ER’lP.’-\THY. MASS MEDIA. AND MODERNEZATION:
A STUDY IN RURAL BRAZIL

Thesis fer the Degree of Ph. D.
MIGHEGAN STATE UNEVERSETY
GORDON. C. wz‘mms
1957

Imam.

0-169

This is to certify that the

thesis entitled

EI‘IPATHY, MASS MEDIA, AND MODERNIZATION:

A STUDY IN RURAL BRAZIL
presented by

Gordon C. Whiting

has been accepted towards fulfillment
of the requirements for

Ph.D. Communication

//7/\

// / LLfiz z.” /’ z /
" Major professor

degree in

.7 / .

W é? ‘2
Date,[;;'7:"{'si’i -. 417’ 1 > / c- '
r u v /

 

 

MATHY, MASS MEDIA, AND MODERNIZALI‘ION:

A STUDY IN RURAL BRAZIL

BY

Gordon C. Whiting

AN ABSTRACT OF A THESIS

Submitted to
Nflchigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Communication

1967

ABSTRACT

EMPATBY, MASS MEDIA, AND MODERNIZATION:
A STUDY IN RURAL BRAZIL

by Gordon C. Whiting

This study focuses on the process of modernization of individuals
in develOping nations and particularly on the role of empathy in this
process. Empathy is viewed as a capacity for discriminating subtle
cues and imaginatively and flexibly'utilizing these to optimize goal
attainment. Three traditional meanings of empathy are examined
and criticized. Two newer approaches are presented which tie empathy
more closely to the modernization process. The first approach suggests
that the more empathic individual develops a more differentiated
generalized-impersonal-other which helps him learn from and interact
with strangers. The second approach suggests empathy as an instance
of hypothetical thought or general cognitive flexibility. Other
links are discussed between empathy and modernization,and a model
is presented of empathy's contribution.

The focus of empirical investigation is narrowed to three
general areas of concern: (1) the nature of empathy as a variable,

i.e. its dimensionality, stability over time, relative importance

as a predictor, and the possibility of a new approach to its measure-
ment; (2) the place of more empathic individuals in communities

of greater or lesser traditionalism; (3) empathy's relations with
mass media and urban contact, especially whether aspects of empathy

mediate the impact of messages in these channels.

Gordon C. Whiting

Data were gathered in two waves of interviewing during 1966
in rural Minas Gerais, Brazil. Indices of variables were develoPed
through factor score procedures and through Guttman scaling. Tests
of hypotheses involved comparisons of appropriate correlations and/or
regression coefficients. The dimensionality of empathy was investi-
gated through factor analysis. Multiple regression routines helped
evaluate empathy's relative usefulness as a predictor.

Factor analysis of 18 empathy items produced three factors:
(1) interaction with strangers and role-knowledge, (2) skill with
hypotheticals, and (3) imaginativeness. When factor analyzed in a
larger pool with an additional 57 items, only components of the first
and last factor achieved satisfactory loadings. When.aggregated to a
community level, empathy loaded well on a factor which included other
indications of skill in interpersonal interaction.

As a predictor of modern attitudes and knowledge imaginativeness
was superior to hypotheticals; both were decidedly inferior to mass
media exposure, intelligence, and literacy. A three-item empathy
index composed of role-taking questions had a test-retest reliability
of .29 over six months. In terms of predictiveness, ease of coding,
and possibly reliability a better measure might be obtained from
interviewer ratings of respondents' performance on a variety of hypo-
thetical items.

Other results:

1. Empathy shows no differential relationship to the individual's
social role, integration, or communication position in communities of

greater or lesser traditionalism.

Gordon C. Whiting

2. Imaginativeness is correlated significantly more with print
exposure than with radio exposure.

3. Print exposure is correlated significantly'more with belief

in work, political knowledge, and attitude toward change than is
radio exposure.

A. The improvement in correlations among the preceding variables
Obtained by cross-classifying on imaginativeness is significantly
greater than the improvement obtained by cross-classifying on hypo-
theticals.

5. The pattern of regression slopes suggests that the more empathic
either increase the modernity of their attitudes and knowledge more
rapidly per unit exposure to mass media or start higher on these
variables when exposed to the media. If both lepes and intercepts
are allowed freedom, the resulting pattern sometimes suggests a take—
off phenomenon in which individuals in the middle levels of empathy
learn more rapidly than those at either extreme. This result is
consistent with the hypothesis that empathy facilitates decoding the
modernization import of the media.

6. There is no evidence that empathy facilitated attraction to the

city due to urban exposure.

EMPATHY, MASS MEDIA, AND MODERNIZATION:

A STUDY IN RURAL BRAZIL

By

Gordon 0"? Whiting

A.THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Communication

1967

{; Li‘BS'O Q

‘1

.3 ,7 ,555

©COPYRIGHTED BY

GORDON CARL WHITING

1968

ACKNOWLEDGEMENTS

The author wishes to acknowledge the assistance received in
the preparation of this thesis from the Department of Communication
and the Computer Center at Michigan State University, from.the
Project on the Diffusion of Innovations in Rural Societies, and from
the FEderal University of Minas Gerais, the Rural University of
Minas Gerais, and the Agency for Credit and Rural Assistance, Minas
Gerais. More particularly, Dr. Hideya Kumata, Dr. David K. Berlo,

Dr. Frederick B. Waisanen, Dr. Bruce L. Smith, Dr. Everett M; Rogers,
Dr. Dennis Gilliland, and Mr. William L. Ruble provided encouragement,

advice, and assistance of a unique nature.

11

CHAPTER
I

II

TABLE OF CONTENTS

IMROHJCTIONANDTHEORY .....

Modernity.
Lerner'sModel.......
Specifications of the Concept, "Empathy" . .
Some Earlier Conceptualizations of Empathy .
Ehnpathy in the Modernization Process . . . .
A Model of Empathy in the Modernization Process
Objectives of the Study . . . . . . . . . . . .
A Summary of the Hypotheses of the Study . . .

METHOIDLOGYW... .......

Beekground of the Study . . . . . . . . . . . . . .
Selection of the Sample . . . . . . . . . . . . . .
Instrument Construction, Interviewer Training,
Data Collection and Coding: Phase I . . . . . . .
Instrument Construction, Interviewer Training,
Data Collection and Coding: Phase II
Scaling and Index Construction . . . .
Analysis Procedures and Rationale . . .
Concluding Comments . . . . . . . . . .

III mum O O O O O O O O O O O O O O O O I O O O O O I

The Factor Analysis of Empathy and Other Measures .
The Reliability of Measures of Empathy Over Time .
The Feasibility of a New Measure of Empathy . . . .
The Empathic Person in More and Less Traditional
Comunities...................
The Relationship of Imaginativeness to Exposure to
RadioandExposuretoPrint. . . . . . . . . . .
Regression Tests on Mass Media Variables . . . . .
Regression on Trips to the City of Preference for
the City, With and Without Cross-Classification
onAspectsofEmpathy. . . . . . . . . . . . . .
Evidence on the Predictive Utility of Two Aspects

ofEmpathy...................

IV SW, DISCUSSION, AND CONCLUSIONS . . . . . . . .

Summary of Rationale and Procedures . . . . . .
Results: The Nature of Empathy . . . . . . . .
Empathy and Degree of Community Traditionalism
EmpathyandtheMassMedia . . . . . . . . . .
ConcludingComments. . . . . . . . . . . . . .

APPENDIX

Page

\J‘II-‘P

SSE—q
\

132
132+
137

137
139
150
151 w
158

160
163

10

12

13

LIST OF TABLES

Guttman scale of skill in hypotheticals on
Lerner-type empathy items: scoring'based
on interviewer Judgments . . . . . . . . . . . .

Guttman scale for skill in hypotheticals
having to do with economic items: scoring
based on interviewer Judgments . . . . . . . . .

Loadings and weights on the three-factor solution
for 18 items tapping aspects of empathy . . . .

"Thinking" or "imaginativeness" factor from.the
six-factor solution of a factor analysis of 75
items which include 18 empathy items . . . . . .

Factor loadings and items defining a community
measure of interpersonal skill and group harmony
among informal leaders at the community level. .

First factor from.a three-factor solution to a
factor analysis of 26 indices constructed for
testing hypotheses in this study . . . . . . . .

Second and third factor from a three-factor
solution to a factor analysis of indices
constructed for testing hypotheses in this
study . . . . . . . . . . . . . . . . . . . . .

Fourth factor from a four-factor solution to
a factor analysis of indices constructed for
testing hypotheses in this study . . . . . . . .

Test-retest reliability of 17 variables over a
Bil'lnonth PeriOd o o c c o o c c c c c o c o c 0

Rank order of deletion of three predictors of
six.mbdernization'behaviors . . . . . . . . . .

Rank order of deletion of six predictors of
six modernization behaviors . . . . . . . . . .

The prediction of adoption from three measures
of hypothetical skill . . . . . . . . . . . . .

Items and factor loadings from a factor analysis
of measures of community development . . . . . .

iv

Page

63

75

78

79

81

82

83

86

89

9O

91

93

TABLE Page

1h Correlations of three aspects of empathy with

seven measures of integration, status, and

influence in the community . . . . . . . . . . . 95
15 Two-way analysis of variance of the degree to which

cross-classification improves the predictive

power of mass media variables for three modern-
ization orientations, belief in work, political
knowledge, and positive attitude toward

change . . . . . . . . . . . . . . . . . . . . . 128

16 Relative adequacy of two aspects of empathy --
skill with hypotheticals and imaginativeness --
in competition with other predictors of eight
modernization orientations . . . . . . . . . . . 135

LIST OF FIGURES

FIGURE Page
1 A.mode1 of empathy in the modernization process . . 31
2 The hypothesized relationship between trips to

the city, urban attractiveness, and empathy . . . Ah

3 The structure of varimax rotated solutions for
18 variables intended to measure aspects of

apathy O O O O O O O O O O C O O O O O O O O O O 7 3

h Belief in work predicted by exposure to mass
media: data classified on five levels of
imaginativeness . . . . . . . . . . . . . . . . . 10h

5 Belief in work predicted'by exposure to radio:
data classified on imaginativeness, lepes
mac e qua]- O O O O O O O O O O O O O O O O O 0 0 107

6 Belief in work predicted by exposure to radio:
data classified on five levels of imaginative-
ness, intercepts made equal . . . . . . . . . . . 108

7 Belief in work predicted by mass media exposure:
data classified on five levels of skill with
hmthet i c 8-18 0 O O O O O O O O O O O O O O O O I no

8 Political knowledge predicted by mass media
exposure: data classified on five levels of
imaginativeness . . . . . . . . . . . . . . . . . 11A

9 Favorable attitude toward change predicted by
mass media exposure: data classified on five
levels of’imaginativeness . . . . . . . . . . . . 120

10 Favorable attitude toward change predicted by

exposure to radio: data classified on five

levels of imaginativeness . . . . . . . . . . . . 122
ll Favorable attitude toward change predicted by

mass media exposure: data classified on five
levels of skill with hypotheticals . . . . . . . 125

12 Favorable attitude toward change predicted by
exposure to radio: data classified on five
levels of skill with hypotheticals . . . . . . . 127

CHAPTER I

INTRODUCTION AND THEORY

One of the more fashionable areas of theorizing and research
in contemporary social science is that of social change. Most social
scientists interested in cross-cultural work pay considerable atten-
tion to social change leading toward the modernization of emerging
nations. The sociology of this interest would itself be interesting,
but is not the focus of present concern. The modernization and
development of nations is. In general, communication as it inter-
acts with characteristics of individuals is the focus here, not the
more frequently researched question of aggregates of communication
flow and their correlates with other indices of development. By
choice, the aspect of modernization selected for attention here is

modernization of the individual by "changing his mind."

I. MODERNITY

Smith and Inkeles point out that modernity has both a social-
system.and individual—system referents (Smith and Inkeles, 1966).
For the social system such characteristics as increased urbanization,
education, industrialization, and social mobility help to define
change toward modernity. Indices of this nature have been shown to
bear a high relationship to measures of communication flow (pieces
of mail processed, etc.) and to the presence of the channel mechanisms
needed for mediated communication (radio sets, cinema seats, etc.).
Social systems have also been called more modern to the extent that
they are more structurally differentiated, are capable of processing

1

larger quantities of information more rapidly and with greater fidel-
ity, or tend to utilize to a greater extent the criteria of achieve-
ment rather than ascription in recruitment to given roles. These
criteria of social~system modernity are drawn from different traditions
of analysis and only give the flavor of the social~system approach to

a definition. Inkeles summarizes this area by noting that the time-
orientation of modern societies is toward the future, that they give
great emphasis to adaptation and change, often at the expense of
consolidation and conservation of tradition, and, most strikingly,

that modern societies are energetically organized for continuous,

 

self-initiated change.

 

The second referent of modernity centers on the individual
and his characteristics. It is widely conceded that different sets
of values, orientations, skills, and behaviors are needed for
successful participation in a modern society from those required in a
traditional. Smith and Inkeles believe that these personal character-
istics are to a degree prerequisites for successful functioning in
the roles fashioned by modern institutions. Recruits for these roles
must have some inclination, training, or experience which enables
them to adopt the attitudes, values and orientations the modern roles
require. They also need at least minimal levels of modern skills if
they are to be useful and productive. As these recruits participate
in the institutions and the total matrix of modern society they further
learn, and receive reinforcement of, modern orientations and skills.
Traditional ways are difficult to overlay with modern insti-
tutions. Innumerable constraints are exercised by the adOption of the

new patterns of activity involved in modern institutions. A steel

factory simply does not produce steel dependably if the siesta system
is maintained without modification. The relationship between instiv
tutions and individuals is certainly an interdependent one, wherein
each modifies the other.

Granting that some of the characteristics of individual
modernity may be learned on the Job and that all of them receive
reinforcement there, the problem.remains of where traditional persons
receive the needed training and necessary resocialization for initial
entry into modern roles. This is particularly true for the rural
populace, that portion of the citizenry which is typically most
difficult to convert from traditions of subsistence, dependence, and
subservience toward participation -- economic, political, and social --
in the modern life. Perhaps this is partly because agriculture, more
readily than industry, can produce something without adaption of
modern institutional arrangements. The constraints of modern, or at
least semi-modern, agricultural technology on traditional behavior
are much less powerful than those of industrial technology. Neverthe-
less, they exist, and for modern technology of any sort to be successr
fully utilized, many changes in behavior, values, attitudes, and
skills are needed. This is particularly true in the habits of mind.

{The need is for a profound shift to a change orientation, since the

 

modern technology of today will not be the modern technology of two
years from.now. The modernity of a farmer's enterprise might perhaps
be better indicated by the rapidity with which he is able and willing
to adapt newer technology than by the present state of his Operations.
In other words, the need is not for the adoption of a one-shot round

of technological improvements as much as for the creation in the rural

populace of responsiveness to Opportunities for continual change, and
often for a modicum of speed in making the change.

Crucial to the development of a widespread change orientation
is the deve10pment of the necessary styles of information-processing
and of the skills needed for utilization of messages sent over modern
information channels. The problem of developing a change orientation
is not solvable simply by increasing the supply of information to the
people, although this is necessary.

From what institutions or experiences can we expect the re-
socialization of traditional peoples for entry into modern roles?

The family, church, and traditional primary education are unlikely

to do much to prepare for such recruitment. Their main push is in the
opposite, traditional direction. One of the basic insights of recent
years is that preparation for modern roles may be almost entirely a
by-product of activities which have other ostensible Justifications.
One of the most important of these is utilization of mass media
exposure. The media have their prime apparent utility as information
suppliers and their prime attractiveness as sources of entertainment
and diversion, yet they seem to communicate incidentally a great deal
about modern role-specifications. Likewise people are sometimes
thrust into other situations that give clues to modern roles, such

as that of soldier. They may also travel, observe, and return with
modern notions. But the incidences of military service and of

*
extensive travel are typically low, especially for the rural populace.

 

*In Brazil, the site of the research to be reported here, the
laws on conscription of youth are not enforced in the rural areas,
only in the towns and cities. A recent study in rural India located

In terms of present and potential incidence the mass media have
overriding importance in extensiveness if not in intensiveness. They
present a model of a different life and, by implication, the suggestion
that this life might be adopted by their traditional audience. Hence

the reaction of the guardians of tradition against them.

II. LERNER'S MODEL

The theory or model of modernization which most explicitly
attempts to link the functions of mass media exposure and information
processing to other modernization phenomena was proposed by Lerner
in 1958 and has been expanded by him and others since then. Central
to his model is the concept of empathy, for which he supplies a
variety of definitions, the most central being "the capacity to see
oneself in the other fellow's situation." (Lerner, 1959, p. 50)
Empathy has, in Lerner's model, the function of a catalyst or
lubricant among the processes of urbanization, literacy, industrial-
ization, national political participation, and use of the mass media.
Briefly reviewed, with the contribution of empathy emphasized, the
theory is as follows: A

(1) Urban development is the starting point. When something

like 7 to 17 per cent of the populace live in cities of

more than 50,000, "take-off" toward modernization occurs.

Since empathy makes physical mobility and integration into

a new milieu more probable, the empathic are more likely to
move to and remain in the cities.

 

only 3 ex-servicemen in a sample of 680 rural cultivators. This
indicates that the servicemen are not recruited from the rural area
or do not return to it, or that the size of the army relative to
total population is quite small.

(2) Literacy is a basic skill underlying the modernization
sequence, since it is needed for the emergence of industrial
complexes and urban living. Literacy and urbanization rates
rise together initially; the initially more empathic are
more likely than the less empathic to become literate and to
do so sooner.

(3) With urbanization and literacy providing an audience and
the skills requisite to their deve10pment, the mass media
expand; the empathic are more "mass media-participan ."

They learn from the media the intellectual skills and pre-
dispositional requisites of modernity. The mass media are
"mobility-multipliers" because they provide mediated "travel"
across the barriers of time and space to worlds of vicarious
experience. Thus, they increase the empathic abilities of
their audiences r- and their initial audiences consist dis-
proportionately of the more empathic individuals.*

(A) With mass media participation comes the capacity and desire
to participate politically and economically, to hold views
and opinions on national and international issues, and to
view as important concerns which formerly were outside one's
traditional competence and purview. With the deve10pment of
the market and industrialization, the spiral of progress rises
upward to ever increasing heights of modernity. Each stage
automatically constitutes the input for succeeding stages.
Lerner made this analysis of empathy so convincing that empathy
has appeared quite regularly as an important variable in numerous
deve10pment studies since 1958. (See for example Eister, 1962;
Frey, 196A; Rao, 1963; Rogers and Herzog, 1966.) Peters recently
provided the testimonial that empathy is "an indispensable personal
skill [aid] for persons moving out of traditional settings." (Peters,

1966, p. 6)

 

*Schramm and Ruggels report research which casts doubt on the
generality of the simple pattern of mass media growth in relation to
other elements of the society suggested by'Lerner's model. Their
research uncovers a variety of patterns. See Schramm and Ruggels,

1967.

III. SPECIFICATIONS OF THE CONCEPT, "EMPATHY"

The conceptualization of empathy calls for some discussion.
Smith's summary of its traditional meaning is relevant to much that
we will cover.

The core idea of empathy is the ability to transpose oneself

imaginatively into the feeling, thinking, and acting of

another. (Smith, 1966, p. 19)

(Smith does not use this definition in his own work, but focuses

instead of presumed similarity as an operationally useful conceptu—

 

alization.) Katz provides an insightful characterization of empathy
as "vicarious introspection." Through an identification process we
introject the other person into ourselves and contemplate him.
Coupled with this identification and introjection is a second stage,

which artists refer to as "empathic distance,‘ wherein we gain added
awareness of our own feelings after the identification and intro-
Jection and in this way presume that we have gained added awareness of
the feelings and probable behavior of the other individual. "We are
still ourselves and it is our own imagination that is active, having
been stimulated by the perceptual cues that came to us from the other
person via our own eyes and our own ears." (Katz, 1963, pp. 93, 9%)
If empathic distance were lost, the barrier between our own identity
and that of the person we are attempting to identify with would be
sacrificed, with potentially embarrassing or otherwise distressing
consequences. It is the ability to Observe ourselves while taking
the role of the other, imaginatively, that constitutes the potential

pay-off from the empathic process, for only thereby can we learn about

the other fellow.

The present study postulates that empathy is important in
the communication process leading to development because empathy
(1) facilitates effective interpersonal interaction, especially between
persons who are strangers or whose backgrounds are markedly different,
(2) makes possible greater fidelity in the transmission of information,
especially of the implicit modernization content of communications
from more to less modern systems, and (3) increases the probability
that, to the extent that choices are available, more modern sources
of information will be selected for attention and decoding. The more
empathic person is more likely to select modern messages and is likely
to decode them more accurately. Furthermore, he will be more likely
to personalize and apply these messages to himself and his own
situation. He will more likely see himself as a person involved in
the changes advocated by the messages.

In other words, the emphasis here is on what the empathic person
is likely t°.S2 with modern information as well as the greater likeli-
hood that he will seek and understand it. Precisely because he is
empathic he will be more likely to apply messages advocating change.
This is because he has deve10ped the capacity and habit of imaginative-
ly generating messages for self—consumption concerning possible alterna-
tive courses of action. This capacity is indexed by his ability to
place himself in other roles when called on to do so, but this
imaginative, role-assuming capacity is only one aspect of a general
cognitive syndrome involving flexibility, skill, and a greater pro-
pensity to contemplate hypothetical situations.

The more empathic individual may be regarded as possessing two

distinctive characteristics: (1) an ability to discriminate more

cues in his information environment and (2) a capacity for recombining
these cues with other, already stored bits of information in a useful
and sometimes novel fashion. His skill in interpersonal interaction
comes, presumably from.his heightened capacity to observe in others
slight indications of mood, intention, affect, frame of reference, etc.
and then to utilize these cues to determine the behaviors on his part
which will most effectively achieve his purposes. Frequently the focus
in discussions of empathy is on those internal processes ("putting
himself in the other man's shoes" etc.) which are involved in deter-
mining useful behaviors. But the cue—gathering stage is of equal
importance. Both occur simultaneously in any given instance. Cues
are being continually discriminated and brought into the information
matrix which is contributing to ongoing behavior.

This more flexible manipulation of information, both stored
from.past experience and being fed into the individual at any given
moment, is the key to his capacity to deal more effectively with
hypotheticals and counterfactuals, relating himself to what his social
milieu may define, atdeast'hypothetiehllyihas’unreal. The contemplation
of intellectual ”heresy" is easier for him.

At the same time, the effective empathizer, in each aspect
of his empathic skill, requires information. Perhaps this is most
vividly illustrated in the imaginativeness aspect of empathy.
Ibmgination in the absence of information ig possible, but its out-
comes would more closely approach autistic reverie than a useful new
‘EPProach to problems in the environment. For imaginative activity to
('bear'a.resemblance or relevance to a given situation or prOblem with-

tnrt completely'by-passing the issue, information is necessary. It

10

is rather difficult to respond meaningfully to a question like "What

 

would you do if you were the 'Big Man' in the moiety on Fui-Fui?"

in the absence of information about the role, 'Big Man', the institu-
tion of a moiety, and the general social and physical system found

on Fui-Fui. Any response other than one equivalent to “I don't know"
is likely to be 'imaginative' primarily because it is quite irrelevant.
Only responses coming from one possessing sufficient relevant infor-
mation about the role, its context, and the range Of normal behavior
in that role can be both imaginative and relevant in the manipulation
of the alternatives Open to individuals in that role.

Thus, empathic skill as a general phenomenon consists of a
heightened capacity for gathering information from the environment
(particularly information presented more subtley), 238 a certain
flexibility in recombining past and present information in both unique
and useful ways. As it relates to individual modernization, the subtle
cues may be those present in the mass media or in the urban milieu.

The empathic individual is more likely to decode these aspects of the
media or the city, and he is more capablecf incorporating them usefully
into his own life. Lerner spoke of this in terms of the empathic
individual's ability to reshuffle his self-system on short notice,

as when thrust into a more modernized environment, e.g. someone

passing rapidly from the farm to the factory. This reshuffling

coupled with a heightened capacity to extract the more subtle and use-
ful cues from one's experiences distinguishes the more empathic
individual from his less empathic peers. The term, "empathic" is
merely a convenient and traditional label for this ability. And the

aspects of the ability to be explored in this study relate to its

11

basic components; discrimination, flexibility in manipulation, and

incorporation of these cues into past elements of information.

IV. SOME EARLIER CONCEPTUALIZATIONS OF EMPATHY

Drawing on Lerner's work and on earlier intellectual traditions
the following aspects of empathy may be distinguished: (1) imagin-
ative role-taking, (2) insight into others' experiences, and (3) skill
in interpersonal perceptiveness. The first two are quite similar,

but they will initially be considered separately.

Empathy as Role-Taking.

 

Coutu defined role-taking as:
...the symbolic process by which a person momentarily pretends
to himself that he is another person, projects himself into
the perceptual field of another person, imaginatively "puts
himself in the other's place" in order that he may get an
insight into the other person's probable behavior in a given
situation. (Coutu, 1951)
G. H. Mead's theory of the manner in which the self develops is heavily
dependent upon this view of the role-taking process. Through role-
taking shared meanings are developed. Additional uses of role-taking
have been proposed as well. Stewart and Hoult, for instance, utilize
deficiencies in role-taking to explaindifferential levels of author-
itarianism, (Stewart and Hoult, 1959).
This role-taking conceptualization of empathy views it as a
process of putting oneself in the other man's shoes in an attempt
to understand and predict his behavior. The popular detective
supposedly works this way, utilizing the counterfactual mode of

thought which language allows to frame questions like "If I were the

l2

bank robber, where would I hide?". In a less conscious or explicit
fashion, Mead believed the development of a self-system grows out of
symbolic interaction with others, and the imaginative re-enactment
Of social roles is crucial to this process. Thereby the individual
comes to understand these roles, and by acting in them.toward himself
somewhat as he finds present incumbents actually acting, he develops
a sense Of self and a self-evaluation. A person viewing an exciting
film.may find himself identified with the hero to a point where his
fists clench, his heart pounds, and his stomach flip-flops. This
example comes close to the kind of thing which occurs when we gain
vicarious insight into the experiences of others by feeling what we
take to be the same emotions that they feel. Although emotional
involvement is not ruled out in imaginative role-taking, and some
would argue that it is a necessary means to gain insight into others'
experiences, the cognitive aspects of the role-taking process tend to
receive primary emphasis. Before either the words empathy or role-
taking were coined, Adam Smith observed:

As we have no immediate experience of what other men feel,

we can form no idea of the manner in which they are affected,

but by conceiving what we ourselves should feel in like

situation. (Quoted in Katz, 1963, p. 9%)
The process of imaginatively putting ourselves in others' roles to
understand their thoughts and feelings and to anticipate their re-
actions to our actions is clearly indicated here. Also highlighted
is the similarity notion, the assumption that others are like us.
_As we will see, this assumption forms an.important link in explaining
the utility of empathy for modernization.

Lerner's main concern is with this role-taking concept of

empathy, and his measurement of it is linked to the individual's

 

l3

ability to describe his activities in terms of modern, high-status
roles. Along with this, Lerner postulates a net increase in human
imaginativeness and greater cognitive flexibility. All this suggests
a larger capacity for interior manipulation, particularly Of the

individual's concept of himself. (Lerner, 1966, pp. 22l-22A).

Empathy as Interpersonal Perceptiveness.

The view of empathy as interpersonal perceptiveness grew out of
the psychology of perception and denotes differential ability in
utilization of cues about others' internal states. This can lead to
improved prediction of others' behavior and greater success in inter-
personal relations. (Cline and Richards, 1958: Cronbach, 1955;
Hastori and Bender, 1952; Foa, 1958) Interestingly, some writers
suggest that this kind Of empathy may be mgge characteristic of tra-
ditional or at least of primitive societies than of modern ones.
Deutsch (1963) believes that traditional men are more interpersonally
perceptive because of their experience in the extended family. Having
a need to differentiate carefully between behavior toward a large;
number Of children and adults who are found in close and constant
interaction in the family, the child develops more capacity to size
up individuals and behave accordingly.

in linewwith this, Margaret Mead (196A) and Hurley (1965) I
suggest that primitive man is already highly imaginative, involved A
in vicarious experiences, and empathic. Hurley claims that modern-
ization wrings this out of the primitive. (Hurley, 1965, p. 3h)
More probably, both the primitive and the modern man are capable Of

role-taking but the modernizing individual learns to suppress or

1A

channel his experiences into acceptable patterns rather than allowing
his imagination free rein. It may be useful to see empirically whether
more traditional individuals differ from less traditional in "net human

imaginativeness."

Empathy as Imaginative Insight into Others' Experiences.

 

Holbart and Fahlberg (1965) claim this is "the" social psycho-
logical approach, although something like it has long been known as
the artistic meaning for empathy. (Langfield, 1920) They believe
this concept centers on feelings, intuition, and perhaps identification
with others and pointto its major difference from the perceptual

approach by the phrase: "knowing (another's feelings, not by seeing

 

them.in him'but by feeling them in oneself." (HObart and Fahlberg,

 

 

1965, p. 596, emphasis mine)

Deutsch provides a brief discussion of the topic which suggests
the interrelatedness of these three approaches. He begins by dis-
tinguishing between understanding a person's situation and under-
standing his outlook. In the former we are able to put ourselves into
someone else's place while in the latter we not only do this but also
have in our imagination

a model of his mind, with his memories and values, that is

sufficiently accurate for us to predict -- and perhaps

experience emotionally by empathy -- how he would act with

his mind and his personality under conditions in which we

ourselves might act quite differently. (Deutsch, 1963, p. 12)
The relation to perceptual psychology is obvious in the beginning of
this quotation. Deutsch brings in the feelingful aspect at the end
and reserves to it the name of empathy. A bit later he makes the

further distinction between the two types on the following basis:

15

a rational reconstruction of the personality, culture, or
cognitive map of the actors concerned

versus

an emotional simulation of their feelings by an imaginative
manipulation of our own minds. (Deutsch, 1963, p. 13)

 

 

 

I believe that Lerner includes both of these components in his
conceptualization of empathy, and certainly they are not mutually
exclusive. While we may distinguish them, and perhaps hOpe that

for certain ends one type may predominate, they easily shade into each
other. The distinctions between cognition and emotion, between role-
playing and interpersonal perceptiveness are particularly difficult to
formulate Operationally for research purposes in terms that traditional
persons can handle. (See Cronbach, 1958, p. 353) Though distinguish-
able, these three traditions Of empathy may blend into one another
when it comes to Operations. Two recent approaches which do not

attempt to distinguish cognition from emotion avoid this prOblem.

The Impersonal Other.

 

Waisanen and others in a discussion of literacy recommend that
Lerner's notions of empathy'be'looked at from the vantage point of
symbolic interaction and that this older theory be expanded to include

"taking the role of the Impersonal Other." This approach not only

 

links the variable of empathy to an earlier body of theory'but provides
a cogent explanation of the observed relationships between empathy
and modernization. Skill in taking the role of the Impersonal Other

...enables the person to approach new social situations with
confidence and to participate in them productively; it broadens
social horizons and enables efficient interaction with
strangers; it frees the self from.the bounds of what is
immediate and particular and, by the mechanism of abstraction
and generalization, enables Eicfl COping with the complex
interplay of societal demands which touch time, authority,
aspirations, reward preferences, the world as change and the
self as potentiality. (Lassey, Mendez, and Waisanen, p. 6,7)

16

The crux of this view of empathy is that it shows hey the empathic
person does better in new social situations and interacts more
efficiently with strangers.

As Waisanen puts it, it seems clear that in all societies part
of the process of child development is an increasing capacity to take

the role of the Generalized other -- that general referent group of

 

significant others whose Judgments of oneself and of one's actions are
internalized and shape one's self-concept.

It is not at all clear, however, that all individuals living in
a peasant society always have much need for or Opportunity to develop
capacity to interact impersonally in a limited and efficient manner
with strangers. Of course, reaction patterns to strangers are learned,
but these patterns do not include taking the stranger's role in a
limited and effectively non-involved manner which suffices for effective
interaction.

For the traditional, rural individual, the concept of stranger
is not likely to be very well differentiated. For the urban, modernized
individual, there are many varieties and types of strangers. To the
extent that differentiation leads to meaningfulness, we may say that
the concept of stranger is more "meaningful," and less generally
threatening or exotic for the urban, modernized man than for his
traditional counterpart.

If we assume that one of the consequences of modernization and
urbanization is bureaucratization then the link to skill in impersonal
interaction is more apparent. Success through continuous face-tO-face
interaction in small groups may not develop the skills needed for

success in a larger society mediated by impersonal systems, paperwork,

l7

and regulations. The essence of bureaucracy is impersonality while
relationships in traditional systems tend to be personalized.

The man who has more contact with urban centers is more likely
to develop both generalized and differentiated skills in dealing with
strangers. He, more than his stay-at-home neighbor, constructs a
generalized Impersonal Other, and comes to take this role with
alacrity. Urban contact, then leads to empathy.* This position is
particularly interesting in view of Rogers' belief that neither mass
media exposure nor trips to the city probably have much impact on the
non-empathic person. Perhaps the contrast in vieWpoints is not
complete, for Rogers seems to say that without a modicum of empathy
to begin with a "mental isolator" is present which immunizes the
peasant from cosmopolitan influence. (Rogers, 1965, p.53) ‘Waisanen
and others appear to assert that trips to the city tend to bring about

a modicum of empathy.

Empathy as an Instance of Hypothetical Thought.

 

Requiring people to provide information about their actions in
a role they have never experienced is to require them to think hypo-
thetically about themselves and to describe their hypothetical
behaviors. Such a process is similar to the thought processes which
accompany uncoerced change -- EEK alternatives are brought into a

decision matrix for consideration. In other words, a person who can

 

*This appears somewhat at variance with the summary of Lerner's
model on page 6, but the divergence is only apparent. There we suggest
that the more empathic are more likely to have urban contact. Here we
indicate that such contact increases empathy. The two are not
mutually exclusive.

18

tell you in some detail what he would do if he were the president of
the country may also be able to imagine what he would do if credit
became available or if he had a better seed to plant.

This way of looking at empathy emphasizes its importance because
the practice of empathy is like the making of a hypothesis. The
individual is able to think about relationships among actions he has
never taken,and(results whichihe has neverwexperiencedo A person
who analyzes hypothetical situations well may cognitively manipulate
and examine both his means and his goals with a mind toward change.
Lerner has moved toward this point of view recently in calling empathy
the "poor man's creativity." (Lerner, 1966, p. 2A3) I believe the
referent for this may be the hypothesizing character of empathy.

This approach to empathy's correlations with other indicators
of a modern orientation has the advantage of a very direct link to
modernizing behavior. The link to the original role-playing questions
is also Obvious. The success a person has in handling such questions
should be directly related to the success he experiences in handling
hypothetical situations that make a difference in his world. He should
be more capable Of entertaining questions like "How much harvest
would I expect to get if I tried this new seed?" or "Suppose I were
living in town instead of out here -- would I be better Off?" It
is not only interesting to know whether a person SEE handle these
kind of thoughts if asked about them.in an interview, but also whether
he spontaneously and customarily composes them in the absence of inter-
viewers. My limited experience with traditional farmers leads me to
believe that few of them.have the habit of intellectually manipulating

hypothetical situations to evaluate outcomes that might lead to different

19

and improved results for themselves and their families. .At least such
hypothetical thoughts as occur do not appear to be well articulated

or easily verbalized.

saw-1am-

Five different approaches to empathy appear relevant to
modernization. Three of them represent traditional conceptualizations.
The fourth incorporates empathy into a well elaborated theoretical
viewpoint, symbolic interaction. The fifth utilizes the similarity
between skills required for empathy and skills involved in decision-
making to account for linkages to modernization. The latter two can
be viewed as ways of cutting across previous difficulties and getting

at the root of the matter as far as national development is concerned.

V. EMPATHY IN THE MODERNIZATION PROCESS

Possible linkages between the locations of individuals in the
process of modernization and the levels of their empathic abilities
may be subsumed under three general headings. These three have to

do with different aspects of empathy.

Similarity,

Some scholars stress the assumption of similarity among inter-
acting individuals which is implicit in empathy. Smith approaches
Operationalization Of empathy as "the tendency of a person to assume
that another person's feelings, thoughts, and behavior are similar to
his own." (Smith, 1966, p. 19) Katz makes the point that we "under-
stand" best those whom we resemble. Analogous to this, we think we

"understand" (or can ”learn to understand") those whom we think we

20

resemble. Empathy follows presumed similarity. Mead asserted that
the role-taking process was essential in attaining commonality of
meaning among humans, and since communication with understanding is
possible only insofar as meanings are shared, empathy is, in this
sense, essential to successful communication.

The assumption of similarity implies a reduction in social
distance among individuals. Rao has emphasized, perhaps overmuch,
the mobility assumption underlying a person's ability to describe his
behavior in another role. (Rao, 1963, p. 277 ff.) If I can say what
I would do as the president I can almost imagine myself becoming
president or at least I can assume that the current incumbent of that
role is not such a very different fellow from.myself and that my views
on what is proper behavior therein are not completely presumptuous.
One bedfellow of reduced social distance is relatively high social
mobility and a positive valuation of it. This is one facet of the
modernized personality.

The assumption of similarity implies an ability to enter into
interaction with strangers. The more I perceive outsiders and strangers,
even foreigners, as being "really" very similar to me, the more
readily I can learn from them, the more I can approach interaction
with them, and the less I will feel distrust and other barriers.

The assumption of basic similarity facilitates my becoming more aware
of the areas of difference existing between the stranger and myself,
and perhaps I can therefore begin to differentiate the processes or
means whereby these areas may be bridged. I become more tolerant in
my awareness of the variety in human beliefs. I learn of the possi-

bility that other forms of thought and action, other life-styles, may

21

have merit. In essence, the argument is that empathy, as assumed
similarity, involves awareness of an underlying human unity and relates
to the individual's ability to feel relatively unthreatened as he
interacts with and learns from.strangers and outsiders. This makes
for a heightened awareness of the variety of human beliefs and of the

alternatives implicitly available to the individual.

COgnitive Flexibility.

 

The more empathic person is less rigidly attached to convention
and precedent, for he can entertain, at least in his imagination,
alternative courses of behavior or alternative beliefs. He is
relatively less preoccupied with the concrete and more with the
hypothetical. In a sense, he is able t°.EEE rather than.merely.rgagt.
0n the intellectual level, he thinks more_§bgut things, rather than
Just thinking things. (Thorndike, cited in Berle, 1960, p. 121)

The non-empathic cultivator who responded in frustration to an inter-
viewer's questions about his Opinions:

Look, when I'm hungry I eat; when I'm sleepy I sleep dammit
...I don't think about things

was indicating an almost animal attachment to the concrete and immediate.
Katz argues that because the more empathic person not only

has stored a larger number of identifications from his past but is

also able to move flexibly from one identification to another he finds

himself less threatened by the strains of transition to a modern

society. He has the role-taking skill needed to achieve a better

understanding of what is going on. His concept of self is likely to

have more continuity as he moves from traditional social groups to

others where his identity must be established anew. He finds through

22

the role-taking process the necessary social support and understanding
of what is expected of him. In addition, he may be better prepared
for the transition by his having played the roles imaginatively during
his exposure to them as depicted in the mass media. Through the media
and his application of their messages to his life, he may be incor-
porated into the larger national life, come to identify with it, and
be concerned with matters outside himself and the community or valley
in which he was born and raised. \His self-image and world view are
broad enough to include more of the variety the modern world presents.
Not only does he arrive more rapidly and with less trouble in the
modern institutions and roles, he also continues his adjustment with
fewer prdblems once there. According to Katz,

Everyone living in a mobile, mass society must be able to make
rapid adjustments in attitude and to maintain self-esteem in the
face of many threats. The poor empathizer is simply more
vulnerable than the average citizen. He is less qualified to
cope with conflicts and is less defended against the anxieties
of living in a society of increasing alienation. (Katz, 1963,

p. llO)
Katz's explanation of why this is so is in terms of the non-empathic
person's supposed difficulty in generating and interpreting feedback
sufficient to construct an appropriate self—image.

He misses the feedback of the data he needs to assess his own
behavior and lacks the sense of mutuality which he requires in
order to maintain his self-esteem. The judgments he makes of
others are out of harmony with the judgments of other group
members or associates. (Katz, p. 110-111)

In sum, following Katz, we may indict the poor empathizer for (l)
difficulty in communicating ideas and feelings, (2) misreading feed-
back needed to assess his behavior's impact on others, (3) a consequent

lack of a sense of mutuality required to maintain self—esteem, (h) being

out-of-step with other group members in Judgments, (5) erroneous

23

interpretations of others' attitudes and evaluations, and (6) a con-
sequent feeling that others do not respond appropriately. In brief,
he is liable to become frustrated in attempting to manipulate social
reality and alienated in his affective relationships with others. The
swirl and flux of change in the modernizing environment confuse and
annoy him. He is led toward isolation rather than toward participation,
and his identity is eroded without the accompanying substitution of a
newer, more useful self-concept.

Further, the more empathic person has energy for matters other
than himself. He has less need to be preoccupied with self, being
more successful in maintaining his identity and less threatened by new
situations and new individuals. Here the bridge between the similarity
and the cognitive flexibility notions becomes explicit; the more
cognitively flexible person is less hypersensitive with strangers
because he feels less suspicious and less threatened. Therefore, he

is more able to learn from and about them.

The Construction of Alternatives.

A person able to say what he would do if put in another man's
shoes is not only demonstrating his belief in the similarity'between
himself and that other man, or his belief in the possibility of
playing to that man‘s role himself or at least learning something from
persons in it; he is also demonstrating that he can contemplate change
with a calm eye, evaluating its benefits and costs. If he can place
hhmself in my hypothetical situation, then he can construct hypothetical
situations involving himself, or at least apply them directly to him-
self when they are presented, ready-made, by others, e.g. via the

mass media. Those who think about the decision-process assume that

21+

such contemplation typically takes place. But the energy required to
consider acting differently and the flexibility involved in such
behavior may not always be present. Those individuals in develOping
societies who have something of the capacity to ask the "what if..."
type of question have a definite advantage when it comes to making
changes (either in a modernizing or traditionalizing direction) in

their living habits and activities.

Empathy as an Interaction—Enhancer or as Identification.

There is another possible line of cleavage along which the
contribution of empathy to modernization may be envisioned. It is
essentially contained in some of the considerations already outlined.
On one side, empathy may be regarded as a generalized "interpersonal-
interaction-enhancer." This fits the psychologists' interest in
interpersonal perceptiveness and the arguments for empathy's utility
in interaction with strangers. Presumably the more perceptive
individual is better able to adjust his messages (both before and
during their emission) to achieve his purposes with his receiver.

The weakness with this approach is that, except for the concern with
interaction with strangers, there is little reason to assume that a
more traditional person is necessarily less empathic than the more
nmdern person. Certainly within his own social group the traditional
nun.is well able to play the necessary roles.

Empathy might, however, be defined not in the general terms
just stated, but specifically as an increased tendency toward
identification with modern role models. .These models are presented
Primarily through the mass media, although some contact may occur on

a face-to-face basis or as a member of a crowd. The individual's

25

ability to describe himself in another modern high-status role, may
merely be symptomatic of greater identification with persons in that
role, and presumably with the values and attitudes they embody. Under
this definition, the person scoring as more empathic on a set of
questions simply is indicating that he has identified to a greater
extent with modern roles. It is hardly surprising, then, that he
has also picked up some of the other cognitive and affective accouter-
ments of those roles and that these show up in his scores on other
measures of modernity.

A partial summary of the aspects of empathy distinguished

conceptually and their linkages to modernity may be useful at this

 

 

point.
Aspects of Empathy Links to Modernization

1. Capacity for imaginative 1. Identification with and adoption of
role-taking attributes of modern role-models.

2. Greater understanding of those in
modern roles and ability to interact
more effectively with them.

3. Assumption of similarity, with
implications of reduced social dis-
tance and of increasing social
mobility.

l—l

2. Ability to handle . Useful in contemplating alternative
hypotheticals courses of action.
2. Useful in adjusting to changes which
are being imposed by forces over
which one has no control.

3. Capacity to cognize and 1. Useful in all forms of interaction
predict the behavior of and particularly in maintaining
others with accuracy-- adaptation in a changing social
interpersonal percep- situation. This ability is not
tiveness necessarily more deveIOped among

the more modern.

h. Capacity for an affec- l. Relates to introspection, sympathy,
tive, emotionally based, and an understanding of the impact
understanding of others of events upon others, with a
and their points of view. consequent concern for their welfare

and rights.

26

5. Skill in taking the role 1. This skill may be negatively related
of the impersonal other. to the tendencies outlined in the last
item. It is useful in dealing with
"segments" of others whom one has no
need or desire to understand in their
entirety.

2. Particularly useful in a'bureaucratized,
urban society where individual identity
may'be submerged in the interest of
large scale operations and getting on
with the business of production._

3. Facilitates productive interaction
with strangers and the possibility
of learning from them.

There is no intention in what has gone before to promote empathy
as the most important variable in the individual's modernization
process. Many other variables have been distinguished and their
value is not disputed by the present study's focus on empathy. The
variables of literacy and education are two of particular importance
and interest. Literacy has close and, perhaps, more direct ties to
modernization. It trains individuals in a skill essential to dezoding
messages in all the print channels of communication. Of perhaps
even greater importance, it may produce a higher awareness of symbolic
manipulations while training individuals in relatively mundane
reading and writing skills. This aspect of literacy is of greatest
interest to those trying to explain the modernization process. This
aspect ties literacy directly to the development of one of the
requisites of a modern social system -- actors capable of dealing with
abstractions and of handling themselves in the richer symbolic environ-
ment brought on by increases in technology and urbanization.

Literacy is clearly closely tied to utilization of the mass
media, as already indicated. Additionally, the ability to read and
write is frequently a concomitant of social status in developing

nations, and status implies the potential to ottain access to the media

27

channels, e.g. attend movies, buy newspapers, purchase a radio set,
etc. ‘Those able to educate their children to read and write are,
typically, more affluent and lodged in positions of greater importance
in the society. This is particularly the case for the rural sector
of developing nations.

When investigated empirically, literacy relates closely to
‘modernization and mass media utilization for yet another reason.
Aside from its centrality in the causal nexus, it is considerably
easier to measure than a skill such as empathy. Its outward manifes-
tations are more easily recognized and can‘be elicited on demand.
A person can be asked to actually read a selection, and he can'be.
scored on the accuracy of his reading and tested on his comprehension
of what he has read. It is more difficult to elicit, in a testing
situation, an objective, behavioral measure of empathy. However, when
we go behind the external manifestations of literacy and require
evidence of increased capacity for symbol manipulation, the measurement
difficulties for the two variables are more similar. From.an ability
to read a paragraph and interpret its content, heightened abilities
in symbol manipulation are typically inferred. Similarly, from an
ability to recount behaviors in a role one has never experienced and
to do so in a novel and relevant fashion, empathic abilities are
typically inferred.

Both capacities ultimately descend into the unobservable,
covert, symbolic realms of human behavior. To the extent they are

distinct, their causal relations are difficult to unravel.* Lerner's

 

. *Empathy can perhaps be best distinguished from symbol manip-
ulation in general on the grounds of greater flexibility in such

28

model, as noted, implies that the more empathic are more likely to
seek literacy and endure the process of acquiring and maintaining the
necessary skills. Likewise, one is safe in assuming that, since
literacy Opens an immense volume of information to the individual,
the literate individual would have more opportunity to gather bits of
information for inclusion in his decisions, i.e., he would have more
alternatives available for manipulation. Even more to the point, if
the inculcation of the skills of literacy in some fashion heightens
the ability to manipulate symbols to useful ends, the literate person
not only has access to more cues but is likely to be more capable of
manipulating them in useful and novel ways. Thus, literacy should
lead quite directly to increasing empathy. Just which causal thrust
is strongest is unknown, although it appears that a better case can
be made for literacy bringing about increments in empathy than for
empathy bringing about increments in literacy (or in symbol manip-
ulation, which is more to the point). The main argument here is that
the two variables are mutually causal. They thus would require analysis
through a systems approach rather than a simple, one-way model.

It is possible that, at least in some instances, literacy may
be achieved without concomitant increment in important aspects of

empathy. The few EurOpeans literate during the medieval period were

 

manipulation and'by noting its dependence upon the discrimination of
subtle cues in the environment. One can conceive of a person highly
skilled in symbol manipulation along lines which rigidly control his
thought who, at the same time, seems unable to discriminate subtle
cues about the feelings, intentions, and reactions of those about him
or is unable to perceive evidence which might lead him to adjust his
too rigid frame of reference. Such a person would be low in empathy
though quite high in symbol manipulation.

29

not universally noted for their differential empathic skills. As a
more modern example, Communist China has reputedly doubled its literacy
level during the past five years but, at least so far as can be Ob-
served, increments in empathy or openness tO thought transcending the
party-line seem.absent.*

Formal education appears to function much as does literacy,
perhaps because its primary impact, at least in Brazil, is the in-
culcation of literacy skills and little else (Bostian and Oliveria,
1965). Education may be regarded as an instance Of a coerced, if
benevolent, Opening Of the mind and an expansion of the alternatives,‘
or skills to perceive alternatives, which a person can use in decision-
making. As such it doubtless contributes tO modernization. However,
education in the rural areas Of Brazil frequently includes heavy doses
Of indoctrination in traditional values and attitudes. Except for
teaching literacy and some minimal arithmetic abilities, current rural
education probably makes little contribution to the modernization Of
most individuals exposed to it. Of course, the impact of learning
these skills is not to be minimized, and if the child continues on
in the education system, something which is much more likely if his
parents move to the larger towns or cities or if they are relatively

affluent, then a greater modernization Of views becomes likely and

*The report Of two correspondents from the West who visited
China for a month or more and had Opportunities to talk with Chinese
students is particularly striking in this regard. After several hours
of questioning the students about conditions in China and their points
of view, the correspondents asked if there were any questions about the
West which the students would like to pose. There were none. Interest
in the world outside seemed totally lacking. Perhaps this was a
function Of the sanctions of the social situation, but the correspondents
felt that lack of interest in the outside Was genuine.

30

capacity for change probably increases.

Despite the importance Of literacy, education, and other variables
which had earlier been neglected, this study limits itself to an
examination of empathy. Except for occasional references to these

other variables, we will not consider them further.

VI. A MODEL OF EMPATHY IN THE MODERNIZATION PROCESS

Having reviewed possible linkages between empathic ability and
modernization, it may be useful to attempt tO array these into a
model which suggests some sequence and time order. The model is
shown in Figure l. The arrows indicate the general flow through time
of one probable sequence Of events. Something Of the interactive
nature Of the process is suggested by the back and forth pattern Of
the arrows, particularly under the heading of mass media exposure,

urban experience, and strangers.

Genesis.

Different levels of empathy or of capacity to develop empathy
under given kinds Of exposure to modern influences, have no very
adequate and certainly no tested explanations. Perhaps genetic,
hygienic, and nutritional factors play an important role in early
life. Perhaps differences in perceptual acuity or tendencies toward
introspection favor such deve10pment. Perhaps the social situation
is of greatest importance. Opportunities to Observe a variety of
roles early in childhood, parental support for role play, position in
the family, and degree of isolation from other social influences may be

central. Deutsch suggests that experience in the extended family

31

Figure l: A model Of empathy in the modernization process

Genetic differences, differential
hygiene, nutrition, social exper-~5§_ Differential levels Of a capacity
iences, primary education, and to develop empathy when exposed to

unknown determinants ééf////////

Mass media, urban experience,
strangers More exposure teaches more about
“\\\\~N‘I“fi>» the conventions and skills of the

Increased empathy facilitates ”,_media or the interaction sit-

the decoding Of the modernization uation, among them the skill Of

import of new messages \\N\‘\\l\\£§’ empathy
The mass media, urban experience,

The more empathic identify more éélaeand contacts with strangers

 

easily with these role models provide new role models
and are more likely to learn.
from.them. \E“‘S>‘ The modern orientations of these

models are more easily and
accurately assimilated by the

The more empathic adOpt the ‘EI’E”'more empathic

modern orientations more

completely. h\\\\\\‘j§> The more empathic tend to seek
out more messages from.modern
sources

Relative Dissatisfaction

 

Migration possible?
Yes, a more modern éf////’ \ES .32, blocked by inadequate
setting sought Opportunity-structure

t
Mbre empathic enter with greater The individual experiences
assurance and less trauma difficulties with traditionals

l’ and tradition
More capable Of adaptations Of Receptive to change and begin
the self-concept to meet the to utilize increased skill with
demands of new situations and hypotheticals to examine and
associations imaginatively work out alter-

natives ////
25//

Expand identifications, reference groups to include
national and cosmopolitan.linkages ///
¢, A,

Fit more adequately into the modern world /

4/

it
Take action to adapt to or implement change
toward the model of modernity feasible in
the particular situation

32

develops higher levels Of interpersonal perceptiveness. Hagen‘s
notion Of the importance Of strains induced in the child-rearing
practices of parents caught in the modernization net might in some
fashion help explain differences in empathic ability or in propensity
for its development. Primary school experiences with peers and with
Older children and adults may play an important role.

No data which can be construed as bearing on the question Of
the genesis Of empathy or of predisposing characteristics are available
from this study. In their absence we will have to ignore the problem
Of origins and merely assume that some individuals have differentially
higher empathic abilities which they bring with them when they en-
counter the modernizing stimuli Of mass-media exposure, urban exper-
ience, and interaction with strangers representative of the modern

sector.

Mass Media Exposure, Urban Experience, and Interaction with Strangers.
These stimuli are most effective with those whoalready possess
a modicum of empathic ability. Initially, exposure helps teach the
conventions and skills needed to decode the messages coming through
these channels. Empathy is one Of the skills which are differentially
inculcated in those so exposed. One Of the initial results of mass
media exposure is the creation Of doubts in the minds of receivers
about the value of traditional ways, life-styles and behaviors.
When the doubts are raised receivers may become less capable (at
least temporarily) of decoding the messages being sent. The doubts
raise barriers to understanding, a sort of "noise" in the receiver
which tends to distort if not bar further modernizing messages.

The empathic, because of their capacity to lay aside temporarily

33

the strictures Of tradition, are more able tO attend tO modern
messages. They are, thus, more likely to learn from.modern messages
and from those who embody'modern roles.

The mass media, interaction with strangers, and exposure to
the urban world help to teach the ability to interact impersonally
with others. As traditional persons increasingly identify with
modern role models, they tend to assimilate the orientations of these
roles and imitate, as far as practical, the behavior Of the role
models. This process is easier for the more empathic, since they
are more capable of imagining change in themselves.

It is also likely that the more empathic Obtain more enjoyment
from.modern messages and consequently they not only tend to under-
stand them better when exposed to them, but also, insofar as message
alternatives exist, they tend to seek out those which are more
modern.

Through occasional contacts in the city or with strangers an
understanding is brought about Of the impersonality necessitated by
the grouping together Of large masses of human beings. The con-
ventions and rules needed to regulate behavior are better apprehended
and are more likely to be tolerated by the empathic. Increased ability
develOps to interact with others in a way which is superficial, but
sufficient to the attainment Of goals. The more empathic learn
more easily from such experiences and come to differentiate more
completely the category Of "stranger." They learn that at least

some strangers may be useful to fulfillment of their goals.

3h

Dissatisfaction With Traditional Restraints

 

One of the consequences of adopting more modern orientations
will be dissatisfaction with at least some aspects Of the traditional
way Of doing things. Perhaps many aspects of tradition will continue
to be valued, but certain habits and procedures will be viewed as
constraints, confining the person to fewer alternatives than he
feels comfortable with, particularly with his newly found awareness
of modern possibilities. A partial dissatisfaction with the
traditionality Of his surroundings will be aroused sooner in the more
empathic man. If he is in a position Of influence and can affect
his community, it will become less traditional. If not, he may tend
toward marginality and begin to look outside the community for his
models, information, and attachments. He will tend to relinquish
friendships and ties within the community to the extent that he tires
of its restraints and Obligations. He may get to a point where he will
consider migration to more modern settings, typically to the city.

If the Opportunity-structure allows this, he will enter the more
modern setting with less trauma and greater assurance. The more
empathic individual will there adopt more easily a new self-concept.
He will establish with greater ease the new reference groups and
associations needed to uphold his sense Of identity and to provide
social feedback. In short, he should fit more adequately into the
modern world.

This would also be the case if we could compare persons con-
strained tO move to the city who differ markedly in their levels Of
empathy upon making the move. While urban experiences will probably

increase empathic abilities, particularly the ability to interact

35

impersonally, those who arrive with more empathy in their kits of
cognitive and affective tools should adjust more easily.

If the opportunity-structure blocks movement out of the
traditional setting, the individual may come to experience some
difficulties with traditionals. As noted, he may tend toward
marginality in the community, or at least his focus Of attention
may shift to relatively more cosmopolitan concerns. He might aspire
to more education and hope to see his sons in more modern Occupations.
If he must remain a farmer he is likely to seek to become a modern

one .

Modernization Efforts.

 

With an increased capacity for venturing into modern settings
and interacting successfully in them, and with greater skill in
hypothetical thought and more Of a tendency to plan, patterned after
the modern role models with which he has identified, the more empathic
individual will be more likely to modernize his life successfully.

His influence and ex.mple will tend to upgrade the level of modernity
Of his community. For himself and his family he will provide a better
fit into a changing world. He will be more receptive to change programs
proferred by governmental and private agencies.

This model has been presented with some doubts. While it
constitutes a very imperfect departure point for empirical tests,
even in its current crude state it suggests more possible areas Of
research than can be dealt with in a single study. Therefore, the
limited purposes Of this study need to be spelled out with precision,

and detailed statements of its hypotheses are required.

36

VII. OBJECTIVES OF THE STUDY
The present study aims to:
Explore through factor analysis the dimensionality Of Lerner's
conceptualization Of empathy and provide other useful evidence
about the nature Of the variable. The following are included
in this analysis:

a. a comparison Of four aspects of empathy: role-taking;
ease and skill in impersonal situations; imaginativeness;
and facility in dealing with hypothetical situations;

b. indications Of the stability Of empathy, as usually
measured, over time;

c. an analysis of the feasibility of a new approach to
empathy's measurement which requires no specific items
and does away with coding prOblems.

Provide evidence about the differential placement and functions
Of the relatively empathic individual in social systems Of
relatively greater and lesser traditionalism. The following
are included in this analysis:

a. relative empathy and the likelihood Of being indicated as
an information source or friend in more and less
traditional communities;

b. relative empathy and agricultural Opinion leadership
in more and less traditional communities;

c. relative empathy and social status in more and less
traditional communities;

d. relative empathy and integration into the social life Of

more and less traditional communities.

37

3. Investigate the relationship between empathy and modernization
variables such as mass media use, cosmOpOliteness, knowledge,
and modern orientations such as the relatively favorable
evaluation Of change, planning and modern orientations to work

as a means for achieving one's goals.

Measures of Empathy

 

A variety of aSpects Of the concept of empathy have already
been distinguished. Obviously when we speak of empathy it is useful
and necessary to specify exactly the aspects and indices Of empathy
we have in mind. The question naturally arises: are these aspects
Operationally distinguishable in research on traditional persons?

This study seeks to provide the beginnings of an answer to
this question by a factor analysis Of 18 items intended to measure
various aspects Of empathy. The items cover the aspects we have
labelled role-taking, ease in impersonal interaction, imaginativeness,
and facility with hypotheticals. The factor analysis procedure

utilized and its outcome will be discussed in subsequent chapters.

Measures Of Empathy's Stability.

 

Is empathy, as it has usually been measured in survey work in
deveIOping countries, a fleeting, shifting phenomenon, or can one
be reasonably sure that the empathy score assigned a person will be
essentially the same as the score he will receive a month or a year
later? The reliability Of the measure over time is important in

judging its usefulness.

38

A New Measure of Empathy.

 

The important aspect of the Lerner type Of role-taking item
may be its measurement Of a general aptitude for hypothetical thinking.
It is clear that one thing involved in Lerner's items is also involved
in any other item which asks a person to imagine himself in a
hypothetical situation. If the adequacy of his performance in
imagining himself in such a situation could be measured in some
other way, it might be possible to eliminate the role-taking questions
altogether. A score on general aptitude for hypothetical thinking could
replace them.

Coding the responses to Open-ended, role-taking, questions is
a tedious task. Training of coders is relatively arduous. Even
among trained coders the impression persists that the response written
in the blank is a very poor indication Of the adequacy with which the
respondent placed himself in the role in question. A well-trained
coder could do a much better job if he had access to the complete
response, verbal and non-verbal, rather than that fraction which the
interviewer preserves on paper. Something similar could be said, of
course, for other Open-ended questions, but the role-taking nature of
the items makes it particularly crucial for empathy.

Suppose that interviewers were trained to estimate the degree

 

to which the respondent placed himself in the situation? And suppose
that not only questions on the order Of "What would you do if you
were president?" were thus rated by the interviewer, but also other
hypothetical questions, the main purpose Of which was the measure-
ment of other variables Of interest. A distinct methodological

advance could be claimed if it were possible to eliminate the coding of

39

empathy items or to eliminate the empathy items themselves by Obtaining
the necessary information from interviewer ratings Of the success with
which the individual placed himself in hypothetical situations called

for in 23225 items. This study explores this possibility. The

resulting measure is compared with the usual approach to measuring

of empathy in terms Of the degrees to which the two predict modernization
behavior in the areas Of mass—media exposure, urban contact, and
adaption Of farm practices. Their relative performance as predictors

are used as a means to decide among them. The possibility Of their

use in tandem is also examined.

The Empathic Person in More and Less Traditional Communities
Two contradictory lines Of thinking stimulate investigation
into this area. On the assumption that empathy is useful to and
developed from interpersonal interaction at the local level, one would
expect the empathic person to have more friends than others and per-
haps to be nominated more Often as an Opinion leader by his peers.
0n the other hand, to the extent that his aspirations and interests
orient him to activities which are primarily outside the community,
and particularly to the extent he finds himself without a chance of
effecting community change, one might anticipate that he would be a
marginal man, have few friends in the community, and not be nominated
as an Opinion leader. He might still, however, be regarded as a
source of information about what is going on outside the community.
Stating these notions as hypotheses:
l. The more empathic person will more Often be a marginal man in

communities which are more traditional, i.e., he will have fewer

to

local friends and be less integrated socially. But he will be
relatively well integrated and have more local friends in
communities which show less traditionalism.

2. The empathic person will be recognized as a useful source of
information about the outside world in.both types of communities
but more so in less traditional communities.

3. The empathic person will have higher status in less traditional
communities than in more traditional communities.

A. The empathic person will receive more peer nominations as Opinion

leader in less traditional communities.

Empathy_and the Mass Media.

 

As stated, what is important about empathy in national modern-

ization may be the individual's capacity toihink hypothetically.

The ability and propensity to think outside Of the ordinary tO what

may be strange, improbable, and surprising, lies behind attempting

the changes involved in modernization. However, some sort Of control
over imaginative flights Of fancy must be connected with the hypothetical
thinking if it is to be productive.

The mass media may have such a controlling effect on their
consumers. Lerner likes to call the media "mobility multipliers,"
thereby emphasizing that they make possible the expansion Of imagination
to distant times, lands, and peOple. (Lerner, 1958, p. 52 ff.) But
the mass media may also route imagination into common channels,
exercising a homogenizing influence upon those exposed to them.

Aside from a general shaping of imagination, the media may
provide direct and pragmatic control over specific imaginative schemes.

The print media may do this more than the electronic. Print is able

kl

to aim.messages in detail at specific audiences and to lay out step
by step instructions. The conventions Of literacy, more than the
conventions Of the electronic media, must make the semi-literate
painfully aware of the control exercised over him. In contrast,
messages in the electronic media are transitory, difficult to repeat,
and unsuited, as Spector found in his field experiment in Ecuador
(Spector, gt. al., 1963), for giving detailed instruction.

Assuming that electronic media are best at exciting the
imagination while print media are more suited to providing control
over it, what kinds Of differential relations with empathy are
possible? One simple relation would be a high correlation between
exposure to electronic media and those components Of empathy that
emphasize imaginativeness. This relationship is investigated in this
Istudy.

The relationship between empathy and exposure to mass media
may be considered from another standpoint. Instead of distinguishing
between types of media and their control capabilities, we may look at
the value of empathy in decoding media messages. The more empathic
should be more capable Of drawing modernization import from messages
in the mass media. In addition, the more empathic should apply these
messages to themselves to a greater extent.

Satisfaction. Mass media exposure on the part of the empathic

 

person who is in a traditional setting is likely to make him dis-
satisfied. The media present pictures Of different and possibly
better ways Of life. The more empathic decode this and apply it to
themselves more completely. Therefore, they should be more dis-

satisfied as a result Of media exposure.

1+2

Yet dissatisfaction has proven surprisingly weak as an
explanatory variable in past research with populations Of developing
countries. (See Bordenave, 1966, pp. l6h-l65) Perhaps this is because
a person may be dissatisfied for many reasons, not all Of them.making
for modernization. Perhaps it is also because a dissatisfied person
with the power to effect changes may report satisfaction with his
progress toward a goal; he feels he is making headway. It may be
that what should be considered is not whether a person currently
reports himself to be satisfied but whether he values'being satis-
fied, i.e. feels one should not strive over much for anything beyond
bare physical subsistence. Thus, a value for dissatisfaction may be
conducive to change and something taught by the mass media. The
teaching of resignation is traditional; the teaching Of restless
seeking for change and progress modern.

Aspirations. If dissatisfied and blocked by circumstances

 

from leaving the land, a farmer may hope that his children will
receive more education than he received, and that they will go into
a higher—status occupation. This hOpe will be more likely to the
extent he is empathic, and the high-status occupation selected by
the more empathic will more likely be modern rather than traditional.
Thus, the correlation between mass media exposure and aspirations
should be mediated by empathy, for the empathic more completely de-
code the implicit message Of intergenerational change and mObility.

‘Ealuation Of planning, attitude toward change, and concern with

 

prOblem.. To the extent that a person is seeking change because of
dissatisfaction and aspirations for improvement, he should also

positively value change, approve Of planning rather than Of leaving

1+3

events to chance, and be concerned with community prOblems and
their solutions. This he learns from media exposure,and he learns
it best if he is more empathic.

Political knowledge. The relationship between political

 

knowledge and media exposure should, again, be stronger for the
more empathic than for the less empathic. The less empathic
would, assumedly, use the media for other purposes than to gather
information about affairs which they feel are no concern of theirs.
Consequently, they would not become so knowledgeable politically.
A number of variables have been presented. In each case
it has been hypothesized that empathy mediates the relationship
between mass media exposure and the acquisition of modern orien-
tations or information. This mediation is related to improved de-

coding Of the modernizing content Of the messages.

[Cosmopoliteness and Empathy.

 

The relationship between empathy and this variable in the
modernization process should be similar to the relationship between
empathy and mass media exposure. Simply stated, more empathy should
make for more effective decoding Of urban experiences and consequently
for greater attraction to urban ways.

Analysis of cosmopoliteness in relation to empathy presents
a rare opportunity to prObe a possible antecedent and consequent
Of empathy. Two referents for cosmopoliteness may be identified:

(1) frequency Of trips to the city and (2) level Of attraction to
city life. We may suppose that in the beginning trips to the city

are undertaken for practical reasons. For some individuals urban

M4

contact leads to greater empathy. Mere empathy, in turn, should
make urban life more attractive and motivate efforts to visit the
city later. Other things being equal, a relationship between number
Of trips to the city and the felt attractiveness Of urban life is
to be expected. However, this relationship should be significantly
smaller for the less empathic if empathy is an important mediator of

urban experience. This hypothesis is outlined in Figure 2.

 

Figure 2. The hypothesized relationship between trips to the

city, urban attractiveness, and empathy.

Time 1: Trips to the city undertaken for whatever reason result,
for some, in more empathy which results in more attraction
to city ways.

Time 2: Trips to the city for the less empathic are due mainly
to practical constraints while for the empathic they are

due also to the feeling Of attraction for the city.

Therefore, the correlation between trips and urban attractiveness

is larger for the more empathic than for the less empathic.

 

Trips-—~——~> Empathy _%>Urban Attractiveness

 

VIII. A SUMMARY OF THE HYPOTHESES OF THE STUDY

Empathy and Social Structure

 

l. The more empathic person will typically have fewer local friends
if he resides in a more traditional community than will his

empathic counterpart residing in a more modern community.

#5

The more empathic person will more Often be regarded as a use-
ful source Of information about the outside world if he resides
in a less traditional community than if he resides in a more
traditional community.

He will have higher status if he resides in a community of lesser
traditionalism.

He will receive more nominations from.his peers as Opinion

leader if he resides in a community of lesser traditionalism.

Empathy; Media Exposure, and Other Variables

1.

Empathy items heavily loaded on imaginativeness will correlate
more highly with exposure to the electronic media than with
exposure to the print media.

Empathy will relate positively to a rejection of satisfaction
as a value. Cross-classifying on empathy will result in a
Significantly stronger relationship between mass media exposure
and satisfaction for the higher empathy sub-classes than for
the lower empathy sub-classes.

The same will hold true for media exposure and aspirations

for social ascent through one's children.

Positive attitudes toward change, higher valuation Of planning,
and higher concern for community prOblems and their solutions
will relate positively to higher empathy and higher media
exposure. When cross-classified on empathy the relationship

of these variables with mass media exposure will be stronger

for the high-empathy sub-class.

ho

5. Mass media exposure and political knowledge will be higher for

the more empathic than for the less empathic.

Empathy and Cosm0politeness

 

1. The relationship between frequency of trips to the city and the
attractiveness of city ways will be mediated by the individual's

level of empathy.

CHAPTER II
.METHDDOLOGY

I. BACKGROUND OF THE STUDY

In December, 1961}, the United States Agency for International
Davelopnent (AID) granted a four-year contract to Michigan State
university's Department of Commication to conduct a large-scale
study of the diffusion of innovations in developing countries.

Everett M. Rogers was the director of the project, and under his
leadership research sites were established in Brazil, Nigeria, and
India during 1965. The theoretic aim of the study was to determine
the degree of applicability to developing nations of research findings
obtained in the United States and other developed countries concerning
' diffusion of innovations. In anticipation that other variables would
be of particular interest and importance, theories of modernization,
particularly of the rural and agricultural sector of society, were
examined and the results of previous studies in underdeveloped areas
were considered. Pragmaticajly, the main purpose of the study was to
provide information which would accelerate the adoption of useful new
ideas in the rural sectors of developing nations.

The study was planned in three phases. The first considered
the village or community as the unit of analysis. A large sample of
villages was selected and their characteristics were described.*

The second phase considered the individual farmer as the unit of

 

4".l'he results for Brazil are available in a Phase I report.
(Whiting, et. al., 1967)

1+7

#8

analysis. Interviews with all farmers in selected villages or
communities were obtained in each country under study and an effort
was made to investigate the characteristics making farmers more or
less prone to early adoption of innovations and more or less likely
to be regarded as opinion leaders'by their peers. The third phase
of the larger study involves the administration of different commun-
ication treatments in controlled field experiments designed to
obtain evidence about the relative effectiveness and utility of some
practical and theoretically interesting alternatives Open to change
agencies.*

From.July, 1965 to December, 1966, the writer was resident
in.Brazil at the research headquarters in the city of'Belo Horizonte,
in the State of Minas Gerais, directing data-gathering and other
preparations for this research project. The data for the first phase
were gathered in January and February, 1966. The writer was allowed
to incorporate some of his measurements of empathy and its correlates
into the interview schedule constructed to measure the characteristics
of the informal community leaders in each of the communities sur-
veyed. The manner of their selection will be described shortly.

In July and August, 1966, the data for the second phase of
the larger study were gathered. Again, empathy was included as a
variable in the interview schedule, and this allowed the investigation
of its reliability and its relationship to the mass media'behavior

and innovativeness of a broader cross-section of individual farmers.

 

*In addition to the report already cited, a number of working
papers, progress reports, and resumes of the larger study are avail-
able. See the Bibliography for details.

#9

Empathy and the writer's particular research interests were not in
major focus during this phase, however, and the majority of the
analysis reported here is based upon data collected in the first

phase.

II. SELECTION OF THE SAMPLE

The sample of cultivators was drawn primarily for purposes
Of the larger study. This led to a somewhat complex system of
selection that guaranteed a large geographical variation and selection
from a large number Of communities, 76. In the course of describing
a community's characteristics, it was necessary to assess the character
of its leadership.

First, to Obtain a selection of communities where change
programs had been operative long enough to have had some impact,
the researchers constructed a list of all the local offices of ACAR,
the State Extension Agency, which had been in Operation for three
years or more. There were 78 such offices. Then, through consultation
with ACAR personnel, we determined that these Offices were scattered
through three regions, definable in terms of farm, market, and sub-
cultural characteristics. These three regions appeared sufficiently
different to justify, for the purposes of the larger study, strat-
ificatiOn of the sample to Obtain approximately equal numbers of
communities within them. Completing this stratification, 13, 1k,
and 13 local offices were randomly selected in each of the geographic-
economic-cultural areas of the State of Minas Gerais which could, on

the criteria of three years change agency activity, be included in the

50

sample.

Within each of these LO offices we asked the local agent to
designate the community or sector of work in the Office which was
the most responsive to the ACAR program and the community which was
the least responsive. Thus, we obtained initially 80 communities,
scattered across the state and representative of those locales where
the change agency had chosen to labor.*

In each local office we interviewed the local ACAR change
agent and the more important formal leaders from the two communities
he designated. The majority of these formal leaders lived in the
towns and market centers, but school teachers in the communities theme
selves were included, as well as a few important store-keepers. We
labelled these people "formal" leaders because they could be identified
by their jObs. While a part of the rationale for this approach was
to inform them of our purposes in visiting the communities, we also
requested all formal leaders to nominate individuals whom they con-
sidered influential in the diffusion process in the communities.

These same nominations were requested of the local ACAR agent. Following
a tabulation of these nominations, we selected about 10 individuals

for interview within each community, attempting always to interview
those with the largest number of nominations. These were designated

as "informal" leaders for the purposes of the larger study. They

 

*In one instance the ACAR change agent resigned the day'before
we arrived to interview; Without information from him the rest of the
data collected in the jurisdiction of his office was not comparable
with the data collected elsewhere and was, therefore, drOpped. In
another Office we found that the ACAR change agent had worked in only
a single community. Consequently there was no community for com-
parison and this local office was likewise deleted. 'We ended up
with 76 commities .

51

themselves were asked to designate influential farmers in the
community and a record was kept Of these votes. While about 10

per cent of those nominated by the formal leaders received no name
inations from.their peers, the majority did receive such nominations.
On the basis of a more complete census conducted six months later

in the second phase Of the research in 20 of the 80 communities, we
were able to determine that the correlation.between the number Of
Opinion leader nominations received from.peers during the first
phase and the number received from peers in the second phase was .6h.
Consequently, we may be fairly certain the individuals interviewed
in the first phase represented the leadership portion of the community.
All were cultivators, all were landowners, and many had had contacts
with the change agent.

In two of the ho local offices unforeseen events forced us to
throw out the data collected. This reduced the number of informal
leaders interviewed from 806 to 775 and the number of communities
represented from 80 to 76. The number of informal leaders interviewed
in each of the communities that remained ranged from 13 to eight,
the majority of the communities having over ten. This sample of
community leaders, then, may be regarded as representing the more
influential and established farmers in 76 communities located in three
different regions Of the State of Minas Gerais. Each Of these commun-
ities has‘been a site of change-efforts by the state extension agency,
ACAR.

The second phase of the study, already alluded to, selected 20
Of the original 80 communities for more intensive study. This time

the unit Of analysis was the individual farmer. Because Of our

52

interest in the Opinion leadership structure of these communities,

an effort was made to Obtain a census of all the cultivators who

were involved in the social life of the community and who owned farm
property. This excluded the few absentee landlords and all families
headed by farm laborers. The 20 communities were selected purposefully
rather than randomly out of the original 80 according to a set of
criteria intended to maximize their comparability and enable them to

be utilized in the field experiment which followed. These selection
criteria included the presence of a center in the community where
meetings could be held, the reception Of a common broadcast signal, and
non-interference with other programs planned by the change agency.

Only one community was chosen per local office; hence the spread was
about half as large as in the first phase.

Interviews were conducted in these 20 communities during.July
and August, 1966, and 1,302 usable interview schedules were prOperly
coded and available at the time the analysis reported here was run.
Seven schedules were eliminated due to interviewer or coder errors,
but these in no way effect the representativeness of the remaining

data.

III. INSTRUMENT CONSTRUCTION, INTERVIEWER TRAINING, DmTA

COLLECTION AND CODING: PHASE I

Observation and contacts in the rural areas of Minas Gerais
were made from.July to September, 1965. IDuring this time the writer
participated with Luiz Fonseca in testing and refining part Of the

instrument to be utilized in a study of rural milk producers in one

53

municipality of Minas Gerais (Fonseca, 1966). Indications Of
political knowledge and risk orientation were subjected to Guttman
scaling procedures and later incorporated, in part, into the inter-
view schedule for Phase I. The sample was also selected.

From OctOber through NOvember, 1965, the interview schedules
(originally five separate schedules, later four) were drafted in
English and translated into Portuguese. In the process the ideas of
other members of the diffusion project staff, particularly those
present in Brazil, were incorporated. In addition the schedules drew
upon earlier field studies done in Colombia (Rogers and Herzog, 1966),
India (Roy, 1965), and some of the ideas developed by Inkeles for his
'multi-nation study of modernization orientations (Smith & Inkeles, 1966).

The translated versions of the questionnaires were pre-tested
in December, 1965 and January, 1966. Three such pre-tests were con-
ducted, with extensive revisions following each of them, The original
versions were shortened to less than an hour average interview time,
their formats were improved for interviewer and respondent ease, and
numerous items were dropped which had turned out to be either incompre-
hensible to the farmers, essentially consensus items, or highly
correlated with other items intended to measure the same variable.
Continual effort was made to render the Portuguese of the instrument
in the dialect of the rural area.

The first two pre-tests were conducted'by Brazilian personnel
brought into the project for training as interviewer supervisors.

They were all graduates of four-year colleges. About 25 farmers were
interviewed by these men. The third pre-test functioned, in part, as

training for 20 interviewers recruited and trained during the latter

5h

part Of January, 1966, and about another 25 farmers were interviewed.
Everett Rogers participated in the final revision of the interview
schedules and gave much valuable advice.

The interviewers were recruited by circulating information about
the project and the opportunities it presented among students in the
Faculdade de Economia, Faculdade de Filosofia, and the Veterinary
School of the Federal University of Minas Gerais, students at the
Rural university of Minas Gerais, and students at the agricultural
college in Lavras, Minas Gerais. we asked those interested to fill
out forms listing their rural experience, age, academic background,
prior training in social science interviewing, and other necessary
information. Out of over 120 applicants, 29 were chosen for training.
The training process further eliminated a few. Eventually, five
teams were formed, provided with a vehicle and supervisor, and sent
into the field.

The data were collected during the last week of January and
the first three weeks of February, 1966. It was necessary to pause
for five days toward the end Of February for carnival. Following
this we were able to send several of the teams back to complete the

collection of data.*

 

IAbout ten weeks before we entered the field in Phase I, the
Camelot scandle broke in Latin America. Considerable feeling was
generated by this and many Latin scholars indicated they would no
longer OOOperate with UtS. researchers in their country. As far as
we could tell, information about Camelot did not reach the individuals
or institutions contacted concerning interviewing until after the
completion Of the Phase I data gathering. In the spring of 1966 the
Ramparts article, linking Michigan State university to C.I.A. activities
in South Viet Nam, reached into Belo HOrizonte. We thereby'became
linked in the minds of many of the students, and perhaps some of the
professors, with'both the Camelot Project and the C.I.A. generally.

55

During March, April, and May, 1966, the data were coded.
Coders were recruited initially from some of the original interviewers,
but this was not completely satisfactory because of scheduling
difficulties. Others were found to work at it on a more regular
basis. The majority of the items had been precoded and most were
fixed response. Some, however, were open-ended. PrObably the most
troublesome of these for the coders were the items requiring respond-
ents to specify their actions if they were to find themselves in the
roles Of President, Prefect, ACAR agent, or the poorest person in the
community.

Four coders were selected for training in the coding of
empathy. About 200 of the questionnaires were selected and answers
were culled for each of the empathy questions and grouped under a
single heading where they were essentially identical. Four training
sessions were held, each of them followed by independent coding Of

the items culled from the questionnaires. An abbreviated version of

 

This led to a noticeable decrement in interviewer motivation in

Phase II as contrasted with the situation in Phase I when it had

been remarkably good. It also made our day-to-day interaction with
Brazilian intellectuals somewhat touchy. Fortunately, we had several
things going for us, and the administrations of the Brazilian institu-
tions we were linked to did not seek to oust us but rather to improve
our communication with relevant others in their institutions.

A more ticklish prOblem.in the Phase I data collection, so far
as the rural populace was concerned, involved agitation over the
activities of the Brazilian Institute for Agrarian Reform (Instituto
Brasileiro de Reforms Agraria, "IBRA"). Farmers were very suspicious
of IBRA’s intentions. IBRA began its state-wide survey only a few
weeks before we began our interviews. Despite the farmers' natural
suspicions, we are relatively certain that the majority of them did
not associate us with IBRA after our initial explanations. Inter-
viewers introduced themselves as university students and spoke of
the project's link to ACAR. ACAR is a highly trusted organization,
as government organizations go, in the rural areas.

56

the criteria given coders for judging the responses follows:
If the answer is: code as:

Don't know, foolish responses, responses

that show respondent's inability to think

Of himself in the role, no response 0

Very general responses that show some

understanding Of the situation and the

actions he could take 1

Mbre specific responses with an indication

that the respondent was actually thinking

of himself in the role called for 2

Very specific responses where the individual

really puts himself into the role, and

makes good suggestions about carrying out

the Obligations of that role 3
After the fourth training session an.average inter-judge reliability
Of .80 was Obtained on the sum of the four empathy items.

All the data were transferred to data-sheets and sent to

Michigan State for key-punching in June, 1966. Error checks were
run on them during the summer and fall of 1966, eliminating illegal,

contradictory, and imprObable codes. A clean deck of data was

available by Nevember, 1966.

IV. INSTRUMENT CONSTRUCTION, INTERVIEWER TRAINING, DATA

COLLECTION AND CODING: PHASE II

Preparations for the construction of the Phase II instrument
were greatly facilitated by the production in April, 1966, of a document
(Keith and Rogers, 1966) outlining the core variables and suggesting
specific items for utilization in the instrument. Gustavo Quesada,
a.Brazilian project associate, utilized this document to draw up in

Portuguese the first rough draft of the Phase II questionnaire. Other

57

staff members contributed their ideas and the questionnaire was
pro-tested on about five farmers. Following this pre-test, the
questionnaire was revised and in early'June, 1966, taken to the field
for a week in a pilot study. From this pilot study Guttman scale
analyses were done for many Of the items and the wording and flow of
the questionnaire were reworked. While the pilot study was being
analyzed, two staff members were sent out to Obtain lists Of farmers
in the 20 selected communities eligible for inclusion in the inter-
views. These two also made rough maps Of the communities and located
each person to be interviewed. In addition, they pre-tested some new
items and some new versions Of Old items on an informal'basis with
the farmers. From these sources the final interview schedule was
drawn up. It required, on average, an hour to complete. It was an
improvement over the Phase I schedule because it utilized suggestions
on items supplied by project headquarters personnel, was drafted
originally in Portuguese rather than English (avoiding thereby both
the translation prOblem.and the difficulties of subtle biases assumed
when working in one language rather than another), went through a
pilot study (which was analyzed through Guttman scaling procedures
rather than simpler methods) rather than only pre-tests, and was the
only schedule in need Of preparation rather than being one of four,
as had been the case in Phase I.

Interviewers were recruited by contacting the more satisfactory
interviewers utilized earlier and making most of them supervisors.
They were asked to contact friends or acquaintances whom they judged
apt for the kind of work that they had experienced with us in Phase I.

Although a slight amount of nepotism and cliquishness resulted, the

58

caliber of interviewers was generally superior to what we Obtained
in Phase I. Interviewers were trained in the use of the question-
naire, both in classroom exercises and in field experience. Data
collection began in early July and extended through the first half
Of August. The completed schedules were coded over the next six
months.

Again, difficulty was experienced with the coding of empathy,
and no amount of training with the Old system of coding seemed to
improve inter-judge reliability to an acceptable level. After four
training sessions the coders reached an average agreement Of .65.
Therefore, a new set of instructions was drawn up, making more
explicit the stages of decision in the coding process:

Instructions to Coders on Codinngmpathy'Items
1. First, ask yourself if there is any evidence at all that the
respondent placed himself in the role called for. If not, code
his response as zero. He will receive zero if, for example:

a. he fails to respond or says he doesn't know

b. he answers in terms so general as to apply to any role

or to any human being, e.g., "I‘d be a good person."
(Nete: these usually are given in terms of cultural
values or attitudes generally approved in the society)

c. he answers in a completely irrelevant manner or in a way

which is clearly intended as a joke, e.g., "I'd take a
vacation."
2. If his response was general but does not indicate any
particular knowledge of the activities of the role, then he
receives a code of one for the response. You have, in other
words, only minimal evidence that he took the role as
directed. Examples of this kind of response are:
a. "I'd do what he's doing right now."
b. "I'd fu1fill his obligation and do his work."

c. "I'd act like he does."

59

One other type of response should also be coded as 1. This
will be discussed later.

3. If he provides evidence that he understands the general area
of work, or the type of thing that a person in the role does,
then he receives a code of two. Responses of this type are
considerably more specific than responses receiving only a code
of one. For example:

a. "If I were the Prefect of the municipio, I'd open more
schools and improve the roads."

b. "If I were President, I'd work to combat inflation."

c. "If I were in charge of a factory, I'd improve the
working conditions and increase production."

d. "If I were the ACAR supervisor I'd try to get more
credit and have it available all year."

NOTE THAT THE RESPONDENT DOES NOT SAY £193 as WOULD GO ABOUT
DOING THESE THINGS.

1+. If, in addition to these fairly specific statements of an
area Of activity, he indicates how he would accomplish the task,
he receives a code Of three. For example:

"If I were the Prefect, I'd help the farmers by getting
some tractors and cultivators for them to rent and by
trying to get some other seed companies interested in Open-
ing stores here."

5. There are two cautions: Don't let your own opinion of the
wisdom or correctness of the ideas expressed enter into your
judgment. If he says that he as President would, for instance,
do away with a certain political party or opponent, only con-
sider whether this action is within the role of the President,
not whether you would approve of it or not. This leads to the
second caution. If the person says he would do something fairly
specific in a given role but that activity is quite inappropriate
to that role (usually because it is not appropriate to the

scope Of activities in the role) then he should receive a code
of 1, even though his response is quite specific. He has shown
ability to take another role, but doesn't know enough about it
to take it accurately.

For example:

a. "If I were the Prefect, I'd raise the minimum price
supports on corn."

This is something which a Prefect has no control over; hence
the response is coded l.

60

In summary:
Code as 0 if there is no evidence the person took the role.

Code as 1 if there is evidence he took it but only very
generally or inaccurately.

Code as 2 if there is evidence he took it and mentioned a
fairly specific and appropriate activity.

Code as 3 if there is evidence he took it, mentioned a fairly
specific and appropriate activity, and mentioned how'he would
go about accomplishing it.

Don't let your own judgment of the wisdom or value of his

ideas enter into your coding.

After a single training session the average inter-judge
reliability among three coders on a random.sample of 80 questionnaires
was .92 for the sum Of four empathy items.

The data arrived at Michigan State for key-punching in February
and March, 1967. Error checks on all the variables reported on here

eliminated respondents when illegal or imprObable codes were encountered.

V. SCALING AND INDEX CONSTRUCTION

Mbst Of the variables involved in this thesis were grouped
into indices through factor analysis of conceptually distinct areas
and the calculation of factor scores on the basis of the varimax
factor loadings from an early and explicable factor solution. Six
such factor analyses were run. The factor score procedure will be
outlined in more detail when the results are discussed. It served
to provide Optimal combinations Of items to measure particular
dimensions. For the data from.Phase I, 20 of the 26 measures utilized
here are factor scores and one is a score from.a four-item Guttman

scale.

61

For Phase II data no factor-weighted indices were constructed.
Instead a simple sum of the empathy items, the actual magnitudes Of
responses in the area of mass-media exposure, and a simple sum.of the
number of innovations adopted'by the individual, were used. Hewever,
Guttman scaling of the interviewer judgments of skill with hypotheticals
was undertaken.

‘groducingGuttman Scales of Skill In Hypothetical Situations From
Interviewer Estimates

Thirteen items were selected for special attention from.those
utilized in Phase II. These were drawn largely from items intended
to measure economic knowledge, empathy, need-achievement, the attractive-
ness Of the city, and risk-orientation. In every instance the item
asked the respondent to imagine himself in a situation outside his
experience or to project himself into the future. The interviewers
were instructed to indicate whether they felt sure the respondent
placed himself in the situation in answering the question, felt sure
he did not place himself in the situation, or were unsure. The scores
Of 2, 0, and 1 respectively were given himey the interviewer in a
space provided to the left nf each of the items. After two brief
explanations one.rccuring before field experience with the question-
naire an: one, following such experience, during the interviewer
.jrdéining sessions, the interviewers appeared to understand the
instructions. Interviewer supervisors were especially asked to check
to be sure the interviewers were dutifully making the required judg-
ments. Upon completion of the interviewing,80 questionnaires were
randomly selected from the total sample and the ratings were sub-

jected to Guttman scaling procedures. Utilizing the original total

62

score of interviewer judgments as a basis for ordering individuals,
five Of the 13 items survived the Guttman scaling procedure. Included
among them.were all four of the role-playing items and one item
utilized in Obtaining a measure of need-achievement. The Character-
istics and content of the scale are given in Table 1.

After the five items which scaled on the initial try were
eliminated, the other eight were utilized to Obtain a new total raw
score for each individual. Individuals were ordered on this total
score and the scaling procedure was repeated. A second Guttman scale
emerged which also consisted of five items with a reasonably high co-
efficient Of reproducibility. The characteristics of this scale are
shown in.Table 2.

Three items did not fit into either of these two scales.

These were:

1. "If you had the Opportunity, would you choose to begin
life again in the city or in the country?"

2. "If a farmer had ten alqueires of corn and a million
cruzeiros to spend do you think he should spent it to
increase his production by buying more land to plant more
or buying fertilizer for the ten alqueires he already
has?"

3. "Let's play a game and look at this house plan. As you can
see, it has three rooms and in each Of these we have four
boxes. In the first room three of the boxes have Cr$50,000
each, in the second roam two of the boxes have Cr$200,000
and in the third room only one of the boxes has Cr$500,000.
wa, you don't know which are the boxes that have the money,
and in order to participate in the game you have to pay
Cr$5,000; then you will be able to enter one of the rooms,
and open only one of the‘boxes in it and keep the money you
find in the'box, if it is a prize one. Do you understand?
WOuld you like to participate in the game and pay the
Cr$5,000? If so, which room would you enter?"

The two scales correlate .hS with each other.

63

Table l. Guttman scale of skill in hypotheticals on Lerner-type

empathy items: scoring based on interviewer judgments *

 

Item Scale Cutt - Per cent Dichotomized
____, errors point correct between
"If you were in charge of a
factory, what would you do? 7 35% hh$ 0 & 1
"If you.were the ACAR agent,
what would you do?" h 50% 50% 0.& 1
"If you were the President
of Brazil, what would you
do? 8 62% 60% o a 1
"If you were the Prefect
Of the Municipality, what
would you do?" 7 69% 60% 1 a 2
"What do you intend to do
in the next three years?" 9 78% 60% 1.& 2

Coefficient of reproducibility = .91
Minimmn Marginal reproducibility = .60

Improvement over chance = 77%

 

*Based on a sample of 80 questionnaires. Items are listed
in order of increasing difficulty. NOte that interviewers scored
most of the respondents as placing themselves in the situation
called for.

**Per cent of the scale.

6h

Table 2. Guttman scale for skill in hypotheticals having to do

with economic items: scoring based on interviewer judgments*

Scale Cutting-
EEEE errors point

"DO you think that if all

farmers succeeded in in-

creasing their production

rapidly all would make a

better profit?" 9 32%

"If many people moved rapid-

1y from the farms to the large

cities and you stayed here,

would you receive better or

poorer prices fOr your

products?" 11 59%

"Let's suppose that every-
body in Minas had a larger
harvest than usual; would
this mean that you.would
make more money, less money,
or about the same amount of

money?" 12 6h%

"What would you do if you
had another 25 head of
cattle?" 8 79%

"new, let's suppose that I

gave you Cr$20,000 ($9) to be

used in buying lottery tickets,

and only for this; would you

prefer to buy 6 different

tickets, two blocks Of three

different tickets, or one

block of only one ticket?" h 89%

Per cent
correct

hl$

60%

60%

70%

80%

Dichotomized
between

 

O & 1

l & 2

1 & 2

l & 2

l.& 2

Coefficient of Reproducibility = .89 (.906 if item 3 is dropped)

Mdnimum.Marginal Reproducibility = .65 (.6h if item 3 is dropped)

Improvement over chance = 69%

 

IBased on a sample of 80 questionnaires.
order Of increasing difficulty.

**Per cent or the scale.

Items are listed in

65

VI. ANALYSIS PROCEDURES AND RATIONALE

The analysis will be presented in the same order and under

the same general headings as the Objectives Of the study.

The Measurement of Empathy

Factor analysis was selected to provide an indication Of the
empirically distinguishable aspects of empathy represented in the
data collected. The varimax rotation procedure was utilized since
the first principal axis solution did not account for a sufficiently
large portion Of the variance to warrant acceptance Of the single
factor solution. The varimax procedure works to maximize each item's
loading on a single factor so that each factor can be described by a
unique set of items. The criteria for the selection of a particular
factor solution included the meaningfulness and parsimony of the
factor clusters, the purity of the factor structure, its stability
across successive rotations, the reliability of the particular
structure as gauged by comparison with two sub-factor analyses where
the data were randomly divided and each half independently factor-
analyzed, and the per cent of the total variance explained.

The analysis Of the test-retest reliability Of empathy over
the five to six month period between the two data collections involves
only simple correlations.

The analysis Of the feasibility of a new measure Of empathy
utilizes the multiple correlation approach with successive deletion
of predictors to aid in determining the relative value Of the various
indices Of empathy obtained in Phase II. This approach takes each of

the empathy items or indices and forms a multiple correlation with a

66

particular dependent variable, e.g., innovativeness. The program
then examines the value of each of the predictors in fOrming the
multiple correlation and determines which is making the least con-
tribution. This one is deleted and a new multiple correlation is
calculated. The process is continued until a significance criterion
is reached. This criterion specifies the point at which the deletion
Of another predictor would result in a significant reduction of the
size Of the multiple correlation. The criterion chosen.here was .05.
Since six dependent variables are utilized it is possible to compare
the survival rate of each of the predictors across the six. In
addition to an intuitive analysis of their relative value, a Friedman
one-way analysis Of variance of the rank order of deletion Of the
variables is possible. With only six comparisons, the chances of

showing significant differences are slim, but worth testing.

The Empgthic Person in Mbre and.Less Traditional Communities

This analysis depends on Obtaining criteria to sabdivide the
76 communities of the sample into two clusters, according to their
traditionality. Then the correlation of empathy and the various
dependent variables (Opinion leadership, number of friends, etc.)
for individuals in the more traditional communities can be contrasted
with the same correlations for individuals in the less traditional
communities. The appropriate statistical test for the difference in
correlations utilizes Fisher's transformation for the difference be-
tween two correlations from.independent samples.

Empathy and the Mass Media

 

The relationship between a measure of empathy which emphasizes

imagination and exposure to electronic versus print mass media is best

67

tested'by partialling out effects of print exposure from.the
correlation between this aspect of empathy and radio exposure and
partialling out effects of radio exposure from.the correlation between
this aspect of empathy and print exposure. Then the difference be-
tween two partial correlations is checked for significance utilizing
the standard 3 test for the difference between two correlations with
a common variable from the same sample.

The analysis of the mediating effects of empathy in the process
of decoding modernization messages gleaned fromumass media exposure
and visits to the city is a bit more complex. Since the hypotheses
imply a causal direction (e.g., mass media exposure causes more
positive attitudes toward change) regression analysis is more appropriate
than correlation analysis. As Blalock points out, in working with
causal models

our interest in the magnitude of‘r is only as an indicator Of

how well we have estimated the slope or as a measure of how'

‘much error or dispersion we have. (Blalock, 196k, p. 101)
Were we to cross-classify the total sample on level of empathy and
compare the magnitude of the correlations between mass media exposure
and modernization attitudes and orientations under the hypothesis
that the correlations would be higher for the more empathic, we would
be testing, in effect, the supposition that the ggodness-of-fit Of
the data around a least squares line is better for the more empathic
than.fOr the less empathic, not that the impact Of units of mass media
exposure on modernization orientations is greater for the more
empathic. It is the regression line which indicates the amount Of
change on the dependent variable which can‘be expected fOr each unit

change on the independent variable. Therefore, the regression co-

68

efficients of separate sub-samples obtained by cross-classifying on
empathy should be compared in terms of their slopes and intercepts
with the expectation that the more empathic will gain more from each
unit of mass media exposure (have a steeper slope) and/or intercept
the ybaxis at a higher level.

In addition to more adequately testing the hypothesis that a
mediation effect is attributable to empathy level, the regression co-
efficient is less sensitive to restriction Of variance in independent
variable as we move across levels of empathy, and less sensitive to
unknown and unmeasured nuisance variables. On this point Blalock
writes:

A.major advantage in comparing slopes, as contrasted with
correlations, is that even though two samples may not differ
significantly with respect to variations in the independent
variable, they may do so with respect to unknown nuisance
variables. If one sample happens to be more heterogeneous
with respect to such variables, the size of r will be
relatively smaller, but the magnitude of'b s ould be un-
affected, except that the sampling error 11 be large.
(Blalock, 1961+, p. 118)
Blalock further argues that if the nature of the relationship among
variables is at issue, then attention should be focused on comparison
Of the apprOpriate regression coefficients rather than on correlations,
which only provide an estimate "of scatter about the least squares
equation and hence an accurate estimate of the true slope."
(Blalock, l96h, p. 51)

The advantages and disadvantages of testing hypotheses through
the use Of correlations versus the use of regression coefficients are
complex.and not completely understood. While Blalock's treatment is
the most thorough available, it does not provide complete guidelines

for error-free decisions. Since the hypothesis Of empathy as a

69

facilitator of the decoding of modern messages appears to involve
the assertion that the nature of the relationship of the dependent
and independent variables differs at different levels of empathy,
and since regression coefficients are likely to be less sensitive to
certain artifacts, the comparison of regression slopes was selected

to test the hypotheses.

VII. CCNCLUDENG COMMENTS

The hypotheses of the first chapter are tested by data gathered
through the survey method in rural Brazil. The advantages of the
survey method in terms of representativeness and generalizability'to
a real population may be overshadowed by the disadvantages of being
unable to (1) actually manipulate independent variables and examine
the effect on dependent variables, (2) check the outcome under different
levels of induced empathy, and (3) subject it to rigid controls for
extraneous variables. While these shortcomings are serious, and.must
make the results very tentative, the manner in which a wholly appropriate
experiment could be conducted is not clear to me. Perhaps with more
time, a larger budget, and considerable experimentation with portions
of the hypotheses or analogues of the variables in controlled laboratory
situations, an effective field test of the hypotheses could have been
mounted. Perhaps initial exploration with the sample survey technique
allows a useful starting point.

There may be some Objection to the utilization of parametric
statistics on data for which interval levels of measurement have not
been established. Perhaps the best defense is that most of the

analyses of greatest interest would not have been possible with

70

available non-parametric techniques. .A second point is that many of
the departures from.measurement assumptions probably worked to make
it more rather than less difficult to demonstrate hypothesized
relationships. If false positives are more dangerous than false
negatives, the error in this instance was in the proper direction.
The inadequacies of available alternative approaches should not be
ignored. Blalock's comments in this area are apt.
While it is true that the use of’Pearsonian correlations

and regression coefficients is not strictly legitimate unless

one has at least an interval scale, it may turn out that it

is no more misleading to make use of dubious assumptions about

level of measurement than it is to make use of data involving

arbitrary cut-points or ordinal scales that obscure differences

in amount of variation. For exploratory purposes it may not be

entirely unwise to make use of the rationale develOped in

connection with higher levels of’measurement, even where actual

measurement cannot be nearly so precise. (Blalock, l96h, p. 9h)

It may also be objected that the assumptions of linearity'

utilized in the analysis may be unrealistic, and indeed, we will see
that some of the outcomes seem to be more explicable in terms of
non-linear models. However, the main effort in this exploratory study
is to determine how well the data fit the simpler linear model; this
does not imply a'belief that the world is exclusively linear, Just a
belief that it is easier to start by seeing to what extent that model
fits. Similar comments could be made to Justify the use of orthogonal
varimax rotations rather than the oblique rotations which probably are

more isomorphic to underlying realities.

CHAPTER III

RESULTS
I. THE FACTOR ANALYSES OF EMPATHY AND OTHER MEASURES

Factor Analysis of 18 Items Intended to Tap Aspects of Empathy
The items involved, their scoring, and the distributions of
marginals are given in Appendix A. There were four role-playingg#<::)
items, e.g. "What would you do if you were President?" which required
coding by Judges after the answers were recorded by interviewers.
In addition we asked the interviewee how'mmeh he knew about the role
he had been asked to play and whether he had given much thought to
this kind of question before. A number of questions dealt with his
(:Ei/ ‘interaetign_with a foreigner or stranger. Some quizzed him.about the
extent to which he had day-dreamed about himself in other roles.
His ability to visualize a situation where his cows suddenly started
producing unbelievable amounts of‘milk was inferred from.the
originality and presumable utility of'his plans for taking advantage
of this windfall. He was also asked whether he could imagine someone
(::> believing differently than he did about some matter of importance to
figflJfitihimm All of these questions were intended to gauge the flexibility
o

5

6,90 of his thinking and the extent to which he was willing or able to

 

think outside the ordinary, place himself in new roles, and react to
new situations. In addition, ratings of his cooperativeness with the
interviewers (an indication of his behavior with strangers), and of

the degree to which he had opinions he was willing to express on the

questions asked him in the other portions of the questionnaire, were

71

72

utilized in the factor analysis.

The correlation matrix for these 18 items was based on 775
observations, hence the correlations are relatively stable.* The
first principal-axis factor accounted for only 16 per cent of the
variance. Extracting more factors and rotating them according to
the varimax.procedure a five-factor solution was reached before

stopping at the KielAWrigley criterion. This solution explained

 

hh.5 per cent of the total variance. For several reasons this was
not the solution chosen. These reasons can be made more understand-
able by reference to Figure 3, where the structure of the factor
solutions is depicted. While the two-factor solution leaves five

of the items unexplained, in the sense that all their loadings are
smaller than .hO, the three-factor solution picks up all the variables
except fer items 7 and 1%. Furthermore, the extra factor formed in
the three-factor solution remains stable from there on out. The
first factor shown in the two-factor solution also remains stable,
although it begins to have substantial loadings on the fourth factor
when the four-factor solution is reached. This kind of contamination
is not found in the three-factor solution. In short, the three-
factor solution was chosen because (1) it explained as many of the
variables as were explained in any of the solutions up to the Kiel-
Wrigley criterion, and (2) it contained the stable factor structures
before they begin to merge into each other or break apart. There

was a third reason for preferring the three-factor solution. When

 

*The .05 confidence limits are about:t.07 for the Fisher trans-
formed r. Since most of the correlations are below .25, the trans-
formation has little effect.

73

Figure 3. The structure of varimax rotated solutions for 18 variables

intended to measure aspects of empathy

Two-factor solution
(2h$ of the variance

 

 

 

explained)
Factor Factor
Net e lained I *QEL
'(l')'(2§E-7')(8)Tlh) K9)(lo)(ll)(12)(l3j 3 5 6 l 16 l 18

 

Three-factor solution
of the variance
explained)

   

Factor Factor

Four-factor solution
(39% of the variance
explained)
Factor Factor Factor
I

NOt e lained I 41y
ll)11§;

. 10 . ll 12 13

---q- ---------—----—

 

 

       

 

Five-factor solution I
{55.5% of the variance
explained)
Factor Factor Factor FactO' Factor

 

 

N t e lained I IV II, Y
W- K9) (105]!K12)|EK11713)I [12)(16)£17)I VIZ2)(3)(1+)(8)| Kl)(fl(6)(7)l

Note: The numbers in parentheses are the variables. The boxes represent
factors. Dotted lines indicate variables with loadings in more
than one factor. The arrows show the paths of individual variables
or clusters of variables from one solution to another. Variables
not explained are those with loadings of less than .ho on all
factors. The numbering of the items in Figure 3 is identical to
the numbering of them in Appendix.A.

7h

the data were randomly divided into two equal halves and these halves
separately factor analyzed the results for the two halves were similar
to each other and to the overall factor analysis in the structure of
their factors and approximate magnitude of their factor loadings
until the four-factor solution was reached. At this point the two
halves seemed to begin to diverge from each other and from the over-
all analysis. Thus, we may be relatively secure in the assumption
that the structure emerging in the three-factor solution is not random,
but beyond this doubts may arise. The three-factor solution accounts
fer only 32 per cent of the total variance. It was Judged better to
forego the extra 12 per cent which would have been added'by taking
the Kiel#Wrigley criterion rather than put up with a structure which
was less reliable, required two more factors and was more difficult
to interpret. Table 3 presents the variables constituting the three
factors, their factor loadings and factor weights utilized in the
construction of factor scores. These weights were calculated by the
formula: 'W = r/l-r2 where r is the factor loading. This is an
approximation of a much more complex calculation routine where the
loadings would be utilized as predictors to obtain the best possible
multiple prediction of the underlying hypothetical dimension
constituting the factor. According to Burt this procedure does a
better Job of estimating the contribution of each item to this pre-

diction than.any alternative short of the exact solution.

Factor Analysis of 15 Items, Among Them the 18 Empathy Items
This analysis includes the 18 empathy items in a larger pool
with an additional 57 items, many of which were utilized in.measure~

ment of other variables. Within this context all the empathy items

75

Table 3. Loadings and weights on the three-factor solution

for 18 items tapping aspects of empathy

Factor 1: Interaction with strangers and role knowledge

 

Factor

 

 

 

*
Item Loading Voight

"would you know what to say to himR"

(a foreigner you.met for the first time) .72 1.h7l
"Do you know what the duties of the President
of'Brazil are?" .67 1.220
"Do you know what the duties of the Prefect
of the Municipality are?" .65 1.133
"Would you'be very interested in getting to know
him (a foreigner or stranger) better?" (if yes)
Do you'believe that, given sufficient time,
you would understand why this foreigner thought
and acted differently than you?" .58 .882
"Let's suppose that you.met a foreigner for
the first time. If he looked friendly...do you
think that you would feel at ease with this
stranger?" .52 .712
Factor 2: Facility in dealing with hypotheticals
"If your cows suddenly started giving ho liters
of milk at each milking, what would you do?" .39 .467
"If you were President onBrazil, what would
you do?" .h6 .591
"If you were Prefect of this Municipality,
what would you do?" .50 .668
"If you were the ACAR supervisor, what would
you do to improve the work?" .h7 .609
"How cooperative was he?" (Asked of the
interviewer and scored 1-7) .h8 .621
Number of opinions expressed on 13 issues .55 .796

Number of Opinions expressed on 7 hypothetical
situations .57 .839

76

Table 3 continued.

 

 

Factor *

Item Load! 'Wei t.
Factor 3: Imaginativeness
"Have you ever thought what it would be like to
live in your great-grandfather's time?“ .63 1.036
"Have you ever imagined yourself being a
great hero?" .71 1.h08
"Have you ever imagined yourself being
President of’Brazil?" .61 .972
"Have you ever thought of questions like
these before?" (asked after the role-playing
questions) .57 .833

Items not included in any factor

 

"If you were the poorest person around here,
what would you do?" (Highest loading: .28 on
factor 2)

"N w think of something which you strongly
believe. (PAUSE) Have you ever thought that
someone could believe about it differently?"
(Highest loading: .2h on factor 1)

 

*Weight is defined by the formula W - r/l-r2 where r is the factor
loading.

 

tend to cluster into a single factor'by the time the Kiel-wrigley
criterion is reached, but the factor is defined much more by the
imaginativeness items taken together than by the Lerner-type role-
playing items alone. There are some exceptions. The degree of role
knowledge a person claims loads off on the first (and strongest)

factor together with education, political knowledge, frequency of

77

reading, memberships in organizations, and two small intelligence
tests, one utilizing arithmetic problems and other antonyms. While
claims to role knowledge do have their largest secondary loading with
the factor including the empathy items, they appear to be much more
a function of education and general knowledge and skill levels.

Another exception is the score on Opinionatedness, which
loads off away from the empathy factor (to the extent it loads any-
where) and attaches more to a factor having to do with contacts with
the ACAR change agent, loans made by him to the farmer, confidence in
him as measured by'a Cantril ladder scale, the individual's behavior
if he were the ACAR agent, and the ratings received from the inter-
viewer on c00perativeness, comprehension, and intelligence.

If we choose a factor loading of .hO as a lower limit fer
saying that an item is explained'by a factor, then none of the Lerner-
type empathy items and none of the interaction-with-strangers items
are explained at the Kiel-wrigley criterion in this factor analysis,
even though they have their highest loadings with the items indexing
imagination. Even in going one solution beyond the.KielAWrigley
criterion we do not produce high loadings for the Lerner-type items.
Instead they function as the weaker sisters of a factor which, though
itself weak, appears to consist of imagination or, more crudely,
reflective thought. The items defining this factor, and their loadings,
are shown in Table h. The Table would have to be considerably extended
to take in the small loadings for items like "shows ability to take
Prefect's role" (.2h), "shows ability to take the role of the poorest
person" (.21), and "says would feel at ease with a stranger" (.23)

even though these are the highest loadings these three variables
achieve in this factor analysis.

78

Table h. "Thinking" or "imaginativeness" factor from the sixefactor
solution of a factor analysis of 75 items which

include 18 empathy items

 

Item Lease
Has imagined self as a great hero .52
Thinks about prOblems .h9
Daydreams about great-grandfather's time .h?
Has imagined self as President of’Brazil .hS
Thinks about solutions to problems .hh*
Knows the Prefect's duties .37
Shows ability to take the President's role .36 5‘
Knows the President's duties .35
Has thought about role-taking questions befbre .33‘
'Wbuld like to get to know a stranger better .33
Thinks himself influential in the community .31
Says he would know what to say to a stranger .29

 

*Has a higher loading elsewhere in the factor analysis

 

'Empathy as a CommunitybLevel Characteristic

While the four customary Lerner-type empathy items did not
appear to do well when considered individually and at the level of the
individual respondent, they did emerge as an important predictor when
added together into an index and aggregated to the community level.
When this score was submitted to factor analysis together with other
scores similarly aggregated from the original data, which were taken
to be descriptive of the characteristics of the community's leaders,
empathy emerged as an important predictor on the second factor. This
factor involved other variables indicative of skill in interpersonal
relations and a higher level of community integration. The first

factor in this analysis consisted of the variables of literacy,

79

education, political knowledge, frequency of reading and so forth--
the familiar modern-knowledge-and-skills syndrome. The content

of the second factor is shown in Table 5. The sample size is 76

 

Table 5. Factor loadings and items defining a community
measure of interpersonal skill and group harmony among

informal leaders at the community level

Item Loadigg

Average informal leader score on trust .67
Average informal leader score on empathy .6h
Average informal leader score on opinionatedness .6h
Average informal leader utilization of radio for *

farm news .57
Average informal leader evaluation of modern infor-

mation sources as more credible than traditional .57
Average number of best-friend nominations exchanged

among informal leaders .55
Average feeling of self-efficacy in controlling

community change .h8
Average interviewer estimate of respondent

cOOperation .hl

 

*The utilization of radio for farm news shows a high
loading on this factor. When the data were analyzed on an
individual basis rather than community, the individual level
of radio use showed no sizable loading anyWhere. Its best
loading did appear with the variables composing the factor
shown in Table h, but the factor loadings were only .2h for
utilization of radio for farm news and .21 for general radio
use. Strikingly, and consistent with other research, radio
utilization refused to cluster with the utilization of other
ferms of mass-media.

 

rather than 775 and the measures typically are appropriate central

80
tendencies aggregated from individual measures.*

Factor Analysis of 26 Indices, Including the Three Indices on Aspects
of Empathy

Purely in the interest of a search for relationships, the indices
constructed out of the responses of the 775 persons interviewed in
February and march of 1966 were subjected to factor analysis. Twentye
six variables are involved and the majority of them were formed by
factor weighting procedures. For example, there are the mass media
indices from.a factor analysis of items in this area, the scores on
modernization orientations from.their factor analysis, scores on
literacy and intelligence from a factor analysis of items in these
areas, scores on the three aspects of empathy selected from.the factor
analysis already described, etc.

The Kiel-Wrigley criterion for st0pping rotation was reached
at the six-factor solution. Had we excluded variables which lack a
loading of .hO or greater on any factor, the three-factor solution
which accounts for 37 per cent of the total variance would have been
the KieleWrigley stOpping point. In the interests of parsimony we
will examine this first with only a few later references to the four-
factor solution.

The first factor is composed almost exclusively of'mass media
exposure. This is partly artifactual in that several scores were
constructed utilizing the data from the earlier mass media analysis
and some of these scores had components common to them. Granting that
the formation of this factor may be somewhat predetermined thereby,

it is still interesting to note which variables are pulled away to

 

~- *~*Fo£ furtherdetail about the comirmnity'138-33“d analysis, see
Whiting, SE.- §_l_., 1967-

81

load together with mass media. This is shown in Table 6. It is worth

 

Table 6. First factor from a three-factor solution to a factor
analysis of 26 indices constructed fer testing

hypotheses in this study

 

Item Loading
Utilization of literacy and cinema .93
General media use .93
Mass media exposure .91
Favorable to change .68
Objective status .60
Perceived status .h9
Intelligence .h3
Social memberships and social integration .hO

 

noting that intelligence, while it has its highest loading on this
factor, has other loadings elsewhere which are almost as high. In
other words, intelligence is factorially complex in this particular
context of variables. The two measures of status and the measure
of attitude toward change show their highest loadings on this mass media
factor.

The second factor consists entirely of contact with the ACAR
change agent and sociometric popularity. It is shown in Table 7.

The third factor appears to pick up some kind of cognitive
dimension involving facility in dealing with strangers and a'belief
in the efficacy of work rather than luck in getting on in the world.
Its loadings are alas Shown in Table 7.

If we were to move to a four-factor solution (explaining

82

Table 7. Second and third factor from a three-factor solution
to a factor analysis of indices constructed for testing

hypothesesrin this study

Second factor:

 

Item Loadigg

Number of nominations received from.peers
as a source of information about the

 

 

outside world .75
Number of nominations received from peers

as best friend .72
Number of nominations received from.peers

as opinion leader .68
Involvement in ACAR programs .67
Knowledge about and trust of the ACAR agent .50

Third factor:

Belief in work rather than luck .71
Facility in dealing with strangers and

role knowledge .71
Political knowledge .I+7
Imaginativeness .h5
Skill with hypotheticals .ho

 

h2.5 per cent of the variance) the fourth factor would be formed by
splitting the first two items factor three away from.the rest. They
would then have higher loadings, but the loadings of the other three
items in factor three would also become higher (except for political
knowledge, which would now be spread about equally across three factors,
the first, the third, and the fourth). In addition, this new fourth
factor would pick two variables out of the group of variables unexplained

in the three-factor solution: the use of mass media for agricultural

83

information and functional literacy. It would also largely steal
away from the first factor the variable of intelligence. The degree
of knowledge about and trust of the ACAR agent would also be stolen
from the second factor. Many variables would continue to have low
loadings on all factors. These include the number of godchildren the
person reports having, his aspirations, his value for dissatisfaction,
his exposure to urban living, his preference for the city life, and
his tendency to think about community problems. The composition of

this fourth factor is given in Table 8.

 

Table 8. Fourth factor from a four-factor solution to a factor
analysis of indices constructed for testing hypotheses

in this study*

 

Item Loading

Skill with hypotheticals .72
Knowledge and trust of individual ACAR

agent .58
Intelligence .57
Imaginativeness .h7
Literacy .h6
Use of radio for agricultural information .h2
Political knowledge .37

 

*

This factor consists mostly of portions of factor
three from.the three-factor solution and variables which
were formerly not well explained.

 

Were it not for the fact that the third factor in the four-factor

solution has only two variables with relatively high loadings, the four-

8%

factor solution would be preferable to the three-factor solution

since it picks up literacy and manages to load skill with hypotheticals
quite purely onto the last factor. Political knowledge would then
load on three factors, the first (mass media), the second (belief

in work, skill with strangers and role knowledge), and the third
(hypotheticals, trust of ACAR agent, and intelligence). It is included
in Table 8 partly because as successive factors are extracted it

achieves its highest loading together with the variables in.Table 8.

II. THE RELIABILITY OF MEASURES OF MATH! OVER TIME

In this section we consider how well individual measures of
empathy hold up over time, how well an index composed of three
empathy items intercorrelates with itself over time, and the test-
retest reliability of empathy compared with that of other variables
such as age, education, literacy, etc. The sample size for all these
comparisons is 102, comprising individuals reinterviewed in 20 of
the original 76 communities after a lapse of six months.

There were three items tapping empathy'both times:

1. What would you do if you were President olerazil?

2. What would you do if you were Prefect of this Municipality?

3. What would you do if you were the ACAR agent?
The context of the items changed somewhat in the two questionnaires
and the scoring procedures for the second set were changed somewhat.
The test-retest correlations for these three items are: (1) if
President, .17, (2) if Prefect, .3’-|-, and (3) if ACAR agent, .29.
When the three items are summed into indices the intercorrelation of

the two indices is .29.

85

For this sample size any correlation larger than .195 is
significant at the .05 level. However, a correlation accounting for
about nine per cent of the variance between essentially identical
measures at two points in time would seem to doom the usefulness of
that measure. If the variable itself fluctuates this much over time,
then it cannot be very useful in predicting anything.

In order to obtain some perspective in this, it is useful to
examine the test-retest correlation of other variables measured at
both times. The test-retest reliability coefficients for 17 other
variables from these two surveys are given in Table 9.*

Looking down Table 9, we find that measures of mass media
utilization are relatively reliable. Indeed, attendance at movies
happens to be as reliable as such matters as age or possession of a car.
The estimate of education is surprisingly unreliable as is the estimate
of literacy, even though this was based on a test performed by the
interviewer with the subject. If the respondent's word is taken for it,
the reliability of literacy estimates is very low indeed. Although
there is fair consistency in claims of belonging to a cOOperative,
the number of memberships claimed in formal organizations barely reaches
significance. The modernity of occupational aspirations for children
does not reach significance.

In general, with the exception of the estimates of mass media

exposure, the most reliable items appear to be those which are very

 

*It should be noted that the correlation for trust and cooperative-
ness was derived from.indices containing different items at the two
points in time. The correlation is based on standard scores from
each of the two indexes.

86

Table 9. Test-retest reliability of 17 variables

over a six-month period

 

Test-retest
Variable correlations
Index of trust and c00perativeness .09
Modernity of occupational aspirations for
children .10
Claims to be able to read a newspaper .18
Number of memberships claimed in formal
organizations .21
Literacy test based on the correct reading
of a paragraph .39
Number of conversations reported with the
ACAR agent .hh
Number of years of education .h8
Innovativeness score based on the adOption
of six practices .5h

Political knowledge based on a four—item

Guttman scale dealing with factual

knowledge of political events .59
Claims to belong to a cooPerative .
Claims to have resided outside the community .61
Number of magazines or newspapers read

per month .63
Has piped-in water in his home .7h
Has electric lights .75
Has a car .82

Age in years .85
Number of movies attended per year .

 

concrete and which require scarcely more than a yes-no response. One
exception to this is the reliability of self-professed ability to
read. Among the variables for which we have data, the empathy does
better than any of the subjective variables and better than the sub-
Ject's estimates of the number of his memberships, although the latter
difference is not significant. The reliability of the literacy test

is not significantly different from.the reliability of empathy'but

87
the number of years of education is significantly more reliable.
III. THE FEASIBILITY OF A NEW MEASURE 0F EMPATHY

Two Guttman scales were constructed from interviewer estimates
of the respondent's ability to handle counterfactual questions. The
first scale is dependent upon interviewer estimates of interviewee
performance on questions used to obtain an.index of empathy. The
second scale utilized hypothetical economic items. Each scale had five
items. The intercorrelation of the two Guttman scales is .h5. The
first scale, based primarily on interviewer Judgments of performance
on empathy items, correlates .70 with the empathy index derived from
coder Judgments of these items. The second scale, based on inter-
viewer Judgments of performance on other items correlates .38 with the
empathy index. (The sample size for all correlations presented in this
section is 1,302.)

Our empathy index, with which the scales will be compared, is
composed of the sum of scores given‘by coders to responses to four
questions:

What would you do if you.were President of'Brazil?

What would you do if you were Prefect of this Municipality?
What would you do if you.were the manager of a large factory?
What would you do if you were the ACAR agent?

To test the relative adequacy of these scales as compared to the
usual measure of empathy, we placed the three as predictors in multiple
regression equations predicting six criterion variables in turn:

(1) the number of visits made to the city, (2) the frequency of radio
listening, (3) the frequency of television viewing, (h) the number of

newspapers and magazines read, (5) the number of movies attended, and

88

(6) the number of agricultural innovations adopted by the individual.
The comparative adequacy of the scales versus the index was Judged on
the basis of the order in which they were deleted as successive
predictors were dropped from the multiple regression equations.

Table 10 presents the rank order of deletion of each of the
three predictor variables on each of the six modern'behaviors. In
addition, it indicates the number of variables left when deletion
stopped at a .05 level of significance and the multiple correlation
after deletion was stopped. Predictors which are retained longer have
higher ranks in Table 10. The sum of these ranks is given on the
right hand side of the Table.

Unfortunately, the interpretation of Table 10 is not unequivocal.
Scale A and the empathy index turn out to be equally important in
terms of the point at which they are deleted in the prediction of
these six variables. One interesting difference is that Scale A is
retained in the final solution every time whereas the empathy index is
twice dropped. Scale B is dropped three times but happens to survive
longest in the multiple prediction of the dependent variable with the
highest multiple correlation coefficient, that predicting adoption.
Another difference between Scale A and the empathy index is that Scale A
seems to have more usefulness on variables for which relatively small
multiple correlations are develOped, whereas the empathy index does a
little better with three mass media variables. If the two scales are

compared with the individual items composing the empathy index a.more

 

decisive pattern emerges. Though it is somewhat unfair to compare the
scales with individual items as we can see from.Table 11, which out-

lines the six predictors and the six criterion variables in the same

89

Table 10. Rank order of deletion of three predictors of six

modernization.behaviors

Modernization variables

Predictor Films City Papers Radio TV Adop- Sum of
variables attended visits read hours hours tion ranks

 

Scale A (based on empathy
items) 3 3* 2* 2* 2* 1* 13

Scale B (based on other
hypothetical items) 2 2* l l 1 3* 10

Empathy index (based on
the sum of coder scores
responses to h empathy
items) 1 1 3 3* 3* 2* 13

Multiple correlation
after deletion to the
minimum significance
criterion of .05 .12 .20 .22 .26 .26 .ho

 

*This variable was retained in the solution

 

faShion as Table 10, the fact that Scale A survives to the last on all
predictions should not be ignored. But of perhaps greater interest

are the contributions coming from the individual items composing the
empathy index. Items having to do with the roles of factory manager
and ACAR.agent seem to be more useful predictors than the two political

roles -- Prefect of the Municipality and.President of'Brazil. The two

Table ll. Rank order of deletion of six predictors of six

Predictor
variables

Scale A

Scale B

If factory manager

If ACAR agent

If Prefect

If President
MMltiple correlation
after deletion to the

minimum significance
criterion of .05

 

9O

modernization behaviors

Modernization variables

 

Films City Papers Radio TV
attended visits read hours hours
6* 6* 6* 6* 6*
1+ 5* 3 3 3
5* 1+ 5* 2 5*

3 1 h* 5* h
2 3 1 li* 2
l 2 2 l 1

. 11+ . 2o . 22 . 27 . 27

*This variable was retained in the solution.

Adop-

.1222.

6*
h.”-

HUOU'I

Sum of
ranks

36
22
23
22

15

 

V

political roles are dropped from.the prediction 11 times out of 12

whereas the two non-political roles are dropped only six times out of

12. The difference is significant on a Friedman one-way analysis of

variance in which only the ranks of these four variables are considered.

Although it is not the focus of this section to examine the level

91

of prediction, it is interesting to find that three measures of
hypothetical thought manage to explain 16 per cent of the variance in
the adOption of agricultural practices. Table 12 shows the pertinent

characteristics of this prediction.

 

Table 12. The prediction of adoption from three measures

of hypothetical skill

 

 

Contribution
Predictor Correlation to multiple R
Scale A (interviewer Judgments of
the degree of role-taking on empathy
items) .36 38%
Scale B (interviewer Judgments of
the degree of assumption of the
hypothetical situation on economic
items) .29 25%
Empathy index (the simple sum of
four empathy items) ~35 37%

Multiple correlation :- .hO

 

»' pIn percentage terms, Scale A and the empathy index make about
equal contributions to the multiple prediction of adoptions despite the

high level of intercorrelation'between them.

IV. THE EMPATHIC PERSON IN MORE AND LESS TRADITIONAL COMMUNITIES

In order to test the hypotheses of this section, criteria were

needed for subdividing the sample of 76 communities, from which the

92

775 respondents were drawn, into communities high and low on tradition-
alism. To accomplish this, thirty-three variables purporting to provide
measures and estimates of community wide deve10pment levels were sub-
mitted to factor analysis. A three-factor solution was selected as
intelligible and reasonably stable. The items, factor loadings, and
factor weights utilized in construction of factor scores for communities
are shown in Table 13. Since the items were first normalized and then
summed according to the weights derived from the factor analysis it is
difficult to paint a composite picture of the 'typical' traditional
community under a given set of items or of the range of traditionality
within our'sample. Table 13 can.best be read in terms offlmpgg
traditional communities having fewer institutional facilities, fewer
channels of communication with the outside, less developed medical
facilities and fewer medical personnel, more health problems, more
dependence upon a single cr0p, and a larger percentage of cultivators
who are not landowners. The lggg traditional communities would have
more facilities, channels, access to a wider geographic area, somewhat
better health and educational facilities, more diversification of
crOpping,and more land-owning cultivators. The difference between

more and less traditional communities is one of degree. None of the
communities involved could be termed modern as a‘U.S. rural community is
modern with extensive paved roads, electrification, telephone system,
bussing of ahildren to consolidated schools, etc. Thus, this discussion
consistently refers to degrees of traditionality rather than to a

traditional-modern dichotomy. In addition, the communities range across

 

points on a continuum.rather than clustering at the extremes. This

helps account for the fact that three separate dimensions appear to be

93

Table 13. Items and factor loadings from.a factor analysis

of measures of community development

 

 

Item Loadigg
Factor I: Institutional develgpment
Community has a government office .77
Community has a postal service .70
Community has a'bar .67
More kinds of work done in the community .65
Community has a plaza .65
Community has a barbershop .62
Community has telephone service to outside _ .60
Community has a football field .50
Community'has bus service to the outside .h2
Community has a school .h2
Distance to the most used commercial center .36
Community has a store .3h

Factor II: Medical development and non-dgpendence on farming

 

Number of visits by doctors per month .65
Per cent of community not involved in farming .56
Per cent of community having sanitary toilet facilities .55
Distance to nearest large factory -.52
Infant survival rate .51
Per cent of farm income from a single crop .h5
Per cent of community using filtered water .h5
Per cent of homes without a hunger problem .h8
Community receives more than one TV station .38
Per cent of cr0ps sold annually .37
Number of months community is isolated by weather -.31

Factor III: Urban contact

Per cent of inhabitants who have parasites .62
Community lacks a school .59
Most farmers have visited the capital at least once in

their life .53
Community lacks a store .h8
Someone visits the capital monthly .h6
Community has motor transport for getting crops out .h5
Community has postal service .hh
Most farmers are not primarily renters but owners .h3
Time needed to get to big city by bus .hl

Most farmers have relatives in the city .39

91+

necessary to explain an adequate amount of variance in the factor
analysis.

The first factor to emerge was named "Institutional.Development."
Items for it were drawn largely from.the work of Frank and Ruth Young.
(Young and Young, 1962) The second factor had its highest loadings on
indices of medical facilities and on indications of non-dependence on
farming. The third factor seemed to be composed mostly of some kind
of urban contact dimension. It also included a high loading on parasite
infestation level which, curiously, loads off away from the medical
items in the second factor.

Following construction of the factor scores, the communities were
arranged in order of their scores on each of the factors separately
and these arrangements dichotomized at their medians.* Since the
three factors are orthogonal and the factor scores uncorrelated, the
three criteria for dichotomization were independent.

With the sample of respondents dichotomized three times according
to three independent criteria, correlations were calculated for upper
and lower halves of each dichotomy between hypothetical thinking and
(1) number of godchildren, (2) number of peers who designated a respondent
as a friend, (3) social memberships, (h) the number of nominations
received as a person who brings in information from the outside world,
(5) status as self-perceived, (6) status as measured from.obJective

criteria, and (7) number of opinion leadership nominations from.peers.

 

*The sample split almost exactly in half each time, with sample
size of 385 for the top group on institutional development and 390 for
the bottom group, 388 for the top group on medical resources and non-
dependence on farming versus 387 for the bottom.group, and 393 in the
top half of the split on urban contact versus 382 in the bottom.half.

95

Despite the degree of precision afforded by the relatively large
sample sizes, only one of the 21 comparisons of correlations reaches
significance at an acceptable level. The correlation between scales on
hypothetical thinking and self-perceived status for individuals in
communities relatively high on institutional development reaches .33,
while the same relationship for those in communities low on institutional
develOpment is .19. The difference is significant. However, one
significant difference in 21 is exactly the sort of finding that
statistical inference anticipates, hence it seems inappropriate to
attempt to interpret this lone positive result.

Table 1% presents the correlations of our measures of aspects
of empathy with the seven variables considered in this section. The
total sample, without dichotomization on traditionality;is the basis

for the correlations in Table 1h.

 

Table 1%. Correlations of three aspects of empathy with seven
measures of integration, status, and influence in the community

Correlation.with
Hypothet- Imagin- Strang-

Variable icals ation ers
Number of godchildren claimed .00 .03 .02
Number of nominations as a friend .05 .01 .06
Number of memberships in social groups .10* .1h* .05
Number of nominations as an information
source .15* .10* .13*
Self-perceived status .27* .29* .13*
ObJectively determined status .11* .1h* .16*
Opinion leader nominations from.peers .15* .01 .06~

 

*Significantly different from.zero at the .05 level of
confidence: N = 775.

96

Although the maJority of the correlations can confidently be
regarded as representing real relationships, it is difficult to
generate enthusiasm over only two, four, or eight per cent common

variance represented by most of the data in Table 1%.

V. THE RELATIONSHIP OF IMAGINATIVENESS TO EXPOSURE TO RKDIO

AND EXPOSURE TO PRINT

It was hypothesized that the imaginativeness aspect of empathy
would relate most strongly to exposure to radio and less strongly to
exposure to print. The factor analysis of exposure to mass media
produced a factor which appeared to be composed primarily of utilization
of literacy. Items selected for a factor score on exposure to print are
given below; the numbers in parentheses indicate the weights given
each item in forming the factor score:

1. H w often per month do you read (or does someone read

to you) magazines and newspapers? (2.299)
2. Do you read news about agriculture in the magazines

and newspapers? (0.79%)
3. How many letters do you write (or ask to have written

for you) per year? (1.301)
M. How many times a year do you go to the movies?* 1.325)

We shall refer to this score as "print exposure."
The exposure to radio score was composed of the following

items:

 

*It was surprising to find movie attendance showing more
relationship to exposure to print than to exposure to radio. It seems
likely that we are dealing with an economic restraint. The radio is
the poor man's means of entertainment. If he can scrape together
enough for the purchase, he has little expense thereafter. Further-
more, literacy or travel to a distant center is not necessary. Then
too, many films shown in rural Brazil require literacy since they are
produced in other countries and only provided with Portuguese sub-
titles, not dubbed voices.

97

1. How many hours a week do you listen to the radio
attentatively? (1.206)
2. Do you listen to agricultural news on the radio? (2.h60

The correlation between imaginativeness and print exposure is
.32. With radio exposure partialled out, the correlation is ”39. The
correlation between imaginativeness and radio exposure is .18. ‘With
print exposure partialled out it drops to ._3, The difference between
these two partials is significant at better than the .0005 1ewel.*

The result is in the direction opposite that hypothesized earlier.
Exposure to print shows aeignificantly greater relationship to imagin-
ativeness than exposure to radio. This is true both for the comparison
of apprOpriate partials and for the zero order correlations. The inter-
correlation of the two factor scores on print and radio exposure is,
incidentally, only .19. (This is largely a function of the orthogonal
factor analysis and does not necessarily represent the degree of
relationship between exposure to one of these media and exposure to the
other.)

The zero order correlation of skill-with-hypotheticals and print
exposure is .23 which is identical to its correlation with radio
exposure. The correlation.between interaction with strangers and print
exposure is .15, while its correlation with radio exposure is .06. This

difference is also significant and favors the print media.**

 

*t = 3.768 with 771 degrees of freedom (an extra degree was lost

since the test was a comparison of first order partials rather than
zero order correlations).

*ft 8 2.1798 with 772 degrees of freedom.

97A

VI. REGRESSION TESTS 0N MASS MEDIA VARIABLES

Two aspects of empathy are considered separately as classif-
icatory variables in this analysis: skill-with-hypotheticals (often
shortened to "hypotheticals" hereafter) and imaginativeness. Both
are operationallzed as normalized factor weighted scores. The items
composing hypotheticals were listed in Table 3 under the discussion of
the factor analysis of empathy items. There they emerged as factor two.
Their weights for factor score construction were likewise given there.
Since each item was cast into a normal distribution before weighting,
the scoring system associated with the raw items is not important. The
items composing imaginativeness were also given in Table 3, together
with their weights, as the third factor from.the factor analysis of
empathy.

For this analysis the 775 subJects were first cross-classified
by their scores on hypotheticals and all regression analyses performed;
then they were cross-classified by their scores on imaginativeness and
these analyses repeated. Since the correlation between the two aspects
of empathy was .20, the cross-classifications were not identical.

Two measures of mass media exposure are used separately in this
section. This was necessary because the factor analysis of’mass
media items indicated that radio exposure, and particularly radio
exposure for farm information, tended to load away from.exposure of
other types. Since the first principal axis factor did not contain
high loadings for exposure to radio, the decision was made to invest-
igate radio exposure separately. This seemed particularly needful in

view of the fact that hypotheticals were more closely associated with

98

radio exposure than with the measure of general mass media exposure
derived from the first principal axis solution. Also imaginativeness
was correlated at a significantly higher level with general.mass media
exposure than with exposure to radio.

The mass media exposure index was formed from the variables
given'below:

1. How often per month do you read (or does someone read

to you) magazines and newspapers? (2.125)
2. How many hours a week do you listen to the radio

attentatively? (0 . 389;
3. How many times a year do you go to the movies? ’ 0.880

This score was similar to but not identical with the "print exposure"
score described in Section V. The radio exposure score from.Section‘V
was utilized again here.

In general, the procedure of cross-classifying the sample on
scores on either hypotheticals or imaginativeness resulted in a moderate
progression of the means of the groups on mass media variables roughly
corresponding to the level of empathy. The five groups were of equal
size (155 each); hence the regression estimates derived from.them were
equally stable, although not equally good fits to the clusters of data.

The correlations of both types of mass media exposure failed
to reach significance with the dependent variables of (1) aspirations,
(2) value for dissatisfaction, and (3) a tendency to think about
problems. N0r did the multiple correlation obtained utilizing levels
of mass media classified on the two aspects of empathy result in
significance for these variables. Therefore, we must regard these
variables as not clearing the first hurdle of this analysis by failing
to show a significant relationship to either variety of mass media

exposure, even under cross-classification conditions. This probably

99

results from poor measurement of the three dependent variables. Two
of the three were dependent on single items and the third relied on
only two items. The three variables will not be considered further
in this section.

Three of the dependent variables did clear the first hurdle
by showing significant relationships to exposure to mass media when
levels of aspects of empathy were utilized to cross-classify the data.
5 These three were (1) belief in the efficacy of work rather than a
fatalistic reliance on luck (hereafter called "belief in work"),

(2) political knowledge, and (3) a favorable attitude toward change.

Possible Statistical Outcomes from Regression Analysis
Cross-classifying the sample on five levels of an aspect of
empathy allows each level of the independent variable to act as a
separate, independent predictor in a multiple correlation. Instead of
merely having the correlation between, say, exposure to radio and belief
in work we have the multiple correlation of five different independent
variables -- exposure to radio-lowest empathy, exposure to radio-low
empathy, exposure to radio-average empathy, exposure to radio-high
empathy, exposure to radio-highest empathy -- and a single dependent
variable, belief in work. Any improvement over the simple correlation
due to this cross-classification can be a result of (1) allowing
freedom in the points of intercept for the five independent variables,
(2) allowing freedom in the slopes of the five, or (3) allowing
freedom in both intercepts and slopes. In.most studies, allowing
freedom to the intercepts is somewhat more important than allowing it
to the slopes; this is consistently the case in the analyses presented

here.

100

If a restriction is introduced in the slopes of the five independent
predictors making them all equal, this restriction can.be tested to
determine whether it significantly reduces the predictiveness of the
"unrestricted" solution, i.e. the solution in which.both slopes and
intercepts are free to vary. It can also be tested to determine whether
it is a significant improvement over the completely restricted solution,
i.e. the solution which ignores cross-classification altogether. The
first test is known as the test of the hypothesis of a common slope.

If a restriction is introduced which forces the intercepts of
the five predictors through a common point on the ordinate, the degree
to which this restriction reduces the unrestricted solution can be
tested for significance. This test is known as the test of the hypothesis
of a common intercept. Likewise, the partially restricted solution with
a common intercept can be tested to determine whether it is significantly
superior to the completely restricted solution.

The fo110wing types of outcomes exhaust the relevant possibilities:*
I. The multiple correlation of the unrestricted solution is
significantly larger than the multiple correlations of the
partially restricted solutions and the correlation of the
completely restricted solution.
II. The multiple correlation of the unrestricted solution is
significantly larger than that of the partially restricted
solution with a common intercept and the correlation of the
completely restricted solution; but it is not significantly
larger than the multiple correlation of the partially restricted

solution in which a common slope and different intercgpts are
utilized.

 

*It would be possible, of course, for the correlation of the un-
restricted solution to be significantly larger than that of the completely
restricted solution while the correlations of the partially restricted
solutions were 223 significantly different from.either the unrestricted
or the completely restricted solutions. This did not occur in this
study and so is excluded from the list given here as an unnecessary
complication to an already complicated tepic.

101

III. The multiple correlation of the unrestricted solution is
significantly larger than that of the partially restricted
solution with a common lepe and the correlation of the
completely restricted solution; but it is not significantly
larger than the multiple correlation of the partially restricted
solution in which different slopes and a common intercept are
utilized.

IV. The multiple correlation of the unrestricted solution is
significantly larger than the correlation of the completely
restricted solution; but it is not significantly larger than
the multiple correlations of either of the partially restricted
solutions, i.e. the solutions in which either the slopes or
the intercepts are free to vary .

V. The multiple correlation of the unrestricted solution is not
significantly larger than the correlation of the completely
restricted solution.*

These outcomes characterize the analyses that follow and will be

referred to therein by their Roman numerals.

It may not be clear why one might be interested in the partially
restricted solutions. Basically the reason is parsimony. Frequently,
when no restrictions are laid on the formation of the multiple regression
equation the regression planes are laid through the data in a math-
ematically optimal but theoretically unintelligible fashion. In
this event, the partially restricted solutions may recommend them-
selves. Clearly they do so more strongly when they are not signif-
icantly inferior to the unrestricted solution although still superior
to the completely restricted solution. But even when the difference
between the unrestricted and one of the partially restricted solutions
is statistically significant, we may, for reasons of intelligibility,
prefer to interpret one of the partially restricted solutions. In so

doing we are placing higher value on our ability to make some sense

 

*This occurred with three of the independent variables, as al-
ready noted. Its occurrence eliminated them from.further consideration.

102

out of the outcome than on the optimization of the goodness-of-fit
of the data to a mathematical model.

One other comment is necessary. In the Figures included in
this section results will be graphed in two dimensions. Strictly
speaking this distorts the outcome since with five independent pre-
dictors a sixedimensional space is necessary to represent the results.
However, the Figures are a permissible simplification since the five
predictors are merely subgroups of the same variable and hence fall

on the same dimension for graphing purposes.

Belief in the Efficacy of Work as a Dependent Variable
This variable is formed out of the sum of the following items:

1. What is the most important for the future of the

people of this community -- hard work or luck? (2.05h)*
2. What has most to do with why some families are

rich -- hard work or luck? (1.897;
3. What's the most distant place you've ever visited? 0.h55

The first two items were scored such that people who indicated more
belief in work received higher scores; the third item gave higher
scores to people who had visited more distant places. As the weights
show, about four-fifths of the variable comes from the first two items.
Mass media exposure classified on imaginativeness. The simple
correlation.between'belief in work and exposure to mass media is .21.
Cross-classifying the sample on five levels of imaginativeness, a

significantly larger multiple correlation of .28 is formed.** The

 

*The weights applied to the items as derived from a factor
analysis of’modernization orientations are shown here.

**significant at better than the .0005 level of confidence.

103

result of the cross-classification is an instance of a type IV out-
come. In other words, it is permissible to assume in turn that (1) all
the levels have a common slope and examine only the pattern of their
intercepts or (2) all the levels have a common intercept and examine
only the pattern of their slopes.

With the slopes restricted to equality and the intercepts
free to vary we have a lepe of .252 and a generally increasing
pattern of intercepts.* In other words, if we assume that everyone
increases the favorableness of his attitude toward work at the same
rate per unit exposure to the mass media, then the best fit to the
data occurs where the more imaginative groups of people start at a
highs; level of belief in work than do the less imaginative.

On the other hand, we can also assume a common intercept.
Then we find a rising pattern of slopes as imaginativeness increases.
In other words, if everyone started at the same level of belief in
work, the best fit to the data occurs where the rate of increase per
unit mass media exposure is greater for the more imaginative than
for the less.** Under the assumption of a common origin, the more
imaginative appear to derive more'belief in work per unit mass media
exposure than do the less imaginative.

If we place no restrictions on either the slopes or intercepts

we obtain the pattern graphed in Figure h. This pattern is not easy

 

*These start at 580 for the lowest level and go to 609, 60%,
632, and 6h8 respectively for levels two through five. Except fer
the small reversal for levels two and three, the intercepts rise as
the level of imaginativeness rises.

*TThe slope for the lowest level is .18 followed by'.2h, .23,
.27, and .30. Except for the slight reversal between levels two
and three, the data conform to the hypothesized pattern.

10h

Figure h. Belief in work predicted by exposure to mass media:

data classified on five levels of imaginativeness

 

 

 

‘ thyk
700 . “Slat
3rd4
Belief 660 -
in 620 a
Work 580 1
5&0 7 R - .2759**
K2nd
500 -
Exposure to Mass Media*
Regression Analysis
Level of Control Intercept Slope*

E 686 .189

675 .178

3 559 -333

2 #96 .h59

1 562 .28h

 

*The independent variable in the graph has been divided by'a
constant of 3 to display the slopes more effectively.

**Significantly greater at better than the .0005 level than the
correlation of these variables without consideration of levels.
That correlation is .2129.

105

to interpret. Ignoring small internal reverses, the intercepts form
somewhat the same pattern as found earlier -- those with less imagin-
ativeness tend to start lowest in their belief in work while those
with more start somewhat higher. The pattern of slopes is more
Jumbled, particularly since the sloPe for the first level is more
similar to that of the third level than that of the fourth. The data
are presented below Figure h. The best rough interpretation is that
persons who have more imaginativeness start at a higher level of belief
in work but increase on this belief more slowly per unit exposure
than do persons low'in imaginativeness. 0n the other hand, the most
unimaginative (level one) increase their belief in work more slowly
than do the somewhat more imaginative (level two). These show the
most rapid increase per unit exposure, even though they start much
lower.

Expggure to radio classified on imaginativeness. The simple
correlation of belief in work and exposure to radio is .07.* Cross-
classifying the sample on five levels of imaginativeness, a signifi-
cantly larger multiple correlation of .23 is formed.**' The result
of the cross-classification is an instance of a type IV outcome. In
other words, it is permissible to examine in turn only the pattern
of intercepts or only the pattern of slopes.

With the slopes restricted to equality at .056 and the inter-

 

*Since the correlation between mass media exposure and radio
exposure is low (.20) and the sample size large, the simple correla-
tion for radio exposure is significantly smaller than that previously
reported for mass media exposure, .21 (t.= 3.1h with 773 degrees of
freedom).

**Significant at better than the .0005 level of confidence.

106

cepts free to vary, the pattern of increase corresponds to expectations.
The only exception is a slight reversal for levels two and three. The
data are presented and graphed in Figure 5. It appears that if we
assume no difference in rate of increase per unit exposure to radio,
the best fit to the data occurs where the more imaginative start with
more'belief in work.

With slopes free to vary and a common origin selected on the
ordinate the pattern shown in Figure 6 occurs. If everyone started
with the same level of belief in work, the best fit to the data
would occur where the rate of increase in belief in work is greater
for the more imaginative. There is a slight reversal of pattern for
levels two and three, but aside from this the results conform to the
hypothesis.

If we place no restrictions on either the slopes or the inter-
cepts we obtain an uninterpretable pattern. Since this is so and since
the unrestricted solution is not significantly superior to either of
the partially restricted solutions graphed in Figures 5 and 6, we
will ignore it.*

Mass media exposure classified on hypotheticals. The simple
correlation of belief in work and mass media exposure is .21, as
alreadyzeported. Cross-classifying the sample on five levels of
hypotheticals produces a significantly larger correlation of .25.**

The result of cross-classification is an instance of a type IV

 

*The pattern of slopes for this unrestricted solution is .103
for the lowest level followed by .038, -.008, .092, and .053. The
pattern of intercepts is 6M7 for the lowest level followed by 720,
7117, 71k, and 762.

**Significant at better than the .0005 level of confidence.

107

Figure 5. Belief in work predicted by exposure to radio:

data classified on imaginativeness, slopes made equal

 

 

 

 

760
Belief
in 7113
Work
720
700 ‘
680 "
Exposure to Radio*
Regression Analysis
Level of Control Intercept Slope*
5 760 . 056
b, 7A2 . 056
3 703 .056
2 708 . 056
l 678 . 056

 

TThe independent variable in the graph has been divided by a

constant of 10 to display the sloPes more effectively.

“Significantly greater at better than the .0005 level than the
correlation of these variables without the consideration of levels.

That correlation is .0737.

108

Figure 6. Belief in work predicted by exposure to radio:
data classified on five levels of imaginativeness,

intercepts made equal

 

800 - 5th”
hth
780 .
Belief 2nd \
in 760 a
3rd 9.
work

7&0 4 R = .2216**
720 late”

 

Exposure to radioI

Regression analysis

 

Level of control Intercept Slope*
5 713 .120
h 713 .091
3 713 .oho
2 713 . 0M8
l 713 .006

 

*The independent variable in the graph has been divided by a
constant of 15 in order to display the slopes more effectively.

**Significantly greater at better than the .0005 level than
the correlation of these variables without the consideration of
levels. That correlation is .0737.

109

outcome. In other words, it is permissible to examine in turn only
the pattern of intercepts or only the pattern of lepes.

‘With slopes restricted to equality at .296 the now familiar
pattern of increases in intercepts with increases in levels of
.hypotheticals is found.* Those higher on hypotheticals start with
more belief in work when the rate of increase per unit mass media
exposure is held constant.

‘With the intercepts constant we find the familiar pattern of
slopes increasing as hypotheticals increase.** This indicates that
if everyone started at the same origin those more skilled with hypo-
theticals would increase their belief in work more rapidly per unit
exposure to the mass media. This outcome agrees with the hypothesis.

If we place no restrictions on either the slopes or intercepts
we obtain the pattern graphed in Figure 7. In this pattern the inter-
cepts generally rise (with two reversals, one from levels four to
five and the other from levels one to two). 0n the other hand, the
slopes generally drop in Figure 7 (with a single reversal from
levels four to five). This indicates that while those with more
hypothetical skill start higher at the present time, their rate of
increase is a little slower per unit mass media exposure.

Exposure to radio classified on hypotheticals. The simple

 

correlation of belief in work and exposure to radio is .07, as already

 

*The intercept for the lowest level is 576 followed by 569,
588, 605, and 610. The multiple correlation produced by this solution
is .2h. A minor reversal in pattern occurs moving from level one to
level two.

**The slope for the lowest level is .27h followed by .262, .292,
.318, and .327. The common intercept is 590. There is a reversal
of pattern between levels one and two.

110

Figure 7. Belief in work predicted by mass media exposure:

data classified on five levels of skill with hypotheticals

 

 

 

 

7h0
700
Belief
in 660
work
620
580
5h0
500
Mass media exposure*
Regression Analysis
Level of Control Intercept Slope*

5 639 .2h6

h 677 .170

3 607 .263

2 A90 .h37

l 505 .h26

 

*The independent variable in the graph has been divided by
2 to display the slopes more effectively.

**Significantly greater at better than the .0005 level than
the correlation of these variables without the consideration of
levels. That correlation is .2129.

111

reported. Cross-classifying the sample on five levels of hypotheticals
produces a significantly larger multiple correlation of .17.* Again,
the cross-classification results in a type IV outcome. In other words,
it is permissible to examine in turn only the pattern of intercepts

or only the pattern of slopes.

With the slopes equal at .0615 the intercepts rise as the levels
of hypotheticals rise.** In general, the more hypothetically skilled
start with greater belief in work when the rate of change per unit of
exposure to radio is held constant. With the intercepts constant at
711 the slopes generally increase as the levels of hypotheticals
rise:.*** If everyone started at the same point those more skilled
with hypotheticals would increase their'belief in work more rapidly
per unit mass media exposure. This conforms to the initial hypothesis.

When both slopes and intercepts are free to vary both patterns
disappear and a Jumble of regression lines is found.**** Since this
unrestricted solution does not significantly improve the correlation
between the independent and dependent variables over either of the
partially restricted (and intelligible) solutions, we will not

consider it further.

 

*Significant at the .02h level of confidence.

*fiThe intercept for the lowest levels is 698 followed by 692,
710, 732, and 7H2. There is a slight reversal of pattern between
levels one and two.

***The slope for the lowest level is .OhO, followed by .037,
.060, .090, and .105.

****In order from lowest to highest levels, the slopes are .067,
.033,h.02h, .197, and -.007; the intercepts are 738, 752, 736, 600,
and 7 3.

112

Political Knowledge as a Dependent variable
This variable was measured as the sum of correct answers to the
following four questions:
1. Who is the present governor of'Mlnas Gerais?
2. To what American country did Brazil send troops last year?
3. Who was the Brazilian President who was deposed by the

revolution two years ago?
h. What Latin American country turned communistic a few years

ago?

These items were selected by Guttman scaling procedures out of a pool
of 11 items. The four survivors displayed suitable scaling character-
istics.* All four items received equal weight in the index of political
knowledgeability.

Since this variable was not converted to normalized form and
made into a factor weighted score its regression slope on variables
so structured appears to be very low; This is largely an artifact
of score construction. It is removed in the Figures used in this
section, but the regression coefficients are reported as they were
produced by the computer.

Ezposure to mass media classified on imaginativeness. The

 

simple correlation of political knowledge and mass media exposure is
.37. Cross-classifying the sample on five levels of imaginativeness
a significantly larger multiple correlation of .hh is obtained. Re-
strictions on slopes closely approach significance (.06) while re-
strictions on intercepts reach it at .02. Strictly speaking, the
result of cross-classification is an instance of a type II outcome.

Statistically speaking, then, it is permissible to examine the pattern

 

*The measure was also relatively stable over time with a test-
retest reliability of .66.

113

of the intercepts under the restriction of equal slopes but not
permissible to examine the pattern of slopes under the restriction

of equal intercepts. On the other hand, since the latter solution
gagg_represent a significant improvement over the completely restricted
solution it may be worth holding in abeyance our knowledge that the
restriction to a common intercept involves a statistically significant
reduction and examine the pattern of slopes as well.

With a common slope of .0005h a pattern of increasing inter-
cepts is found.* Therefore, if persons at all levels of imaginative-
ness increased their political knowledge at the same rate per unit
mass media exposure, the more imaginative would start at a higher
level of knowledge.

If we start everyone at the same level of knowledge, then the
lepes of those high on imaginativeness are steeper than the slopes
of those low on imaginativeness.** This fit of regression slepes to
data clusters conforms to hypothesis.

When we examine the unrestricted solution a more confusing
pattern emerges. It is graphed in Figure 8. Persons in the second
and third levels start considerably lower and increase more rapidly;
those in the first level start higher‘but increase more slowly;
those in the two highest levels increase somewhat the same as those
on the first level but intercept the ordinate at a higher points.

This pattern seems to suggest a take-off phenomenon. Those with

 

*For the lowest level of imaginativeness the intercept is -.87.
This is followed‘by -.58, -.h5, .Oh, and -.03. There is a small
reversal of pattern for levels four and five.

*IFor the lowest group the lepe is .000h3. This is followed
by .000h9, .00051, .00059, and .00058.

11h

Figure 8. Political knowledge predicted by mass media exposure: data

classified on five levels of imaginativeness.

11thN
1.00
‘ 5th (
O 60 I
lst‘N
Political ~20 .

knowledge

 

 

 

Mass media exposurer

Regression analysis

Level of control Intercept SlOpe*
5 .h7 .000h5
1+ .73 .000h2
3 -2.17 .00085
2 -2.12 .00082
1 "' o 15 39 o OOOLI'O

 

*The independent variable was divided by-1000 to display the
slopes more effectively.

**Significant1y greater at better than the .0005 level than the
correlation of these variables without the consideration of levels.
That correlation is .3702.

115

enough imaginativeness begin to learn political information rather
rapidly from exposure to mass media while those who have not reached
a certain threshold of imaginativeness increase somewhat more slowly
and those in the highest levels of imaginativeness also increase
slowly.

Expgsure to radio classified on imaginativeness. The simple
correlation of political knowledge and exposure to radio is .18.*
Cross-classifying the sample on five levels of imaginativeness a
significantly larger correlation of .35 is obtained.** The result
of cross-classification is a type IV outcome. It is permissible to
examine both the pattern of intercepts and the pattern of slopes.

With a common slope of .000217 a pattern of increasing inter-
cepts is found.*** With a common intercept the lepes likewise
increase as hypothesized.**** The more imaginative increase their
political knowledge more rapidly per unit exposure to radio.

The unrestricted solution is sufficiently difficult to inter-

 

*Since the intercorrelation between exposure to mass media
and exposure to radio is low (.20) and the sample size large the
simple correlation for radio is significantly larger than that
previously reported for mass media exposure, .37 (t_= h.50 with 772
degrees of freedom).

**Significant at better than the .0005 level of confidence.

***For the lowest level of imaginativeness the intercept is .59.
This is followed by .89, 1.00, 1.73, and 1.68. There is a slight
reversal of pattern between the intercepts of the two higher levels.

"”“IFor the lowest level the slope is .OOOlh. This is followed
by .00019, .00020, .00030, and .00030.

116

pret that we will prefer either of the alternative, partially
restricted,solutions.*

Mass media exposure classified on hypotheticals. The simple
correlation of political knowledge and mass media exposure is, as
already reported, .37. Cross-classifying the sample on five levels
of hypotheticals, a significantly larger multiple correlation of .hO
is obtained.** The cross-classification results in a type VI out-
come. Thus, we may with statistical propriety examine either the
pattern of intercepts or the pattern of slopes.

With all slopes at .000605 the familiar increasing pattern of
intercepts is found.***; If individuals at each level of hypotheticals
increase their political knowledge at a uniform rate per unit mass
media exposure, then those with more hypotheticals start at a higher
level of initial knowledge. With all intercepts set a -.73 the
pattern of slopes is also one of increase as hypotheticals increase.
If everyone starts at the same point in political knowledge those with
more hypotheticals increase more rapidly per unit exposure to the mass
media.

The unrestricted solution is difficult to interpret,and since

it is not a significant improvement over the partially restricted

 

*From lowest to highest level the intercepts are .09, .27,
2.03, 2.70, and .73. The sloPes, in the same order, are .000292,
.000308, .000069, .000077, and .000350.

**Significant at the .022 level of confidence.

*ZfiThe lowest intercept is -l.07 followed by -.88, -.78, -.5h,
and -0 9s

****The lowest level's slope is .0005h5 followed by .000579,
.00059h, .0006h0, and .0006u2.

117
*
solutions Just presented,no attempt will be made to interpret it.

Exposure to radio classified on hypotheticals. The simple
correlation of political knowledge and exposure to radio is, as already
reported, .18. Cross-classifying the sample on five levels of skill
with hypotheticals, a significantly larger multiple correlation of
.26 is obtained.** Cross-classification results in a type IV outcome.
It is, therefore, permissible to examine the pattern of intercepts
or the pattern of slopes.

With all slopes at .000222 the intercepts increase as we
pass from.the lowest to the highest levels.*** If individuals at
each level of hypotheticals increase their political knowledge at
the same rate per unit exposure to radio, then those with more
hypotheticals start at a higher level of initial knowledge. With all
intercepts set at 1.07 the pattern of slopes is likewise one of

II”! If everyone

increase with increasing levels of hypotheticals.

starts at the same point on political knowledge, then those with

more hypotheticals increase more rapidly per unit exposure to radio.
'With'both sloPes and intercepts free to vary a Jumble of

patterns occurs. This Jumble is particularly due to the fifth level,

 

*In order from lowest to highest level, the intercepts are
-l.33, -.90, —.30, -l.16, and -.l7; the slopes are .00065, .00061,
.00052, .00071, .00055. Except fOr the reversal of pattern in the
intercept of the fourth level, the intercepts at least would form a
pattern of increase.

**Significant at the .01 level of confidence.
*H'The lowest intercept is .79 followed by .98, 1.05, 1.38, and 1.52.

*""""‘1‘he lowest level has a slope of .000183 followed by .000212,
.000221, .000262, and .00028h.

118

which finds its intercept midway between the third and fourth levels
and has a very gradual slope. ‘Were it not for this level, the inter-
cepts, at least, would fall into an interpretable pattern.* However,
since the completely unrestricted solution is no improvement over
either of the partially restricted solutions there is no need to

interpret the unrestricted solution.

Favorable Attitude Toward Change as a.Dependent Variable
.This variable is formed out of’the sum.of'the following items:

1. Whom do you trust most when it comes to new agri-
cultural ideas, educated persons (2) or persons

with practical experience (0)? (.h75)
2. If you could change things around here would you
leave them as they are (0) or'make changes (2)? (.hhh)

3. Compared with your father's time would you say that
farming has changed (yes and that's good = 3, yes
and that's'bad = 0) or has stayed the same (yes

and that's good = 1, yes and that's bad = 2)? (.619)
h. Is the respect shown'by youth for the aged changing?

Is this good or bad? (same codes as 3 above) (1.169)
5. Do you think changes bring more problems (0) or

more improvements (2)? (1.3h3)
6. Do you have much influence in the way things change

around here? (.607)

The weights applied to the items are from.a factor analysis of‘modern-
ization orientations. They are given in parentheses next to the items.

Exposure to mass media classified on imaginativeness. A favor-

 

able attitude toward change correlates .59 with exposure to mass media.
Cross-classifying on the five levels of imaginativeness produces a

significantly larger multiple correlation of .62.** The cross-

 

*In order from lowest to highest level, the intercepts are
.h72, .390, .636, 2.830, and 1.755; the slopes are .000271, .000309,
.000018, .000189.

**Significant at better than the .0005 level.

119

classification results in a type I outcome.* Nevertheless, the
consideration of this data classified by levels while imposing either
a common regression slope or a common intercept results in a signifi-
cant improvement over the completely restricted solution. It happens
that the unrestricted situation is readily interpretable. As can‘be
seen in Figure 9, the intercepts rise regularly and the slopes fall
regularly as we move across levels of imaginativeness. This indicates
that persons who are more imaginative start with a more favorable
attitude toward change and increase their favorableness somewhat more
slowly per unit mass media exposure.**

[Expgsure to radio classified on imgginativeness. The simple
correlation of a favorable attitude toward change and exposure to the
radio is .18.*** Cross-classifying the sample on five levels of

imaginativeness a significantly larger correlation of .36 is found.”"""

 

*The imposition of restriction on slopes is significant at the
.018 level while the imposition of restrictions in intercepts is
significant at the .002 level. Hence, both the hypothesis of a
common regression slope and the hypothesis of a common intercept must
be reJected.

**Were we to force all slopes to be equal the intercepts would
continue to fall into a pattern of increase similar to that shown in
Figure 9. Their range, however, would be restricted to only ten
scale points rather than the 71 scale points they now cover. The
lowest level would have an intercept of 261 followed by 262, 265, 271,
and 271. 'Were we to use a common origin on the ordinate for all five
levels the more imaginative would increase their favorableness toward
change more rapidly (rather than less rapidly, as is the case with
the unrestricted solution). The regression coefficients would be
.181 for the lowest level followed by .183, .188, .198, and .197.

***Since the intercorrelation.between exposure to mass media and
exposure to radio is low (.20) and the sample size large the corre-
lation with radio is significantly smaller than that previously noted
fer'mass media exposure generally, .59 (3.: 11.39 with 772 degrees of
freedom). Apparently a favorable attitude toward change is more
related to exposure to mass media in general than to exposure to radio
for agricultural information

**"'*Significant at better than the .0005 level of confidence.

120

   

Figure 9. Favorable attitude toward change predicted by mass media
exposure: data classified on five levels of imaginativeness
5th-s
nth},
290
3rd‘u
275 .
Favorable
attitude 260 -
2nd-3,
toward
change 2&5 '
230
R = .6169**
215

 

 

 

Mass media exposure*

Regression analysis

 

Level of control Intercept

288
280
270
2111
217

FJHJUJ$r\n

Slope*

.162

- 175
.181
. 229
. 270

 

*The independent variable was divided by'2 to display the lepes

more effectively.

**Significantly greater at better than the .0005 level than the
correlation of these variables without the consideration of levels.
That correlation is .5933.

121

Cross-classification gives a type I outcome. However, both partly
restricted solutions do a considerably better Job of predicting the
dependent variable than would be the case if levels were ignored
altogether. Thus, we might be willing to look at the pattern of
slopes or the pattern of intercepts even in the face of knowledge
that a significantly better prediction occurs when we consider‘both
simultaneously, because a likewise significantly better prediction
occurs when at least one of them is free to vary.

The unrestricted solution is difficult to interpret. Figure
10 depicts it. ‘We see particularly that the intercept and slope of
the second level depart from an otherwise regular pattern of increasing
intercepts and decreasing slopes.

Since the interpretation of the unrestricted solution is not
completely straight-forward, we will look briefly at the solutions
with restrictions, even though we know these restrictions make the
correlation significantly smaller than that found in the unrestricted
case. With a common slope of .Ohl (which restriction is significant
at the .01 level) a multiple correlation of .3M is produced and the
intercepts have a regularly increasing pattern.* If a common inter-
cept of 3hh is set (which restriction is significant at the .001
level) a multiple correlation of .33 is produced and a regularly
increasing pattern of slopes is foundfa'It would appear that if either

an artificial starting point could be imposed or respondents at all

 

*They start at 336 and mount to 338, 31m, 355, and 357 as
we move up levels.

**The slope of the lowest level is .030 followed by'.03h, .037,
O 057’ ma 0 058 O

122

Figure 10. Favorable attitude toward change predicted by exposure

to radio: data classified on five levels of imaginativeness

 

 

 

 

 

Loo
Favorable 380
attitude
1;: ;C 360
toward
change 3&0
320
300 . _ **
, R - .3601
Exposure to radio;_—
Regression analysis
Level of Control Intercept Slope*
5 381 .0068
h 372 .0169
3 359 .01h0
2 295 .10h7
1 33h .0h33

 

*The independent variable was divided by 10 to display the
slopes more effectively.

**Significantly greater at better than the .0005 level than
the correlation of these variables without the consideration of
levels. That correlation is .1789.

123

levels of imaginativeness forced to increase the favorableness of
their attitudes toward change at a uniform rate per unit exposure to
the radio, then the more imaginative would either start at a higher
level of favorableness or increase more rapidly. However, we know
both conditions to be artificial and must defer for later investi-
gation an explanation of the pattern of slopes and intercepts found
in the unrestricted case. This explanation may involve an early
take-off followed by more gradual levels of increase.

Mass media expgsure classified on hypotheticals. The simple

 

correlation of a favorable attitude toward change and exposure to
mass media is, as already reported, .59. Cross-classifying the sample
on five levels of skill with hypotheticals, a significantly larger
multiple correlation of .62 is Obtained.* Cross-classification gives
a type I outcome.** HOwever, either partially restricted solution
is significantly superior to the completely restricted solution.
Thus, we might be willing to look at the pattern of slopes or the
pattern of intercepts, even in the face of knowledge that a signifi-
cantly better prediction occurs when we consider both simultaneously,
because a likewise significantly better prediction occurs when at
least one of them is left free to vary.

The interpretation of the unrestricted case is not too
difficult if we ignore the reversals of position of the top three

levels when it comes to intercepts. Since the three are relatively

 

*Significant at better than the .0005 level of confidence.

**Restrictions on lepes reach significance at .039 while re-
strictions on intercepts reach significance at .005.

12%

close together this may be permissible. The steeper lepes are
associated with the first two levels followed by the fifth level.
The relationships are shown in Figure ll.*

Exposure to radio classified on hypotheticals. The simple
correlation of a favorable attitude toward change and exposure to-
radio is, as already noted, .18. Cross-classifying the sample on
five levels of skill with hypotheticals a significantly larger
correlation of .30 is produced.** Cross-classification gives a
type I outcome. However, either partially restricted solution is
significantly superior to the completely restricted solution. Thus,
we might be willing to look at the pattern of lepes or the pattern
of intercepts, even in the face of knowledge that a significantly
better prediction occurs when we consider both simultaneously.

The interpretation of the unrestricted solution is decidedly
difficult. The first level has both the lowest intercept and the
highest slope but it is followed by the second level with a rather
high intercept (above that of the fifth and third levels) and a very

low slope. The fourth level has the highest intercept but a slightly

 

*Were we to allow ourselves the restrictions of a common slope
set at .120 the pattern of intercepts would be one of consistent
increase. The lowest level has an intercept of 252 followed by 263,
26%, 26h, and 266. Apparently, if everyone increased the favorableness
of his attitude toward change at the same rate per unit exposure to
mass media the more hypothetically skilled would start with a more
favorable attitude. If we allowed the even stronger restriction of
a common intercept set at 262, the pattern of lepes would rise with
only one small reversal as the levels of hypotheticals increased. The
lepe for the lowest level would be .1812 followed by .1997, .2008,
.2003, and .20h0. As noted, either of these solutions would result
in a significantly better prediction of attitude toward change from
exposure to mass media than that found in the completely restricted
solution.

**Significant at the .0005 level of confidence.

125

Figure 11. Favorable attitude toward change predicted by mass media

exposure: data classified on five levels of skill with hypotheticals

 

 

 

 

300
280
Favorable
260
attitude
toward 2&0
change
220
200 .
Mass media exposure:
Regression analysis
Level of control Intercept Slope*
5 273 -187
A 276 .172
3 281 .1670
2 253 -215
l 216 .263

 

*The independent variable was divided by 2 to display the
slopes more effectively.

**Significantly greater at better than the .0005 level than the
correlation of these variables without the consideration of levels.
That correlation is .5933.

126

negative slope. An explanation of this pattern would be unusually
unparsimonious, requiring several accelerations and decelerations of
slope plus, apparently, a negative sloPe, somewhere toward the higher

levels. The problem is shown in Figure 12.*

sales:
If we consider the improvement in goodness-of-fit brought about

by cross-classifying the data Just presented on the relationship of

mass media to the individual's orientations and values, we can achieve

a.more compact notion of the value of the two aspects of empathy.

A score can be derived for each of the 12 analyses Just discussed by

determining the improvement in predictiveness (the extra proportion of

variance accounted for) attributable to the unrestricted solution resulting

from cross-classification on a particular aspect of empathy.
Cross-classification on the aspect of imaginativeness added

an average of 5.75 per cent to the variance accounted for; cross-

 

*If we entertain the restriction of a common sloPe set at
.039 (which restriction is significant at the .035 level) the pattern
of the intercepts is one of increase as we move up levels of hypo-
theticals. The lowest level has an intercept of 335 followed by
3A6, 3A7, 350, and 356. Clearly, if everyone increased the favor-
ableness of his attitude toward change at the same rate per unit
exposure to radio, those with more hypotheticals would start with a
higher initial attitude.

Likewise if the restriction were allowed of a common origin
of 3h5 (which restriction is significant at the .ooh level) the
pattern of sloPes would be one of regular increase. The lepe of the
lowest level would be .026 followed by .oul, .0u2, .0h6, and .05h.
In this instance, if everyone were to begin with the same level of
attitude toward change but different levels of hypotheticals, those
with more hypotheticals would increase the favorableness of their
attitude more rapidly per unit exposure to radio.

As noted, either of these solutions would result in a signifi-
cantly'better prediction of attitude toward change from.exposure to
radio than the completely restricted solution which ignores hypotheticals.

127

Figure 12. Favorable attitude toward change predicted by exposure
to radio: data classified on five levels of

skill with hypotheticals

 

 

 

 

hOO _
Favorable
380 _ hth‘}
attitude 2nd\
toward 360 - 3rd]
change ‘ {3)
31m 4 5
320i lat.3 R = .3043H
300
Exposure to radio*’
Regression analysis
Level of control Intercept Slope*
5 31a .0509
h 378 -.0002
3 350 -03h3
2 36h .0123
l 301 .0903

 

*The independent variable was divided by 10 to display the
slopes more effectively.

**Significantly greater at better than the .0005 level than the
correlation of these variables without the consideration of levels.
That correlation is .1789.

128

classification on the aspect of skill with hypotheticals added an
average of 3.18 per cent. Six separate scores are involved in each
average and the types of dependent variables and of mass media are
confounded therein. Looking at the relationship another way, the
predictive adequacy of exposure to radio was improved an average
of 5.79 per cent while the predictive adequacy of exposure to general
mass media rose an average of 3.1h per cent due to cross-classification
on some variety of empathy. In these latter instances, the aspect
of empathy and the three dependent variables were confounded.

Entering the twelve scores in two-way analysis of variance
the significance of these differences plus the possibility of an
interaction effect between the aspect of empathy utilized in cross-
classification and the type of mass media can be investigated. Table 15

shows the resulting analysis of variance.

 

Table 15. Two-way analysis of variance of the degree to which cross-
classification improves the predictive power of mass media variables
for three modernization orientations, belief in work,
political knowledge, and positive attitude toward change.*

Source of variance Mean-square ‘2

.00222 6.19**
.00207 5.78**
.000h5 1.26
.00036

 

Type of mass media
Aspect of empathy
Mass media-~empathy interaction
Within
Total

tfloah'haki H:

 

*The three dependent variables are, of necessity, confounded
in the analysis.

**Significant at the .05 level of confidence. The E value for
l and 8 degrees of freedom at this level is 5.32.

129

We can'be somewhat confident that, as the variables were
measured in this study, cross—classification on imaginativeness leads
to a greater improvement in prediction of these three dependent
variables than cross-classification on skill with hypotheticals.
Likewise, we may conclude that such cross-classification helps increase
the strength of the relationship between exposure to radio and the
dependent variables more than it helps increase the strength of the
relationship between general exposure to mass media and these same
dependent variables.*' The interaction of mass media type and the
aspect of empathy utilized in cross-classification is not significant.

Turning now to the regression estimates for these variables,
we found, generally, that when we restricted all the slopes of the five
levels to equality, in other words, when we accept the hypothesis of
a common slope,*‘the intercepts of the five levels show a definite
pattern of increase as we move up levels of a particular aspect of
empathy. 'With the data cross-classified on imaginativeness there were
only four reversals in this progression for the six comparisons.

With the data cross-classified on skill with hypotheticals, only three

reversals in the pattern of rising intercepts occurred. In none of the

 

*This might be a function of the lower initial relationship
between radio and the dependent variables, i.e., there was more room
for improvement with radio.

**This restriction was statistically acceptable with the first two
dependent variables, belief in work and political knowledge, but not
with the third, favorable attitude toward change.

***Under the hypothesis of random patterns, 30 such reversals would
have been expected for the six comparisons. With more than five reversals
per comparison, the opposite pattern of decreasing intercepts as levels
increased would be a.better fit to the data. Thus, the most random assort-
ment possible is that involving five reversals for each of six analyses.
One example would be a pattern like 5,1,2,h,3.

130

12 comparisons did more than one reversal occur. The reversals were
mostly slight and happened at the extremes, i.e., between levels one
and two or levels four and five.

It seems safe to conclude, then, that if the rate of increase
per unit of exposure is made constant for all levels of empathy, the
more empathic will start at a somewhat higher point on the dependent
variable. In other words, if everyone learned at the same rate from
exposure then the more empathic persons began at somewhat higher levels
on these three variables.

If we impose the stronger restriction of a common intercept
for all levels of empathy,* i.e., if we accept the hypothesis of a
common origin for each of the five groups, the pattern of slopes is
generally one of increase as we move up levels of empathy. Cross-
classifying on imaginativeness there were only four reversals of this
pattern and cross-classifying on hypotheticals only three. In a
random situation 30 would have been expected in each instance. In none
of the 12 comparisons did more than one reversal occur. The reversals
were mostly slight with no particular pattern in their location.

It seems safe to conclude, then, that if all individuals started
at the same point those in the groups higher on aspects of empathy
would increase on the dependent variable at a somewhat faster rate than
those in the less empathic groups, i.e., the more empathic became more

modern on these three dependent variables somewhat more rapidly per

 

*This restriction was statistically acceptable with the dependent
variable of belief in work, acceptable with a single exception (imagin-
ativeness cross-classifying the relationship with mass media) for
political knowledge, and unacceptable for the four comparisons involving
favorable attitude toward change.

131

unit of media exposure.

If we impose no restrictions on the data, the pattern for inter-
cepts generally remains one of increase whereas the pattern of slopes
generally reverses and becomes one of decrease as we move up levels of
empathy. With the data cross-classified on imaginativeness there
are 11 reversals of this pattern of intercepts and 16 reversals of
this pattern of slopes. In no instance would the opposite patterns
provide a better fit. For one of the comparisons we come close, hows
ever, with four reversals in both the slepes and theintercepts.* In
another we have a perfect fit with no reversals in either slapes or
intercepts.** Cross-classifying on skill with hypotheticals we find
17 reversals of the intercept pattern and 19 reversals of the slope
pattern.

In general, if we stay with a linear model, the best approximate
description of the typical unrestricted situation consists of the more
empathic groups starting at higher levels on the dependent variables
and increasing somewhat more slowly than the less empathic as they
receive more units of exposure. However, the linear model is frequently
a poor fit to the data and at least occasionally a model postulating a
take-off phenomenon would fit better.

It may be noted that the regression coefficients in these
analyses are typically small. While this is artifactual with the

variable of political knowledge, the artifact there can be largely

 

*This is for political knowledge predicted from general exposure
to mass media.

**This is for favorable attitude toward change predicted from
general exposure to mass media.

132

removed by'multiplying the coefficients by 1000. After doing this

the coefficients are still quite small. Since both dependent and
independent variables are standardized in the analysis this indicates
that rates of increase per unit of media exposure are very low; We
may have confidence that they exist, but we are looking at a long range
phenomenon involving cumulative, reinforcing effects rather than a
one-shot change in modernity of attitudes and knowledge resulting from
a single experience with the media. At least, this is likely in the
vast maJority of instances.

If we consider 2512 the unrestricted situation, the general
hypothesis that levels of empathy differentially facilitate the
decoding of the modernization import of the mass media is not supported
if we attend to the slopes (rate of increase per unit of exposure),
but is supported if we attend to the intercepts (points of origin).
Since we have attempted to stop a process by slicing through it at
one point in time, a plausible explanation for this outcome can be
devised which allows us to retain, with modification the hypothesis
that levels of empathy contribute differentially to the decoding of
the mass media. This modification will be discussed in the next

chapter.

VII. REGRESSION ON TRIPS TO THE CITY OF PREFERENCE FOR THE CITY, WITH

AND WITHOUT CROSS-CLASSIFICATION ON ASPECTS OF EMPATHYV ,

The same aspects of empathy are considered in this section as
in the previous section. However, the independent and dependent
variables are different. From a factor analysis of modernization

orientations two factors were selected, one having to do with a

133

preference for living in the city or the attractiveness of the city
to the individual and the other composed of travel to the city and
a negative pole of fatalism or distrust of modern sources of ideas.
These factors were named "preference for city" and "urban exposure"
respectively. The items composing them and the factor weights utilized
in obtaining scores are shown below:
Preference for living in the city
1. Under the same conditions would you prefer to live here
or in the city?
la. Would you like to visit the city (country) from
time to time? (2.157)
2. If you made twice as much in the city (half as much
in the city) would you still prefer to stay here
(stay there)? (2.903)

Urban exposure

 

1. What do you trust when it comes to new agricultural

ideas, education (2) or practical experience (0)? (-.35h)
2. Do you think persons' actions and character are deter-

mined mostly by what they get from.their parents

or by themselves? -.520)
3. In the last year, how often did you go to a large city? .837)
h. Have you visited a friend or relative in the city

during the past month? El.051)
5. What's the most distant point you've ever visited? .h20)

The correlation of preference for living in the city and urban
exposure was .2h. Cross-classifying on skill with hypotheticals raised
the correlation to .25. The increase is not significant (a type V
outcome) and hence it is not useful to examine the pattern of relation-
ship for various levels of skill with hypotheticals.

Cross-classifying on imaginativeness the correlation was raised
to .27. This increase approaches significance at the .066 level.
Strictly speaking, a common regression line may be accepted for this
relationship. Since the patterns of intercepts and slopes in the

unrestricted solution show no interpretable relationship we will accept

13h

the verdict of no significant improvement through consideration of

levels of imaginativeness.

‘VIII. EVIDENCE ON THE PREDICTIVE UTILITY OF TWO ASPECTS OF EMPATEY

In order to get another kind of evidence about the value of
the aspects of empathy which have been under scrutiny in the last
two sections, skill-with-hypotheticals and imaginativeness, these
aspects, as defined by their factor scores, were included with six
other predictors in a series of least squares routines predicting eight
modernization variables, viz. aspirations for children, value for
dissatisfaction, belief in the efficacy of work as opposed to luck,
favorable attitude toward change, urban exposure, preference for
the city, degree to which the individual ponders problems affecting
his community, and political knowledge. The other six predictors
were; claimed use of radio for agricultural information, claimed use
of other mass media, involvement in ACAR programs, trust of the ACAR
extension.agent, level of literacy, and intelligence. Each of these
variables was a factor score derived by appropriate procedures.

The data are shown in Table 16. Predictors retained longest
have higher ranks in Table 16. The sum of these ranks appears on
the right-hand side of the Table. we sequuite clearly that the three
strongest predictors are mass media exposure, literacy, and intelligence,
each of which has a sum of ranks across the eight dependent variables
of #5. These variables are retained in the final prediction equations
five, three, and two times respectively. They are also predominant
in those multiple correlations which reach the most satisfactory

levels. In contrast, imaginativeness is retained twice, once where

135

amva Ho. menanm maoncunw on n20 anemone 0m msennr% s: meHH tHnr revonrmnHomHn can
HssmHsan<oammm u: H: noseannHoo tHnr anrmn enmaHononu on onrn BoamnoHNnnHon onHmonmnHosm$

KGGNwZHN>HH02 §>NH>urmm

 

 

 

 

 

 

wwmanaow <eHcmm umHHmm anon: HUHoxm

<>wH>mem >aanu awuumnn Ho wn<onc exec- ermu swoon onHnHooH was on
anosu Hummnnwos tour ornamm acne anN MMOUHmBm xsothamm waste

mxHHH anr

rvansmnHomHm m» a H b H N N H uo

HammHoenH<m=mmu N m me u u o H we we

wmQHo cum H a N N u H N w No

Onrmn amaHm can a H Ne me mi me u we hm

Ho<oH<mBmon H:

>o>w enomnmau N u u ¢$ a u a as we

Hucmn om >n>w u at m H N N a N Nb

HHanmnw a N or N Ne w m mt am

HsanHHmmsom w N a we a a m we am

 

chnHva non-
HmHmnHo: mmnmn .HN .om .Nm .om .wo .Ho .oo .mm
amHmnHo:

 

ezsavonu H: nun euch Hananmnm are near onamn on amHmnHos on unmaHnnonmm Hnnmmn scavmnu 3mm: arm
enmaHnnon too nonsHsma Hosmon. >nnoHHuwom acetone tons nmnnwsma Ha arm mHsmH nochHoo.

136

it is the best predictor of belief in work, and once where it aids
in predicting political knowledge. Skill with hypotheticals, which
is based on the role-taking items, is retained only once. It is the
lone surviving predictor of aspirations with a correlation of .12.
The only predictor in the group which is inferior to the two aspects
of empathy for prediction of these dependent variables is the use
of radio for agricultural information. This variable has a sum.of
ranks of only 20 and never survives the deletion process.

Skill with hypotheticals performs best (in the deletion sense)
where the multiple correlation is low. If the eight modernization
variables are dichotomized into those which have high multiple
correlations and those which have low multiple correlations, skill with
hypotheticals obtains 77 per cent of its sum of ranks on the lower
half. In contrast the best predictor, exposure to mass media (exclusive
of radio), has 64 per cent of its larger sum of ranks on the upper

half.

CHAPTER IV

SUMMARY, DISCUSSION, AND CONCLUSIONS

I. SUMMARY OF RATIONALE AND PROCEDURES

This study focused on the modernization of the traditional
(and particularly, the rural) individual in developing nations. It
further focused on the contribution of empathy in the modern-
ization process. Empathy was viewed as a capacity for discriminating
subtle cues in the environment and utilizing these cues in a flexible,
imaginative fashion together with other information to optimize the
attainment of goals. The traditional meanings of empathy were examined
and criticized, viz. empathy as insight into the feelings of others,
as interpersonal perceptiveness, and as capacity to place oneself
in others' roles. Two newer approaches were presented which tied
empathy more closely to the modernization process. The first approach,
an extension of Mead's theory of symbolic interaction, suggests that
the more empathic individual deve10ps a more adequate and differentiated

generalized impersonal other which helps him learn from.and interact

 

with strangers, particularly in an urban milieu. The second approach
suggests empathy as an instance of hypothetical thought or general
cognitive flexibility. The more empathic individual is seen as

more capable of examining the components of the decision process,
incorporating relevant new information into the decision matrix, and
working through the consequences of departures from the usual way

of doing things.

Other links were discussed between empathy and modernization,

137

138

and a model was presented of empathy's development and contribution to
the transformation of the traditional individual. ‘We noted that
empathy was only one variable of potential importance in the process
and that other variables, such as literacy (with its presumed impact
on facility with abstract, symbolic manipulation), might make more
direct contributions to the process than did empathy.

Since the discussion of empathy uncovered many'more implications
than could be tested in a single study, we narrowed our focus by
specifying three general areas of concern: (1) the nature of empathy
as a variable, i.e. its dimensionality, stability over time, the
possibility of a new system for measuring empathy, and its relative
importance as a predictor; (2) the place of the more empathic individual
in communities of greater or lesser traditionalism; (3) the relations
between mass media or urban contact and empathy, e.g., whether
aspects of empathy play mediating roles in the impact of messages
coming through these channels.

To investigate these concerns data were gathered in two waves
of interviewing during 1966 in the rural sector of the State of
Minas Gerais, Brazil. From.the first wave 775 usable schedules
resulted; from the second wave 1302 resulted. Many indices of
variables were developed through factor-score procedures; some were
develOped through Guttman scaling. The tests of the hypotheses
involved primarily comparisons of appropriate correlations and/or
regression coefficients. The dimensionality of items tapping aspects
of empathy was investigated through factor analysis. Other factor
analyses were run to investigate empathy's relationship with other

variables of interest. Multiple regression equations were utilized

139

to evaluate empathy's value as a predictor of modern orientations.

II. RESULTS: THE NATURE OF EMPATHY

Factor Analysis

 

The results of the factor analysis of 18 items intended to tap
empathy generally indicated that aspects of empathy can and should be
distinguished. The first principal axis factor accounted for only
16 per cent of the variance, indicating that more than one factor was
necessary to explain the correlation matrix. The behavior of two of
the aspects distinguished subsequently indicated that there may be
some utility in making the distinction.

The Lerner-type role-taking items tended to be more impurely
loaded and to have somewhat lower loadings than those items which
asked about the individual's role knowledge, his interaction with
strangers, or his imaginativeness. The mere fact of counterfactual
format in the Lerner-type item (i.e., "What would you do if you
were______?") did not assure that all items with this format would
load together on a factor. For instance, when the blank was filled
by "the poorest person around here" or "the ACAR agent" the items
did not reliably appear with others calling on the individual to play
the role of President or Prefect. This would appear to discourage
the notion that it is the hypothesizing character of the role-taking
question which is paramount. Rather, what is paramount may be the
substance of the question (e.g., the social distance from.the respondent
of the role in which he is to place himself).

This consideration would appear to cast doubt on the uni-

dimensionality of the role-taking type of question or its utility

1h0

as a measure of an important dimension of modernization. They
consequently discourage the interpretation of these particular role-
taking questions as one major form of hypothetical thought and tend

to suggest that the role-content specified makes most of the difference,
not the counterfactual format.

This conclusion is bolstered by the fact that the person's
ability to imagine others believing differently about something he
strongly believes to be true shows no relationship to any of the
empathy items. Indeed, it has no strong relationship to any other
item in the study.* While the item was badly skewed in the positive
direction (78 per cent said they could imagine someone believing
differently), its lack of relationship to other empathy items seems
to indicate that the area of empathy so measured may not be sub-
sumed so easily under the rubric of "skill with hypotheticals."

0n the other hand, the emergence of self-professed imagination
tr "thinking" dimension) as an important variable may represent a
useful extension of our understanding of mentality of the rural populace
in Brazil. This dimension appears to be somewhat independent of formal
education and measures of intelligence. We will later examine its
relationship to the prediction of other modern orientations. It will
be remembered that imaginativeness generally was more useful as a
classifying variable in the regression analysis than was skill with
hypotheticals.

There is one contradictory piece of evidence which should

be introduced in defense of the role-taking or hypothetical skills

 

*Its highest correlation was with political knowledge (r = .16).

lhl

variable. When the Lerner-type empathy items were summed for each
individual, and these sums were aggregated for the community in the
form of a mean, and this score was submitted to factor analysis
together with other scores similarly aggregated to describe the
characteristics of the community's leaders, empathy emerged as an
important variable in the second factor. This factor appeared to
involve other variables indicating skill in interpersonal relations
at the community level. It should be emphasized that this analysis
is of community rather than individual characteristics. This evidence
would appear to encourage the notion that empathy as measured with
the typical role-taking question is morexelevant to interpersonal
perceptiveness and group harmony than to skill with hypotheticals.

Reliability

 

Over a sixrmonth period the test-retest reliability of an index
of empathy composed of three role-taking items was 29.* In traditional
testing and measurement circles, studies based on this level of
reliability would be discarded. An examination of the other indices
indicated that this aspect of empathy was not the worst in terms

of test-retest reliability and that among the more subjective measures

 

*There is a possibility that certain individuals were mistakenly
paired in this analysis. The pairing process was accomplished by
coders who could read the handwriting of the interviewers and was only
indirectly supervised and checked by the author. Some outright coder
error could have occurred, or two persons with the same name, perhaps
Junior and senior, might have been paired. This latter possibility was
considered in examining the absolute discrepancies between ages of the
respondents over the six months and eight individuals who "became"
more than nine years older or younger during this period were excluded
from the analysis. This exclusion did nothing for the reliability
of the empathy measure. The higher reliability coefficients reached
by the mass media exposure indices, which are likewise somewhat
subjective, tend to discourage the explanation of low reliability of
empathy as a function of coder error.

1&2

it fared rather well. This, of course, does not say much for the
other subjective measures.

It is interesting to note that the reliability of the three-
item index (.29) is not greater than the reliability of the highest
item in that index -- taking the role of the Prefect (.3h). The test-
retest correlations for the other two items were .17 for the President
and .29 for the ACAR agent. If we accept the dictum that additional
items improve the reliability of an index only if the items tap the
same underlying dimension and thus contribute a small portion of true
measurement which is retained in their summation with other similar
items while the error or unique variance is randomly cancelled out,
this appears as evidence that these three items do not necessarily
lie on the same dimension.

To the extent that role-taking ability’amounts to a projective
test, the rationale used by devotees of projective testing to explain
the typically low reliability of projective tests over time might
apply; Low reliability over time is to be expected when the phenomena
under study fluctuate wildly. If peOple with manic-depressive
tendencies, for example, happen to be tested for depression with a
projective test on a day when they are manic, there will be no necessary
correlation a few months later when some have passed into a depressive
state and others have not. If a person's tendency to take a role
adequately when called on is highly dependent upon his breakfast that
morning or his latest visit to the movies or some other idiosyncratic
event, then the score may be quite valid for the moment in which it is
taken, without there being any sizable correlation with a score on

the same measure at a later time.

ll+3

However, to the extent this posture is assumed, the ability
to make predictions about trends for types of individuals on the
basis of empathy is undermined. Unless an individual, clinical
approach is utilized and the person is studied over time with repeated
measures of his empathy to gain a notion of its periodicity or the
events to which it is reacting, the variable has little to recommend
it. And neither our measurement devices nor our research resources
were at a level to justify such intensive study of a single variable.

It is also possible that the differences introduced in coding
empathy between time one and time two made for a lower correlation.
Certainly, had the system finally utilized in Phase II been used with
the Phase I data as well, a somewhat higher test-retest correlation
would be expected. Even if we use the upper confidence limit of the
Obtained correlation (.hO) as the measure that would possibly have been
obtained if the second, superior coding system.had been utilized each
time, the correlation would still not satisfy measurement and testing
specialists.

Thus, we may conclude that empathy, as measured by the survey
techniques here used, is not a very reliable attribute, and it finds
itself in company with even less reliable measures of other important
subjective attributes. The reliability ceiling in these two surveys
would appear to be about 75 per cent common variance. This level was
attained'by reports of age. If the reliability of age is taken as a
criterion of excellence, then empathy indices composed of three items
are about one-eighth as reliable as age in these surveys. The
reliability of empathy could undoubtedly be increased through the

utilization of more items if these were drawn from the same under-

1hh

lying dimension. If the hints contained in the relative reliabilities
of the individual items are not misleading, it would appear that
estimates on the adequacy of role-taking for roles in closer contact
with the individual (the ACAR agent and the Prefect of the Municipality)
are more reliable than estimates based on high-status roles such as

that of the president.

The Adequacy of a New Measure USing Interviewers' Judgments
We found evidence that two new measures of empathy which
utilized interviewer judgments of the degree to which individuals
took the roles called for in 13 items in the Phase II instrument
were equal or superior to the measure utilizing the sum of the empathy
items as coded in the office. One of the new measures correlated .70
with the traditional coder-derived scores on empathy, the other .38.
They were at least as effective as the traditional measure in predicting
modernization behaviors. Indeed, if the two new scales were utilized
together, they outperformed the traditional measure on some variables.
The two new scales were composed of different kinds of items.
The first scale was composed of interviewer-estimates primarily for
role-playing items. The second scale took interviewer-estimates
primarily for items having to do with economic eventualities and
situations. Thus neither one provided a content-free measure of skill
with hypotheticals. However, the ease with which they can be collected,
their apparent utility, and the possibility that they will demonstrate
considerably greater reliability, if only because more items can easily
be included in them, leads to a glimmer of hope about the measurement

problem.

111.5

What may we conclude? It is usually less effort to ask the
interviewers to judge the degree to which subjects placed themselves
into the role-playing questions than to train coders in the reliable
coding of the portion of the response preserved in the questionnaire,
and it is much less time consuming. The scale based on interviewer
estimates of’Lerner-type role-taking items correlates .70 with the
empathy index'based on coder judgments. This would seem to be a
quick way of getting a good measure of empathy.* At the same time,

a somewhat independent estimate of hypothetical thinking can be derived
from.interviewer judgments of the degree to which the respondents
placed themselves in the roles or situations called for in other items
in the questionnaire. This scale or index apparently would make an
important contribution to predicting behaviors like the adoption of
agricultural practices. H0wever, such a scale might become confounded
with the variables the items from which it is derived were originally
intended to measure. Thus, a better approach would seem to lie along
the following lines:

1. If interested in empathy, ask the traditional role-taking
empathy questions.

2. Ask the interviewers to judge whether or not the
respondents placed themselves in the roles called for
(2 if "yes," 1 if "not sure," and 0 if "no" .

3. Sum these interviewer judgments as a measure of empathy.

R. If there is energy and time available, scale them for
greater precision.

 

*This will be particularly true if the judgments of the inter-
viewers are submitted to Guttman scaling procedures; but very probably
the trichotomous scoring system used would correlate highly with the
Guttman scale which refines it, and it may not be worth the extra
effort of scaling.

1M6

If our experience is any guide, this should give an estimate
which is as good as, if not better than,the original index derived
from the training of coders_ in terms of predicting other modern
behaviors. Since this approach puts less burden on the researcher
at the point of the coding of data, more empathy items might be
included to improve reliability and increase variance. Very possibly
items of a non-political nature would make a‘better contribution than
the political items which have come into greater currency in recent
studies. This hunch will require more study before it can‘be said
to have empirical support, however, since the results with the fOur
items we used were based on an analysis suggested‘by the data rather
than derived from a prior theoretical rationale and subjected to

testing.

Empathy‘s Value as a Predictor

 

The measures of empathy derived from interviewer ratings and
the empathy index produced a multiple correlation of .40 with average
adoption levels from the Phase II data. When the individual items
composing the empathy index were compared for predictive utility with
the two Guttman scales formed from interviewer judgments we found that
the non-political items from the index tended to survive longer. This
suggests the possibility that the use of only political items in
empathy indices subsequent to Lerner's initial presentation of
empathy may lower empathy's predictive value. ‘We would urge that
future operationalizations include non-political roles. Other kinds

of hypothetical items might also be considered.*

 

*It may be worth noting that the original index utilized by
Lerner contained only three explicit role-taking items. The other

1&7

The aspects of empathy derived from Phase I data by factor
analysis did not fare well in comparison to literacy, intelligence,
and mass media exposure when all were placed in eight separate least-
spares-deletion routines. Mass media exposure, literacy, and intelligence
survived deletion longest and were prominent in solutions with more
satisfactory levels of multiple correlation. Imaginativeness was
retained twice and skill with hypotheticals only once in the eight
tests. Apparently empathy is a legs important consideration in the
prediction of at least the eight variables we considered than intelligence,
literacy, or mass media exposure.

We can tentatively draw a number of conclusions about the
empirical nature of empathy as usually measured, about its utility
as a predictor, and about substitute measures for the concept. These
are summarized below:

1. Empathy as skill with hypotheticals or role-taking can be
distinguished from other empathy type items as a separate

factor in a factor analysis.

a. Its loadings, stability and purity will not be as
adequate as those for other factors.

b. Not all items in the counter-factual role-taking
format (i.e. "What would you do if you were ?")
will load together.

2. Two other aspects of empathy may be distinguished:

 

six items were culled from a variety of areas, some of them involving
estimates about mass media penetration, others imagination of oneself
in another country, etc. Perhaps because the theoretic description of
the index given by Lerner centered on the role-taking character of the
index, or perhaps because the name "empathy" adopted for it led back
to role-taking conceptualizations, the role-taking type of item was
the only one to survive in subsequent uses of the empathy index

(e.g., Frey, Eister, Rao, and Rogers and Herzog). The conceptual-
ization of the index as involving skill with hypotheticals or capacity
for imagination might have avoided this concentration on a single

type of item.

1A8

(l) imaginativeness and (2) role knowledge plus
facility in dealing with strangers.

a. The latter aspect may be further subdivided into two
components.

b. The latter aspect emerges earlier in the factor
analysis, and remains stable until it subdivides.

3. When empathy items are factor analyzed together with a large
pool of items intended to measure other characteristics, they
tend to cluster together around the imaginativeness items.

a. The role-knowledge items are exceptions; they load with
other knowledge indices.

b. Role-taking items and facility with strangers have lower
loadings than the imaginativeness items.

c. Items on thinking about community problems and their
solutions load in as important components of the factor.

k. If empathy is considered on a community level, with data
aggregated to represent the characteristics of community leader-
ship, then it loads together with indices of trust, exchanges
of best-friend nominations, and utilization of radio. This
appears to imply an interperson-perceptiveness utility among
the role-taking items rather than a skill-with-hypotheticals
dimension.

5. As a predictor of other modernization orientations and attitudes
the measure of empathy focusing on imaginativeness is retained
more often in the prediction equations than the measure
focusing on skill with hypotheticals, but neither is as
important as mass-media exposure, literacy, or intelligence.

6. Empathy as role-taking has low test-retest reliability.

7. A more adequate measure of empathic role-taking may well be
obtained from interviewer ratings than from.eoder judgments,
and the saving in coders' time might justify the inclusion
of more interviewer-rated role-taking or hypothetical items,
thereby possibly increasing reliability of the measure.

Critique of the Conclusions

 

There are a number of possible artifacts which may explain the
relatively poor showing of the role-taking items. First, the coding
of the items may not have focused on the adequacy of the role-taking

represented by the responses preserved in the questionnaires. The

1&9

coding system was subsequently changed to record degrees of role-
taking more adequately. The coding systems utilized the degree of
specificity of the response as a criterion of more thorough role-
taking. The respondent who said more (or whose interviewer bothered
to record more) was almost certain to obtain a higher score.

Second, a factor analysis can produce factors only for the
data that are put into it. The context of the analysis is highly
important. We saw this when we ran the factor analysis with a larger
pool of items or when we first aggregated the data to the community
level. Perhaps more careful thinking-out of the items to include in
the factor analysis of empathy would have resulted in a more powerful
role-taking factor.

In the process of trying to cover many variables and yet keep
the research instrument brief, we were led to include, in some cases,
only a single item or two items to measure a particular area.

Instead of cursory measures of many areas, the result was a series of
inadequate measures which were of no value whatsoever. Rather than
eliminate items because they showed correlation with other items, we
should have eliminated variables of less central concern and interest
and included more correlated items tapping the same conceptual areas.

Despite all this, there appears to be sufficient evidence to
make us reconsider the value of the usual role-taking items. Their
unidimensionality is not secure, they are troublesome to code, and
they are of less utility, apparently, than other variables for pre-
dicting orientations and characteristics at the individual level. At
an aggregate, community level, where they may function to indicate

some kind of interpersonal skill, they may have more value. We will

150
examine them further as we continue.
III. EMPATHY AND DEGREE OF COMMUNITY TRADITIONALISM

Our results here are a washout. We have no evidence to show
that the more empathic individual is any more or less integrated,
befriended, respected, or sought for information in the less traditional
community than he is in the more traditional community. This may be
because of contradictory forces working to nullify any difference.

It may be an artifact of poor measurement, or of inadequate variance
in degrees of traditionality. It may have resulted from.the crude
dichotomization of the sample and from comparisons of correlations
where results for the extremely less traditional or more traditional
communities were obliterated by the large middle range of communities.

A more probable explanation may be sought in terms of an in-
adequate conceptualization and thinking-through of the links among
community development, degrees of modernization and individual
characteristics. Young and Young have proposed that scales of
institutional development and of community articulation with the
national life may be useful indicators of modernity. (Young and
Young, 1962) And we were able to generate Guttman scales for the
Brazilian communities represented in the sample by following their
procedures. Despite this, we have not succeeded in relating these
scales to anything of interest.* The possibility exists that we have
measured as much the institutions of tradition as the institutions

of modernity. In other words, the presence of a bar, barber shop,

 

*For more details, see Whiting 33. 21., 1967.

151

or rural school may ngt'be an indication that the forces of modernity
are invading the community. The supposition that institutional deve10p-
ment in the community means more modern development should be re-
examined.

Since three bases for dichotomization were obtained, and since
such ostensible indicators of modernity as health situation, level
of farm.power-supply, and isolation were included in these indices,
some more clear-cut results should have emerged if any strong relation-
ship existed, in the manner hypothesized, between empathy and tradition-
ality. Tentatively, the general hypothesis that a more empathic
individual has a somewhat different position and role in a more

modern community must be judged to be without support.

IV. EMPATHY AND THE MASS MEDIA

Correlations with Imaginativeness

Our finding here represents a complete reversal of our hypothesis.
It is surprising in view of the factor analysis results where exposure
to radio tended to load more highly with imaginativeness than with
print exposure. Apparently even though radio exposure relates some-
what more to imaginativeness than to anything else, print exposure
relates to imaginativeness even more strongly. ‘We cannot attempt to
sort out the possible causal nexus of this relationship here. The
outcome does require a reworking of our thinking about the two types
of media exposure and imaginativeness.

In preparation for the next section, and as a possible explanation
of the reversal of the hypothesis found in this section, it may be worth

noting that not much of what is available in most br0adcast media has

152

much direct utility for the rural dweller. Music, most of it of the
rock-and-roll variety, pervades most of the airways. News is avail-
able periodically and a government-controlled review of current and
forthcoming decrees occupies two hours of prime evening time. .A
farm program.of one half hour is available just before this review.
unfortunately, we have no direct evidence of farmer preferences in
radio programs, but if the farmers in our study are like others studied
in.Brazil, the farm programs available are among the least popular
programs on the radio. (See Whiting and Guimaraes, 1967, Chapter VII
for details.)

The print media present a greater variety of contents, even
in rural areas. There is no reason to believe that the slick
illustrated magazines and sensational daily press could not stimulate
and fire the imaginations of their readers. However, we have no data
on the actual types of print materials reaching the farmers. Mbst
available cinema is, of course, unadulterated entertainment and
escapism, with less value for modernization than for instilling

tendencies to reverie.

Empathy as a Mediating Variable of Mass Media Effects

We eliminated, first off, three of the dependent variables --
aspirations, value for dissatisfaction, and tendency to think about
problems -- because mass media exposure did not have any significant
correlation with these variables, even when we cross-classified on
aspects of empathy. Three dependent variables remained: belief in
work, political knowledge, and attitude toward change. Two varieties
of mass media exposure were used as independent variables and two

aspects of empathy were used as a basis for cross-classifying the

153

independent variable into five levels.

Let us first consider the strength of the relationships
between the dependent and independent variables. Without classification
on any aspect of empathy, the correlations between radio exposure and
the dependent variables were generally lower than the correlations
between mass media exposure and the dependent variables. The average
correlation with mass media exposure was .hl while the average
correlation with radio exposure was .15.* When we classified on
aspects of empathy and allowed both intercepts and slopes to vary
freely, a moderate improvement occurred in the size of the correlations.
This improvement was significantly larger for radio than for mass
media and significantly more attributable to cross-classification on
imaginativeness than on skill with hypotheticals. The interaction of
type of cross-classification with type of media was not significant.

We may conclude that, as these variables were measured in this
study, cross-classification on imagination is more valuable than
cross-classification on skill with hypotheticals. We may also conclude
that such cross-classification will help increase the strength of the
relationship between radio exposure and the dependent variables more
than it will help increase the strength of the relationship between
print exposure and the dependent variables.

If we consider only the unrestricted situation, the general
hypothesis that empathy facilitates the decoding of the modernization

import of the media is not supported if we attend to the slopes but

 

*For print the correlations were .59, .37, and .21 while for
radio they were .18, .18, and .07. The averages were calculated using
a Fisher g transformation.

15h

is supported if we attend to the intercepts. We may modify our
hypothesis'by suggesting that the more empathic do indeed decode the
modernization import of the media more effectively, that is how they
became more modern on these variables. But that at the point in time
at which we are considering them their rate of increase has slowed
down, simply because they have hit the top of our scales or because
more units of exposure no longer make much difference to them. Those
of lesser empathy, on the other hand, are still increasing somewhat
more rapidly the modernity of their knowledge and orientations per
unit of'mass media exposure, but they start from.a lower level because
they have been less exposed in the past. This interpretation makes
sense in a take-off model, where initially the slope for the least
empathic is low (perhaps they are still learning the conventions of
the media, including the convention of empathy) and then, after their
skills with these conventions reach a certain point, the modernity

of their attitudes accelerates as they receive more messages through
these media. Their empathy also increases but after a time the effects
on the attitudes from increases in media exposure slow down.

Thus, the fault with the earlier hypothesis was oversimplification
with respect to its time-dimension. When the assumption of a common
starting point at the moment in time when the data were collected is
made, the hypothesis is upheld -— the more empathic do appear to
increase the modernity of their orientations more rapidly, But this
obscures the relationship, for the less and the more empathic do 223
all start from.the same post at the moment we study them. Since the
more empathic have already been exposed to mass media when we arrive

to measure them, they are already at the top of the measure and their

155

slope accelerates less than that of their less empathic neighbors;
they are more similar to the unempathic who are being influenced very
little more for each additional unit of media exposure.

A note of caution is necessary here. We have not actually
been able to manipulate the same individuals' characteristics and
note the changes explicit in the hypothesis. Instead, we have, in
effect, been pairing individuals who are alike on two variables and
different on the third so as to infer the causal effects. Clearly
there is danger in this and we are on different grounds when we shift
scores around among people by mathematical operations than when we
set out to change the actual values of the hypothesized determinants.
Blalock gives an instructive example of this difference but it is
too lengthy to be repeated here. (See Blalock, l96h, pp. LO-he.)

There are two possible objections to this interpretation.
First, the whole analysis of the patterns of regression lepes and
intercepts may be attacked on the basis that we are interpreting a
‘regression artifact, similar to the familiar artifact of regression
toward the mean which is a danger in experimental studies utilizing
change scores and constructing groups on the basis of scores on an
imperfectly correlated variable (see Campbell and Stanley, 1963,
pp. 10-12 for a lucid discussion). Some of the unrestricted relation-
ships graphed in the last chapter do look strikingly as if they are
regressing toward a grand mean, and it is also the case that the
individuals were classified according to the levels of their scores
on a variable which is known to be 3251 imperfectly correlated.
However, we are not dealing with an.experimental situation, a fact

which we just bemoaned but now can console ourselves with, and we

 

156

are not utilizing change scores. Instead we have a more complex
relational measure which is being compared, not among the extremes
only but across the entire range of the data. The circumstances under
which the patterns of relationships found here could be generated by
a regression artifact are very complex, so much so that they are not
worth the effort of trying to work out mathematically. They could

be obtained only by Monte Carlo methods. An explanation of these
results as artifacts of regression toward the mean is highly
unparsimonous.*

Another objection may be that the results presented do not
constitute evidence that empathy aids in the decoding processes, for
the same kinds of results would have‘been obtained had guy variable
having positive correlations with both the independent and dependent
variable been used to cross-classify the data. If the latter state-
ment is true, we may cheerfully note that there has been no intent to
imply that empathy was the 2211 variable which mediated the decoding

process or even the most important one: more importantly, the validation

 

process in science works on the principle of proposing a relationship
and seeing whether the data lead to a rejection of the proposal.
Models of relationships can only be discarded by empirical test, not
definitively adopted. When they are adopted it is only because
superior models have not yet been found. Whether or not empathy is
532111 mediating the decoding process is certainly not ascertainable

from these data or any data. All we can say is that the results still

 

*This is a summary of the opinions expressed by Dr. Dennis
Gilliland of the Department of Statistics and Probability, Michigan
State University, with whom I consulted on this point.

met

._._'__-——_= ‘-—.-¢.—-——.

O _ :"_.'_.:

 

157 ff '-

admit that possibility.

The relationship between travel to the city and attraction to
the city was significant, though small, but was not improved by cross-
classification on either aspect of empathy. Hence it is of little
interest in the discussion of empathy’s role in the modernization
process.

To the list of conclusions offered on pages 1&7 and 1&8 we
can now add the following:

1. Empathy shows no differential relationship to the social
roles, integration, or function of individuals in communities
of greater and lesser traditionalism.

2. The relationship between the "imaginativeness" aspect of
empathy and exposure to print is significantly stronger
than the relationship of this aspect with exposure to radio.

3. Mass media exposure is significantly more related to three
dependent variables: (1) high valuation of belief in work,
(2) political knowledge, and (3) a favorable attitude toward
change, than is radio exposure.

A. In relationships involving both types of media exposure, the
improvement obtained by cross-classifying on the imaginative-
ness aspect of empathy is significantly greater than the

improvement obtained by cross—classifying on the skill with
hypotheticals aspect.

 

 

5. With few exceptions, the pattern of intercepts of the
regression of modern orientations on exposure to media
rises as we move up levels of empathy, when a common slope
is employed.

6. With few exceptions, the pattern of lepes of the regression
of modern orientations on exposure to media rises as we move
up levels of empathy, when a common intercept is employed.

7. With numerous exceptions, the pattern of intercepts rises
and the pattern of slopes falls for the regression of modern
orientations on exposure to media as we move up levels of
empathy, when both intercepts and slopes vary freely.
Frequently a curvilinear model employing a take-off effect
might more adequately represent some of the data.

8. Such a take-off pattern would involve a modification of the
hypothesis of empathy's mediation of the decoding of the

 

158

modernization import of the media to include (1) an initial
learning of the conventions of the media, including the
inculcation of empathy, (2) a somewhat more rapid growth

in modernity of orientations as units of media exposure are
added in the context of moderate levels of empathy, and

(3) a leveling out when, with higher empathy, rates of 3;
additional growth in modernity of orientation'become slower.

V. CONCLUDING COMMENTS

These findings require much more research before any confidence
can be placed in them. Some of the lines of research possible on the
data collected already have been intimated. It appears there might be
profit in investigating the possibilities of a take-off phenomenon;
in investigating, through partial regression coefficients, the relative
importance of the empathy variables to the total set of variables that
may assist in the decoding process; and, possibly, in abandoning equal
sizes of groups, classifying individuals according to the extremity
of their empathy scores and looking for discontinuities in the scale
scores to make cutting points, taking into consideration the normality
of the distribution. In a separate area, further analyses with the
new measure of empathy could give indications of its predictive
adequacy in its unscaled form and with a wider variety of dependent
variables.

It would seem.wise in the future to pay attention to the aspect
of empathy here called imaginativeness, since this was stronger, more
stable, and more effective than the aspect we dubbed skill with
hypotheticals. However, a new measure of empathy involving interviewer
ratings might tap the latter more than the former, and with improve-
ment in measurement stronger relationships with hypothetical skills

might be found.

n‘ifi ins-i..- an...

159

The inadequacies of the data gathered for this study point to
the need for another round of empirical evidence to re-test and amplify
these findings. The lack of positive results for the relationship
between empathy and the respondent's roles in communities of varying
traditionalism.is particularly challenging.

Perhaps the empirical outcomes of the study appears to fall far
short of expectations. Certainly they do not confimm every hypothesis
advanced, and the hypotheses tested were only a small selection from
the broader concerns with the linkages of empathy to modernization
that were outlined in the first chapter. In rationalizing these meager
results, the words of Campbell and Stanley relative to disappointment
with experimental designs are apropos:

we must instill in our students the expectation of tedium.

and disappointment and the duty of thorough persistence, by
now so well achieved in the biological and physical sciences.
We must expand our students' vow of poverty to include not

only the willingness to accept poverty of finances, but also
a poverty of experimental results. (Campbell and Stanley,

1963: P' 3)

‘m!! I,“ ->___i
'J

BIBLIOGRAPHY

Berlo, David K. , The Process of Communication, New York, Holt, Rinehart
and Winston, 1960.

Blalock, Hurbert M. Jr. , Causal Inferences in Nonexperimental Research,
Chapel Hill, North Carolina, university of North Carolina Presé}“ig6h.

Bordenave, Juan Diaz, "The Search for Instrumental Information among
Farmers of the Brazilian Northeast," Ph.D. Dissertation, Department
of Communication, Michigan State University, 1966.

Bostian, Lloyd R. , and Fernando C. Oliveira, "Relationships of Literacy
and Education to Communication and to Social and Economic Conditions
on Small Fame in Two Municipios of Southern Brazil," Paper read
at the 1965 meeting of the Rural Sociological Society, Chicago, Ill.

Campbell, Donald T. , and Julian C. Stanley, Eerimental and Quasi-
Experimental Designs for Research, Chicago, Rand McNally 8: Co. ,

1963.

Cline, Victor B. and J. M. Richards, Variables Related to Accuracy in
Integegsonal Perception, Group Psychology Branch, Office of
Naval Research, University of Utah, 1958.

 

Coutu, Walter, Emergent Human Nature, a Symbolic Field Interpretation,
New York, A.A. Knopf, 1919.

Cronbach, Lee, "Processes Affecting Scores on 'Understanding of Others'
and 'Assumed Similarity'," Psychological Bulletin, 52 (1955),

pp. 177-193-

, "Proposals Leading to Analytic Treatment of Social Perception
Scores ," in Person Perception and In}erpersonal Behavior, Renato
Tagiuri and Luigi Petrquo (eds. ), Stanford, California, Stanford
University Press, 1958, pp. 353-379.

 

Department of Connnunication, "Annual Progress Report for Research
Project on Diffusion of Innovations in Rural Societies ," July 1,
1967 , Mimeo.

Deutsch, Karl W. , The Nerves of Government: Models of Political
Communication and Control, New York, The Free Press of Glencoe,

1961+.

Eister, Allan Wardell, "Critical Factors in the Modernization Process
in West Pakistan," Year Book of the American Philosophical Society,

1962. pp- 362-365-

 

 

160

161

Foa, U.G. , "Empathy or Behavioral Transparency?" Journal of Abnormal
and Social Psychology, 56 (1958), pp. 62-66.

 

Frey, Frederick W. , "The Mass Media and Rural Development in Turkey,"
Report No. 3, Rural Development Research Project, Cambridge
Massachusetts, Massachusetts Institute of Technology, 1966.

Hagen, Everett E. , 0n the Theory of Social Change, Homewood, Ill. ,
Dorsey Press, 1962.

Hastori, A. H. and I. E. Bender, "A Caution Respecting the Measurement
of Empathic Ability," Journal of Abnormal and Social Psychology,

#7 (1952). pp. 57h-5761

Herzog, A. William, "Selection of the Sample for Phase One," Mimeo,
September 1+, 1965.

\
Hobart, Charles W. , and Nancy Fahlberg, "The Measurement of Empathy)" )'
American Journal of Sociology, 70 (1965), pp. 595-603.

Hurley, Neil P. , "The Role of Commmications Technology in Democracy,"
in Communications Explosion, Paper Number 9, Program of Policy
Studies in Science and Technology, George Washington University, 1965.

Katz, Robert L. , Empathy: Its Nature and Uses, New York, The Free Press
of Glencoe, 1963.

Keith, Robert F. , and Everett M. Rogers, "Phase II Concepts: Definitions
and Suggested Measurement," February, 1966.

Langfield, Herbert S. , The Aesthetic Attitude, New York, Harcort,
Brace and Co. , 1920.

Lassey, William R. , Alfredo Mendez, and F.B. Waisanen, "The Consequences
and Meaning of Literacy for Attitudes and Social Behavior,"
preliminary draft, no date.

Lerner, Daniel, "Enlightenment and Communication," in Comparative
Theories of Social Change, Hollis W. Peter, ed., Ann Arbor,
Michigan, Foundation for Research on Human Behavior, 1966.

, The Passing of Traditional Society: Modernizing the Middle
East, New York, The Free Press of Glencoe, 1958.

Mead, Margaret, Continuities in Cultural Evolution, New Haven, Yale
University Press, 1961+.

Rao, Y.V. Lakshmana, "Communication and Development, A Study of Two
Indian Villages," Ph.D. dissertation, Department of Journalism,
University of Minnesota, 1963.

 

 

 

 

162

Rogers, Everett M. , "Diffusion of Innovations in Rural Societies:
Resume of a Research Project Conducted by Michigan State University
Under Contract with the U.S. Agency for International Development,"
1965, Mimeo.

, "Motivations, Values, and Attitudes of Subsistence Farmers:
Toward a Subculture of Peasantry," Paper presented at the Agri-
cultural Development Council's Conference on Subsistence and
Peasant Economics, East-West Center, University of Hawaii, Honolulu,
Hawaii, February 28-March 6, 1965.

Rogers, Everett M. and William A. Herzog, "Functional Literacy among
Colombian Peasants ," Economic Develgpment and Cultural Change, 11!»

(1966)) PP: 190’2030

Schramm, Wilbur and W. Lee Ruggels, "How Mass Media Systems Grow,"
in Communication and Change in the Developing Countries, Daniel
Lerner and Wilbur Schramm (eds.), pp. 57-75.

Smith, David, and Alex Inkeles, "The OM Scale: A Comparative Socio-
Psychological Measure of Individual Modernity," Sociometry, 29

(1966). pp. 353-377.
Smith, Henry Clay, Sensitivity_to People, New York, McGraw-Hill, 1966.

Spector, Paul, 32. 11. , "Communication and Motivation in Community

Development: An Experiment, " Washington D.C. , Institute for
International Services, 1963.

Steward, Don, and Thomas Hoult, "A Social Psychological Theory of the
Authoritarian Personality," American Journal of Sociology, 65

(1959). pp. 27h-279.

Whiting, Gordon C. , "A Summary of Data Collection on Phase One of the
Project on Diffusion of Innovations in Rural Societies: Brazil
Operations," Mimeo, March, 1966.

Whiting, Gordon 0. and Lytton Guimaraes (eds. ), Communicac‘ao de
Inovanes: Pesquisas Aplicaves ao Brasil, Rio de Janeiro,
USAID, forthcoming in 1968.

Whiting, Gordon C., William A. Herzog, Gustavo M. Quesada, and
David H. Stanfield, Innovation in Brazil: Success and Failure of
Agricultural Programs in 76 Minas Gerais Commities, Department
of Communication, Michigan State University, East Lansing, Michigan,
November, 1967.

Young, Frank W. and Ruth C. Young, "The Sequence and Direction of
Community Growth: A Cross-cultural Generalization," Rural Sociology,

27 (1962). pp. 37h-386.

 

 

 

163

infirm“
vié-q

APPENDIX A

ITEMS USED IN THE FACTOR ANALYSIS OF EMPATHY

"If you could change things around here in (COMMUNITY)
would you change a few things or leave things as they are?

 

1. "Let's suppose, for instance, that cows gave #0 liters of milk
at each milking; what would you do if the world were like this?

(CODES: 0 Don't know, doesn't respond, responds negatively

saying he wouldn't like it or think it good.

1 = Be satisfied, continue as usual, use the extra
money for luxuries, travel, or other consumatory
activities.

2 3 Buy more cattle, improve the farm, try to make
progress and other responses indicating an instru-
mental orientation.but lacking in originality or
specificity.

3 = Original and useful reactions such as modernizing
the dairy Operation, reduce the number of cows to
more effectively utilize pasture, turning former
pasture lands to cropping, etc.

2. "Have you ever thought what it would be like to live in your great-
grandfather's time?"
0. No Yes
1. Don't know How often?
1. Seldom
2. Often

3. "Have you ever imagined yourself being a great hero?"
(Same codes as 2 above)

h. "Have you ever imagined yourself being the President of Brazil?"
(same codes as 2 above)

. "If you were the President of Brazil, what would you do?"

 

. "If you were the Prefect of this county, what would you do?"

 

5
6
7. "If you were the poorest person around here, what would you do?"
8

. "Have you ever thought of questions like these?"
(same codes as 2 above)

9. "Do you know what the duties of the President of'Brazil are?"
0. thhing l. A little 2. A good deal

10. "Do you know what the duties of the prefect of this county are?"
0. Nothing 1. A little 2. A good deal

 

12.

13.

1h.

15.

16.

17.

18.

16h

"Let's suppose that you met a foreigner for the first time.
If he looked friendly...do you think that you would feel
at ease with this stranger?

0. No 2. Yes (1. Don't know)

"Would you know what to say to him?"
0. No 2. Yes (1. Don't know)

"Would you be very interested in getting to know him better?"
0. No Yes Don't know
"Do you believe that, given sufficient
time, you would understand why this
foreigner thought and acted differently
from you?"
1. No 2. Maybe 3. Yes

"New think of something which you strongly believe. (PAUSE)
Have you ever thought that someone could believe about it
differently?

0. No 1. Maybe or don't know 2. Yes

"If you were the ACAR supervisor, what would you do to improve
the work?"

"How cooperative was he?" (Asked of the interviewer)
(Code: a rating from 0 to 6)

Number of times the respondent had opinions on 13 Opinion items
concerning issues.

Number of times the respondent had opinions (did not say "don't
know" on items 5,6,7,1l,12,l3,1h) on hypotheticals.

 

"I7'1?NN's/ililflfli'i‘flliflr