THE EFFECT OF PERCEIVED CHOICE AND IMPORTANCE
ON THE RETENTION OF INFORMATION

Thesis for the Dome of M. A.
MICHIGAN STATE UNIVERSITY
David Ken? Kline
I966

 

 

 

 
     
     

I" LIBRARY
' Ivlic’nigan State

University

  

 

'3

‘3; U. I [I‘

’ \ 3 d u
f x
. . .
\ . |
' A
1

 

 

ABSIRACI

THE EFFECT OF PERCEIVED CHOICE AND

IMPORT.¥Cfi Cfi Tnn RthNIION OF INFORMATICN

by David Kent Kline

A number of weaknesses in dissonance theory formula-
tions and eXperiments have recently been pointed out. The
purpose of this exoeriment was to test the validity of some
of the basic propositions of dissonance theory under rigorous
conditions.

It was hypothesized that subjects committing them-
selves to a volitional choice of one of two equally attractive
but qualitatively dissimilar article titles, would retain more
information from the article chosen to be read than would sub-
jects reading the same article without a prior choice. In
addition, the effect or choice on retention would be greater
for subjects choosing between highly important titles than for
subjects choosing between titles of low importance. A final
prediction was that differences in retention would disappear
after dissonance was reduced.

Each subject was given a large manilla packet contain-
ing an instruction sheet and two sealed medium size packets.
The two medium size packets contained the same article with
different titles, and a small sealed packet in which was a 14
question multiple choice test on the article. Half of the

subjects chose between articles of either high importance or

WTV- :“_ ‘fi—H —-r— -' ,_-..

__.' ___-.__'__ .._ __ i .i , - -i—w__—._—._—.

David Kent Kline
low importance. The other half of the subjects were assigned
to read an article of either high importance or low importance.
After reading the chosen or assigned article the subjects
indicated their answers to the 14 multiple choice questions
on an IBM answer sheet. One week later the subjects took the
multiple choice test for the second time. In addition, the
subjects rated their perception of the eXperimental manipulations.

The analysis of variance revealed that the main effect
of choice and the interaction between choice and importance
were not significant. The main effect of importance and the
main effect of test-retest were significant.

The analysis of the scales measuring the subject's per-
ception of the experimental manipulations revealed that the
choice should have resulted in the arousal of dissonance. In
addition, it was determined that aroused dissonance should
have been reduced by the addition of new information. Since
no significant results were obtained which supported the
dissonance theory predictions, it was assumed that dissonance
was not aroused. Two possible eXplanations were offered for
the null finding: reinforcement theory and the lack of

qualitative dissimilarity in the choice alternatives.
, o
A? ' ”
Appmm- lewd /i his
Cgmmittee Chairman

Date: JAZMM 27,15?“

\

 

Thesis Committee:
Alfred G. Dietze, Chairman
James L. Phillips
William T. Stellwagen
Charles F. Wrigley

THE Eb‘FbC‘i‘ Or“ PERCLIVEJJ CnOlCt. AND IIVLPOKI'AL‘ECE ON 'fritl

KETEN’I‘IOLQ OF IUFOL‘L‘AALION

BY

David Kent Kline

A I‘i—LLSIS

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

.MASTER.OF ARTS

Department of Psychology

1966

Acknowledgments

The author wishes to express his gratitude to Dr.
Alfred Dietze, Dr. William Stellwagen, and Dr. Charles
Wrigley for the thoughtful guidance which they offered
during the preparation of this thesis. In addition, the
author expresses his appreciation to Dr. James Phillips
for his invaluable assistance in writing this thesis. The
author also thanks his wife, Marilyn, for her generous
assistance and patient understanding which made this

accomplishment possible.

11

LIST OF TABLES . .
LIST OF APPENDICES
BACKGROUND . . . .
EWDBLEM . . . . .
METHOD . . . . . .
RESULTS . . . . .
DISCUSSION . . . .
BIBLIOGRAPHY . . .

APPENDIX . . . . .

TABLE

OF CONTENTS

iii

Page

iv

61
67
75
84
93

96

l1“

h.‘l. (‘ I :1; "I :1”.
LIQJ. o 1.95.44.1322

bxoerimental groups into wnich subjects were
partitioned O O O O O O O O O O O O O O O I 0

mean differences between experimental condi-
tions on three rating scales . . . . . . . .

Jean difference in test scores for two
sections of advanced general psychology

courst". O O O O O O C C O O O O O I O I O O O

flaw score distribution for immediate teSt and
delayed test . . . . . . . . . . . . . . . .

means and standard deviations of number of
correct reSponses on the multiple choice

eXa'TIo o o o o o o o o o o o o o o o o o o o 0

analysis of variance of the number of correct
responses on the multiple choice exam . . . .

article titles constructed for different

importance categories . . . . . . . . . . . .

Z scores for paired comparison choices be-
tween 16 pairs of article titles . . . . . .

Correlated t values for differences in
importance within eight article title pairs .

Correlated t values for differences in
importance between article pairs . . . . . .

iv

77

77

60

pl

LIST OF APPENDICES

Appendix Page
A Method for selecting the experimental
article titles . . . . . . . . . . . . . . 97
B Measurement devices for pre-experiment . . 105
Section 1. Paired comparison form
for measuring article
title attractiveness.
Section 2. Importance ranking form.
C Verbal instructions for the eXperiment . . 108
D Packet instruction forms . . . . . . . . . 111
Section 1. Choice——high importance
form.
Section 2. No-choice-—high importance
form.
Section 3. Choice--low importance form.
Section 4. No—Choice-—low importance
form.
5 ArtiCleS O O O O O O O O O O O O O 0 O O O 115
Section 1. High importance article.
Section 2. Low importance article.
F Multiple choice questionnaires . . . . . . 123
Section 1. High importance questionnaire.
Section 2. Low importance questionnaire.
G Rating scales of subjects' perception of
experimental manipulations . . . . . . . . 130

BACKGROUND

A simple and heuristic attempt to explain the rela-
tion between cognition and overt behavior is Festinger's
“Theory of Cognitive Dissonance" (1957). Festinger defines
a cognition, or cognitive element, as any knowledge, Opinion
or belief that a person may possess about his self, his
behavior or his environment. Some examples of cognitive
elements are: a person's knowledge that he likes to smoke,
a person's belief that he should have good health, and a
person's knowledge that scientific evidence indicates that
smoking has a relation to some types of cancer. An analysis
of these cognitive elements reveals that some elements in-
volve value judgments (e.g., good health is desirable).
Other cognitive elements contain an affective relation (e.g.,
a person likes to smoke). On the other hand, a cognitive
element may have neither a value judgment nor an affective
relation, but only a factual component (e.g., smoking has a
relation to some types of cancer). The one thing that all
these cognitive elements have in common is that they are
general propositions about what a person does, believes,
feels, or knows. These general propositions, or cognitive
elements, may be irrelevant to each other, or some relevant
relation may exist between them. This relevant relation may
be one of dissonance or consonance. A dissonant relation is
one in which "the obverse of one element would follow from

1

2
the other” (Festinger, 1957). In order to understand the
exact nature of the dissonant relationship, the terms obverse
and follow from will be further defined.

According to Festinger the obversel of an element is
that which is not the element. In other words, the negation
of the element x is not-x. For example, the negation of the
knowledge that smoking has a relation to some types of cancer
is that smoking does not have a relation to some types of
cancer. With the assumption that a person desires good
health, the cognitive element which would logically "follow
from" the knowledge that a person likes smoking is that
smoking does not have a relation to some types of cancer. The
knowledge that smoking is related to some types of cancer,
therefore, is dissonant with the knowledge that one likes
smoking.

It is the meaning of the term "follow from“ which
Specifies the process by which dissonance arises. Four

different interpretations of "follow from" are discussed below.

 

1The term obverse, as it is used in logic, does not
imply the contradictory relationship between two propositions
which Festinger implies in his use of the term. Searles (1956)
in “Logic and Scientific Methods" defines obversion as "a form
of immediate inference in which the quality of the proposition
is changed from affirmative to negative or from negative to
affirmative, and the predicate of the original proposition
is contradicted”. According to this definition the obverse
of the proposition "All A is B" is "No A is non-B". There is
nothing contradictory between these two propositions since
they are both stating the same fact. The proposition "No A
is B", however, would be contradictory to "All A is B". This
is the type of relationship between two propositions which
Festinger calls obverse. The relationship between the
Pr0positions “All A.is B" and "No A is B" is better defined by
the terms negation or contradiction. In this paper we will
use the term negation to Specify this relationship.

 

First, dissonance could arise due to a logical in-
consistency. In the smoking example given above, Festinger
would say that the person's belief that smoking is not
related to some forms of cancer logically followed from the
knowledge that the person liked to smoke. The knowledge
that smoking is related to some forms of cancer, therefore,
is logically inconsistent with the knowledge that the person
liked to smOke.

Second, dissonance may arise due to cultural mores.

In America, the knowledge that one has more than one wife is
dissonant with the knowledge of cultural mores against polygamy.
However, in a different culture dissonance may not occur
between these cognitive elements.

Third, dissonance could arise because a Specific
opinion is included, by definition, in a more general Opinion.
If a person is a Catholic but attends a protestant ChurCh
phere probably is dissonance between these two elements of
knowledge. This dissonance arises because the concept of
Catholic usually carries with it the Opinion that one should
attend the Catholic church.

Fourth, dissonance may arise due to past eXperience.
PeOple learn to expect the occurrence of certain events
because of their systematic happening. For example, if a
person threw a ball up into the air and.it did not return to
earth, dissonance would probably occur. There would be a
dissonant relation between knowing that the'ball‘had'been

thrown up into the air, and knowing, ghrough past experience,

that the ball should, but did not, return to earth.

4

An examination of all the above examples of disson-
ance situations reveals that each Specification of the occur—
rence of dissonance rests on some implicit assumption about
related cognitive elements. Dissonance in the smoking
situation depended on the person desiring good health. In
the polygamous situation dissonance was dependent on the
premise that the individual accepts the cultural mores of
his society. The dissonant situation for the Catholic assumes
that the person desires to obey the directives of the
Catholic church or that there is a Catholic church in the
area in which the person lives. Finally, the occurrence of
dissonance in the case of throwing the ball up into the air
assumes that no extraneous factors, such as the ball getting
stuck in a tree, entered into the situation. In this con-
nection, Festinger points out that a person's needs, goals,
and beliefs are important factors in determining whether or
not two cognitive elements are dissonant.

The definition of consonance is quite easy to state
once dissonance has been defined. A relation of consonance
between a pair of cognitive elements exists when either
cognitive element in a pair does follow from the other
element. For example, the knowledge that one does not smoke
is consonant with the fact that smoking is related to some
forms of cancer, and the converse relation is also consonant.

If neither a certain cognitive element nor its

negation follows from another cognitive element in a pair,

 

5
then the relation between these elements is irrelevant. For
example, a person's knowledge that he is a man is irrelevant
to the knowledge that the sky is blue.

Festinger pr0poses three hypotheses based on the

concepts defined above.

1. The existence of dissonance is psydhologically
uncomfortable and will motivate the individual
to attempt to reduce dissonance and adhieve
consonance.

2. In states of dissonance the individual will
actively avoid situations and information whiCh
would probably increase dissonance.

3. The strength of the pressure to reduce dissonance
in the individual is a function of the magnitude
of the dissonance.

The following paragraphs will attempt to examine these hypo-
theses and their implications more fully. First, the vari-
ables whiCh determine the magnitude of dissonance will be
discussed. Second, an analysis will be made of the different
methods available to the individual for the reduction of
dissonance. Third, the conditions will be reviewed in which

a person will attempt to avoid situations increasing dissonance.
Fourth, the implications of dissonance theory for choice
behavior will be examined.

Magnitude of dissonance. The magnitude of aroused

dissonance is not the same for all dissonant relations. In

some conditions the amount of dissonance may be very large,

6
while in other situations an almost imperceptible amount of
dissonance may be aroused. Differences in the magnitude of
dissonance are determined by the importance of the cognitive
elements in a dissonant relation and the prOportion of
dissonant to consonant cognitive elements. An analysis of
these two determinants of the magnitude of dissonance is
presented below.

Festinger states that the greater the importance of
the two cognitive elements in a dissonant relation, the
greater the magnitude of the dissonance. No specific
theoretical or operational definition of importance is given
by Festinger, but he does imply that he means--of value or
consequence to a person. For example, most middle class
Americans, who value security and money, would find a
choice between which house to buy to be of greater importance
than a Choice between which brand of cigarettes to buy.

Some of the cognitive elements associated with buying a
house might be: a house must suit the needs of all the
family, a house must provide shelter and security for years
to come, and a house is one of the largest investments a
family makes. These cognitive elements are generally of
more consequence and value to a person than such cognitive
elements as: a cigarette may provide temporary enjoyment
for one person, a pack of cigarettes is a minor investment,
and a pack of cigarettes may last for a day.

In order to test the validity of the importance hypo-

theses a number of experiments manipulating importance have

been performed. Four of these experiments are described
below.

Cohen, Brehm and Latane (1959) told subjects they
were going to participate in a two person betting card
game. The subjects were informed that the probability for
winning was not the same for each of the two players and,
therefore, each subject could choose the side on which he
would play. The subjects were then divided into two groups.
Subjects in the high importance condition were told that
their name, choice of side, and amount of winnings would be
published in the undergraduate newSpaper. The subjects in
the low importance condition did not have to reveal their
name nor were they told that their choices or amount of
winnings would be published. The high importance manipula-
tion was empirically verified by keeping count of the sub-
jects who made written calculations before choosing the side
on which they would play. The eXpectation was that the more
important the situation was for the individual the greater
the amount of written calculations he would make to aid him
in choosing. The results indicated that the subjects in the
high importance condition made significantly more written
calculations than the subjects in the low importance group.
Before starting the experiment, each subject was told he
could change sides once during the course of the game but
only after paying a penalty. Thirty hands were dealt in a
game and after the twelfth hand play was stOpped. Dissonance

was expected to occur between a person‘s choice of sides and

 

8
the amount of his present winnings or losses. The subjects
that had been winning or had hardly lost any money, experi-
enced little or no dissonance. The participants that lost a
small amount of money were in a dissonant state but still
eXpected to win. However, when a large amount of money had
been lost, the persons experienced quite a bit of dissonance
but did not expect to win. Next, a graph was presented to
each subject showing the probabilities of winning for each
side. In all cases close examination of the graph revealed
to a subject that the side he had chosen would probably lose.
The eXperimenter recorded the amount of time each subject
Spent looking at the graph.

The eXperimental hypothesis was that the dissonant
subjects in the public condition would Spend more time
inSpecting the graph than the dissonant subjects in the private
condition. The data support this hypothesis for the small
loss group at the 10 percent level of significance, and for
the large loss subjects at the 5 percent Significance level.

Mills, Aronson and Robinson (1959) asked college
students to choose whether they wanted to take an objective
or an essay examination in their psychology course. The
Choice was to produce dissonance which would result in the
Subjects showing a preference for information that supported
their decision. The importance of the choice was eXperiment-
ally varied by making the exam worth 70 percent or 5 percent
of the student's grade for the course. In order to empirically

Validate the fact that a difference in importance was created,

 

9

each subject rated, on a six point scale, how difficult it
was to make the decision. The difference in difficulty
between importance conditions was significant at beyond the
.02 level. After each subject had indicated his exam choice,
a list of six article titles was distributed to each partici-
pant. The list consisted of three titles about objective
tests and three titles about essay exams. The titles con-
tained positive information about the two types of exams for
half the subjects. Negative information relating to the
exams was in the titles presented to the other half of the
subjects. The subjects were asked to rank the six titles in
terms of how much they would prefer to read articles
correSponding to the titles. The mean rank for the positive
article titles was different in the right direction from the
expected mean rank beyond the .001 level of significance.
However, no significant effects were obtained for either the
importance manipulation or the avoidance of negative article
titles.

Rosen (1961) repeated the eXperiment just described
with a modification in the procedure. The use of the words
“objective“ and "essay“ in the article titles was counter-
balanced between two lists. Each list contained three
fictitious pro-objective article titles and three fictitious
pro-essay article titles. In addition, one of the six titles
showed the advantage of essay over objective exams, while
another showed the converse. Importance was varied and

empirically verified in the same fashion as in the Mills, et al.

__.___ ___.._.— .,_' - .r— v - ~~~--- ‘— ' -

lO
experiment. The results failed to reveal a significant
difference in the importance manipulation. Rosen suggested
that the high importance condition may have been trivial
since only one-third of the subjects in this condition
scored in that half of the scale which indicated it was quite
difficult to make the decision. The resultant article ranks
showed a significant preference for selecting supporting
articles. No preference was revealed for avoiding dissonant
information.

A fourth experiment on dissonant behavior which
investigated the importance variable was conducted by Deutsch,
Krauss and Rosenau (1962). The subjects began by rating six
different kinds of jam on two nine point bipolar scales.

One scale was a measure of the over-all quality of each jam

as compared with other jams the subject had used. The second
scale was an evaluation of the flavor of the jams. A female
eXperimenter was then introduced to the subjects as an

eXpert on "food selection factors", who was going to conduct

a panel interview on "reasons behind food preferences". At
this point the subjects were given the importance manipulation.
The high importance group received a message stressing the
relationship between peoples' ability to judge subtle differ-
ences in the quality of foods and their creative judgment
ability. Creative judgment ability was suggested to be
related to leadership aptitude, executive potential, and
artistic judgment. The low importance condition did not
receive the importance creating message. Next, the experi-

menter selected two jams that on the over-all rating had

1”— """"' '-""" _,_ " *'

 

ll
received equivalent scale values. If possible, the jams
selected had rating values near the middle of the scale.
The experimental subjects then chose one of these two jams
to take home with them. Finally, all subjects rerated the
six jams on the two rating scales. The control group only
rated the six jams twice.

The eXperimenters proposed that post-decision
dissonance would result in a greater change in the rating
values in the high importance condition than in the low
importance condition. The results revealed a Significant
difference in the amount of rating change between the two
importance conditions for the flavor rating scale.

One other experiment that obtained greater dissonance
effects for the higher importance condition was Zimbardo's
(1960) eXperiment on communication discrepance. Importance
was defined as the individual‘s concern with the consequences
of his action.

Importance does seem to have an effect on the
magnitude of dissonance in some situations. The failure to
obtain significant results in all cases may partially be due
to the lack of a Specific operational definition of the term.

In determining the magnitude of dissonance it is
necessary to look at the relationship between all relevant
cognitive elements as well as to analyze the importance of
two relevant elements. Given the fact that all relevant
cognitive elements are equal in importance, Festinger states

that the magnitude of dissonance is a function of the

12
proportion of relevant cognitive elements that are dissonant
with a given cognitive element.

Brehm (1956) investigated the ratio of dissonant to
consonant cognitive elements using a group of college
students. The subjects were asked to rate the desirability
of eight different consumer articles (e.g., a toaster, a
painting, etc.). Each subject was told to judge the desir—
ability of the objects in terms of the objects' attractive-
ness, quality and the subject's need for the object. The
experimenter selected two of the rated products and had the
subjects in the experimental condition choose one of them.
After the subjects in the experimental condition had chosen
the article they desired, each subject rerated the eight
consumer articles. In the control condition the subjects
rated the articles twice without making a choice. In the
high dissonance condition the two articles were nearly equal
in attractiveness. In the low dissonance condition, the two
articles were quite different in attractiveness. It was
eXpected that the less the discrepancy in attractiveness
between consumer articles, the more cognitive elements that
were dissonant with the high attractive article and, there-
fore, the greater the proportion of dissonant to consonant
cognitive elements. The hypothesis was that the greater the
Pr0portion of dissonant elements the larger the magnitude of
the dissonance. Dissonance could be reduced by increasing
the desirability of the chosen article and decreasing the

desirability of the unchosen article. In addition, the ratio

13
of dissonant to consonant cognitive elements could be
reduced by the addition of consonant elements about the
chosen alternative and the unchosen alternatives. Half of
the subjects in each dissonance condition, therefore, read a
research article, which listed good and bad points about
four of the consumer products, including the two products
involved in the previous choice. The other half of the
subjects read research reports about four non-choice products.
The former group was called the information condition and the
latter group the no information condition.

The results indicated that the more equally attrac-
tive choice condition produced a greater enhancement of the
Chosen article and a greater devaluation of the unchosen
article. This effect was significant, at the .05 level, for
only those subjects reading the research article which
included information about the chosen product. The eXpect—
ation that there would be greater dissonance reduction in
the information condition than in the no information condition
was not supported.

A somewhat different approach to testing the effect
of the ratio of dissonant to consonant cognitive elements on
the magnitude of dissonance was made by Brehm and Cohen
(1959b). The main premise was that the magnitude of
dissonance is a function of what one has to give up compared
to what one obtains. Sixth grade students were asked to

indicate their liking for 16 toys. The subjects then chose

14
the toy they liked the best. Half of the subjects chose
between toys with similar qualities (e.g., sets of table
games or metal craft sets). The other half of the subjects
chose between toys with dissimilar qualities (e.g., swimming
fins and a model of a ship). As can be seen in the examples,
the similarity — dissimilarity dimension is defined in terms
of the use made of the toys. Each of these conditions was
further subdivided into two groups. One sub-group chose
between two toys. The other sub-group chose between four
toys. The hypothesis was two-fold. First, the less the
similarity (i.e., the more attributes not held in common) of
the toys involved in the choice the larger the number of
cognitive elements that can be dissonant with the chosen
alternative and, therefore, the greater the dissonance.
Second, the greater the number of toys involved in a choice
the larger the number of cognitive elements that can be
dissonant with the chosen toy and, therefore, the greater the
dissonance. The magnitude of dissonance was expected to be
reflected in an enhancement of liking for the chosen toy and
a devaluation of liking for the unchosen toys.

The results revealed that beyond the .01 level of
Significance there was a greater change in liking scores, in
the eXpected direction, for the subjects choosing between
dissimilar toys, as compared to the subjects choosing between
Similar alternatives. The predicted difference in liking

Scores for the two groups choosing between different numbers

15
of alternatives was significant at the .02 level. Brehm and
Cohen, however, point out that the latter difference could
have been due to an extraneous factor. The difference in
initial liking for the chosen toy and most liked unchosen
toy was not equated for the two and four alternative con-
ditions. There was a difference in the equality of
attractiveness, therefore, for the two and the four alter-
native groups.

Empirical evidence has been given to demonstrate the
effects of importance and the proportion of dissonant to
consonant cognitive elements on the magnitude of dissonance.
Rarely does one of these variables affect the strength of
dissonance independently of the other variable. Festinger
proposed, therefore, that the total magnitude of dissonance
is a product of the importance of the relevant cognitive
elements and the ratio of dissonant to consonant cognitions.
In other words, dissonance increases as the number and/or
the importance of dissonant cognitive elements increases in
prOportion to the number and/or importance of consonant
cognitive elements.

Beduction of Dissonance. One of the basic proposi-
tions of dissonance theory is that the greater the magnitude
of the dissonance the more intense the pressure to reduce
dissonance. .As the pressure to reduce dissonance becomes
greater the individual is “motivated" to find a way to
reduce dissonance. Festinger states two basic propositions

for the reduction of dissonance.

16

1. The individual can reduce his dissonance by
decreasing the importance of all relevant
cognitive elements.

2. The individual can reduce his dissonance by
decreasing the proportion and/or the importance
of the dissonant cognitive elements (Festinger,
1957, p. 264).

It can be seen in these two propositions that the reduction
of dissonance involves manipulating the variables which
increase the magnitude of dissonance.

There are three methods by which the individual can
implement the above two prOpositions in order to reduce
dissonance: change the behavioral cognitive element, change
the environmental cognitive element, or add new cognitive
elements.

First, the individual can reduce his dissonance by
changing the behavioral cognitive element. For example, an
individual may have a behavioral cognitive element correspond-
ing to his desire to fly an airplane on a certain day. The
environmental cognitive element, however, may correspond to
the fact that the weather that day is very bad. In this
Situation dissonance probably exists between these two know-
1edges. The individual may reduce his dissonance by deciding
that the flight he wanted to make was not very important
after all, and therefore, he will not fly. The above
illustration is an example of reducing dissonance by decreas-

ing the importance of the behavioral cognitive element.

17

In most eXperiments designed to test this reduction
method the change in the behavioral cognitive element is
reflected in a revaluation of the choice alternatives. This
revaluation usually involves a change in both the importance
and proportion of dissonant elements. For example, Brehm's
(1956) experiment on the change in desirability of consumer
articles resulted in an enhancement of the chosen product and
a devaluation of the unchosen article. The devaluation of
the unchosen product can partially be attributed to a
decrease in importance, while the enlarged discrepancy be-
tween the chosen and unchosen product can be interpreted as
resulting in a smaller proportion of dissonant elements. This
Same method of dissonance reduction is illustrated in the
experiments, previously cited, by Brehm and Cohen (1959b)
and Deutsch, et a1. (1962).

If a person has sufficient control over his environ-
ment he may, secondly, reduce dissonance by changing the
environmental cognitive element. In the illustration of the
person wanting to fly when the weather was bad, this would
have meant changing the weather to acceptable conditions for
f1Ying. Due to man's lack of control over his physical
environment he rarely can change the reality corresponding
to the environmental cognitive element. However, in some
cases man can ignore the physical reality in order to reduce
dissonance. Even though changes in the physical environment
are difficult to make, changes in the social environment are

often possible. The person who thinks of himself as a great

18
tennis player but continually loses may surround himself
with poor tennis players that praise his skill. The addi-
tion of friends in agreement with a person's self concept
would reduce the proportion of dissonant cognitions by
adding consonant elements.

A third technique for reducing dissonance is the
adding of new consonant cognitive elements. In the gambling
eXperiment by Cohen, et a1. (1959) most subjects in the high
importance condition attempted to reduce dissonance by seek-
ing consonant information from a probability graph. The
acquisition of cognitive elements consonant with the subject's
Choice was eXpected to reduce the proportion of dissonant
elements. Subjects in the public condition also could have
tried to reduce dissonance by adding the belief that few
people would see their names in the undergraduate newspaper:
thereby reducing the importance of the choice.

Ehrlich, Guttman, Schdnbach, and Mills (1957) pre-
dicted that, as a result of the dissonance aroused by the
decision to buy a car, new car owners would read more ads
about the car they had just purchased than ads about other

cars.

In addition, it was predicted that after a decision,

new car owners would read more ads about the car they owned

than would be read by old car owners about their cars.

Sixty-five males who had purchased a new automobile in the

Past four to six weeks and 60 adult males who had owned

their cars for three or more years were interviewed. The

19

experimenter found out, prior to the interview, what
magazines and neWSpapers each subject read regularly and
brought these to the interview. The subjects were asked,
at the interview, to indicate which automobile ads they had
noticed and read in the selected magazines and newSpaperS.
Finally, the interviewer aSKed the subject which makes of
car he had seriously considered before making his purchase.
The results indicated that new car owners read more ads
about the automobiles they purchased than ads about other
cars. The difference was Significant beyond the .01 level.
In addition, beyond the .01 level of significance, a greater
proportion of new car owners, than old car owners, read ads
about the car they purchased.

Jecker (Festinger, 1964) paid subjects to partici-
pate in a competitive war game involving strategy and
tactics. Each subject played in a group of three people:
the other two players being confederates of the experimenter.
Each subject in the group was given a test measuring the
relationship between personality factors and problem solving.
The two confederate "subjects“ left the room, supposedly to
take the test, but did not return. From this point on in the
experiment two different treatment conditions were imple-
mented.

Half of the subjects were instructed to Choose one of
the other "subjects" as their partner (Choice condition).
The other half of the subjects indicated which one of the

other "subjects" they would like to play with but were told

20

that their choice was subject to agreement by the person
chosen (uncertain condition). It was possible, therefore,
in the latter condition that the subject would not get his
choice of partners. The subject then viewed 6 slides con-
taining the alledged description of the problem solving
behavior of each confederate. Three of the Slides contained
information indicating good ability in problem solving and
three of the slides contained information indicating poor
problem solving ability. The experimenter recorded the
amount of time a subject spent viewing the slides. Finally,
a test was administered to measure the subjects' retention
of the descriptive information on the Slides.

Dissonance was expected to be reduced by adding con-
sonant cognitive elements from the slides. The hypothesis
was that the choice subjects would spend more time viewing
the slides than the uncertain subjects. This hypothesis was
based on Festinger's revised proposition that dissonance is
the result of making a choice of consequence (this concept
is more fully explained in the next section of this chapter).

An analysis of the data revealed that the choice
subjects Spent significantly (p=.05) more time viewing the
slides than the uncertain subjects. In addition, the results
Showed that the choice subjects retained a greater amount of
supporting information than the uncertain subjects. The
mean difference test was significant at the .05 level.

A study by Adams (1961) contributes a different kind

of information relative to reducing dissonance by adding

21

cognitive elements. Mothers of first grade students were
interviewed on the subject of hereditary versus environmental
factors in Child develOpment. The mothers were asked to
indicate which of the two factors had the greater influence
on child development. The experimenter then requested the
subjects to listen to an 18 minute tape recorded talk by an
alledged expert. For the experimental group the talk
favored either the heredity or environment concept: which
ever was contrary to the mothers' eXpressed opinion. In the
control group the talk was in agreement with the mothers'
expressed opinion. Subjects were then told that they could
hear further discussions on either hereditary or environ-
mental approaches at the university the following week.
Since both topics were scheduled for the same evening the
mothers were asked to indicate if they would like to attend
one of the sessions and, if so, whiCh one? At the conclu-
sion of the interview the subjects were aSked to again
scale their opinion on the subject of child development.

Adams proposed two hypotheses derived from dissonance
theory. First, a person exposed to an authoritative com-
munication expressing a contrary opinion, on an important
subject, is more likely to seek information than a person
eXposed to a compatible communication. The expectation was
that being eXposed to an opposing point of view on an
important subject creates dissonance by increasing the

proportion of dissonant to consonant cognitive elements.

 

22
This dissonance can be reduced by adding cognitive elements
whiCh are consonant with one‘s opinion. A comparison of
the subjects in the experimental condition, as opposed to
the control condition, revealed a preference for hearing
further discussion in the experimental group. The differ-
ence was significant of the .02 level. Adams concludes that
under conditions of free Choice, subjects attempt to reduce
dissonance by seeking information.

Adams' second hypothesis was that a person
experiencing dissonance as a result of eXposure to a con.
trary communication seeks information agreeing with his
opinion. This hypothesis was not supported by the data.

The experiment by Brehm (1956), reported earlier,
does not lend support to the hypothesis that dissonance can
be reduced by adding cognitive elements. After subjects had
chosen between consumer products they were asked to read a
research article containing favorable and unfavorable
information about the chosen product. The results indicated
that the reading of the information did not facilitate the
reduction of dissonance. This outcome may have been due to
offsetting effects of reading both favorable and unfavorable
information.

The previous paragraphs have discussed and shown
three different ways for reducing dissonance. In order to
empirically test the propositions of dissonance theory, how-

ever, it is necessary to be able to predict which method of

23
dissonance reduction will be used in a given situation. If
an individual wants to reduce dissonance by Changing a
cognitive element Festinger predicts that the cognitive
element which is most likely to change is the one with the
least resistance to change. One way of finding out the
resistance to change of a cognitive element is to analyze
the determinants of the resistance. Festinger proposes two
determinants of the resistance to change of a cognitive
element. First, the more new dissonance that will be pro-
duced by a change in a given element the greater the
resistance to change of that element. New dissonance is
determined by the proportion of present cognitive elements
that will be dissonant with the Changed or modified cognitive
element.

The second determinant of the resistance to change
of a cognitive element is the responsiveness of a cognitive
element to reality and the difficulty of Changing that
reality. The less control the individual has over his
environment and the more veridical the cognitive element is
in representing reality the harder it is to Change the
cognitive element.

The individual may want to reduce dissonance by
adding new c0gnitive elements instead of Changing old

elements. There are mainly two factors which determine the

Probability of using this method of dissonance reduction.

First, the more new dissonance which will be produced by the

24
new information the less likely it is that the subject will

add that new information. Second, if the new information is
not available to the subject then the subject can not make
use of the information. If the new information is
consonant information and readily available to the subject
it is most likely that he will reduce‘his dissonance by
adding cognitive elements.

The previous illustration of the pilot who wants to
fly might help to clarify the problem of determining whiCh
method of dissonance reduction to use. In this illustration
dissonance exists between the pilots desire to fly and the
fact that the weather is very bad. Dissonance could be
reduced by changing the weather to acceptable conditions for
flying. The pilot, however, probably is unable to change
this environmental cognitive element. Another means of
reducing the dissonance would be for the pilot to change the
behavioral cognition and decide not to go flying. It could
be, however, that the flight is imperative and the pilot
must make it if at all possible. The pilot then might
contact the control tower and add the cognition that the
storm is localized and that he can fly above it. The
addition of this new cognition would then reduce the dis-
sonance between flying and the bad condition of the weather.

It is also possible to use more than one method of
dissonance reduction in a certain situation. Steiner and
Johnson (1964) studied the effect of using two different

types of dissonance reduction methods. These researchers

25
suggested that dissonance is aroused when a person's belief
is contradicted by the expressed judgments of other people.
It was stated that this dissonance could be reduced in four
ways: conforming to the opinion of the other person(s),
rejecting the credibility of the other person(s) to make such
a judgment, underrecalling (repressing) the disagreements,
or minimizing the importance of the topics about which there
was disagreement. Steiner and Johnson proposed to force
subjects to use two of these reduction methods--conformity
and rejection. The expectation was that if conformity and
rejection are additive then the subjects should make less
use of the other two methods of reduction than if conformity
and rejection are not additive. Subjects were told they
would participate as a member of a two person group. The
second member of the group was a confederate of the
eXperimenters. The subjects were to verbally answer 32
questions, but they were not to reSpond until after the
confederate had answered the question. The acceptance-
rejection condition was Split into two groups. Half of the
subjects were given a set to accept the confederate by being
told the confederate was a good student and would be an
asset to the subject in the eXperiment. The other half of
the subjects were given a set to reject the confederate by
telling them the confederate was a poor student and would
not be of much help to them in the experiment. Each of
these two groups was further split into conformity—non—

conformity groups. The subjects in the conformity group were

26
told to reSpond to eight questions with answers similar to
the confederate reSponses, and to give the best possible
answer to the remaining 24 questions. The non-conformity
subjects were to give eight answers disagreeing with the
confederates responses, and give the best possible answer
to the other 24 questions. The confederate always gave the
same answers to the 32 questions. Consequently, the
subjects in the four conditions should have found themselves
in disagreement (dissonance) with the confederate approximately
half of the time. The hypothesis was that the reject-conform
subjects would use both rejection and conformity to reduce
dissonance and thereby make the least use of the other
reduction methods. The accept-nonconform subjects could use
neither rejection or conformity to reduce dissonance and,
therefore, should make the greatest use of the other reduc-
tion methods. The reject-nonconform and accept—conform
groups could each make use of one method of reduction and
should use the other methods to an intermediate degree. The
results supported these hypotheses at: the .001 level of
Significance.

It has been shown that dissonance can be reduced by
one method or a combination of reduction methods. In order
to predict the use of a single method of dissonance reduc-
tion the individual must control the use of all methods
except one. If this is not done one cannot be sure

dissonance was reduced in only one manner.

27

‘Ayoidance_gf Dissonance. The second basic

 

theoretical hypothesis preposed by Festinger was that persons
in a state of dissonance will actively avoid situations and
information that would likely increase dissonance. This
hypothesis has received the least eXperimental support of

the dissonance theory hypotheses.

In the previously cited eXperiments by Mills, et a1.
(1959), and by Rosen (1961), subjects indicated their
preference for reading articles about types of exams by
ranking the article titles. The results of both eXperiments
indicated that the titles containing information dissonant
with a subject's choice were not avoided. Other experiments
not revealing a tendency for dissonant persons to avoid
dissonance increasing situations are those by Adams (1961),
and Ehrlich, et a1. (1957).

A non—significant trend supporting the avoidance
hypothesis is revealed in the gambling experiment of Cohen,
et a1. (1959). The subjects experiencing dissonance from
large losings showed a tendency to spend less time viewing
the probability graph. The experimenters interpreted this
behavior as the avoidance of information in the probability
graph that would confirm the fact that the subject was losing.

A recent eXperiment by Mills (1965) does lend some
SUpport to the avoidance hypothesis. Subjects ranked their
preference for ten different consumer products. The eXperi—
menter then asked the subjects to choose one of two of the

products to keep. Next, the subjects rated their interest

28
for reading ads about each of the ten products. In order to
remove differences in interest due to unequal initial
desirability for the products, the eXperimenters adjusted
the interest ratings by linear regression. A linear
regression was calculated separately for each subjects'
interest ratings for the eight products not involved in the
choice. Predicted interest ratings were then calculated for
the chosen and rejected products using these regression
equations. On the basis of this statistical procedure Mills
was able to Show a significant (p .01) avoidance of dis-
sonant information.

Theoretical implications for choice behavior.
Festinger concludes, from his definition of dissonance, that
making a choice or a decision results in dissonance. This
dissonance arises between the knowledge of having made a
choice and the positive attributes of the unchosen alter-
native and/or the negative attributes of the chosen alter-
native. The individual who has just purChased a car has a
cognitive element corresponding to the fact that he chose
that car in preference to a number of other automobiles.

The cognitive elements which follow from this choice are the
positive attributes of the chosen car and the negative
attributes of the unChosen car(s). The cognitive elements
whiCh are the negation of the consonant elements just
described and, therefore, are dissonant with the Choice, are
the negative attributes of the chosen car and the positive

attributes of the unchosen car(s).

 

 

29

As previously stated, the magnitude of dissonance is
a function of importance and the ratio of dissonant to con-
sonant cognitions. In choice situations the effect of the
ratio of dissonant cognitive elements on the magnitude of
dissonance is manifested in two different manners. First,
the more equal in attractiveness are the alternatives in a
choice situation, the greater is the number of positive
attributes in the unchosen alternatives. According to
Festinger's hypothesis, therefore, the more equally attractive
are the alternatives in a choice Situation, the greater is
the dissonance resulting from a Choice. Brehm (1956) tested
this hypothesis in the experiment previously cited, by
having subjects choose between two different consumer
products. In one condition the two products were approx-
imately equal in attractiveness: in the second condition the
two products were unequal in attractiveness. The results
indicated a significantly greater amount of dissonance
reduction after choosing between approximately equally
attractive products.

A second factor influencing the ratio of dissonant
to consonant cognitive elements is cognitive overlap. The
amount of cognitive overlap between two alternatives is the
Proportion of common cognitive elements which they share.
The more elements shared by two alternatives the fewer the
number of dissonant elements that can exist. Festinger

Proposes that with the attractiveness and importance of two

Ch

.016

 

30
alternatives held constant, the greater the amount of cog—
nitive overlap between two alternatives the less the
magnitude of dissonance. For example, choosing between a
Ford with bucket seats and.the same Ford without buCket seats
would probably not create as much dissonance as choosing
between a Ford and a Chevrolet. In this example the two
Fords have many cognitive elements in common and, therefore, a
large amount of cognitive overlap exists.

Brehm and Cohen (1959b), in an experiment previously
cited, tested the cognitive overlap hypothesis by having
sixth grade students choose between qualitatively dissimilar
or qualitatively similar toys. The subjects choosing between
dissimilar toys showed a significantly greater amount of
dissonance reduction than the subjects choosing between
similar toys.

Four methods are available by which the individual
can reduce the dissonance resulting from a choice. First,
the individual can change or revoke his choice. Using this
method, the person who has recently purChased a new car can
reduce his dissonance by deciding he doesn't want the car
and, therefore, returning the automobile. The person might
then decide to buy a different car or he may not buy a new
car at all.

.A second means of reducing post~choice dissonance is
by changing the attractiveness of the Choice alternatives.
Changes in attractiveness usually take the form of a re-

valuation of the cognitive elements relevant to a choice.

31
This revaluation could be a change in the desirability of
the choice objects, or a Change in the liking for the
choice alternatives. The experiments of Brehm (1956),
Brehm and Cohen (1959b), and Deutsch, et a1. (1962) overtly
measured this revaluation by having subjects scale the
attractiveness of alternatives before and after a choice
between the alternatives. The pre-post difference in the
ratings was taken as an indicator of the change in
attractiveness. Significant changes in attractiveness were
obtained in all three of the above mentioned eXperiments.

A third method by which the individual can reduce
post—choice dissonance is to decrease the importance of the
relevant cognitive elements. In the Deutsch, et al. (1962)
experiment the high important subjects could have decided
that creative judgment ability was not an ability which
they desired to possess. This reduction in the importance
of the "self-involving" message would have reduced any
dissonance aroused after hearing the message.

The final method for reducing post-choice dissonance
is to add new cognitive elements. The persons eXperiencing
dissonance after the purchase of a new car in the Ehrlich,
et al., (1957) experiment reduced their dissonance by read-
ing advertisements relevant to their choice. Mills, et a1.
(1959) and Rosen (1961) reported that after individuals
chose between two types of exams they preferred to read
articles consonant with the exam chosen. Adams (1961) con-

cluded that persons experiencing dissonance, subsequent to

32
making a voluntary choice, attempt to reduce their dissonance
by seeking information.

The first section of this chapter has attempted to
review the basic postulates and implications of Festinger's
original dissonance theory formulation. In addition,
relevant evidence has been presented which both lends
support and fails to support these notions. The next
section of this chapter will discuss the modifications to
dissonance theory that have been proposed as a result of
this evidence.

Modifications in dissonance theogy

The large amount of research that has tested
dissonance theory has pointed out ambiguities in certain
theoretical formulations. These ambiguities center around
the lack of explicit Specification of the conditions under
which dissonance occurs and increases. For the most part,
Brehm and Cohen (1962) have examined these uncertainties
and have suggested possible solutions. The three formula-
tions in question are: the definition of dissonance, the
manner in which importance affects the magnitude of
dissonance, and the role of commitment and volition in post—
choice dissonance arousal.

Definition of dissonance. Festinger's original
definition of dissonance is an all-or-nothing proposition.
Either a cognitive element is the negation of another

cognitive element or it is not the negation of that element.

33
There is no variation in the degree to which a cognitive
element is in a state of negation: it is a discrete classifi-
cation. Brehm and Cohen, however, state that some of the
variables involved in a dissonant relationship have a con-
tinuous distribution. In the gambling eXperiment by Cohen,
et a1. (1959) for example, dissonance was declared to exist
between the person's choice of sides and the loss of money.
A person could lose varying amounts of money, however, and
according to the original definition of dissonance there is
no way to determine at what point this loss is great enough
to become dissonant (i.e., the negation of not losing money).
The eXperimenters' assumed that possibly no dissonance
existed for those individuals who lost only a very small
amount of money. Dissonance, instead, occurred when the
person's losses were still somewhat small but significant.
There is no way in which this assumption could have been made
on the basis of the original definition of dissonance.

In order to account for the existence of continuous-
ly distributed variables Brehm and Cohen prOpose a re-
definition of dissonance. Dissonance is defined to exist
when any variation from one cognitive element, in the direc-
tion or the negation1 of that element, follows from another
Cognitive element. According to this definition, the subjects
in the gambling experiment acquired some dissonance at the

moment they lost any money.

—_

lBrehm and Cohen, following Festinger, use the
lOgically incorrect term obverse instead of negation.

34

Importance. As previously stated, Festinger did not
explain what he meant by the importance of cognitive elements.
The assumption was made by this writer that an important
cognitive element was one which was of value or consequence
to an individual.

In the previous discussion of the manner in which
importance affects the magnitude of dissonance four
experiments were described. The experiments by Cohen, et a1.
(1959) and Deutsch, et a1. (1962) reported a significantly
greater amount of dissonance reduction in the higher
importance conditions. Mills, et a1. (1959) eXperiment and
the replication by Rosen (1961), however, found no differ-
ence in dissonance reduction between the importance condi-
tions. Brehm and Cohen (1962) suggest that the operational
definition of importance in the Mills, et a1. (1959) and
Rosen (1961) eXperiments accounts for their failure to
achieve significant results. The explanation given by Brehm
and Cohen is that exams may be commonplace to the student
and, therefore, even if they'make up a large prOportion of
the students' grade they will not be perceived as highly
important. There is still the problem in these experiments,
however, of why the two importance conditions were statis-
tically different on the overt measure of importance. A
possible explanation is that since importance was measured
bY the difficulty subjects had in choosing between exams, it
was not an uncontaminated measure of importance. Importance

undoubtedly contributes to difficulty but difficulty in

35

choosing could also be due to equal attractiveness or
failure to understand the purpose of the Choice. The other
two experiments manipulating importance operationally
defined the term in a different manner. Cohen, et al.
(1959) induced high importance by publicly announcing a
subject's name and his winnings or losses. The low import-
ant condition did not have this information made public.
Deutsch, et al. (1962) presented their high important sub-
jects with a message indicating they had creative judgment
ability if they could make subtle distinctions in the
quality and flavor of food. This message was not read to the
low importance subjects. Both of these experiments obtained
significant differences in dissonance reduction between
importance conditions. These two experiments differed from
the previous eXperiments in that they used importance in a
self-involving manner. The public announcement of winnings
and the suggestion of having or not having creative judg-
ment ability both reflect on the competence of the individual.

In accordance with the results obtained in the
Deutsch, et al. (1962) and the Cohen, et al. (1959)
experiments, Brehm and Cohen state that in order for import—
ance to influence the magnitude of dissonance it must
manipulate cognitive elements whiCh are related to the
individual's self-esteem or self-concept.

Commitment and volition. The act of making a choice
was previously stated to almost inevitably result in

dissonance. In the eXperiment by Jecker (Festinger, 1964)

36
an exception to this prediction was reported. The partici—
pants whose choice of a partner was subject to the approval
of the person chosen did not reveal any arousal of dissonance.
This result was interpreted by the experimenter as indicat-
ing that dissonance is not aroused by a choice which has no
immediate consequence. The subjects who were told they
would participate in the eXperiment with the partner they
chose did eXperience dissonance.

In the Deutsch, et al. (1962), Brehm (1956), and
Brehm and Cohen (1959b) SXperiments the subjects were told
they could keep the chosen article. The choices in these
experiments, therefore, had definite consequences for the
individual and thereby specified the choice at a given point
in time. significant results were obtained in all of these
experiments.

This Specification of a choice that has definite
consequences is what Brehm and Cohen (1962) call commitment.
They state that it is commitment which Specifies the
dissonant relation and thereby arouses dissonance. A person
is committed when he

has decided to do or not do a certain thing,

when he has chosen one (or more) alternatives

and thereby rejected one (or more) alternatives,

when he actively engages in a given behavior or

has engaged in a given behavior. Any one or a

combination of these behaviors can be considered

a commitment. (Brehm and Cohen, 1962, p. 7)
another aSpect of commitment might be that the more difficult

it is to revoke a choice the greater the commitment to that

choice. With this assumption the prediction is that the

37
greater the commitment to a choice the larger the magnitude
of the aroused dissonance.

In order to demonstrate the necessity of commitment
to the arousal of dissonance :rehm and Leventnal (Srehm and
Cohen, 1962) conducted an eXperiment varying the degree of
commitment. Subjects sat in front of a black curtain with
one of their hands extended under the curtain. The experi-
menter was on the other side of the curtain and told the
subjects he was going to place 15 different weights in
their hands and they were to assign a value to each weight.
In addition, after assigning a value the subjects were to
predict the average value of the series of 15 weights. In
actuality, the subject was only presented with 11 weights,
but it was necessary for him to believe he would make
judgments for more than 11 weights. a five dollar gift
certificate was offered as a prize for achieving a certain
level of accuracy in the prediction of the average weight
of the series. In order to anchor the subjects‘ judgments
the values 50 and 30 were assigned, reSpectively, to tne
first two weights. The discrepancy in grams between any two
weights was over four times as large between the tenth and
eleventh weights as between any other two weights. half of
the subjects followed the procedure described above. The
commitment variable was inserted for the rest of the sub-
jects between the judgment for the tenth and eleventh weight.
The commitment subjects were told, after assigning a value

to the tenth weight, that this next estimate of the average

 

 

38

weight of the 15 weights was the crucial prediction which
would determine whether or not they won the five dollar prize.
After making this estimate the subjects lifted the discrepant
eleventh weight and assigned a value to it. Finally, the
subjects again predicted the average weight for the series of
15 weights. The subjects were then told that there were only
11 weights instead of 15 and, therefore, the experiment was
completed. Dissonance was eXpected to be aroused by the
large discrepancy of the eleventh weight. The distortion in
the prediction of the average weight of the series after lift-
ing the eleventh weight was interpreted as a sign of dissonance
reduction. The results indicate a significant (p=.05)
distortion in the prediction of the average for the high
commitment condition but not for the low commitment condition.
Brehm and Cohen interpret this result as an indication of
the necessity of commitment for the arousal of dissonance.

Finally, Brehm and Cohen point out two main benefits
from the use of commitment to arouse dissonance. First,
commitment provides the Specific act of consequence at a
given point in time with which cognitive elements are
dissonant or consonant. This allows the experimenter to pre-
dict the relevant cognitive elements in a choice situation.
The second benefit of commitment is that it increases the
resistance to change of the cognitive elements associated
with the committed behavior. This greater resistance to
change of the committed behavior allows a more accurate pre-
diction of the means of dissonance reduction by reducing the

field of possible reduction methodS.

 

 

39

According to the above discussion, whenever a per-
son commits himself to a Specific choice, dissonance should
be aroused. Brehm and Cohen (1959a), however, conducted a
choice experiment with commitment and found no evidence of
dissonance reduction in one condition.

In this experiment subjects committed themselves
to participating in a boring, tedious task. None of the
experimental subjects received payment but each subject was
told that the other subjects received either one dollar or
ten dollars for participating in the experiment. The sub-
jects were told that a random procedure was used for
selecting the subjects who would be paid and they were not
selected. The expectation was that the larger the depriva-
tion in payment the greater the dissonance. The amount of
freedom each subject had in determining whether or not he
would participate in the experiment was varied by telling
some subjects they could leave with approval. The rest of
the participants were not given an opportunity to leave.
The degree of freedom was measured by assessing each sub-
ject's perception of how difficult it would be to get out of
participation. The measure of dissonance reduction was the
degree of satisfaction with the boring task. The results
showed that with low freedom the greater the deprivation in
payment the less the satisfaction. The subjects perceiving
high freedom of movement, however, revealed greater satis-
faction with an increase in deprivation of payment. Brehm

and Cohen (1962) interpreted this result as an indication

40
that even with commitment dissonance will occur only to the
extent the subject perceives himself to have prior freedom
of choice.

The name given to this concept, by Brehm and Cohen,
is volition. Volition is defined as "the extent to which a
person feels that he controls his own behavior (including
reSponses, emotions, motivations, etc.)“. (Brehm and Cohen,
1962, p. 201).

From the above discussion it would seem that volition,
in addition to commitment, Specifies certain conditions for
the arousal of dissonance. In addition, volition affects
the magnitude of dissonance.

Consider, for example, a two—alternative choice
situation. As the two alternatives become more
nearly equal in attractiveness, the proportion

of cognitions dissonant with choice of the more
attractive alternative increases, and, hence,

the magnitude of dissonance should increase.

But the same would be true of the degree of
volition, since, as the choice alternatives be-
come more nearly equal in attractiveness, the
individual must become more concerned with which
alternative he will choose. When one alternative
is clearly more attractive than the other, the
degree of volition would ordinarily be low,

since the individual would select the attractive
alternative without much consideration. The fact
that he could choose the less attractive alter-
native would be of little consequence. In general,
then, the degree of volition will frequently be
positively correlated with the ratio of dissonant
to consonant cognitions. Brehm and Cohen, 1962,
p. 203).

Both volition and commitment appear to be important
Variables in determining the arousal and magnitude of

dissonance. Brehm and Cohen have attempted to clarify the

41
interaction between these two concepts. These authors

state that

Two kinds of assumptions may be necessary:

one concerning commitment, the other concern—
ing volition. The interpretation of each
study requires that a commitment be identified
in order for there to be a clear case of
dissonant cognitions. Once a commitment is
identified, and assuming the existence of one
or more important cognitions discrepant with
that commitment, the degree of volition will
control the magnitude of dissonance, at least
in part. Where the degree of volition is low,
the magnitude of dissonance will necessarily
be small. Where the degree of volition is high,
the magnitude of dissonance will be large and
will vary as a function of other factors such
as the general importance of the relevant
cognitive elements, and the ratio of dissonant
to consonant elements. (Brehm and Cohen,
1962, p. 218)

Festinger (1964), in a re-assessment of dissonance
theory, admits that the evidence makes clear that
dissonance is not the inevitable result of almost all
choices. The commitment proposal of Brehm and Cohen is
accepted by Festinger and he states that a person must be
committed to a choice in order for it to produce dissonance.
The modifications to dissonance theory suggested by Brehm
and Cohen alter some of the basic formulations of dissonance
theory. At this point, therefore, a restatement of the
modified formulations of dissonance seems called for.

l. Dissonance exists when any variation in a
cognitive element in the direction of being the
negation of that element follows from another

cognitive element.

42

2. The existence of dissonance is psydhologically
uncomfortable and will motivate the individual
to attempt to reduce dissonance and achieve
consonance.

3. The strength of the pressures to reduce
dissonance is a function of the magnitude of the
dissonance.

4. In order for importance to affect the magnitude
of dissonance it must manipulate cognitive
elements which are related to an individual's
self-esteem or self-concept.

5. The magnitude of dissonance increases as the
number and/or importance of dissonant cognitive
elements increase in proportion to the number
and/or importance of consonant cognitive elements.

6. The total magnitude of dissonance is, in addi-
tion, proportional to the amount of volition a
person perceives himself to have in a given
situation.

7. The individual can reduce his dissonance by
decreasing the importance of all relevant cogni-
tive elements and/or by decreasing the propor-
tion of dissonant cognitive elements.

8. Commitment to a volitional choice results in
dissonance.

E;iticisms of dissonance theory
Dissonance theory has been accepted by a large

number of people as a valid eXplanation of certain kinds of

43

behavior. In addition, a good deal of research has been
generated by the theory. However, a closer look at the
evidence seems to be warranted. In the following sections
the theoretical formulations of dissonance theory will be
examined and then an analysis will be made of the methodo-
logical procedures used in many dissonance theory experiments.

Theoretical criticisms. Festinger‘s definition of a
cognitive element as being a knowledge, opinion, or belief
is as abstract as the concept it is attempting to define.
The components of a cognitive element still are not known.
In an analysis of Festinger's definition of cognitive
elements, Asch (1957) questions Festinger's assertion that
cognitive elements are independent of any affect. Especial-
ly since Festinger states that almost every dissonant
situation involves some assumption about a person‘s needs,
desire, values, etc. In an attempt to account for the
effectual component it was previously suggested that
cognitive elements seemed to be general propositions about
what a person does, believes, feels, or knows. There is
still a need, however, for an Operational definition which
explicitly specifies the components of a cognitive element.

,A second criticism of dissonance theory formulations
pertains to the number of elements which are involved in a
dissonant relation. Festinger states that dissonance is
the result of an inconsistency between Egg cognitive
elements. It would be very convenient if, in reality,

dissonance was the product of the relationship between only

44
two cognitions. EXperimentation on dissonance theory would
be much less equivocal if dissonance could be controlled by
the manipulation of just two cognitive elements. Chapanis
and Chapanis (1964) point out, however, that it is almost
impossible to reduce all the relevant variables in a complex
social situation to two cognitive elements. The best examples
of the fact that dissonance is not just an isolated relation-
ship between two elements are the eXperiments cited in
support of dissonance theory. In the Festinger (1957) and
Cohen, et al. (1959) gambling experiment, dissonance was not
just the result of an inconsistency between a person's Choice
and the amount of his losses. Other related cognitive
elements were the importance of the person's Choices and the
amount of negative information the person perceived in the
probability graph. Brehm and Cohen (1959b) asserted that
dissonance was the result of the inconsistency between the
choice of a toy and the negative attributes of that chosen
toy plus the positive attributes of the non-chosen toy.
There are more than two cognitive elements in this dissonant
relationship. The choice, the different attributes of the
chosen toy, and the various attributes of the non-chosen toy
form a cluster of relevant cognitive elements. Eadh toy is
not a single cognitive element but a group of elements made
up of the different attributes of that toy. Some of the
elements correspond to the attractiveness of the toy, some to
the physical properties, and others to the possible uses of

the toy. The notion of dissonance being the result of

 

45

inconsistency between two cognitive elements is parsimonious,
but not very useful in explaining behavior in real life
situations. It must be realized that dissonant relations
involve clusters of cognitive elements and that all of these
elements must be controlled and investigated in order to
rigorously test dissonance theory.

'lhe third criticism refers to that definition of
”follow from“ wnich states that dissonance is aroused by a
logical contradiction between two beliefs. As an example,
fleetinger offers the two beliefs: Man will in the near
future reach the moon (a): and Man will not be able to leave
the earth's atmOSphere (B). bestinger states that the
negation of s logically follows from A7 therefore, a and B
are logically contradiCtory. Drown (1965) points out that
in order for a logical contradiction to exist between two
propositions, one proposition must negate the other. In the
above example, the negation of A is the belief that man will
not in the near future reach the moon. PrOposition B in the
above example, therefore, is not the logical contradiction
of proposition A. brown states further that dissonance can
never be the result of a logical contradiction. In order for
dissonance to be aroused by a logical contradiction, the
individual would have to believe a certain orcposition and,
also, the negation of that proposition. recole usually do

I

not hold contraciccory beliefs. Brown s conclusion is that

dissonance theory does noc deal with logical inconsistency,

46
but instead, psychological expectancy. The reason that
propositions A and B, in Festinger's example, are usually
dissonant, is people expect that the only way to reach the
moon is to leave the earth's atmOSphere. As Brown suggests,
maybe a way could be found to bring the moon within the
atmosphere of the earth, in which case dissonance would not
occur between the two beliefs in the above example. The
relation between propositions that results in the arousal of
dissonance, therefore, is not a matter of logical contradic-
tion but of psychological expectation.

A fourth criticism refers to Festinger's failure to
adequately define what he means by the attractiveness of
choice alternatives. The experiments testing dissonance
theory usually define the tenn as either the desirability of
an object or the liking for an object. For example, Brehm
(1956) had subjects rate the attractiveness of consumer
products on the basis of the objects' desirability. Brehm
and Cohen (1959b) asked subjects to rate their liking for 16
different toys as a measure of attractiveness. There may be
little difference in the actual process being measured in
these two eXperiments. Desirability and liking may only refer
to different facets of the same process. besirability is a
property of the object and liking is a property of the per-
ceiver, but both to is may refer to a process of evaluating
a given perception. In any case, these two terms seem to

Specify what Festinger means by attractiveness.

47

In relation to the effect of attractiveness on the
magnitude of dissonance, Festinger differentiates between
attractiveness and importance. This would seem to be a
necessary distinction since in some cases the two variables
do differentiately effect behavior. A person may make a
relatively unimportant choice between two attractive alter-
natives. For example, a man may choose to buy one of two
attractive ties. Conversely, a person may make a relatively
important choice between two unattractive alternatives. For
example, an individual may be euite sick and have to choose
between having an Operation or staying in bed a month.

Just as one can think Of ways in which attractive-
ness and importance are uncorrelated, so can examples be
given Of situations in which the two variables are corre-
1ated. The man trying to select a tie may suddenly see two
of the prettiest ties he has ever seen. The act of making a
choice is then very likely to become more important for
this man. In the case of the sick person, the patient might
be told that he is going to die unless he has an Operation
or is confined to bed for a month. This situation has un—
doubtedly become very important to the individual and it is
very likely that the Operation and the confinement to bed
are considerably more attractive to the individual.

It would seem that even though there is some differ-
ence between attractiveness and importance there is also some
common variance accounted for by the two variables. he

mentioned above, attractiveness seems to refer to some

48
evaluative process. This evaluative process may be the pro-
cess which is common to both attractiveness and importance.
Osgood, Suci, and Tannenbaum (1957) in their develOpment of
the semantic differential state that the important-
unimportant dimension has a factor loading of .38 on the
evaluative factor. Even though Festinger distinguishes
between attractiveness and importance the two terms appear
to both be a part of an evaluation process. To the extent
that importance differs from attractiveness, importance
would seem to refer to the consequence, utility, or salience
of the relevant variables in a given situation. This dis-
tinction between attractiveness and importance is only
theoretical, however, and in order to apply it to dissonance
theory research some Operational distinction must be made.
Deutsdh's et al. (1962) experiment operationally distinguiShes
between the effects Of attractiveness and importance. Sub-
jects rated the attractiveness of six jam Spreads. Two
approximately equally rated jams were selected from which
the subjects chose one spread to take with them. Half of
the subjects were then presented with a self-involving
message to induce high importance. The other half of the
subjects did not receive the self-involving message. The
subjects than re—rated the six jam Spreads. Post-Choice
dissonance was reduced by Changing the attractiveness of the
two choice alternatives. The results showed a significantly
greater change in attractiveness in the high importance

condition than in the low importance condition. The

49
important message seems to have introduced some new variable
into the situation which resulted in a greater change in
attractiveness. This new variable had to be something other
than attractiveness, since the article read to the subjects
referred to the individual‘s abilities and not the Choice
Objects. There does seem to be a dimension in which
attractiveness and importance separately influence behavior.
The fact should always be taken into account, however, that
there is also a dimension in which changes in attractiveness
covary with Changes in importance.

Another criticism related to the manipulation Of
importance is that Festinger does not distinguish between the
importance of the Choice Objects, the importance of the
choice itself, and the importance of the cognitive elements.
All three of these factors would seem to affect the magnitude
Of dissonance. Festinger refers only to the importance Of
the cognitive elements. Of course, it can be argued that the
importance of the cognitive elements includes the other two
factors. That is to say, the importance of the choice itself
is determined by the importance of the choice Objects and in
turn the importance of the Choice Objects is determined by
the importance of the cognitive elements. Whatever the case,
in order to have an unequivocal definition of importance,
these three factors of importance should be either differ-
entiated or integrated.

It was previously stated that the magnitude of

dissonance was a product of the importance of the cognitive

50
elements and the proportion of dissonant to consonant elements.
The problem with this concept is that there is no way to
Operationalize it and thereby determine the total magnitude
of dissonance (Brown, 1965). In order to accomplish this
task one would have to be able to assess the number Of
relevant cognitive elements in a situation and the relative
importance of each cognitive element. Since no instrument
is available that can assess all the relevant elements in a
given situation the total magnitude of dissonance is not
definable. This is not to say, however, that the gross
relation between dissonant and consonant elements is not
discernable, nor that gross changes in magnitude can not be
measured.

AnOther criticism is that commitment has been estab-
lished as a necessary ingredient for the arousal of dis-
sonance but a Specific Operational definition is not available.
The present definitions of the term states that commitment
is the act Of making a choice or engaging in a given
behavior, or an act of definite consequence Specified at a
giVen point in time. These definitions do not Show commit-
ment to be much other than choice. Commitment is used in
dissonance theory eXperiments as a variable separate from
Choice and, therefore, an eXplicit Operational definition Of
commitment is needed in order tO fully understand its func-
tion in a dissonance situation (Deutsch and Krauss, 1965).

Deutsch and Krauss (1965) criticize dissonance theory

on the basis that even with the benefit Of commitment it is

51
still difficult to Specify the method of dissonance reduction.
In order to solve this problem the factors which determine
which method of reduction will be used in a given situation
must be explicitly defined. This means examining individual
differences in the use of different methods of dissonance
reduction. In addition, some way must be devised for deter-
mining the resistance to change of cognitive elements or Of
the amount of new dissonance created by adding cognitions.
Very little experimentation has been performed which investi-
gates these factors.

A final criticism of dissonance reduction refers to
some Of the measures used to assess attempts to reduce
dissonance by adding cognitive elements. The previously
described experiments frequently attempted to measure the
addition of cognitive elements by recording the time spent
reading the new information or by having subjects indicate
their relative preference for different material. Cohen,
et al. (1959) in their gambling experiment measured dis-
sonance reduction by recording the amount of time subjects
spent reading a probability graph. Mills, et al. (1959)
and Rosen (1961) had their subjects rank their preference
for a group of articles titles about types of exams. Even
though significant differences between groups were obtained
in these experiments neither time of reading or indication
of preference is a direct indication of dissonance reduc-
tion. Both of these measures are secondary indicators of

the reduction of dissonance.

52

The time spent reading relevant information does not
necessarily indicate the person was trying to reduce dis-
sonance. Instead, differences in reading time may have been
due to differential interest in the material or differential
difficulty in understanding the information. In addition,
it is very difficult to empirically determine the exact
amount of time an individual actually Spends reading a piece
of information. The person may have been distracted part of
the time or he may not overtly indicate exactly when he is
finished reading the material.

The use of a person's indication of his preference
for information is also, under most conditions, a poor
measurement of dissonance reduction. If a person overtly
indicates that he would prefer to read a certain type Of
information, but does not believe that he will have to read
the information, it cannot be assumed that his preferences
and his behavior coincide. There is no commitment to action
in just the indication Of preference. The person must
believe that his choice in stating a preference leads to some
behavioral consequence.

The fact that dissonance has been aroused in a
situation is usually inferred from the occurrence of attempts
to reduce dissonance. It is difficult to determine whether
or not dissonance has been produced if the measure of dis-
sonance arousal is not directly related to dissonance vari-
ables. If any understanding is to be obtained of the process

by whiCh dissonance is aroused and reduced primary measures

 

 

 

 

53
Of reduction must be used instead of ambiguous secondary
indicators.

There are other dissonance theory phenomena, which
are important to an understanding of the dissonance process,
that are ambiguiously defined or not defined at all. First,
very little investigation has been made of when dissonance
occurs after a choice, or what extinction process it follows.
Second, Maccoby and.Maccoby (1961) raise the question of
individual differences in tolerance for dissonance. Some
people may be able to endure larger amounts of dissonance
than others. Under certain circumstances people may seek out
dissonance. All of these ambiguities raise questions about
dissonance theory which must be investigated in order for
the theory to remain useful.

Methodological Criticisms. The research reported by
Festinger (1957) to support his original formulations was
criticized by AsCh (1958) for three general weaknesses.

First, there are some studies which yield positive
results for the theory but these results are not consistently
positive. In addition, even though the trends in the data
are statistically significant they are generally weak.

Second, there is a lack of adequate control of psycho-
logical factors that could account for the results as well as
dissonance theory. This criticism applies to some Of the
studies which provide the strongest confirmation for
dissonance theory. Third, there is not a uniform base for

the studies supporting dissonance theory. For example, a

54
lot of information is provided relevant to the effect of
dissonance on smoking behavior, seeking Of information and
Choice behavior. The problem is that most of the informa—
tion comes from different sources of investigation, i.e.,
different experimental designs or methods. AsCh says that
What is needed is the intense study of one type of dissonant
behavior using a uniform method of experimentation.

The research that has been performed from 1957
through 1961 to investigate dissonance theory has been
evaluated by Chapanis and Chapanis (1964). Two general
criticisms of dissonance theory research are offered by
these investigators. First, the eXperimental manipulations
in dissonance theory research are so confounded that it is
difficult at times to determine what variables produced the
changes in the dependent variable. Second, some of the
methods used in analyzing the data are highly suSpect. A
review of these two criticisms as they pertain to the experi-
ments previously described is given below.

Two eXperiments which Chapanis and Chapanis cite for
confounding the variables, beyond the point of clear inter-
pretation, are the gambling eXperiment by Festinger (1957)
and the modification of that experiment by Cohen, et al.
(1959). Three weaknesses are pointed out in these eXperi-
ments. First, Chapanis and Chapanis state that these
eXperimenters confounded their dissonance formulations.

Dissonance was stated to have been aroused, in the winners

55

of the game, between their choice of sides and the informa-
tion in the probability graph which said they were eventually
going to lose. For the losers, dissonance was postulated
to be the result of the inconsistency between the person's
Choice and the fact that he was losing money. Dissonance,
therefore, could be the result of an inconsistency between
choice and information of an inconsistency between Choice and
amount of losings. It is possible, therefore, that the two
sources of dissonance were additive and that dissonance was
increased for the subject who was losing When he acquired
the information telling him he was going to lose more. In
this experiment no attempt was made to separate out the two
different causes of dissonance arousal and consequently it
is not possible to determine the exact cause Of dissonance.

The second weakness which Chapanis and Chapanis
discuss is that the results were not consistent or unequivocal.
Dissonance for both the winners and losers was reflected in
the amount Of time Spent viewing the graph. The results
showed that the winners spent a moderate amount of time view—
ing the graph: the small losers viewed the graph the longest
time: the moderate losers avoided the graph: and the biggest
losers viewed the graph for a moderate amount of time.
These results were interpreted as supporting dissonance theory,
Chapanis and Chapanis point out that the results are not Con~
sistent with eaCh other and that the opposite pattern of
results for the losers could also have been interpreted as

SUpport for dissonance theory.

56

The third and most important criticism is that the
results can be explained by ore-decision behavior as well
as by post—decision behavior. The subjects were told that
they could change sides before the probability graph was
given tO them. The time Spent viewing the graph, therefore,
may have been an indicator Of the subjects attempt to Obtain
information to help him make a choice and not an indicator
Of dissonance reduction. Festinger and Cohen, et al., did
not attempt to control this predecision behavior.

Silverman (1964), in reSponse to the Chapanises'
criticisms of the gambling eXperiment states that the
behavior of the losers was predicted by the hypotheses: and
to that extent the results are relevant. In addition,
Silverman says that the explanation for dissonance in the
winners was made ex post facto. This exolanation, there-
fore, was a prooosal for further research and not a hypo-
thesis Of expected behavior. In the final analysis, however,
Silverman concedes that the study should be partially repli-
cated with dissonance resulting only from the Choice and the
amount of losings.

Three recommendations are made by Chapanis and
Chapanis for reducing the confounding Of variables in future
eXperiments. First, the eXperimental manipulations should be
Simplified. This could be accomplished by using fewer
variables and more rigorous controls in eXperiments. The
second recommendation is that more control groups be used in

order to clarify the interrelationships of the experimental

57
manipulations. Third, more importance should be given to
the subjects perception Of the experimental variables and
situation.

Chapanis and Chapanis Offered two criticisms of the
teChniques used in analyzing data Obtained in dissonance
theory experiments. First, the rejection of a large propor-
tion of subjects after the data is analyzed is a frequent
occurrence in dissonance theory eXperiments.

Brehm and Cohen (1959b) discarded 65 percent Of the
subjects in their study of the attractiveness of choice
alternatives (toys). The reasons given were as follows.
Some subjects liked the chosen toy too much for their to be
any significant increase in liking. Other subjects failed
to rate one toy higher enough than any other toy SO that its
choice could be eXpected. In addition, subjects were dis-
carded as "unreliable" when they failed to Choose the toy
initially marked as most liked. Chapanis and Chapanis point
out that discarding "unreliable" subjects falsely reduces
the error variance and changes the means in such a way as to
enhance the difference in attractiveness. In other words,
rejecting "unreliable" subjects produced the hypothesized
effects. Brehm (1956) discarded 35 percent Of his subjects
for similar reasons to Brehm and Cohen's (1959b).

Another experiment discarding a large number Of sub-
jects is Ehrlich, et al. (1957). The criticism of this

eXperiment is that subjects were thrown out at so many levels

58

of analysis that it is impossible to determine what the final
data means. Fifty percent of the old car owners, in the
category of "cars considered,“ were discarded because either
no advertisements for their car preferences appeared, or
they did not name any cars as "seriously considered.“ One-
third Of all the subjects in the experiment were thrown out
because they did not notice the advertisements which did
appear. Finally, those subjects were discarded who did not
notice at least one advertisement in each car category, or
who read an equal percentage Of advertisements in each cate-
gory. Up to 82 percent of this original sample was thrown
out at different levels Of analysis. Chapanis and Chapanis
state that this biased the results to the extent that they
are indeterminable.

In defense of the rejection Of subjects Silverman
(1964) says that subjects were not discarded in these
experiments on the basis of the dependent variable. Instead,
the subjects analyzed were preselected by a criterion which
enabled the hypothesis to be tested. However, even if pre-
selection of subjects by a criterion is a widely used pro-
cedure, it surely does not add to the validity Of the
results. Silverman does agree that in the Brehm and Cohen
(1959b) eXperiment the rejection of "unreliable“ subjects
Could have accounted for the results.

The second criticism of the technique used in analyz-
ing the data is that some studies are inadequately designed

and analyzed. In many studies no control groups, or an

59
insufficient number of control groups, were used. Without an
adequate number of control groups it is impossible to Obtain
an unequivocal interpretation Of the differences between
conditions. In the Ehrlich, et al., eXperiment no control
group was used tO determine the effect of reading car
advertisements before making a decision. The Mills, et al.
(1959) and Rosen (1961) experiments did not use a control
group in which persons' only scaled their desirability for
the articles but made no choice.

Another weakness in design is that an overall test
Of significance is frequently not used. The use Of a series
of t tests without a prior overall significance test many
times leads to a higher level of significance than is justi—
fied. In addition, without an overall significance test it
is quite difficult to determine if a significant inter-
action occurred. Two Of the previously Cited studies not
using an overall test or significance are Brehm and Cohen
(1959a) and Mills, et al. (1959).

‘rhe Chapanises' final general criticism of the
dissonance theory research is that many times non—significant
trends are reported and interpreted as support for the theory.
For example, Adams (l96l) interprets a .06 level of signifi—
cance as support for his hypOtheses. In the same manner,
Cohen, et al. state that support was Obtained for one Of
their hypotheses at the .10 probability level.

Most of the main propositions of dissonance theory

have been shown to have theoretical weaknesses and

 

6O
insufficiently clear eXperimental support. Further investi-
gation of the dissonance theory formulations under rigorous

conditions seems to be required.

Problem

The purpose of this study was to eXperimentally test
some of the formulations Of dissonance theory under rigorous
conditions. In doing this it was hOped that reliable data
could be Obtained about the propositions Of dissonance
theory.

Two basic propositions of dissonance theory were
tested in this experiment.

1. The existence of dissonance is psychologically
uncomfortable and will motivate a person to
attempt to reduce dissonance and achieve con-
sonance.

2. The greater the magnitude of the dissonance the
greater will be the pressure to reduce
dissonance.

As a test of these two postulates persons were asked to
choose between two alternatives in order to arouse dissonance.
The magnitude of the dissonance was manipulated by varying
the importance of the choice.

Choice. The act of making a choice is proposed to
be dissonant with the positive attributes of the rejected
alternative(s) and the negative attributes Of the Chosen
alternatives. Some of the positive attributes of the
rejected alternative(s), however, must be qualitatively

dissimilar from the attributes of the accepted alternative.

61

62

If the attributes of the two alternatives are quite similar,
i.e., cognitive Overlap exists, then the chooser is not
rejecting much in selecting either of the two alternatives.
If the chooser does not have to give up much Of anything
then there are few cognitive elements that can be dissonant.

In order for dissonance to be aroused the individual
must make a volitional choice to which he is committed. The
individual must believe that the choice he makes will affect
his sub equent behavior and that he had freedom in making
that choice.

Dissonance should be aroused, therefore, only when an
individual commits himself to a volitional choice between two
qualitatively dissimilar alternatives.

Magnitude Of dissonance. When a person perceives

 

himself to have a large amount Of volition in a choice
situation the magnitude of the dissonance is a product of the
ratio Of dissonant to consonant cognitive elements and the
importance Of these elements. The prooortion Of dissonant
cognitive elements in each member of a group of individuals
can be held constant by controlling two variables. First,
the choice must be between equally attractive alternatives,
i.e., the alternatives must be equally desirable and equally
liked. Second, all people must choose between alternatives
which are equally dissimilar in content. With these two
variables held constant the magnitude of the dissonance would
be the result of variations in the degree to which the

individuals perceived the situation as involving factors of

63

importance for them. The more important these factors the
greater the resulting dissonance. As previously pointed out,
attractiveness and importance are correlated. Variations in
importance, therefore, will produce some change in the
attractiveness of the alternatives. These changes will
equally affect both choice alternatives, however, and conse-
quently the two choice alternatives will still be equally
attractive.

Reduction of dissonance. One way in which the indivi-
dual can attempt to reduce the dissonance resulting from a
choice is for him to revoke or change his choice. If the
individual has publicly committed himself to a choice, how-
ever, it becomes very difficult for him to revoke or change
that Choice. In addition, the environmental condition may
make it almost impossible for a person to revoke or change
his decision. For example, a car dealer may have a policy
that no car can be returned once it has been purchased.

Secondly, the individual could attempt to reduce post-
choice dissonance by changing the attractiveness Of the
choice alternatives. The attractiveness of the choice alter-
natives can be changed by reducing the desirability of the
dissonant cognitive elements, and/or by reducing the liking
for the dissonant cognitions. In addition, increasing the
desirability Of the consonant cognitions and/or increasing
the liking for the consonant elements will reduce dissonance
through changing the attractiveness Of alternatives. These

procedures would be very difficult to implement if the

CH

CH

Msbu

64
alternatives were initially equally attractive and if the
attractiveness was substantial.

A third means by which the individual could attempt
to reduce post—choice dissonance is by reducing the import-
ance of the relevant cognitive elements. It would be quite
difficult to reduce the importance of these elements, how—
ever, if they were initially quite important.

If the above methods of dissonance reduction were
made very difficult or impossible to implement in a given
situation the person would have to use some other method to
reduce dissonance. If immediately after a choice, a person
was presented with an article containing information relevant
to his choice, he would probably reduce his dissonance by
adding the relevant information in the article.

Measures Of attempts to reduce dissonance. In pre-

 

vious eXperiments the attempts to reduce dissonance by add—
ing cognitive elements were usually measured by the time
spent reading the information or by an indication of prefer-
ence for relevant reading material or situations. as was
previously pointed out (p. 51), these measurements are only
ambiguous secondary indicators of attempts at dissonance
reduction. There should be a measure Of attempts at dis-
sonance reduction which is more directly related to the
variables manipulated in a dissonance experiment.

In the eXperiment by Jecker (Festinger, 1964) some
subjects were allowed to choose their own partner in a war

game while other subjects were told their selection had to be

65
agreed upon by the person selected. The problem solving
abilities Of the person not chosen were dissonant with the
subjects' choice Of the other person. In addition, the
attributes Of the chosen partner which were detrimental to
problem solving were dissonant with the choice of that
partner. The measure of attempted dissonance reduction was
the number OI items Of information retained by the subjects
about their partner's problem solving ability. The results

' making a choice retained a

indicated that the subjects
greater amount Of information about their partners than the
subjects whose selection was not assured Of having any con-
sequence. This is a measure Of an attempt at dissonance
reduction which is directly related to dissonance variables.
When the subject eXperiencing dissonance was presented with
information about the chosen partner he acquired that infor—
mation. Once this information was acquired it was accessible
to measurement through measuring retention. The more dis-
sonance the person possessed, the more cognitive elements
he added to reduce the dissonance and, therefore, the more
items of information he retained. It can also be argued that
as the dissonance is reduced the acquired information is no
longer needed to reduce dissonance and retention is reduced.
The differences in retention due to dissonance, therefore,
should disappear over time.

In summary, the formal hypotheses Of this experiment
are:

1. Subjects who have volitionally chose and committed

themselves to one of two equally attractive but

66
qualitatively dissimilar alternatives will re—
tain more Of the relevant information to which
they are exposed than subjects who are exposed
to this same information without having made a
prior choice.

2. The effect of choice on the retention Of
information will be greater when the choice is
between alternatives highly important to the
individual than when the choice is between
alternatives of low importance.

3. There will be no differences in the amount of
information retained between subjects making a
choice and subjects not making a choice after
the dissonance reduction process is complete.

In addition to the above hypotheses this eXperiment
will attempt to overcome the methodological flaws pointed
out in the previous Chapter. Overt measurements will be
made of the subject's perception Of the eXperimental situation.
The subjects not making a choice will serve as a control
group for the subjects m king a choice. Only two independent
variables will be manipulated in the experiment. These
variables are choice and importance. In the analysis of
results no subjects will be rejected after the data is analyzed
and the data will be analyzed by an overall test of signifi—
cance, Finally, an attempt will be made not to draw con-

clusions beyond the significance of the results.

hethod

The subjects in this experiment were 145 under-
graduate stuuents at michigan State University. During the
Spring term Of 1965 these subjects were enrolled in two
sections of an advanced general psychology course. The sub-
jects were members Of all four college classes, i.e., fresh-
man, SOphomore, junior, and senior. Four subjects were dis-
carded because they did not follow the instructions for the
experiment. The remaining 144 experimental subjects con-
sisted Of o1 males and 63 females.

1681C!“

V

*rhe design was a 2x2 factorial with a repeated
measurement for each subject. The first factor was divided
into a choice level and a nO-choice level.

‘ihe subjects in the choice group were presented with
one Of two pairs Of article titles. The two titles in each
pair were equally attractive but qualitatively dissimilar.
each title in a pair was a summary label for an article. The
subjects in the choice group were to choose the article they
would prefer to read by selecting one of the two article
titles in the pair.

The subjects in the nO—choice group were presented

with the same pairs of article titles as given to the choice

6:
group. Ihe difference between the two groups was that the
subjects in the no-choice condition were not allowed to
select the article they would read. Instead, the no-choice
subjects were told which one of the two articles in a pair
to read. The no-choice subjects thus served as the control
group for the subjects making a choice between article
titles.

The choice and no—choice groups were each further
partitioned into a high importance group and a low importance
group. ‘rhe subjects in the high importance condition chose
or were assigned to read an article determined to be of high
importance to them. An article determined to be of low
importance was chosen or assigned to be read by the subjects
in the low importance condition. ihe pair of article titles
for the high importance condition was:

1. Premarital sex and legalized abortion.

2. 'rhe changing View toward premarital sex.
lhe pair of article titles for the low importance condition
was:

1. Principles of learning and their relation to
acquisition.

2. :he best way to prepare for a test.
rhe pre—eXperiment that was performed in order to select
these two pairs of article titles is described in Appendix A.
After the subjects in each of the four conditions

had read the chosen or assigned article they were given a

69
Table 1

Experimental groups into which subjects were partitioned

 

 

 

 

 

 

 

 

—‘Indepenoent Condition
Choice ho—choice
Dependent high Low High Low
Condition Importance Importance Importance lImportance
Immediate test Group 1 Group 2 Group 1 Group 4
Delayed testl Group l' Group 2' Group 3' Group 4'

 

 

 

 

 

Repeated measurement on the same subjects as in the immedi-
ate test condition.

multiple—choice test. She test was designed to measure each
subjects retention of certain facts contained in each
article. One week later the same multiple choice test was
readministered to the subjects.

A tabular description of the design or the eXperiment
is given in Table 1.
Construction of the exoerimental materials

Articles had to be written which would go with the
titles selected for each importance category. One article
was written for each importance category. Ihe two titles in
each importance condition were both given to the article
written for that importance condition. By giving both
titles to the same article, subjects within each importance
condition received identical information no matter which
article they chose or were assigned to read. ”The content

of the articles in different importance categories had to

7O
differ but an attempt was made to equate the structure of
the two articles. First, the article for the high import-
ance condition was written: then the eXperimenter wrote the
low importance article. The low important article used the
same sentence structure, paragraphing, and length as the
article of high importance. The only difference between the
two articles was the content. The article for each import-
ance category is presented in Appendix E.

In order to measure the amount of information each
subject retained from the article read, a multiple choice
exam was constructed. The test consisted of 14 questions,
each question having 5 or 6 answer alternatives. First, the
test for the high importance condition was constructed. The
low importance condition test was then constructed using the
same structure and number of questions, and the identical
order and number of answer alternatives for each question.

A copy of the two sets of multiple choice questions is
presented in Appendix F.
Procedure

The eXperimenter presented the same verbal instruc-
tions to both sections of the advanced general psychology
course. A copy of these verbal instructions is presented
in Appendix C. The results for each section were kept
Separate and were subsequently analyzed for any significant
differences between sections.

The subjects were randomly assigned to the four

eXperimental conditions. The subjects in all four conditions

71
participated in the eXperiment at the same time. The
following method was used in order to simultaneously ad-
minister the experiment to all four conditions. Three
different sizes of sealable manilla packets were used. Each
size packet was able to fit inside the next larger size
packet. The largest packet was sealed and contained an
instruction sheet, and IBM answer sheet and two sealed
medium size packets lettered A and B. Each of the medium
size packets contained the article for the importance cate-
gory to which the subject had been randomly assigned. The
article in each packet had a different title. In addition,
each of the two medium size packets contained a small packet
in which was sealed the apprOpriate multiple choice questions
for the article.

The large packets containing all of the items listed
above were shuffled and distributed to the subjects as they
entered the classroom. The subjects were first informed
that none of the individual results of the eXperiment would
be publicly disclosed. In addition, the subjects were aSKed
not to talk to anyone else during the eXperiment nor to look
at the material in anyone else's packet. The eXperimenter
then asked the subjects to print their name and student
number on the outside of the largest packet. The subjects
then opened the largest packet and.removed its contents. All
of the participants were instructed to read the instruction
sheet and perform the operations specified on it. First,

the instruction sheet informed each subject about the general

72
purpose of the experiment. In addition on each instruction
sheet was listed a pair of article titles. Each article
title was preceded by the letter A or B. This letter
signified the medium size packet containing the apprOpriate
article. In half of the cases A referred to one of the
article titles and in the remaining half of the cases A
referred to the other article title. .A box was provided in
front of each of the letters, A and B, in which the subject
could check the article he was going to read. The subjects
in the choice condition were instructed to choose which one
of the two articles they would prefer to read and indicate
their choice by checking the appropriate box. The subjects
in the no-choice condition were asked to check the box pre—
ceding the letter A if the number on the outside of their
largest padket was odd. The packet number was always odd
for the no-choice subjects. The instruction sheet is pre-
sented in Appendix D.

After all of the subjects had read and performed
their instructions, the experimenter asked the participants
to return the medium size packet containing the article not
chosen, and the instruction sheet to the largest packet.
The large packet was placed on the floor beside the subject.
The subjects were then asked to raise one of their hands
according to which article (A or B) they were going to read.

Next, each subject opened the medium size packet

with the letter correSponding to the letter checked on the

73

instruCtion sheet and emoved tn: enclosed article- :he

experimenter instructed the participants not to open the

U l

snall cacret which we: inside the sane medium size pacze'.
The subjects were given five minutes to read the article.
After the subj:Cts had finished reaping the article, it was
returned to the medium size secret and the small packet con-
taining the Questionnaire was removed from the medium size
packet The medium sized paCket, now containing only the
article, was sealed and returned to the largest packet. The
subjeCts were given six minutes1 to answer the 14 multiple-
cnoice questions on the article they had just read. The
answers to the queStion: were recorded on an 181 answer

snect. After the subjeCts had answered all l4 questions

(I

they were asked to return all of the material to the largest

O
Q!
0
/\
('J
(1
DJ
:3
0,
Fr.
:3
f!
13
(r

“eal that pacret. finally, the experimenter
informed the subjects that the final stage of the experiment
was to be conducted in exactly one week.

One week later the experimenter returned and re-
administered the multiple choice test to the subjects. ihe
structure of the delayed test differed from the structure of
the immediate test. The order of the questions and the order
of the possible answers was shuffled on the delayed test in

order to negate any position effects.

 

 

l_.. . . . . . . .
'rnis time limit was determined by practice trials
on comparable subjects outside the experimental class before
the administration of the actual GXperiment.

74
after the subjects had taken the delayed multiple
choice test, they rated themselves on four scales. rhese
four scales were designed to assess the subjects perception
of various asoects of the exoeriment. rhese rating scales
are presented in Aooeneix G. rhe four questions asKed on
the rating scale form are listed below.
1. now much do you like the article you read last
Week?
2. .cOW difficult to understand was the article you
read?
3. how much freedom die you feel you had in the
selection of the article read?
4. have you talked to anyone about the nature or
content of this exoerinent since last weec?
Finally, the subjects were thanked for their parti-
cipation. In addition, the nature and purpose of the
exteriment was exolained to them. The participants were
also told that the results of the exocriment would eventually

be available for their examination.

Results

Scales measuring subjects Egrggntion of exnerimental variahles

 

t.
f)"
(D

three scales which measured the subjects pares;—
tion of the excerimental manipulations were analyzed. ihe
puroose in analyzing these scales was to obtain a quantita-
tive measure for uetermining the success of the experimental
manipulations. since the scale values were oroinal, a

t test was used for determining differences between experi-
mental conoitions. No assumptions were made as to the homo-
geneity of the variance in scale reSponse. lhe t test used,
therefore, had a conservative estimate of the degrees of

- l
freecom

(Walker and Lev, 1353). iable 2 lists the t values
for the three rating scales.

'rhe subjects in the choice group were compared with
the subjects in the no~choice group for differences in eer-
ceived freeuom in Choosing the article to be read. 'rhe t
value in iable 2 reveals that the choice subjects perceivea

themselves to have significantly more freeeom of choice than

the no-choice subjects.

—- “.D- hfl~fi --

 

.-. “.__--.—-- - . .. 0-...— ,.-._.._ o- --<-— ~-

‘ihe formula for oetenmining the conservative degrees

of freedom is: ,2 S2 The resulting value is
rounQeQ to o;_ + 2* the nearest integer value.
n n
l 2
-te~- ~v-w-—---- —2
CA 52
:1.., + ._.._;2__
r n +1 n n +1

75

M

76

The second rating scale was analyzed to determine if
there were any differences between the subjects in the
choice and no—choice condition in their liking for the
article read. An examination of Table 2 reveals that there
was not a difference in perceived liking for the article
read between the subjects making a choice and the subjects
who did not choose between article titles.

As stated previously, one of the objectives when
writing the two different articles was that the articles be
the same in all characteristics except topic. One measure
of the degree to which the structure of the two articles was
the same is the subjects' perceived difficulty in under-
standing the two articles. Table 2 shows that subjects in
the low importance condition perceived themselves to have
significantly more difficulty in understanding their article
than did the subjects in the high importance condition.
Tests measuring_§ubjects' retention of information

Since the retention scores for the immediate test
and the delayed test were obtained from two sections of a
class, they were analyzed for differences in reSponse be-
tween the two sections. The results of the t tests, shown
in Table 3, reveal no significant differences between the
reSponses of the two sections on either of the two tests.
The scores of the two groups were pooled for further analy-
sis.

The distribution of the raw scores for the immediate

test and the delayed test is given in Table 4. The raw

77

Table 2

Mean differences between eXperimental conditions on three
rating scalesc

 

 

 

Scale Condition M s.d. dfe td
Choice 3.03 1.19 b
Freedom No-choice 1.55 1.01 67 7'728
. . Choice 4.43 1.20
lelng No-choice 4.06 1.12 67 1°851
- . High Importance 1.34 .54 _ a
DitflCUIty Low Importance 1.58 .74 67 2'147

 

 

 

 

 

ap less than
be less than
cN equals 133

.05.
.001.

Table 3

dtwo-sided t test

econservative degrees of
freedom.

Mean difference in test scores for two sections of advanced
general psychology coursea

 

 

 

 

 

 

Section 1 Section 2 b
Test M s.d. M s.d. t
Immediate test 9.24 . 1.90 9.10 2.20 .467c
Delayed test 8.16 1.99 7.87 2.29 .801C

 

 

 

 

an equals 144.

bconservative df equals 67.
Cnon-significant at .05 level.

7s
scores for the immediate test range from 3 through 14 and
for the delayed test the range is from 2 through 14. No
subject got all of the questions wrong on either of the two
tests. Only one subject on each of the two tests got all of
the guestions correct. The distributions for both tests
were somewhat skewed to the left and also platykurtic, how-
ever, they did possess some of the characteriscics of the
normal distribution. approximately 75 percent of the scores
fell within : 2 standard deviations of the mean.

In order to determine if the distribution of scores
possessed enough normal distribution characteristics to use
an analysis of variance (ADV) further investigation was made
of the means and Standard deviations for each experimental
condition. These means and standard deviations are presented
in Table 5. The means for the eight conditions in Table 5
range from 6.80 to 10.03. The standard deviations range
from 1.78 to 2.26 for the eight groups. A study of the
relationship between the means and standard deviations does
not reveal any significant correlation between these two
statistics. The F—max test showed the variances to be homo—
geneous (Walker and Lev, 1953). The value for F-max was 1.94,
which was well under the critical value at the .05 level of
significance (K=8, df=36). Since the variances were homo—
geneous, and the means and standard deviations were not
correlated, and the distribution of scores approached normal-

ity, it was determined that the AQV could be used.

Raw score distribution for immediate test and

 

-

-..-...———- ——__._—-.

 

haw score

Table 4

Immediate tee

,~_- ..._-. ._.

delayed test

 

Delayed test

 

 

 

 

 

f Cum. f f Cum. f

14 1 l 1 1

13 7 8 2 3

12 11 19 7 10

11 21 40 14 24

10 26 66 13 37

9 23 59 25 62

a 23 112 23 90

7 17 129 22 112

6 5 134 14 126

5 6 140 11 137

4 2 142 3 140

3 2 144 3 143

2 1 144
lean 9.14 8.02
8.3. 14.14 2.3].

 

 

 

.Hawu HwQ mGOHHMUHaflmu mm

.wva masses Z

 

 

 

 

 

 

 

 

 

 

 

 

“New monoco oz

 

Aadv moflocu

L
_

m
_

om.m an.n em.m on.m mo.m se.s Ho.m oe.m Amos ewes

- a Jm>mamn

m3.m no.n mh.a em.m an.a >v.n 0H.N mO.OH AHUV ummy

- mumHUmEEH

we .d.m E .o.m .2 .U.n . z .0.m E

Amnv mocmuuonfiH Aanv wocmsnonfiH Amnv mucmqudEH Aanv mocmuHOQEH

. 30A coax 30A cmflm

 

 

new so mmnCOQmmH nomuuoo m0 HsQEdd no ndOHumH>mU Unmodmuw

.m

emxm nuance mHOHuHsE

m mHQmB

ocm mcnmz

Tabl e 6

Analysis of variance of the number of correct reSponses on
the multiple choice exam

 

 

 

Source of variance df mean square F
Choice (A) 1 7.670 1.092
Importance (a) 1 218.753 31.145b
Tests (c) 1 90.003 62.815b
.A X B 1 2.920 .416
A X C 1 .781 .545
B X C 1 .031 .022
A X B X C 1 .067 .060
Within errora 140 1.433

Between error 140 7.024

Total within error 144
Total between error 143

Grand Total 287

 

 

 

 

aThe mean square for factors and interactions involv-
ing level C are divided by the within error mean square. The
other factors and interactions are divided by the between
error mean square.

bp less than .005.

82

The A07 mocel used was an AXBXCXSS factorial, with
two levels per factor and 36 replications per cell. The
results of the ADV are presented in Table 6. Since the
immediate test-delayed test condition constituted a repeated
measure on the same subjects, the two levels of this factor
were not orthogonal. Two different error mean squares,
therefore, had to be calculated for determining the F ratios.
The within error mean square was used in calculating the F
ratio for all main effects and interactions involving the
test factor (C). The between error mean square was used
in determining the F ratio for all main effects and inter-
actions involving only orthogonal factors (A and B)l. The
mean squares for the main effects and interactions were
obtained from an ADV computer program written for the CDC
3600 at Michigan State University (Ruble, et al., 1963).

Inspection of Table 6 reveals that there were two
significant main effects. First, there was a significantly

greater retention of information in the high importance

 

1The two different error mean squares were obtained
by separately calculating a CXSs AOV for each of the four
orthogonal conditions, i.e., A(l)B(l): A(l)3(2), A(2)B(l),
and A(2)B(2). The within error sum of squares was obtained
by adding together the sums of squares in each of the four
conditions for the interaction between factor C and the
replication factor (85). In order to obtain the within
error mean square, this total sum of squares was divided
by the sum of the degrees of freedom for the interactions
in each condition. The within error sum of squares was sub-
tracted from the total error sum of squares to obtain the
between error sum of squares.

83
group than in the low importance group. secondly, a signi-
ficant decrement in retention occurred between the immediate
test and the delayed test. neither of these main effects,
however, offer any support for the hypotheses of this
eXperiment.

A significant F ratio for the choice main effect or
for any of the interactions would have offered some eXperi—
mental support for the hypotheses. None of these F ratios
were significant, however, nor were their p values close

enough to significance to suggest any trends.

Discussion

As was previously stated, dissonance is aroused when
a person commits himself to a volitional choice between one
of two or more qualitatively dissimilar alternatives.
Choice, therefore, is a necessary but not a sufficient con-
dition for the arousal of dissonance. Volition, commitment
and a qualitative dissimilarity in the choice alternatives
are also required ingredients for the arousal of dissonance.
An examination was made of the procedure and results of the
eXperiment in order to determine whether or not each of
these three conditions were satisfied. First, the analysis
of the subjects perception of how much freedom he had in
choosing the article he would read revealed a significantly
greater perception of freedom for the choice subjects. This
difference was interpreted as indicating that the choice sub—
JECtS perceived themselves to have a substantially larger
amount of volition than the no-choice subjects. Second, in
the Cohen, et a1. (1959) gambling experiment, which was
described earlier, greater dissonance was obtained for the
subjects who made a public commitment to their choice than
for subjects who made a private commitment. In the eXperi-
ment described in this paper, therefore, commitment was
introduced by having subjects publicly signify their choice

of articles by raising one of their hands. Third, although

84

85
no empirical validation was obtained for the qualitative
dissimilarity between the article titles in a pair, the
titles were dissimilar to some degree. each title in a pair
referred to a different topic. The two topics, within each
importance condition, were not identical but only related to
the same category of information. The content of the two
titles in each pair, therefore, was not the same. More will
be said about this problem later in this chapter.

‘fhe second independent variable to be manipulated
was importance. It was previously stated that two factors
had to be controlled in order for importance to determine
differences in the magnitude of dissonance. First, subjects
had to choose between equally attractive alternatives.
second, the two articles titles in each importance condition
had to be equally dissimilar in content. In the pre—experi-
ment for selecting the title pairs, each pair of article
titles was tested for equal attractiveness. The two pairs
of titles used in the experiment proved to be equally
attractive in both the ratio by which each title in a pair
was selected and in the scaling of the importance of each
title. In the actual experiment the subjects in the choice-
high importance condition chose each article in the high
importance pair 18 times. The choice—low importance sub—
Jects, however, Split 11 and 25 on the two titles in their
pair. The article on "the best way to prepare for a test"
was selected over twice as many times as the article on

learning. This difference might have been sufficient to

86
interfere with the arousal of dissonance in this one condi-
tion. In the Opinion of the eXperimenter, however, the
absence of such interference, if the interference did in
fact exist, could not have appreciably modified the obtained
results.

The second factor which had to be controlled in
order for importance to determine the magnitude of the dis-
sonance was the dissimilarity in content between the two
titles in a pair. The dissimilarity between the two titles
in a pair had to be equal for each of the two pairs of titles.
As previously mentioned, no measure was taken of the degree
of qualitative dissimilarity between titles in a pair. As
a result, no empirical evidence is available to determine
whether or not the titles in each of the two pairs were
equally dissimilar. However, even if the two title pairs
were not equally dissimilar this discrepance could not have
prevented the arousal of dissonance. The inequality in dis-
similarity would instead only have increased the difference
between importance conditions, in the magnitude of previously
aroused dissonance. For the purposes of the present dis-
cussion, therefore, it will be assumed that the two pairs of
article titles were equally dissimilar.

The fact that a true difference in importance was
created is supported by the results of the pre-experiment
and the actualeXperiment. In the pre-eXperiment a signifi-

cant difference was found in the mean importance rank for the

87
two pairs of article titles. In the actual experiment a
significant main effect for importance was obtained in the
ADV.

One factor which may have confounded the importance
effect was the degree of difficulty the subjects had in
understanding the two articles. Even though the articles
were similarly structured the analysis of the rating scales
showed that the subjects preceived the low importance
article to be significantly more difficult to understand.

If this difference in difficulty was large enough it may

have contributed to the importance main effect. On the other
hand, the difference in difficulty of understanding may not
have been due to a difference in the construction of the
articles but instead to a variation in importance. It seems
very plausible, that the subjects reading an article of con-
siderable importance, would find it easier to understand

that article than the subjects reading an article of low
importance. People usually concentrate and work the hardest
on tasks that are the most important to them.

Since all of the required conditions seem to have
been satisfied, dissonance theory predicts that the choice
subjects would retain more information about the articles
they read than the no-choice subjects would retain about
their articles. No support was obtained for this hypothesis
Since the choice main effect in the AOV was not significant.
In addition, if the choice between article titles resulted

in the arousal of dissonance then dissonance theory would

88
predict that importance would interact with choice to in-
crease the magnitude of the dissonance. It is possible that
a significant choice effect would not be obtained and yet a
significant choice and importance interaction would occur.
The choice and importance interaction in the ADV, however,
was not significant. The previously mentioned non-equal
division of choices in the low importance condition would
have affected this interaction. However, it seems unlikely
that a more equal division of choices would have given any
support to the dissonance predictions. The choice-low
importance subjects were predicted to retain less information
than the no choice—high importance group. Since the high
importance groups were virtually equal, an equal choice
split in the choice—low importance group would have made
little, if any, change in the obtained results.

The conclusion that choice and importance did not
interact to produce dissonance rests on the assumption that
the subjects would have tried to reduce their dissonance by
adding cognitive elements. It is possible that some other
means than adding cognitive elements was used to reduce
dissonance.

As was previously stated dissonance can be reduced
by the individual in four wayS. A re-exanination of these
four methods of dissonance reduction and the precautions
taken to control them seems to be called for. First, dis-
sonance could have been reduced by changing or revoking

one's choice. An empirical check of whether or not a subject

89
changed his choice was available. In describing the proce-
dure used in this eXperiment it was stated that the subjects
Specified their choice on the instruction sheet. The
instruction sheet was then returned to its packet and the
packet was placed on the floor beside the subject. 'The
subjects were carefully observed. no subject changed his
choice. In addition, the subject could have revoked his
choice by refusing to continue to participate in the eXperi-
ment. There were no subjects who drOpped out of the experi—
ment at any point. It was assumed, therefore, that no sub—
jects reduced their dissonance by changing or revoking their
choice.

A second method by which the individual could have
reduced dissonance was by changing the attractiveness of
the choice alternatives. In addition, related to changes in
attractiveness is a third method of dissonance reduction.
The third method the individual could have used was to re—
duce the importance of the relevant cognitive elements in a
choice situation. As previously pointed out (p. 48),
attractiveness and importance are correlated on an evaluative
factor. To the excent that attractiveness and importance
are correlated, changes in either variable should be
measurable by an evaluative scale. A scale which measures a
Person's perception of how much he likes the chosen alter-
native would be an evaluative scale of this type. The
reduction of post—choice dissonance by either a change in

attractiveness or a reduction in importance, therefore,

90

should be reflected in a difference in the liking of the
choice subjects and the no—choice subjects for the chosen
article. The analysis of the like scale revealed no signi-
ficant difference in liking between the choice and no-choice
subjects. It would seem that dissonance was not reduced by
either changing the attractiveness of the choice alternatives
or by reducing the importance of the relevant cognitive
elements. It was thus assumed that if dissonance had been
aroused it would have been reduced by adding cognitive
elements. Since no evidence was obtained which would suggest
that any cognitive elements were added it was concluded that
dissonance was not aroused.

There was also an attempt in this eXperiment to
apply rigorous methods to the analysis of the data. First,
the no-choice condition served as a control group against
which the choice subjects could be compared for determining
the effect of choice on retention of information. Second,
no subjects were thrown out after the data had been analyzed.
The four subjects that were discarded were thrown out before
the analysis of the data. Third, an overall test of Signi—
ficance (Analysis of Variance) was used to determine if any
differences existed between the eXperimental groups. Finally,
no attempt was made to draw conclusions which went beyond
the significance of the results. The only interpretation
of the results was that dissonance was not aroused in a
situation where dissonance theory predicts that it should

occur. This result constitutes negative evidence for the

91
theory. Some possible eXplanations for this null finding
are given below.

First, the failure of the results to support dis—
sonance theory might be eXplained by a simple reinforcement
theory of learning. The act of making a choice between
article titles may not have any effect on the retention of
information about the chosen title. Instead, the reading of
the article which goes with the title may serve as a rein-
forcement for the subject's public commitment to reading
that article. The difference in retention in the high and
low importance conditions would be due, under this interpre-
tation, to the greater reward from reading the article which
was of more importance and interest to the subject.

Another possible eXplanation for the failure to
obtain findings wnich support dissonance theory, is that the
subjects may not have been presented with a "real" choice.
As mentioned earlier, the two article titles may not have
been dissimilar enough in content for the subject to be
really giving up anything. If the attributes of the rejected
alternatives were not qualitatively dissimilar from the
attributes of the chosen alternative then the two sets of
attributes could not have been in a dissonant relation.
Since no empirical validation of the qualitative dissimilar-
ity of the two titles in a pair was obtained there is the
possibility that there was not a "real" choice given to the
subjects and, therefore, dissonance was not aroused. This

seems to be the only possible eXplanation not Specifically

92
refuted which makes the data consistent with dissonance
theory.

The final conclusion must be that no support was
obtained for dissonance theory formulations. In addition,
neither were any of the propositions of dissonance theory
diSproved. There is a need which the study does emphasize,
however: more rigorous research on dissonance is imperative
if the theoretical concepts of dissonance are to be un-
equivocal and if the theory is to make any real contribution

to the understanding of human behavior.

BIBLIOGRAPHY

Adams, J. S. Reduction of cognitive dissonance by seeking
consonant information. Q. abnorm. soc. Psychol.,
1961, 62, 74—75.

 

 

Allyn, Jane, & restinger, L. The effectiveness of un-
anticipated persuasive communication. g: agngrm.
SOS;WE§ZEhOl-' 1961, 62, 35-40.

Aronson, E. The effect of effort on the attractiveness of
ewarded and unrewarded stimuli. QiléhQQHEL_§QE;
ES ___:-LCEhOl¢ I 1961, 63, 375-350.

ASCh, s. a. Cacopnonophobia. gpgtemp;_§sycngl., 1955, 3,

 

Brehm, J. w. Postdecision changes in the desirability of
alternativeS. J. abnorm. soc. Psych91., 1956, 52,

Brehm, J. w. Increasing cognitive dissonance by a fait
accompli. g._abngrm._§oc. Psycnol., 1959, 55, 379—
352.

 

brenm, J. W. Attitudinal consequences of commitment to un-
pleasant behavior. J;~abnorg._§oc.-_nf_sychol., 1960,
60, 379-353.

Brehm J. n. & Cohen, A. R. Choice and chance relative

I I
de privation as determinants of cognitive dissonance,
g: abnorm. Soc. Psy chol;, 1959, 58, 353-38/. (a)

Brehm, J. W., & Cohen, A. R. ke—evaluation of choice
alternatives as a function of their number and
Qualitative similarity. J. abnormL soc. ?sycnol.,
1959, So, 373- 378. (b)

 

Brehm, J. W & Cohen, A. R. txnloration:_in cognitive

0 I
dissonance. New York: wiley and Sons, 1962.

 

erenm, J. W., & Lipsner, D. Communicator-communicatee
discrepancy ano perceived communicator trustworthi-
ness. J. Pers., 1959, 27, 352—361.

Brown, R. Social psvcholodv. new York: rhe bree frees,

1965, pp. 55 '4- 0v 94.

94

Chapanis, Natalia P., & Chapanis, A. Cognitive dissonance:
Five years later. Psyghol. Bull., 1964, 61, 1-22.

Cohen, A. R. Attitudinal consequences of induced dis-
crepancies between cognitions and behavior. Publ.
gpin. guart., 1960, 24, 297-318.

 

Cohen, A. R. Communication discrepancy and attitude change:
A dissonance theory approach. J. P§£§so 1959, 27,

Cohen, A. R., Brehm, J. W., & Fleming, W. H. Attitude
change and justification for compliance. J. abnorm.
soc. Psychol., 1958, 56, 276—278.

 

Cohen, A. R., Brehm, J. W., & Latane, B. Choice of strategy
and voluntary eXposure to information under public
and private conditions. J. Pers., 1959, 27, 63-73.

Cohen, A. R., Terry, H. I., & Jones, C. B. Attitudinal
effects of choice in exposure to counter-propaganda.
J. abnorm. soc. Psychol., 1959, 58, 388-391.

Davis, K. B., & Jones, E. E. Changes in inter-personal
perception as a means of reducing cognitive dis—
sonance. J. abnorm. soc. Psychol., 1960, 61, 402-410.

Deutsch, M., & Krauss, R. M. Theories in social psychology.
New York: Basic Books, 1965.

Deutsch, M., Krauss, R. M., & Rosenau, Norah. Dissonance or
defensiveness. J. Pers., 1962, 30, 16-28.

Edwards, A. Technigues ofuattitud§_sca1e_construction.
New York: Appleton-Century-Crofts, 1957.

Ehrlich, D., Guttman, L., Schonbach, P., & Mills, J. Post-
decision eXposure to relevant information. Q: abnorm.

 

soc. PsyChol., 1957, 54, 98-102.

 

Festinger, L., Gerard, H. B., Hymovitch, B., & Kelley, H.H.
The influence process in the presence of extreme
deviate. Human Relat., 1952, 5, 327-346.

Festinger, L. A theory of cogpitive dissonangg. nvanston,
111.: Row & Peterson, 1957.

Festinger, L. The psychological effects of insufficiept

J rewards. American Psychologist, 1961, 16, 1-1 .

ConflictL decision, and dissonance.

Festin er, L. - . .
9 Calif.: Stanford UniverSity Press,

1. 1964.
Stanford,

95

Festinger, L., & Carlsmith, J. M. Cognitive consequences
of forced compliance. J. abnorm. soc. Psychol.,
1959, 58, 203-210.

Kelman, H. C. Attitude change as a function of reSponse
restriction. Human Relat., 1953, 6, 185-214.

 

Maccoby, N., & Maccoby, Eleanor. Homeostatic theory in
attitude change. Publ. Opin. Quart., 1961, 25,

Mills, J. Changes in moral attitudes following temptation.
J. Pers., 1958, 26, 517-531.

Mills, J., Aronson, B., & Robinson, H. Selectivity in ex-
posure to information. J; abnorm. soc. Psychol.,
1959, 59, 250—253.

Osgood, C. B., Suci, G. J., & Tannenbaum, P. N. The measure-
ment of meaning. Urbana, 111.: University of
Illinois Press, 1957.

 

Rabbie, J. M., Brehm, J. w., & Cohen, A. R. Verbalization
and reactions to cognitive dissonance. J. Pers.,
1959, 27, 407-417.

Rosen, S. Postdecision affinity for incompatible informa-
tion. J. Abnorm. soc. Psychol., 1961, 62, 188-190.

Searles, H. L. Logic and scientific methods. New York:
Ronald Press, 1956.

Silverman, I. In defense of dissonance theory: Reply to
Chapanis and Chapanis. Psychol. Bull., 1964, 62,

Steiner, I. D., & Johnson, H. H. Relationships among dis-
sonance reducing reSponses. J. abnorm. soc. Psychol.,
1964, 68, 38-44.

Walker, Helen, & Lev, J. Statistical inference. New York:
Holt, Rinehart, & Winston, 1953.

Yaryan, Ruby, & Festinger, L. Preparatory action and belief
in the probable occurrence of future events. J.
abnorm. soc. Psyghol.,196l, 63, 603-606.

Zimbardo, P. G. Involvement and communication discrepancy
as determinants of Opinion conformity. J. abnorm.
soc. Psychol., 1960, 60, 86-94. —

 

 

APPENDIX

96

Appendix A

Method for selecting the experimental article titles

A pre-eXperiment was conducted on 26 males and 5
females enrolled in an advanced general psychology course at
Michigan State University for the Fall quarter of 1964. in
addition, the same pre-exoeriment was given to 16 males and
7 females enrolled in the same class for the Winter quarter
of 1965. The results of these two pre—eXperiments were com-
bined for selecting the article titles pairs to be used in
the actual experiment.

‘fhe main objective in conducting the pre-eXperiment
was to find two pairs of article titles which met the follow-
ing two requirements.

1. The two titles in each pair must be equally

attractive and yet qualitatively dissimilar.

2. One pair of article titles must be significantly

more important than the other title pair.
In addition, the two titles in a pair were to be titles for
the same article. This duplication of articles was neces-
sary in order to control the information each subject
received. The article titles in each importance category,
therefore, had to be qualitatively different and yet be the
title for the same article. It was decided that in order to

meet this requirement, the pair of article titles had to be

97

98

about similar tOpics. For the high importance category,
therefore, one of the topics was premarital sex and the
other topic was abortion. The low important topics were
learning theory and testing. Three articles titles were con-
structed for each of the two tOpics in each of the two
importance categories. These 12 article titles are listed
in Table A 1.
Equal attractiveness of article titles in a_pair

The first requirement tested was that a pair of
article titles had to be equally attractive and yet quali-
tatively dissimilar. Edwards (1957) states that one method
of determining the degree of equality in the attractiveness
of items is by having subjects make choices between a series
of pairs of the various items. The equality of attractive-
ness of a pair of items is determined by the ratio of the
number of times one item is chosen over the other item. The
smaller the ratio, i.e., the more equal the number of times
each item is chosen, the more equally attractive the two
items in a pair. The significance of this ratio is determined
by obtaining the Z score for the ratio value. This method
of paired comparison choices was thus selected for initially
determining the most equally attractive pairs of article
titles.

Since the two items in a pair of article titles had
to be qualitatively dissimilar as well as equally attractive,

it was decided that one title in a pair should be on one of

 

.mUGMEMOMHmQ 0cm wuaaaflm Op coHpmamH mua paw qCaMmme
.coauamanum on coflumHmH Hams“ paw maficumma Mo mmadfiucfium

.mcflpmmu psonm won on man %Hom£u moacnmma umsz

wm30cx mco was? mcfiusmmme M0 mapme pHHm> m mummy mHfl
.Umwu o How mummmuo on wwS ummfl wca

«when one ma mem Mo mow» not:

.GOHpHOQm omNHHmmwa m0 mohmwmn on?

.coauuoom poNHHmmma use xmm amusememum

99

.cofluuonm omnaammma 30aam m3 pasosn

.mmmSQqu

cmofium.fi co coHusH0>mu HMHOE mes pom xmm Hopflumewum
mucmfiu Ham mmmfiHHmE muouwn mmusooumpca Hmsxom mH

.Xmm Hmuaumsmud pumsou 3ma> msamcmcu use

mmause wausune

.NH
.aH

.OH

.9

 

whomsa mcflcnmmq

mcaumme

mocmuHOQEH 30g

COHflHOQfi

xmm Amufiumsonm

mocmuHOQEH cmflu

mmauommumu menus

mmfluommumu mucmuHOQEH pcmumuwfip How oouusuumsou mmaufiu waofiuum

a a. m Ewe

100
the topics in an importance category and the remaining title
in that pair should be on the second topic in that import-
ance category. For example, one title would be about pre-
marital sex and the other title would refer to abortion.
It was only necessary, therefore, to combine the three
article titles for one topic in an importance category with
each of the three article titles on the other tOpic in the
same importance category. This method of constructing pairs
of article titles resulted in nine pairs for each importance
category, or a total of 18 article title pairs. These 18
pairs of article titles and the method used for obtaining a
subjects choice for each pair is shown in Appendix B, Section
1. ‘Ihe Z scores for the 18 different combinations of
article titles are given in Table A 2.

An examination of Table A 2 reveals that the Z values
for all 18 pairs of article titles fell within 1.96 units of
the normal curve. Only those eight title pairs with a Z
value between .000 and I .200, however, were used for
further analysis. Four of the eight acceptable title pairs
were from the high importance category. The other four
title pairs were from the low importance category.
Qifference in importance_of article titles_

“The second method used in selecting the appropriate
title pairs had to do with the importance of the article
titles. each subjects perception of the relative import-
ance, to himself, of each article title was obtained by

having the subject rate the 12 article titles. The 12 titles

101
were rated on a seven point scale which ranged from "very
unimportant" to "very important". The method used to obtain
each subjects rating of the importance of each article titles
is fully described in Appendix B, Section 2.

Difference in importance within a pairjgf.a£tigle
titles. The analysis of the difference in importance within
title pairs was performed in order to assure the equal
attractiveness of the alternatives in a pair of article
titles. The eight article titles accepted as being equally
attractive were tested for differences in importance within
each pair of titles. A correlated t test (Walker and Lev,
1953) was used to determine if there were any differences in

importance. The t values for these eight title pairs are

Table A 2
2 scores for paired comparison choices between 18 pairs of
article titlesa

(Z score represents the proportion of times tne column title
is picked over the row title)

 

 

 

 

 

 

 

 

Article Title Article Title Number
Number 1 2 3 I7 8 9 9

4 .179 -. 319 .319
5 -.136 -.513 .179
6 .674 1 .000 .513

10 —.465 -.O90 .565

11 .179 .090 .513

12 -.990 -.732 -.090

 

 

 

 

 

 

 

 

aN equals 54.

 

102
presented in Table A 3. An analysis of these t values reveals
no significant differences in importance, at the 5 percent
level of significance, between any pair of article titles.
Article titles in eight pairs, therefore, qualified as being

not significantly different in attractiveness.

Table A.3

Correlated t values for differences in importance within eight
article title pairsQ

(Two-sided t values based on first title number in pair minus
second title number in pair)

 

 

 

Title Pairs —_ _- t values
High importance
1 a 4 .655c
3 a 5 .239C
5 a 1 -.175°
Low Importance
7 a 11 -l.O26b
s a 11 .547C
10 a s -1.2ssb
12 a 9 .339c

 

 

greater than .05.
greater than .10.

D greater than .50.

ovr) q n:
‘ TJ'U

R equals 50.

 

103

Eifferences in 1???chmesses-219.5%;aliasigseflsriiele

 

EEELEEL The final requirement for the selection of two
pairs of article titles was that one of the two pairs had to
be about a tepic of significantly more importance than the
topic of the other pair or titles. The final two pairs of
article titles would have to come from the eight title pairs
satisfying the equal attractiveness tests. ihe Objective at
this point, therefore, was to compare the four pairs of
article titles in each importance category with each other
and determine which pairs, if any, were significantly differ—
ent in importance. In order to determine the difference in
importance between title pairs the importance rating values
for the two titles in a pair were added together. A corre-
lated t test was then used to determine the difference be-
tween the summed importance ratings for each title pair. An
analysis of the t values in Table A 4 reveals that there are
11 combinations of article titles pairs wnich are different
in importance at the .05 level of significance. In one case
the difference in importance is in the wrong direction.

After examining the t values in this table it was decided
that further consideration would be giVen only to those
combinations of title pairs with a p value less than .01.
From these remaining seven combinations of title pairs the
final two pairs of article titles were randomly selected.

The two chosen title pairs are listed below.

J

'.b1e A 4

O

l
‘0

Correlated t values for cifferences in importance between
article pairs

(Two—sided t values based on column variable minus row variable)

 

 

..-.-_.—o--__,_____ _ .__—o‘_ "—.“h—
__. A—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Low Importance ‘"_w_gighfiigportanpe_giwlp_Pai s
. . , r . 1 “'"h‘ m'flw
iltle ‘alrs 1 a 4 i 3 a 5 5 a 1 7 a 2
7 a 11 4.64c 3.soc 4.64C 2.16a
s a 11 3.15b 2.49a 3.07b .90
10 s s 3.05b 2.25a 2.75b 1.09
12 a 9 -.37 -.7o -.29 -2.21a
._ --.__ .1- -1-.. 41.-.---” W J- _ A
a . .
p is less than .05.
bp is less than .01.
Cp is less than .001.
O.

ecuals 47.

‘rhe title pair chosen for the high importance condi-
tion was:
1. Premarital sex and legalized abortion.
2. 'rhe changing view toward premarital sex.
The title pair chosen for the low importance condition was:
1. Principles of learning and their relation to
acquisition.

2. The best way to prepare for a test.

Appendix 5

H
U
H
(D
I
(D
X
G
(D
H
r}
(T)
:3

Measurement “evices fo

Section 1. Paired cgmparison form for measuring article
title attractiveness.

Below are listed titles of articles on various tOpics.
The titles are grouped in pairs, the same title is paired
with other titles but the same two titles are never paired

more than once.

Please indicate for each of the eighteen (18) differ-
ent title pairs which article you would prefer to read. Base

your decision on the titles alone.

Indicate your choice for each pair by putting an "X"
in the box preceding the title you prefer.

r.g., if you prefer to read an article entitled ”The
Lone Ranger Rides Again” instead of ”Here
Comes Mickey house”, you would indicate your
preference as shown below.

rhe Lone Ranger Rides Again

Here Comes Mickey Mouse

{:1

Row choose one from each of the following pairs.

Ts sexual intercourse before marriage all right?

.1.

Pre—marital sex and legalized abortion.

[—.I
71
1.1

1:;

Should we allow legalized abortion?
'The changing view toward pre-marital sex.

N
o

Pre-marital sex and the moral revolution on American

campuses.
'rhe hazards of legalized abortion.

Should we allow legalized abortion?
Is sexual intercourse before marriage all right?

.b

[hafima

Are tests a valid means of measuring what one knows?
Principles of learning and their relation to acquisition.

()1
o

‘ I: .

105

106

Is sexual intercourse before marriage all right?
The hazards of legalized abortion.

0‘
‘
l

7“ .
Li

7’
ED

Principles of learning and their relation to acquisition.
The best way to study for a test.

CD
0

a: as

What type of exam is the best?
Principles of learning and their relation to acquisition.

Should we allow legalized abortion?
Pre-marital sex and the moral revolution on American
campuses.

6

Testing and its relation to ability and performance.
The best way to study for a test.

32

ll,“i.What type of exam is the best?
[::]What learning theory has to say about testing.

lZJumJWhat learning theory has to say about testing.
l_fljAre tests a valid means of measuring what one knows?

lBinlThe best way to prepare for a test.
[::]What learning theory has to say about testing.

14£::'Testing and its relation to ability and performance.
{__ What type of exam is the best?

lSIi:]Pre—marital sex and the moral revolution on American
,‘campuses.
[:Q Pre-marital sex and legalized abortion.

L—fid

l6£_:}The hazards of legalized abortion.
ng'The changing view toward pre-marital sex.

17J;;]The changing view toward pre-marital sex.
[_j Pre-marital sex and legalized abortion.

lBIjj Are tests a valid means of measuring what one knows?
[:3 Testing and its relation to ability and perfonmance.

107
Section 2. Importance ranking form.

Titles suCh as we have been considering are of varying
degrees of significance or importance to different peeple.
Some things may be very important to you, others not so
important. Please try to evaluate the following titles
according to the degree of importance they hold £95 ygg.
Remember the decision is to be made on the basis of its
importance :2; ygg and not its importance for others or
society as a whole.

For the final step in this survey we have listed twelve (12)
article titles below. Please indicate the degree of import-
ance of each title by placing its number at the position on
the scale which best represents its importance £95 ygg.

If two (2) titles are of equal importance both may be placed
between the same two vertical lines on the scale.

E.g., if title number two (2) is quite important for
you and title one (1) is very unimportant while
titles three (3) and four (4) are equally fairly
important you might indicate this as shown below.

 

4
l / L / L 3 J 2 /
very very
unimportant important

Now place each title number on the following scale.

 

l / / / / L 1
very very
unimportant 2 3 4 5 6 important
1 7

What learning theory has to say about testing.

. Predmarital sex and the moral revolution on American
campuses.

The best way to prepare for a test.

The hazards of legalized abortion.

Pre-marital sex and legalized abortion.

The changing view toward pre-marital sex.

What type of exam is the best?

Should we allow legalized abortion?

. Is sexual intercourse before marriage all right?

10. Testing and its relation to ability and performance.
11. Are tests a valid means of measuring what one knows?
12. Principles of learning and their relation to acquisition.

mid
O

Omflmtflhw
O

Appendix C

Verbal instructions for the experimepg

All information in this experiment is private and will
not be disclosed by nane.

Print in large letters your Name and Student Number-~on
the outside of your envelope-—in the upper left hand
corner.

These packets contain various information. All of them,
however, do not have the same contents. Therefore,
please do pg; look at anyone else's information at any
time during this eXperiment or talk to those around you.
If you do you will destroy the validity of the eXperiment.

Open the large packet. Inside you should find:

) An instruction sheet

) An IBM answer sheet

) Two medium size packets labeled g and g in the upper
left hand corner. In the upper right hand corner
is the general packet number which is also on the
outside of your main packet.

LUMP

(
(
(

Fill in the information on the IBM Sheet.

(1) Read the instruction sheet. When you have finished
reading it and have completed what it asks you to do
please raise your hand.

(2) Do pp; Open either packet A or B until I instruct
you to do so.

(3) Go ahead and read.
(4) Are there any questions about the instructions?

Commitment and reading of article.

(1) Return to the main large packet the instruction
sheet and the envelope that has the letter (either
A or B) on it which you did BEE check on the
instruction sheet. You are never to open this
packet for this eXperiment. Place the large
envelOpe on the floor beside you.

(2) How many checked the box for article A? (Raise hand)
How many checked the box for article B? (Raise hand)

108

LI)

10.

110

109

(3) Now Open the packet which has the letter (either A
or B) on it that you did check on the instruction
sheet. Be careful and do not destroy the packet.
It should open easily. Remove only the loose
article. Leave the small envelOpe in the packet.
Do not turn the title page of the article until I
instruct you to.

(4) Does anyone have a title on the cover sheet other
than the one they cheeked on the instruction sheet?
If so, raise your hand and I will come and give you
the right article.

(5) You have five minutes to read the article. Use all
of the time to become familiar with its contents.

Remember - do not look at your neighbor's information.
Go ahead and read. (time them)

Now, take the article — put it back in the packet it came
from. At the same time remove the smallest packet which
was previously left in the enve10pe. Now lick and reseal
the medium sized packet and return it to the large
envelOpe.

Also, obtain the 13w answering sheet from the large
packet.

(1) Open the small packet. Do Egg destroy it. Remove
the questionnaire from it but do not start reading it.
Put the small packet in the large envelope.

Now, place the large packet on the floor. Only the
questionnaire and IBM answer sheet should remain on
your desk.

AA
UJN
VV

(1) Read the instructions and then answer the questions
by indicating the correct answer on the IBM answer
sheet.

(2) Answer the questions on the basis of the information
read in the article and not what you believe to be
true.

(3) Raise your hand when completed.

Put the answer sheet and questionnaire back in the large
packet you put on the floor. Do not fold the IBM
answer sheet.

Close the main packet and claSp it shut.

12.

13.

14.

110
Pass your packet to the left.

Please do 223 talk to anyone inside or outside of the
class about the nature or content of this eXperiment.
This is very important—for there is a second part to

this eXperiment next week and conversation about the

eXperiment will destroy its purpose and validity.

Thank you.

appendix b

Packet instruction fprms

 

Section 1. Choice--high importance form.
ggsrRUCTIoms

l. The purpose of this experiment is to examine the
relation between attitudes and the retention of informa-

tion.

2. You have two packets or smaller envelOpes labeled 5 and
2. Each contains an article on a given subject. You
are to choose from the pair below which article you
would prefer to read and indicate this choice by check-
ing the box preceding that article title and packet

letter.
Packet Title
Fifi A "The Changing View Toward Pre-
Martial Sex”
[:1 B "Pre-Marital Sex and Legalized

Abortion"

3. Do not ppen the chosen packet until instructed to do so
by the experimenter. You will then have five minutes
to read the article.

111

112

Section 2. No-choice--high importance form.

 

INSTRUCTIONS

 

The purpose of this experiment is to examine the
relation between attitudes and the retention of infor-
mation.

You have two packets or smaller envelopes labeled 5 and
g. Each contains an article on a given subject. You
are to choose from the pair below which article you
would prefer to read and indicate this choice by check—
ing the box preceding that article title and packet
letter.

Packet Title
{:3 A ”The best Way to Prepare for a Test"
(f3 3 "Principles of Learning and Their

Relation to Acquisition“

Do not Open the chosen packet until instructed to do so
by the eXperimenter. You will then have five minutes

to read the article.

113

Section 3. Choice1:low_importapce form.

 

 

 

IN STRUCTICNS

 

l. The purpose of this eXperiment is to examine the
relation between attitudes and the retention of infor-
mation.

2. You have two packets or smaller envelopes labeled g and
g. Each contains an article on a given subject. Please
check the number on the outside of the large envelope.

a) If it is an 9_d_C_1 number (i.e., l,3,5,7,etc.).You
are to read the article in the packet labeled g.
b) If it is an even number (i.e., 2,4,6,8,etc.),you
are to read the article in the packet labeled g.

Before you continue please check the box below preceding
the apprOpriate letter and title (determined above) to
indicate which article you will read.

Packep Title
{»f A "Pre—Marital Sex and Legalized
Abortion"
[:7 B “The Changing View Toward Pre4Marital
' Sex"

3. Do not Open the selected packet until instructed to do
so by the eXperimenter. You will then have five minutes
to read the article.

114

Section 4. No-choice-—low importance form.

 

IESTRUCTICNS

 

l. The purpose of this experiment is to examine the
relation between attitudes and the retention of infor-
mation.

2. You have two packets or smaller envelOpes labeled g and
a. Each contains an article on a given subject. Please
check the number on the outside of the large envelOpe.

a) If it is an Egg number (i.e., l,3,5,7,etc.), you
are to read the article in the packet labeled.§.

b) If it is an even number (i.e., 2,4,6,8,etc.). You
are to read the article in the packet labeled B.

Before you continue please check the box below preceding
the apprOpriate letter and title (determined above) to
indicate which article you will read.

Packet Title

(“T A "Principles of Learning and Their
Relation to Acquisition”

_.

. B “The Best Way to Prepare for a Test"

3. Do not ppen the selected packet until instructed to do
so by the experimenter. You will then have five minutes
to read the article.

Appendix E

érticles

Section 1. gigh-importance article.

Sexual ethics is a widely talked about subject to-
day, especially among college students. It is not the in-
tent of this article to solve this problem nor to present a
new point of view but rather to survey some information
about this topic.

First, what truth is there in the report of a signi—
ficant increase in the past few decades in the frequency of
pre—marital sex relations? A recent article in L295
magazine states that in a series of studies done between
1915 and 1959 the frequency of pre-marital sex experience in
males varied between 32% and 73%. For women the range was
from 7% to 57%. Kinsey's data on the American female
reveals that of women born before 1900 only 10% had pre—
marital relations, while 30% or more of the women born in the
1920's and later reported having sexual relations before
marriage. A study made in 1963 on the "Sexual Behavior of
the Human Female” states that 20% of college women have pre-
marital sex relations before graduation. In addition,
Graham Blaine, Jr. in the January, 1964 issue of the Ladies
figme Journal reported that more than 50% of college women

have pre—marital intercourse.

115

116

On the other hand some sources suggest that the
alteration in sexual behavior has been an increase in petting
and non-intercourse love-making rather than in pre-marital
intercourse. Ira Reiss in the article “Sex Behavior in Our
Culture" suggests that in this generation there is actually
more restraint and uncertainty in pre-marital sexual
behavior because the sex role of the single person is not
well defined and is sometimes conflicting for the individual,
while the tOpic of sex receives increasing publicity and
attention.

Whatever the modification in sexual ethics, let us
examine the factors which may have contributed to this
transformation. One suggestion is that the increased
emphasis in American society on the equality of the sexes,
resulting in the female becoming competitive with the male,
has caused a transition in the sexual code. The old code
was that the girl was eXpected to be restrained and resist
the boy's advances. There was a single standard for women,
petting—with-affection, i.e., pre—marital sexual relations
were not justifiable. Only petting was allowed. However,
for the male there was a double standard, i.e., pre—marital
intercourse is all right for one party (the boy) but not for
the person they date (the girl). Men wanted their girls to
be both sexually active and pure. The equalitarian position,
however, forces the women into a more active sex role in

which the result is a code of permissiveness—with—affection.

117
Under this standard pre—marital sex is justifiable as long
as the two peOple involved are in love, “going steady“,
engaged, or in some way affectionately related.

A second suggestion, which might partially account
for the transition is the technical development in the
efficiency and safeness of contraceptives. One of the main
deterrents to pre—marital intercourse, eSpecially for the
woman, is the fear of pregnancy. However, with the high
probability of prevention due to the advent of birth control
pills and similar devices this deterrent has lost its force
and Opened the way for more free sex behavior.

Even with the acclaimed efficiency of contraceptives
some peOple assert that there has been an increase in the
abortion rate due to increased pre—marital intercourse.
Whether this increase is due to overconfidence in contra—
ceptives or ignorance about contraceptives is not clear.

In 1964, the Science fiews Letter reported that in

 

 

the 0.5. there were from 200,000 to 1,200,000 abortions
annually. In 1957, 260 of these resulted in death to the
mother. This abortion rate was stated as being significantly
greater than the rate during the past few decades.

These requests for abortion, however, were reported
to have come equally from married and unmarried women. Also,
one source suggests that the upper class person is more will-
ing to resort to abortion, while the lower class individual

seems to prefer unwed motherhood.

118
Presently, in the U.S., 46 states permit abortions
to save the life of the mother and three states to preserve
the mother's health. All states allow therapeutic abortions
when a psychiatrist believes the mother may commit suicide

rather than have the baby. A report in the Science News

 

Letter states that abortions are no longer very dangerous
when performed by a licensed M.D. who uses sterile procedures.
It goes on further to say that 90% of the abortions today
are done by these M.D.'s. The problem is with the remaining
10% who try self-induced or non-medical abortion.

In conclusion, a comparison of sexual behavior for
different countries shows that France does not permit
abortions or contraceptives, while Japan allows both: yet
both countries have approximately the same abortion rate.

We have not tried to present a solution to either of
these problems in this short discourse, but rather to
present some ideas on the subject for consideration. None-
theless, if any conclusion may have come from this investi-
gation it is that we are not sure exactly what is the nature

or magnitude of a transition in the sexual ethic.

119
Section 2. Low importance article.

Learning is a subject of common interest to the
college student. It is not the intent of this article to
fully discuss this topic nor to present a new point of view
but rather to survey some information about the subject of
"how to learn.“

First, has there been a significant increase in the
past few decades in the number of pepple using organized
methods of study? A recent article in £995 magazine states
that in a series of studies done between 1915 and 1959 the
number of persons making use of organized study methods
varied between 3 % and 73% for males. For women the range
was from 7% to 57%; Ladd's data on study habits reveals
that for persons born before 1900 only 10% used organized
learning methods in their studies, while 30% or more of
those persons born in the 1920's and later applied these con-
cepts when studying. A study made in 1963 on the "Methods
of Study of Today's College Students" states that 20% of
college students have some organized method they use in
studying. In addition, George Blaine, in the January, 1964,

issue of Reader's Digest reported that 50% of college stu-

 

dents use a personally organized study procedure.

On the other hand, some sources suggest that the
change in study habits has been an increase of interest in
learning and a better education rather than an increasing
application of organized study habits to learning. Ira Reiss

in the article "Study Behavior in Our Society" suggests that

120
in this generation there is actually more ambivalence and
uncertainty about study behavior because the role of the
”good" student is not well defined and is sometimes con-
flicting for the individual, while the emphasis on obtaining
more education and high grades is steadily increasing.

Whatever the modification in study behavior, let us
examine the factors which may have contributed to this
transformation. One suggestion is that the increasing in-
fluence of findings in the area of psychology of learning
has caused a transition in study habits. At this point we
shall examine some of the more important findings in this
area.

First, the college student must be motivated. A
primary source of his motivation for studying is the goals
he sets. Another term for goal is "level of aspiration."
The level of aSpiration can have varying effects on the indi-
vidual. If it is set so high that the person cannot attain
the goal, he may become depressed. On the other hand, if
the goal is so low that it takes no effort to reach it,
there is little reinforcement for achieving the goal and
little motivation to try again. Also, an individual may
have a double standard of goal behavior. One goal is set
very high and is the Object a person strives to attain: the
other is set not so high and is the level he actually ex-
pects to reach.

It has been claimed that there has been a significant

elevation in the level at which students are setting their

121
goals. Whether this is due to an increased emphasis on
educational achievement or an increased application of
psychological learning principles is not clear.

An important principle of study behavior is the
superiority of diStributed over massed practice. In 1964,
Munn reported that those persons who took periodic short
rests at specified intervals during study not only learned
the material 12% faster but retained it 26s longer.

'rhe principle of retroactive inhibition suggests
the importance of scheduling when we study certain subjects.
The greater the degree of similarity between the original
subject and the subject studied at a later time the more the
latter subject inhibits the retention of the original
material. This suggests, for example, that a person should
not study French and Spanish one right after the other. In~
stead he should follow a foreign language with math or a
similar subject.

An important technique to make use of in learning is
recitation. In learning Studies it was found that those sub~
jects who Spent 60% of their study time in recitation learned
significantly faster than those Spending less time in
recitation (e.g., 20%, 40%, or 60$). In addition, a report
by Morgan and Deese states that a group trained with recita—
tion retained 75% of what was learned while the non-
recitation group retained only 16%.

A second major fact which may have contributed to

the transformation in study behavior mentioned above may be

122
the increasing emphasis in our society on more and better
education. However, the extent to which this may have in-
fluenced study behavior is difficult to determine.

In conclusion, a comparison of study habits for
different countries shows that England's students are more
organized in their study habits than United States students.

We have not tried to present a complete analysis
of this subject in this short discourse, but rather to
present some ideas on the subject for consideration. None-
theless, if any conclusion may have come from this investi—
gation it is that we are nOt sure exactly what is the

nature or magnitude of a transition in study behavior.

Appendix F

Multiple_choice questionnaires

 

Section 1. High importance questionnaire.

 

 

Lem 00110213

 

Below are 14 multiple-choice questions on the article you
read. Select the alternative which best answers the question.
Indicate your choice on the answer sheet by filling in the
box underneath the number correSponding to the number of the
alternative chosen for each question. You must answer every
gpestion.

l. A double standard of sexual behavior means:

1) both petting and pre-marital intercourse are allow-
able means of love making

2) Pre-marital intercourse is all right for one person
but not for the person's boy (or girl) friend

3) the girl should be restrained and the boy make the
advances

4) both 1 and 2

5) none of the above

2. The code of permissiveness—with-affection means

1) any type of affection, includinggre—marital sex, is
permissible for a boy and girl

2) pre—marital intercourse is justifiable as long as
both parties are agreeable prior to the act

3) pre-marital intercourse is all right provided the
couple are in love, "going steady“, engaged, or
affectionately related

4) all of the above

5) none of the above

3. This article suggests that the transformation, if any,

in the sexual ethic may be due to .
1) an increased emphasis on the equality Of the sexes

2) technical improvement in birth control devices

3) the double standard '
4) the code of permissiveness—with—affection

5) both 1 and 2

123

124

The percentage of abortions done in the U.S. by licensed
MD's is

l) 90%

2) 46%

3) 43%

4) 10%

5) 36

According to this article, in the U.S. the requests for
abortions

1) come more from the upper class person than the lower
class person

2) come equally from married and unmarried women

3) come more from the lower class than the upper class
person

4) come more from unmarried than married women

5) both 1 and 2

6) both 3 & 4 (indicate this choice by marking the boxes
for both alternative 3 & 4)

The abortion rate in the U.S. is

l) declining

2) about 200,000 to 1,200,000 annually
3) staying at about the same rate

4) about 26,000 annually

5) none of the above

Reiss in his article on "Sex Behavior in our Culture"
suggests that the restraint and uncertainty of a Single
person today about pre-marital sex behavior 18 due to

l) a failure to adequately define his sex role
2) a conflict between possible sex roles .
3) increasing publicity and attention to the tOpic of sex

4) all of the above
5) none of the above

The present abortion laws in the U.S. reveal that

1) all states allow abortion to save the mother's life

2) 43 states allow abortion to preserve the mother 5
health

3) 3 states allow therapeutic
by a psychiatrist

4) 46 states allow abortions to sav

5) none of the above

abortions if recommended

e the mother's life
The conclusion reached by the author of this article was:

1) Pre—marital sex is now more widely accepted in the
U.S. than before

10.

11.

12.

13.

14.

125

2) Abortions have increased due to more frequent pre-
marital sex

3) We are not sure how much or in what way sexual ethics
have changed

4) We should allow abortions under most conditions in
the U.S.

5) both 1 and 2

This article suggests that if increased pre-marital
intercourse is resulting in a climbing abortion rate it
is due to

1) over confidence in contraceptives

2) ignorance about contraceptives

3) lack of contraceptives

4) incorrect information about contraceptives
5) it is not clear what the cause is

This article states that one of the main deterrents to
pre-marital intercourse is the

1) fear of pregnancy

2) latent, gradual response of the female to love
making

3) strong moral character of our youth

4) conflict between sex roles

5) both 1 and 2

According to this article, which of the following
countries permits both abortions and contraceptives?

1) France
2) England
3) Italy
4) Japan
5) Russia

Kinsey reported in his study on sex behavior that the
frequency of pre-marital relations

1) varied between 32% and 73% for the American male

2) varied between 7% and 67% for the American femalet

3) varied from 10% for those women born before 1900 o
30% or more for those born in 1920 and later

4) was around 20% for the American female

5) both 1 and 2

Studies made on college women reported that the frequency

of pre—marital intercourse was

1) 20%

2) 50%

3) between 7% and 57%
4) all of the above

5) both 1 and 2

126
Section 2. Low importance questionnaire.

INSTRUCTIONS

 

Below are 14 multiple-choice questions on the article you
read. Select the alternative which best answers the question.
Indicate your choice on the answer sheet by filling in the
box underneath the number corresponding to the number of the
alternative chosen for each question. You must answer every

question.

 

l. A doUble standard of goal behavior means:

1) motivation and purpose are essential to effective
study.

2) one goal is the level of achievement the individual
strives to attain: a lower goal is the level he
actually expects to reach.

3) a person has a certain goal in studying but a differ-
ent one for education

4) both 1 and 2

5) none of the above

2. The term "level of aspiration" refers to:

1) the level actually achieved by the person relative
to the level he was striving to reach

2) the level of depression or reinforcement as a result
of whether the goal was attained or not

3) the goal which an individual sets for himself

4) all of the above

5) none of the above

3. This article suggests that the transformation, if any,
in study behavior may be due to:

1) an increasing influence of findings in the area of
psychology of learning

2) an increasing emphasis in our society on more and
better education

3) the double standard

4) level of aspiration

5) both 1 and 2

4. How much more did the recitation group retain in compari-

son to the non—recitation group in the Morgan and Deese
study:

1) 75% to 16%
2) 80% to 16%
3) 60% to 20%
4) 80% to 20%
5) 60% to 16%

127

According to this article, the principle of retroactive
inhibition states that:

l) the greater the similarity between the original sub-
ject and the subject studied afterwards the more
retroactive inhibition

2) study of French should be followed by study of math
Or a similar subject

3) the greater the similarity between the original sub-
ject and the subject studied afterwards the less
retroactive inhibition

4) study of French should be followed by study of Spanish
or another foreign language

5) both 1 and 2

6) both 3 & 4 (indicate this choice by marking boxes for
both alternatives 3 & 4).

A study comparing the effects of massed and distributed
practice revealed that:

1) there is no significant difference in the two methods

2) those who used distributed practice learned the
material 12% faster and retained it 26% longer

3) the massed practice group performed better than those
using distributed practice

4) those who used distributed practice learned the
material 26% faster and retained it 12% longer

5) none of the above

Riess in his article "Study Behavior in Our Society"
suggests that the ambivalence and uncertainty of the
college student about study behavior is due to:

1) a failure to adequately define the "good" student's
role. .

2) a conflict between possible “good“ student roles

3) increasing emphasis on obtaining more education and
high grades

4) all of the above

5) none of the above

Studies on the effect of recitation on learning show
that:

1) those students spending 75% of their study time in
recitation, learned significantly faster than those
Spending less time.

2) those students Spending 60% of their study time in
recitation learned significantly faster than those
spending 20% or 40%

3) those students spending 90% of their study time in
recitation learned significantly faster than those
Spending 60%, 40%|Or 20%

11.

13.

4)

5)

12d

those students pending 80% of their study time in
recitation learned significantly faster than those
spending less time.

none of the above

‘Ihe conclusion reached by the author of this article
was:

1)

learning principles are more widely used in the U.S.
than before

2) there has been a change in study behavior due to an
increased emphasis on education

3) we are not sure how much or in what way study be—
havior has changed

4) we should put more emphasis on the use of learning
principles in studying

5) boch 1 and 2

'rnis article suggests that if an elevation is occurring

in the level at which students are setting their goals

it is due to:

1) an increased (monas1a on ed.ucational achit vement

2) an increased aaplicaticn of psychological learning
principles

3) too much emphasis on doing better than the next
were“

4) too muCh information about the importance of goals

5) it is not clear what the cause is

:his article States that a primary source of motivation

for the college student 'S:

l) the goals he sets

2) the cesire to achieve

3) the strong competitive nature of our education system

4) conflict between “good” Student roles

5) both 1 and 2

According to this tic cle, whicn of th.: following coun—

tries'

SCUCC' TIES

w
H,
(1

m1 most organize:1 in their study

habits?

1) francs

2) Onitco Stat-s

3) Italy

4) Lngland

5) Russia

Latl reportcc in his anal/sis of Ltucy raoite that tne
hunter or pectl; u~ing orcaniz ed learninC methOts:

1) varied b;twetn 32m and l3w far .males

2) varied between 7% and 67; for 1al

14.

129
3) varied from 10% for those persons born before 1900
to 30% or more for those born in 1920 and later.

4) was around 20% for college students
5) both 1 and 2

Studies made on college students reported that the fre-
quency of using organized methods in studying was:

1) 20%

2) 50%

3) between 7% and 57%
4) all of the above
5) both 1 and 2

Impendix<3

ating scales of subjects' perception

)

of exoerimental manipulations

31)

Put a check mark in the Space above the term(s) which best
describes your answer to each of the following questions.

e.g. If your reSponse to the question "how do you
feel today?“ is "Very well“, you would indi-
cate it as below.

A A L / / /
Very bad uuite bad SomeWhat somewhat uuite Very
bad well well well

1. How much do you like the article you read last week?

 

A._ L /_ / / /
DiSliked disliked Disliked Liked Liked Liked
a lot somewhat a little a little somewhat a lot

2. How difficult to understand was the article you read?

 

L / / / / /
hot at all a little Somewhat uuite Very
difficult difficult difficult difficult difficult

3. flow much freedom did you feel you had in the selection
of the article read?

 

no freedom A little some Quite a lot very much
freedom freedom of freedom freedom

4. Have you talked to anyone about the nature or content of

this experiment since last week?

M
(D
m
L-
O

130

M1': HIGAN STATE UNIVERSITY LIBRARIES

I”| III III III

98