.«.-<<.<<<. .- o . . _
~'. 0 _. ...~~._ .. ... .....v. ......o......‘-. . .... . . ..o.. .. .... . . . ... .. .. . . . .. . _ . . . . ...o- ...... . nov‘.o.s.l...l. .
.o .. .. .. .. . .... .. . ..‘_.. . _ . ... .. ...,. .. .. . ., _. . .. .. . ... ...... o .
... . ..~ .. . ... u. . ... ,.. . ... ._ ... . . . , , . . . . 4. o o, ..o
. ... . . .. ... .. . . . . ....v._v . ... . . . .. . . _ . .. . , . . . . . . . ‘.. o ...-t...
. u ... .... _.. .. .‘ o, . ... V..Y. ...x. .. . . . . . . . 0.. . a I
.
1 .

_. 1.
. . . . ..a... . v

v “hug-1.1.! '- Q-Ochju"fofctfltfi'0m.q~c“-“‘-‘ "II

  

 
  

  
  

    

  
  
 

y N. , -. . . . _.
. .. . c . . . . . .. ..
' .
4 . . v
a An M .
u ‘ v . . . o. .
‘. _ . .
. I o . - l .
. v ‘ g r l l A
. . 88. N. B: . . .. . .
l . uI '
. #5. U. . -
E .8 . V - . _ . .
i . . I A
.6 AU . ...! . . ,
~ . , . _ I. . a n I I p
a A ‘ e T . . .. .
. . . O .
. .n. u m .S - . . . - .
. ‘ r. . . _ . . . .. .
—. ¢ . . . . .
Q r I. - .
. r
v S In..."
. f
.3 . . .
.I O I
o
M r l l
. . .
. I .
. .
.
v
,.
7
o
_.
.
*
a
.
O
t a . u
. I O
l. . . -
r . _ . . . .
.v n. ... . . u... . .... . .. ‘. . . . ,. .... .. . , . .. _ .. . u . o . . . . . . u . .
. . _. ......p. v . . . . . . , . . . . a o . c..r.:.. . 0.. 0.... . . ‘ . n c
. . .. . . o ._ . . ....r. .... '.,..o. . .. .. . .. . . .. .. ...1 .....r . .4‘ n .. . ... E . AV.. 1‘... . . .oooo)-..'tn.c.o\ . 1 ..
a ..1 1... (”fl. ... . ..t- . .... ...o...» ., : _¢.:c_o .... o. _. ... . o. _ ,o.....O’ o. ... .... ... I... ...... . .’X.. o. . ‘3 :2 r . a. g ,t’oLf ..A ‘ ‘ l a I 3 . not ... . .‘9. -u ‘
W.” o . O . . . .. . ... . .. . . .. a. . I ... .. ... ..w. .. I I . ......I. ..v.... ..4~...1...v.... . n.” H2»...c:x.- ..wfo ...,”a. .~ ..rmcw‘.. H: ._. .b . our ... m..m.. .....i. .n . ..l ...Hsu n ,. 1. H 11).)...”qu {wt
. u . d ‘ o. .n O .0 s ' 1,..“'.OQO Q‘u-oQ . .l'. 0'-

 

Unlin-

 

 

ATTITUDE CHANGE AND VALUES

BY

WILLIAM JENNINGS BROWN JR.

AN ABSTRACT OF A THESIS

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

MASTER OF ARTS

Department of Psychology

1972

ABSTRACT

ATTITUDE CHANGE AND VALUES
BY

WILLIAM JENNINGS BROWN JR.

This thesis tested the generality of a paradigm
introduced by Rokeach for inducing attitude and value change.
Rokeach's persuasive message said to college students that
their high rank for the value freedom and their low rank for
the value eguality meant that they valued freedom for them-
selves but not for others. He drove the argument home by
noting that those who oppose civil rights show the greatest
disparity in their values. Two alternate versions of
Rokeach's experiment were run. In one treatment Rokeach's
exact argument was used with the concept "labor unions"

replacing "civil rights" and with the value happiness replac-

 

ing freedom. This group was not significantly different

from the control group on any dependent variable. A second
treatment was obtained by making only the replacement of
happiness for freedom. This group showed a significant
increase in their value for eguality and no increase in their
attitude toward civil rights over a period of six weeks--
results which are in accord with Rokeach's findings. Since

change was dependent on the presence of "civil rights" in the

William Jennings Brown Jr.

argument, the results were taken to contradict Rokeach's hypo-
thesis that his results were a function of his paradigm rather
than the particular values and attitudes involved. Instead
the results were hypothesized to have been produced by the
strong feeling of selfishness produced by the use of civil

rights in the argument.

ATTITUDE CHANGE AND VALUES

By

WILLIAM JENNINGS BROWN JR.

A THESIS

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

MASTER OF ARTS
Department of Psychology

1972

To those to whom no one has ever made a dedication.

ii

ACKNOWLEDGEMENTS

I wish to express my most sincere appreciation to
Dr. J. E. Hunter, my committee chairman and academic advisor,
who through his patience, insightfulness, hospitality,
encouragement and countless hours of unselfish consultation
has contributed immeasureably to this thesis.

Thanks are also due Drs. T. M. Allen and M. Rokeach
without whose critical commentary this thesis would certainly
have been a lesser achievement.

I also wish to eXpress appreciation to Rhonda Hunter
and Judy Ridge for their assistance in proofreading the
final manuscript and Rhonda especially for the many other
instances where she has gone out of her way to be of help,
not the least of which has been the provision of good food

during long consulting sessions with Dr. Hunter.

iii

TABLE OF CONTENTS

DEDICATION . . . . . . . . .
ACKNOWLEDGEMENTS o o o o o; i o o o o o 0
LIST OF TABLES . . . . . . . . . . . .

LIST OF APPENDICES

Chapter
I. INTRODUCTION . . . . . . . . . .
II. PRELIMINARY CONSIDERATIONS . . . . .
A Preliminary Study to Find Alternate

Attitude-Value Configurations for Use
in Rokeach's Paradigm .

The Selection of the Bipolar Adjectives
and the Resulting Reliabilities of the

Semantic Differential Scales . . .
The Civil Rights and Labor Unions

Scales . . . . . . . . . .
Miscellaneous Calculations: Appendices

B and K . . . . . . . . . .

III. THE DESIGN AND PROCEDURE FOR THE MAIN
STUDY 0 O C O O O O O O I O 0

Purpose and Strategy . . . . . .

General Design . . . . . . . .

subjects 0 O O O O O O O 0 0

Procedure . . . . . . . . . .

IV. RESULTS . . . . . . . . . . .
Attitude and Value Change Results . .

Rokeach's Results . . . . . . .

V. DISCUSSION . . . . . . . . . .
REFERENCES 0 O O O O O O O O O O O O
APPENDICES O O O O O O O O O O O O O

Page
ii

iii

vi

11

11

13
13
14
14
15
25

25
28

34
39

41

LIST OF TABLES

The data table presented by Rokeach as
part of his persuasive message . . .

The data table presented to the LU group
as part of the persuasive message . .

The data table presented to the CR group
as part of the persuasive message . .

Intercorrelations of bipolar adjectives
selected for use with institutions
(averaged over 18 semantic differential
concepts). . . . . . . . . . .

The data table from Rokeach (1968, 1971)
presented to the CR and LU groups as
part of the persuasive message . . .

The data table presented to the LU group
as part of the persuasive message . .

The average change in attitudes and values
observed in the main study . . . . .

The average attitude change and value
change reported by Rokeach (1968) in his
first use of the paradigm used in the
main study here . . . . . . . .

The average change in attitudes and values
reported by Rokeach (1971) in his second
study using the paradigm used in the
main study here . . . . . . . .

Page

19

22

26

29

31

LIST OF APPENDICES

CLUSTER ANALYSIS OF ATTITUDES . . . . . .
CLUSTER ANALYSIS OF VALUES . . . . . . .

A DETAILED COMPARISON OF THE LU AND CONTROL
GROUP S O O O C O C O O O O O O .

VALUE AND ATTITUDE CHANGE: TREATMENT BY
SUBJECT INTERACTIONS AND OTHER CORRELATES OF
C HAN GE 0 O C O O O O O O C C O O

ATTITUDE AND VALUE CHANGE WITHIN THE CR GROUP
ATTITUDE AND VALUE INSTRUMENTS . . . . .
CONTROL GROUP INSTRUMENTS . . . . . . .

EXPERIMENTAL MATERIALS AND INSTRUMENTS USED
WITH THE LU GROUP , . . . . . . . . .
EXPERIMENTAL MATERIALS AND INSTRUMENTS USED
WITH THE LU GROUP . . . . . . . . . .
EXPERIMENTAL MATERIALS AND INSTRUMENTS USED
BY ROKEACH IN PREVIOUS RESEARCH . . . . .

THE TEST-RETEST CORRELATIONS FOR THE ATTITUDE
SCALES AND THE VALUES . . . . . . . .

vi

Page
42

48

58

62

75

83

90

93

100

102

104

CHAPTER I
INTRODUCTION

In the past, attitude change studies have consisted
primarily of the presentation of an attitude related communi-
cation and the evaluation of the effect(s) of that communica-
tion on the attitude. Rokeach's development of his Value
Survey (Rokeach, 1968) and his subsequent work with attitudes
and values has added a fresh dimension to this area of
investigation. He has turned from the traditional context of
persuasion to that of education and re-education.

The present chapter will first discuss, in general
terms, the experimental paradigm that Rokeach has used to

effect attitude and value change1

and some of the theory
underlying these efforts. Second, more Specific attention
will be given to the attitude and values that have been his
primary focus and a discussion of some pertinent results.
The final section will concern itself with the relationship
between the present effort and that of Rokeach.

[The Rokeach Value Survey is based on the assumption ’

that a person's value system is hierarchial with respect to

the "importance" of the values to him. That is, if we could

 

1Throughout this eXposition "value change" should be
taken to mean changed in the rank order of the values on the
Rokeach Value Survey.

get a complete list of the values in a person's value scheme,
this list could be rank ordered along a dimension of impor-
tance for that person. If values were the sole determinants
of behavior, then when two values come into conflict in any
situation, the value the individual holds to be more impor-
tant as a "guiding principle" in his life would determine
what action the person will take.

Rokeach holds further that there are aSpects of the
attitude and value systems of most peOple that are inconsis-
tent. For example, a person may profess a quite liberal atti-
tude toward civil rights, but we may find that equality and
brotherhood are ranked near the bottom on the Value Survey.

If these elements have never been discussed in the same con-
text, then the subject would not be aware of the inconsistency
and it would go unchallenged for years. The essence of the
paradigm that Rokeach has used successfully in his attitude
and value change work is to make such inconsistencies salient.

For example, Rokeach (1968, 1971) has studied the
relation between civil rights attitudes and the rankings of
the pair of values, equality and freedom. The empirical
evidence that is presented to g in the eXperimental setting
indicates that people who are pro-civil rights rank freedom
high but equality low. The interpretation placed on these

results are as follows:

This raises the question as to whether those who are
against civil rights are really saying that they care a
great deal about their own freedom but are indifferent
to other peOple's freedom. Those who are for civil
rights are perhaps really saying they not only want
freedom for themselves, but for other peOple too.

 

The results from these experiments indicate that this
brief experimental treatment does in fact produce positive
change in civil rights attitudes and higher rankings of the
values equality and freedom. The changes in these three
variables remained measurably significant for as long as
seventeen months after the experimental session. (Appendix
J contains a copy of the printed eXperimental materials used
by Rokeach.)

The present study was motivated by the power of this
simple eXperiment. Empirically the question that came to
mind was: will the power of Rokeach's paradigm extend to
other attitudes and other variables? This can also be stated
more theoretically. Rokeach hypothesized that the agent of
change in his studies is the juxtaposition of attitude and
value, i.e., the inherent inconsistency of the values
revealed. If this is true, then the results stem from his
method and the results should generalize to other attitude-
value configurations. By the same reasoning, if his paradigm
does not generalize to other attitude-value combinations,
then his explanation is immediately brought into question.

One simple solution to this problem would be to
select some concept such as labor unions that can be 123;:
gally substituted for civil rights into Rokeach's persuasive
message. The substitution could then be made and the study
conducted without further ado. However, this substitution
would probably falsify the persuasive message in whole or in

part, i.e., would constitute a deliberate deception of the

subject. Might it not be possible to find a concept which
is not only logically substitutable but for which the
empirical relation cited in the argument would also hold?
Table 1 shows the relationship among freedom,
equality and civil rights as Rokeach has presented them to
SS in his studies. The data in this table were actually
collected in one of Rokeach's early studies on the relation
between attitudes and values. The preliminary study to be
described next was a search for an alternate attitude-value
configuration that would yield comparable empirical results.
The main study then assessed the attitude and value change
produced by that alternate configuration.
TABLE l.--The data table presented by Rokeach as part of his
persuasive message (Rokeach, 1968, 1971). It is

actual data although the phrase "average rankings"
actually means "rank of average value."

 

Average Rankings of Freedom and Equality by MSU
Students For and Against Civil Rights

No, Not
Yes, Yes, But Sympathetic
and Have Have Not to

Participated Participated Civil Rights

 

Freedom 6 1 2
Equality 5 ll 17

 

Difference +1 -10 -15

 

CHAPTER II

PRELIMINARY CONSIDERATIONS

A Preliminary Study to Find Alternate Attitude-Value
Configurations for Use in.Rokeachjs Paradigm

 

The initial aim of the preliminary study was to find
an attitude object that had a relationship with equality and
freedom parallel to that of civil rights. If such an atti-
tude were found, then it would be possible to repeat Rokeach's
paradigm with his exact words and conclusions; only the name
of the attitude would be changed. Sixteen concepts were
selected for study and suitable bipolar scales were devised
for use in semantic differential instruments for those con-
cepts. The concepts were police, law-and-order, large cor-
porations, computers, small businessmen, the draft, disrup-
tive protests, wire tapping, hippies, labor unions, strikes,
boycotts, civil rights, migrant workers, psychologists and
pollution.

Questionnaires were administered to 114 male and
female undergraduates from introductory psychology courses at
Michigan State University. §$ were told the purpose of the
study was to establish relationships between attitudes and
values and to improve the questionnaires. Critical comments
were invited. gs completed Form E of the Rokeach Value Sur-

vey, a 16 concept semantic differential and a 20 item,

5

pretested and cross validated scale on attitudes toward labor
unions (Berger, 1969).

A detailed search of the sixteen concepts failed to
reveal a concept that had the same relationship to freedom
and equality as civil rights. However, diligent search

revealed the fact that the two values happiness and equality

 

did show a relation to attitudes toward labor unions that
was parallel to that shown in Table 1. This data is pre-
sented in Table 2. This suggested a replication of Rokeach's

study with happiness and labor unions replacing freedom and

 

civil rights. However, if both substitutions were made, this
would confound the comparison to Rokeach's results since a
failure to replicate Rokeach's results could be a function of

either substitution. Since the labor union, freedom, equality

 

configuration did not conform to Table l, the only possible
single substitution configuration remaining is civil rights,

happiness,‘equality. Fortunately, personal communication

TABLE 2.--The data table presented to the LU group as part
of the persuasive message. Although these were
the findings of the preliminary study, the phrase
"average ranking" would more accurately be stated
"rank of the average value."

 

Average Rankings of Happiness and Equality by MSU
Students For and Against Labor Unions

Pro-labor unions Neutral Anti-labor unions

 

Happiness 5 ' 4 4
Equality 7 ll 13

Difference -2 -7 -9

 

with Rokeach revealed that his gs produced responses on
happiness, equalipy and civil rights that were highly similar
to those shown in Table l. Rokeach's data on these variables

are presented in Table 3.

TABLE 3.--The data table presented to the CR group as part
of the persuasive message. These are the results
found by Rokeach, but again the phrase "average
rankings" actually should read "rank of the
average value."

 

Average Rankings of Happiness and Equality by MSU
Students For and Against Civil Rights

 

 

Yes, Yes, But No, Not
and Have Have Not Sympathetic to
Participated Participated Civil Rights
Happiness 4 2 l
Equality 5 ll 17
Difference -l -9 -l6

 

The preliminary study also provided ancillary data
about the questionnaire per se. In particular it appeared
that the two semantic differential concepts "strikes" and
"labor unions" could replace the scale of 20 items. The cor-
relations of the sum of the 20 items with the "strikes" and
"labor unions" concepts was .70 and .75 respectively. There-
fore the main study need not use Berger's scale.

The data from the preliminary study also confirmed
the selection of the particular bipolar adjective pairs used
in the semantic differential. This will be considered in

greater detail in the following section.

The Selection of the Bipolar Adjectives
and the Resulting Reliabilities of the
Semantic Differential Scales

 

 

 

The attitude objects in the present study were social
institutions and social issues. The bipolar adjective pairs
for the semantic differential typically found in the litera—
ture do not lend themselves to use with these topics. There-
fore a more appropriate set of evaluative and non-evaluative
or filler adjective pairs was used. In order to test the
empirical appropriateness of the scales selected and in order
to calculate the reliability of the evaluative scale for
each concept, the responses to each individual bipolar
adjective pair were correlated. The correlations presented
below were calculated for the pretest measures in the main
study. Table 4 lists the adjective pairs and their inter—
correlations, averaged across 18 concepts. Twenty concepts
were measured in the study, but the attitudes toward "psycho-
logists" and "pollution" were very homogenous. This abnor-
mally low variance produced abnormally low correlations so
these concepts were not used in the present analysis.

In the a priori construction of the semantic differ—
ential, five pairs were constructed to be evaluative and
five were intended to be nonevaluative fillers. The first
five adjective pairs in Table 4 are those predicted to be
evaluative and the last five are those predicted to be non-
evaluative. The average correlation is .605 for the evalua-

tive adjective pairs. Thus the internal reliability

 

 

oo.a we. SH. mm. mo. no. OH. AH. ma. 0H. mcflcflaomaumcfl3ouo
me. oo.a Hm. mm. om. om. mm. AN. mm. mm. ucmuuomsH
mmmqlmuoz mcHEoomm
SH. Hm. oo.H mm. mm. mm. mm. me. pm. pm. mmooumum>flum>occH
mm. mm. om. oo.H om. km. as. am. we. oa. ucmflofimmmcHuuaonAumm
mo. om. mm. om. oo.a Ne. ms. me. me. me. m>flufloamxm
nomufluflmm oHHnsm
so. om. mm. km. NS. oo.H om. mm. om. Hm. ummconmflnuummcom
0H. mm. mm. se. me. om. oo.H mm. mm. em. m>fluosuummaum>fluosuoum
Ha. pm. me. mm. we. mm. mm. oo.H mm. on. ammucooo
ma. mm. Em. ms. ms. om. mm. mm. oo.H HE. mmmasuuoz-manmsfim>
OH. mm. km. ov. me. am. so. on. an. oo.a manmuflmmccaumanmufimmo
m>aum9am>m|coz m>fiumsam>m

 

Amummocoo HmfipcwumMMHo caucmEmm
ma Hm>o ommmuw>mv mcowusuflumcfl nuw3 mms How

omuomamm mm>fluomflom unaomwn mo mcoflumamuuooumchls.v mqmda

10

(coefficient alpha) for the evaluative scale in the averaged
matrix is .85. The reliabilities of the evaluative scales)
for each of the 18 concepts were calculated individually.
They averaged .88 and ranged from .79 to .94.

The correlations among the evaluative adjectives are
all higher than the correlations between the evaluative and
non-evaluative adjectives or the correlations among non-
evaluative adjectives. Among the evaluative adjective pairs,
"desirable-undesirable" is the strongest pair and "honest-
dishonest" is the weakest pair. Across the concepts,
honesty was stronger when the role of peOple was evident in
the institution, e.g. police, but was weak if people were not
involved, e.g. computers.

All of the non-evaluative adjectives could have been
used as evaluative adjectives for one or more concepts,
judging from the correlations taken concept by concept. The
magnitude of the averaged correlations between the evalua-
tive and non-evaluative adjectives in Table 4 adequately

reflect the proportion of instances where this is true, i.e.,

 

"public spirited—exploitive" could most often have been used
as evaluative and "growing-declining" least often.

In general the magnitude of the intercorrelations and
reliabilities for the evaluative adjectives indicate that the
selection of these five bipolar pairs was in fact highly

appropriate for social and institutional concepts.

11

The Civil Rights and Labor Unions Scales

Before the main study was done, there was some con-
cern that a single semantic differential concept like
"Civil Rights" might provide too narrow a base for measuring
the civil rights attitude. Therefore the concept "The
Civil Rights Movement" and four additional concepts were
used. These four concepts were taken from Rokeach's attitude
questionnaire used to measure the same attitude and changed
only as much as necessary to make them appropriate for a
semantic differential format. They are: (1) "Open housing
laws that forbid discrimination against Negroes," (2) "The
use of sit-in, lie-in and picket line tactics by civil rights
demonstrators," (3) "Inter-racial marriage," and (4) "School
‘busing' to achieve racial integration."

For similar reasons, labor unions were not only
represented by the concept "Labor Unions" but by "Strikes"
and "Boycotts" as well.

In order to test the scales, a cluster analysis of
the twenty concepts was performed. The two scales were
clearly confirmed. Most of the remaining concepts fell into
two clusters related to attitudes toward the establishment
and toward dissenters. The full clusters analysis of the

twenty attitudes is presented in Appendix A.

Miscellaneous Calculations: Appendices B and K

 

In making a full exploration of the question of value

change, it became necessary to greatly increase the

12

reliability of difference scores for values. Therefore a
cluster analysis of both terminal and instrumental values was
performed. This cluster analysis is presented in Appendix B.
Although the treatment was intended to produce real
change in attitudes and values, the test-retest correlations
still give conservative estimates of the coefficient of
stability (Cronbach, 1960) or "test-retest" reliability of
each instrument. The average test-retest correlation between
single sessions (about 3 weeks) was .84 for the four attitude
scales, .64 for the 18 terminal values, and .59 for the 18
instrumental values. The average test—retest correlation
from session 1 to session 3 (6 weeks) was .78 for the four
attitude scales, .57 for the terminal values, and .54 for the
instrumental values. The fact that the test-retest correla-
tions for 6 weeks are only slightly lower than those for 3
weeks means that most of the variance in the observed change

scores is due to unreliability.

CHAPTER III

THE DESIGN AND PROCEDURE FOR THE MAIN STUDY

Purpose and Strategy

 

Rokeach studied the change produced by juxtaposing
the values freedom and equality and the civil rights atti-
tude. The central question for this study was whether
Rokeach's results would still be obtained if other attitudes
and values were substituted in his paradigm. The preliminary
study established the fact that the empirical relationships

that Rokeach capatalized on between freedom, equality, and

 

civil rights also hold for happiness, equality, civil rights

 

and for happiness, equality, labor unions but not for freedom,

 

equality, labor unions. Thus without deceiving the subjects,
two new treatment conditions can be defined within Rokeach's

paradigm. 'The substitution of happiness for freedom produces

 

the configuration happiness, equality, civil rights and will

 

be referred to hereafter as the "civil rights" or CR treat-

ment. The substitution of both happiness for freedom and

 

labor unions for civil rights produces the configuration

happiness, equalipy, labor unions and will be referred to

 

hereafter as the "labor unions" or LU treatment.

13

14

General Design

 

The study consists of three groups: LU Group, CR
Group and Control Group. All gs participated in three testing
sessions. Session 2 was conducted two weeks following Session
1 and Session 3 was six weeks after Session 1.

Session 1: The initial session consisted of pre-tests
on the twenty attitudes measured by the semantic differential,
Form E of the Rokeach Value Survey, and the experimental
treatment for the LU and CR Groups. There was no immediate
post-test.

Session 2: The first post-test on the attitudes and
values was administered. §§ were also asked to complete a
personality inventory at the request of a colleague doing
related research.

Session 3: The second post-test on the attitudes
and values was administered. Just prior to Session 2 the
shooting of the students at Kent State University occurred.
Some violence and a general student strike (boycotting of
classes) had occurred on this campus. Therefore the oppor-
tunity was taken during Session 3 to assess each subject's

involvement in the strike.

Subjects

gs were male and female undergraduates from introduc-
tory psychology courses at Michigan State University. They
were required as part of their course work to participate in

psychology eXperiments and signed up for the present study

15

knowing only that it related to "Social Issues and Personal
Values." Each §_was contacted by telephone prior to each
session and reminded of the time and place of the meeting.
In Sessions 1, 2, and 3 there were 119, 118, and 116 SS

respectively.

Procedure

 

The three testing sessions were conducted in a class-
room setting in groups of approximately ten to twenty sub-
jects. LU, CR and control groups were treated differently in
Session 1, therefore questionnaires and eXperimental proce—
dures were administered separately to the various groups for
that session. During Sessions 2 and 3 all subjects received
the same information and completed the same questionnaires
and no attempt was made to retest intact Session 1 groups.

Subjects signing up for the study were given an oppor-
tunity to choose the time and date of Session 1 from several
possible testing sessions over a two week period. All sub-
jects choosing a particular time and date were taken as a
single testing group for Session 1. Each of these groups was
assigned to a particular treatment condition (LU, CR or con-
trol) in a semi-random manner. The first testing group was
assigned randomly to one of the three. The second testing
group was assigned randomly to one of the two conditions
remaining,and the third testing group was assigned to the
remaining treatment condition. The same was done for the

fourth, fifth and sixth testing groups, and so forth. This

16

was subject to two restrictions: (1) no two successive
testing groups were assigned to the same treatment condi-
tion, and (2) when it was clear that the number of subjects
participating was going to be less than anticipated, the
assignment of a testing group to the control group was skipped
every other time, resulting in one assignment of a testing
group to the control group out of every five assignments.
Scheduling for Sessions 2 and 3 Was accomplished at the
beginning of Sessions 1 and 2 respectively. At these times,
subjects were given the time and location of meetings that
would be held either two weeks (Session 1 to Session 2) or
four weeks (Session 2 to Session 3) in the future. They were
asked to choose a time that was convenient for them but were
encouraged to choose one that was exactly two or four weeks
from that present time or as near to that as possible.
Session 1—-Procedures and Instructions
Common to All Groups

Questionnaires were distributed face down prior to
the arrival of the subjects and early arrivals were instructed
not to turn them over until we were ready to start. At the
beginning of the session subjects were told that the study
concerned attitude and value systems in pe0ple and was being
conducted in an attempt to get a better understanding of how
these systems interrelated. An elaborate prepared statement
to this effect was read to each testing group at Session 1.
The same statement was used for all groups. Subjects were

reminded that the study entailed three testing sessions to be

17

spaced out over six weeks and the procedure for arranging
the times for Sessions 2 and 3 explained. They were then
asked to choose a time for Session 2 and write it on the top
of part I of the questionnaire along with their names and
student numbers and were assured that their personal responses
would be held strictly confidential throughout and after the
study.

The remainder of the first half of Session 1 was
devoted to completing the 20-concept semantic differential

and Form E of the Rokeach Value Survey (See Appendix F).

Session 1--Control Group

 

After completing the first test booklet, the subjects
in the control group left the room until everyone was caught
up. After they returned, they filled out a second test book-
let which is unrelated to the present study but is shown in

Appendix G.

Session l--Civil Rights Group

 

Before taking a break, subjects in the civil rights
group were asked to COpy their rank order of the 18 terminal
values onto the second test booklet. After copying their
responses, the first booklet was turned in to E, following
the procedure used by Rokeach. Subjects were then asked to
rank-order the 18 terminal values as they "think M.S.U.
students on the average would rank them." They were then
told to stop and wait for everyone to catch up before pro-

ceeding. Most subjects waited in the hall.

18

After the break, the subject was confronted with the
experimental materials. Each page displays a table that
presents some empirically derived evidence about relevant
values and attitudes. The text accompanying these tables
serves to describe the tables, place emphasis on certain
points and to make interpretations of the results displayed
in the tables, all with the expressed aim of sharing some
interesting information with the subjects and with the unex-
pressed aim of providing an attitude-and-value-change
inducing message. The text was read aloud by E'to insure
that the information was not glossed over by §3°

The first table, presented here in Table 5, was the
rank order of the 18 terminal values for 298 "students in
Psychology 151." This was followed by the accompanying text:

One of the most interesting findings shown in Table 1

is that the students on the average felt that Happiness
was very important--they ranked it 2; but they felt that
Equality was considerably less important--they ranked it
11. Apparently, Michigan State students value Happiness
far more highly than they value E ualit . This suggests
that MSU students in general are much more interested in

their own happiness than they are in happiness for other
people.

 

 

The next page was the change inducing message and is

cited in its entirety:

We have one other finding which we think is unusually

interesting. In order to make this finding more mean-
ingful and relevant to you personally, you should first
answer honestly the following question on civil rights:

Are you sympathetic with the aims of the civil rights
demonstrators?

Yes, and I have personally participated in a civil
rights demonstration.

Yes, but I have not participated in a civil rights
demonstration.

No.

19

TABLE 5.--The data table from Rokeach (1968, 1971)
presented to the CR and LU groups as part
of the persuasive message. The numbers
are the rank order of the means found by
Rokeach in earlier research.

 

Rank Order of Importance to 298
Michigan State Students

 

13 A Comfortable Life
12 An Exciting Life

6 A Sense of Accomplishment
10 A World at Peace
17 A World of Beauty
11 Equality

9 Family Security

1 Freedom

2 Happiness

8 Inner Harmony

5 Mature Love

16 National Security
18 Pleasure

14 Salvation
15 Social Recognition
4 Self-Respect

7 True Friendship

3 Wisdom

 

20

The MSU students who participated in a previous study
of value systems were asked a similar question. They
were divided into three groups, according to how they
responded. Table 2 shows the average rankings of

Happiness and Equality for each of these three groups.

Table 2.--Average rankings of Happiness and Equality
by MSU students for and against civil rights

 

Yes, Yes, But No, Not
and Have Have Not Sympathetic to
Participated Participated Civil Rights

 

 

Happiness 4 2 l
Equality 5 ll 17
Difference -l -9 -l6

 

Notice that in Table 2:

l. Pro-and anti-civil rights students all value
Happiness relatively highly. Of 18 values all
groups rank Happiness among the top four.

 

 

2. Students who are strongly for civil rights value
Equality rather highly--they ranked it 5th; but
those a ainst civil rights place a much lower value
on EquaIity--they ranked it 17th in importance.
Those who are sympathetic but non-participants
ranked Equality 11th.

3. The distance between Happiness and Equality is -1
for the strong civil rights group, —9 for the middle
group, and —16 for the anti-civil rights group.

Apparently both Happiness and Equality are important to
some people, wHile to others Happiness is very important
but Equality is not.

This raises the question as to whether those who are
against civil rights are really saying that they care a
great deal about their own happiness but are indifferent
to other peOple's happiness. Those who are for civil
rights are perhaps really saying they not onIy—want
happiness for themselves, but for other peOple too.

 

 

What do you think? (Please circle one number)

1 2 ’ 3 4 5 6 7
I agree I'm not I disagree
strongly with this sure strongly with this

interpretation interpretation

21

Before you go on to the last part of this questionnaire,
please spend a few minutes comparing your own rankings
from the first page with these results. Then go on to
the next page.

After completing the page containing the message,
subjects were instructed to complete the remainder of the
questionnaire at their own pace. The next page asked gs to
respond on a 1—7 Likert type scale to the following four
questions about the experiment.

1) Did you find it thought-provoking?

2) Do you think this eXperience will lead you to do some
more thinking about your own values?

3) Do you feel that your responses were somewhat hypo-
critical?

4) Right now, how satisfied do you feel about the way
you ranked the values?

The last page required gs to place an "X" beside each
of a list of the 18 terminal values indicating their satis-
faction or dissatisfaction with the way that they had ranked
each of the values, then indicate with a "yes" or "no"
response whether they thought the findings that had been
described to them were valid and to explain why "they answered
the previous questions the way (they) did." gs were then
invited to make any written comments they wished about the
study. This concluded Session 1 for the CR Groups. (A com-

plete copy of the second test booklet is shown in Appendix H.)

Session l--LU Group

 

The procedure for the LU group was identical to that

for the CR group, only the content of the experimental

22

message was altered. Here two things were done. First, the
phrase "labor unions" was systematically substituted for the
phrase "civil rights" throughout. Second, the numbers in
"Table 2" were changed to those given earlier in Table 2 and
reproduced here for comparison in Table 6 (See Appendix I for
the complete text of these materials).
TABLE 6.--The data table presented to the LU group as part

of the persuasive message. The phrase "average

ranking" should have read "rank of the average
value.“ A

 

Average Rankings of Happiness and Equality by MSU
Students For and Against Labor Unions

Pro-labor unions Neutral Anti-labor unions

 

 

Happiness 5 4 4
Equality 7 11 13
Difference -2 —7 -9

 

Session 2

 

Sessions 2 and 3 were post-testing sessions on atti—
tudes and values and no further "treatment" was applied to
the various experimental groups at these times. All subjects
received the same information and questionnaires. There was
no attempt to retest intact Session 1 groups.

Two questionnaires were administered during Session 2.
The first was a post-test on attitudes (semantic differen-
tial) and values. The second was a personality inventory,
which was administered at the request of a colleague who was
doing research on changes in the ranking of the value Equality

as a function of personality traits.

23

At the beginning of Session 2, subjects were asked
to select a time and date from a list of possible times for
Session 3 and to write this at the top of the first ques-
tionnaire along with their name and student number. Subjects
were told that there were two questionnaires and that
instructions as to their manner of completion were different.
The instructions for the semantic differential were read
again and subjects were admonished to read all instructions
carefully and to ask for clarification if there was the
slightest doubt as to what was eXpected. Each subject was
allowed to work at his own rate and to complete both ques-
tionnaires without interruption. When both questionnaires
were completed he was allowed to leave, thus completing

Session 2.

Session 3

 

Two questionnaires were administered during Session
3. The first was a post-test on attitudes and values. The
second was a sixty-item questionnaire which queried the sub-
ject's involvement in the general student strike which
occurred on campus following the shooting of the students at
Kent State University.

At the beginning of the session it was explained that
following the completion of the questionnaire that there
would be a debriefing period during which the aims of the
study would be explained and any questions would be answered.

Since §$ would be finishing the questionnaire at different

24

times and there might be a considerable wait for some people,
they were told that the debriefing was not a mandatory part
of their volunteer commitment. However, they were encouraged
to remain for it.

The instructions for the semantic differential were
then reviewed, again with the admonition to read all instruc—
tions carefully and to ask questions if there was the least
uncertainty. They were allowed to complete the question-
naires at their own speed. When all questionnaires were com-
pleted and turned in, the debriefing period was held. This

concluded Session 3 and the study.

CHAPTER IV

RESULTS

Attitude and Value Change Results

 

Table 7 presents the means and standard deviations
of attitude and value change for the three groups. Table 7a
shows the changes from Session 1 to Session 2 (two weeks)
and Table 7b the changes from Session 1 to Session 3 (six
weeks).

The Control Group received no attitude or value change
oriented information. No significant change was expected and
none occurred.

On the other hand, the results for the LU group came
as a surprise. The only variable on which there was signifi-
cant change was freedom. However, this was also the variable
on which the control group showed the most change. And in
fact, the amount of change on freedom for the LU group is not
significantly higher than the control group for either two
weeks or six weeks. Thus the LU group is not significantly
different trom the control group on any variable.

The CR group showed a significant increase on only
three variables. First, the change on freedom was signifi-
cant. However as was true for the LU group, the value of

freedom for the CR group did not increase by significantly

25

26

 

 

.mo. a pm unmofimwcmflm mmamno mmumoflocH A
Loo. v .om. Roe. v Ho.u Amp. V ma. mcofico Hogan
1mm. o mo. Amp. o oo. loo. V oH.- munoflm Hfl>flo
Amo.oo we. Lao.mv oo. Amm.mv oH.- mmmasmmmm
Aoo.ov .o~.~ Amo.mv mo. Amp.~v oo. spasmsom
loo.mv .mo.H Imo.ov .oo.H 1mm.mv mm. soommum
m coflmmom on H coflmmmm Scum mmcm:uau.nh
1mm. v .om. Am». C Ho.- Aom. 3 mo.- mooflco uonoq
rem. V «H. ”Ho. 1 No. Aom. V mo.- muzmflm HA>Ao
loo.ov oo.H Akm.mo om. 1k~.mv oo.u mmmoflmmmm
Iom.mv .oo.a Loo.mo om.u Ina.mc ma.n muAHmsom
ioh.mv .ok.a Amm.oo .Hm.a Ako.mv mo. soommum
m sawmmmm on a coflmmmm Eoum chmnvln.wh
om ax om ox am am

mmuz

gnome musmflm Hw>flo

nmuz

macaw coflco umnmq

msouw Houucoo

hNuZ

 

.Axcmu cw mmmmnomo

m .m.wv msaw> aw wmmmuocw cm momma Hogans m>fluwmom m

umnu om ommum>mu mmz mmcmno msam> mo swam one
man cw om>ummno mmsam> 0cm mopsuwuum CH mmcmno monum>m msanl.h mqmda

.mosum came

27

more than did the Control group. Second, there was an
increase in attitudes toward labor unions (but none toward
civil rights). However there is some reason to question this
result. The two week post test means for the Control and CR
groups are .785 and .760 which are essentially identical.

At six weeks, they are .814 and .993 which are not signifi-
cantly different. The difference occurs on the pretest:
.517 and .808. It seems entirely possible that the pretest
mean was a "fluke" and that the apparent change simply
reflected the means for the two week and six week measure-
ments returning to their actual level. Finally the CR group
showed a significant increase on equality at two weeks and
increased still further between Sessions 2 and 3.

A large systematic search for treatment by subject
interactions was made. At the same time the question of
correlated attitude and value change was considered. Most
of these results are of questionable validity and will not be
presented in the main body. Appendix C presents "A Detailed
Comparison of the LU and Control Groups" and gives a rationale
for pooling them into one large control group for further
analysis. Appendix D presents "Value and Attitude Change:
Treatment by Subject Interactions and Other Correlates of
Change" comparing the pooled control group with the CR Group.
Appendix E presents "Attitude and Value Change within the CR
Group" and some mean changes for various sub-groups for com-

parison with similar results reported by Rokeach (1968).

28

Rokeach's Results

 

Table 8 presents the results from Rokeach's (1968)
first empirical study in this area. His main experimental
group is identical to the CR group except that the persua-

sive message used freedom instead of happiness.

 

After being shown the rank order of the terminal
values for 298 students,the "weak treatment" group saw only
the short written message that interpreted the relative
ranking of freedom and equality as indicating "freedom for
me, but not for others." They did not receive the table per—
taining to civil rights nor the extended argument. The data
reported show no significant difference between the two
experimental groups and hence this discussion will assume
them to be pooled. The experimental groups show a signifi—
cant increase on freedom of .74 after three weeks which
decays to .47 after three months. The control group decreases
on freedom by -.47 after three weeks and still further to
-l.l9 after three months, with only the latter decrease being
significant. Are the experimental groups significantly dif-
ferent from the control group in their changes on freedom?
Neither standard deviations nor any equivalent statistics
were reported, and so a formal test cannot be performed. If
the standard deviations from the present study are used as a
first estimate, the critical ratios are 1.94 for three weeks
and 2.24 for three months, i.e. borderline significant at the

.05 level. The fact that the control group showed a decrease

29

owuuommu no: u u:
mmHSmmmE pwumamuuoo new ummuuu .Hoo. m *«s
monomwme omumeHHoo MOM ammulu .Ho. m ««

mmHDmMGE QOHMHTHHOO HON #mmuflu. emOo am i

 

 

 

 

A¢om.a A: no Hook.au no no munofim HA>Ao
«ssmw.a skshv.a we. s¢«NB.H «*¥h¢.H m5. muflamsvm
«we. we. «smH.HI egos. «tmh. b¢.I Eocmmnm
omauzv Amouzv Ammuzv Aokauzv Ammauzv Asouzv
HmucmEHHmmxm HmucmEfiummxm Houucou Hmucofiwummxm Hmucweflummxm Houucou
xmmz xmmz
mnpcoz m mxmmz m

 

.muws mosum some may

cw poms Eqflpmnmm mnu mo wmo umuwm was so Ammmav nommxom
an pounommu mmcmno moam> can mmnmno monufiuum mmmnm>m onslu.m mamas

30

of -.72 over the interval from three weeks to three months
may mean that the decrease in the eXperimental groups was not
a decay in the induced attitude change but simply additional
change produced by the generally negative messages to which
the group as a whole was eXposed outside the experiment.

On equality both experimental groups showed a signi-
ficant increase after three weeks and no change from Session
2 to Session 3. The control group showed an insignificant
increase of .79 which decayed to .44. Again the lack of
standard deviations prevents an exact test of the difference
between the two groups. However, using the standard devia-
tions from the present study, the critical ratios are 1.45
for three weeks and 1.73 for three months, i.e. the experi-
mental grOUps did pep increase significantly more than did
the control group on equality.

Unfortunately, almost none of the data for attitudes
towards civil rights was reported. Even the means displayed
in this table were indirectly calculated from another table.
Thus little can be said about the relation between the con-
trol and experimental groups. Furthermore without standard
deviations, no within groups test can be made. The fact that
there was an initial fluctuatiOn of —1.78 followed by one
of +1.56 strongly suggests that there was no significant
change in attitude toward civil rights produced in Rokeach's
first experiment.

Table 9 presents the data from Rokeach's (1971)

second study. In that study the control group was precisely

31

pmuuommu uo: H:
monummme Umumamuuoo new ummunu .Hoo. m «««

mmusmmme omumawuuoo How umoulu .Ho. m es

 

 

 

 

 

...os.~ om. ..oo.~ om. se.- oo.- musoflm Hfl>flo
«yamm.~ mm. «*«om.~ an. «««Hm.a mm. hufiamsom
«samm.a mm. «gmH.H Hm. ««mv.a om. Eoowmum
Amhauzv Ahvauzv Ahmanzv Aomauzv Ausuzv Aucuzv
HmucmEHmexm Houucou Hmucmefiummxm Houummw kucmEmnmmxm Houucou
mango: hanma mango: mum mxmog m

 

.mnmn modum some map so com:
Emflomnmm on» mcwms mosum oncomm was so Aahmav somwxom
an omuuommu mmsHm> can moospwuum a“ mmsmno mmmnm>m msall.m mqmde

32

equivalent to that in the present study and again the experi-
mental group differed from the present CR group only in that

the argument was in terms of freedom rather than happiness.

 

The sample sizes in Rokeach's second study are so large that
the results are unequivocal. After three weeks, the value
the subjects placed on freedom had inCreased by 1.48 and this
change was maintained throughout the entire time period
studied. The control group initially increased by .20 and
remained constant thereafter. Although no standard devia-
tions are reported, the standard deviations from the present
study yield critical ratios of about 2.0 for each time period,
i.e. borderlinq significance at the .05 level.

The experimental group initially increased its value
for equality by 1.91 and increased slightly to about 2.74
thereafter. The control group initially increases by .68 but
then eventually this decays to .32. If the standard devia-
tions from the present study are used in lieu of those not
reported by Rokeach, then the critical ratios range from
perhaps 2.0 at three weeks (the sample size for three weeks
was not reported) to 5.90 at 15-17 months, the difference
between the groups is unarguable.

In terms of their attitudes toward civil rights,
both groups show an initial nonsignificant decrease. However,
after three months the eXperimental group showed an increase
of 2.09 and this increased still further to 2.79 thereafter.
Unfortunately, the standard deviations were not reported and

hence no comparison of the control and eXperimental groups is

33

possible. Furthermore, the instrument in the present study
is only correlationally equivalent to Rokeach's measure and
hence no estimate of his standard deviation can be made.

It is worth noting that the change in the control group
increases from -.69 to .20 to .86 throughout the time inter-
val. Thus some of the increase in change observed in the
experimental group is due to positive post eXperimental

messages rather than being the product of the experiment.

CHAPTER V
DISCUSSION

In both studies, Rokeach focused on the same con-
figuration: attitude toward civil rights and the values
placed on freedom and equality. The purpose of the present
study was to see if Rokeach's technique for producing value
and attitude change would work for other combinations of

values and attitudes. When happiness was substituted for

 

freedom and labor unions for civil rights, the treatment for
the LU group was produced. This group was not significantly
different from the control group. On the other hand, if
happiness was substituted, but civil rights was retained,
then the design generated the treatment given the CR group.
For the Cr group, there was a significant increase in
equality.

At first glance this suggests that the present
results are at considerable variance with those reported by
Rokeach. However, a closer examination shows smaller dif-
ferences than at first appear. The change in equality for
the CR group was about 2 points, which is slightly greater
than Rokeach (1968) and slightly less than Rokeach (1971).
Also there was no change in attitudes toward civil rights in
Rokeach (1968) except for one subgroup of 8 subjects out of

the 460 subjects run. Furthermore, there was no attitude

34

35

change in Rokeach (1971) for 3 weeks, only for 3—5 months
and more. The duration of the present study did not allow
the effects of such a lengthy time lapse to be assessed.

Thus the main difference between the present study
and Rokeach's studies is the results for freedom. For experi-
mental groups, the present study found an averaged increase
of 1.6 while in his two studies Rokeach found averaged
increases of .6 and 1.5. However for the control groups, the
present study found an increase of .7 while in his two studies
Rokeach found -.7 and .2. Thus the differences in eXperi—
mental control comparisons between studies reflect differences
in the amount of change in the control groups rather than dif-
ferences in the experimental groups. Actually one might well
ask: why should freedom change? Freedom was never mentioned

in the present study, the arguments centered on happiness.

 

But then consider the actual arguments in Rokeach's studies.
The persuasive arguments did not state that the value subjects
placed on freedom was too low, but that the value placed on
freedom was too high! The question then in Rokeach's studies
is why didn't freedom decrease? Perhaps the answer lies in
the manner in which Rokeach presents rank order data. In the
table showing the "rank order of the values for 298 Michigan
State students," the rank shown for freedom is 1. This repre-
sents the fact that the mean rank for freedom is higher than
the mean rank for any other value. The actual mean rank of
frerdom in the present study was 5.09. Thus what most

noticed was that their own rank for freedom was much lower

36

than the rank assigned by "Michigan State students." Thus
this table told most students that their primary reference
group ranked freedom much higher than they did.

Finally, the last relevant variable is happiness.

 

In the present study, freedom was never actually mentioned in

the main argument. Instead its role was played by happiness.

 

The change in happiness averaged over sessions was -.12 in

 

the Control group and .80 in the CR group. The difference
between the eXperimental and control group is thus .92. This
is in the same direction as Rokeach's results for freedom
but only two-thirds as big. Note that analagous to freedom
in Rokeach's studies, the persuasive message in the present

study argued that the value given to happiness was too high,

 

i.e. that the value placed on happiness should be lower.

 

However, where the message assigned a rank for happiness of

 

2 to Michigan State students, the actual mean rank in this
study was 6.87. Thus as was true of freedom in Rokeach's
experiment, most subjects saw a considerable gap between

their own rank for happiness and the rank given it by their

 

primary reference group. The fact that happiness increased

 

by less in the present study than freedom did in Rokeach's
studies probably reflects the fact that the basic argument in
the persuasive message is that the subject's values are
selfish. The charge of selfishness would presumably be

stronger in the case of happiness than in the case of freedom.

 

Further evidence for this interpretation is the fact that in

the present study, freedom increased by more than happiness

 

37

did. Although the reference group discrepancy for freedom

was only 5.09-1 = 4.09 and was less than that for happiness,

 

6.87-2 = 4.87, the subject's value for freedom was not
attacked.
It is important to differentiate between the two

roles played by happiness in the present study. On the one

 

hand, the subject's value for happiness is a dependent vari-

 

able. In this role the data for happiness were only roughly

 

comparable to the results for freedom in Rokeach's studies.

But on the other hand, happiness also played the role of the

 

second value in the persuasive message for both experimental
groups. In the case of the civil rights group, the results
produced by that message were every bit as strong as those
produced by freedom.

Rokeach's work can be looked at in two ways: as
studies of the particular configuration of civil rights atti-
tude and values for freedom and equality or as studies of a
new method of producing attitude change. The goal of the
present study was to test the hypothesis that Rokeach's
results were produced by his method of juxtaposing values and
attitudes. If his results reflect method rather than con-
tent, then any configuration of attitudes and values that can
be logically substituted in his argument should produce the
same pattern of results. This study considered two such sub-
stitutions: happiness for freedom and labor unions for
civil rights.

What is the result of the substitutions? If

38

happiness is substituted for freedom, the pattern of results
is unchanged. But it appears that the civil rights attitude
is critical. If it is removed, then there is no change '
beyond that in the control group. Thus Rokeach's results do
depend on the specific attitude that he studied. This sug—
gests that the active ingredient in Rokeach's experiments

is not the juxtaposition of values and attitudes. The actual
agent of change in Rokeach's work is only a matter of specu-
lation at present. But one strong hypothesis is that the
change was produced by the effectiveness of his charge that

those who are anti civil rights are selfish.

REFERENCES

39

REFERENCES

Anderson, N. H. "Test of a Model for Opinion Change."
Journal of Abnormal and Social Psychology, 1959,
59, 371—381.

Anderson, N. H. and Havland, C. I. "The Representation of
Order Effects in Communications Research." In
Carl I. Havland (Ed.), The Order of Presentation in
Persuasion.‘ New Haven: Yale University Press,
1957, pp. 158-169.

 

 

Berger, W. G. "Some Correlates of Attitude Change, Retention
of the Attitude Message, and Evaluation of the Com-
munication." Unpublished doctoral dissertation,
Michigan State University, 1969.

Cronbach, L. J. Essentials of Psychological Testing. New
York: Harper and Brothers, 1960.

Hunter, J. E. "Methods of Reordering the Correlation Matrix
to Facilitate Visual Inspection and Preliminary
Cluster Analysis." J. Educ. Meas., 1973, (in press).

 

Hunter, J. E. and Cohen, S. H. "PACKAGE: A System of Com-
puter Routines for the Analysis of Correlational
Data." Ed. Psych. Meas., 1969, 22, 697-700.

 

Hunter, J. E. and Cohen, S. H. "Mathematical Models of
Attitude Change in the Passive Communication Context."
Unpublished manuscript, Michigan State University,
1972.

Rokeach, M. Beliefs, Attitudes, and Values. San Francisco:
Jossey-Bass, I968.

Rokeach, M. "Long-range Modification of Values, Attitudes,
and Behavior." American Psychologist, 1971, 26-5,
453—459.

 

40

APPENDICES

41

APPENDIX A

CLUSTER ANALYSIS OF ATTITUDES

A cluster analysis was performed on the subject's
attitudes toward 20 concepts during Session 1 (N=116). The
primary purpose of this analysis was to assess the internal
consistency of the concepts that were to make up the labor
unions and civil rights attitude scales.

Initially the 20x20 correlation matrix was computed
and reordered using the ORDER and ARRANGE (Hunter, 1973)
programs provided in PACKAGE (Hunter and Cohen, 1969).

After inSpecting the two reordered correlation matrices and
the matrix of similarity coefficients, a set of four clusters
was formed. An oblique multiple groups factor analysis (with
and without communalities) was computed for these four
clusters and a residual set of two concepts. The clusters
were then reevaluated and the final clustering consisted of
four main clusters and a residual set of three concepts.

Table A1 lists the names that have been given to the
four clusters and the concepts that make up that cluster.
Table A2 presents the correlation matrix for the 20 concepts
ordered by clusters, the factor intercorrelations, and factor
loading matrix that resulted from the oblique multiple
groups factor analysis. The numerical labels beside the con-
cept and cluster names in Table A1 are the same as those in

42

43

TABLE Al.--Summary of Content on the Four Attitude
Clusters and Residual Items and Coefficient
Alpha for Each Cluster

 

 

Alpha

 

501 Establishment .81
‘ 17) Police

9) Law—and Order

15) Large Corporations

7) Computers

12) Small Businessmen

502 Repression .66
2) The Draft
11) Disruptive Protests (Reflected)
13) Wire Tapping
20) Hippies (Reflected)

503 Labor Union .71
14) Labor Unions
6) Strikes
l8) Boycotts

504 Civil Rights .70
4) Open Housing laws that forbid
discrimination against Negroes
8) The use of sit-in, lie-in and
picket line tactics by civil
rights demonstrators
10) Inter-racial Marriage
16) School "busing" to achieve
racial integration
19) The Civil Rights Movement

505 Residual Set
1) Migrant Workers
3) Psychologists
5) Pollution

 

Table A2 and may be used as reference for content when
looking at the correlations. It will be noted that Table A2
has two factor loading matrices and two matrices of factor
intercorrelations. The set marked "corrected" (for attenua-
tion) is the result of calculating the multiple groups with

communalities in the diagonal of the correlation matrix and

44

TABLE A2.~~The final cluster analysis of the twenty attitudes

assessed in the main study. The entries in the diagonal

of the table are the estimated communalities for each con-
cept. The factor loadings across the bottom of the table
and the first set of factor loadings to the right were cal—
culated with communalities and have therefore been corrected
for attenuation. The second set of factor loadings on the
right was calculated with ones in the diagonal, i.e. these

factors are the average standard score for the attitudes in
The two 4x4 correlation matrices
at the bottom right of the table are the correlations
between factors;

the corresponding cluster.

corrected for attenuation.

the first table is the same as the second

The correlations are based on
the responses of 116 subjects to the pretest.

 

 

 

 

 

 

 

 

 

 

 

/ ’ Labor

Establishment Repression Unions 11

17 9 15 7 'I2 2 I1 713 20 14' 6 TB

17 Police (71) 72 52 51 41 44 49 41 47 ~2 ~25 ~32

9 Law-and-order 72 (58) 51 45 33 39 43 40 38 3 ~18 ~28

15 Large Corpora~

tions 52 51 (42) 57 19 26 23 35 18 11 ~11 ~25

7 Computers 51 45 57 (50} 36 30 27 24 24 17 ~6 ~18

12 Small Business 41 33 19 36 (19V 1 31 ~l 3 9 ~4 2

2 The Draft 44 39 26 30 1 '31‘ 28 34 35 4 ~15 ~28
11 Disruptive ,

Protests 49 43 23 27 31 28 /23 26 32 ~19 ~34 ~33
l3 Wire Tapping 41 40 35 24 ~1 34 26 r373 43 2 ~11 ~33
20 Hippies 47 38 18 24 3 35 32 43 44\~12 ~28 ~45

~14 Labor Unions ~2 3 ll 17 9 4 ~19 2 ~12 31 59 21
6 Strikes ~25 ~18 ~11 ~6 ~4 ~15 ~34 ~11 ~28 59 95 54
18 Boycotts ~32 ~28 ~25 ~18 2 ~28 ~33 ~33 ~45 21 54 26

~ 4 Open Housing* ~5 ~15 ~7 ~14 ~4 ~8 ~12 ~22 ~13 ~2 7 14
8 CR Demonstra-

tion

Tactics* ~24 ~29 ~4 ~19 ~8 ~30 ~43 ~13 ~29 9 35 37
10 Inter-racial

Marriage ~14 ~20 ~23 ~17 15 ~9 0 ~30 ~27 ~10 14 31
16 Busing* ~11 ~10 ~13 ~19 ~15 ~4 ~27 ~10 ~24 ~4 22 26
19 CR Movement ~12 ~12 ~10 ~21 ~16 ~12 ~16 ~11 ~21 8 18 13

l Migrant Workers 3 3 4 2 2 -l ~5 ~21 ~16 4 23 26

3 Psychologists 7 7 12 8 21 ~16 5 0 ~11 9 22 10

5 Pollution ~19 ~17 ~5 ~34 ~29 ~9 -6 1 ~3 ~12 ~6 -3
'501 Establishment 84 76 65 70 44 41 51 41 38 11 ~19 ~30
-502 Repression 78 69 44 45 15 56 47 61 67 ~11 ~38 ~60
503 Labor Union ~29 ~21 ~12 ~4 3 ~19 ~42 ~21 ~41 54 101 49
504 Civil Rights ~23 ~30 ~20 ~31 ~10 ~22 ~34 ~30 ~40 l 34 42

* The full statement of the concept is given in Table A1

45

 

 

 

 

 

 

 

 

 

 

 

Corrected
Residual For
Civil Rights Set Attenuation Uncorrected
4 8 10 16 19 l 3 5 501 502 503 504 501 502 503 504
~5 ~24 ~14 ~11 ~12 3 7 ~19 84 78 ~29 ~23 84 64 ~25 ~19
~15 ~29 ~20 ~10 ~12 3 7 ~17 76 69 ~21 ~30 80 57 ~18 ~25
~7 .~4 ~23 ~13 ~10 4 12 ~5 65 44 ~12 ~20 74 36 ~10 ~17
~14 ~19 ~17 ~19 ~21 2 8 ~34 70 45 ~4 ~31 77 37 ~3 ~26
~4 ~8 15 ~15 ~16 2 21 ~29 44 15 3 ~10 61 12 3 ~8
~8 ~30 ~9 ~4 ~12 ~1 ~16 ~9 41 56 ~19 ~22 37 70 ~17 ~19
~12 ~43 0 ~27 ~16 ~5 5 ~6 51 47 ~42 ~34 46 66 ~36 ~29
~22 ~13 ~30 ~10 ~ll ~21 0 1 41 61 ~21 ~30 37 72 ~18 ~25
_~l3 ~~29 ~27 ~24 ~21 ~16 ~11 ~3 38 67 ~41 ~40 34 74 ~36 ~34
~2 9 ~10 ~4 8 4 9 ~12 11 ~11 54 1 10 ~9 76 0
7 35 14 22 18 23 22 ~6 ~19 ~38 101 34 ~17 ~31 89 28
14 37 31 26 13 26 10 ~3 ~30 ~60 49 42 ~27 ~49 73 36
26 18 28 32 41 ~24 16 2 ~13 ~24 10 50 ~12 ~19 8 65
18 24 17 31 50 ~9 .21 17 ~25 ~50 40 49 ~22 ~41 34 64
28 ‘17 20 28 36 29 24 ~0 ~17 ~29 17 45 ~16 ~23 15 62
32 31 28 36 44 15 27 ~1 ~20 ~29 22 59 ~18 ~23 19 69
y_4l 50 36 44 67 8 27 6 ~21 ~26 19 83 ~19 ~21 16 80
24 ~9 29 15 8 ** l6 6 4 ~19 26 23 4 ~16 23 20
16 21 24 27 27 16 ** ~9 16 ~9 20 40 15 ~8 18 '34
2 17 ~0 ~1 6 6 ~9 ** ~31 ~7 ~10 8 ~28 ~6 ~9 7
~13 ~25 ~17 $20 ~21 4 16 ~31 100 74 ~18 ~34 100 55 ~14 ~26
~24 _~50 ~29 ~29 ~26 ~19 ~9 ~7 74 100 ~54 ~55 55 100 ~38 ~38
10 40 17 22 19 26 20 ~10 ~18 ~54 100 37 ~14 ~38 100 27
50 49 45 59 83 23 40 8 ~34 ~55 37 100 ~26 ~38 27 100

 

46

the set marked "uncorrected" has been calculated without
using communalities, i.e., with 1's in the diagonal. The
diagonal entries in the correlation matrix are the communali-
ties of the concepts within the cluster of which they are
members.

The Establishment cluster has a coefficient alpha of
.81. The cluster has good internal consistency although
attitude- toward "Large Corporations" only correlates .19
with attitude toward "Small Businessmen." "Police" is the
strongest variable internally and "Small Businessmen" the
weakest variable, both internally and externally. There is
a strong-weak gradient running from variable 17 to 12 within
the cluster that is generally evident in its relationship to
the other clusters, although the relationship of each vari—
able within the civil rights cluster tends to be more nearly
equal. Externally the cluster has a low to moderate negative
relationship with the Labor Union and Civil Rights clusters
and a high positive relationship to Repression cluster.

Cluster 502 is the Repression cluster and has a
coefficient alpha of .66. Internally the correlations are
flat and consistent and the level of the relationships is
moderate. Externally the concepts are parallel in their rela~
tions to other clusters, although "The Draft" and "Wire
Tapping" are weaker with the Labor Union and Civil Rights
clusters. Here, as with the Establishment cluster, the
relationship to Labor Union and Civil Rights are negative.

One might also note that if "The Draft" and "Wire Tapping"

47

had been reflected instead of "Disruptive Protests" and
"Hippies," the cluster might well have been named "Anti-
Establishment."

Establishment and Repression clusters can be differ-
entiated in two ways. First, the four issues in the second
cluster make up the core of the anti-establishment feelings
on campus. Second, the second cluster correlates much more
highly with the Civil Rights and Labor Unions clusters.
'Finally, it should be noted that they only correlate .74 with
other when corrected for attenuation.

Cluster 503 is the Labor Union cluster and has a
coefficient alpha of .71.

Cluster 504 is the Civil Rights cluster and has a
coefficient alpha of .70. Internally the item "The Civil
Rights Movement" is clearly the strongest and most general.
The other four items are the more specific in content and
have lower correlations. Indeed the issues selected show
considerable specificity in attitude. The items are parallel
in their correlations with the other clusters except for
item 8 which deals with the demonstration tactics of civil
rights demonstrators. It shares common content with other
items that deal with protests, e.g., "Strikes," "Boycotts,"

"Disruptive Protests."

APPENDIX B
CLUSTER ANALYSIS OF VALUES

The reliability of difference scores for individual
values in this study was extremely low. Therefore the
values were clustered and difference scores for the clusters
were investigated. In order to maximize the reliability,
the cluster analysis was based on average value scores.

Table Bl shows the results of a cluster analysis of
the instrumental values. The diagonal entry for each vari-
able is its communality in that cluster, i.e., an estimate
of how much variance the value shares with the cluster to
which it belongs. The correlations between values are much
higher than for single time measurements since three time
scores were averaged to increase the reliability. The corre-
lations between values and clusters or between clusters and
clusters are all computed on the matrix with communalities

and have therefore been corrected for attenuation.

Submissiveness

 

The Submissiveness cluster has a coefficient alpha of

.71 and consists of the values obedient, polite, clean,

 

 

responsible, broadminded (reflected), intellectual (reflected)

 

 

 

and imaginative (reflected). Thus a person obtains a high

 

score on this cluster if he assigns a high value to those

48

49

TABLE B1.~~A cluster analysis of Rokeach's 18 instrumental

The value score correlated was formed by
averaging the three ranks assigned to that value

on the three administrations and is therefore
much more reliable than a single ranking.
diagonal entries in the table are the communali~
ties of each value determined by the cluster to

values.

which it is assigned.

The

The factors were calcu~

lated with communalities and therefore the
factor loadings and the factor correlations have
been corrected for attenuation.
were used in the correlations.

All 116 subjects

 

 

 

 

 

 

 

Submissiveness Benevolence

33 34 23 35 20 30 28 26 25 32

Obedient 47 51 29 25 18 35 42 ~18 ~5 ~19
Polite 51 37 25 l6 17 34 39 ~3 9 3
Clean 29 25 26 16 36 15 36 ~29 ~22 ~19
Responsible 25 16 16 24 33 21 42 ~15 ~26 ~11
Broadminded (R)* 18 17 36 33 12 ~3 13 ~15 ~8 22
Intellectual (R) 35 34 15 21 ~3 12 9 14 21 -l
Imaginative(R) 42 39 36 42 13 9 37 ~13 ~10 ~19
Helpful ~18 ~3 ~29 ~15 ~15 14 ~13 36 39 26
Forgiving ~5 9 ~22 ~26 ~8 21 ~10 39 40 28
Loving ~19 3 ~19 ~11 22 ~l ~19 26 28 21
Capable ~l ~30 0 17 3 ~1 ~2 ~19 ~44 ~26
Independent ~12 ~27 5 ~8 ~3 ~10 ~9 ~33 ~38 ~21
Honest (R) ~5 ~5 ll 3 1 3 ~19 ~2 ~31 0
Ambitious ~0 ~7 24 19 23 5 14 ~36 ~36 ~33
Self-controlled 21 7 6 25 6 5 38 ~33 ~31 ~27
Cheerful (R) 19 ~0 12 20 6 ~17 21 ~28 ~20 ~13
Logical ~2 ~5 ~2 ~l ~10 ~20 ’1 ~34 ~31 ~16
Courageous ~28 ~27 ~18 ~13 ~2 ~6 ~5 6 3 ~2
Submissiveness 69 61 51 49 35 34 60 ~22 ~12 ~12
Benevolence ~25 6 ~42 ~31 ~0 20 ~25 60 64 44
Ambition ’9 ~34 20 16 12 ~2 ~8 ~45 ~74 ~40
Self control 23 1 9 26 1 ~19 34 ~57 ~49 ~33
Courageous ~28 ~27 ~18 ~13 ~2 ~6 ~5 6 3 ~2

 

 

* (R) Indicates that the variable has been reflected or

reverse scored to conform to other variables in
the cluster.

50

 

 

 

 

 

 

 

 

 

 

Self Factors Corrected
Ambition Control for Attenuation

21 29 27 19 36 22 31 24 501 502 503 504 505
~l ~12 ~5 ~0 21 19 ~2 ~28 69 ~25 ~9 23 ~28
~30 ~27 ~5 ~7 7 ~0 ~5 ~27 61 6 ~34 1 ~27
0 5 ll 24 6 12 ~2 ~18 51 ~42 20 9 ~18
l7 ~8 3 19 25 20 ~l ~13 49 ~31 16 26 ~13
3 ~3 l 23 6 6 ~10 ~2 35 ~0 12 l ~2
~1 ~10 3 5 5 ~17 ~20 ~6 34 20 ~2 ~19 ~6
~2 ~9 ~19 14 38 21 ~l -5 60 ~25 ~8 34 ~5
~19 ~33 ~2 ~36 ~33 ~28 ~34 6 ~22 60 ~45 ~57 6
~44 ~38 ~31 ~36 ~31 ~20 ~31 3 ~12 64 ~74 ~49 3
~26 ~21 0 ~33 ~27 ~13 ~16 ~2 ~12 44 ~40 ~33 ~2
17 15 28 23 l l4 l9 4 ~4 ~53 41 20 4
15 19 31 22 13 9 ~0 ~7 ~17 ~54 43 13 ~7
28 31 43 32 7 ~l 15 ~6 ~3 ~20 66 12 ~6
23 22 32 26 17 19 ~l ~5 22 ~62 51 21 ~5
1 13 7 17 34 38 25 -19 3o -54 19 58 -19
14 9 ~l 19 38 42 29 5 17 ~36 20 65 5
.119 ~0 15 ~l 25 29 21 ~28 ~12 ~49 16 44 ~28
4 ~7 ~6 ~5 ~19 5 ~28 100 ~28 4 ~7 ~25 100
~4 ~17 ~3 22 30 17 ~12 ~28 100 ~27 ~l 21 ~28
~53 ~54 ~20 ~62 ~54 ~36 ~49 4 ~27 100 ~94 ~83 4
41 43 66 51 19 20 16 ~7 ~1 ~94 100 33 ~7
20 13 12 21 58 65 44 ~25 21 ~83 33 100 ~25
4 ~7 ~6 ~5 ~19 5 ~28 100 ~28 4 ~7 ~25 100

 

51

attributes most likely to offend others: intellectual,

imaginative, broadminded.

Benevolence

 

The Benevolence cluster has a coefficient alpha of
.57 and consists of the values helpful, forgiving and loving.
Ambition

The Ambition cluster has a coefficient alpha of .57
and consists of the values capable, independent, honest

(reflected) and ambitious. The striking thing about this

 

cluster is that it shows that if a person places a high value

on capable, independent and ambitions, then he is likely to

 

place a very low value on honesty.
Self Control

The Self Control cluster has a coefficient alpha of
.57 and consists of the values self control, cheerful

(reflected) and logical.

Courageous

 

The value Courageous did not go well with any of the
instrumental value clusters and so was left as a singleton.
It has low negative correlations with the Submissiveness and
Self Control clusters implying that people who place a high
value on courage tend to see themselves as being somewhat

unconventional and impulsive.

52

There are two large correlations between various
instrumental value clusters. PeOple who place a high value
on Ambition tend to place a low value on Benevolence (and
Vice versa). People who place a high value on Self Control
also tend to place a low value on Benevolence (and vice
versa). PeOple who place a high value on both Ambition and
Self Control pay almost no heed at all to Benevolence.

Table B2 presents a cluster analysis of the terminal
values. There is clearly far less correlation among the
terminal values than among the instrumental values. This
may reflect a greater degree of independence among the values
or it may reflect the much higher degree of abstraction in
these terms, i.e. greater susceptibility to semantic contami-

nation. In any case, the clusters are generally much poorer.

UtOpianism

 

The Utopianism cluster has a coefficient alpha of

.70 and consists of the values a world at peace, a world of

 

beauty and equality

Hedonism

The Hedonism cluster has a coefficient alpha of .63

and consists of the values a comfortable life, happiness and

 

 

pleasure.

Misogamist

 

The Misogamist cluster has a coefficient alpha of .41

53

TABLE B2.-~A cluster analysis of Rokeach's 18 terminal
values. The value score correlated was formed
by averaging the three ranks assigned to that
value on the three administrations and is
therefore much more reliable than a single
ranking. The diagonal entries in the table are
the communalities of each value determined by
the cluster to which it is assigned. The
factors were calculated with communalities and
therefore the factor loadings and the factor
correlations have been corrected for attenua-
tion. All 116 subjects were used in the correla-

tions.

 

(I

 

 

 

 

 

 

 

 

 

 

 

M.
Utopianism Hedonism Misogenist L.
4 5 6 l 9 13 18 7 17 11

A world at peace 66 48 52 ~26 ~31 ~29 ~2 24 ~13 ~26
A world of beauty 48 34 33 ~27 ~18 ~9 9 16 ~10 ~14
Equality 52 33 38 ~39 ~34 ~35 8 l7 ~5 ~20
A comfortable life-26 ~27 ~39 29 26 41 ~33 ~19 ~17 ~3
Happiness ~31 ~18 ~34 26 29 41 ~21 ~10 ~16 2
Pleasure ~29 ~9 ~35 41 41 56 ~25 ~12 ~2 l7
Wisdom ~2 9 8 ~33 ~21 ~25 17 20 16 ~4
Family security

(R)* 24 16 17 ~19 ~10 ~12 20 26 21 ~15
True friendship ~13 ~10 ~S ~17 ~16 ~2 16 21 18 6
Mature love ~26 ~14 ~20 ~3 2 17 ~4 ~15 6 100
A sense of

accomplishment ~0 ~7 ~5 7 ~21 ~22 4 3 ~17 ~23
Self respect ~33 ~22 ~19 ~12 ~11 ~13 6 8 ll 1
Social recogni-

tion ~35 ~37 ~31 36 l 26 ~0 ~0 ll 4
Freedom 19 ~6 18 ~11 ~10 ~12 3 31 2 ~17
Inner harmony (R) 21 l 15 20 ~2 6 ~9 ~7 ~3 ~14
Salvation ~16 ~21 ~3 ~25 ~11 ~31 ~12 ~3 ~8 ~12
National security 11 ~15 9 ~12 ~2 ~27 ~15 ~1 ~14 ~29'
Exciting life (R) ~5 4 24 ~18 ~5 ~10 7 ~20 ~0 3
Utopianism 83 57 61 ~46 ~42 ~36 7 28 ~14 ~30
Hedonism ~48 ~30 ~59 53 53 76 ~43 ~22 ~19 9
Misogamist 7 ll 16 ~52 ~36 ~29 40 50 42 ~9
Mature love ~26 ~14 ~20 ~3 2 17 ~4 ~15 6 100
Status ~74 ~71 ~59 33 ~33 ~8 ll 11 6 ~20

. Egocentrism 36 ~5 30 8 ~10 ~5 ~5 21 ~l ~28
g Conservatism —7 ~22 19 ~36 ~12 ~45 ~14 ~15 ~15 ~25

 

* (R) Indicates that the variable has been reflected or

1n 1.1.-

r9172:- 29 cnnrofil +.~-.

.~Am5nrm 4-n nd-knv- "a...' “-1..-

...1-

54

 

 

 

 

 

 

 

 

 

 

 

 

Egocen- Residual
Status trism Set Factors

3 15 16 8 10 14 12 2 501 502 503 504 505 506 507
-0 -33 -35 19 21 -16 11 -5 83 -48 7 -26 -74 36 -7
-7 -22 -37 -6 1 -21 -15 4 57 -30 11 -14 -71 -5 -22
-5 -19 -31 18 15 -3 9 24 61 -59 16 ~20 ~59 30 19
7 -12 36 -11 20 -25 412 -18 -46 53 -52 -3 33 8 -36
-21 -11 1 ~10 -2 911 ~2 -5 -42 53 -36 2 -33 ~10 ~12
-22 -13 26 -12 6 -31 -27 -10 -36 76 -29 17 -8 -5 -45
4 6 -0 3 -9 -12._15 7 7 -43 40 -4 11 -5 -14

3 8 -0 31 -7 -3 ~1 -20 28 -22 50 ~15 11 21 -15
-17 11 11 2 -3 -8 ~14 -0 -14 -19 42 6 6 -1 -15
-23 1 4 -17 -14 -12 -29 3 -30 9 -9 100 ~20 ~28 ~25
18 9 13 ~18 -7 -8 -18 -9 -6 -20 -7 -23 43 -23 ~23
9 5 5 -5 -8 7 6 18 ~37 -19 19 1 19 -12 20
13 5 10 -15 1 -15 -15 -0 -51 35 8 4 30 -12 -20
-18 -5 -15 33 29 -0 13 -5 15 -18 27 -17 -40 56 5
-7 -8 1 29 33 5 19 -15 19 14 -14 -14 -16 56 6
-8 7 -15 -0 5 30 28 25 -20 -37 -17 ~12 ~18 4 55
-18 6 -15 13 19 28 26 23 2 -23 -23 -29 ~29 29 51
-9 18 -0 -5 -15 25 23 22 11 -18 -10 3 10 ~18 46
-6 -37 -51 15 19 -20 2 11 100 -68 17 -30-102 30 -5
-20 -19 35 -18 14 -37 -23 -18 -68 100 -64 9 -5 -4 -51
-7 19 8 27 -14 -17 -23 -10 17 ~64 100 -9 21 12 -33
-23 1 4 -17 -14 -12 -29 3 -30 9 -9 100 -20 -28 -25
43 19 30 -40 -16 -18 -29 10 -102 -5 21 -20 100 -51 -24
-23 -12 -12 56 56 4 29 -18 30 -4 12 -28 -51 100 10
-23 20 -20 5 6 55 51 46 -5 -51 -33 -25 -24 10 100

 

55

and consists of the values wisdom, family security (reflected)

 

and true friendship. The striking thing about this cluster

 

is that subjects who assign a high value to both wisdom and
true friendship are likely to assign a low value to family
security. Is there a subset of the subjects who would pro~
duce such a pattern of correlations? Well consider homo-
sexuals, members of the Women's Lib movement, confirmed miso-

gynists, men contemplating priesthood, etc.

Mature Love

 

The value mature love had only low negative correla~

 

tions with the other clusters. Therefore it was left as a

single item.

Status

The Status cluster has a coefficient alpha of .23 and

consists of the values a sense of accomplishment, self respect

 

 

and social recognition. Obviously this cluster was largely

 

determined by two criteria: content and the elimination of
alternatives. However, all three values do show a tendency

to correlate similarly with other values.

Egocentrism

 

The Egocentrism cluster has a coefficient alpha of

.45 and consists of the values freedom and inner harmony

 

(reflected). The fact that freedom and inner harmony are

 

negatively correlated means that those who place a high

56

 

value on freedom tend to place a low value on inner harmony.
Who would place a low value on the need for inner harmony?
Presumably those who never suffer from inner discord. But
the only people who never suffer from discord are those for
whom there is no inner conflict, i.e. those with no con-
science. Stated another way, those whose inner harmony stems
from perfect concordance between conscience and behavior have

no need for external freedom.

Conservatism

 

The Conservatism cluster has a coefficient alpha of

.50 and consists of the values salvation, national security

 

 

and an exciting life (reflected).

 

In considering the correlations between clusters, it
is important to recall that since communalities were used,
the correlations between clusters have been corrected for
attenuation. In particular this means that the correlations
for Status are grossly overestimated.

The large correlations all center about Hedonism.
Hedonism is negatively correlated with Utopianism (r-.68),
with Misogamist (r-.64), and with Conservatism (r-.51), i.e.
all are strongly in the expected direction.

There are three small correlations. Utopianism cor-
relates negatively with Mature Love (r—.30) and positively
with Egocentrism (r-+.30) as one might predict from the some—

what slower social development among intellectuals. Finally

57

Conservatism correlates negatively with Misogamist (r*.33) as

one might well expect.

APPENDIX C

A DETAILED COMPARISON OF THE LU

AND CONTROL GROUPS

The first point which emerges from a detailed con-
sideration of the data is that the experimental manipulation
had little or no effect on the LU group. The statistical
support for this assertion is shown in Table Cl. This table
contains point biserial correlations between the change
scores and all the various Bernoulli variables that could be
defined by a 2x2x2x2x2 analysis of variance with factors of
control-experimental and the initial levels on the values

equality and happiness, and the initial levels on attitudes

 

towards civil rights and labor unions. Each initial vari-
able was divided at the mean to provide a high-low contrast.
Each point biserial correlation corresponds exactly to the
named "effect" in the analysis of variance. The square of the
correlation is approximately the percent of variance explained
by that effect and would have been exact if all the cells had
had exactly the same number of subjects. A significant posi~
tive correlation in the T column would mean that the LU group
is significantly higher on the row variable, i.e. had a sig~
nificantly more positive mean change score. A significant
negative correlation in the E column would mean that sub-

jects who initially placed a high value on equality had a
58

59

TABLE Cl.~~The evidence favoring the view that the Labor Union
Group is not different from the Control Group:
biserial correlations between variable change scores

(rows) and Bernoulli variables defined by dichotomiz-
ing the initial levels of the values e ualit and
happiness, the attitudes towards civil rigfits and

Point

labor unions, and the treatment (LU group, +1; control

group,

-l).

The interactions of these five variables

were also calculated using Bernoulli variables.
the CR group was not relevant to this analysis the
sample size for these correlations is only 64.

Since

 

 

 

4..)

c

w

E

P U A

m x x

m m m (J .3 o A A m m

u C! g I: m x x >< x x x x x

a a: U A m ca a: m m U m m
Establishment ~17 6 12 8 ~2 ~3 ~l 13 ~0 12 ~1 3 11
Labor unions ~5 8 6 ~2 ~36 0 ~23 ~19 ~8 17 ~4 ~3 15
Civil rights 6 ~8 ~13 ~50 2 ~5 ~23 ~10 ~14 ~5 ~4 ~14 2
Repression 6 l3 l9 3 17 26 11 21 5 ~0 10 15 ~19
Utopianism 9 ~7 ~23 ~3 ~l ~20 19 ll 7 ~23 ~11 3 l7
Hedonism -6 l ~4 ~13 5 0 ~9 ~5 ~3 9 13 6 ~17
Misogamist ~15 3 ~15 3 3 ~12 ~20 ~2 l4 9 ~13 ~0 0
Mature love ~2 ~14 ~1 ~2 ~22 ~l l 4 ~5 ~27 ~3 ~9 ~11
Status ~l ~6 ~6 ~5 14 ~20 9 ~8 ~2 20 ~3 ~5 ~2
Egocentrism 9 ~16 7 ~16 ~6 ~15 9 8 ~1 l7 7 7 39
Conservatism -7 17 23 ~21 5 9 ~27 8 l ~2 ~9 6 27
Submissiveness 2 6 9 12 3 16 7 0 ~2 7 ~0 12 4
Benevolence ~23 ~8 ~9 8 7 ~17 2 ~8 ~5 ~9 ~4 ~16 ~25
Ambition ~9 9 3 l7 5 12 l 5 23 30 ~15 18 12
Self control 14 ~4 17 7 16 14 11 0 20 ~2 ~3 18 21
Courageous ~11 ~9 ~2 ~13 ~2 ~6 3 ~16 ~14 7 3 ~2 2

 

60

 

 

 

:2

A U A A A 0

>4 :4 >4 x x x

A o a: U A U A A :r: a: U o m
x x x x x x :4 x x x x x x
U :n: o A m m m :1: a: U m m m a: m
x x x x >4 x x >4 :4 x x :4 x x x
:1: ca E—u [-4 £4 £4 B E—« a [-4 a a E! B E-c
~13 - ~8 ~3 5 16 ll 2 19 ~23 2

~15 ~29 3 -1 -13 ~15 11 -12 11
10 -5 -5 9 6 -5 ~24 28 20

5 4 -5 12 16 ~21 -0 11 -1 8 23
9 -6 -3 7 9 16 5 -6 ~28 23 15
1 ~16 -1 ~20 ~17 ~10 ~23 10 14 0 6
2 -5 13 -4 -4 14 -1 -4 8 ~21 ~27

-15 6 -2 10 ~25 10 ~16 ~23 13 -2 13

13 7 23 ~6 ~1 -11 ~10 4 -9 18 25
-3 -9 -5 1 -2 ~10 ~20 -9 ~13 ~28 ~15
l3 2 -5 -4 12 0 14 ~4 3 12 -1
11 9 15 -1 -9 -6 1 l6 6 15 12
~12 ~26 5 10 -0 ~4 ~11 16 7 8 -3

 

61

significantly more negative mean change score on the row vari-
able than did subjects who placed a "low" value on equality.
The interpretation of a