
personal communication activities, but they were merely assumed.

In future studies of a similar nature, more useful results could
probably be obtained by directly measuring these dyadic rela-
tionships within the social system, and including them as another
variable for consideration.

5. In the preliminary discussions regarding the present
study, several variables from Phase One of the Diffusion Project
were considered for inclusion. These variables included number
of visits to the village by the agricultural extension agent,
concentration of opinion leadership within the village, consensus
among the change agents regarding the most important problems
facing the village at the time of the survey, and other theoretically

promising variables. However, logistical problems connected with

 

   

109

including these variables in the analysis were prohibitive, so they
were dropped from consideration. The problems included differences

in number of respondents in the two phases, difference in systems of
coding used in the two phases, difference in the form submitted to

the computer center for analysis, etc. We can only regret that it

was not possible to include these variables, since the present results
would probably have been more rewarding by their inclusion.

6. An analysis for curvilinearity computed for the relation-
ship of each of the 34 independent variables to innovativeness in-
dicated that many of the relationships were highly curvilinear.
Since a product-moment correlation computed between variables in
which the relationship is actually curvilinear will always be less
than the true relationship, the findings of the present study rep-
resent an under-measurement of the actual relationship between
these independent variables and innovativeness. Thus, higher zero-
order correlations could have been obtained had the analyses been
based on the correlation ratio, instead of the product-moment cor-

relation coefficient.

Implications for Change Agent Strategy
What implications does the present investigation have for
change agencies seeking to introduce innovations and modernizing tech-
nology into a less developed country such as Nigeria? Several sug-
gestions seem to be warranted.
1. Although the evidence for system effects and interaction

effects is not overwhelmingly powerful in the present investigation,

 

 

110

the findings do suggest that an individual farmer's innovativeness is
influenced by something more than individual characteristics alone.

A change agency desirous of introducing an innovation into a given
social system may therefore expect to find the adoption of that in-
novation affected by characteristics of the social system as well as
of the individual receiver. The system characteristics should there-
fore be taken into account in the mapping of a strategy for change.

2. One of the concepts used by development theorists is
"climate for development," usually defined as a concentration of
social and psychological conditions within a social system which are
conducive to modernization and development. The system effects found
in the present study would seem to corroborate the value of taking
such a concept into account in strategy planning. If the change agency
desires to introduce innovations into a target social system, the
system effects within that social system can reinforce and augment
the process of modernization initiated by the individual moderniza-
tion characteristic of the individuals. Since the mean level of a
given modernization characteristics within a social system is a
function of the number of individuals within that social system who
possess that characteristic, the system variables which are positively

related to innovativeness can serve as a "multiplier" for social change.

Implications for Future Research
The present form of analysis could well be expanded into other

fruitful areas of research.

 

 

111

(1) One such area that might be investigated would be the
determination of the relation between some selected characteristic of
an individual (.e.g, innovativeness) and the level of selected in-
dependent variables within some type of social system other than a
village. Human beings interact with, and are members of, various
social systems, including the family group, the larger kinship group,
cliques within the village or community, and perhaps special interest
groups or reference groups. One might hypothesize, for instance,
that individuals are influenced to adopt or not adopt various innova-
tions by their interaction with members of the clique of which they
are members. What happens in such situations when the clique members
have belief or attitudes which are antagonistic to the norms of the
larger social system? Would an individual be more innovative if he
were a member of a special interest group in his community if that
special interest group were more innovative than the average member
of the larger social system? Such questions might reveal previously
unstudied relationships between sub‘systems and the larger social
system, as well as between individuals within the sub-system itself.

(2) The amount of variance in innovativeness explained in the
present study might have been increased by selection of a more
promising set of independent variables. The variables utilized in the
present investigation were chosen because of the conceptual relevance
to innovativeness, as indicated by previous research. However, the
list of possible independent variables was by no means exhaustive.

If this study should be repeated, higher correlations with individual

   

 

112

innovativeness might be produced by such system variables as:

The presence or absence of institutions in
the village

The number of social cliques in the village

The population of the village

The geographical dispersion of the village
population

The presence or absence of communications media

An index of usage of communications media

Distance from nearest urban center

An index of accessibility (paved roads,
remoteness, etc.)

An index of commercial development

One advantage of including system variables such as these, in
addition to or in place of aggregated individual level variables, is
that such non-aggregated variables are more easily manipulable. Ag-
gregated individual level variables arelfil definition a function of the
level of these variables on the part of individuals, so that little
can be done, at least in the short run, to alter the level of these
variables at the system level. On the other hand, a variable such
as the presence or absence Of an educational institution can be
manipulated, since such a facility could be physically erected, so that
its impact upon the village or other social system could be measured
with some rigor.

(3) The relationship of many of the independent variables in
the present analysis to the communication process has been implied,
rather than operationalized and measured. The correct knowledge of
agricultural innovations within the system, for example, undoubtedly

has impact upon individual innovativeness because of the communication

process in operation within the system. However, no measure of such

 

113

interpersonal communication processes was included in the present
analysis. If system variables have influence upon an individual’s
innovativeness, as the present study indicates, then a fruitful
opportunity for future research lies in tapping the communication
networks between individuals within the system, to determine Egg.
the system variables exert their influence. What are the communica-
tion processes that bring an individual into contact with system
level achievement motivation, for instance? Or again, what are the
communication linkages between the deviant and other members of
the social system (e.g., a low innovative individual in a high
innovative social system, or a high innovative individual in a low
innovative social system)? Such questions were not considered in
the present analysis, and offer an opportunity for productive
future research.

(4) The present investigation hopefully will encourage
other investigations utilizing system variables in addition to
individual variables in their analyses. Although the investigation
of the influence of individual characteristics is not nearly com-
pleted, perhaps equally or even more fruitful results could be
obtained by expanding the scope of diffusion research to include more
system analyses, at least to some extent. One of the notable
lecunae in previous diffusion research has been an extremely narrow
focus upon the individual and his characteristics, with hardly any
attention to the individual's relationship to the social system of

which he is a part. Both reason and research, however, indicate

 

 

11”

that the individual is markedly influenced by his environment. The
present study investigated individual and aggregated social, psy-
chological, and behavioral characteristics in analyzing innovativeness.
Perhaps even more enlightenment could be gained by including
variables chosen from the economic, religious, political, philosophical,
or other realms.

(5) A clear opportunity exists for further investigation of the

joint influence of individual and system level variables upon innova-

 

tiveness, using an even larger number of villages or social systems
as the locus of the study. Added generalizability could be gained by
conducting corroborative or similar research under similar research
conditions, but in different social settings (perhaps another less
developed country). Are the findings of the present study confined
only to less deveIOped countries, or do they hold true for more
developed countries as well? Could more reliable and precise results
be obtained by including a larger number of social systems in the
analysis?

Perhaps partial answers to the question regarding the general-
izability of the findings of the present study are already being com-
piled. In a contemporary study conducted by Saxena (1968) in the
Department of Communication, Michigan State University, of the influence
of system variables upon agricultural innovativeness in India, results
very similar to those of the present study were obtained. Saxena
investigated the relation of 15 individual and aggregated modernization

characteristics to the individual innovativeness of residents of

 

115

eight Indian villages, and found that the amount of variance in inno-
vativeness explained by taking both individual and system level
variables into account was lu percent greater than including individual
level variables only.

Saxena also extended the analysis of system variables by inves-
tigating the relevance of Newcomb's theory than an individual will
tend to maintain minimal discrepancy between his own attitude toward
an innovation and that of the system of which he is a part, depending
upon the valence that is jointly attributed to change by the individual
and by the system.

Such an investigation represents one extension of the present
analysis. Others dealing with balance theories (i.e., cognitive
dissonance, congruity, etc.) are possible, and offer opportunities for
potentially rewarding research.

Thus, with this initial entry of system variables into diffusion
research, perhaps the analytical techniques described by Durkheim a
century ago in sociological investigations can increase our understand-
ing of the relevance of system effects and produce even more imposing

results than those already compiled.

   

BIBLIOGRAPHY

Aurabac, Herbert A. (1955) "A Guttman Scale for Measuring Isolation."
Rural Sociology, 20, pp. 1u2—u5.

Becker, Selwyn A. and Frank Stafford (1967), "Some Determinants of
Organizational Success," Journal of Business, #0, pp. 511-18.

Becker, Selwyn A. and Thomas A. Whisler (1967), "The Innovative
Organization: A Selective View of Current Theory and Research."
Journal of Business, 40, pp. 462-69.

Berelson, Bernard, Paul F. Lazarsfeld, and W.N. McPhee (195u), Voting:
A Study of Opinion Formation in a Presidential Campaign.
Chicago, Illinois: University of Chicago Press.

 

Blalock, Hubert M., Jr. Causal Inferences in Non—Experimental Research.
Chapel Hill, N.C.: University of North Carolina Press, 1961.

 

 

Blau, Peter M. (1957), ”Formal Organization: Dimensions of Analysis."
American Journal of Sociology, 63, p. 6M.

 

Blau, Peter M. (1960), "Structural Effects.” American Sociological
Review, 25, No. 2, pp. 178—93.

 

Bose, Santi Pryia and Dasgupta, Satadal (1962), The Adoption Process.
Calcutta, West Bengal Department of Agriculture, Socio-Agro—
Economic Research Organization, Extension Bulletin No. 1.

 

Campbell, Ernest Q. and C. Normal Alexander (1965), "Structural
Effects and Interpersonal Relationships." American Journal
of Sociology, 71, pp. 284—89.

Carroll, Jean (1967), "A Note on Departmental Autonomy and Innovation
in Medical Schools." Journal of Business, 40, pp. ”31—34.

 

Cattell, R.B. (1951), "Types of Group Characteristics," in Harold
Guetzkow (ed.), Groups, Leadership, and Men. Pittsburgh:
Carnegie Press. pp. 16-52.

 

Coleman, James S. (1964) Introduction of Mathematical Sociology.
Glencoe, Illinois: The Free Press.

 

116

 

 

117

Coughenour, C. Milton (l96u), "The Rate of Technological Diffusion
Among Locality Groups." American Journal of Sociology, 69,
pp. 325—39.

 

Dasgupta. Satadal, (1968), "Relative Predictability of Five Indices
of Adoption of Recommended Farm Practices." ‘Sociologia
Ruralis, 8: 1—22.

Davis, James A., Joe L. Spaeth, and Carolyn Huson (1961), "A Tech-
nique for Analyzing the Effects of Group Composition."
American Sociological Review, 26, No. 2, pp. 215-25.

 

Deutschmann, Paul J. and Fals Borda, Orlando (1962). Communication
and Adoption Patterns in an Andean Village. San Jose, Costa
Rica, Programma Interamericano de Information Popular.

 

 

Duncan, James A. and Burton W. Kreitlow (195M), "Selected Cultural
Characteristics and the Acceptance of Educational Programs
and Practices." Rural Sociology, 19, pp. 3N9-57.

Durkheim, Emile (1897), Suicide. (Translated by James A. Spaulding
and George Simpson, 1951), Glencoe, Illinois: The Free Press.

Durkheim, Emile, The Rules of Sociological Method. (Translated by
George Simpson, 1950), Glencoe, Illinois: The Free Press.

 

Economic Development Institute (1967), Phase I: Preliminary Report
of Selected Descriptive Findings for Leaders and Progressive
Farmers in 71 Eastern Nigerian Villages. USAID/MSU Diffusion
of Innovations Research Report, University of Nigeria, Enugu
Campus (Mimeo).

 

 

 

 

Economic Development Institute (1967), Phase II: Preliminary Report
of Selected Descriptive Findings for 13u7 Rural Farmers and
Innovators in 18 Eastern Nigeria Villages. USAID/MSU Diffusion
of Innovations Research Report No. 72, University of Nigeria,
Enugu Campus (Mimeo).

 

 

 

Evans, William A. and Guy N. Black, (1967), "Innovation in Business
Organizations: Some Factors Associated with Success or Failure
of Staff Proposals." The Journal of Business, #0, pp. 519-30.

 

Faris, R.E.L. and H.W. Dunham (1939), Mental Disorders in Urban Areas.
Chicago, Illinois: The University of Chicago Press, pp. 101-23.

 

Fliegel, Frederick C. (1957), "Farm Income and Adoption of Farm
Practices." Rural Sociology, 22, pp. 159—62.

’5

 

118

Forde, D. and G.I. Jones (1950), The Ibo and Ibibio-Spgaking Peoples
of Southeastern Nigeria. London: International African
Institute.

 

Frey, Frederick W. (1963), "Surveying Peasant Attitudes in Turkey."
Public Opinion Quarterly, Vol. 27: 335-355.

 

Garrett, Henry E. (1964), Statistics in Psychology and Education. New
York: David McKay Company, Inc.

Goldsen, Rose K. and Ralis, Max (1957), Factors Related to Acceptance
of Innovations in Bang Chan, Thailand. Ithaca, N.Y.: Cornell
University, Southeast Asia Program Data Paper 25.

 

 

 

Green, M.M. (1947), Ibo Village Affairs. London: Sidgwick and
Jackson, Ltd.

Groves, Ernest R. and William F. Ogburn (1928), American Marriage and
Family Relationships. New York: Henry Holt and Co., pp. 193-205.

 

 

Gusfield, Joseph R. (1967), "Tradition and Modernity: Misplaced
Polarities in the Study of Social Change," American Journal
of Sociology, 72, pp. 351-362.

Keith, Robert F. (1968), An Investigation of Information and Moderni-
zation Among Eastern Nigerian Farmers. East Lansing, Michigan:
Department of Communication, Michigan State University, Technical
Report No. 4.

 

 

Kendall, Patricia and Paul F. Lazarsfeld (1950), "Problems of Survey
Analysis," in Robert K. Merton and Paul F. Lazarsfeld, editors,
Continuities in Social Research: Studies in the Scope and
Method of 'The American Soldier.r Glencoe, Illinois: The Free
Press, pp. 195-96.

 

 

Lazarsfeld, Paul F. and Herbert Menzel (1961), "On the Relation Be—
tween Individual and Collective Properties," in A. Etzinoi
(ed.) Complex Organizations. New York: Holt, Rinehart,
and Winston, p. 425.

 

Lipset, S.M., M.A. Trow, and J.S. Coleman (1956) Union Democracy,
Glencoe, Illinois: The Free Press, pp. 163-171.

Loomis, Charles P. and J. Allan Beegle, (1964) Rural Sociology, The
Strategy of Change. Englewood Cliffs, N.J.: Prentice-Hall,
Inc.

 

 

119
McClelland, David C. (1961), The Achieving Society. New York: The
Free Press.

McNemar, Quinn, (1965), Psychological Statistics. New York: John
Wiley and Sons, Inc.

 

Marsh, C. Paul and A. Lee Coleman (1954), "The Relation of the
Neighborhood of Residence to Adoption of Recommended Farm
. .» _Practices." Rural Sociology, 19, pp. 385—90.

Merton, Robert K. (1957), Social Theory and Social Structure. New
York: The Free Press.

 

Neill, Ralph E. and Rogers, Everett M. (1963), "Measuring Achievement
Motivation Among Farmers," Columbus, Ohio: Agricultural
Experiment Station Departmental Series AB 346.

Niehoff, Arthur H. (1966), A Casebook of Social Change. Chicago:
’Aldine Publishing Company.

 

North Central Rural Sociology Subcommittee for the Study of the
Diffusion of Farm Practices (1961), Adopters of New Farm
Ideas: Characteristics and Communicative Behavior. East
Lansing, Michigan: Michigan State University, North Central
Regional Extension Bulletin No. 13.

Ottenberg, Simon. ds (1959), "Ibo Receptivity to Change," in M.
Herskovits and W. Bascom, Continuity and Change in African
Culture. Chicago: University of Chicago Press, pp. 130-43.

Qadir, Syed Abdul (1966), Adoption of Technological Change in the
Rural Philippines-~An Analysis of Compositional Effects. Ph.D.
Thesis, Cornell University. Published by University Microfilm,
Inc., Ann Arbor, Michigan.

Rahim, S.A. (1961), The Diffusion and Adoption of Agricultural Practices:
A Study in a Village in East Pakistan. Comilla, Pakistan
Academy for Village Development.

Riley, Matilda White (1963), Sociological Research: I. A Case
Approach. New York: Harcourt, Brace, and World, Inc.

Robinson, W.S. (1950), "Ecological Correlations and Behavior of
Individuals," American Sociological Review, 15, pp. 351-57.

 

Rogers, Everett M. (1962), The Diffusion of Innovations, New York:
The Free Press.

 

 

120

Rogers, Everett M. (1965), Revised Operational Plan for Research Project
on Diffusion of Innovations in Rural Societies. East Lansing,
Michigan: ’Michigan State University, Department of Communication
(Mimeo).

Rogers, EverettM . (1966), Mass Media Exposure and Modernization Among
Colombian Peasants." Public Opinion Quarterly, 29:614—625.

 

Rogers,Everett M. (1967), Bibliography on the Diffusion of Innovations,
East Lansing, Michigan: Michigan State University, Department
of Communication (Mimeo).

Rogers, Everett M. and Rabel J. Burdge, (1961), Community Norms,
Opinion Leadership, and Innovativeness Among Truck Growers,
Wooster, Ohio: The Agricultural Experiment Station Research
Bulletin No. 912. '

 

Rogers, Everett M. with F. Floyd Shoemaker (forthcoming) The Diffusion
of Innovations: A Cross-Cultural and Communication Approach.
Glencoe, N.Y.: The Free Press.

Rogers, Everett M. and David Stanfield (1966), "Adoption and Diffusion
of New Products: Emerging Generalizations and Hypotheses."
Department of Communication, Michigan State University, Mimeo—
graphed Departmental Publication.

Ryan, Bryce and Neal C. Gross, (1943), "The Diffusion of Hybrid Seed
Corn in Two Iowa Communities," Rural Sociology, 8:15-24.

Sapolsky, Harvey M. (1967), "Organizational Structure and Innovation,"
Journal of Business, 40, pp. 297-510.

Saxena, Anant, (1968), System Variables and Individual Innovativeness
in Indian Villages. Unpublished Ph.D. Dissertation, Department
of Communication, Michigan State University.

Scheuch, Erwin, (1965), "Cross—National Comparisons Using Aggregate
Data: Some Substantive and Methodological Problems," R.
Merrit and S. Rokkan (editors), in Comparing Nations. New
Haven: Yale University Press, pp. 143-167.

Schramm, Wilbur (1965), quoted in Tom Carroll and R. Vincent Farace,
System Analysis, Computer Simulation, and Survey Research:
Applications to Social Research in Developing Countries,
Computer Institute for Social Science Research, Michigan State
University.

Selvin, Hanan C. (1958), "Durkheim's Suicide and Problems of Empirical
Research," American Journal of Sociology, 63, pp. 607-19.

 

 

121

Selvin, Hanan C. and W.O. Hagstrom (1963), "The Empirical Classification
of Formal Group." American Sociological Review, 28, pp. 399-411.

 

Shepard, Herbert A. (1967), "Innovation-Resisting and Innovation-
Producing Organizations." Journal of Business, 40, pp. 470-77.

 

Sills, David L. (1961), "Three Climate of Opinion Studies," Public
Opinion Quarterly, 25, pp. 571-73.

Steiner, G.A (ed.) (1965), The Creative Organization. Chicago: The
University of Chicago Press.

Stouffer, S.A., et al. (1949), Studies in Social Psychology in World
War II, Vol. 1. The American Soldier During Army Life, Princeton,
New Jersey: Princeton University Press.

 

 

 

 

Tannenbaum, Arnold S. and Jerald G. Bachman (1964), "Structural Vs.
Individual Effects," American Journal of Sociology, 69, pp. 585-95.

 

Uchendu, Victor C. (1965), The Igbo of Southeastern Nigeria, New York:
Holt, Rinehart and Winston.

 

Valkonen, Tapani (1966), Individual and Structural Effects in Ecological
Research," Paper prepared for the Symposium on Quantitative
Ecological Analysis in the Social Sciences, Evian France.

Van den Ban, Anne Willem (1960), "Locality Group Differences in the
Adoption of New Farm Practices." Rural Sociology, 25, pp. 308-320.

. Webster“s New Collegiate Dictionary, (1958), Springfield,
Massachusetts: G.C. Merriam Co.

Yeracaris, C.A° ( ), "Social Factors Associated with Acceptance of
Medical Innovations: A Pilot Study." Unpublished Paper,
New York: University of Buffalo.

Young, Frank W. and Ruth C.Young (1962), "The Sequence and Direction
of Community Growth: A Cross Cultural Generalization." Rural
Sociology, 27, pp. 374-386.

Young, James N. and A. Lee Coleman (1959), "Neighborhood Norms and the
Adoption of Farm Practices." Rural Sociology, 24, pp. 372-83.

 

 

 

APPENDICES

 

APPENDIX A

Correcting Multiple R for Inflation

A multiple correlation coefficient, R.computed from a sample
always tends to be somewhat "inflated" with respect to the population
R, owing to the accumulation of chance errors which tend to pile up
since R is always taken as positive. The inflation of a multiple R
is most pronounced when N is small and the number of variables in the
problem quite large.

The multiple correlation coefficients obtained in the present
analysis are not likely highly inflated for the following reasons:

(1) The sampling rate within each of the villages studied was
quite high, ranging from 27 percent of the total eligible respondents
to a complete census (100 percent sample) in two of the villages
(see Table 4).

(2) The sample size used in the computation of the multiple
correlation coefficient was quite large (N = 1,142).

(3) Although a total of 34 independent variables was utilized
in one analysis, 23 in another, and 16 in another, the number of inde-
pendent variables was >nly a small proportion of the sample population,
in each analysis.

Formula for correcting an inflated multiple R: An obtained

 

multiple correlation coefficient can be corrected or "shrunken" to
give a better measure of the population R by the use of the following

formula:
123

 

124

2- -2M
Rc-lk (N—m)

where:
N = size of the sample
m = number of variables in the problem
(N - m) = degrees of freedom
2=(l-R2)
In the findings of the present investigation, R = .68, in the

analysis computed for 23 significant individual and system level

 

variables, N = 1,142, and m = 23 (not including 11 variables not reach-
ing the minimum significance criterion of .05); k2 = (l — .682) or

.53. Substituting these values into the formula, we have:

3 (1,141)

2
R = 1-.5
c (1,115?

which yields:

R2 = .46 and R = .678
C C

and the correction is negligible.

 

APPENDIX B

Alternative Indices of Innovativeness

During the course of the present investigation, we were concerned
with questions regarding the reliability and validity of the index of
innovativeness utilized in this study. These questions focused on
the "non-unidimensionality" of the index, as indicated by (l) the
seeming competitiveness of various innovations in the index with other
innovations, (2) various rationales apparently utilized by Nigerian
farmers for adopting or not adopting each of the innovations, and
(3) the wide disparity between innovations regarding agricultural skill
required, amount of investment required, and relative advantage for
the farmer.

At least two other indices of innovativeness have recently been
utilized in conducting similar analyses of these data: (1) a three-item
index, utilizing the most widely adopted innovations in the list:
fertilizer, Aldrin dust, and NS—l maize (Keith, 1968); and (2) a six-
item index, including cocoa, rice, cassava, vegetable seeds, poultry,
and NS-l maize, constructed by factor analysis of all fourteen innova-
tions, and selecting the "cleanest" factor which explained the largest
amount of variance in innovativeness (Salcedo, 1968).

These two additional indices of innovativeness were included
in the present analysis, for information purposes, with the results
shown in Table 18.

125

 

126

 

Table 18: Comparison of Zero-Order Correlations of 34
Individual and System Level Variables with
Three Indices of Innovativeness (N = 1,142)

 

II.

Variable
I. Individual
Cosmopoliteness
Education
Familism

Educational Aspiration

Empathy

Knowledge of Agricultural
Workers

Economic Aspiration

Literacy

Newspaper Exposure

Radio Exposure

Film Exposure

Agricultural Media Exposure

Credit Orientation

Correct Agricultural
Knowledge

Formal Social Participation

Achievement Motivation

System

Cosmopoliteness

Education

Familism

Educational Aspiration

Achievement Motivation

Empathy

Knowledge of Agricultural
Workers

Economic Aspiration

Literacy

Newspaper Exposure

Radio Exposure

Film Exposure

Agricultural Media Exposure

Fourteen
Item

Indexa

.53

.35
.03

.03
.14

.09
.06
.16
.11

.11
.13
.11
.08
.17
.22

Six Three

Item Item
Index Index
.24 .25
.32 .33
-.03 -.09
.18 .26
.25 .35
.29 .39
.10 .12
.30 .30
.32 .24
.27 .35
.22 .22
.43 .55
-.02 .04
.42 .57
.23 .29
-.03 .06
-.008 -.004
.04 .07
—.ll -.27
.08 .19
.04 .06
.09 .16
.l4 .16
.12 -.O2
.06 .02
.04 .02
.006 .09
.15 .06
.16 .21

 

127

Table l8--Continued.

_H, Fourteen Six Three
Variable' Item Item Item
Indexa Index Index
Credit Orientation -.05 -.06 -.05
Level of Living ' .13 .02 .05
Correct Agricultural .28 .22 .33
Knowledge ‘ ‘
Formal Social Participation .18 .08 .12

 

aThe fourteen-item index of innovativeness is the index utilized
in the present analysis.

 

Several explanatory comments are in order regarding these alter-
native indices of innovativeness:

(1) As indicated by Table 18, the fourteen-item index of innova-
tiveness and the three-item index of innovativeness yielded very similar
zero-order correlations, and the six-item index of innovativeness
yielded lower correlation coefficients in almost every instance.

(2) The six—item index of innovativeness was constructed from
a factor analysis of all 14 innovations, and was attended by some
degree of arbitrariness regarding the number of rotations to be
utilized, minimum level of loading to be taken into account in deter—
mining communalities, and amount of intercorrelation with items in
other factors to be tolerated.

In view of our desire to utilize as realistic an index of

innovativeness as possible, our decision was to utilize the l4—item

 

128

index of innovativeness, since its wider range of innovations offered
several obvious advantages. In our considered opinion, the use of
either of the other two indices here described would not have appre-
ciably improved the reliability of the findings, nor significantly
improved the coefficient of determination. Although less than per-
fect, the fourteen-item index is believed to have been the best choice
among the three alternatives, from a theoretical, statistical, and
practical point of view.

Table 19 shows the intercorrelations between the three alter-

native indices of innovativeness.

 

Table 19: Intercorrelations Between l4—Item,
6-Item, and 3-Item Indices of
Innovativeness (N = 1,142).

 

6-Item Index 3-Item Index
l4-Item Index .78 .69
6-Item Index ——— .48

3-Item Index -—- _--

 

 

APPENDIX C

Analysis of Curvilinearity

As outlined in Chapter II, one of the assumptions on which
the present analyses rest is linearity of the data. When the means
of the arrays of the successive columns and rows in a correlation
table follow straight lines (at least approximately), the regression
is said to be linear or straight-line. Regression lines which
"best fit" the means of the successive columns and rows in the table
can be calculated, and used as the basis for prediction of the
dependent variable.

When the drift or trend of the means of the arrays (rows or
columns) cannot be well described by a straight line, but can be
represented by a curve of some kind, the regression is said to be
curvilinear, or in general, nonlinear. When the regression is non-
linear, a curve joining the means of successive arrays will fit
these mean values more exactly than will a straight line. Hence,
should a truly curvilinear relationship be described by a straight
line, the scatter or spread of the paired values about the regression
line will be greater than the scatter about the better-fitting regres-
sion curve. The smaller the spread of the paired scores about the
regression line or the regression curve, the higher the relationship

between the two variables. For this reason, a zero-order product-

 

moment correlation calculated from a correlation table in which the

 

129

 

130

relationship is curvilinear will always be less than the true rela-
tionship. If the regression is significantly nonlinear, it makes
considerable difference whether the correlation ratio (eta) or the
product-moment correlation (r) is the measure of relationship. But
if the correlation is low and the regression is not significantly
curvilinear, theproduct—moment correlation (r) will give as adequate
a measure of relationship as the correlation ratio (eta).

Examining Table 20, we see that considerable differences exist
for every independent variable between the product-moment correlation
(r) and the correlation ratio (eta), indicating that the relationship
of many of these independent variables to innovativeness is highly
curvilinear. Since the relationships of these independent variables
to innovativeness is curvilinear, the correlations computed on the
assumption that the relationships are linear are not a satisfactory
basis for predicting innovativeness. The curvilinear relationships
also explain the unexpectedly low correlation coefficients we obtained
in our analyses, since the zero-order correlations were based on the

assumption of linearity.

 

131

 

Table 20: Comparison of Zero-Order Product-Moment
Correlation (r) and Coefficient of Non-
linear Relationship (n) for 34 Indivi-
dual and System Level Independent Variables

 

 

 

 

and Innovativeness (N = 1,142).
Zero—Order Correlation,of
Product-Moment Nonlinear
Correlation Relationship
(r) (eta)
Variable Individual System Individual System
Level Level Level Level
Cosmopoliteness .26 .03 .36 .18
Education .35 .14 .46 .26
Familism -006 -e 18 .21 .3”
Educational Aspiration .23 .09 .30 .26
Achievement Motivation .03 .06 .21 .24
Empathy .29 .16 .38 .21
Knowledge of Agricul- .34 .11 .43 .28
tural Workers
Economic Aspiration .12 .11 .24 .23
Literacy .34 .13 .43 .23
Newspaper Exposure .32 .11 .51 .22
Radio Exposure .31 .08 .41 .21
Film Exposure .24 .17 .38 .26
Agricultural Media .52 .22 .59 .33
Exposure
Credit Orientation -.O2 -.05 .22 .19
Level of Living .38 .13 .45 .20

 

132

Table 20--Continued.

 

 

 

Zero-Order Correlation of
Product-Moment Nonlinear
Correlation Relationship
(r) (eta)
Variable Individual System Individual System
Level Level Level Level
Correct Agricultural .53 .28 .63 .42
Knowledge
Formal Social .35 .18 .41 .26
Participation

 

 

APPENDIX D

 

Table 21: Mean Values and Standard Deviations for

17 Individual Level, 17 System Level
Variables (N = 1,142).a

 

Variable

Cosmopoliteness

Education

Familism

Educational Aspiration

Empathy

Knowledge of Agricultural
Workers

Economic Aspiration

Literacy

Newspaper Exposure

Radio Exposure

Film Exposure

Agricultural Media Exposure

Credit Orientation

Level of Living

Correct Agricultural
Knowledge

Formal Social Participation

Achievement Motivation

 

Individual System
Mean Standard Mean Standard
Value Deviation Value Deviation
3.07 3.57 3.07 1.19
0.55 0.77 0.55 0.24
0.63 0.96 0.63 0.37
2.33 0.81 2.33 0.24
3.98 2.51 3.98 0.77
0.31 0.46 0.31 0.20
2.43 1.48 2.43 0.21
6.66 13.01 6.66 2.98
1.28 4.46 1.28 0.93
2.66 4.53 2.66 0.92
0.76 1.49 0.76 0.40
3.04 3.36 3.04 1.09
3.20 2.17 3.20 0.67
2.58 2.75 2.58 0.90
11.32 6.39 11.32 3.30
2.65 1.81 2.65 1.05
5.09 1.44 5.09 0.20

aSee Chapter III for operationalizations of these variables.

 

133

 

 

 

 

 

 

 

 

"11111111111111111115