THE RELATIVE ACCURACY AND
RELIABILITY OF THE CONGRUITY AND
BELIEF CONGRUENCE MODELS AS
PREDICTORS OF COGNITIVE INTERACTION:
A REP‘LICATION AND EXTENSION

Thesis for me Degree of Ph. D.
MICHIGAN STATE UNIVERSITY
WILLIAM ELROY TEDRICK
1968

 

THuoib

This is to certify that the

thesis entitled

The Relative Accuracy and Reliability of the
Congruity and Belief Congruence Models as
Predictors of Cognitive Interaction: A
Replication and Extension

presented by

William E. Tedrick

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Communication
A
/ 7—7 (7
- 7M]? *1 / ’7// :j/

Date _

O~169

 

 

 

 

 

 

 

 

w:

I-II’FDAIA v‘. ‘1

Jiffhif-Ian S j

U11" vcrsity

«you Mr '
,.

   
 

' amnmc an ”5"
TIME & SUNS'
BOOK IINIIERY INC.

' """Y BINDERS
II! [In-I‘ll

      

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

fulfillment of the requirements for the Doctor of Philosophy degree.

Director of Thesis
Guidance Cal-littee: waif CE Zflfi/Qt , Chairman

4" I A.“

II ’/ ,

Jigie

WW %;

 

‘_.L

ABSTRACT
THE RELATIVE ACCURACY AND RELIABILITY OF THE
CONGRUITY AND BELIEF CONGRUENCE MODELS AS

PREDICTORS OF COGNITIVE INTERACTION: A
REPLICATION AND EXTENSION

by William Elroy Tedrick

The two major objectives of this investigation were to
test the relative accuracy and the relative reliability of
the congruity and belief congruence models as predictors of
cognitive interaction resulting from certain word combinations.

Cognitive interaction is defined as the process by
which a single evaluative meaning emerges as a result of com-
bining two stimuli, each having their separate meaning.
Previous research indicated rather clearly that the belief
congruence principles represent a superior explanation of the
underlying cognitive processes inherent in cognitive inter-
action. In a sense, the present investigation represents a
replication of the earlier study, as well as incorporating
several important extensions designed to provide further
empirical evidence relative to the two predictive models that
were tested. Among these extensions were: the test of
reliability, a research design improvement, the use of a
certain type of word combinations not generally tested be-
fore, and certain recommendations that may help account for
a Specific type of overassimilation for which the models, as

presently formulated cannot account.

William Eery Tedrick

Thirty-six assertions of a specific type not generally
tested before were administered to 619 high school age stu-
dents who had been randomly assigned to 12 different test
groups. The study was designed so that six of the 12 test
groups rated assertions in which the subject was held con-
stant and the characterization was varied. The remaining
six test groups rated the same set of assertions with the
subject of the assertion varied and the characterization
held constant.

Appropriate semantic differential scales of the Evalu-
ative Factor were used to obtain the relevant connotative
meaning measurements required by the congruity and belief con—
gruence procedures.

Results were analyzed in terms of predicted and

 

obtained scores. Predicted scores were calculated by apply—

 

ing the semantic differential scores and the relative import-
ance rating to the congruity and belief congruence formuli.
The obtained score represented the actual evaluative meaning
respondents assigned to the assertions as determined by the
semantic differential scaling procedure.

The results indicate that:

l. The belief congruence model is significantly
superior to the congruity model as a predictor of cognitive
interaction resulting from certain word combinations. Based

on gross mean error of prediction, the congruity model showed

William Elroy Tedrick

.75 scale units of error per assertion while the belief con-
gruence model manifested only .33 scale units per assertion.
When compared to the congruity model, the belief congruence
model predicted more than twice as many assertions accurately
(congruity model, 25 percent of the 72 assertions; belief
congruence model, 67 percent of the 72 assertions).

2. When tested by the equivalent group technique,
the belief congruence model is significantly more reliable
than the congruity model as a predictor of cognitive inter-
action.

3. For the assertions tested in this study, the
relative influence of the characterization component is sig-
nificantly greater than the influence of the subject com-
ponent. A positive correlation of .76 was observed between
the evaluative scores assigned to the characterization com-
ponent in isolation and the evaluative score assigned to
the combined word combination (assertion).

In fact, the evaluative scores assigned to the charac-
terization component are themselves a reasonably good indicator
of interaction effect for certain word combinations.

4. The present belief congruence procedures do not
take into account a specific type of overassimilation. In order
to do so, it is apparently necessary to obtain a measurement of
the direction of the overassimilation relative to the character-
ization component, independent of the evaluative scores assigned

to the characterization in isolation.

THE RELATIVE ACCURACY AND RELIABILITY OF THE
CONGRUITY AND BELIEF CONGRUENCE MODELS AS
PREDICTORS OF COGNITIVE INTERACTION: A

REPLICATION AND EXTENSION

by

William Elroy Tedrick

A THESIS

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

DOCTOR OF PHILOSOPHY
Department of Communication

1968

ACKNOWLEDGEMENTS

I find it extremely difficult to express adequately
the appreciation deserved by so many individuals who Spent
hours in my behalf throughout my advanced degree program.

My deepest appreciation goes to Dr. Gerald Miller for the
expert professional guidance and assistance offered as
Chairman of my doctoral committee and Director of the
dissertation. Also, to my friend and former colleague,

Dr. Gary Seevers, for a great amount of help on the statis-
tical analysis so vital to the success of the study.

For continued assistance throughout my degree program,
I would like to thank Doctors Milton Rokeach, Randall Harrison,
Santo Camilleri, Everett Rogers, and Mason Miller. Their
counsel and guidance as members of my doctoral committee were
a helpful and rewarding experience.

A note of appreciation is extended to the Michigan
Cooperative Extension Service, its administrators and my
fellow colleagues for their understanding support. Special
thanks is reserved for those in the College of Agriculture
and Natural Resources who supported and made possible my
sabbatical leave and financial support that helped to make
it all a reality.

For typing this manuscript, I would like to express
my appreciation to Mrs. Margaret Hudson. Her cheerful and

efficient work made the many revisions more acceptable.

ii

A sincere debt of gratitude is extended to my wife,
Wanda, and the children, who all too often were required to
defer their fondest wishes so that their father could go about

his academic endeavors.

iii

CHAPTER

I

II

III

TABLE OF CONTENTS

INTRODUCTION 0 (.3 O O O O O O O O O O O

The Congruity Principle and
Cognitive Interaction . . . . . . .
The Belief Congruence Principle
and Cognitive Interaction . . . . .

Empirical Test of the Congruity Versus

the Belief Congruence Principles as
Predictors of Cognitive Interaction
The Type of Assertion and Cognitive
Interaction . . . . . . . . . . . .
The Objectives of the Investigation
and Relevant Hypotheses . . . . . .

Objective 1 . . . . . . . . . . .

Objective 2 . . . . . . . . . . .

METHOD I I O O O O O O O O O O O O O 0

subjects. 0 O O O O O O O O O I O O
The Assertions and Their Components
Research Design . . . . . . . . . .

The Measurement of Evaluative Meaning

The Measurement of Relative Importance
and the Computation of Cognitive Inter-

action by the Belief Congruence Principle

Computing Cognitive Interaction by
the Congruity Principle . . . . . .
Administering the Test Materials. .
Preparation of Data for Machine

Processing. . . . . . . . . . . . .

RESULTS 0 O O O O O O I O O O O O O 0

Test Group Equivalency. . . . . . .

Test of the Hypotheses. . . . . . .
Objective 1 . . . . . . . . . . .
Objective 2 . . . . . . . . . .

The Relationship of the Assertion (CS)
Components to the Ability of the Congruity
and Belief Congruence Models to Predict

Cognitive Interaction Accurately. .
The Belief Congruence Model and
Overassimilation. . . . . . . . . .

iv

Page

10
ll
11
12
l4
14
15
18
20
23

28
30

32
33
33
42

42
59

62

65

CHAPTER

IV

BIBLIOGRAPHY
APPENDIX A .
Part I . .
Part II. .

APPENDIX B .

TABLE OF CONTENTS (Cont.)

CONCLUSIONS, DISCUSSION, AND
IMPLICATIONS FOR FUTURE RESEARCH . . .

Conclusions Relative to Accuracy of
Prediction of Cognitive Interaction

by the Congruity and Belief Congruence

Models . . . . . . .

Conclusion Relativ to the Reliability
of Prediction of Cognitive Interaction by
the Congruity and Belief Congruence Models
Conclusions Relative to the Theoretical
Issues Underlying the Cognitive Inter-

action Processes and the Operational
Dynamics of the Congruity and Belief
Congruence Models. . . . . . . .‘. .
Implications for Future Research . .

VERBAL INSTRUCTION GIVEN VIA CCTV. . . . . . . .

Page

88

88

90

91
93

97
99
99
107
114

114

LIST OF TABLES
TABLE Page

1. The Relationship between the Number of
Subjects Administered the ApprOpriate Test
Materials and the Number of Subjects
Eliminated from the Final Analysis 15

2° Summary of the Subjects (8) and Character-
izations (C) Used to Structure the Thirty—
Six Verbal Assertions (CS) Employed in the
Investigation, and the Mean Evaluative
Meaning Assigned to each by Subjects 17

3. Significance Test for Difference in Mean
Evaluation Scores Between Equivalent Test
Groups for Subjects (S), Assertions (CS) and
Characterizations (C) on all 36 Combined
Configurations 34-40

4a-f. Mean Error of Prediction and Significance
Tests of Difference Between Mean Obtained
Evaluation Scores for Combined Concept and
Mean Predicted by the Congruity Model and
the Belief Congruence Model When Subject (S)
is Constant and Characterization (C) is Varied 43-48

Sa-f. Mean Error of Prediction and Significance
Tests of Difference Between Obtained Evalu-
ation Scores for Combined Concept and Mean
Predicted by the Congruity Model and the
Belief Congruence Model Whem.Subject (S) is
Varied and Characterization (C) is Constant 49-54

6. The Total Absolute Error of Prediction, the
Average Absolute Error of Prediction per
Assertion, and the Percent and Number of
Assertions Predicted Accurately by the Con-
gruity and Belief Congruence Models across all
Test Groups and Assertions 56

7. Total Mean Error of Prediction, the Mean Error
of Prediction Per Assertion, and the Number
and Percent of Assertions Predicted Accurately
by the Congruity and Belief Congruence Models
When the Subject (8) and the Characterization

vi

TABLE

lOa-f.

lla-f.

LIST OF TABLES (Cont.)
Page

(C) are either Held Constant or Varied
in the Combined CS Configuration 60

The Frequency and Percentage Distribution

of the Failures Observed in the Application

of the Congruity and Belief Congruence Models

to Predict Cognitive Interaction as a Function
of the Subject (S) and Characterization (C)
Relative to the Combined CS Configurations 64

The Percent of Respondents Who Indicated a
Stronger Feeling Towards the Characterized
Subject (CS) than They Felt Towards Other

Peeple Characterized in the Same Way (Over-
assimilation) 66

Mean Difference Between the Evaluation of

the Combined Configuration and each of the

Two Components When the Subject (S) is Held
Constant and the Characterization (C) is

Varied 68-73

Mean Difference Between the Evaluation of

the Combined Configuration and each of the

Two Components When the Subject (S) is Varied

and the Characterization (C) is Held Constant 74-79

vii

FIGURE

LIST OF FIGURES

Systematic Assertion Structure Using
Six Subject Components and Six
Characterization Components . . . . . . . .

Systematic Assignment of Assertions (CS)
and their CorreSponding S and C Components
to the Twelve Test Groups Employed in the
Research Design . . . . . . . . . . . . . .

An Illustration of the General Application
of the Three Semantic Differential Scales
as a Measurement of Evaluative Meaning of
S, C, and CS as Concepts. . . . . . . . . .

An Illustration of the General Application
of the Relative Importance Rating as a
Function of the Belief Congruence Principle

A Schematic Illustration of Overassimilation

not Accounted for by the Belief Congruence
MOdel O O O O 0 O O O O I O O O O O I O O 0

Recommended Revision of the Relative
Importance Rating Form to obtain the
direction of the overassimilation in
relationship to the characterization
component in the combined CS configuration.

viii

Page

19

21

22

24-25

83

85-86

CHAPTER I

INTRODUCTION

The student of communication theory is especially
interested in predictive models that increase his ability to
explain behavioral changes resulting from a certain communi-
cative act. Central to communication theory are the pro-
cesses by which one assigns connotative meaning to stimuli
that are encountered in his environment.

One rather narrow aspect of this process deals directly
with the assignment of evaluative meaning to a single symbol
or combination of symbols. For example, if one perceives the
word NEGRO typed on a sheet of paper, he will assign certain
evaluative meanings to it° Should one perceive the words
DEMONSTRATING FOR CIVIL RIGHTS apart from the word NEGRO,

he would probably assign evaluative meaning to the former

—___

 

. -v~.......-.._-—-__..-

independent of the latter. A central problem to communication

theory, then, is how the association of such words as A NEGRO,

~~ *— fir, “- 4.. _

with such phrases as DEMONSTRATING FOR CIVIL RIGHTS, will alter
the meaning of the two different sets of words as perceived.
independently. Furthermore, if’a cognitive interaction effect
occurs, is it predictable from knowledge of the evaluative
meaning that one has assigned to the two different components
of the assertion, namely, A NEGRO and DEMONSTRATING FOR CIVIL

RIGHTS, or is the resultant meaning a function of some gestalt?

The ability to predict such cognitive interactions resulting

from the combination of verbal stimuli is directly related to
the study of certain communicative acts, particularly from the
VieWpoint of attitude change and measurement.

The function of assigning connotative meaning to verbal
stimuli is a cognitive process that can be Operationally de-
fined in terms of the semantic differential measurement pro-
cedure deveIOped by Osgood, Suci and Tannenbaum (9). Cog-
nitive interaction can thus be defined, "...as the process
by means of which a single evaluative meaning emerges as a
result of combining two stimuli, each having their separate
meaning (12)."

The ability to predict cognitive interaction is, in
one way or another, the goal of the so-called "balance
theories." Included in these theories are Festinger's (3)
work on cognitive dissonance, Heider's (5) interpersonal
relationship studies,and Newcomb's (7) strain towards
symmetry. Fishbein and Hunter (16) investigated the relative
virtues of the summation and balance model in attitude organi-
zation and change. However, the congruity and the belief
congruence models attempt to make specific quantitative pre-
dictions regarding the outcome of cognitive interaction when

verbal stimuli are associated.

The Congruitnyrinciple_and
Cognitive Interaction”

 

 

The congruity principlelis an additive model that
attempts to predict the outcome of cognitive interaction
solely from knowledge of the direction and degree of polar-
ization of the two stimuli considered in isolation. The re-
duction of incongruity between the source and object (in the
case of source-object assertions) is said to be achieved by
a compromise in which the source and object both change to-
ward or away from one another in inverse proportion to their
respective degrees of polarity.2 It follows, then, that the
congruity principle is essentially a compromise model, the
only exception being when an extremely polarized stimulus is
positively associated with a neutral stimulus. In this situ-

ation the meaning of the combined stimuli is assimilated to

 

1Although the discussion presented here relative to
the congruity and belief congruence principles has been
slightly altered to fit the present situation, for all
practical purposes it should be considered as verbatim quo-
tations from the Rokeach and Rothman (12) study.

2The formula Osgood, Suci and Tannenbaum use to pre-
dict the outcome of cognitive interaction is:

dm = /da/ (da) + - _/dn/ (dn) where
/da/ + /dn/ /da/ + /dfi/'

/d/ is deviation or polarization from neutrality on the scales
regardless of sign, d is deviation from neutrality with respect
to sign. (3, p. 278).

 

 

that of the extremely polarized stimulus. In addition, the
congruity model formally posits an assertion constant in the
case of source-objEct assertions, in order to correct for

the greater force assumed to be acting on the object. In

all cases,.the congruity principle assumes complete credulity,
and uses quantitative corrections for relevance-nonrelevance,
derogation-nonderogation and adjective-noun combinations

to modify the predictive outcome of the model.

The Belief Congruence Principle and
Cognitive Interaction

 

The belief congruence principle asserts that the cog-
nitive comparison process cannot be activated until the two
stimuli are linked together to form a unique gestalt (l, 6).
Thus,the outcome of cognitive interaction cannot be accurately
predicted solely_from a knowledge of the direction and inten—
sity of the two stimuli considered separately. Rokeach points
out that all types of assertions associating two stimuli have
something in common:

...they are unique configurations cognitively
representing a characterized subject (CS)--a per-
son, thing, or idea Characterized or qualified in
some unique way. The unique configuration consists
of two components: a subject (S), capable of being
characterized in many ways, and a characterization
(C), capable of being applied to many subjects.
(12, p. 129)

Rokeach goes on to state the principle of belief congruence
in this way:

Whenever two stimuli are brought into association
with one another through an assertion they form

a unique configuration activating two kinds of
comparison processes: the stimuli will first be

compared for mutual relevance, and if they are

perceived to be at least partially relevant for

one another, they will then be compared for

relative importance. (12, p. 129)

The belief congruence principle thus asserts that in-
congruity arises not from the psychological disparity between
C and S but from the disparity between C and CS, or between
S and CS, or both. In contrast to the congruity principle,
the belief congruence model allows for various degrees of
compromise, and for assimilation and overassimilation, de-
pending on the relative importance of C versus 8 and of C
versus CS, in the context CS, regardless of the degree of
polarization of C and S when considered separately. A
detailed statement of the comparisons suggested above and
the relevant formuli are given in Chapter 2.

Obviously, the development of the two different prin-
ciples leads to instances of contradictory predictions, and
to a_concern as to which model is a more accurate predictor
of cognitive interaction.

Em irical Test of the Congruity Versus
the BeIief Congruence Principles as I=

Predictors of Cognitive Interaction

 

 

The theoretical rationale and the empirical evidence
supporting the congruity principle and its application to the
prediction of cognitive interaction are presented by Osgood,
Suci, and Tannenbaum (8, 9). Their work deals with the com-
bining of verbal stimuli commonly classified as nouns and

adjectives. Specifically, their study investigated the ability

of the congruity principle to predict the interaction effects
of combining eight nouns: NURSE, SCIENTIST, THUG, PROSTITUTE,
HUSBAND, COMEDIAN, IMP, and SECRETARY with eight adjectives:
ARTISTIC, HAIRY, LISTLESS, AVERAGE, SINCERE, sny, TREACHEROUS,
and BREEZY.

These eight nouns and adjectives were combined in all
possible ways (64 possible combinations) and submitted to
eight different test groups. Respondents rated all 16 con-
cepts on three semantic differential scales, selected for
their loadings on the Evaluative, Potency, and Activity factors.
After rating each concept separately, the subject rated one of
the eight nouns in combination with each of the adjectives;
that is, each of the eight test groups rated only one series
of possible noun-adjective combinations for their combined
interaction meaning. Ratings were obtained on the same set
of scales employed for the single noun and adjective list.

The predictive accuracy of the congruity formula was
determined relative to various criteria, but perhaps the most
rigorous one used was the mean error of prediction across the
64 combinations on the Evaluation, Potency, and Activity
factors. The average units of error reported for the Evalu—
ative factor was .92, or nearly one scale unit per noun-
adjective combination. This error of prediction was well
beyond the reliability estimate assumed on an a priori basis,
and was approximately three times greater than the error of

prediction for the Potency and Activity factors. In their

summary, Osgood,.Suci, and Tannenbaum state:

The results of this study show that the semantic
effects of word combinations are neither haphazard
nor unique. In terms of the average meaning of
the word combinations, semantic effects follow
expectations from a congruity principle quite
closely. Analysis of the data from individual
subjects, however, reveals consistent errors

in prediction with the congruity formula: parti-
cularly on the evaluation scale, the measured
meanings of combinations regularly deviate by
being more unfavorable (bad, weak, and passive)
than predicted. It was also shown that the con-
gruity formula predicted less and less well as
the angular displacement of word components in
the semantic space increases. In other words,
the less comparable two signs that are put in
combination, in terms of sharing of characteristic
attributes, the less congruity they display; and
the failure of congruity under these conditions
typically appears as a dominance of the unfavor-
ably evaluated component. (9, p. 283)

In 1965, Rokeach and Rothman (12) reported the results
of an investigation designed to test the theoretical impli-
cations of the belief congruence model and to compare the
relative accuracy of the belief congruence and congruity
models as predictors of cognitive interaction resulting from
various word combinations. For their study, Rokeach and
Rothman selected 12 different word combinations representing
three of the four types of assertions defined by Osgood pp 31.
(9, p. 202) Specifically, the assertions included simple
linguistic qualification, statements of classification, and
source-object assertions as follows: 'A WHITE PERSON who is
a COMMUNIST'; 'A WHITE PERSON who is an ATHEIST'; 'A NEGRO
who believes in GOD'; 'A NEGRO who is an ANTICOMMUNIST'; 'My ’

MOTHER is INSINCERE'; 'UNIVERSITY PROFESSOR favors EXTRAMARITAL

SEXUAL RELATIONS'; 'CLARK CABLE was in favor of FIDEL CASTRO';
'DISHONEST ATHLETE'; 'NIKITA KRUSHCHEV advocates CLOSE FAMILY
TIES'; 'UNFAITHFUL ROMANCE';"RUSSIA extends FREEDOM OF THE
PRESS'; 'A PROSTITUTE who looks like GRACE KELLY' .

The 12 assertions listed above generated 22 different
concepts (the word or series of words in caps make up the
two component parts of each assertion; i.e., A WHITE PERSON
and COMMUNIST in the case of Assertion 1) that were rated
across the same three item semantic differential scale of the
Evaluative dimension used by Osgood and Ferguson and reported
in Osgood, Suci, and Tannenbaum (9). In addition, a second
supposedly comparable sample of subjects rated each assertion
for relative importance of the component parts. Although
both sets of subjects were university students enrolled in
introductory psychology courses at Michigan State University,
this methological procedure creates design problems that are
dealt with in the present investigation and that are dis-
cussed in greater detail in this chapter.

The values predicted by the two models were then com-
pared to the actual ratings assigned to each of the 12

assertions by the subjects. Differences between the obtained

 

mean evaluation scores for each assertion and the predicted

 

mean evaluation scores for each of the models were treated
as "error of prediction."
Tolerance for acceptable error of prediction was

assumed to be no greater than "chance error" and was

operationally defined as a function of a t-test for matched
or correlated groups (4). That is, if the difference between

the predicted evaluative meaning scores and the obtained

 

 

evaluative meaning scores was greater than zero at the .05
level, it was assumed that the model was a poor predictor of
the interaction effect between the two component parts of the
assertions. Although Rokeach and Rothman (12) considered
several different criteria to determine accuracy of prediction,
they concluded that a criterion based on the amount of abso-

lute error between the obtained scores and the predicted

 

 

scores as computed for each model was the most rigorous test
of accuracy. Using this test of accuracy, then, it was
assumed that when the difference between the obtained mean
evaluation scores and the predicted mean evaluation scores
for the two models was large enough to be evaluated as sig-
nificant at the .05 level, the two models differed in their
predictive accuracy.

Based on this criterion, the results revealed that the
belief congruence principle reduced the average error of pre—
diction over the 12 assertions by about two-thirds when com-
pared to the congruity principle. Rokeach and Rothman
reported an average error of 1.07 units per assertion for
the congruity model, but only .34 units of error for the
belief congruence model. In addition, the belief congruence
model correctly predicted nine of the 12 assertions, while

the congruity principle correctly predicted only four of the

10

12 assertions. Thus, the belief congruence principle not only
reduced the total error of prediction, it predicted over twice
as many assertions correctly.

The Type of Assertion and
Cognitive InteractiOn

 

 

As indicated above, Osgood 23 EA (9) have specified
four types of assertions in which one may perceive a re-
lationship between two stimuli, each with their own eval-
uative meaning. It is further assumed that the assertion
associates the two stimuli in such a way that a new evalua-
tive meaning is assigned to them. The assertions might be
in the form of verbal statements written or printed on
sheets of paper or of nonverbal symbols associated in some
manner; e.g., a photograph of George Wallace, former gover-
nor of Alabama, shaking hands with Cassius Clay, former
heavyweight boxing champion. No matter what their form,
Osgood pp 31 suggest that the assertions can be categorized
into four general types: (1) simple linguistic qualification,
(2) simple perceptual contiguity (as suggested above with
Wallace shaking hands with Clay), (3) statements of classif-
ication, and (4) source-object assertions. The studies
reviewed above, as well as the present study, deal only with
word combinations utilizing assertions that associate two
independent sets of printed verbal stimuli of types 1, 3, and 4.

Rokeach and Rothman (12) viewed the assertion as composed

of two component parts, knoWn as the subject (S) and the

11

characterization (C); thereforeIthe assertion was defined as

 

the subject characterized (CS) in some unique manner. This

 

general formulation is used in the present study.

However, one form of assertion was not used at all in
the Osgood and Ferguson Study, and was used only moderately
by Rokeach and Rothman. This is the form in which the c0pula
or action itself is capable of being assigned evaluative
meaning apart from the assertion's associative or disassoci-
ative function. For instance, the assertion, A NEGRO DEMON-
STRATING FOR CIVIL RIGHTS illustrates how the COpula or

assertive action (DEMONSTRATING) has a meaning apart from

 

the assertion A NEGRO IS FOR CIVIL RIGHTS. In a general sense,
the c0pula 'DEMONSTRATING' places the subject in a specific
activity or situation that serves to characterize him some—
what differently than most of the assertions employed by
Rokeach and Rothman. This distinction is central to the

present study, since only assertions of the activity or

 

situational form were employed in the research.

 

The Objectives of the Investigation
and Relevant Hypotheses

 

 

The present investigation sought to achieve two broad
objectives. In addition, three hypotheses based on the the-
oretical rationale and empirical results reported in the 1965
study by Rokeach and Rothman (12) were tested.

Objective 1

 

The first objective was to test the relative accuracy

of the congruity and belief congruence models as predictors

12

of cognitive interaction resulting from word combinations of

a Specific type not generally tested before; i.e., combinations
in which the copula or action itself is capable of being
assigned evaluative meaning. It was assumed that the results
of this test would either support or refute the earlier
reported (Rokeach and Rothman, 12) general superiority of

the belief congruence model over the congruity model.
Specifically, the following two hypotheses were tested:
Hypothesis No. 1:

When assertions that express a situation or activity
are tested under identical conditions, the belief congruence
model will manifest significantly less mean error of pre-
diction per word combination than will the congruity model.
Hypothesis No. 2:

When assertions that express a situation or activity
are tested under identical conditions, the belief congruence
model will accurately predict a significantly greater number

of word combinations than will the congruity model.

Objective 2

 

A second objective of this investigation was to test
the relative reliability of the congruity and belief con-
gruence models as predictors of cognitive interaction result-
ing from word combinations, when the same combinations are
tested across equivalent groups of subjects. Obviously, the
ability of a model to predict accurately across different
groups is a useful test of reliability. Specifically, the

following hypothesis was tested:

13

Hypothesis No. 3:

When assertions that express a situation or activity
are tested under identical conditions, the belief congruence
principle will manifest Significantly greater reliability of
prediction than the congruity principle when the same word
combinations are tested across different but equivalent
groups.

CHAPTER 2

METHOD

Subjects
Subjects were 849 high school age youth attending the

State 4-H Club Week program at Michigan State University in
June, 1967. Most subjects were junior and senior level high
school students selected to attend the event from about every
Michigan county. The ratio of girls to boys was about 2 to 1.
Prior to completing the test materials, the 849 sub-
jects were randomly assigned to 12 different test groups.
There was considerable attrition of subjects. Of the 849
who responded to the test materials, 230 were eliminated be-
cause they obviously did not understand the instructions.
Most of the difficulty stemmed from the relative importance
rating required by the belief congruence model: Either sub-
jects left one or more of the questions unanswered, the two
required percentage estimates did not add up to 100, or sub-
jects answered question lb when they should have answered
question 2c, or vice versa. Table 1 summarizes the actual
breakdown on subject assignment to test groups and the num-
ber of usable tests obtained from subjects in each group.
While the loss of 27.1 percent of the subjects may
seem excessive, it does not differ greatly from the nearly
20 percent loss reported by Rokeach and Rothman (12, p. 129).

Unlike the present study,.Rokeach and Rothman were working

14

15

with college level subjects who might be expected to deal more
effectively with the rather difficult relative importance
rating. Also Rokeach and Rothman administered their test
materials in a classroom situation, while in the present in-
vestigation, the materials were administered using closed
circuit television and room monitors. These considerations
further explain the relatively large number of imprOperly
marked questionnaires.

Table 1. The Relationship between the Number of Subjects

Administered the Appropriate Test Materials and
the Number of Subjects Eliminated from the Final

 

 

 

Analysis
Test Subjects Subjects Percent of
Group Assigned Completing Subjects
Number Test Forms Usable Forms Eliminated
l 68 55 19.2
2 69 43 37.7
3 66 - 50 24.3
4 70 49 30.0
5 73 47 ' 35.6
6 69 54 21.8
7 78 - 55 29.5
8 72 56 22.2
9 69 55 20.3
10 71 55 22.6
11 71 52 26.8
12 __13 __48 24.3
Total 849 619 Ave. 27.1

The Assertions and Their Components

 

Deve10pment of the assertions used in this study was

guided by three major criteria: (1) the component parts--

16

i.e.,the subject (S) and the characterization (C)--had to be
selected so that each component part could logically be com-
bined or associated with all other component parts; (2) each
assertion had to be designed so that the subject (S) was

always combined with a characterization (C) which Specified

an action or situation, and (3) the subject (S) and charac-

 

terization (C) components had to generate a reasonably wide
range of connotative meanings. As was the case in prior
word combination studies, the word combinations (assertions)
used in this investigation were assumed to have complete
credulity (8, 9, 12).

The six subjects (S) and Six characterizations (C)
included in this investigation are summarized in Table 2.

A pilot study revealed that the 36 different assertions (CS'S)
structured from the set of six subjects (S) and six charac—
terizations (C) satisfied the three criteria given above
reasonably well._

The 36 assertions (CSFS) were formed by systematically
associating each of the Six subjects (S) with each of the
characterizations (C). AS Table 2 shows, it was possible to
associate a highly positive S with a relatively negative C
[MY FATHER (+2.56) with USING A HABIT FORMING DRUG (-2.57)]
or a relatively negative S with a relatively positive C
[A COMMUNIST (-1.93) with PUNISHING A CHILD (+1.73)]. The
remaining CS'S form associations falling in between these

two extremes.

17

Table 2. Summary of the Subjects (S) and Characterizations
(C) Used to Structure the Thirty-Six Verbal
Assertions (CS) Employed in the Investigation,
and the Mean Evaluative Meaning Assigned to each
by Subjects

Assertion Mean Evaluative Number of

Components Meaning Assigned~ Subjects
by Subjects Rating Rating
Component* Component

 

SUBJECTS (S)

 

MY FATHER 6.53 376
A PROTESTANT 5.59 368
A CATHOLIC 5.49 375
A WHITE PERSON .5.26 364
A NEGRO 5.20 371
A COMMUNIST 2.61 370

CHARACTERIZATIONS (C)

 

PUNISHING A CHILD 5.73 353
ATTENDING A FUNERAL 5.26 350
FIGHTING THE VIET CONG 5.03 354
PROMOTING BIRTH CONTROL 4.61 353
DEMONSTRATING FOR CIVIL RIGHTS 3.87 353
USING A HABIT FORMING DRUG 1.43 346

*The components were rated on a three item semantic differen-
tial scale, using bipolar adjectives loading highly on the
evaluative factor. Mean scores were calculated on a 1 through
7 basis which is equivalent to a range from -3 to +3. The +
or - scale score can be obtained by subtracting 4.00 from the
scores in the table.

 

The mean evaluation score was obtained by averaging the
individual scores assigned by the respondents who actually
rated the concept in the investigation. For example, 376 of
the 849 respondents who participated in the study rated the

concept MY FATHER.-

 

18

Although the evaluative scores assigned to the six
subjects (8) ranged from -1.39 to +2.53, four of the six
clustered between +1.20 to +1.59, which does not represent
as much variation as would be optimally desirable. The
range of evaluative meaning scores assigned to the charac-
terizations (C) was more satisfactory, except that the most
positive characterization was at the +1.73 level, represent-
ing an upper positive limit that was less extreme than op-

timally desirable.

Research Design

 

The-36 different assertions used in this study are
summarized in Figure 1. Each assertion is structured for
each cell by combining the corresponding S and C. For ex-
ample, Assertion C151 is MY FATHER PROMOTING BIRTH CONTROL,
and Assertion C686 is A CATHOLIC USING A HABIT FORMING DRUG.
All other cells are read in a like manner.

In Figure l the assertions across rows have the S com-
ponent held constant, while the assertions across columns
have the C component held constant. This particular design
made it possible to assign the six assertions in each of the
six rows in Figure l to six different, but equivalent test
groups, and to assign each of the Six assertions in each of
the six columns in Figure 1 to six different, but equivalent
test groups. Thus, a total of 12 different, but equivalent

test groups were employed in the study design. The-design

is illustrated more clearly in Figure 2.

19

Characterization Component**

C2

C281

C282

C253

C254

C255

C286

Subject
Component*
C1
81 C151
S2 ClS2
S3 C153
S4 C134
85 C185
S6 C156
* s1 - My Father
$2 - A White Person
S3 - A Negro

S4 - A Communist
S - A Protestant

S - A Catholic

3 1 4
C352 C4
C3S3 C4
C3S4 C4
C335 4
c336 c

** C1
C2
C3
C4
C5
C6

5 6
C531 C651
C532 C652
c533 c633
C554 C654
C555 C655
C556 C656

Promoting Birth
Control

Fighting the Viet
Cong
Demonstrating for
Civil Rights
Punishing a Child

Attending a Funeral

Using a Habit Form-
ing Drug

Figure 1. Systematic Assertion Structure Using
Six Subject Components and Six
Characterization Components.

From Figure 2, the S and C components and the corres-

ponding CS'S that were presented to each test group can be

determined. For example, subjects in Test Group 1 rated S1

(MY FATHER) and Cl

through C

6

(PROMOTING BIRTH CONTROL,

FIGHTING VIET CONG, DEMONSTRATING FOR CIVIL RIGHTS, PUNISHING

A CHILD, ATTENDING A FUNERAL, USING A HABIT FORMING DRUG) as

independent concepts.

They then rated the CS configuration

20

when S1 was associated with each characterization C through

1

C6' The latter measurement was defined Operationally as the
obtained evaluative meaning score, and served as the basis

for determining the accuracy of the pgedicted scores. The

 

predictive scores were obtained by the application of the S
and C evaluation scores secured independently of each other
and independently of the CS rating, and applied to the con-
gruity and belief congruence formuli. Each cell in Figure.

2 can be interpreted in the same manner. It should be noted,
however, that Test Groups 1 through 6 have the S constant
and the C varied, while Test Groups 7 through 12 have the S
varied and the C constant.

It should also be noted that in this design each
assertion and its corresponding components are rated by two
independent but equivalent test groups. Thus, one can deter-
mine the relative reliability of the two models in predicting
cognitive interaction across independently selected groups
as well as within a single group. This design characteristic

provides the basis for dealing with Objective 2 of the study.

The Measurement of Evaluative Meaning_

 

As stated in Chapter 1, one purpose of the present
research was to attempt to replicate the results obtained in
two earlier studies on word combinations and cognitive inter-
action. In both the Osgood (9) and the Rokeach and Rothman

(12) studies, semantic differential scales were used to obtain

21

Test group Components and Corresponding
Assertions Rated by the Test Group*

1 C151 C281 C381 C451 C581 C681
2 C182 C282 C382 C482 C582 C682
3 C183 C283 C333 C483 C583 683
4 C154 C254 C384 C484 C584 654
5 C185 C285 C385 C485 C585 C655
6 C186 C286 C356 C4S6- C586 C686
7 C181 C182 C183 C184 C185 C186
8 C281 C282 C283 C284 C285 02S6
9 Césl C382 C353 C384 C385 C386

10 C481 C482 C483 C454 C435 C486

11 C551 C582 C583 C584 C585 C586
12 C681 C682 C653 C654 C685 C636

* S1 - My Father *Cl - Promoting Birth Control
82 - A White Person C2 - Fighting the Viet Cong
S3 - A Negro C3 - Demonstrating for Civil
Rights

S4 - A Communist C4 - Punishing a Child

S5 - A Protestant C5 - Attending a Funeral

$6 - A Catholic C6 - Using a Habit Forming

Drug

Figure 2. Systematic Assignment of Assertions (CS)
and their Corresponding S and C Components
to the Twelve Test Groups Employed in the

Research Design.

22

measurements of the evaluative meanings assigned to the con-
cepts studied. These researchers used three scales loading
highly on the evaluative factor: good-bad, deplorable-
admirable, and worthless-valuable, and employed the same
procedural techniques used in the present study.

In the present investigation the concepts rated were
the six different C's, the six different 8'5, and their cor-
responding 36 different CS'S. The general form of the semantic
differential scale used is given in Figure 3. The complete
test booklet is found in Appendix A.

MY FATHER (S)

good : : : : : bad

 

 

deplorable : : 3. .: : admirable

 

 

valuable : ~: : : worthless

 

 

PROMOTING BIRTH CONTROL (C)

 

 

 

 

good : : : : : bad
deplorable : : : : : .admirable
valuable : : : : : worthless

L

MY FATHER PROMOTING BIRTH CONTROL (CS)

 

 

 

bad : : : : : good
deplorable : : : : : admirable
worthless .: : : : : valuable

 

Figure 3. An Illustration of the General Application
of the Three Semantic Differential Scales
as a Measurement of Evaluative Meaning of
S, C, and CS as Concepts.

23

The Measurement of Relative Importance
and the Computation of Cognitivek Inter-
action by ther BeIief Congruence Principle

 

 

 

To fulfill the requirements of the belief congruence

principle, it was necessary to measure the relative importance

 

the respondent assigned to the subject (S) and characterization
(C), both in relationship to one another, and also in the con-
text of CS versus C. The procedure fer obtaining these
reSponses was developed by Rokeach and Rothman (12) and was
employed in the present investigation with two minor alter-
ations that increased the subject's ability to understand

how to respond to the instrument.

Figure 4 illustrates the general form of this measure-
ment. As can be noted from the complete test booklet in
Appendix A, the Relative Importance rating was labeled as
Part II and was preceded by appropriate verbal instructions
and by the CS rating pages. The form was so structured that
the respondent was forced to again indicate how he felt to-
wards each_assertion (CS) on ail to 7 scale. Rokeach and
Rothman (12) reported that the addition of the evaluative
rating of CS immediately prior to the required comparison
ratings increased Significantly the ability of the subjects
to understand the instructions on how to make the comparisons.

Question 2 sets the framework which enables the res-
pondents to make the C yg S comparison rating. If the res-
pondent felt that C did not completely influence his feelings

about the complete CS configuration, he simply moved to Question

24

How do you feel about: MY FATHER PROMOTING BIRTH CONTROL

STRONGLY : : : : : : :STRONGLY
DISAPPROVE I 2 3 _47 5 6 7 APPROVE

 

 

In rating MY FATHER PROMOTING BIRTH CONTROL the way you'
did, how important, that is, how much weight did the

words PROMOTING BIRTH CONTROL have in determining your
rating of the statement, MY FATHER PROMOTING BIRTH CONTROL?
Did you feel about the combination MY FATHER PROMOTING
BIRTH CONTROL the same way you felt about the words PRO-
MOTING BIRTH CONTROL or the same way you felt about MY ’
FATHER? Estimate how much weight PROMOTING BIRTH CONTROL
and MY FATHER had in determining the way you actually
rated the combination MY FATHER PROMOTING BIRTH CONTROL.

a. My feelings about PROMOTING BIRTH CONTROL completely
(100%) determined the way I rated MY FATHER PROMOTING
BIRTH CONTROL. (Check the one that applies).

YES Now go to question b. 23 not answer question c.
NO Now go to question c. 23 not answer question b.

b. In fact, my feelings about MY FATHER PROMOTING BIRTH
CONTROL are even more extreme than my feelings about
other peOple PROMOTING BIRTH CONTROL.

Answer those that apply to you:

No, my feelings about MY FATHER PROMOTING BIRTH
CONTROL and PROMOTING BIRTH CONTROL are about
equal strength.

Yes, my feelings about MY FATHER PROMOTING BIRTH
CONTROL are even stronger than my feelings about
other peeple PROMOTING BIRTH CONTROL.

How much stronger? Check the one that would be
closest to your best guess:
Slightly stronger (1% stronger)?

Quite a bit stronger (50% stronger)?
Much, much stronger (100% stronger)?

c. My feelings about PROMOTING BIRTH CONTROL did not
completely determine my rating of the combination
MY FATHER PROMOTING BIRTH CONTROL. Indicate on
the scales below how much you think each part of

25

the statement influenced you:

Assume that each scale equals 100%. Check the point

on each scale which you think best describes the

amount of influence that part of the statement had

on your rating of the statement on the whole. Remember
that the percentage checked on each scale when added
together must be equal to 109 percent.

MY FATHER: : : : : : : : : : : :
T‘IU‘TG‘IUTU‘S‘G‘G‘G—W—FGTUW
PROMOTING BIRTH CONTROL

:Tg‘fig76:'3'6:_4'6:"§'6:"66:"7'6:‘8—6.—§'6.I'66:

Please Check: Add the two percentages together that you have
checked. Do they equal 100%? If not, readjust
your rating so that they sum to 100%.

Figure 4. An Illustration of the General Application
of the Relative Importance Rating as a
Function of the Belief Congruence Principle.

26

c where he indicated on the two scales what percent influence
the C and S contributed to his overall feeling about CS. The
addition of the equal interval scale used in this study repre-
sents a change from the Rokeach and Rothman format. They
requested the respondent to indicate the percent he assigned
to the C and S in the combined CS configuration. It was felt
that the scale used in this study would facilitate understand—
ing on the part of respondents. The second modification of
Rokeach and Rothman's procedures consisted of the addition

of the instructions in Question 2a following the response
categories "Yes" and "No." The respondent was instructed

app to answer Question 3 if he had responded "Yes" to Question
2a, or app to answer Question E if he had responded "No" to
Question 2a. This alteration was suggested by the results

of a pilot study using high school youth, and appeared to
increase the ability of the youth to respond appropriately

to the forms.

Next consider the obtained ratings of relative import-
ance in relationship to the formuli used to compute the inter-
action evaluative meaning assigned to the CS from the indi—
vidual ratings of C and S.

First, consider formula one:

dCS = (w) dc + (l-w) ds [1]
where (w) and (l-w) refer to the perceived importance of dc

and dS in the context of CS. Operationally, then, if a

27

respondent checked Question 2a as "No“ and moved to 2c and
indicated a 40 percent importance for the S and a 60 percent
importance for the C, dc was weighted by the value of .6 and
dS = (l-.6) or .4. Thus, by inserting the evaluative mean-
ing value the respondent assigned to the C and the S, it was
possible to compute the dcs (CS) and obtain the predicted
value based on the belief congruence principle.

Second, consider formula two:

dcs = dc + (v) dc [2]

where (v) is the coefficient assigned according to how much
more negatively or positively the respondent indicated he
felt toward the CS than he did toward the C alone. This-
measurement was used when the respondent checked "No" to
Question 2a and "Yes" to Question 2b, with the coefficient
being either .1, .5, or 1.0. If the respondent indicated
"No" on Question 2b, (v) becomes 0 and dCS = dc' Operation-
ally, then, the evaluative rating for dC (C) was inserted in
the formulaeand the weighted value of dc’ obtained from
Question 2b, was added to account for the roverassimilation
towards the C component of the CS configuration. It must be
noted, however, that the (overassimilation is restricted to
the extremes of the measurement scale being employed. In
the present study, the scale was based onIa —3 to +3 range;
thus,the weighted evaluative meaning of the dcs cannot exceed

-3 or +3 due to limitations imposed by the measurement

procedures.

28

Predicted CS scores were computed on the basis of
either formula 1 or 2 above for each subject for each assert-
ion rated. These individual scores were then averaged over
all subjects rating each assertion (CS).

Computing Cognitiye Interaction
by the Congruity Principle

 

Operationally, the congruity principle requires only
the measurement of the evaluative meaning the subject assigns
to the C and S in the CS configuration. These measurements
are obtained by the semantic differential method discussed
above.

Assuming that no corrective constants are employed,
the evaluative meaning ratings are inserted into the formula
below:

d = /dc/

d
cs —7— (d ) + j/ 81 (d )
/dc/ + 755 C 7dc/ + /ds/ S

 

where /d/ is deviation or polarization from neutrality on the
scale regardless of sign, and (d) is deviation from neutrality
with respect to sign (1, p. 287). Assuming that across the
three scales the respondent had assigned the evaluative mean—
ing ratings of -l.5 to the dc and +2.3 to the (13 then:

d = 1.5

 

 

CS ‘I1S + 2.3 I'1°5) T “1 6': 243+ (+2.3)
Therefore,dcs = .86 as a function of cognitive interaction

between the C and S components of the assertion based on the

congruity principle.

29

Predicted CS scores were computed for each subject
for each assertion rated. These individual predicted scores
were then averaged over all subjects rating each assertion
(CS).

Given the present research design, it was necessary
to construct 12 different sets of test materials, one for
each of the 12 different test groups. Each set was designed

to enable the subject to rate 13 different concepts on the

 

same three semantic differential scales, with each concept
rated independently of the others insofar as possible. In
addition, each subject was required to make relative import-
ance ratings of the same six assertions.(CS) he had already;
rated on the semantic differential scales.

The problem of keeping each concept rating independent

of the others was handled by placing only one concept at a

 

time on a half Sheet of paper. Thus,each set of test materials
contained 13 half Sheets of paper.

To control for order effects, the half sheets of paper
were rotated in a systematic order. That is, in the case of
subject (S) and characterization (C), about one-seventh of
the respondents received the test materials with the concepts
presented in the same order. In the case of the CS ratings,
about one-Sixth of the respondents received the test sets
with the CS'S in the same order. This ordering effect was
accomplished by numbering the half-Sheets from one to 13 and
then rotating the order in which they were assembled to con-

form to the control specifications.

30

The relative importance rating Sheets were also rotated
on a one-sixth basis, and in the same order used in the case
of the CS'S rated as concepts on the semantic differential
scales.

The usual procedure of alternating the semantic dif-
ferential adjective pairs was carried out, with one set of
adjectives being reversed for each of the 13 sets of scales.
Since the random assignment of adjective pairs both in terms
of polarity and vertical location in the set of three scales
would have made machine processing very difficult, the scales
were not rotated vertically.

Overall, then, the complete test booklet was constructed
as follows: The standard semantic differential instructions
appeared on the first page (blue) of the test booklet. These
instructions were followed by 13 half-sheets of paper on which
the individual concepts appeared along with the three semantic
differential scales. Immediately following the 13 concept
rating forms, a yellow instruction sheet for Part II was

inserted. This sheet was followed by the six relative

 

importance rating sheets. The Test Group number and individual

 

respondent identification number were placed on the cover page.

Administeringjthe Test Materials

 

In the organization of the Club Week Program, the 849
subjects had been randomly placed in 30 different sub-groups

and asSigned separate rooms in Bessey Hall. Each room was

31

equipped with closed circuit television sets.

The investigator recruited room monitors and trained
them to assist with the administration of the test materials
used in the investigation. These room monitors were for the
most part professional Extension 4-H - Youth agents assisting
with the programs.

The test materials were divided into 30 different
packets in which approximately the same number of each of
the 12 different sets were placed. The room monitors were
instructed to hand the test materials out randomly to the
subjects.

Prior to the actual administration of the test, the
investigator made a videotape recording of the instructions
for completing the test forms. This recording enabled the
investigator to administer the test materials to all groups
under the same conditions and to administer test materials
to a relatively large number of subjects in a period of about
one hour.

Room monitors were in position to give additional assist-
ance to individual subjects if they did not fully understand
the video presentation. However, it was the investigator's
observation that little additional assistance was required by
the subjects, and the monitors' main role was to distribute'

the test forms and collect them when the subjects had finished.

32

Preparation of Data forjMachine
Processing

 

The 849 forms were reviewed for completeness to
determine if the respondent had followed the instructions.
From this review a total of 619 forms were found to be com-
plete and were retained for analysis.

Each form was then re-assembled into its base form
so that the data processing card column numbers pre-assigned
to the forms were in consecutive order. Each form was coded
according to the format required for punching the data cards.
The data were then transferred to punch cards and processed

by the Computer Center at Michigan State University.

CHAPTER 3

RESULTS

Test Group Equivalency

 

In the study design, subjects in two independently
composed groups assigned evaluation scores to each assertion
and its component parts as illustrated in Table 3. To some
extent, the analysis presented in this chapter depends upon
the equivalency of the mean evaluation scores for the two
groups. Equivalency is particularly essential in the test
of Hypothesis 3, the hypothesis dealing with relative reli-
ability of the two predictive models.

Table 3 contains the mean evaluation scores assigned
by respondents in the two groups to the subjects (8), charac-
terizations (CL and combined configurations (CS) for all 36
assertions employed in the study. A simple analysis of
variance was computed for each pair of group mean scores to
determine if they were significantly different. The results
of the analysis appear in Table 3. Pairs for which the mean
evaluation scores differ at the .05 level are viewed as
Significantly different, while pairs which do not differ at
the .05 level are considered equivalent.

The design allowed a total of 108 comparisons of mean
scores: 36 comparisons between the Six subject (S) components
used in the study, 36 comparisons between the six character-
ization (C) components used in the study, and 36 comparisons

between the 36 different combined (CS) configurations tested.
33

34

 

 

mz qa.m om.m
mz oo.m mq.m
mz mm 30.0 mm om.©
mz wo.m o©.m
mz mm.m m¢.m
mz mm Hm.© mm om.c
mz Hm.m mo.¢
mz qw.m No.m
mz mm mo.© mm om.©
mz wo.¢ mm.¢
mz mm.q mw.¢
mz on mq.o mm om.c
mz wm.¢ mm.¢
mz Hw.m Hw.m
mz mm 08.0 mm om.o
m z .mme z Icmmm
ucmumcoo ooflum>
cowumnwuouomumno cowumwfiuouomumzo

ooeem> boomeam

ucmumcoo uoomosm

wQHOUW COM HNSHW>M GMT:

A<Mw23h < GZHDZmHH<
A<MMZDm < UZHDZNHH< mmmfi<h NZ
mmmH<m w:

QAHmU < UszmHZDm
QAHmU < UZHmmHZDm MmmH<m NZ
mmmfi<h wz

mHmUHM AH>HU mom UZHH<MHmZOZMQ
mHmUHm AH>HQ.mom UZHH<mHmZOZma mmmH<h w:
mmme<m $2

@200 HMH> mmH ozHHmUHm
UZOU HmH> mzH OzHHmOHm MmmH<m >2
mmm8<m W2

AOMHZOQ meHm UZHHOZOMQ
AOMHZOU mHMHm OZHHOZOMm MmmH<m M2
Mmme<h WE

ucocoesoo
pom cowuuomm<

 

mcohuouswwmcoo pmcaosoo on HHS co ADV mcoeumueuouomumno paw Amov mcquuomm< .Amv muoonosm you
maoouo umoH ucmam>fioom coosuom monoum cowumoam>m cooz SH oocouowmfio Low umoH mocmowmwcwwm

35

 

 

mz ea.m oo.m
mz m~.m ma.m
monv mm mm.m me ew.e
mz mo.m mm.m
mz em.m em.m
monv mm Ne.m me ee.e
mz Aa.m mA.e
mz mm.e ae.e
moav mm mm.m me em.e
mz mo.e wo.m
mz o~.m Rm.m
8v on 36 me see
mz om.e oo.m
mz Am.e No.e
mz mm ne.m me ew.e
mz mm.A ee.e
mz am.a ee.a
mz we oe.o mm om.o
m 2 com: 2 com:

so >m >o em

OHOUW GOH uwSHQ>M 5mm:

A<Mmzmm < UZHQZMHH<
A<mm23h < QZHQZMHH< ZOmMmm MHHmS <
ZOmMmm MHHm3 <

QAHmU < UZHmmHZDm
QAHmU < qummHZDm zommwm MHHmz <
zommmm MHHE». <

mHmUHM AH>HU mom UZHH<MHmZOSmn
mHmUHm AH>HU mom OZHH<mHmZOZMQ zomMMm MHHmS <
zomMmm NHHmS <

Ozoo HMH> mmH wzHHmuHm
Ozoo HMH> mmH UZHHmUHm zommmm MHHmS <
zommmm MHHE3 <

domHzoo mHmHm UZHHOZOMA
AOMHZOU mHMHm UZHHOZGmm zomMmm MHHmz <
ZOmMNA MHHES <

03mm UZHZMOm HHm¢m < wszD
wDMQ GZHZMOm hHm<m < qumD mmmfi<m $2
Mmma<m #2

.HH

.OH

.0

ucocoesoo
pom cowuuomo<

A.o:oov m oeooe

36

 

 

mz ee.m oe.m
mz He.m ao.m
mo.v mm ma.e om ma.m
mz we.m Re.m
mz Ne.m ee.m
moev mm om.m om mk.m
mz Aa.m Ho.e
mz mm.e mA.e
menu mm am.m om mk.m
mo. wo.e om.m
8V RR 86
8V on 86 em m:
mz em.e oA.m
mz om.e oA.m
8V R ER om ohm
mz mm.a me.a
mz oo.a oo.a
mz we wa.m me eo.e
m z cmoz z com:

so >m >o em

ouoom coauooao>m coo:

A<Mmzsm < UZHQZMHH<
A¢Mmznm < UZHQZMHH< Qmumz <
omumz <

QAHmU < UszmHZDm
QAHmU < wszmHZDm Odomz <
omumz <

mHmUHM AH>HU mom UZHH<MHmZQ2MQ
mHmUHM AH>HU mom UZHH<MHmzozmn omumz <
omwmz ¢

@200 HMH> NEH UZHHmuHm
0200 HMH> mma UZHHmuHm omwmz 4
omomz <

aoeezoo memHm ozHeozoem
aomezoo mazes ozHeozoee omomz <
_oeomz <

Damn GZHZMOM HHm<m d oszD
USMC GZHZmOm HHm<m < UZHwD ZOmMmm MHHSB <
ZOmmmm MHHmS <

.mH

.oH

.mH

.SH

.mH

.NH

ucocoesou
poo cowuuomm<

A.oeoov m ofiooe

37

 

 

mz sA.m Aa.a
mz wa.s aa.s
mz Na sa.~ as ao.~
mz wo.a aa.a
mz Na.s aa.s
mz ma am.~ as ao.~
mz Aa.a aa.a
aouv aa.~ ae.a
mz mm AA.N as ao.~
mz wo.s Ne.s
mz om.a aa.s
mz ea aa.~ as ao.~
mz ea.s aa.s
mz ao.s Na.s
mz ma Na.~ as Ne.~
mz aa.a aa.a
mz ea.a Na.A
menu as sa.s on ea.a
m z com: 2 com:

so Im >o Rm

ouoom coflumSHm>m com:

A<MMZDm < QZHQZMHH<
A<mm23m < UZHDZMHH< HmHZDZZOU <
HmHZDZXCU ¢

DAHSU < Uszmstm
QAHmo < OZHmmHZDm HwHZDZZCU <
HmHZDZZCU <

mBmUHM AH>HU mom DZHH<MHmzozmo
mHmUHM AH>HU mom OZHH<MHmZOZMG HmHZDZZOU <
HmHZDSEOU <

UZOU HMH> mmH UZHHmeh
Uzoo HHH> mmH OZHHmuHm HmHZDzzoo 4
HmHZDZZOU <

AOMHZOU mHMHm ozHHozomm
AOMHZOU mHMHm QZHHOEGMm HmHZDZEOU <
HmHZDZSOU <

UDMG UZHZMOm HHm<m d qumD
Damn OZHzmom HHm<m < wszD omwmz 4
Qmumz 4

.mm

.NN

.HN

.ON

.mH

.mH

ucocoesoo
pom cowuuomm<

a.oooov a oahme

38

 

 

mz «H.m m¢.m
mz w~.m m~.m
mz Nm oo.m we mm.m
mz wo.m mw.m
mz nm.m om.m
mz mm am.m as mm.m
mz Hm.m mm.m
mz mH.¢ no.m
mz mm mo.m as mm.m
mz wc.¢ m~.m
mz mH.m H¢.m
mz cm Om.m as mm.m
mz wm.¢ mH.¢
mz o~.¢ ma.m
mz mm so.m no mm.m
mz mm.~ No.H
mz am.a m¢.~
mz we mm.~ as mc.N
m 2 com: 2 com:

x0 >m >0 xm

OHOUW QOHUNDHQ>N GNU:

A4mMZDh 4 UZHQZMHH4
A4mmzsm 4 UZHDZMHH4 HZ4HmMHOMm 4

Bz4HmMHomm 4

QAHmU 4 UszmHZDm
QAHmU 4 UZHmmHZDm Hz4HmMHomm 4
Hz4HmMHomm 4

mHmOHM AH>HU mom UZHH4MHmZOZmQ
mHmem AH>HU mom OZHH4MHmZQEMQ Hz4HmMHOMm 4
HZ4HmMHOMm 4

AOMHZOU mHmHm GZHHOZOMm
AOMHZOO mHmHm UZHHOZOMm Hz4HmMBOMm 4
Hz4HmMHOMm 4

0200 HMH> mmH QZHHmem
ozoo HMH> mmH UZHHmUHm Hz4HmMHOMm 4
HZ4HmMHOMm 4

UDMQ UszmOh HHm4m 4 UszD
03mm OZHZmOh HHm4m 4 oszD HmHZDZSOU 4
HmHZDZZOU 4

.om

.mm

.mm

.oN

.mN

.4N

ucwcomsoo
pom cowuuoom4

a.oeoov a means

39

 

 

mz «H.m NN.m
wz mm.m ms.m
mz mm N¢.m om om.m
mz wo.m mw.m
mz N¢.m mn.m
mz mm mm.m on om.m
mz Hm.m mm.m
mz o~.¢ Hm.m
mz mm Om.m em o~.m
mz wb.¢ mo.m
mz mm.m sm.m
mz om om.m on om.m
mz ww.¢ om.¢
mz om.m oo.¢
m2 mm Ho.m gm om.m
mz mm.H so.H
mz so.H HN.H
mz we ww.m no mm.m
m z cmmz z cmoz

x0 >m . >0 xm

OHOUW 50.." UMDHN>W waz

A4MMZDm 4 UZHQZMHH4
A4dm22m 4 UanzmHH4 UHAomH4U 4

UHaomH4o 4

QAHmU 4 qummHZDm
QAHmU 4 qummHZDm OHAomH4U 4
UHaomH4U 4

meUHM AH>HO mom UZHH4MHmZOZma
mHmuHm AH>HU mom UZHH4MHmZQZMQ UHAomH4U 4
UHAomH4U 4

0200 HmH> mmH ozHHmuHm
ozoo HmH> mmH UZHHmem UHAQEH4U 4
UHAOEH4U 4

AOMHZOU mHMHm quHOZOmm
AOMHZOU mHMHm QZHHOZUMA UHaomH4U 4
UHAomH4U 4

USMC UZHzmom HHm4m 4 OZHmD
Gama UZHZMOM HHm4m 4 OszD Hz4HmMHOMm 4
Hz4HmHHQmm 4

.mm

.qm

.mm

.Nm

.Hm

.om

ucocomEoo
pom cowuuomm4

a.oooov m oaoee

40

 

 

mz mm.a o¢.~
mz Hm.H om.H
mz we n¢.m gm 0~.m
m z cmoz 2 seem

so >m . . >0 xm

muoom cowuoSHo>m cow:

Gama OZHZmom HHm4m 4 UZHmD
00mm UZHZmom HHm4m 4 oszD 0H40§H40 4

8.855 s. .3
ucwco
paw SOHDLMWWW

A.uc00v m anwH

41

As Table 3 indicates, 13 of the 108 group mean comparisons
yielded significant differences. Overall, only two of the
36 combined CS configurations differed Significantly. In
just one instance did the mean'evaluation scores on the
characterization (C) component differ Significantly, but
scores assigned to the subject (S) components varied sig-
nificantly in 10 of the 36 comparisons tested.

The subjects, A WHITE PERSON, and A NEGRO,accounted
for all 10 of the Significant differences obtained across
all subjects tested, while the characterization, FIGHTING
THE VIET CONG, accounted for the only significant difference
obtained between characterizations tested. The subjects,

A NEGRO, and A COMMUNIST, were common to the two assertions
(CS) found to differ significantly between the test groups.

It is not clear why the two test groups varied so
greatly on the subjects, A WHITE PERSON, and A NEGRO, while
demonstrating a high degree of equivalency for the other
subjects and characterizations. On the whole, however, the
two independently drawn test groups rated most of the sub-
jects (S), characterizations (CL and configurations (CS)
quite Similarly. This fact adds considerable strength to

the results reported for Hypotheses l, 2, and 3.

42

Test of the HypotheseS'

 

Objective 1

 

The first objective of this study was to test empiri-
cally the relative accuracy of the congruity and belief con-
gruence models as predictors of cognitive interaction
resulting from word combinations of a specific type not
generally tested before. The results of this empirical
test should either support or refute the general superiority
of the belief congruence model over the congruity model

reported by Rokeach and Rothman (12).

Hypothesis 1:

When assertions that express a situation or activity
are tested under identical conditions, the belief congruence
model will manifest significantly less mean error of pre-
diction per word combination than will the congruity model.

The basic data comparing the two models are contained
in Tables 4a through 4f and Tables 5a through 5f. Tables 4a
through 4f contain data relevant to the Situation in which
the characterization (C) was varied and the subject (8) was
held constant, while Tables Sa-f contain data relative to
the Situation in which the subject (S) was varied and the
characterization (C) was held constant in each of the 36
different word combinations or assertions employed in the
study. In all tables, the data consist of a comparison of

the predicted scores for each model with the obtained scores

resulting from the actual evaluative meaning scores assigned

43

mowv um ucmowmwcwwmns

 

 

mo.~ na.m oocosuwcoo wowaob ”pODUHpmum

Smm N «a be. m¢.H pocwmubo 00mm oszmom HHm4m 4 osz:

«04 ma nm N mm. 00.4 xuwouwcoo "pODOAUOHA MmmH4m NZ
Nq.a N0.m mocosuwcoo mowaob "pouowpoum

«H H ad mm.H mo.m . pocflmubo A4mmzam 4 ozHazmaH4

sag m mm mm. 0m.o muwsuwcoo .npouowpoum mmma4m M2
mm.H Hm.m mocmsuwcoo wowaob "pmuowpoum

«Am N as mN.H as.m ooeamooo oaHmo a oszmHzem

two m mm on. Hm.0 >uwsuwcoo upmuowomum MmmH4m >2
o~.~ wo.¢ mocosuwcoo onHOQ upmuowpoum

ass a Re A Aa.a Ho.a ooesoooo memoHe AH>Ho mom ozHasmemzozmo

Rom 0H mm N 0H.H sw.m mueouwcoo “pouoapoum Mmma4m NZ
H¢.H mm.m mocosuwcoo «champ "wouowpoum

sa~.m. 0m ma.H mw.¢ oocwmuoo 0z00 HMH> mma ozHHmUHm

Ro~.o am A em. -.o aosoeweoo "ooooeooem emmeee >2
aw.H om.¢ mucosuwcoo moHHob upouofipoum

see 0 mo H Hm.H Hw.m pocwmuoo AOMHZOU mHMHm UZHHQEQMm

4mm OH NN N 0a. No.0 huwauwcoo "pouowpoum Mmma4m w:

u ummwm mm 2 cowuuomm4

powum> me on cofiumuwuouomumno pcm ucmumcoo we Amv uoonoom Suez Hopoz mmuz

mucosuwcoo wowaom ozu 0cm Hopoz >ufisumc00 OLD ho pcuowpoum cow: was uaoocoo pocwosoo now mmuoom
cowumSHm>m pocfimubo cmoz cooBuom mocmpmmmwa mo mummH cosmoamacwwm pom cofiuowpoum mo uouum cooz .oe pomH

44

mo.v no ooooEAewEuR

 

mm.~ ©N.H mocmsuwcoo wowaoo upouowpoum
om HH mH.H mb.H pocfimubo 00mm UZHZMOm HHm4m 4 oszb
ads No mm.~ am.~ zuwouwcoo "oouowpoum ZOmmmm MHHmz 4
qm.H mo.m mocosuwcoo mowaoo Abouofipoum
No 0H mm.~ mH.m pmcwmuoo A4mMZDm 4 oanzmaa4
Hm mo 4H.H NN.m huesumcoo "pouoflpoum ZOmMMm MHHmB 4
Ha. mm.m mocosuwcoo onHob "pmuofipoum
«0 mm. om.H om.m pocwmuoo QAHmU 4 oszmHZDm
mm on ma. ob.m muwsuwcoo upouowooum zomMmm MHHm3 4
no.H wa.q cocooumcoo wmwaoo upouowomum
as. am. ma.A oa.s ooeeeoho memon aH>Ho mom ozHeememzozmo
cm so mo.~ qm.4 muwsuwcoo "pmuowpoum ZOmMMm MHHmz 4
mm.H NH.m occosuwcoo woflaoo upmuoapoum
m4. mm. . .
sa A an a ooaaooeo ozoo emH> may ozHemon
mm mm H4.H NH.m unwouwcoo "pouoapmum zomMmm MHHZS 4
~¢.H 0H.m mocoouwcoo mohaoo upouofipoum
Rae mm Na.a ~m.s ooeeeoeo aomezoo memHm ozHeozoma
Rum mm mm.H o~.m muwsuwcoo "pmuofipoum zomMmm mHHms 4
u umww am 2 cofiuummm4

asuz es edema

45

mOAV um ucmonchHmus

 

om. mm.H cocoouwcoo onHob "pmuoHpoum

a... H om K. Na; eoeafieo open oszmoe See: 4 osz:

soH m Hm H H¢.H MH.m muHsuwcoo "pouoHpoum 04002 4
0H.H m¢.m mocosuwcoo onHob "pouoHooum

«on N 0N H0.H mo.m oockuoo H4MMZDH 4 UZHQZMHH4

umq H mH 00.H ww.m huHsuwcoo "pouoHpoum 04002 4
mm. om.m oocosuwcoo onHob upouoHooum

sow N mm 0N.H Hm.m pmckuoo QHHmo 4 UZHmmstm

«mm m om mo. H~.o muHoumcoo "pouoHpmnm omomz 4
0m.H mm.¢ oocosuwcoo onHoo "ocuoHpoum

3mm 0H 00.H mH.q pochuno mHm0Hm HH>H0 mom 02H94mam20200

«mm 4 mm m0.H 4H.m huHsnwcoo "ocuoHpon omomz 4
mm.H mm.m cocosumcoo onHoo upcuuHooum

*mm m mm mN.H $0.0 UOGHMHno UZOU HHH> mmH UZHHZUHW

cm H 4N NH.H mw.m zuHsumcou "pouoHpoum omumz 4
mm.H m0.m oocosumcoo meHco "poDOHpmum

00 «0 mm.H HH.m pochuoo HOMHZOU mHmHm UZHHOZOMH

Sam m Hm ma. mm.m zuHsuwcoo "oouoHpoum omumz 4

u umww 0m 2 coHuummm4

onuz .04 mHan

46

 

mo .V um ucmonchHmnt

 

no.H 0a.H mocosumcoo meHoo "omuuHooum

IAN a ma oo.H as.~ emceeooo open oszmoe aHmem < oszo

sum m ow om. mo.H huHouwcoo "oouoHpmum HmHzozzoo 4
om.H Na.¢ cocoouwcoo HoHHoo “wouoHpoum

sac m we mm.H mm.m kuHauwcoo ”pouoHooum HmHzozzoo 4
om.H oH.m oocoouwcoo HoHHoo "wouoHooum

mm H mm 0m.H w~.o pochubo oHHmo 4 02H2mH20m

4am a sm os.H oN.s aoaoeweoo "soooeooum HmHzezzoo <
om.H so.m cocoouwcoo moHHoo "omuoHoon

ma Hm Ne.H mo.a oozeooeo memon HH>Ho mom ozHaemamzozmo

Ram N on ms.H aw.~ muHouwcoo "wouoHooum HmHzozzoo 4
mo.H 0H.4 mocosuwcoo HOHHob "oouoHpcum

mo Ho om.H NH.s oochDbo 0200 HMH> 02H 02H920Hm

mm H mo mm.H mo.m muHsuwcoo "wouoHooum HmHzozzoo 4
om.H mm.4 mucosuwcoo HmHHob "wouoHpoum

mH H so NH.N mm.s oochuoo HOMHzoo mHmHm 02HHozomm

Asa a ma Ha.e wa.a aheauweoo "soooeooea amHzozzoo <

u ummw 0m 2 coHuuomm4

m¢fl2 .04 oHnma

47

mo .V be acooEchH mus

 

NH.H -05.H mucosuwcoo moHHoo "omuoHomum

Ho 00 om.H 0N.H oochuoo 0020 02Hzmom HHm42 4 02Hmo

«04 q 0H H mm.H 00.N auHsuwcoo upEDOHooum 824HmMHomm 4
oH.H N¢.m cocosuwcoo onHoo "ocuoHooum

two N so mm.H Nm.m huHsuwcoo ”oouoHomHm H24Hmmeomm 4
00. 40.m cocoopwcoo moHHoo upouoHpoum

%N® N N4 N4 . H NM . m. UOfiHMUn—O DAM—“mo < UZHEWHZDM

400 N Hm mH.H 00.m muHouwcoo "wouoHooum 824Hmmaomm 4
H0.H 00.m mocosuwcoo moHHoo "oouoHooum

mo no mm.H mm.¢ pochuoo 0820Hm HH>H0 mom 02HH429020200

Sam 4 mm N0.H Hm.o muHouwcoo "pouoHooum 8249009020 4
mm.H 0N.m mocosumcoo HoHHoo "ocuoaooum

am mH 04.H Ho.m oochuoo 0200 HMH> 0:8 02HH20Hm

as H Nm oH.H ma.m muHsuwcou "wouoHooum H24Hmmyomm 4
00.H 0m.o mocosuwcoo HoHHoo "pcuoHooum

Rmm.N mm. NN.H mm.m pochuoo 4029200 mHmHm 02HH0202m

.40m 4 00 H Hm.H ma.¢ zuHsuwcoo "vouoHpoum 824Hmmaomm 4

u ummw 0m 2 coHuuomm4

mfinz .m4 oHan

48

m0 .V on ucmonchHmus

 

NH.H NN.H oocosuwcoo moHHmo "oouoHooum

04 no MN.H mm.H vochDoo 0020 02H2202 HHm42 4 02Hmo

«40 4 mm 04.H 4N.N muHsuwcoo "poDUHpoum 0HH02940 4
oH.H 0N.m cocosuwcoo moHHoo "omuoHooum

mm H 0H 0H.H 04.m pochuoo H4mmzom 4 02H0209H4

mm HH HN.H 00.0 huHsuwcoo "pouoHooum 0H402H40 4
NH.H m0.n mucosuwcoo onHmo "oouoHooum

mm H ON 00. 00.0 muHsuwcoo "oouoHooum 0H402840 4
00.H mm.m cocoouwcoo onHmo ”ocuoHooum

am so sw.H Ha.m eoeemoeo memoHe AH>Ho mom ozHaememzozuo

RNN N 04 4m.H mm.4 huHsuwcoo "pouoHpoum 0H402940 4
mN.H NH.m cocoouwcoo moHHoo “oouoHooum

sa H am oo.H sm.m ooeemooo ozoo HmH> use ozHemon

0H m0 00.H Hm.m muHsuwcoo ”omuoHpoum 0HH02H40 4
aw.H 04.4 cocoouwcoo meHoo “ooDOHooum

00 H 4m 00.H 00.4 oochuoo H02H200 mhmHm 02HH0202m

Row 4 0H H NN.H 0H.m zuHsuwcoo "wouoHooum 0HH02840 4

u umww 00 2 coHuuomm4

40-2 .04 pooH

49

mo.uV um ucoonchHmuIn

 

 

mm.H Hm.4 omcoauwcoo moHHob ”oouoHooum

s04 m mm m0.N 0o.m oochuoo HO0H200 2HOH0 02H902000

404 0 H4 H 4m.H mm.n auHsuwcoo “oouoHooum 00002940 4
40.H 0N.4 oocoouwcoo onHoo "oouoHooum

Ros N Na aw.H s~.s ooeemooo aomezoo mHaHm ozHeozoma

«H0 4 NN H Hm.H 04.0 huHsuwcoo uoOuOHooum Hz4Hm0Homm 4
N0.H 00.m cocoouwcoo mmHHob "oouoHooum

m H H4 4N.H 00.4 occhuoo 0009200 0HOH0 02H902000

440 m 00 N0.H H4.m huHsuwcoo “wouoHooum HmHzozzoo 4
00.H 00.4 mucosuwcoo meHoo "oouoHooum

0N 40 m0.H 00.4 oochuoo 000H200 mamHm 020802000

s0N N 04 04.H no.0 huHsuwcoo “oouoHooum 00002 4
4m.H 40.4 mocosuwcoo onHoo "omuoHoopm

on 00 mm.H 00.4 oochuoo HO0H200 2HOH0 02HH02000

«0N N 0m 44.H 00.4 muHsumcoo "wouOHooum 200000 0HH03 4
H0.H 40.0 oocmsuwcoo onHoo ”owuoHooum

st 0 MN H 00.H Hw.m oochuoo 0009200 0HOH0 02HH02000

S00 0 NH N 00.H 00.0 auHauucoo “vouoHooum 000H40 02

u ummw 00 2 .coHuuomm4

ucmumcoo mH A00 coHumuHuouomumo0 0cm ooHuo> mH Amy uoonoam c003 mnuz

Hmooz cocooumcoo moHHcm osu 0cm Hoooz zuHsuwcoo sou ho omuoHpoum cow: was unaccoU oocHosoo pom
mouoom coHumon>0 omchDoo cooauom mucou000H0 mo mummy mnemoHMchHw 0cm coHuoHooum mo uouu0 coo: .mm oHooH

50

00 .V um ucoonHcmH mus

 

00.H 00.4 cocmsumcoo moHHwo HomDUHomu0

smN N 00 N0.H 00.0 oocHouno 0200 HOH> 00H 02HH0000

04 HH N0.H NN.0 muHsuwcoo "oouoHoou0 0H000940 4
00.H 00.4 mocmouwcoo mmHHoo "ooDOHoou0

0N H 0N 04.H 0H.0 oochubo 0200 HOH> 009 02H000H0

00 0N 00.H 00.0 muHsuwcoo "oouoHoou0 5249008000 4
N0.H 00.4 oocmonwcoo mmHHmo "oouoHoou0

00 0N N0.H 00.0 oocHouno 0200 HOH> 00H 020900H0

aH so sa.H sw.a ashaoweoo “oooohooem amHzezzoo s
00.H 00.4 mucosumcoo mmHHoo “oouoHomu0

«No a mo as.H Na.a ooeroeo ozoo HSH> new ozHHmon

*00 N H0 00.H Hm.4 muHsuwcoo "oouowvou0 00002 4
N0.H 00.4 mocoouwcou moHHoo "oouoHoou0

00 H 04 00.H 0N.0 oochubo 0200 HOH> 000 02H000H0

N0 00 00.H NN.0 zuHsuwcoo HomuOHomu0 200000 0HH03 4
40.N 00.0 mucosuwcoo moHHoo "oouoHomu0

Inc a Ha Hm.H am.s ooaeooeo ozoo HMH> may ozHHmon

4H0 0 40 H 0H.H 00.0 0uHsu0coo "oouoHoou0 000H40 02

u ummw 00 2 coHuummm4

oanz .on oHoee

51

no .V no 23323 on...

 

44.H N0.4 wocmsuwcoo meHo0 "omuoHoou0

so A oH Ho.H oH.s ooahoooo mamon AH>Ho mom ozHaseHmzozmo

004 4 00 00.H 00.0 0uHsnwcoo ”omuoHomu0 0H000840 4
00.H H4.4 mocmauwcoo moHHon "oouoHomu0

No H mm so.H aH.s ooehoooo mamon HH>Ho mom ozHaemHmzozmo

«00 0 00 00.H No.0 kuHsuwcoo ”omuoHoou0 HZ4HO0H000 4
00.H 00.0 mucosuwcoo meHmo "omuoHoou0

«so a as mo.H aa.~ ooaaeooo mamoHe AH>Ho mom oZHHsmemzozmo

no 4N 00.H NH.0 zuHsuwcoo "oouoHoop0 80H202200 4
00.H H0.4 mucosuwcoo moHHoo "omuoHoou0

mm H «N om.H ma.s ooeaeooo memon AH>Ho mom ozHeemamzozmo

00 H 0N H0.H N0.4 zuwouwcoo "oouoHomu0 00002 4
00.H 04.4 oocoouwcoo moHHon "wouoHoou0

om mo oo.H no.4 ooaoeooo mamoHe AH>Ho mom ozHasmemzozeo

«HN 0 m0 04.H 40.0 muHsuwcoo "wouoHoou0 200000 0HH03 4
N0.H HH.0 mocmdumcou meHon "oouowomu0

«mm s ma H oH.H ao.a ooeeoooo msmon HH>Ho mom ozHeememzozmo

*00 0H 00 N 0N.H 00.0 muHsuwcoo "wouoHoou0 000840 >2

u ummw 00 2 coHuuomm4

00n2 .00 anmH

52

00 .V no osmonchHmus

 

N0.H 00.0 oocoauwcoo onHmo wowuoHomum

4H0 0 00 H0. H0.0 muwsumcoo "wouoHoou0 00000840 4
40.H 00.0 oocmsuwcoo moHHoo "omuoHoou0

R00 0 00 00. 00.0 howsuwcoo "000000000 8248008000 4
00.H 00.0 mocoouwcoo mmHHoo "omuoHomu0

0H H 0H 00.H 00.4 ooSHmuoo 00000 4 02HO0H200

sa4 N 00 H0.H 04.4 huHsuwcoo "oouoHoou0 800202200 4
00.H H0.0 mucosumcoo moHHob "oouoHoou0

00 H 0N H0.H 04.0 omcHouoo 00000 4 020000200

400 N 04 00. N0.0 huHsuwcoo "oouoHoou0 00002 4
40.H 00.0 mocmoumcoo onHoo "omuoHomu0

00 H 0H 0H.H 00.0 oochuno 00000 4 020000200

s04.N 00 00. 00.0 0u0oumcoo ”oouoHomu0 200000 08003 4
00. 00.0 oocoouwcoo mmHHoo "omuowomu0

400 N N0 HN.H 00.0 oocHouoo 00000 4 020000200

4H0 0 00 NN. 40.0 0u0=uwcoo "oouoHomu0 000840 02

u ummw 00 2 coHuuomm4

mauz .om oHooa

53

00 .V um ucmo000cm0 mayo

 

0N.H 00.0 moawzuwaoo 000H00 uvmuo0vmu0

”q mo N¢.H mm.m UNCHQUQO A<mmzph < UZHQZWHH<

*00 N Nq 00. 00.0 0u0suwcoo "0mu00vmu0 00000040 4
oH.H 00.0 mucmauwaoo 0m0Hmn "vmuo0wmu0

00 Ho 00.H 0N.0 0mc0mu00 0400200 4 020020004

*oN m 00 00. 00.0 0u0=uwcou "0mu00vmu0 0240000000 4
00.H H0.4 woCmsnmaoo 000Hmn "vau000mu0

mo MO 00.H wm.¢ VQCHGHDO A<MWZDM < UZHQZWHH¢

*N0 m 00 00.H H0.0 0u0=u0coo "vmu00vmu0 000202200 4
oH.H 0H.0 mocmsuwaoo 0m00m0 "vmuo0vmu0

0¢ H 0N 00.H N¢.0 wmc0mu00 0400200 4 020020004

NH N0 00.0 q¢.0 0u0suwaou uvmu00vmu0 00002 4
HN.H 00.0 wucmsuwcoo mmHHmn "vmuo00mu0

mm OH mm.H QN.W VQGHMUDO A<MMZDW < UZHQZNHH<

*00 N #0 «0.0 00.0 0u0=uwaoo uvmu000mu0 200000 00003 4
0H.H 00.0 mocmsuwaou 0m0Hm0 «00u000mu0

0H 00 0N.H 00.0 vaHMuno 0400200 4 020020004

*00 q 00 00. 00.0 0u0=uwcou "vmuUvau0 000040 02

u ummw 00 z :00uummm4

Nmuz .mm mfinma

-.

54

00wv um ucm0000cw0mu*

 

 

mm. 00.H mocwsuwcoo mm0Hmn uwmuo0vmu0
00 00 N0.H HN.H 0w00wu00 0000 0202000 00040 4 02000
«00 0 «N H NN.H 0m.N 0u0=u0coo uvmu000mu0 00000040 4
0H.H 00.H mocmauwcoo 0m0Hmn ukuo0vmu0
qo 8 2.0 $0 8538 8% ”02058 SE ¢ 025:
30 w R 0 2.0 36 30338 6338.5 0.23.3080 <
#0. N0.H mucmsnwaoo 0m00m0 uvmuo0wmu0
«3 c mm 2.0 3.0 8338 was 94058.0 0:2: < 90.0.:
mm H #0 00. 00.H 0u0suwcoo "wmuo00mu0 000202200 4
00.H H0.H mucmsuwaoo 0o0Hmn "vmuo0vmu0
No 2 8.0 mm; 8538 8% “020208 0020 < 308
«00 q 00 N0.H m¢.N 0u0auwcoo "vmu000mu0 00002 4
40.H «0.0 wucmsuwcoo 0mHHmn "0mu000mu0
0N ¢0 0H.H 00.H 0mc0mu00 0000 0202000 00040 4 02000
«00 0 N0 H «0.H N0.N 0u0suwaoo uvmuo00mu0 200000 00003 4
00.H 0N.N mucwsuwcou 000000 ukuo0va0
*¢0.m 00 . . ,
00 00 H 0ma0mu00 0000 0202000 00040 4 02000
*00 0H 00 N N0. N0.m 0u0=uwcou “vmuu00mu0 000040 02
u ummwm 00 2 a00uummm4

0¢u2 .00 mHnw0

55

by the respondents to each of the'36 assertions. In both
instances, the t test for correlated measures (4) was
used to determine if the mean difference between predicted
and obtained scores was greater than zero for both models.
The resulting 5 value and its significance is indicated for
each of the 36 different assertions resulting from the two
test situations.

Table 6 presents a composite of the data contained in
Tables 4a-f and 5a-f without regard to the individual
assertions. These results are relevant to the testing of
Hypothesis 1. The data provide support for the hypothesis
that the belief congruence model manifests significantly
less total error of prediction, thua,less error per assertion.
'Table 6 indicates that the congruity model predictions re-
sulted in over two times as much error as did predictions
based on the belief congruence model. The congruity model
error totaled 54.37 scale units compared with 23.95 scale
units for the belief congruence model. When averaged across
the 72 assertions, the average error of prediction is .75
scale units and .33 scale units for the congruity and belief
.congruence models respectively. The average error of pre-
diction for the belief congruence model parallels very closely
that reported by Rokeach and Rothman (12). These researchers
reported an average error per assertion of .34 for the belief
congruence model and 1.07 for the congruity model. Thus,

average error of prediction for the belief congruence model

56

.umvuo ucmum0000 m 00 u50 .0 005ousu H 005000 0000 00 0wumu mumz

mm NH £w5ou0u m 005ouw ummu 00 0mumu wc0m0 mGO0uumwmm mama wnu £u03 .mc00uummmm 0 0mumu 05ouw ummu nom0
.0Hmumu5uom uommmo :OHuomuoucH mnu Umuo00mu0 Hmwos msu umsu 0085mmm m0 u0 .umwuuu msu 00 mocmuo00000

:0: m0 cu 0mo0auwuw0 m0 mocmum0000 came mnu :m03

...V US“ m. mQHDmH CH fiwCHMUGOU 0“ $0000 COHUMDHN>Q

came mnu 00m vmuo00mu0 muoom 000um5Hm>m came msu cmmzumn monouomm00 came mLu ou 000000m m0 0005mmwa

 

 

 

 

0mum0muuoo now ammuuu 0:0 cmnz UmGHMuno muH5mou co vmmmn 00 :00uo0vmu0 00 0omu5oom now soHHmwmmwuwnHH
N0 No.00 00 0.0N 0H 00. 00. 00.0N 00.00 0H0 Hmuo0
0 0.00 q 0.0H H 0N. 0N.H 00.H 00.0 00 NH
0 0.00H 0 0.0H H 00. H0. 00. no.0 N0 HH
0 0.00 0 0.0 0 0N. 00. 00.0 00.N 00 00
0 0.00 q N.mm N 00. 00. 00.N 00.0 00 0
0 0.00 m 0.00 0 00. «0. 00.0 N0.N 00 0
0 0.00 m 0.0 0 N0. 00.H HH.0 00.0 00 m
0 0:00H 0 0.00 0 0H. 00. 00.H 00.N 00 0
0 0.00 q 0.00 H mm. 00. 00.0 00.0 00 0
0 0.00 0 00.0H H 0N. .00. 00.H 00.0 00 q
0 0.00 m N000 N 0N. 00” 00.0 00.0 00 m
0 0.00 0 0.00 q NN. 0N. 00.H N0.H 00 N
0 0.0H 0 0.0 0 00. 00.H HN.¢ 0H.00 00 H
mGOHuummm4 N .02 N .02 .\0000 000500 .00000 0u05uw
H 00 um0552 mocm5uwsoo 0m0Hw0 0u05uwco0 0m0Hm0 -000 0m0Ho0 1000
0kumu5oo4 0muo00mu0 ch0uummm4 0000 :00u000 muumflmdm “@0852

H

nummm4 000 00000
mu5Homn4 m0muo>4

umu0 mo uouu0

mu5Hom04 Hmuo0

00 05000

um0852 umm0

 

mso0uuwmm4 cam 005000 umm0 HHm mmouom mHovo2 moco5uwcoo

0m0Hm0 0cm 0u05u0500 onu 00 0Hmumu5oo4 0QUo00ou0 chHuummm4 00 um0552 0cm udmouw0 mnu 05w
.:00uummm4 uma s00u000mu0 00 uouu0 mu5Homn4 mwmuw>4 mzu .c00uo00ou0 mo uouu0 mu5Homn4 Hmuo0 wn0

we magma

57

varied only .01 scale units in the two studies, while average
error for the congruity model varied .32 scale units. On the
other hand, when compared with the Rokeach and Rothman find-
ings, the congruity model showed a 40 percent decrease in
average error of prediction in the present study. Although
the reduction of .32 scale units per assertion is substantial,
it evidently had little or no effect on the overall ability
of the congruity model to predict cognitive interaction
accurately, as is shown under Hypothesis 2.

The design of the study made it possible to test
whether varying or holding constant either subject (S) or
characterization (C) in regard to the total CS configuration
affected the average error of prediction per assertion for
both models. Evidently, differences in the variable and con-
stant elements of the CS configuration do not affect the
dynamics of either model. As Table 7 shows,the average error
of prediction per assertion between the two test situations
were not significantly different. The congruity model showed
a .74 and .76 average error of prediction per assertion, while
the belief congruence model showed a .32 and .34 error per
assertion.

Thus, the present data clearly support the results re-
ported by Rokeach and Rothman (12) illustrating significant
superiority of the belief congruence model over the congruity
model as predictors of cognitive interaction. Moreover, the

data clearly support Hypothesis 1: that the belief congruence

 

 

58

model will manifest significantly less average error of pre-
diction per assertion than will the congruity model when
tested under comparable situations.
Hypothesis 2:

When assertions that express a situation or activity
are tested under identical conditions, the belief congruence a“;
model will predict accurately a significantly greater number { 1
of word combinations than will the congruity model.

Table 6 indicates that for the 72 different tests of

the two models the relative superiority of the belief con-

 

gruence model is readily apparent. Each of the 12 test groups .p}
responded to six assertions and their appropriate subject (S)
and characterization (C) components. Thus, it was possible
to obtain six independent tests of both models in each test
group. If the mean difference between the predicted and ob-
tained model scores as listed in Tables 4a-f and 5a-f did not
differ at <f.05, it was assumed that the model predicted the
cognitive interaction accurately. Hypothesis 2 is designed
to test directly the relative efficacy of the two models
based on accuracy of prediction across assertions.

The results summarized in Table 6 support Hypothesis 2.
Of the 72 independent tests, the belief congruence model
accurately predicted 48, or 66.02 percent of the assertions,
while the congruity model accurately predicted only 19, or
26.3 percent. This relative predictive superiority of the
belief congruence model is not significantly altered when the

structure of the assertion is taken into consideration.

59

Table 7 shows that the belief congruence model predicted
equally well for both types of configurations; i.e., 24 of
the 36 predictions were correct when the subject (S) was
varied and the characterization (C) was held constant and 24
of the 36 predictions were correct when the characterization
(C) was varied and the subject (S) was held constant.

When compared with the Rokeach-Rothman (12) results,
the percentage of accurate predictions by the two models is
similar. Rokeach and Rothman found that the belief congruence
model predicted accurately 75 percent of the assertions they
tested while the congruity model predicted accurately only 25
percent of assertions tested. Their findings compare almost
exactly to the congruity model results reported above and
differ only slightly from the 66.02 percent accurate predictions

of the belief congruence model obtained in this study.

Objective 2

 

The second objective of this investigation was to test
the relative reliability of the congruity and belief congruence
models as predictors of cognitive interaction resulting from
word combinations when the same combinations are tested across
equivalent groups of subjects. Obviously, the ability of a
model to accurately predict across different groups is a sig-

nificant test of reliability.

60

 

 

 

 

 

0.00 0N 00. 00.NH 0.00 0N N0. 00.HH 0000000000
000H00
93 m 00. 2.: mam S .00. 00.00 300080
0 .00 a... .00 .IIIEB:
0000 .0000 0000 .0000
0H00005oo< u0000< 00 0H00005ou< :00000 00
000000000 000 00000 000000000 000 00000
0000000000 00000 0002 0000000000 00000 0002
00 000852 0002 H000H 00 000052 0002 H000H
0000005000000 00.00000000 00 0000005000000 00 00000800 00
00000000 0H0: ‘0 .00000> m 00000> 0 .00000000 0H0: 0
000008 00000000< 00 0>000H00 000005000 0000 00 000002
0000005000000 00 00000000 000 00 00000>
00 00000000 0H0: 000000 000 on 0000000000000000 000 000 A00 0000050 000 0003 0H0002
0000500000 000H00 000 000500000 000 00 0H00005oo< 000000000 0000000000 00 0000000 000
000552 000 000 .00000000< 000 0000000000 00 00000 0002 000 .0000000000 00 00000 0002 H0009 .0 0H009

61

Hypothesis 3:

When assertions that express a situation or activity
are tested under identical conditions, the belief congruence
principle will manifest significantly greater reliability of
prediction when compared to the congruity principle.

To test Hypothesis 3, it was necessary to assume that
test groups in which the assertion subject (S) was held con-
stant and the characterization (C) was varied were equivalent
to the test groups in which the subject (S) was varied and
the characterization (C) was held constant. Since it has
been pointed out that this change in the assertion structure
had little or no effect on the relative ability of the two
models to predict cognitive interaction accurately, or on
the average error per assertion, it seemed reasonable to make
the equivalence assumption.

As Tables 4a-f.and Sa-f indicate, only five (26.3% of
the 19 accurate predictions were common to both test groups
for the congruity model, while 40 (83.3%) of the 48 accurately
predicted assertions were common to both test groups for the
belief congruence model. In other words, only five of the 11
assertions predicted accurately by the congruity model in the
test groups where the assertion subject (8) was held constant
and the characterization (C) varied were again predicted
accurately when the reverse was true. However, in the case
of the belief congruence model, 20 of the 24 assertions pre-

dicted accurately in the first test situation were again pre-

dicted accurately in the second test situation. This result

'I. 1.x: ‘ (1

V'-

 

7
m:-_

62

supports the position that the belief congruence model is sig-
nificantly more reliable in predicting across equivalent test
groups, a finding consistent with Hypothesis 3.

The Relationship of the Assertion (CS) .

Components to the Ability of the Congruity

and Belief Congruence Models to Predict
Cognitive Interaction Accurately

 

 

 

 

As reported above, the congruity model predicted
accurately only 19 of the 72 word combinations tested, while
the belief congruence model predicted 48 of the 72 word com-
binations correctly. The present discussion examines how
the component parts of the CS configurations are related to
the models' failure to predict cognitive interaction. Obviously,
the congruity model failed to predict correctly on 53 occasions,
while the belief congruence model failed only 24 times. Table
10 contains a summary of how these failures to predict are re-
lated to the subject (S) and characterization (C) components
of the assertions tested.

Examination of Table 10 reveals that the nature of the
components of the assertions appears to be related to the ability
of the models to predict accurately the cognitive interaction
outcome. For example, in the assertion where the subject (S)
is MY FATHER, neither model was very successful in predicting
cognitive interaction. The congruity model failed 100 percent
of the time, and the belief congruence model 83 percent. On
the other hand, in the assertions where the subject (S) com-

ponent is A PROTESTANT, the belief congruence model failed

0
000

 

EE—‘A—

63

25 percent of the time, but the congruity model showed an 83
percent failure.

The data show that the 12 assertions in which MY FATHER
serves as the subject (S) contributed nearly twice as much to
the total predictive failures of the belief congruence model.
(42%) as they did to the failures of the congruity model (22%).
This difference is further illustrated by the fact that the
percentage contribution to total predictive errors by the
subject (S) component ranges from four percent, (A WHITE
PERSON) to 42 percent, (MY FATHER) in the case of the belief
congruence model, but from only 13 percent, (A WHITE PERSON),
to 22 percent, (MY FATHER) in the case of the congruity model.
It is interesting to note that the same subject (S) components
contributed both the largest and smallest percentages to the
overall number of predictive failures in both models.

When individual characterizations are considered, it
is evident that PROMOTING BIRTH CONTROL was the most-difficult
characterization for both models to predict accurately. The
congruity model failed on all 12 assertions (100% failure) in
which this characterization appeared, while the belief con-
gruence predicted only six (50% failure) correctly.

As further indicated in Table 8, when the character—
ization (C) components were considered, the relative differences
of rate of failure between the two models are not as pronounced
as they were in the case of the subject (S) component. Still,

in only one instance did the congruity model equal or exceed

 

t ~
~
’3'
V
l
1

.4

13 .'Q:._- . q
.
lv'

64

.00>00000 00050000 00 000550 00000 000 00 05000 00 0050000 0000000000
0 0003 0000000000 003 A0 00 00 000000500 00000000 0 00500 00 0000000 000 0000000 050000 000800

.000000500 0500

000 00 00000 000000000 N0 00 0000000 00 000000 00005 000 00500 00 0000000 000 0000000 050000 0009*

 

 

 

 

000 xx 00 000 xx mm 00000
00 mm 01. 00 Na 00 0500 0202000 00000 0 02005
0 0 0 00 00 0 0000200 0 020020000
00 00 0 00 mm 00 00000 0 020000250
00 mm m 00 m0 0 000000 00>00 000 0200000020200
00 00 m 0 mm m 0200 000> 000 02000000
mm on 0 00 000 NH 0000200 00000 020002000
000 2000000000000000
000 xx 00 000 xx 00 00000
N0 00 mu 00 00 .0: 00000000 0
00 mm ,m 00 mm 00 0200000000 0
00 mm m 00 00 0 000252200 0
00 mm q 00 00 0 00002 0
q 0 0 00 00 0 200000 00003 0
00 mm 00 «N 000 00 000000 02
000000 0.00. 000000 0.00
000 000000 000000 0.00 N0 000 000000 000000 0.00 00

00000000 00 00050000 00050000 00000000 00 00050000 00000000
00 000000000 00 00000000 000552 00 000000000 00 00000000 000502

 

00002 0000500000 000000 00002 000000000

 

0000005000000 00 000 00 0000 003 0000 00 000000500 0003
00000000000 0>0000000 0000000 00 000000 000002 00509 00 0000000 000 000552

000 0000050

000000500

 

00000005000000 00 00000500 000 00 0>000000 ADV 0000000000000000 000 Amy 000n00m
000 00 00000050 0 00 00000000000 0>0000000 0000000 00 000002 0000000000 000000 000 000500000
000 00 00000000000 000 00 00>0000o 00050000 000 00 000050000000 0000000000 000 000000000 000 "0 00009

65

The ability of the belief congruence model to predict cognitive

interaction. This was in the instance when FIGHTING THE VIET
CONG served as the characterization in the assertions. In
this case, the characterization contributed only six percent
to the total failures manifested by the congruity model, while
it contributed 21 percent of the failures of the belief con-
gruence model. Unlike the subject (S) components, the overall
range in percentage of failures contributed by the character-
izations is nearly the same for both models: six percent to
22 percent for the congruity model and four percent to 25
percent for the belief congruence model.

The Belief Congruence Model
and Overassimilation

 

Rokeach and Rothman (12, p. 130) refer to a.process
whereby a stimulus not only takes on the valence of another
stimulus with which it is associated, but in addition,takes
on an even stronger valence. This process is sometimes re-
ferred to as overassimilation. Although the belief congruence
model is designed to account for this possibility, there are
evidently certain cases where the present formulation does
not function adequately.

The data in Table 9 clearly show that after the CS com-
parisons are carried out under the belief congruence principle
the respondent may overassimilate towards either the subject
(S) or the characterization (C) component. Based on the rela-
tive influence of the C component, it appears that overassim-

ilation is most often in the direction of the evaluation score

 

66

“Munfi uf' kl, . r
“o. All. .I‘d‘ .— 0‘ ...- #2

.A0000000500 mu 0> U 000 00 o 000 0000300
00000005000000>0v 0000000 00000 w00005000 000000 00000 0000300 0000 0000 0000 0000000 00000 w000o5o00
000000 00000 0000300 20000000 0000 0000 2000 0000 000000000 0000000 00 00 00 .0020200 20200 020002000
mmmH<0 MS .000000000 000 w00000 00000000000 000 000 00 .0005000 000 .5000 0000000050 0>00000m 000 00

0000000500 mu 0> U 000 w00x05 0003 U 000 0000300 0000005000000>o 003 00000000000 00 000550 000 20
000000000 0000 w00000 00000000000 00 000550 00000 000 w0000>00 00 00000000 0003 00w00000000¥

 

 

 

m 0 m w 0 0 000000<U 4
N0 0 N 0 m 0 0200000000 0
m w w 00 00 0 000202200 <
m N N N0 m 00 00002 0
00 m 0 m m 0 200000 00003 0
mm 00 0N m0 0m ow mmmH<0 w:
*ufimoumm
0020 0202200 0020200 00020 000002 00>00 0200 000> 0020200
00042 0 < 000 000 00200 000000000
< 02000 020020000 020200200 0200020020200 02002000 020002020 0000050

 

000000500 0000000000000000

 

000000005000000>ov
003 0500 000 00 0000000000000 000000 00000 0000300 0000 2000 0000 Amov 000n05m
0000000000000 000 0000300 0000000 00m0000m 0 000000000 003 00000000000 00 0000000 000 .0 00000

67

assigned to that component. This observation is supported to

some extent by the relatively high correlation coefficient

(r=.76) observed between the C and the combined CS configuration.2

In other words, the evaluation score assigned to the character-
ization by the respondents is in itself highly related.to the
CS evaluation regardless of what subject (8) is associated

with the characterization (C). In fact, data in Tables lOa-f
and lla-f indicate that if the C evaluation scores are compared
with the CS, applying the same criterion used to determine
accuracy of prediction between the two models, 43 of the 72
assertions are predicted accurately. This represents only five
fewer accurate predictions than were achieved by use of the
entire belief congruence procedures.

Based on a criterion of total mean error of prediction
across all 72 assertions (36 tested twice), the congruity model
error totaled 54.37 units and the belief congruence model
23.96 units. When the characterization evaluation scores only
are used as the predictor, the total mean error of prediction
is 25.62 units. Thus, for this set of data, the application

of the overall belief congruence procedures reduced the total

 

2A Pearson Product-Moment correlation was used to deter-
mine the relationship between the C and S, the S and CS, and
the C and CS. The resulting r's were: r=.04; r=.l3 and 40.76
respectively. Correlations were computed over the entire 3,679

observations without regard to assertion or group differentiation.

 

n..x|u~mu.' . 5- F
1

V": ’
l

68

 

¢¢.H
oo oo ¢¢.H
«mm mm NH m om.o
«m.m
om mo m¢.m
«¢N m mH H om.o
oo.m
Hm NH mq.m
$mm o 50 H om.o
mo.¢
*om m nq No.m
«No oH ¢o N 9m.o
ma.¢
«qq mo mw.q
*Hm 0 H5 H om.o
mm.¢
«ac q mm Hm.m
*oq m we N om.c
.wan
u cum: 2

momv um ucmoHMchHun %

UDMQ Uszmom HHm<m 4 wszD
63mm UZHZMOM HHm<m < UZHmD mmmH<m w:
mmmfi<m w:

A<mm23m ¢ QZHQZMHH<
A<Mm23m < QZHQZMHH< mmmH<m W2
MmmH<m W2

QAHmU < OZHmmHZDm
QAHmo 4 qummstm Mmmfi<h M2
mmmH<m $2

mHmUHM AH>HU mom UZHH<MHmZOZHD
mHmUHm AH>HU mom UzHfldmemzozma MmmH<m M2
mmmfi<h NZ

UZOO HMH> mmH UZHHmUHm
Ozoo HMH> mmH UZHHmem mmmH¢h wz
Mmmfifim WE

AOMHZOU mHmHm UZHHOZCMm
AOMHZOU :HMHm OZHHQZOMm MmmH<m W2
mmmH<m WE

 

mmuz omHum> mH ADV GOHumNHuwuumumnu mnu van ucmumcoo vam mH va uomflnsm mnu cmzz mucmcomEoo
03H mnu mo comm wcm coHumustmcoo vwcHnEoo mcu mo aowumsHm>m mnu cmmBumm mocmumwan cmmz .moH mHan

69

 

mq.H

mm H mm 00.H

*mo «H wH qw.¢

oo.m

mm H oH oH.m

mo H mm #w.q

mm.m

mm.H mH. «m.m

*qo N on qw.¢

mH.¢

mm H «m mq.q

0N H mm qw.q

wo.m

Hm H mm mm.m

*mm N mm qw.¢

oo.m

oq H wH Nw.q

mo No «m.q
u .uwHQ ‘2

cmwz

mo.V um ”:83chme u y»

03mm OZHZMOm HHm<m < Osz:
mea wszmom HHm<m < UZHmD ZOmmmm NHHmZ <
ZOmmmm MHHm3 <

A<mm23h < UZHQZMHH<
A<Mmz2m < wZHQZMHH< ZOmMmm MHHm3 ¢
ZOmmmm MHHm3 <

QAHEU < OszmHZDm
QAHmu < oszmHZDm ZOmmmm MHHSS 4
ZOmmmm mHHmS 4

mHmUHM AH>HU mom QZHH<MHmzozmn
mHIOHm AH>HU mom UZHH<mHmZOzmo ZOmMMm mHHmS <
ZOmmmm MHHmS <

0200 HMH> mmH UZHHmem
wzoo HMH> mmH OZHHSUHM ZOmmmm MHHmz 4
ZOmmmm MHHm3 d

AQMHZOU mHMHm UZHHOZCMm
HOMHZOU EHMHm OZHHOZOMm ZOmMMm MHHmB <
ZOmMMm MHHE3 d

mqnz noH mHan

70

 

 

mm.

mq no Nm.
«mm NN c¢ ¢ mm.
o¢.

*mm N mN mo.
mm mo mm.
mm.

*mN N NN ON.
mm mo wm.
Ho.

om NH mH.
*mN o cm H mm.
om.

%NN N mm mo.
Nq H Hm mm.
0H.

00 00 OH.
¥Hq N mm mm.
u .men z

cmmz

mo.V um unmoHMchHm u an

UDMQ QZHZMOM HHm4= 4 oszD
wbma 62H2mom HHm4m 4 qum3 omwmz 4
Omumz 4

A4MNZDm 4 UZHQZMHH4
A4MMZDm 4 UZHQZMHH4 omomz 4
omwmz 4

QAHmU 4 UszmHZDm
QAHmU 4 UZHmmHZDm OMUMZ 4
Qmumz 4

mHmem AH>HU mom UZHH4MHmZOZMQ
mHmem HH>HU mom DZHH4MHmZOEmn omwmz 4
omwmz 4

0200 emH> mma wzHemon

ozoo emH> mug quamuHm omomz ¢
nomomz.<_

Homazoo mHmHm quHQZOMm

HOMHZOU mHMHm UZHHOZOMm omumz 4
omomz 4

omuz .ooH «HQMH

71

 

 

N0.H

*mq mH H m¢.N
mm «N no.N
Hm.m

¢Nm «N mm.q
*qN oa H No.N
mm.m

*HN Nw. NN.¢
*HN oH N mo.m
NN.m

H¢ HH Nc.m
«mH mm No.N
Nm.q

«wH mo NH.¢
«Hm on H N0.N
mm.¢

mN .NN hmtq
«No me H N0.N

u .meo z
cmmz

mo.V um unmoHMchHm u .1

USMC wszmom HHm4m 4 OZHmD
03mm UZHZMOM HHm4m 4 UZHmD HmHZDZZOU 4
HmHZDZZGU 4

A4mm23m 4 UZHQZMHH4
A4MMZDM 4 UzHozmHH4 HmHZDZZCU 4
HmHZDZZCU 4

QAHmU 4 UZHSmHZDm
QAHmU 4 qummHZDm HmHZDZZOU 4
HmHZDZZOU 4

mHmem AH>HU mom UZHH4mHmZOZma
mHmon AH>HU mom QZHH4MHmZOZHQ HmHZDZZOU 4
HmHZDZZUU 4

0200 HMH> mmH UZHHEUHm
OZOU HMH> WEB GZHHmem HmHZDZZOU 4
HmHZDZZOU 4

AOMHZOD mHMHm DZHHOZUMm
AOMHZOU mHmHm GZHHOZOMm HmHZDZZOU 4
HMHZDEEOU 4

mquz .voH mHnma

72

 

mm. N . q¢.H
H N HN.H
*Nm HH No m mm.m
m¢.m

mm mH wN.m
HN mo mm.m
mw.m

*ow N mq om.m
oH mo mm.m
mm.m

Nq mo No.m
*Ho m we H mm.m
mN.m

«m mH H¢.m
oN mo mm.m
mH.¢

*Hm N qq mm.m
*mm m mm H mm.n

u .mmHo 2
sum:

mo.uv um unmoHMchHm u «

was oszmom Sea 44 was
was ozHZmom EBB < osz: azfimmeomm <
Bzfimmaomm <

A4MMZDm 4 UZHQZMHH4
H4mm23m 4 UZHQZMHH4 Hz4HmMHOMm 4
Hz4HwMHOMm 4

QAHmU 4 wszmHZDm
QAHEU 4 UZHmmHZDm Hz4HmMHOMm 4
Hz4HmMHOMm 4

mHmUHm AH>HU mom OZHH4MHmZOZMQ
mHmUHm AH>HU Mom UZHH4MHmZQZMQ Hz4HmmHomm 4
HZ4HmNHQMm 4

0200 HMH> mmH UzHHmUHm
Uzoo HMH> mmH ozHHmon HZ4HmmHOMm 4
Hz4HmMHomm 4

AOMHZOU meHm GZHHOZCMm
HOMHZOU mHmHm UZHHOZOmm HZ4HmMHOMm 4
HZ4HmMHOMm 4

squz .moH mHan

73

 

o¢.H

*Ha N 04 om.H

*mo NH oq m 0N.m

NN.m

No N mN md.m

MN H HN 0N.m

ww.m

mm mo mN.n

*mq N mm 0N.m

mm.m

mm H Nm H¢.m

*wo m mm H 0N.m

mo.m

*MN N m¢ qw.m

No H wN oN.m

om.¢

Nm H «N oo.¢

*mm m 0N H 0N.m
u .mmHa 2

sum:

mo.V um ucmoHMchHm u an

mama ozHZMOh HHm4= 4 UZHmD
02mm wZHZMOm HHm4m 4 qumD UHAomH4U 4
UHAomH4U 4

A4mm23m 4 QZHQZMHH4
A4MmZDm 4 UZHQZMHH4 UHaomH4U 4
oHaomH4U 4

QAHmU 4 OszmHZDm
QAHEU 4 UszmHZDm UHAomH4U 4
UHAomH4U 4

mHmUHM AH>HU mom wzHH4MHmzoZmn
mHmem HH>HU mom QZHH4deZOZmQ UHHOZH4U 4
CHAOmH4U 4

0200 HmH> mmH UZHHmeh
D200 HMH> mmH UZHHmUHh UHHO$H4U 4
UHAomH4U 4

AOMHZOU EHMHm OZHHOZOMm
AOMHZOU mHmHm UZHHOEOMM UHAOEH4U 4
UqumH4U 4

«muz .moH mHQMH

74

 

8V 3 £833:me u u...

 

wm.¢ Homazoo mHmHm quaozomN
Now N «a «o.m gouazoo mHmHm quaozomm oHaoma<u <
*qo m No H Ho.m 0Haomy<u 4
mm.q Homazoo mamHm ozHHozcmm
mm H «m ¢N.q Homezou mamHm ozHaozomm Hz¢ammeomm <
*mq q oq H «o.m ez<ammaomm 4
mm.q Homazoo mHmHm quaozomN
«Nm N mq No.¢ aomazoo mamHm quaozcmm amHzazzco <
*mo m Nq H Nm.N amstzzco <
mm.q Homazoo mamHm ozHanomm
NH No om.q Homezoo mamHm quaozcmm omumz ¢
mm.q nomazoo mHmHm wzHeozomm
«0 Ho Nm.q Homazoo mamHm wzHeozomm zomMmm meHmz <
*Nm N am NH.m zommmm mHHms <
mm.q Homazoo mamHm quHozcmN
*mN m NN Hm.m Homazoo memHm ozHaozomm mmme<m N2
«NN.0H mo.N .
oq o mmma<m w:
in. it: z.|
cum:

 

on ..o c .. . .o . . . .. .o o.. o. .. . . .. .c-T ...o...‘oU
I

039 mnu mo 50mm was coHumuswwmaou vmcHnaou wcu mo :oHummHm>m mnu :mwzuon mocmummea cmmz .mHH mHan

75

 

 

 

mo.q

”HAHN no. $6
No mH om.m
mo.¢

Hm H Nq mH.m
mm H mm om.m
mo.¢

«qo m mm ow.m
*qm N am mm.N
mo.¢

*mN m «w Nm.m
%o¢ N Nm oo.m
wo.¢

«om N 00 wN.m
qq oH mm.m
wo.¢

No H mN mm.¢
*Nm o oH N m¢.o

u .MMHQ 2
sum:

mo..V um unmUHMchHm u .4.

wzoo HMH> mmH GZHHmuHm
0200 HMH> mmH ozHHmUHm UHAomH4U 4
UHaomH4U 4

0200 HMH> mmH UZHHmem
Uzoo HMH> mmH UZHHmuHm HZ4HmMHOMm 4
HZ4HmMHOMm 4

0200 HMH> NIH UZHHmUHm
Ozoo HMH> mmH wzHHmem HmHZDSZGU 4
HmHZDZZOU 4

6200 HMH> mmH UZHHmUHh
uzoo HMH> mmH quHmUHm OMOMZ 4
omomz 4

0200 HMH> NEH UZHHEUHm
UZOU HMH> Ema UZHHmUHm ZOmmmm MHHm3 4
ZOmMMm MHHm3 4

0200 HMH> mmH OZHHmuHh
U200 HMH> mmH UzHHmOHm MMIH4m M2
mmmH4m wt

omuz .nHH oHan

76

 

Hm.m

mm H mN oH.¢
%ON 0 «m H oN.m
Hm.m

wN mN mH.¢
*NN n «q H mo.m
Hm.m

*om n ma mm.N
mm NN HN.N
. Ho.m
*oH ¢ Ho Nm.¢
«No N mm NN.m
Hm.m

*mm N ¢¢ mm.q
*Qm q mN H mm.m
Hm.m

Hq No ¢w.m
*mm HH mm N ¢o.o

u .HMHQ 2
sum:

moV um “:83ch: u «

mHEOHM AH>HU mom UZHH4¢HmZOEmn
mHmUHM AH>HU mom UZHH4MHmZOzmQ UHAomH4U 4
UHAOEH4U 4

mHmUHm AH>HU mom OZHH4MHmZOZMQ
mHmUHm AH>HU mom QZHH4MHmZQEmo HZ4HmNHOMm 4
HZ4HmMHOMm 4

mHmUHm AH>HU mom UZHH4MHmZOzmo
mHmUHm AH>HU mom UZHH4MHmZQZMQ HmHZDZZCU 4
HmHZDZZOU 4

mHmUHM AH>HU mom UZHH4MHmZQZmQ
mHmem AH>HO mom UZHH4mHmzonQ omumz 4
omwmz 4

mHmem AH>HU mom UZHH4MHmzozmn
mHmuHm AH>HO mom UZHH4MHmZOEMQ ZOmMmm MHHmS 4
zommmm WHHm3 4

mHmUHM AH>HU mom OZHH4MHmZOme
mHmem HH>HU mom QZHH4MHmZOzmQ MmmH4m M2
MmmH4m M2

mmuz .oHH wHQMH

77

 

1F“;

 

wo.m

mo 0N em.m
m¢ HH mm.m
wc.m

00 HH mm.m
mo 00 mm.m
wo.m

«mm 0N Nm.¢
*HH No N mw.N
wo.m

we 0N N¢.m
we NH om.m
wo.m

Nw oH mm.m
mm oH Nq.m
wo.m

we no mN.m
¥¢H on Hm.o

u .mmHo z
cmmz

mo.V um unmonchHm u an

QAHmU 4 UZHmmHZDm
QAHmo 4 QZHEMHZDm OqumH4U 4
UHHOSH4U 4

QAHmU 4 QZHmmstm
QAHmU 4 UZHmmHzam Hz4HmMHomm 4
Hz4HmMHOMm 4

QAHmU 4 UZHmmHZDm
QAHEU 4 oszmHZDm HmHZDZZOU 4
HmHZDZZOU 4

QAHmo 4 UZHmmHZDm
QAHEU 4 UZHmmHZDm omumz 4
omomz 4

QAHmU 4 UZHmmHZDm
QAHEU 4 wszmHZDm ZOmMmm MBHm3 4
zommmm MHHm3 4

QAHSU 4 UZHmmHZDm
QAHSU 4 GZHmmHZDm Mm=H4m NZ
Mmmfi4m w:

mmuz .UHH mHan

78

 

¢H.

qN mN mm.

NH mo Nq.

qH.

0N qH wN.

«q Nm 00.

«H.

¥¢o on mm.

*Hq «o N «m.

¢H.

mm mN Hq.

mm mq mm.

«H.

mm mo MN.

qq NH mm.

«H.

*om oq co.

*mo «0 H do.

u .mea 2
saw:

mo.V um unmoHMchHm n .1

A4MMZDh 4 UZHszHH4
A4MMZDm 4 OZHQZMHH4 UHAomH4U 4
UHAomH4U 4

A4Mm23m 4 UZHQZMHH4
A4MMZDm 4 UZHQZMHH4 Hz4HmMHOMm 4
HZ4HmMHCMm 4

A4MMZDh 4 UanzmHH4
A4Mmzam 4 UZHQZMHH4 HmHZDZZCU 4
HmHZDZZOU 4

A4MMZDm 4 UZHQZMHH4
44mmzam 4 UZHQZMHH4 omomz 4
HOmwmz 4

A4MMZDm 4 UZHszHH4
A4MMZDm 4 OZHszHH4 ZOmMMm MHHm3 4
ZOmMmm MHHI3 4

44mmznm 4 UZHQZMHH4
44mm22m 4 DZHQZMHH4.MmmH4m ME
MmmH4m M2

Nmuz .mHH mHQMH

79

 

 

mm.H

*no m on HN.H
¥NN mH on m Nq.m
mme

*mm N mN «o.H
*mm 0N qN q ww.m
mm.H

Nm No N¢.H
mm Nm mN.N
mm.H

Ho N HN om‘H
*mm mH mm m qm.q
mm.H

«mN N mN 00.H
«mo MH mm m wH.m
mm.H

NN «o mm.H
*mH «N Ho m o¢.o

u .mmHo 2
saw:

mo.V um ucmoHMchHm u «.

wade GZHZMOm HHm4m 4 QZHmD
UDMD UZHZMOh HHm4m 4 oszD UHAO$H4O 4
UHAOEH4U 4

USMC UZHZMOh HHm4m 4 UZHmD
03mm UZHZMOh HHm4m 4 UZHmD Hz4HmMHomm 4
Hz4HmMHOmm 4

UDMD UZHzmom HHm4m 4 UZHmD
UDMQ UZHZMOh HHm4m 4 UZHmD HmHZDZZOU 4
HmHZDZZOU 4

Damn UZHzmom HHm4m 4 GZHmD
03mm GZHZMOK HHm4: 4 UZHmD omumz 4
OMUMZ 4

comm UZHZmom HHm4m 4 OZHmD
03mm UZHzMom HHm4: 4 oszb ZOmmmm MHHmz 4
ZOmMmm MHHZS 4

03mm OZHZmom HHm4m 4 OZHmD
undo Uszmom HHm4m 4 UZHmD MmmH4m NZ
mmma4m $2

m¢uz .wHH mHan

80

mean error by only-1.67 units over 72 assertions tested. How-
ever, this small reduction did result in five additional ac-
curate~predictions.

Given the relatively high correlation between the C
and CS, it seemed appropriate to determine if the C evaluation
score could be improved upon as a predictor of cognitive inter-
action by generating some constant correction factor. The
data were submitted to a simple linear regression technique
in order to obtain a constant factor by the least squares
method. This factor was then applied to the group mean evalu-
ation scores assigned to the C components in Tables 10a-f and
lla-f. Contrary to expectations, the total mean error of-pre-
diction actually increased slightly after the correction factor
was applied. It appeared, then, that the error was not a con-
stant factor, as might be assumed from a cursory examination
of the data.

It was further observed that in eight of the 12 tests
made, the mean evaluation score assigned to the C component
would have accurately predicted the cognitive interaction re-7
sulting in the assertions where MY FATHER served as the subject
(S) component. For these assertions, this rate of predictive
accuracy far exceeded both the congruity model, which failed
100 percent of the time, and the belief congruence model, which
failed 83 percent of the time.

As indicated in Table 9, the respondents showed the

greatest tendency to overassimilate towards the C component

my!”
L.

.l k” H?-

81

for the set of assertions where MY FATHER functioned as the
subject (S) component. It was apparent that the procedures
used in the belief congruence model were not adequately
accounting for this overassimilation.

If one considers the formula designed to account for
overassimilation in the belief congruence model, it becomes
apparent why this set of statements produced a number of
errors.

The formula states that:
dcs-dc + (v)dc [2]
where dC represents the effect of the first comparison process
(C versus S) and where (v)dc represents the additional effect
of the second comparison process (CS versus C), and where v
represents the extent to which the person attaches greater
importance to the CS than he does to C. When the person
judges CS to exceed C in importance, v will equal some coef-
ficient expressing the extent to which CS is perceived to
exceed C in importance, and dCS will exceed dC by the amount
(v)dc: When dC is positive, dCS will be more positive, and
when dC is negative dcs will be more negative.

The type of overassimilation referred to above is illus-
trated by the data taken from Table 10 for Assertion l, and
visualized in Figure 5. In this case, the subject, MY FATHER,
was evaluated +2.56 and the characterization, PROMOTING BIRTH
CONTROL,was evaluated +0.55. The combined CS was evaluated

-0.19 which is 0.74 scale units more negative than the

82

characterization, PROMOTING BIRTH CONTROL, and 2.75 scale
units more negative than the subject, MY FATHER.

When the evaluation scores were assigned by the belief
congruence procedure, the predicted evaluative score was +0.90.
This obviously is 1.09 scale units above the obtained CS
evaluation score of —0.l9. The discrepancy results at least
in part from the inability of the belief congruence model to
take into account the direction of the overassimilation
relative to the characterization component. Formula [2] above
posits that:

When a person judges CS to equal C in importance,
v=0 and d =d . When the person judges CS to
exceed C i3 importance, v will equal some coef—
ficient expressing the extent to which CS is
perceived to exceed C in importance, and d s
will exceed d by the amount (v)d . When 8

is positive, 8 will be more positive; whefi

d is negative fies will be more negative.

($2, p. 131).

Based on the data given for Assertion 1 above, it is
possible for a person to judge the CS as being more negative
than a positive dC or more positive than a negative dc' When
this type of assimilation is manifested,it is essential that
a valence be assigned to v to indicate the direction of the
importance of the db in relationship to the CS. Thus, in the
case illustrated in Figure 5, the v would have been assigned
a negative valence and subtracted from the positive dc evalu-

ation, rather than added to it, as is the case in the present

procedure.

83

.Hopoz
wocooumcou moHHom mSu hp How pmucsooo4 uoc coHumHHEHmmmum>o mo coHumuumsHHH oHumEonom 4 .m .me

.wuspmooum wcHumm mucouHOQEH mnu ou HmHucommm mH

ucocodaoo U onu ou pocmemm coHumsHm>o onu mo ucmpcodmch GOHumHHEHmmmuo>o mnu Ho coHuooqu
onu mo ucmsousmmoa m wchHmubo mo ponuoe m umnu consHocoo orb Ou mpmmH uomw mHnH .o>onw
pmumuumsHHH coHumHHEHmmmum>o mo memo ozu woummecmE poummu mcoHuuommm NN mnu mo m¢ mo Hmuou
m upsum ucommud mnu cH .ucmcodEoo o opp ou pocmemm cOHumsHm>o mnu mo cOHuomqu msu

cH othHm mH > ou pmcmemm mocme> onuoumnumwmesmmm uH mmsmomn o pom b mocHH somauon

moum ocu ucsooom OucH oxmu uoccmo pA>v + p n p mHosuow one .ucoGOQEoo U woman Am.+v
ka>HuHmom m Eoum coHuowqu Ao>HuHm0d mmev o>Humwmc m cH coHumHHEHmmmuo>o onu mucomoudmu

o pom b mocHH comauob moum msH .Homezou mHmHm ozHHozcmm Mmme4m w: .H coHuummm4 mo

memo mnu cH coHuoHpmum mo uouum Hmuou mnu ucwmoudmu o pom m mcHH cmo3uob muHcs onom mnH

 

 

 

 

mHmom
Ml
NI
HI
mu
H3 ANS 0 3.25328 Hofizoo Ea; 025985 swag >2 u . e
0
3V. . . . . . . .. 33 H8 40%on Ea: oonzomm u e
mo
H3 95 H+ 930383 85on FEB quHozSE mmmafi w: u e
N+
m
. 3N3 H8 mag: E u u

m+

84

One possible way to correct this weakness would be to
have the reSpondents indicate the direction of their feelings
on the importance rating form. Figure 6 illustrates how the
present procedure could be altered so that, in effect, it
would determine the valence to be assigned to the v, regard-
less of the dc evaluation score. If the respondent feels
more negatively towards CS than towards C, v is negative, if.
he feels more positively towards the CS than towards the C,

v is positive. In the first instance, the negative v coef-
ficient multiplied by the positive dc would be subtracted from
dc thus movingthe prediction towards the less positive position,
and thus accounting for at least part of the type of over-
assimilation illustrated above.

Based on the results of this investigation, certain
alterations in the construction of the Importance Rating Form
used with the belief congruence model are suggested in Figure
6. Included in this suggested form is a question for obtain-
ing measurements of the direction of the respondents' feelings
toward the characterization independent of their feelings
toward the characterization in isolation. It appears that
the suggested revision illustrated in Figure 6 might also re-
duce the number of subjects eliminated because they could not
Operationally respond to the importance rating instrument.

Overall the present study provides substantial support
for the basic theoretical rationale underlying the belief con-

gruence principle. It is evident that the congruity model,

1.

disapprove : : :

85

How do you feel about MY FATHER PROMOTING BIRTH CONTROL?

Strongly Strongly
approve

 

2 3 4 5

In rating MY FATHER PROMOTING BIRTH CONTROL the way you
did, indicate on the scale below how important, that is
how much weight did the words PROMOTING BIRTH CONTROL
and MY FATHER have on how you felt about MY FATHER PRO-
MOTING BIRTH CONTROL? Indicate just how much influence
you feel each part of the statement had on how you
actually rated MY FATHER PROMOTING BIRTH CONTROL.

MY FATHER

o 10 20 30 4o 50 60 7o 80 90 100 Percent

 

PROMOTING BIRTH CONTROL

:T:W:TU:—§U:70:_50:—60:_76._§'6:_§U.mzpercent

Please check: Add the two percentages together. Do they

total to 100 percent? If not, readjust your
ratings so they total 100 percent when added
together.

If you indicated above that PROMOTING BIRTH CONTROL
completely (100%) determined the way you rated MY FATHER
PROMOTING BIRTH CONTROL, go to question 3 below. Other—
wise continue to the next rating sheet.

Since you felt that the words, PROMOTING BIRTH CONTROL,
completely determined how you feel about MY FATHER PRO-
MOTING BIRTH CONTROL, do you feel even more strongly
towards MY FATHER PROMOTING BIRTH CONTROL than you feel
about other people PROMOTING BIRTH CONTROL?

a. Yes___my feelings about MY FATHER PROMOTING BIRTH
CONTROL are even stronger than my feelings
about other pe0ple PROMOTING BIRTH CONTROL.
(complete c and d below)

b. No ___my feelings about MY FATHER PROMOTING BIRTH
CONTROL and PROMOTING BIRTH CONTROL are about
of equal strength. (omit c and d below and
go to next rating sheet)

c. Since you do feel more strongly towards MY FATHER
PROMOTING BIRTH CONTROL than you feel towards other

86

people PROMOTING BIRTH CONTROL, do you feel more
favorable or less favorable ' towards MY FATHER
PROMOTING BIRTH CONTROL?

d. How much more strongly do you feel?

Slightly Much,much
stronger: : : : : : : : : : :more
'1 2 3 4 5 6 7 8 9 10 strongly

Fig. 6 Recommended Revision of the Relative Importance
Rating Form to obtain the direction of the over-
assimilation in relationship to the character-
ization component in the combined CS configuration.

 

87

which assumes that cognitive interaction resulting from word
combinations can be predicted from an evaluation of the words
in isolation, does not predict as well as the more gestalt-
like belief congruence model. It is obvious that any model
which purports to predict cognitive interactions of the type
investigated in this study must allow for.various degrees of
assimilation and overassimilation much in the manner that it
is manifested by the belief congruence model.

Although the Rokeach and Rothman (12) work and the. V

 

present study provide considerable empirical evidence in sup-
port of the dynamics of the processes involved in cognitive
interaction under the belief congruence model, the model
appears to need further procedural refinements to improve its
accuracy. In particular, procedures for dealing with the
overassimilation phenomenon should be used in future tests of

the model.

88

CHAPTER 4
CONCLUSIONS, DISCUSSION, AND IMPLICATIONS
FOR FUTURE RESEARCH
The purpose of the present investigation was to test

empirically the relative accuracy and reliability of the con-

gruity and belief congruence models in predicting cognitive
interaction resulting from certain word combinations. The
investigation was guided by two broad objectives and a series

of three working hypotheses designed to make specific the

 

1r

empirical data and their relationship to the conclusions
drawn from the analysis.

It is believed that when considered within the limi—-
tations of the characteristics of the respondents and word’
combinations employed in this investigation, the conclusions.
presented in this chapter are reasonable in light of the data
presented in Chapter 3.

Conclusions Relative to Accuracy
of Prediction of Cognitive~1nteraction.

59 the Congruity and BeliefCCOngruence
Models

 

 

 

 

(1) It is concluded on the basis of the present investi-
gation that the belief congruence model is significantly su-
perior to the congruity model as a predictor of cognitive
interaction resulting from certain word combinations, both in
terms of total mean error of prediction per word combination
tested and in total number of word combinations predicted

accurately.

89

These findings and conclusions are consistent with those
reported by Rokeach and Rothman (12) in their 1965 study which
used a different set of word combinations and a different type
of respondents. The fact that comparable results were obtained
under the two different situations gives additional support to

the general superiority of the belief congruence procedure as

"f
, ~w

a more accurate predictor of cognitive interaction.
It would appear that the underlying theoretical ration-

ale of the belief congruence principle--which takes into ac—

a,
0-'.'|? . ‘

count various degrees of assimilation and overassimilation
towards the characterization component of the combined CS
configuration--is the major reason for the increased accuracy
of prediction manifested by the belief congruence model. How-
ever, as was pointed out above, the present belief congruence
formulation apparently does not deal adequately with at least
one type of overassimilation manifested in the present
investigation.

(2) It is further concluded that whether the subject
(S) or the characterization (C) is varied in a series of com-
bined CS configurations has no effect on the mean evaluation
scores assigned to the CS, nor does it affect the predictive
outcome of the two models. In other words, it makes little
difference to the respondent whether or not he is responding
to a series of assertions in which the subject (S) of the
assertion is the same or varied in relationship to the charac-
terization (C) applied to the subject (S) of the assertion (CS)

being tested.

 

90

This particular finding may be of methodological
utility in future research dealing with cognitive interaction.
In the series being assessed for evaluative meaning, it appears
that the subject is responding to each assertion rather inde-
pendently of the others. The fact that subjects do apparently
respond to a variety of word combinations without biasing
interaction effect is of potential methodological importance.
Conclusion Relative to the Reliability.

of Prediction of Cognitive Interaction by_
the Congruityand Belief Congruence Models

 

(1) It is concluded that the belief congruence model
is significantly more reliable than the congruity model in
predicting cognitive interaction resulting from certain word
combinations.

This conclusion is based on the results obtained when
identical word combinations were tested across independent,
but equivalent test groups. Although the total number of
assertions predicted accurately by the congruity and belief
congruence models varied widely (19 vs. 48 out of 72 tested)
the reliability test showed that only five of the word com-
binations were common between the two test groups for the
congruity model, while 20 of the 24 word combinations pre-
dicted accurately in one test group were also predicted
accurately in the second test group by the belief congruence
model. This result indicates that the belief congruence form-
ulation represents a more reliable explanation of the under-

lying COgnitive interaction processes that are apparently

91

common to different but equivalent groups of subjects who are

asked to assess the same word combinations.

Conclusions Relative to the Theoretical
Issues Underlyifig the Cognitive Interaction
Processes and the Operational Dynamics of.thev
Congruity and Belief Congruence MOdels

 

 

(1) It is concluded that the characteristics of the F
component parts of the combined CS configuration are sig-

nificantly related to the ability of either model to

 

accurately predict cognitive interaction.

This conclusion is based on the results which show

 

that the models could more accurately predict the cognitive
interaction outcomes on some assertions than on others used
in this study. Two subject (S) components were particularly
difficult for both models to predict accurately; namely, MY
FATHER and A WHITE PERSON. The characterization (C), PROf
MOTING BIRTH CONTROL, was found to contribute to prediction
error most often in both models. In other words, almost
without exception, any word combination that contained the
subjects (S's) MY FATHER or A WHITE PERSON in association
with any of the six characterizations was likely to result
in cognitive interaction different than the outcome predicted
by the models. This same relationship was also noted in the
case of the characterization (C) PROMOTING BIRTH CONTROL.

(2) It is concluded that at least for certain word
combinations tested the major reason for the models' inability

to predict cognitive interaction accurately is directly related

92

to a Specific type of overassimilation that cannot be adequately
accomodated by the present procedures.

Since, from an operational vieWpoint, the present pro-
cedures are basically additive, neither model can adequately
deal with overassimilation in which the respondent assigns an
evaluative meaning score to the CS configuration that is less
positive than a positively rated characterization (C),.or less
negative than a negatively rated characterization (C). The
data clearly indicate that in many of the word combinations
(48 of 72) the mean CS evaluation scores assigned were either
less positive than a positive rating assigned to the character-
ization in isolation, or more positive than the negative
evaluative scores assigned to the characterization in isolation.

This situation makes it essential that the direction of
the overassimilation be obtained independently of the evaluative
rating assigned to the characterization (C) in isolation. The
possibility of using such a corrective measure in conjunction
with the belief congruence procedure is discussed on Pages 82
and 84 in Chapter 3.

(3) A high positive correlation between the character-
ization (C) evaluative scores and the combined CS configuration
evaluative scores leads to the conclusion that the character-
ization evaluative scores alone are reasonably good predictors

of cognitive interaction between certain word combinations.

 

93

In the present investigation, a significant correlation
of .76 was found between the C and CS evaluation scores. Al-
though an attempt to generate a corrective constant by linear
regression methods failed to increase the predictive accuracy
of the characterization rating alone, the relatively high
relationship is of interest and merits more investigation. F

In general, what this finding indicates is that when I 1
compared with the characterization (C) component, the sub-

ject (S) of the word combination is of relatively little

 

value in determining cognitive interaction effects. The- ?
fact that the subject (S) has a very low correlation (r=.13)
with the combined CS evaluation.scores further supports

these apparent differences in the importance of the two com-
ponents. Evidently, it makes little difference what subject
is associated with a particular issue, activity, or event as
determined by the characterization (C) in the combined CS
configuration; instead, the relative influence of the
characterization evaluation will apparently dominate the
interaction effect. Although the exact correlations of the
Rokeach and Rothman study were not available to the author,
it would appear that comparable relationships were also mani-

fested in their data.

Implications for Future Research
The present investigation extended the results of the

Rokeach and Rothman study in several significant ways.

94

In addition to replicating their results, the present study:
(1) tested the relative reliability of the two models by
employing an equal group technique; (2) strengthened the
research design by obtaining all relevant measurements on
the same subjects3; (3) recommended a possible measurement
technique that will deal with one type of overassimilation
which the present belief congruence procedures do not take
into account; (4) defined more specifically the relative
influence of the subject (S) and characterization (C) com-
ponents on cognitive interaction resulting from certain word
combinations, and (5) tested a series of word combinations
of the type not generally tested in prior word combination
studies; i.e., combinations in which the c0pula itself is
capable of eliciting evaluative meaning.

The study suggests that cognitive interaction, even
of the least complex types, remains a complicated problem.
In terms of predictive accuracy, the various attempts to
quantify such behavior have left much to be desired. Most
certainly one must give credit to the originators of the
congruity and belief congruence principles for the inroads
and new insight they have contributed to the area of cognitive

interaction research.

 

3The Rokeach and Rothman study design employed an equi-

valent group technique in which one group of subjects assigned
evaluative ratings to the assertions and their component parts,
while another group of subjects rated the assertions and their
component parts for relative importance. The present study
obtained both measurements on the same subjects.

‘ ”.1"?
r

at,
'54.“ J -

95

Both the prior research reported in this study and the
present investigation itself have been devoted to only a very
small aspect of the broader dimensions of cognitive intere
action. However, it seems appropriate to suggest that if the
congruity or belief congruence principles can be empirically

supported in the narrow aspects of the cognitive research

‘2' ‘~ In .

area, it might well be the first step toward generalization

\Jibnmw’.‘

of the principles.

The findings of the present study indicate that P

 

additional research is needed to reformulate the belief con- L
gruence model as an Operational, predictive model. The
present operational procedures appear to rely on the presence
at the time of measurement of the cognitive interaction event
to be predicted. The present investigation offers no specific
suggestions or leads as to how this might be accomplished,
except to point out that there is a relatively high correlation
between the characterization (C) component of the combined CS
configuration and the actual cognitive interaction effect
assigned by the 619 respondents employed in the study. This
finding suggests the possibility of future research to deter-
mine if the close relationship exists across a variety of
word combinations and populations, and how the relationship
may manifest itself under different test situations.

The operational value of the close relationship observed
between the C and CS evaluative scores lies in the fact that

it is relatively simple to obtain measurement on a single

96

component as compared to the total Operational procedure re-
quired by the belief congruence model. The data of this
study reveal that the C evaluative scores closely approximate
the predictive outcome of the entire belief congruence pro-
cedure, and thus appear to give a reasonably good estimate of
cognitive interaction effects. Based on this finding, it
seems appropriate to suggest that future research could well
be directed toward determining how generalized this relation-
ship may be across different types of word combinations, and
to explore ways to increase the methodological utility of
this single measurement as an estimate of cognitive inter—
action.

In addition, the present investigation suggests the
possibility of an empirical study to test the corrective
measurement technique suggested to help account for a certain
type of overassimilation that the present belief congruence
procedure does not take into account. This technique is dis-
ecussed in detail on pages 84 Rte 87 in Chapter 3. Does the
addition of the direction measurement of the overassimilation
on the C vs. CS comparison increase the predictive accuracy
of the belief congruence model? Questions such as this one

are fair game for future research.

10.

11.

12.

13.

14.

97

BIBLIOGRAPHY

 

Asch, S. E. Social Psychology, New York: Prentice—
Hall, 1952.

 

Edwards, A. L. Experimental Design in Psychological
Research, _New York: Holt, 1960.

1
Festinger, L.” A Theory of Cognitive Dissonance.
Standard, California. Stanford Un1versity Press, 1957.

Games, P. A.) and Klare, G. R. Elemtary_Stati§tics.w
New York: McGraw-Hill, 1967.

 

Heider, F. The Psychology of Interpersonal Relations._
New York:. John Wiley and Sons, 1958.

 

Lewin, K. Field theory and experiment in social psy-
chology. In Field theory in social science, Harper,
1951.

 

Newcomb, T. An Approach to the Study of Communication
Acts. Psychological Review, 1958, 393-404.

 

Osgood, C. E. and Tannenbaum, P. H. The Principle of
Congruity in the Prediction of Attitude Change. Psy-
chological Review, 1955, 62, 42-55.

 

Osgood, C. E. Suci, G. J., and Tannenbaum, P. H.
The Measurement of Meaning. ~Urbana: _University of
Illinois Press, 1957.

 

Rokeach, M., The Nature of Attitudes. In International
Encyclopedia of Social Science. New York: McMillan, I965.

 

 

Rokeach, M. The Open and Closed Mind. New York: Basic
Books, 1960.

 

Rokeach, M. and Rothman, G. The Principle of Belief
Congruency and the Congruity Principle as Models of Cog-
nitive Interaction. Psychological Review, 1965, 72,
128-142.

 

Rokeach, M. The Three Christs of Ypsilanti: a Psycholo-
gical Study. New York: KnOpf, 1964.

 

 

 

Sherif, C., Sherif, M., and Nebergall, R. Attitude and
Attitude Change. Philadelphia: Saunders, 1965.

 

 

15.

16.

Walker, H. H. and Lem,J.
York: Holt and Company,

Fishbein, M. and Hunter,
in Attitude Organization
Social Psychology, 1964,

98

Statistical Inference.“ New
1953.

R.. Summation Versus Balance
and Change. Journal of Abnormal
69, 505-510.

APPENDICES

V '_” '3 fl“

 

APPENDIX A

'1'? N0. ’
“-2 - 305*: to.
ma 1 c3-7

ASTUUYOPWORDMEMING

The purpose of the study is to explore ways people think about individual words
and codinetions of words within a single statement. In PART I of this POEM you will
be presented a series of ideas made up of either two words or a series of words (a
phrase). We want you to consider each of these ideas separately and to judge thee
against a series of descriptive scales. We want you to make your judgaents on the
basis of what these ideas mean to yg_. There are no right or qu93 answers. it is
your peggonal 92 Mi that you are to use to make your Judgments.
bel Under each idea to be judged you‘will find a set of three scales like the one

ow:

MY FRIEND
IMPORTANT : : 2 : : 2 UNINPORIANT
Here's how it works:

First look at the adjectives at each end of the scale and decide which one fits
best. If you felt that MY FRIEND was very IMPORTANT, you would put an ”X‘ in the
space closest to IMPORTANT. If you felt that MY FRIEND was very UHIMPOPTANT, you
would place the "X‘ in the space closest to UHIMIORTANT.

The middle space along the scale stands for NO OPINION. and the other spaces
seen stronger and stronger feelings as they get closer to the ends of the scale.

_ You eight think of the spaces as being labeled like this:

MY FRIEND
very quite a little neutral a little quite very

.-

* ..m/W \. )1 ..

IMPORTANT: (1) Consider each idea separately, and check the three scales under it
. and nova to the next. Do not try to reaenber how you check.sinilar
items earlier in the test. Make each idea a separate and independent

jfimts

 

(2) Place your checkraark ("x ) in the middle of the space, not over the
colone or near the boundaries:

This Int This
: x : : : I : :

 

 

(3) be sure to check all three scales for each idea...do not onit any.

(4) Never put more than‘ggg,check~nark on any single scale.

If any of the instructions were not clear please request the person who handed
out the teat FORM to help you. If not, go ahead and complete PART I of this FORM.
After you finish PART I, please wait for further instructions before turning to
PART II. You‘will have about l0 minutes to complete PART l.

95?

12-1

ADMIRABLE

WORTHLESS

GOOD

12-2

ADMIRABLE

VALUABLE

BAD

USING

O. I.
00

1'00

HABIT FORMING DRUG

MY FATHER

DEPLORABLE c8
VALUABfiE c9

BAD clO

DEPLORABLE C11
WORTHLESS 012

GOOD 013

12-3

DEPLORABLE

VALUABLE

GOOD

12-4

ADMIRABLE

A WHITE

10.1.

PERSON

ADMIRABLE-

 

WORTHLESS

 

'BAD

 

A NEGRO

DEPLORABLE

 

WORTHLESS

VALUABLE

 

GOOD

BAD

 

C14
c15
016

C17
C18
019

102

 

 

 

 

 

12-5
A COMMUNIST
ADMIRABLE : : : : : : DEPLORABLE
VALUABLE : : : : : : WORTHLESS
BAD : : : : : : GOOD
12-6
A PROTESTANT
DEPLORABLE : : : : : : ADMIRABLE
VALUABLE : : : : : : WORTHLESS
GOOD : : : : : : BAD

 

020
c21

c22

023
c24

c25

 

103.

 

 

 

 

 

 

12-7
A CATHOLIC
ADMIRABLE .= : : : . : DEPLORABLE 026 ,
WORTHLESS : : : : : : VALUABLE 027
GOOD : : : : : : BAD 028
12-8
MY FATHER USING A HABIT FORMING DRUG
ADMIRABLE : ° : : : : DEPLORABLE 029 7
VALUABLE : : : : : : FWORTHLESS 030
BAD : : : : : : GOOD 031

 

103

 

 

 

 

 

 

 

12-9
A WHITE PERSON USING A HABIT FORMING DRUG
DEPLORABLE : : . : : : BADMIRABLE 032
VALUABLE : _3 : : : : WORTHLESS 033 ‘
GOOD : : : : : : BAD 034 i
!
12-10
A NEGRO USING A HABIT FORMING DRUG
ADMIRABLE : : . : : : DEPLORABLE 035
WORTHLESS : : : : : : VALUABLE 036
GOOD : : : : : : BAD C37

 

105

 

 

 

 

 

12-11
A COMMUNIST USING A HABIT FORMING DRUG
ADMIRABLE : : : : : : ‘DEPLORABLE
VALUABLE : : : : : : WORTHLESS
BAD : : : : : : _+GOOD
12-12
A PROTESTANT USING A HABIT FORMING DRUG
DEPLORABLE : : : : : -ADMIRABLE
VALUABLE : : : : : : WORTHLESS
GOOD : : : : : : BAD

 

038
039
040

041
042
043

106

 

 

12-13
A CATHOLIC USING A HABIT FORMING DRUG
ADMIRABLE : : : : : : ‘DEPLORABLE 044
WORTHLESS : : : : : : VALUABLE 045
GOOD : : : : : BAD 046

 

.10?
PART II-

In PART I you rated a series of statements to indicate
how you felt about their meaning. PART II is somewhat more
difficult to understand and do. It will take your careful
attention to the instructions.

At the tOp of the next six pages, the same statements
are given again that you just completed rating in PART I.
This time, however, we want you to judge the statements in a
somewhat different way:

STEP I

First, we want you to consider each statement again
and indicate on the scale below it how strong you approve or
disapprove of the idea. Just place a check-mark a ove t e
number on the scale that you feel best describes how you feel
about the idea. If you check "1" that means you strongly
disapprove and if you check "7" that means you strongly a rove.
The numbers between "1" and "7" may be used to indicate degrees
between strongly disapprove and strongly approve. Again he
sure to cheek each scale, and check each scale only once.

 

 

 

 

STEP II

After you have indicated how you approve or disapprove
of the statement, move immediately to number 2 on the same page.
To help you complete this part of the rating, we think it will
be helpful if you will think of each statement as having two
parts. The first part is made up of the first two words and
the second part is made up of_the remaining words in the
statement.

 

 

It will become more clear when you read the first para-
graph for number 2 as to just exactly what kind of a rating
you are to make. It may be necessary for you to re-read this
first paragraph two or three times to get the full understand-
ing of how to proceed with a. b. or 0. parts in number 2.

Now turn to the next page. Complete each page in turn.
Please consider each statement separately, even though they
seem to be related. Do not look back to your former ratings
as this might confuse you on this part.

Thank you very much for your cooperation in this study.

32-11;

1.

2.

108

How do you feel about: MT FATHER mm A HABIT towns: URN

Brannon: :___:____:___:___:__:__8 813083“ 147‘-
DISAPPROVB‘1234567APPROVI °

In rating MY PATBR 03m: A HABIT roman DRUG the way you did, how
important, that is, how such weight did the words 03m: A “311' Pm .
DRUG have in determining your rating of the statement, )4! BATTER USDA}

A HABIT FORMIm DRUG? Did you feel about the combination HY FATHER

081m A HABIT PORNIM DRUG the same way you felt about the words USIM

A HABIT PORHIK: DRUG or the sane way you felt about HT FATHER? Rstinste
how such weight USING A HABIT rm new and H! FATHER had in determining
the way you actually rated the combination MY FATHER USDA} A HABIT PORMIDB

a. My feelings about mm: A nun roam sens m (1001.) determined
the way 1 rated )1! rams usms A man- roenms mm.c1uck the one

 

that applies. .
TBS Now go to question b. m answer question c. 0&3-
so Now go to question c. m answer question b.

 

b. In fact, ny feelings about HT FATHER UBING A HABIT FORMING DRUG are
even nore extreme than my feelings about other people 03m: A HABIT
mus DRUG.

Answer those that apply to you:
'0, my feelings about lfl FATHER USING A HABIT MG 181135

and. USING A HABIT MG mus are about of equal strength. 0:!

____Tes, lay feelings about HY FATHER USDA: A HABIT mus
URN are even stronger than my feelings about other
. people USIIB A HABIT mun DRUG.

How nuch stronger? Check the m that would be closest to
your best guess:

_____Blightly stronger (ll stronger)?
Quite a bit stronger (501 stronger)? 050..—
___Huch, such stronger (1001 stronger)!

c. Hy feelings about USING A HABIT PORHIm DRUG did not completely deter-ins
ay rating of the, combination MY PAM usnas A HABIT FORMING DRUG.
Indicate on the scales below how such you think each part of the statement
influenced you: .

Assume that each scale equals 100%. Check the point on each scale which
you think best describes the amount of influence-that part of the state-
nent had on your rating of the statement on the whole. em: that
the percentage checked on each scale when added together eust be equal

toLQQpercent.
HTPATHBR::-:::::::: 051-—

. 0" 1'0 "2'0 '35 40 so'so'iB-sb' 901 '
munimnromneus
_::___:__::::::-:: 052

o 10 20 3076-5655676'3'6—9336

W: Add the two percentages together that you have
checked. Do they equal 100% If not, readjust
your rating so that they sun to 1002. .

--E

 

2-15

1.

l()9

How do you feel about A wan-s mason usmc A mum roams um

STROMLY : : : : : : 45mm“
DISAPPROVE l 2 3 l: 5 6 7 APPROVE.

053..

In rating A warn: mason usms A mu tannins onus the way you did, how
important, that is, how much weight did the words mm A HABIT Forums ,
MUG have in determining your rating of the statement A WHITE PERSON
USING A HABIT FORMIhG DRUG? Did you feel about the combination A WHITE
PERSON USIm A HABIT FORMING DRUG the same way you felt about USING A
HABIT FORMING DRUG or the same way you felt about A WHITE PERSON?
Estimate how such weight USING A HABIT roenm: DRUG and A WHITE PERSON
had in determining the way you actually rated the combination A WHITE
PERSON USIM A HABIT PORMIM; DEW.

a. my feelings about usnn A man- rommc onus mm (1001)
determined the way 1 rated A warn: mason USING A man rename new.
(Check the one that applies)

TES Now go to question b. Do not answer question c. czh_ __
I
NO ___'_Now go to question c. Do not answer question b.

b. In fact, my feelings about A WHITE PERSON USING A HABIT PORHIm DRUG
are even more extreme than my feelings about other people name A
HABIT resume DRUG. Answer those that apply to you.

No. my feelings about A warm msou name A nAsrr ms mus
and USING A HABIT FORMING mm: are of about equal strength .

- c55_
Yes, my feelings about A WHITE PERSON name A HABIT mamas DRUG
are even stronger than my feelings about other people new A HABIT

 

resume DEM.

How such stronger? Check the m that would be closest to your .
best guess.
Slightly stronger (17. stronger)?

l

Quite a bit stronger (501 stronger)? (:56...
Much, much stronger (1001 stronger)?

c.‘ Hy feelings about 081m A HABIT roman DRUG did not completely
determine my rating of the combination A HHITE PERSON 081m A HABIT
mm: BRIE. Indicate on the scales below how such you think each part
of the statement influenced you:

Assue that each scale equals 100%. Check the point on each scale
which you think best describes the amount of influence that part of
the statement had on your rating of the statement on the whole.
Remember that the percentage checked on each scale when added together
met be equal to 193 percent.

A WHITE PERSON _: : : : : : : : :_ : :
. 0 lo 20 3O 40 50 60
name A HABIT roman: new ' °57 -—
’: ' : : : : : : :
0 lo 20 30 40 $0 60 70 SO 90 100
S __
Please Check: Add the two percentages together that you have checked.

Do they equal 1007.? If not readjust your rating so that
those muse sub 1M7 .

o“:—

A '.

 

arr-.1

 

110

 

12~16
1. How do you feel about: A NEGRO USING A HABIT FORMING DRUG
STRONGLY : : : ' : : :- STRONGLY"
DISAPPROVE 1 2 3 I. , 5 6 7 APPROVE 65‘9_

2. In rating A NEGRO USING A HABIT FORMING DRUG the way you did, how important, that
is, how much weight did the words USING A HABIT FORMING DRUG have in determining
your rating of the statement, A NEGRO USING A HABIT FORMING DRUG? Did you feel
about the combination A NEGRO USING A HABIT FORMING DRUG the same way you felt
about the words USING A HABIT FORMING DRUG or the same way you felt about A NEGRO?
Estimate how much weight USING A HABIT FORMING DRUG and A NEGRO had in determining
the way you actually rated the combination A NEGRO USING A HABIT FORMING DRUG.

b.

Hy feelings about USING A HABIT FORMING DRUG cmletely (1002) determined
the way I rated A NEGRO USING A HABIT PORNINGDRUG. (Check the one that
applies.) .

YES Now go to question b. Do not answer question ‘e. 060,,
NO 3 Now go to question c. Do not answer question b.

In fact, my feelings about A NEGRO USING A HABIT FORMING DRUG are even
more extreme than my feelings about other people USING A HABIT BURNING
DRUG. Answer those that apply to you.

No, my feelings about A NECRO USING A HABIT FORMING MUG'and

.*-

USING A HABIT FORMING are'of about equal strength.
obl—

Yes, my feelings about A NEGRO USING A HABIT FORMING DRUG are even
stronger than my feelings about other people USING A HABIT BURNING
RUG.

‘ “How much stronger? Check the _9_n_e;_ that would be closest to your best
guess.

Slightly stronger (l! stronger)?

 

Quite a bit stronger (502 stronger)? . cE

Much, much stronger (100: stronger)?

My feelings about USING A HABIT FORMING DRUG did not completely determine
my rating of the comgination A NEGRO USING A HABIT FORMING DRUG. Indicate

. on the scales below how such you think-each part of the statement

influenced you. ‘

Assume that each scale equals 100:. Check the point on each scale which
you think best .describes the amount of influence that part of the state-
ment had on your rating of the statement on the whole. Rem-ber that
the percentage checked on each scale when added together must be equal
to 1_0_9_ percent.

ANEGRO-::::::::: c63-

._~._-O—m-_~O._m

o '10 20 so 40 so so 70 so so 100
usmo A man roams DRUG ’ ‘
...3....i.._.=-._.3.__3-...3._...3.._.3___.3-..._3-_. - 06h
Olozoaoeosobomsosoioo .
W: Add the two percentages together that you have checked.

Do they equal 100:? If not readjust your rating so that .
EDA, IUD to wae

.e I
I
I

 

12-17
1.

2.

111

How do you feel about: A COPMUNIST USING A HABIT POFMING DRUG

DISAPPROVE APPROVE

STRONGLY : : : : : :_ STRONGLY

”065.

In rating A COMMUNIST USING A HABIT FORMING DRUG the way you did, how important,
that is, how much weight did the words USING A HABIT FORMING DRUG have in deter-
mining your rating of the statement, A COMMUNIST USING A HABIT FORMING DRUG?

Did you feel about the combination A COIBIUNIST USING A HABIT FORMING DRUG the

same way you felt about the words USING A HABIT FORMING DRUG or the same way

you felt about A COMMUNIST? Bstinate how much weight USING A HABIT FORMING

DRUG and A COIMUNIST had in determining the way you actually rated the combination
A COMMUNIST USING A HABIT FORMING DRUG.

My feelings about USING A HABIT FORMING DRUG mletely (1002) determined
the way I rated A (IMMUNIST USING A HABIT FORMING DRUG. Check the one
that applies.

 

YES Now go to question b.‘ Do not answer question c. 067.—
NO Now go to question c. 90 not answer question b.

b.In fact, my feelings about A COMMUNIST USING A MBIT FORMING DRUG are

c.

even more extrmne than my feelings about other peeple USING A HABIT
FORMING DRUG. Answer those that apply to you.

No, my feelings about A COMMUNIST USING A HABIT FCRMING vans and
USING A HABIT FORMING DRUG are of about equal strength. 068,

_Tea, my feelings about A COMMUNIST USING A HABIT FORMING DRUG are
even .tronger than my feelings about other people USING A HABIT
FORMING DRUG.

How much stronger? Check the 99; that would be closest to your
best guess.

Slightly stronger (ll stronger)? 062—
Quite a bit stronger (501 stronger)?
Much, much stronger (1002 stronger)?

My feelings about USING A HABIT FORMING DRUG did not completely deter-
mine my rating of the combination A COMMUNIST USING A HABIT FORMING
DRUG. Indicate,» the scales below how such you think each part of
the statement influenced you.

Assume that each scale equals 1001. ' Check the point on each scale which

you think best describes the amoxmt of influmce that part of the state-

ment had on your rating of the statement on the whole. Remember that the

percentage checked on each scale when added together must be equal to

m percent.

A OOlBRJNIST ___:_____ ' ___:___: :-_:____:____: :_ : °7°____
o 10 To 30% so 60 midso‘W .

USING A HABIT FORMING DRUG

___:___:____:___ . -__!._3..._.
010 20304050607080” 100 071...

LIN: Add the two percentages together that you have checked.
Do they equal 1002? If not readjust your rating so
that they sum to 1001.

112

12-18
' I. How do you feel about: A pmzsnmr USING A man rommc onus
STRONGLY : z 3.. _: :-___:_ STRONGLY Card
DISAPPROV'Bl 2 3 4, 5 6 7mm , n
o72__

2. In rating A PROTESTANT USING A HABIT FORMING DRUG the way you did, how important,
that is , how much weight did the words USING A HABIT FORMING DRUG have in deter-
mining your rating of the statement, A PROTESTANT USING A HABIT FORMING DRUG?

Did you feel about the combination A PROTESTANT USING A HABIT FORMING DRUG the
same way you felt about the words USING A HABIT FORMING DRUG or the same way

you felt about A PROTESTANT? Estimate how much weight USING A HABIT FORMING

DRUG and A PROTESTANT had in determining the way you actually rated the combination
A PROTESTANT USING A HABIT FORMING DRUG.

a. My feelings about USING A HABIT some onus toggletelz (1002) determined
the way I rated A PROTESTANT USING A HABIT FORMING DRUG. Check the one
that applies.

YES Now go to question b. Do not answer question c. 073...
NO Now go to question c. Do not answer question b.
b. In fact, my feelings about A PROTESTANT USING A HABIT FORMING DRUG are

even more extreme than my feelings about other people USING A HABIT
FORMING DRUG. Answer these that apply to you. .

 

_______No,my; feelings about A PROTESTANT USING A HABIT FORMING mus
and USING A HABIT FORMING DRUG are of about equal strength. 07!

Yes. my feelings about A PROTESTANT usn'xc A HABIT roams moo are
even stronger than my feelings about other people USING A HABIT

poms DRUG.
How‘much stronger? Check the 995 that would be closest to your
best guess. A
Slightly stronger (12 stronger)! . 075_

Quite a bit stronger (502 stronger)?

Much, much stronger (1002 stronger)?

 

c. My feelings about USING A HABIT FORMING DRUG did not completely determine
w rating of the combination A PROTESTANT USING A HABIT FORMING DRUG.
' Indicate on the scales below how such you think each part of the statement
influenced you.

Assume that each scale equals 1002. Check the point on each scale which
you think best describes the scent of influence that part of the state-
ment had on your rating of the statement on the whole. Mr that
the percentage checked on each scale when added together must be equal
to m percent.

A PROTITANT : s 3 z 076....

° USING A HABIT FORMING DRUG
____:____:__:___:___:___:___:___x__:___x___ 078“
0102030‘05060108090100

[lgse Chsc : Add the two percentages together that you have checked.
Do they equal 1002? If not readjust your rating so that
they sum to 1002. 079 1

113

 

12 .
:19
1. How do you feel about: A CATHOLIC USING A HABIT FORMING DRUG Card
I 2
STRONGLY : : : : : : STRONGLY
DISAPPROVE l 2 3 4 5 6 7 APPROVE c8

-..

2. In rating A CATHOLIC USING A HABIT FORMING DRUG the way you did, how important,
that is, how much weight did the words USING A HABIT FORMING DRUG have in deter-
mining your rating of the statement, A CATHOLIC USING A HABIT FORMING DRUG? Did
you feel about the combination A CATHOLIC USING A HABIT FORMING DRUG the same
way you felt about the words USING A HABIT FORMING DRUG or the same way you felt
about A CATHOLIC? Estimate how much weight USING A.HABIT FORMING DRUG and
A CATHOLIC had in determining the way you actually rated the combination A
CATHOLIC USING A HABIT FORMING DRUG.

a. My feelings about USING A HABIT FORMING DRUG templately (1002) determined
the way I rated A CATHOLIC USING A HABIT FORMING DRUG. Check the one
that applies.

YES Now go to question b. Do not answer question c. c9

NO Now go to question c. Do nq£_answer question b.

b. In fact, my feelings about A CATHOLIC USING A.HABIT FORMING DRUG are even
more extreme than my feelings about other peeple USING A HABIT FORMING
DRUG. Answer those that apply to you.

_.__._R0. my feelings about A CATHOLIC USING A HABIT roams mm: and
USING A HABIT FORMING DRUG are about of equal strength. clO

_;Ies, my feelings about A CATHOLIC USING A HABIT FORMING DRUG are

even stronger than my feelings about other people USING A HABIT
FORMING DRUG.

How much stronger? Check the one that would be closest to your
best guess.

Slightly stronger (12 stronger)? cll

Quite a bit stronger (502 stronger)?
Much, much stronger (1002 stronger)?

c. my feelings about USING A HABIT FORMING DRUG did not completely deter-
mine my rating of the combination A CATHOLIC USING A HABIT FORMING
DRUG. Indicate on the scales below how much you think each part of the
statement influenced you.

Assume that each scale equals 1002. Check the point on each scale which
you think best describes the amount of influence that part of the state-
ment had on your rating of the statement on the whole. Remember that
the percentage checked on each scale when added together must be equal
to 100 percent.

A.CATHOLIC m: : : : clz___

.o— o—Wo_-—.—--.‘*

O 10 20 30 40 50 60 7O 80 90 100

USING A HABIT FORMING DRUG
c11___
0 10 :20 30 :40 50 60 70 80 90 100

[lease Check: Add the two percentages together that you hawe checked.
Do they equal 1002? If not readjust your rating so
that they sum to 1002.

£1154

APPENDIX B

VERBAL INSTRUCTION GIVEN VIA CCTV

The next few minutes, we will be participating in’a
scientific study on the meaning of words. I am Bill Tedrick,
and besides being a member of the State 4-H - Youth program
staff, I am also a graduate student in the Department of
Communication Arts here at Michigan State University. As
part of my work, I am conducting some research which I hope
you will help me in by participating and completing the
materials that have been passed out to you in your viewing
rooms just after the news broadcast. First of all, I want
to say a few words about these test materials. I hope that
you will take these test materials and consider them on a
page by page basis--rather than looking back and forth
through the materials that you would take each page as it
comes. The materials have two parts. The instructions for
the first part I will give you at this time. This is the
blue sheet which is on tap of the materials that you have.
Then, about halfway through, you will find a yellow sheet
which is the instructions for the second part of the materials.
We won't be sticking directly to the words on the printed
sheet, but will be following it as our guide to instruction.
It is very important that you follow closely with me as we
go through the instructions so that you will be able to respond
to the test materials in the way they are intended. So, with-

out further information about the test materials themselves,

.115

I will proceed now and we will.go.through the instructions for
Part 1. The purpose of this study is to explore ways peOple
think about individual words and combination of words within
a single statement. In Part 1 of this form, you will be pre-
sented a series of ideas made up of either two words or a
series of words. In other words, a phrase. We want you to
consider each of these ideas separately and judge them
against a series of descriptive scales. We want you to make
your judgments on the basis of what you feel about these
ideas--what these ideas mean to you. There are no right or
wrong answers. It is your personal opinion that you are to
use to make your judgment about each of these words or
phrases. These materials are not identified in any way so
your responses to them are completely confidential. Let's
look at how this works. Here I have the scale which appears
on the sheet you have in front of you. You will notice at
the t0p of the scale there is the word my friend and on the
left side there is the word important and on the right side
unimportant. There are seven spaces between those two adjec-
tives. Another way to look at this scale is to see it in
this manner. On the page, you will find that another scale
has been devised which sets up the same sort of situation
that we just discussed. The word my friend is at the tOp,

on the left side of the scale is important, on the right

side is unimportant. If you consider the word my friend and

you think that my friend is very important, you would check

116

this marked here. You would.check with an "X" right in the
center of this space. If you think that my friend is very
unimportant, then you would check this space over here. If
you have no feeling about my friend in terms of importance or
unimportance, you would check the middle space which is a
neutral space. The second space in on both sides would be
quite, which would mean quite important, or quite unimportant.
The first space for neutral on either side would be a little
important or a little unimportant. I think when we turn the
page to the first white sheet, and let's do that now, let's
turn the page to the first white sheet, and this is about
what you will see. The word at the top will be different.

In this case, I am using a policeman. Then, there are three
sets of scales under the phrase or word that you have on

your test materials in front of you. The first set would be
admirable or deplorable, or deplorable - admirable, whichever
the case may be. The second would be worthless - Valuable,
or valuable - worthless. The third would be good - bad, or
bad - good. Again, there are seven spaces between each of
these scales. Again, I ask that you consider the word at

the t0p of the three scales and then consider each one of
these, placing an "X" at the point on the scale which you
feel would best indicate how you feel about the word or
phrase at the top of the scale. There are three or four im-
portant things that you need to consider here. First of all,

you check each of these scales. In other words, you should

117

have a check for the first, second, and third scales for each
of the wordsor series of words that.are listed on the pages
in your test materials. There should only be one check on
each one of these. In other words, if you feel that a police-
man is very admirable, you would check here; if you thought
he was very valuable, you would check here; if you thought he
was very good, you would check here; or, if you felt neutral,
you would check in the middle spaces. Remember now, one
check on each of these three scales. Now, there are people
in your room who are able to help you individually if you do
not understand the instructions entirely. The leader develOpers
would be able to help you,and there will also be room monitors
stepping into your room who can answer questions if you have
individual problems in responding to the material as I have
explained it here in the studio. I believe that this is all
the instructionf you need for this part, so please start
cheking each of the items on the white pages in your material
until you come to the yellow page. At that point, please wait
until I return to give you instructions for Part 2. You will
have ample time--5 or 6 minutes is all it will take you to
complete this part of the material. Thank you and go ahead
and start on your work.

Now that you have had time.to complete Part 1, you must
move to Part 2. If, by chance, you did not have time to com-
plete all of the first part, please stop where you are and

listen to the instruction; for the second part, get that, and

118

then go back and pick up on the first part wherever you left
off. However, I believe most everyone is completed. Now,
Part 2. Everybody turn to the yellow page, if you are not
there, and we will begin on this. Part 2 is a little dif-
ferent. In Part 1, you rated a series of statements to
indicate how you felt about their meaning. Part 2 is a
little more difficult to understand, perhaps, but I am sure
that as we work through this, it will become clear to you in
the next few seconds. I'd like for you to turn the yellow
page over and you will find, then, on the first white sheet
something that looks very much like this one. In other
words, at the tOp you will find No. 1, then you will find a
No. 2, and then under No. 2 you will find an A, B, and C.
This sheet is like the one that you have in your test materials
and we will need to now consider each part of this so that you
can proceed. So, let's first of all consider the Part 1 on
Part 2. Merely consider this as a statement, how do you feel
about something, in your materials. In this case, I have a
policeman fighting crime. There are a series of scales from
strongly disapproved to strongly approved and in your booklet
you will find that there is a l, 2, 3, 4, S, 6, and 7 under
this scale. If you strongly disapprove, you would check at
this end, and if you strongly approve of this statement up
here, you would check at this end. If you neither approve or
disapprove, you would check about the 4 in the center of the

scale. There is no problem on this. It is just a matter of

119.

checking how you feel in terms of disapproving or approving
of the statement that is at the tOp of the scale. Now, let's
move to Part 2. This one is a little more difficult, perhaps,
to understand and I think we can best do this by just reading
through one of these. I.will read one which is not in any of
the test materials which you have in your room. It is like
the ones you have in your test materials, but it is in form,
but it is not in content. Let's take a look then at the

No. 2 and read through one of these. This is what mine says:
In rating a pliceman fighting crime the way you did, how
important, that is, how much weight does the word fighting
crime have in determining your rating of the statement, "A;
policeman fighting crime"? See, we are asking a question here
for you to compare the second part of the statement with the
total statement. The second question in this part is: Did
you feel about the combination, a policeman fighting crime,
the same way you felt about the words a policeman?“ See, now
we have asked the second question, do you feel about the
fighting crime part of the statement the same way you feel
about the a policeman part of the statement. Estimate how
much weight fighting crime and a policeman had in determining
the way you actually rated the combination a policeman fight—
ing crime. Let's turn to Part A of this section. ‘Here, we
find that Part A says my feeling about fighting crime, or
whatever statement you have in your materials, completely,

that is, 100 percent, determined the way I rated a policeman

120

fighting crime. Now, here you have two alternatives--you
could say, yes, fighting crime did completely, 100 percent,
determine how I rated the statement, a policeman fighting
crime, or you could say, no, and if you checked yes, then
you would go to Part B and you would not answer Part C. If
you checked no for this section, for this particular state-
ment, you would go to Question C and not answer Question B.
This is pretty important that you get this clear. If you
check yes, move to Question B. If you check no, move to
Question C below. Let's take a look at the B part of this
section. In this section, the statement is: In fact, my
feelings about a policeman fighting crime.are even more
extreme than my feelings about other people fighting crime.
Here,again, you have two alternatives. You can say, no, my
feeling about A POLICEMAN FIGHTING and FIGHTING CRIME are
about as equal strength, or yes, my feeling about a police-
man fighting crime is even stronger than my feeling about
other pe0ple fighting crime. Now, remember, if you checked
yes above, you would be working on Part B, and if you checked
no, you would have completed this. In other words, you would
say that these two parts of the sentence are equal. If you
would say, yes, I feel even stronger about the policeman than
I do other pe0ple fighting crime, then you would have to
indicate whether you feel slightly stronger, quite a bit
stronger, much, much stronger in one of the three spaces be-

low that. I indicated that if you checked no in Part A under 2,

121

that you would need to move to Part C. So, let's take a look
at Part C and see what it looks like. Here we have the state-
ment: My feeling about fighting crime did not completely
determine my rating of the combination a policeman fighting
crime. Indicate on the scale below how much you think each
part of the statement influenced you. In other words, we

have now two parts of the statement. We have a fighting crime
on one of these scales and a policeman on the other scale.
Each one of these scales go from zero to 100. Now, assuming
that you had checked no under 2, you would need, then, to
decide how much weight the policeman, or the first part of
your statement, had and the second part of your statement had
in determining how you rated the statement at the top of the
page. So, you would determine which one of these percentages
you think is the best guess as to how important it was in

your rating of the statement. If you put an "X" at 20 on a
policeman, then you would need to put an "X" at 80 on the
:Eighting crime segment of the statement in order to make these
two scales total to 100 percent. I think that this completes
the information that you need to go through this. Again, I
will indicate to you that there are people in your room who
can help you understand this, if you do not completely under-
stand it, and all you would need to do is hOld up your hand
and one of the leader developers or one of the room monitors
would help you individually. So, go right ahead now and

complete Part 2 of the materials. After you have completed

1.212

these, check them in with your leader develOper and they will
look through them so that they will know that you have com-
pleted them in the way expected.

Thank you very much for cooperating with me on this

study. I appreciate it very much.

unjyumnm‘

7

WW