ABSTRACT

AN ANALYSIS OF THE EFFECTIVENESS OF POST-TRANSACTION
COMMUNICATIONS IN REDUCING COGNITIVE DISSONANCE

BY
Shelby D. Hunt

The purpose of the research was to investigate some
of the decision oriented marketing implications of Leon
Festinger's theory of cognitive dissonance. The study used
an "after only with control" design to examine whether cer-
tain post-transaction messages from a retailer to recent
purchasers would be effective in assisting the consumer in
reducing dissonance, and, if the messages were effective in
reducing dissonance, what benefits would accrue to the
retailer for providing the messages.

A sample population consisting of purchasers of
refrigerators from a major department store in the Detroit
Metropolitan Area was randomly divided into a control and
two experimental groups with a total sample size of one
hundred and fifty-two households. One experimental group
received a letter from the department store giving the con-
sumers post-transaction reassurances that buying a specific
appliance was an excellent decision. The second experimen—

tal group received a telephone call from the department

Shelby D. Hunt

store expressing the same types of reassurances as were
given in the letter. The control group received no post-
transaction message.

Interviews were conducted with all subjects approxi-
mately 7 days after the transaction.

To determine whether the messages were effective in
reducing dissonance, the extent of post-transaction doubt
concerning the purchase decision was used as an indicator of
the magnitude of dissonance the subjects were experiencing.
The questions were constructed using a seven step rating
scale similar to a semantic differential type of instrument.

The major guiding hypothesis of the study was that
the dissonance scores of the two experimental groups would
be lower, less dissonant, than the scores of the subjects in
the control group. The findings of the study revealed that
the subjects in the group which received the post-transaction
letter were less dissonant than the control group, but the
subjects in the group which received the telephone call had
higher dissonance scores than did the control group.

The second guiding hypothesis was that the groups
which received the post-transaction messages would have more
favorable attitudes toward the store than the control group.
The findings revealed a general tendency for the group which
received the letter to have more favorable attitudes toward

the store than the control group. Contrary to the hypothesis,

Shelby D. Hunt

the group which received the telephone call had less favor-
able attitudes toward the store than the control group.

The third guiding hypothesis was that the groups
which received the post-transaction reassurances would be
more predisposed to purchase at the store than the control
group. The findings of the study revealed that the subjects
who received the letter tended to have higher eXpectations
of future purchases than did the subjects in the control
group. Contrary to hypothesis, the subjects in the tele-
phone call group tended to have lower expectations of future
purchases than did the control group.

The fourth guiding hypothesis concerned whether cer-
tain types of consumers would be more likely to experience
dissonance than others. The findings showed no significant
relationship between dissonance and any of the demographic
variables investigated.

Since one type of post-transaction message, the
telephone call, tended to increase dissonance, the study
concludes that a general policy recommendation of providing
post-transaction reassurances to recent purchasers may be
ill-advised. Much more research is needed before defini-
tive policy recommendations can be made using cognitive

dissonance as a theoretical foundation.

AN ANALYSIS OF THE EFFECTIVENESS OF POST-TRANSACTION

COMMUNICATIONS IN REDUCING COGNITIVE DISSONANCE

BY

Shelby D. Hunt

A THESIS

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

DOCTOR OF PHILOSOPHY
Department of Marketing and Transportation

1968

ii

©
Copyright by
SHELBY D. HUNT

1969

ACKNOWLEDGMENTS

A doctoral dissertation is never the work of only

one individual. I would like to express my thanks to those

who assisted me in the study.

Dr.

Dr.

Dr.

The

Bernard J. LaLonde, Professor, Department of Market-
ing and Transportation, who, as chairman of my dis-
sertation committee, provided valuable counsel
throughout all phases of the research project and
gave unselfishly of his time and energy whenever a
crisis arose.

Stanley C. Hollander, Professor, Department of
Marketing and Transportation, whose scholarly
demeanor is an inspiration to all his students.

Hideya Kumata, Professor, Department of Communica-
tion, whose early suggestions considerably strength-
ened the research design.

J. L. Hudson Company, Detroit, Michigan, whose
financial assistance and c00peration made the study
possible. Special thanks go to Howard Kushmar,
Assistant Director of Research, for his technical
assistance.

Finally, I should like to express my deep apprecia-

tion for the efforts of my wife, Suzanne, who not only con-

tributed her typing expertise, but also assisted in the

field work and was a consistent source of encouragement and

understanding.

iii

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . .

Chapter

I.

II.

INTRODUCTION . . . . . . . . . . . . . . .

Purpose of Study . . . . . . . . . . . .
Review of the Theory of Cognitive
Dissonance . . . . . . . . . . . . . .
Implications of Dissonance Theory
to Marketing . . . . . . . . . . .
Statement of the Problem . . . . . . . .
Hypotheses . . . . . . . . . . . . . . .
Methodology . . . . . . . . . . . . . .
Limitations of the Study . . . . . . .
Contributions of the Study to Marketing
Theory . . . . . . . . . . . . . .
Contributions of the Study to Marketing
Practice . . . . . . . . . . . . . . .
Organization . . . . . . . . . . . . . .

CONCEPTUAL FOUNDATIONS: A REVIEW OF THE

LITERATURE . . . . . . . . . . . . . . .

Introduction . . . . . . . . . . . . . .
Modes of Dissonance Reduction . . . . .
Selective Exposure to Information

, as a Mode of Dissonance Reduction
Selective Recall of Information as
a Mode of Dissonance Reduction . .
Perceptual Distortion as a Mode of
Dissonance Reduction . . . . . .
Behavioral Change as a Mode of
Dissonance Reduction . . . . . . .
Effects of Information on Dissonance
Reduction . . . . . . . . . . . . . .
Source Credibility and Dissonance
Reduction . . . . . . . . . . . . . .
Attitude Change and Cognitive Dissonance
Consumer Characteristics and Cognitive
Dissonance . . . . ,,. . . . . . . . .

iv

Page

vii

H

l—‘CDO‘U'IUJ

13
15
l7

l7
l7

19
23
25
26
28

31
32

39

Chapter Page

Cognitive Dissonance Experimentation:

A Commentary . . . . . . . . . . . . . . . 44

III. RESEARCH DESIGN . . . . . . . . . . . . . . . 48
Basic Design . . . . . . . . . . . . . . . . 48
Sample Selection Procedure . . . . . . . . . 49
Treatment Procedure . . . . . . . . . . . . 51
Interview Procedure . . . . . . . . . . . . 52

The Questionnaire . . . . . . . . . . . 54

Part I: Perceived Dissonance . . . . . 56

Part II: Store Image . . . . . . . . . 57

Part III: Predisposition to Buy . . . . 58

Part IV: Shopping Effort . . . . . . . 59

Part V: Demographic Data . . . . . . . 60

Data Collection and Analysis . . . . . . . . 61

IV. THE FINDINGS . . . . . . . . . . . . . . . . . 63
IntrodUction . . . . . . . . . . . . . . . . 63
General Results . . . . . . . . . . . . . . 64

The Population . . . . . . . . . . . 65

The Effectiveness of the Post-Transaction
Communications in Reducing Cognitive

Dissonance . . . . . . . . . . . . . . . . 67
.Attitude Change and Dissonance Reduction . . 7l
Predisposition to Purchase and Dissonance
Reduction . . . . . . . . . . . . . . . . 76
Consumer Characteristics and Cognitive
Dissonance . . . . . . . . . . . . . . . . 78
A Concluding Note . . . . . . . . . . . . . 79
V. SUMMARY AND CONCLUSIONS . . . . . . . . . . . 80
Introduction . . . . . . . . . . . . . . . . 80
Brief Summary of Purpose and Research
Design . . . . . . . . . . . . . . . . . . 80
Review of Hypotheses and Findings . . . . . 82

Guiding Hypothesis No. l: The
Effectiveness of Post—Transaction
Communications in Reducing

Cognitive Dissonance . . . . . . . . . 82
Findings of the Study . . . . . . . . . 82
Guiding Hypothesis No. 2: Attitude

Change and Dissonance Reduction . . . 83
Findings of the Study . . . . . . . . . 83

Chapter
Guiding Hypothesis No. 3:
Predisposition to Purchase and
Dissonance Reduction . . . . . . .
Findings of the Study . . . . . . .
Guiding Hypothesis No. 4: Consumer
Characteristics and Cognitive
Dissonance . . . . . . . . . . . .
Findings of the Study . . . . . . .
Conclusions and Implications for
Marketing . . . . . . . . . . . . . .
Cognitive Dissonance and Marketing:
A Commentary . . . . . . . . . . . . .
Suggestions for Future Research . . . .
Appendix
A. TABLE 12 . . . . . . . . . . . . . . . . .
B. RESEARCH HYPOTHESES . . . . . . . . . . .
C 0 THE QUESTIONS USED 0 o o o o o o o o o o o
D. INTERVIEWER INSTRUCTIONS AND QUESTIONNAIRE
E. THE POST-TRANSACTION LETTER MESSAGE . . .
F. THE POST-TRANSACTION TELEPHONE MESSAGE . .
G. STATISTICAL PROCEDURES . . . . . . . . . .
H. LIST OF SUBURBS INCLUDED IN THE SAMPLE . .

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . .

vi

Page

83
84

84
85
85

88
91

93

95
100
106
122
123
124
129

130

Table

1.

LIST OF TABLES

Comparison of the overall means of the
three attitudes in common between the
present study and R. G. Wyckham's study . . .

Summary of Mann4Whitney U and associated
probabilities with the null hypotheses

that the attitudes of the sample of the
general public taken by R. G. Wyckham

are the same as the attitudes of the

groups composing the present sample . . . . .

Overall mean perceived dissonance by group .

Summary of Kruskal4Wallis H and associated
probabilities with the null hypothesis

that the perceived dissonance scores of

the three groups came from the same
population . . . . . . . . . . . . . . . . .

Summary of Mann4Whitney U and associated
probabilities with the null hypotheses that
each pair of groups comes from the same
population . . . . . . . . . . . . . . . . .

Overall mean for each attitude for each
group O C O O O O O O C O O O O I O O 9 O 0

Summary of KruskalAWallis H and associated
probabilities with the null hypothesis that
for each attitude the scores of the three
groups came from the same population . . . .

Summary of Mann4Whitney U and associated
probabilities with the null hypothesis that
the differences in attitude scores between
pairs of groups come from the same population

Mean predisposition to purchase for each
group O O C I O O O O O O O I O O O O O O O .

vii

Page

66

68

69

7O

71

72

74

75

76

Table

10.

11.

12.

Page

Summary of KruskalAWallis H and associated
probabilities with the null hypothesis that

the predisposition to purchase scores of

the three groups came from the same

population . . . . . . . . . . . . . . . . . . 77

Summary of Mann4Whitney U and associated
probabilities with the null hypothesis

that the differences in predisposition

to purchase between pairs of groups come

from the same population . . . . . . . . . . . 78

Summary of Kruskal4Wallis H and associated
probabilities with the null hypothesis that

the perceived dissonance scores of the

subjects when classified by the various

consumer characteristics came from the

same population . . . . . . . . . . . . . . . . 93

viii

CHAPTER I

INTRODUCTION

Purpose of Study

There have been few theories developed in the behav—
ioral sciences which have sparked as much empirical research
as Leon Festinger's theory of cognitive dissonance [23].
Although the research in the area is voluminous, very little
of it is of direct applicability to marketing. The purpose
of the research is to investigate some marketing implica-
tions derived from the theory of cognitive dissonance.
Before discussing these implications, and the research meth-
odology, it is appropriate to briefly review Festinger's
theory.
Review of the Theory of
Cognitive Dissonance

Definition 1: "[Cognitions are] the things a person

 

knows about himself, about his behavior, and about his sur-
roundings. These elements, then, are "knowledges". . . [23:9].

Definition 2: "Two [cognitions] are in a dissonant
relation if, considering these two alone, the obverse of one

element would follow from the other" [23:13].

Definition 3: "[Two cognitions are consonant] if
. . . one does follow from the other" [23:15].

Definition 4: "[Two cognitions are irrelevant] where

 

one cognitive element implies nothing at all concerning [the]
other element“ [23:11].

To illustrate the above definitions, assume that a
housewife had been considering the merits of a General Elec-
tric washing machine versus a Whirlpool machine, and that
she finally purchased a Whirlpool machine. The cognition
that the General Electric "mini-basket" is a desirable fea—

ture to have on a washing machine would be dissonant with

 

the cognition that the purchaser actually selected a Whirl-
pool washing machine. The cognition that the General Elec-
tric machine did not have a "self cleaning filter" is con-
sonant with the cognition that she selected a Whirlpool.
The cognition that the housewife likes sirloin steak would
be irrelevant to the cognition that she selected a Whirlpool.
With the above definitions in mind, the core of the
theory of cognitive dissonance may be stated:

1. There may exist dissonant or "nonfitting" relations
among cognitive elements.

2. The existence of dissonance gives rise to pressures
to reduce the dissonance and to avoid increases in
dissonance.

3. Manifestations of the operation of these pressures
include behavior changes, changes of cognition, and
circumspect exPosure to new information and new
opinions [23:31].

The basic hypotheses which Festinger promotes are that: "the

existence of dissonance, being psychologically uncomfortable,

will motivate the person to try to reduce dissonance and
achieve consonance" [23:3]. "The magnitude of postdecision
dissonance is an increasing function of the general impor-
tance of the decision and of the relative attractiveness of
the unchosen alternatives" [23:264].

Given that the person is experiencing some magnitude
of cognitive dissonance, there are three major methods by
which the person may reduce this dissonance through time:

1. By changing one or more of the elements involved in
dissonant relations.

2. By adding new cognitive elements that are consonant
with already existing cognitions.

3. By decreasing the importance of the elements in-
volved in the dissonant relations [23:264].

Implications of Dissonance Theory
to Marketygl

If the theory of cognitive dissonance is an accurate
representation of reality, then a multitude of marketing
implications can be derived:

1. Since most products are available in many alterna-
tive stores, a consumer will experience dissonance
as a result of selecting a particular store. There
will be some favorable cognitions with respect to
the alternatives not selected.

2. If several brands are available for a product, then
dissonance will be present if the unchosen brands
have attributes which are superior to the chosen

brand as perceived by the consumer.

3.

A consumer eXperiencing cognitive dissonance as a
result of store selection will seek, and be very
receptive to, advertising by that store as a means
of reducing cognitive dissonance.

A consumer exPeriencing cognitive dissonance as a
result of the brand selection decision will seek,
and be receptive to, the advertising of the manufac-
turer as a means of reducing dissonance. The recep-
tivity of the consumer to manufacturer's messages
will be greater post-decision than pre-decision
since dissonance is a post-decision phenomenon
[24:19,30].

The greater the importance of the purchase decision
(e.g., the higher the price) the greater the disso-
nance--ceteris paribus.

The greater the number of stores (brands) for a
particular purchase, the greater the dissonance.

If a consumer finds the level of dissonance intoler—
able, and discovers no method for reducing the dis—
sonance to a tolerable level, he may negate the
decision by returning the product purchased [31:75].
Since the consumer seeks means to reduce dissonance
after the sale, it may be wise not to "oversell" him
on the product. It is preferable for the product to
exceed consumer expectations rather than merely

satisfy them or fall short of them [50:62].

9. When post-decision dissonance is reduced, the ini-
tial decision becomes difficult to reverse. Thus,
there is a likelihood of complete adoption, that is,
repeat sales [50:63].

10. Post-decision dissonance is likely to be greatest
after the purchase of recently established products

from lesser known companies [50:64].

The above delineated marketing implications have not
gone unnoticed by the authors of many modern marketing text—
books. The following quotation is typical of suggestions
which are frequently made:

The existence of possible negative postpurchase feelings
indicates the marketer might benefit from directing some
of his communications to the recent buyer, rather than
all of them to the potential buyer. The recent buyer
may need assurance that he has made the right choice.

If he is in a dissonant state, he will be looking for
supportive evidence in the form of advertising and other
communications. The appliance manufacturer may take the
opportunity of building assurances into the information
brochures that accompany his product. Manufacturers
also can run advertisements showing recent purchasers
showing satisfaction with their choice, and why. Unless
the seller dispels the dissonance by some positive
efforts, he may lose the customer unnecessarily [31:75].

Statement of the Problem

The theory of cognitive dissonance would lead one to
conclude that the consumer would be very receptive to commu-
nications from the retailer and the manufacturer immediately

after an "important" purchase. The research is designed to

answer the following questions:

1. Do consumers eXperience cognitive dissonance after
the purchase of a "major" appliance?

2. Are some types of consumers more likely than others
'to eXperience dissonance?

3. Are communications from retailers and/or manufactur-
ers effective in assisting consumers in their
efforts to reduce cognitive dissonance?

4. What is the relative effectiveness of various types
of communications in assisting consumers in their
efforts to reduce cognitive dissonance?

5. ;£_the communications are effective in assisting
consumers in their efforts to reduce cognitive dis-
sonance, what benefits will accrue to the retailer

for supplying these communications?

Hypotheses

The research hypotheses are reproduced in detail in
Appendix B. The hypotheses are divided into five sections
with the following guiding hypotheses as the underlying
rationale:

A. The major guiding hypothesis of the research is
that manufacturers and/or retailers can assist consumers in
their efforts to reduce cognitive dissonance by providing
post-transaction reassurances. Furthermore, not all types
of post-transaction messages will be equally effective. The

research hypotheses in section A posit the effectiveness of

the different messages in assisting consumers in their
efforts to reduce cognitive dissonance.

B. A second guiding hypothesis is that one of the
benefits which will accrue to the retailer for providing
dissonance reducing information is that the groups which
receive assistance will then view the store more favorably
than the groups which do not receive assistance. The
research hypotheses in section B posit the differences in
the attitudinal dimensions which comprise the store image of
the three groups.

C. A third guiding hypothesis is that retailers
will benefit from providing dissonance reducing information
because the assisted groups will be more predisposed to pur—
chase at that store than the group which does not receive
assistance. Section C contains hypotheses positing differ—
ences in predisposition to buy among the three groups.

D. Past research has indicated that people who
expend more effort shopping will experience less cognitive
dissonance [13]. Using this as a guiding hypothesis, sec—
tion D posits the relationship between shopping effort and
perceived dissonance for each group.

E. Prior research has indicated that certain per—
sonality types are more likely to experience cognitive dis-
sonance than others [8:14-15]. However, there has been no
research which would suggest that the level of cognitive

dissonance should vary with any of the six demographic

characteristics of age, education, family income, social
class, stage in the family life cycle, and race. The re-
search hypotheses in section E postulate the relationships
between the demographic variables and perceived dissonance

for each group.

Methodology

A department store was asked to c00perate in the re-
search project by supplying lists of the names and addresses
of customers who recently purchased refrigerators. Except
for persons without established credit, the names were avail—
able on the day following the transaction. Names of those
customers without established credit were received as soon
as credit was approved. The customers were divided on a
random basis into three groups:

1. Group 1 was a control group and received no post-
transaction message from the retailer.

2. Group 2 was sent a letter on department store
stationery, and signed by the appliance department
manager, expressing his appreciation for the custom—
er's purchase and reassuring her that her selection
of that particular brand and model was a "good" one.
(See Appendix E for a copy of the letter.)

3. Group 3 received a telephone call from the researcher,
who, with permission, represented himself as a

department store employee. The oral message used

similar copy to that which group 2 received (see

Appendix F).

Interviews were conducted by professional interview-
ers one to three days after the consumer received the post-
transaction message, the five part questionnaire was admin-
istered (see Appendix C). Part I of the questionnaire con-
sisted of six questions on a semantic differential scale,
the mean of which was used as an indicator of perceived
dissonance.

Part II of the questionnaire consisted of semantic
differential questions to measure the five attitudinal dimen-
sions postulated to make up the department store image:

1. Merchandise suitability
2. Store services

3. Store congeniality

4. Sales personnel

5. Locational convenience.

The above attitudinal dimensions were drawn in part
from a study of aggregate department store images by Robert
G. Wyckham [48:84].

One of the benefits hypothesized to accrue to retail-
ers when they assist consumers in their attempts to reduce
dissonance is that the consumers who receive assistance will
have more favorable attitudes toward the store than the con—

sumers who do not receive assistance, the control group.

10

Part III of the questionnaire was to determine how
the consumer perceived his future shopping behavior. It is
hypothesized that the groups which receive assistance from
the retailer will be more predisposed to purchase at the
retailer's store than those who do not receive assistance.

Part IV was to determine whether perceived disso-
nance decreased as shopping effort increased, as has been
reported in the literature [13].

Part V contained demographic data of seven types:

1. Age

2. Education

3. Family income

4. Race

5. Brand eXperience

6. Stage in family life cycle

7. Social class,

The above data were used to determine if there were
any demographic correlates with the level of perceived dis-
sonance, i.e., were certain types of people more likely to
eXperience dissonance than others?

A computer program was then used to test the various
specific research hypotheses using the Mann4Whitney U test
statistic and the KruskalAWallis onedway analysis of vari-

ance (see Appendix G).

11

Limitations pf the Study

results

1.

Caution should be exercised in generalizing the

of the study because of the following reasons:
Since the study was limited to only one type of
purchase (refrigerators), generalizing to other
types of purchases may not be possible.

Since the study was conducted in only one metropol-
itan area (Detroit), generalizing to other geo-
graphic areas may not be possible.

The study used an "after only" measurement; hence,
the determination of causal relationships between
the independent variables (the messages) and the
dependent variables (perceived dissonance, etc.) is
basically inferential rather than eXperimental.
There is no validated technique for measuring dis-

sonance directly.

Contributions of the Study to
Marketing Theory

It is anticipated that the research will be useful

for those who are attempting to develop a meaningful theory

of consumer behavior by placing in proper perspective the

role of

general

cognitive dissonance theory as a component of a

theory of consumer behavior. Possible specific

contributions to marketing theory are as follows:

1.

Cognitive dissonance theory postulates that a person

will experience dissonance after a decision when the

12

decision is "important" and where the unchosen alter-
natives have at least gpmg characteristics which are
superior to the chosen alternative [23:262]. It
would appear that many consumer purchase decisions
could be dissonance producing, and, hence, cognitive
dissonance theory should be an integral part of any

general theory of consumer behavior. The research

[provides information on whether at least one par—

ticular type of purchase decision is dissonance
producing.

If consumers g2 eXperience dissonance after a major
purchase decision, it is important to know whether
some types of consumers are more likely than others
to eXperience dissonance. Some evidence is avail-
able in the literature that certain personality
types are more likely than others to experience
cognitive dissonance. The research provides infor-
mation as to whether the level of dissonance varies
with any of the six demographic variables of age,
education, family income, race, life cycle, and
social class.

Cognitive dissonance theory postulates that certain
types of pre-decision activities will lead to a
lessening of post-decision dissonance. In partic-
ular it is held that the physical act of shopping

is one such activity [13]. The research provides

13

information as to the relationship between shopping
effort and dissonance.
4. Cognitive dissonance theory posits the existence of
several different methods by which people can reduce
dissonance [23:264]. Unfortunately, the theory does
not predict which particular method will be used
under a particular set of circumstances. The re-
search provides information as to whether the con-
sumer can be induced to use a mode of dissonance
reduction which is favorable to the manufacturer
and/or retailer.
Contr$butionspp§ithe Study
to Marketing Practice

To simply theorize that consumers eXperience cogni-
tive dissonance is of little significance to the marketing
practitioner unless he has guidance as to what he can do
about dissonance, what he should do about dissonance, and
what benefits will accrue to him if he does do something
about dissonance. In short, what are the action-oriented
implications of dissonance? Specific potential contribu-
tions are as follows:

1. The study provides information as to whether certain
types of post-transaction activities, i.e., the spe-
cific communications, will be effective in reducing

dissonance. Furthermore, the study indicates the

14

relative effectiveness of alternative actions, i.e.,
the different messages, in reducing dissonance.

A benefit posited to accrue to retailers and/or
manufacturers for providing their customers with
dissonance reducing information is that the recip-
ients of the information will be more predisposed to
purchase from the retailer and/or manufacturer who
provides the information. The study tests whether
the groups who receive assistance are in fact more
prediSposed to purchase than those who do not
receive assistance.

A second benefit which might accrue to manufacturers
and/or retailers for providing dissonance reducing
information is that the assisted customer will have
a more favorable perception of the store's total
offering, i.e., "store image," than the customers
who do not receive assistance. The study provides
information to determine if the above actually
occurs.

It is possible that retailers and/or manufacturers
might want to send post-transaction communications
to only certain types of purchasers. The study pro-
vides information as to which types of customers
would most benefit from dissonance reducing infor-

mation.

15

Organization

The organization of the study will consist of five
chapters:
1. Introduction
2. Conceptual Foundations: A Review of the Literature
3. Research Design
4. The Findings

5. Conclusions and Implications for Marketing.

Chapter I introduces the study by stating its pur-
pose and briefly reviewing the theory of cognitive dissonance
and its implications to marketing. The research hypotheses
are presented along with the research design and the poten-
tial contributions to both marketing theory and marketing
practice.

Chapter II first reviews the literature with respect
to the various methods people employ to reduce dissonance.
The effects of information on dissonance reduction are dis-
cussed as well as the literature on source credibility and
cognitive dissonance. Since some of the research hypotheses
concern attitude change, the literature on attitude change
and dissonance is reviewed, along with the relationships
between dissonance and several consumer characteristics.

Chapter II sets forth the research design of the
study including the basic design, sample selection procedure,

treatment procedure, and the research instrument.

16

Chapter IV presents the findings of the study by
first discussing the general results and the population.
The rest of the findings follow the organization of the
research hypotheses as shown in Appendix C.

Chapter V briefly summarizes the purpose and
research design and then presents a review of the major
findings and conclusions. The chapter concludes with the
implications of the study to marketing and suggests some

directions for future research.

CHAPTER II

CONCEPTUAL FOUNDATIONS: A REVIEW

OF THE LITERATURE

Introduction
Chapter II is an effort to establish the underlying

theoretical structure for the study by reviewing the litera—
ture relating to the research hypotheses. The literature
review will be divided into five sections:

1. Modes of dissonance reduction

2. Effects of information on dissonance reduction

3. Source credibility and dissonance reduction

4. Attitude change and dissonance reduction

5. Consumer characteristics and cognitive dissonance.

Modes of Dissonance Reduction
Given that the consumer may be experiencing some
magnitude of cognitive dissonance after a purchase, Festinger
hypothesizes three major methods by which the consumer may
reduce this dissonance:
1. By changing one or more of the cognitive elements
involved in the dissonant relations.

2. By adding new cognitive elements that are consonant
with already existing cognitions.

17

18

3. By decreasing the importance of the elements
involved in the dissonant relations [23:264].

Brehm and Cohen expand upon the above list by includ-
ing the specific dependent variables measured in various
experiments as indicators of modes of dissonance reduction
[11:306-308].

1. Attitude change

a. Opinion change
b. Evaluation change

2. Selective eXposure to information
3. Selective recall of information
4. Perceptual distortion

5. Behavioral change.

The thrust of the present study is to attempt to
determine whether certain communications from the retailer
or manufacturer would be effective in assisting consumers in
their efforts to reduce dissonance after a major purchase.
This implies that the retailer is attempting to induce the
consumer to adopt a particular mode of dissonance reduction,
i.e., to add new Eggnipifie elements that are consonant with
existing cognitions. It is, therefore, appropriate to
review the literature with respect to the methods people

employ to reduce dissonance.

19

Selective Exposure to Information
as a Mode of Dissonance Reduction

One of the earliest studies concerning selective
perception and dissonance was conducted by Ehrlich, Guttman,
Schonbach and Mills in 1957 [18:98-102]. Ehrlich et al.
used a survey design with a population consisting of people
who had purchased a new automobile within the past two
months and people who purchased a car at least a year earlier
and who owned cars at least three years old.

Two basic kinds of data were obtained: which auto—
mdbile ads were preferred, from a list of heretofore unseen
ads, and the percentage of actual ads that the respondents
had seen and read in recent magazines and newspapers.

Ehrlich et a1. hypothesized that new car owners
would be eXperiencing cognitive dissonance and would, there-
fore: seek out information supportive of their decision
more than old car owners and, also, would avoid information
non-supportive of their decision more than old car owners.
Advertising preference data did not support either of the
two hypotheses. Advertising readerships data weakly sup-
ported the first hypothesis and actually contradicted the
second hypothesis, i.e., ads about cars that had not been
considered at all were read less than ads about cars that
had been considered but not bought.

Another study on selective exposure and cognitive
dissonance was conducted by Mills, Aronson, and Robinson

[34:250-253]. Unlike the Ehrlich et a1. study, the Mills

20

et al. study used an experimental manipulation of dissonance.
Students were given a choice between taking an essay examina-
tion or a multiple choice examination. Dissonance was manip-
ulated by telling some students that the examination would
count 5 percent [low importance-low dissonance] and others

70 percent [high importance-high dissonance].

The Opportunity to adopt selective exposure was pro-
vided by requesting that the students rank order their
preferences to read several articles which were either posi—
tive or negative to essay and multiple choice examinations.
The investigators hypothesized that the high dissonance
group would show a stronger preference for articles support-
ing their decision than the low dissonance group.

The results of the study did not yield evidence in
support of the study. A replication of this study by Rosen
also failed to show evidence of subjects using selective
exposure as a mode of dissonance reduction {40:188-190].

Adams' study attempted to demonstrate that people
who are exposed to counter-attitudinal information would
eXperience dissonance and would, therefore, selectively
expose themselves to pro-attitudinal information [1:74-78].
Mothers were asked to express their opinion as to whether a
child's behavior is mostly inborn or mostly learned. The
mothers were then eXposed to a tape recorded talk by an
“expert" with about half being eXposed to a view Opposing

their own. The mothers were then given the opportunity to

21

eXpose themselves to one of two talks to be held at the
university.

If dissonance and subsequently selective exposure
were operative, then one would eXpect the high dissonance
subjects to prefer consonant information more than low
dissonance subjects. This hypothesis was not confirmed.

Brehm and Cohen stress the importance of commitment
in using cognitive dissonance theory as a predictive theory
[11:7-10]. Sears and Freedman used strength of commitment
as an experimental variable in a study using a mock jury
situation [41]. High commitment (high dissonance) was
obtained by telling some subjects that their first decisions
were irrevocable and were to be made public. Low commitment
(low dissonance) was obtained by telling the subjects that
their first decisions could later be changed and were to
remain confidential.

Once again there was no difference in preferences
between the two groups for supportive information as disso-
nance theory would have predicted, nor did the two groups
differ in duration of exposure to subsequent supportive and
non-supportive communication.

Aaron Lowin proposed that the reason many eXperimen-
tal studies failed to demonstrate selective eXposure on the
part of dissonant subjects was that at least two modes of
dealing with the dissonant information might be in use [33].

Merging approach-avoidance concepts with dissonance theory,

22

he proposed that a subject might either avoid dissonant
information or, alternatively, approach the dissonant
information and attempt to refute it. The choice between
approach or avoidance was proposed to depend, at least in
part, on the ease of message refutation.

Six hypotheses were developed:

1. Strong consonant information is preferred to weak
consonant.

2. Weak dissonant is preferred to strong dissonant.
3. Strong consonant is preferred to strong dissonant.
4. Weak dissonant is preferred to weak consonant.

5. Strong information desired by agreers more than
disagreers.

6. Weak information desired by disagreers more than
agreers.

All six of the hypotheses were strongly supported by
an eXperiment on self exposure to political propoganda exe-
cuted during the Johnson-Goldwater campaign of 1964. Unfor-
tunately, there was no eXperimental manipulation of disso-
nance and, hence, the study does not provide strong support
for the ligk_between dissonance and selective exposure.

Harold H. Kassarjian and Joel B. Cohen used a survey
design to investigate the public's reaction to the surgeon
general's report on smoking and health [30]. They came to
the following conclusion with respect to the confirmed

smoker:

23

With rare exceptions the confirmed smoker appears to

be behaving consistently with his belief system by
continuing to smoke. He has justified its rationality
either by disassociating his responsibility over the
decision; by denying, distorting, misperceiving, or
minimizing the degree of health hazard involved; and/or
by selectively drawing out new cognitions and new infor-
mation that will reduce the inconsistency of his own
behavior and achieve consonance in his own cognitive
world [30:63].

J. F. Engel attempted to replicate in a survey design
[19] some aspects of the earlier study by Ehrlich et a1.
[18]. Two groups of subjects, one who had recently purchased
new Chevrolets and a second who had not, were briefly exposed
to recent automObile advertisements. If subjects responded
that they had seen the advertisement, they were asked to
"play back" as much of it as they could.

The investigators concluded that recent Chevrolet
purchasers had not selectively exposed themselves to poten-
tially dissonance reducing advertisements with the exception
of one advertisement not intentionally included in the study
which listed prices of new Chevrolets.

Selective Recall of Information
as a Mode of Dissonance Reduction

Brehm designed an eXperiment to attempt to determine
whether dissonance would result from a person's receiving
information discrepant with his own, recently given, explic-
itly stated views [11:94]. Subjects were given the oppor-

tunity to selectively recall information so as to reduce

dissonance.

24

The experiment was conducted by asking students to
fill out a personality rating sheet on himself, a close
friend, and a movie star. The sheet consisted of 40 person-
ality traits (lazy, intelligent, etc.) with the following
scale responses: "Not at all," "Moderately," "Very," and
"Extremely." The subjects were then shown rating sheets on
him purportedly made by his friend. Actually, 30 of the 40
ratings were identical with his own ratings of himself and
10 were discrepant by 2 points.

Subjects were then asked to attempt to memorize the
ficticious rating sheet and to recall it immediately and
then recall it after one week.

Since the discrepant information was eXpected to be
dissonant producing, it was hypothesized that a smaller per-
centage of the discrepant information would be recalled.

The results confirmed this hypothesis at the 5 percent sig-
nificance level.

One part of a study by J. F. Engel also investigated
selective recall as a mode of dissonance reduction [19].

Two groups of subjects, one who had recently purchased a
Chevrolet and one who had not recently purchased a Chevrolet
were exposed to lists of statements about Chevrolets and
Plymouths. About 15 minutes after exposure the subjects

were asked to recall as many of the statements as they could.

25

The results of this did not show selective recall to
be operative. Chevrolet owners recalled about as many disso-

nant and consonant statements as did non—Chevrolet owners.

Perceptual Distortion as a
Mode of Dissonance Reduction

Bramel designed a study to determine whether the
defense mechanism of projection would be used as a mode of
dissonance reduction [9:121-129]. Subjects were given fake
"personality" tests and then told on a random basis that the
results were either "favorable" or "unfavorable." The sub—
jects were then hooked up to a fake galvonic skin response
apparatus and were told that a dial in front of them would
register any homosexual tendencies as they were exposed to
pictures of men in various stages of undress. Actually
the dial was manipulated by the investigator to get high
responses to pictures of men with few clothes on. The sub-
jects were asked to estimate the dial position for each
picture for another person whom they had just met.

Dissonance theory predicted that the subjects who
received a "favorable" personality report would be more
dissonant than those who received an "unfavorable" personal-
ity report. Furthermore, the higher dissonance subjects
should "project" higher homosexual scores to other peOple.
The results of the experiment confirmed this hypothesis,

i.e., projection was used as a mode of dissonance reduction.

26

Yaryan and Festinger designed an experiment to deter-
mine if dissonance might produce a perceptual distortion of
the probability of a future event occurring [49]. Subjects
were given the same information concerning the possibility
of having to take an examination over some difficult sym-
bolic definitions. :One group was told to memorize the defi-
nitions (high effort-high dissonance), while the second
group was told to give them a cursory examination (low-
effort-low dissonance).

The hypothesis was that the high dissonance group
would have a stronger belief that they would actually have
to take the examination than would the low dissonance group.
The results of the experiment supported this hypothesis [49:
603-606].

Behavioral Change as a Mode
of Dissonance Reduction

One of the primary objectives for a manufacturer or
retailer for providing dissonance reducing messages would be
to elicit a behavioral change, i.e., to obtain a higher per-
centage of the consumer's future purchases. It is appro-
priate, therefore, to review the literature relating disso-
nance reduction and behavioral changes.

Adams and Rosenbaum used dissonance theory to pre—
dict what would happen to the productivity of workers who

were made to believe they were "overcompensated" relative to

27

their "fair worth" [2:161-164]. Nine subjects were assigned
to each of four categories:
1. Subjects who were hired and informed that they were
pp£_qualified to earn their hourly wage.
2. Subjects who were hired and informed that they wegg
qualified to earn their hourly wage.
3. Subjects who were hired and informed that they were
pp£_qualified to earn their piece rate.
4. Subjects who were hired and informed that they Egg;

qualified to earn their piece rate.

The investigators hypothesized that workers in group
I would experience dissonance because of the disparity
between their "pay" and their "worth" and that this group
would then increase their productivity to reduce dissonance.
The second hypothesis was that workers in group 3 would
similarly be dissonant but that they would decrease their
productivity so as to decrease total earnings in order to
reduce dissonance. The results of the experiment supported
both hypotheses.

Aronson and Carlsmith meshed a theory of eXpecta-
tions with cognitive dissonance theory {11:178-180]. They
theorized that dissonance is aroused when events do not
coincide with expectations and that an expected event is
preferred to other events.' A non-obvious prediction of this
theory is that an undesirable, eXpected event is preferred

over a desirable, unexpected event.

28

The investigators gave a set of 100 cards to the
subjects one card at a time. There were three photographs
on each card and the subjects were asked to attempt to
identify who among the three was a schizophrenic. One group
was told that they were going very poorly as they progressed,
thus causing low expectations of future performance, and one
group was told they were doing very well, thus causing high
expectations of future performance. After the subjects had
looked at their last 20 cards, the "good" subjects were told
that they had done poorly on the last 20 and the "poor" sub—
jects were told that they had done much better. Both groups
were given an opportunity to change their selections on any
of the last 20 cards. The number judgments changed on the
repeated trial was taken as an attempt to reduce dissonance
resulting from not fulfilling expectancies.

The data support the hypothesis that the group whose
performance exceeded their expectations did experience disso-
nance, and did change their answers so as to bring together
performance and eXpectations.

.Effects of Information on
Dissonance Reducpigg

The present research study is concerned in the
implications of providing certain information to the con-
sumer post-transaction. It is hypothesized that this infor-
mation will be effective in assisting the consumer in his

efforts to reduce dissonance. It is, therefore, appropriate

29

to review the literature with respect to the relationship
between information and dissonance reduction. Since selec-
tive exposure to information after the decision is reached
has already been reviewed, this section will deal only with
predecision information.

Jon R. Davidson designed an experiment to examine
the relationship between cognitive familiarity and subse-
quent dissonance reduction [24:45-59]. Subjects were asked
to listen in on a faked interview between a professor and an
imaginary student named "Bill." Afterwards one group was
asked to do a personality rating sheet about "Bill." This
forced the subjects to think about "Bill" and thus estab-
lished high cognitive familiarity. Another group filled out
the same rating sheet, but on themselves instead of "Bill."
This established low cognitive familiarity with "Bill."
Both groups were then asked to indicate how well they liked
"Bill" on another rating sheet. Since "Bill" had been made
to appear rather likable, most of the scores were rather
high.

All subjects were then induced to read a set of
negative statements about "Bill" which, supposedly, "Bill"
could hear. These counter-attitudinal statements were
hypothesized to create dissonance. A post-measure of how
likable "Bill" was was then taken.

The difference between the pre-measure and post-

measure was taken as an indicator of dissonance reduction.

30

To reduce dissonance in this case, subjects would lower

their evaluation of "Bill" to bring their opinions more in

line with their negative statements. The authors conclude:
Clearly, time spent thinking about details of the other
person before the decision (before they even knew there
was going to be a decision) does actually facilitate
post decision dissonance reduction and does, indeed,
seem to be substitutable for time spent after the dis-
sonance already exists [24:54].

Robert J. Holloway also dealt with the issue of pre-
decision information in a study concerning automdbile bat-
tery purchases [29:39-43]. Subjects were asked to play the
role of automobile battery purchasers and asked as a pre-
measure to complete a rating form on each of 12 different
brands of batteries. Each subject was placed in the situa-
tion of having to choose from among three batteries, none of
which he had rated either very high or very low. At the
time of decision, half the subjects were given additional
information about the batteries and half were not. After
the decision a post-measure of the rating of the batteries
was taken.

One of the investigator's hypotheses was:

Individuals to whom additional positive information is
provided to aid their decision-making will experience
less dissonance than those to whom no additional posi-
tive information is provided. In other words, addi-

tional positive information concerning a purchased item

should make a buyer more confident and less dissonant

31

than when he was not given that additional information
[29:41].
Without showing the supportive data, the author con-

cludes:
In the first significant interaction, when high induce-
ment interacted with the two levels of information
[i.e., information and no information], a significantly
different amount of dissonance was measured. Thus,
hypothesis 3 was in part confirmed, as the amount of
positive information did affect dissonance somewhat.

The two studies just reviewed would seem to indicate
that greater familiarity with the product, i.e., greater pre-
decision information, would help the consumer in his efforts
to reduce dissonance. The above does not seem to be entirely
consistent with the following policy recommendation from a
modern marketing textbook:

Since the consumer seeks means to reduce dissonance
after the sale, it may be wise not to "oversell" him on
the product. It is preferable for the product to exceed
consumer eXpectations rather than merely satisfy them or
fall short of them [50:62].

Source Credibility and Dissonance
Reduction

The present study uses what may be assumed to be a
low source credibility agent to provide the post-transaction
messages to the consumers, i.e., the manager of the appliance
department of the retail store at which the appliance was
purchased. It is important, therefore, to examine the rela—

tionship between source credibility and cognitive dissonance.

32

F. A. Powell investigated the attitude change
effects of low and high source credibility agents on six
eXperimental groups composed of Michigan State University
upperclassmen [37]. The experimental groups were exposed to
a persuasive message under conditions of (a) high or low
source credibility, (b) voluntary compliance or noncompli-
ance, (c) high or low discrepancy in the attitude positions
of source and receiver. Using a series of Likert-type atti-
tude items, attitude change was used as an indicator of
dissonance reduction.

One of the hypotheses was that greater attitude
change would accrue from behavioral compliance with the
position of a low credibility source than from compliance
with that of a high credibility source. The hypothesis was
not confirmed by the results of the experiment. The author
attributed the inconclusive results on a failure to achieve
the intended differences in source credibility, i.e., sub-
jects did not perceive the two agents as being different in
source credibility.

Attitude Change and
Cognitive Dissonance

One of the dependent variables in the present study
is "store image" which is hypothesized to consist of the
following attitudinal dimensions:

1. Merchandise Suitability

2. Store Services

33

3. Store Congeniality
4. Sales Personnel

5. Locational Convenience

It is hypothesized that the groups which receive
post-transaction reassurances will have more favorable atti—
tudes toward the store than the group which does not receive
these reassurances. For this reason it is appropriate to
review the literature with respect to attitude change and
cognitive dissonance.

E. E. Smith conducted an experiment using Army
Reservists who, under a guise of a study of survival situa-
tions,were encouraged to eat grasshoppers after filling out
a questionnaire on their liking for grasshoppers as a food
[44:626-639]. Half the subjects were encouraged to eat
grasshoppers by the experimentor in a relaxed, friendly
manner, i.e., positive cognitions. The other half were
ordered to eat the grasshoppers and the eXperimentor was
very gruff, i.e., negative cognitions. After eating the
grasshoppers, the subjects once again filled out a question-
naire on their liking for grasshoppers.

The research hypothesis was that the group exposed
to the negative cognitions would find their behavior to be
more discrepant with their attitude and, hence, would modify
attitude to conform with behavior. The results of the

experiment supported the hypothesis.

34

The previous eXperiment illustrates one of the
ambiguities of using dissonance theory to predict attitude
change. Interpreting the experiment differently, one could
predict just the Opposite results. One could characterize
the situation as being truly a "forced compliance" situation
and since the "negative cognitions" group was actually
ordered to eat grasshoppers, they had no choice, one could
hypothesize that they should experience less, not more dis-
sonance than the group which received positive cognitions.

P. R. Stephenson hypothesized that the purchase of
a major item, an automobile, would be inconsistent with an
unfavorable attitude toward future economic conditions [46].
The investigator used a "before—after with control" design
with economic attitude as the dependent variable as measured
by the questionnaire of the University of Michigan Survey
Research Center.

The results of the study did not confirm the hypoth-
esis that subjects would change their economic attitude in a
favorable direction after the purchase Of an automObile.

Cohen used a "forced compliance" design to determine
whether dissonance could be used to change students atti-
tudes toward "police brutality" [11:74-78]. Soon after a
student "riot" at Yale students were induced to write favor—
able essays On the conduct of the police during the riot by
paying them sums Of money ranging from $0.50 to $10.00. The

mean difference on an attitude scale between the pre-measure

35

and the post-measure was used as an indicator of dissonance
reduction.

The hypothesis was that as the reward for the atti-
tude discrepant essay increased, dissonance would decrease
and, therefore, attitude change would decrease. The results
Of the study confirmed this hypothesis.

Another "forced compliance" experiment was conducted
by Rabbie, Brehm and Cohen. College students who were known
to favor intercollegiate athletics were asked to write essays
in favor of eliminating intercollegiate athletics. One
group of subjects were given minimal rewards for writing the
attitude-discrepant essay, whereas others were given much
greater rewards.

The hypothesis was that those subjects who received
a minimal reward should experience more dissonance and,
hence, should change their attitude toward eliminating inter—
collegiate athletics. The results supported this hypothesis.

L. A. LoSciatO examined the influences Of product
preferences upon dissonance reduction [32]. Subjects were
asked to rate phonograph albums on a 10 point scale on a
criterion of desirability. Subjects were then randomly
assigned to three groups:

1. Group I were told that they could have either their
third or fourth choice [high dissonance].
2. Group II were told that they could have either their

third or eighth choice [low dissonance].

36

3. Group III were told that they were to get their

third choice [control].

Group I [high dissonance] was hypothesized to
attempt to reduce dissonance by revising upward their rating
of the third ranked albums and downward their rating of the
fourth ranked album. Chi-square analyses indicated that
both changes were statistically significant, whereas an
analysis of variance indicated that only the downward change
of the unchosen alternative was significant;

Dissonance theory would predict that a choice among
two or more membership alternatives should product disso-
nance and that the subsequent tension should force a re-
evaluation Of the selected group to a higher level. Richard
J. Dieker attempted to test this derivation [17].

The investigator manipulated the freedom of choice
of belonging to an attractive and a dull discussion group.
It was hypothesized that the more attractive the non-chosen
group, the more dissonance, and, hence, a re-evaluation Of
the chosen group would occur. The results of the study did
not support the hypothesis.

In an unpublished dissertation J. H. Fett examined
the role of volition in attitude change resulting from hear-
ing an attacking speech. It was an attempt to remove some
Of the "coercion" that has accompanied many "free choice"

eXperiments [25].

37

Subjects were pre-measured on 8 contemporary topics
of concern and were asked to indicate their preference for
hearing certain speeches on these topics. Subjects were
then divided into two groups:

1. Group I (the "free choice") were told: "Here is

[a type of a] speech you indicated a strong prefer-

ence to hear."

2. Group II (the "coercion" group) were told: "Because
so many Of you want to hear the same speech, and we

only have a few tapes, you will have to listen to a

speech you indicated you would just as soon not hear."

Although subjects in the free choice situation
changed attitudes more than subjects in the coercion condi-
tion, the difference was not statistically significant.

An eXperiment by Raven and Fishein dealt with changes
in attitude toward extrasensory perception (ESP) [38:188-190].

Subjects were measured on their belief in ESP and
subsequently asked if they were receiving any ESP images.
Upon denial Of the reception of images some of the subjects
were punished by an electric shock and some were not.

Dissonance was hypothesized to occur because of the
incongruity between admitting to not receiving an image and
acceptance of the shock. Subjects could then reduce disso-

nance by increasing their disbelief in ESP. The results of

 

 

the study supported the hypothesis.

38

A study by Brock used as the dependent variable non-
Catholic students' attitudes toward Catholicism [12:264-27l].
Students were given either a high freedom of choice or a low
freedom of choice as to whether to write an essay on: "Why
I would like to become a Catholic."

The results confirmed the hypothesis that attitude
change toward Catholicism increased as freedom of choice to
write the essay increased.

Allyn and Festinger tested the derivation from dis-
sonance theory that pre-exposure knowledge that a communica-
tor will Oppose one's own position will result in reduction
of dissonance by avoidance and defensiveness rather than by
opinion change [3:35-40].

YOung subjects were exposed to a lecture advocating
that the minimum age for driving be raised. A pre-test
showed this to be an attitude discrepant communication.

Half the subjects were foreWarned Of the speaker's bias; the
other half were told that the purpose was to study the per-
sonality of the Speaker.

The results of the study confirmed the hypothesis
that the group not forewarned of the speaker's bias would
exhibit more attitude change than the group that was fore-
warned.

A recent experiment by J. Beijk examined the reduc-
tion of subjective feelings of fatigue due to dissonance

[7:45—54]. Subjects were either given low payment (high

39

dissonance) or high payment (low dissonance) for participat—
ing in the experiment. Subjects were fatigued by exercising
on a bicycle ergometer and then asked their subjective feel-
ings' about how tired they felt.

The low payment group (high dissonance) was hypoth-
esized to attempt to reduce dissonance by decreasing their
feelings Of subjective fatigue. The hypothesis was rejected.
Consumer Characteristics and
Cognitive Dissonance

Of significant importance to marketing is whether
certain types of consumers are more likely to experience dis-
sonance than others. This information is necessary if disso-
nance theory is to be used as a vehicle for market segmenta-
tion. The present study contains hypotheses with regard to
demographic variables, social class variables, shopping
effort and their relationships with cognitive dissonance.

The literature with respect to most of these variables has
been modest, but there has been considerable speculation and
eXperimentation on effort and dissonance.

Aronson tested the derivation from dissonance theory
that the degree of effort expended by a person in attempting
to Obtain a reward has an effect on the relative attractive-
ness of stimuli associated with rewarded versus unrewarded
trials [4:375-380].

Subjects were required to fish for containers of

different colors, some of which had money in them (the

4O

reward). For some of the subjects the task was made very
simple (low effort) and for others very difficult (high
effort). Subjects were asked to rate the relative attrac-
tiveness of the two colors before and after the experiment.

The hypothesis was that the high effort group should
experience more dissonance which should partially reverse
the effect of the reward and bring the relative attractive—
ness of the two colors chosen together. The results showed
that the low effort group increased whereas, the high effort
group showed no change in the relative attractiveness of the
two colors.

A study by Zimbardo also examined the relationship
between effort and dissonance [11:30-31]. Subjects who were
in favor of the adoption of a numerical grading system at
New York University were asked to read a counter-attitudinal
report on numerical grading systems. Effort was manipulated
by having the subjects read the report aloud under differing
conditions of delayed auditory feedback. High effort sub—
jects had a feedback Of .3 seconds delay while low effort
subjects had a .01 seconds delay.

The hypothesis was that high effort subjects would
experience more dissonance than low effort subjects and
would reduce this dissonance by changing their attitude in
a more negative direction. The results of the study sup-

ported the hypothesis.

41

An unpublished Ph.D. dissertation by Richard Cardozo
also dealt with consumer effort in addition to consumer
product expectations [13].

Some of the subjects were asked to simulate a shop—
ping experience by going through a catalogue consisting of
descriptions Of various ball-point pens and to write down
one feature which impressed them for half the items shown
(low effort). The task consumed about 15 minutes. Other
subjects were asked to go through a similar catalogue and
record five different features for each Of the 31 items
(high effort). The task consumed about one hour. Each sub-
ject then received a modest ball-point pen as a reward.

One hypothesis was that the high effort group would
eXperience considerable dissonance because the reward would
not be commensurate with the effort expended. To reduce
dissonance, the subjects would evaluate the pen more highly
than those subjects who expended little effort. This hypoth—
esis was confirmed.

Gerald-D. Bell explored the relationships among a
consumer's confidence, ease of persuasibility, and cognitive
dissonance [8]. The investigator interviewed 234 persons
who had recently purchased an automObile. A measure of the
amount of dissonance the consumer was experiencing was
obtained by asking questions which would reveal the respon-
dent's attitudes and feelings about his purchase, i.e., how

much doubt the consumer was experiencing. Generalized

42

self-confidence was measured using the items originated by
Day and Hamblin [15:499—511]. Persuasibility was measured
by first asking respondents in separate questions to what
extent they were influenced by the salesman on price, pay—
ments, color, etc. and then combining the score on these
same characteristics.

The findings showed no significant relationship
between persuasibility and dissonance until the investigator
controlled for self-confidence. For those who were high on
self-confidence, if they were easily persuaded in purchasing
their new car, they were high on dissonance. Conversely,
those who were low on self—confidence and were easily per—
suaded had little dissonance. The author concludes that the
automobile industry might well increase the amount of adver-
tising directed toward the recent purchaser, especially
those who were easily persuaded.

Lance Canon developed an experiment which also con—
cerned self—confidence and dissonance, although in this case
the amount of self—confidence was experimentally manipulated
[24;83‘95]. Subjects were assigned several problems to
solve. As the subjects were solving the problems, half were
told that their solutions were correct and that most of
their colleagues' answers were not correct (thus building
the subject's self-confidence). The other half were told

that they had given incorrect solutions, but the majority of

43

their colleagues' answers were correct (thus building low
self-confidence).
At the end of the last problem, half the subjects

were told that they would subsequently be forced to rebut a
series Of written probing questions as to why their answers
were correct. The other half were told that they simply had
to write a short essay supporting their positions. Each sub-
ject was then given the Opportunity to rate his preference
for reading a list Of articles, some of which were dissonant
with the subject's position and some of which were consonant.
It was thought that the dissonant articles would be more use-
ful to the group which had to rebut probing questions than
to the group which simply had to write an essay. There were,
then, four groups:

1. High confidence—highly useful

2. High confidence-less useful

3. Low confidence-highly useful

4. Low confidence-less useful.

The two major hypotheses were:

1. The high confidence-highly useful group would prefer
to read the dissonant articles so as to prepare
arguments to rebut the probing questions.

2. The low confidence-less useful group would strongly

prefer consonant articles.

44

The results of the experiment supported the above
major hypotheses.

A consideration Of some of the features of the Open-
closed minded position of Rokeach [39] suggested to Samuel
Fillenbaum that there might be a relationship between dogma-
tism and a person's ability to tolerate dissonance [26:47-50].

The experimental design was essentially the same as
that used in an earlier study by Aronson and Mills [5:177-
181]. In that eXperiment persons who were forced to undergo
an unpleasant initiation in order to gain a group liked the
group more than people who were Spared the initiation.
Fillenbaum modified the experiment by first administering
the Dogmatism scale (form E consisting of 40 items) to all
subjects.

The hypothesis was that there would be a positive
correlation between dogmatism and dissonance, i.e., the very
dogmatic subjects would, therefore, show a higher liking for
the group than the low dogmatic subjects. The results Of
the experiment supported this hypothesis.

Cognitive Dissonance EXperimentation:
A Commentary

The foregoing literature review is not purported to
be an exhaustive review of cognitive dissonance experimenta-
tion. It is, however, designed to be a balanced presentation
of the evidence from prior research concerning the Specific

hypotheses to be examined in the present study. Some of the

45

evidence supported the present research hypotheses, some of
it did not, and much of it was inconclusive.

The ambiguity and inconclusiveness as well as the
sheer volume of experimentation on cognitive dissonance has
prompted several psychologists (notably S. E. Asch, and
Natalia and Alphonse Chapanis [14]) to critically review the
research on dissonance theory. The criticisms of Chapanis
and Chapanis roughly fall into two classes [14]:

l. [Dissonance] experimental manipulations are usually
so complex and the crucial variables so confounded
that no valid conclusions can be drawn from the data.

2. A number of fundamental methodological inadequacies
in the analysis of the results . . . vitiate the

findings.

Chapanis and Chapanis point out that dissonance is
an intervening variable whose existence is inferred from
measuring observable dependent variables. To test dissonance
theory it is necessary to create various degrees of disso-
nance by introducing various discrepant cognitions within
the subject. The dissonance researchers rarely use disso-
nance theory to predict highly structured simple events, but,
rather, to predict complex social events. In order to do
this experimentally, they have constructed extremely compli-
cated situations with elaborate instructions to the subject

to generate dissonance. Under these situations, therefore,

46

how can one be certain that dissonance and dissonance alone

was created?

After discussing five typical cases where the vari-

ables were confounded Chapanis and Chapanis make the follow-

ing recommendations [14:11].

1.

It is possible to design eXperiments so that [the
variables] are not confounded. . . . We recommend
first Of all that the eXperimental manipulations
be simplified.

Our second recommendation is that additional control
groups be included in the design.

Our third recommendation is that a little more
attention be given to discovering the possible
cognitions that a S might have about the situation,
particularly those which might be contrary to dis-
sonance theory.

Chapanis and Chapanis review further the treatment

of the data in dissonance experiments and point out three

major flaws:

1.

2.

Rejection of cases. This was their most important

criticism, not only because it was so basic a flaw,
but also because supporters of dissonance theory do
it so often. In a total of 7 eXperiments the per-

centage of subjects used and subseguently discarded
ranged from 10% to 82%. It is difficult to assess

the meaning of a statistical test on data when many
of the subjects have been systematically discarded

after data collection and analysis.

inadeguate design. Most Of the experiments are some
variant of a 2X2 factorial with unequal, nonpropor-
tional, and generally small N's in each cell. These
restrictions make it impossible for the authors to
carry out ordinary analyses of variance. Instead we
find them using a hodgepodge of "t" tests and a
statistic which they refer to as an "interaction t."
. . . The principal difficulty is that in making
such multiple comparisons [without an overall test
of significance] the experimenter is allowing him-
self a number of opportunities to find an event
Esignificance) which normally occurs infrequently
14:18].

47

3. Straining for significance. . . . Authors tend to
present results as significant and as supporting the
dissonance theory prediction when the probabilities
are greater than the usually accepted value of .05.
. . . It is extremely disconcerting to find these
statistically nonsignificant trends quoted authori-
tatively in subsequent reports and later reviews as
substantiating the theory, without any qualifying
statements.

In conclusion, the author must agree that the more
recent eXperimentation in dissonance theory is entirely con-
sistent with the closing comment of Chapanis and Chapanis,
". . . All of the considerations detailed above lead us to

concur with Asch's evaluation Of the evidence for dissonance

theory and return once more a verdict of not proven [14:21]."

CHAPTER III

RESEARCH DESIGN

Basic Design

The basic research design of the thesis may be clas-
sified as an "eXperimental design--after only, with control
group" [28:111]. It is an experimental design because there
was manipulation of one assumed causal variable, and there
was a random assignment of subjects to experimental and
control groups. It is "after only" because measurements
were not taken prior to the "treatment." The design may

be represented in symbols as:

Experimental Groups

 

1. 2_. 2.
Sample Select1on R1 R2 R3
Experimental Variable X2 X3
After Measurement 01 O2 03

The notation used above is as follows:

R indicates that individuals were selected at random
to receive various treatments.

X represents the exposure of a particular group to
a particular treatment.

0 refers to the measurement taken.

48

49

Therefore, the second column of the above table
would be read, "Group 2, selected at random from the popula-

tion, was eXposed to treatment X and then measurement 0

2’ 2
was taken." The control group (R1) received no treatment.

Although it was not feasible to Obtain a specific
pre-treatment measure, use is made of a prior study which
used a similar measurement tool to determine how different
socio-economic groups perceived department store "images."
This is the closest thing to a pre-test which is practicable
[48].

The lack of a specific pre-test may not be as unfor-

tunate as some believe:

The pre-test is a concept deeply imbedded in the think-
ing of research workers but it is not actually essential
to true eXperimental designs. Almost all the agricul-
tural eXperiments are run without pre-tests. In the
social sciences, however, it seems difficult to give up
the notion that the experiment and control groups might
have been unequal before differential eXperimental
treatment and rely upon randomization to reassure us
that there will be a lack of initial biases between
groups. Perhaps this belief comes from the fact that
we have not run man experiments in the social sciences
and marketing [6:35 . A

Sample Selection Procedure

A major department store in the Detroit metropolitan
area, The J. L. Hudson CO., was asked to participate in the
study. After consultation with the department store, it was

decided to use recent purchasers of refrigerators as subjects

for the following reasons:

50

l. Refrigerators are one of the most expensive items
that department stores sell. Therefore, if price is
a manifestation of "importance," then a refrigerator
purchase decision would be an "important" decision.

2. The department store carried several "brand name"
lines of refrigerators. Therefore, not only will
the customer have to decide in which store to make
the purchase, but also, which brand in which store.
This increases the possibility Of there occurring
dissonant cognitions.

3. There are a bewildering array of special features on
refrigerators, e.g., styles, colors, special com-
partments, revolving shelves, automatic defrosting,
rollers, etc. The consumer in many cases will be
forced to accept a unit which will not have every
specific feature she desires, which will introduce
dissonant cognitions.

4. In the Opinion Of applicance department personnel,
refrigerator purchasers experienced more post-trans-
action anxiety than most purchasers of comparably

priced department store items.

The department store made available a list Of the
names and addresses Of all purchasers of refrigerators. The
list was made available in general on the day following the
transaction except for those who did not have credit estab-

lished. The names of those who did not have credit

51

established were provided on the day immediately following
credit approval.

From the above described lists the following types
of customers were excluded from the population:

1. Corporate, professional, and other business custom-
ers were excluded.

2. Customers who lived outside Detroit or the immediate
suburbs were excluded. (See Appendis F for a list—
ing Of the suburbs included in the sample.)

3. All employees of the department store were excluded.

4. Purchasers of "built in" appliances were excluded.

From this revised list, customers were placed into

three groups on a random basis.

Treatment Procedure

Each Of the three groups received a different post-
transaction message (treatment).

Group I.-—Group I was a control and received no post-
transaction message.

Group II.--Group II received a personal letter with
the following characteristics (see Appendix E for the actual
message):

1. Department store letterhead stationery.
2. Signed by the appliance department manager. It was
hypothesized that the above would increase the

source credibility of the message.

52

3. A sentence expressing appreciation for the custom-
er's purchase.

4. Several sentences reassuring the customer on the
following points:

a. Her selection of the department store was

"correct."
b. Her selection of the particular brand—model was
”correct."

c. The department store "stands behind" the merchan-

dise it sells.

Group III.--Group III received a telephone call from
the researcher, who represented himself as an employee Of the
department store. The message that was used in the letter
sent to group II was delivered orally to group III with a

few modifications (see Appendix F).

Interview Procedure

All interviewing was conducted by a team of profes—
sional interviewers hired from the Market Opinion Research
CO. of Detroit, Michigan. Personal interviews were used
because it was believed that the length and complexity of
the questionnaire required interviewer supervision. Also,
a mailed questionnaire would have allowed the respondent
too much time to deliberate on the questions, the returns
would have been low, and control of the "time frame" would

have been lost. Since dissonance was expected to decrease

53

through time, it was important to keep the time from deci-
sion to interview as constant as possible.

An attempt was made to interview each subject no
sooner than three days after the subject's name was received.
The three day delay was necessary so that the subjects in
groups II and III could receive the messages.

Interviewers were instructed to telephone subjects
in advance to arrange appointments preferably on the third
day after the subject's name was received. If an appoint-
ment could not be Obtained on that particular day, then the
interviewer was instructed to attempt to Obtain an appoint-
ment on either Of the two immediately following days. If an
appointment could not be made on any Of the three above
designated days, then the subject was dropped from the list.

Telephone calls were made at various times during
the day so as to avoid a "working wife" bias.

In an attempt to increase the percentage completion
rate, interviewers were permitted to cluster subjects and
make some "cold" calls. However, any subjects who were not
at home on the "cold" calls were telephoned for an appoint-
ment.

The above procedures were designed to attempt to
minimize and hold constant the time from transaction to
interview. Since cognitive dissonance is eXpected to
decrease through time, it is desirable to hold the time from

transaction to measurement as uniform as possible.

54

Interviewers were instructed to interview the female
head of the household. If the household consisted of a
single male, then he was interviewed.

The above procedure was continued until a total

sample size of 150 completed interviews were Obtained.

Theiguestionnaire

The basic measurement tool selected for the research
was a modified form of semantic differential. The semantic
differential is a tool which has been widely used to measure
the meaning of words since its introduction by Osgood, Suci,
and Tennenbaum [36].

Since its introduction the semantic differential has
been used in at least the following primary applications:

1. Comparing corporate images, both among suppliers of
particular product classes and against an "ideal" or
standard image of what respondents think a company
should be [48].

2. Comparing brands and services of competing suppliers.

3. Determining the attitudinal characteristics of pur-
chasers of particular product classes or brands
within a product group.

4. Analyzing the effectiveness Of advertising and other
promotional stimuli on attitude change [28:202].

The original semantic differential scale requires
respondents to make repeated judgments of how they perceive

a particular concept along a seven point interval-ordinal

scale between two bipolar adjectives. For example:

Concept Z

Cold: : : : : : : :Hot
Very Quite Slightly Neutral Slightly Quite Very

55

An "X" in the apprOpriate interval then indicates both the
direction and the intensity of the respondent's feeling
toward the concept.

Several important modifications of the original
semantic differential scale have been proposed to make it
more useful in marketing research [35:28-29]. Mindak has
suggested the use of descriptive nouns and phrases (in addi-
tion to bipolar adjectives) to make the tool more applicable
to marketing problems.

The research employed phrases and sometimes complete
sentences instead of simply bipolar adjectives.

The semantic differential has the following advan-
tageS:

1. It is a quick, efficient means of getting in readily
quantifiable form for large samples not only the
direction but intensity of opinions and attitudes
toward a concept. . . .

2. It provides a comprehensive picture Of the "image"
or meaning Of a product or personality.

3. It represents a standardized technique for getting
at the multitude of factors which go to make up a
brand or product "image."

4. It is easily repeatable and quite reliable.

5. It avoids stereotyped responses and allows for
individual frames of reference.

6. It eliminates some of the problems of question
phrasing . . . [35:28-29].

The Questionnaire, which is reproduced in Appendix C, is
divided into five parts: I. Perceived Dissonance; II. Store

Image; III. Predisposition to Purchase; IV. Shopping Effort

Expended; V. Demographic and Other Data.

56

Part I: Perceived Dissonance

There are at least three methods by which one might
attempt to measure the level of cognitive dissonance a sub—
ject might be experiencing. One possibility would be to
attempt to measure certain physiological correlates of dis-
sonance [15:392-399]. However, submitting the subject to a
polygraph type of apparatus in his home was not a viable
alternative for the research.

A second possibility would use cognitive dissonance
theory prOper as a guide. Dissonance theory states: "The
total amount of dissonance that exists between two clusters
of cognitive elements is a function Of the weighted propor—
tion of all relevant relations between the two clusters that
are dissonant. The term "weighted proportion" is used
because each relevant relation would be weighted according
to the importance of the elements involved in that relation"
[23:18].

It might, then, be possible to:

1. Determine all the relevant cognitions for Eggm_
consumer.

2. Determine which ones are dissonant and which are
consonant.

3. Determine the "relative importance" of each cognition.

4. Use the weighted average of the proportion of disso-
nant cognitions to consonant cognitions as an indi-

cator Of the level of cognitive dissonance.

57

Unfortunately, steps (1) and (3) pose problems with
no ready resolution and, hence, this method was not adOpted.

A third method of deriving an indicator of the level
of cognitive dissonance, which was the method selected for
use, is to ask the subject questions concerning how much
doubt the respondent has about his decision. A similar
method was used by Gerald Bell in a study of automObile
purchasers [8:14].

Essentially, the subject could be eXperiencing dis—
sonance because of cognitions which were inconsistent with
two decisions:

1. The selection of the store (store dissonance).

2. The selection Of the brand (brand dissonance).

Questions 1 and 3 under Part I are indicators Of
store selection dissonance, and questions 2 and 4 are indi-
cators of brand dissonance. Questions 5 and 6 have elements
of BREE store and brand dissonance.

The seven point scale was numbered for each question
in such a manner that the higher the number, the higher the
dissonance. The meangperceived dissonance was then defined

as the mean score of the subject's answers to questions 1-6

in Part I.

IPart II: Store Image

One Of the hypotheses is that as a result Of the

£>Ost-transaction communication by the retailer, the subject

58

would view the store more favorably. In other words, those
subjects who were assisted by the retailer in their efforts
to reduce cognitive dissonance will have more favorable
attitudes toward the retailer than those who were not
assisted by the retailer.
I Part II of the questionnaire breaks store image into

five attitudinal dimensions:

1. Merchandise Suitability

2. Store Services

3. Store Congeniality

4. Sales Personnel

5. Locational Convenience.

Dimensions 1, 3, 4, and 5 are the attitudes used by
Robert G. Wyckham in his study Of aggregate department store
images [48:84]. They were adapted from a previous article
by George Fisk [27:5]. The dimension of Store Services was
added for purposed of the present research, and the original
questionnaire utilized by Wyckham to measure the attitudes
was modified to better suit the needs Of the study.

For each question under the five attitudinal dimen-
sions a "1" was a "most favorable" response as viewed by

the store.

Part III. Predisposition to Buy
Part III contains a question to attempt to determine

how the customer perceived his future shopping behavior and

59

to test the hypothesis that the groups which received post-
transaction communications would be more predisposed to buy

at that store than the control group.

Part IV. Shopping Effort

Part IV contains questions to attempt to determine
whether cognitive dissonance decreased as shopping effort
increased, as has been reported [13]. The following method
was used to translate the answers to high, medium, and low

shopping efforts:

 

 

 

High Effort Medium Effort Low Effort
(IV-1—b)+(IV-2-c) (IV-l-a)+(IV-2-c) (IV-l-a)+(IV-l-a)
(IV-1—c)+(IV-2-a) (IV-l-b)+(IV-2-a) (IV-l-a)+(IV-l-b)
(IV-l—c)+(IV-2-b) (IV-l—b)+(IV-2—b)

(IV—l—c)+(IV-2—c)

The first line would read: "If the respondent
checked ”b" to question IV-l, and "c" tO question IV-2, Ehgm
he would be classed as high shopping effort, i.e., he visited
two stores and compared three or more brands."

Questions 1, 7, and 8 concerning age, marital status,
and presence of children were used to determine the stage of

the family life cycle. The following stages were used:

Young, single

Young, married, no children
Young, married, with children
Older, married, with children
Older, married, no children
Older, single.

OLD-photo!“

60

By "young" is meant that the respondent is under 40
years of age. A person over 40 years of age is classified
as "Older.“ By single is meant that the reSpondent is not
married, widowed, divorced, or separated. By "children"
is meant children living at home or away at school who are
supported by the respondent.

Questions 9, 10, ll, 12, 13, 14, and 15 were used to

estimate the social class of the respondent.

Part V: Demographic Data
Part V contains demographic data of seven types:
1. Age
2. Education
3. Family income
4. Race
5. Brand experience
6. State in the family life cycle

7. Social class

The data were used to determine if there were sig-
nificant demographic correlates with high, medium, and low
dissonance. Also, the brand experience data were used to
determine whether people who have had extensive experience
with a brand would have significantly lower perceived disso-

nance.

61

Data Collection and Analysis

Each questionnaire was scored by hand and the data

transferred to computer punch cards for the following vari-

ables:
1. Mean
2. Mean
3. Mean
4. Mean
5. Mean

6. Mean

perceived dissonance
merchandise suitability
store services
congeniality

sales personnel

locational convenience

7. Predisposition to buy

8. Shopping effort category

9. Age category

10. Educational category

11. Income category

12. Race

category

13. Brand experience category

14. Stage in the family life cycle

15. Social class.

A computer program was then employed to determine

the Mann4Whitney U statistic and a KruskalAWallis one-way

analysis Of variance. The Mann4Whitney U statistic is a

non-parametric test to determine the probability that two

independent samples, which may be of unequal size, have come

62

from identical populations. The procedure for the calcula-
tion of the statistic is described in Appendix G.

The KruskalSWallis one-way analysis of variance by
ranks is a test for determining whether "k" independent
samples are from different populations. The procedure for

determining the KruskalAWallis H is described in Appendix G.

CHAPTER IV

THE FINDINGS

Introduction

The findings Of the experiment on the effectiveness
Of post-transaction communications in reducing cognitive dis-
sonance are presented in Chapter IV.

The chapter begins with some of the general results
of the interviewing, including the number Of interviews
attempted, the percentage Of interviews completed, the
length Of time for the typical interview, and the average
length of time from transaction to interview.

The second part Of the chapter compares the attitu-
dinal characteristics of the population under study with the
attitudinal characteristics of the much larger sample of
the general public Obtained by R. G. Wyckham [48].

The rest of the findings in Chapter IV follow the
organization of the research hypotheses (described in detail
in Appendix B):

l. The effectiveness Of post transaction communications
in reducing cognitive dissonance.

2. Attitude change and dissonance reduction.

63

64

3. Predisposition to purchase and dissonance reduction

4. Consumer characteristics and cognitive dissonance

General Results

A total Of 152 interviews were Obtained between
June 21, 1968 and July 16, 1968. Since the original sample
consisted Of 335 subjects, the completion rate was 45 per—
cent. Of those who were not interviewed, 48 percent were
"not at home." A large variety of other reasons accounted
for the other 52 percent, including: I'tOO busy," "don't
want to be bothered," "illness in the family," and a general
suspicion of all interviewers.

Two additional interviews were Obtained but had to
be discarded. In one case the subject had cancelled her
order and had replaced it with a refrigerator from another
department store. The responses were directed toward the
other department store and, therefore, the subject was
dropped. The second interview which was discarded was Of
an employee of the department store.

Interviewing time averaged about 35 minutes. Inter-
viewers indicated that subjects seemed to understand the
questions well and were very willing to talk about their
purchases.

The average time from transaction to interview was

as follows:

65

1. Control Group = 7.6 days
2. "Letter Group"l = 7.1 days

3. "Telephone Group" = 7.4 days.

The Population

As discussed in the research design chapter, the
design of the study was an "after onlydwith control group."
Since no pretest was possible, it is appropriate to investi-
gate whether the attitudes toward the department store of
the sample were similar to the attitudes of the general
public as found in the much larger survey design of R. G.
Wyckham [48]. Both studies used the same department store.

Table 1 shows the overall means Of the three atti—
tudes which the two studies had in common. The numbers
represent the arithmetic average of the mean scores for each
person for each attitude in each group. The "overall means"
are presented for descriptive purposes only and no statisti-
cal tests will be done on them. All statistical tests will
be conducted on the underlying data in accordance with the
procedures as outlined in Appendix G.

Since the scales were constructed such that higher
numbers reflect more unfavorable attitudes than lower num-

bers, Table 1 would indicate that the overall population had

lHenceforth in the report, the group which received
the letter message after the transaction will be referred to
as the "letter group" and the group which received the post-
transaction telephone call will be referred to as the "tele-

Phone group . "

66

 

HH.N mm.H hm.H mm.m

coflumooq muoum
Gmmz HHmHm>O

 

mm.m om.a wa.m mw.m amccomumm moamm
cam: Hamum>o
om.m oo.~ mo.m n¢.~ suuaanmum mmuncmaoumz
and: Hamum>o
mdouo OOOSQOHOB msouo Hmuumq QOOHO Honucoo muaommm mopsufluu<

m .228»: . o . m

 

mooum ucommum mo muasmom

 

 

”meg mooum m.Emnxo>3 .O .m cam hooum ucmmoum
on» somzuon OOEEOO CH moosufluum mounu wsu mo momma Hamno>o may mo conflumgfiou .H magma

67

less favorable attitudes toward the department store than
the present sample under study. The finding is reasonable,
since one would eXpect that a department store's own custom-
ers would have more favorable attitudes toward the store
than would a sample of the general population.

Table 2 uses the basic data to test the null hypoth-
esis that the various groups come from the same population,
i.e., does the sample under study have significantly more
favorable attitudes toward the department store than a sam-
ple drawn from the general public? In each case the null
hypothesis that the two samples came from the same population
is rejected at a significance level substantially less than
the .01 level. It would seem clear that, at least for the
three attitudinal dimensions, the present sample is not
representative of the general public in the Detroit metro-
politan area.

The Effectiveness Of the Post-
Transaction Communications in
Reducing Cognitive Dissonance

The major guiding hypothesis of the study was that
post—transaction reassurances from a retailer to recent
purchasers would be effective in assisting the consumer in
reducing dissonance (see Appendix B for the specific research
hyptotheses). The hypothesis was that the mean perceived
dissonance scores, operationally defined as the mean of the
responses to questions Al through A6 of Appendix C, for the

group which received the telephone call would be lower, less

68

.mwmmsuomwn 005: mouu mo mm0u00030300m

Q

.om cmnu Hmummum NC £003 D mmcuficzwccmz mo msam>m

 

 

00. 0000.0 00000 00000 000000009 000000 0000000 00000000
mHOum

00. 0000.0 Nvooa @5000 Hmuumq OHHQom Hmnmcmw cowumooq
muoum

00. 0000.0 00000 00000 0000000 000000 0000000 00000000
mnoum

HO. NOOO .0 Manama QDOHO OCOEQTHTB Oflafldm HMHOCOO HOCCOmHmm
moamm

00. 0000.0 0000 00000 000000 000000 0000000 000000000
moamm

00. 0000.0 00000 n00000 0000000 000000 0000000 000000000
mmamm

00. 0000.0 0000 00000 000000000 000000 0000000 00000000000
mmflocmnonmz

00. 0000.0 0000 00000 000000 000000 0000000 00000000000
mmHOCmSOHOE

00. 0000.0 00000 00000 0000000 000000 0000000 00000000000
mmHchnoumz

OOCm00w0cmHm Am 0D @5000 @5000 mosuwuu<

mo Hm>mq

OHQmHO>mm muoz

manmuo>mm mmmq

 

 

1000mm: 00:: may 0003 mm0u00000900m Omumfloommm pow mmcu0£31ccmz mo xumfifiom

mamEmm uOOmwum
mnu mC0momEOo mmsoum may mo moosuwuum 0:0 mm mEmm mfiu mum Hm0a Emnxohz
.O .m an cmxmu O0HQOQ Hmuwcmm 000 mo mHmEmm mnu mo mopsu0uum 000 0000 00mm

.N OHQmB

69

dissonant, than the scores of the group which received the
letter, which would in turn be lower than the control group.

Table 3 shows the overall mean perceived dissonance

for each group.

Table 3. Overall mean perceived dissonance by group

 

 

Overall Mean

 

Group N Perceived Dissonance
Control Group 66 2.11
Letter Group 43 1.90
Telephone Group 43 2.31

 

Although no statistical analysis will be performed
on the overall group means, the means do serve as useful
descriptive statistics to demonstrate the trends of the data.

Contrary to eXpectations, the overall group means
would indicate that the group which received reassurances
via the telephone was experiencing greater dissonance than
either the control group or the group which received the
letter. However, the letter group did have a lower overall
mean perceived dissonance score than did the control group,
which is in accordance to eXpectations.

To test the overall significance Of the trends, the

underlying data were subjected to a KruskalAWallis one way

7O

analysis of variance by ranks (see Appendix G for statis—
tical procedure explanation). The results Of the test are

summarized in Table 4.

Table 4. Summary Of Kruskal4Wallis H and associated prOba-
bilities with the null hypothesis that the per-
ceived dissonance scores Of the three groups came
from the same population

 

Kruskal4Wallis Ha Pb Level of Significance

 

5.3513 .0689 NS

 

aSee Appendix G for statistical procedures.

bProbability of a true null hypothesis.

Since the test statistic does not quite meet the
customary .05 level of significance, the above findings as
to the relationships Of the dissonance among groups must be
taken to be tentative.

Further analysis of the basic perceived dissonance
data was conducted by examining the relationship between
each group and every other group, two at a time, using a
Mann Whitney U test statistic (see Appendix G for statis-
tical procedures). Table 5 summarizes the findings of the
analysis.

Table 5 supports the results of the earlier descrip-
tive statistics in that the dissonance of the telephone

group was greater than the dissonance of the control group,

71

Table 5. Summary of Mann4Whitney U and associated proba-
bilities with the null hypotheses that each pair
of groups comes from the same population

 

 

 

More Dissonant Less Dissonant a b Level of
Group Group U P Significance
Control Group Letter Group 1236.5 0.1279 NS
Telephone Group Control Group 1185.5 0.0733 NS
Telephone Group Letter Group 663.5 0.0119 .05

 

aValue of Mann0Whitney U with n greater than 20.

2
bProbability Of a true null hypothesis.

which, in turn, was greater than the dissonance of the
letter group. However, only the differences between the
telephone group and the letter group reached the .05 level
Of significance.

Attitude Change and
Dissonance Reduction

 

 

A second guiding hypothesis concerns one of the ben-
efits posited to accrue to retailers for providing dissonance
reducing information. It was hypothesized that the groups
which received assistance in reducing dissonance would sub-
sequently View the store more favorably than the group which
did not receive assistance (see Appendix B for the specific
research hypotheses).

Table 6 shows the overall group means for each of
the attitudes Operationally defined to comprise the "store

image":

72

1. Overall mean merchandise suitability
2. Overall mean store services

3. Overall mean sales personnel

4. Overall mean store congeniality

5. Overall mean store location.

Table 6. Overall mean for each attitude for each group

 

 

Overall

Grou’? N Group Mean

 

Merchandise Suitability:

Control 66 2.03
Letter 43 2.00
Telephone 43 2.26

Store Services:

Control 66 1.88
Letter 43 1.94
Telephone 43 2.19

Sales Personnel:

Control 66 2.14
Letter 43 1.90
Telephone 43 2.35

Store Congeniality:

Control 66 1.95
Letter 43 1.85
Telephone 43 2.40

Store Location:

Control 66 1.97
Letter 43 1.93
Telephone 43 2.11

 

aThe arithmetic average Of the mean scores of all
subjects for each attitude in each group.

73

The overall means serve only as descriptive statis-
tics, with all tests Of significance conducted on the under-
lying data.

In general the results paralleled the dissonance
findings. The telephone group, contrary to expectations,
had the most unfavorable position for each of the five atti—
tudinal dimensions comprising "store image." The scales
were constructed so that high numbers are more unfavorable
than low numbers. The group which received the letter did
react according to eXpectations and had the most favorable
position for every attitude with the exception Of store ser—
vices. For the attitude of store services the tendency was
for the control group to be slightly more favorable than the
group which received the letter.

Table 7 is a summary of the Krushal4Wallis one way
analysis of variance on the underlying data to determine
whether the overall differences among the groups for each
attitude were significant. Since the results show that only
the differences among the groups for the attitude of store
congeniality were significant at the .05 level, the above
findings that the telephone group had the most unfavorable
attitudes, and that the letter group had the most favorable
attitudes, must be considered as tentative.

Table 7 also shows that the least significant of all
differences was for the attitude Of store location. Since

store location is probably the most "objective" of all the

74

Table 7. Summary of KruskalAWallis H and associated proba-
bilities with the null hypothesis that for each
attitude the scores Of the three groups came from
the same population

 

 

Kruskal-Wallis Level of

 

Attitude Ha Pb Significance
Merchandise

Suitability 2.7712 0.2502 NS
Store Services 5.3079 0.0704 NS
Sales Personnel 4.4020 0.1106 NS
Store Congeniality 8.6811 0.0130 .05
Store Location 0.9597 0.6189 NS

 

aSee Appendix G for statistical procedures.

bProbability of a true null hypothesis.

attitudes, it would, therefore, be expected to be the most
resistant to change.

Further analysis of the data on attitudes was con-
ducted by examining differences between groups, taken two at
a time, using the MannAWhitney U test statistic. Table 8
summarizes the results of the tests.

The data support the overall means in that the tele-
phone group consistently had less favorable attitudes toward
the store than the other two groups, and that the letter
<group generally had more favorable attitudes toward the store
than either the control or telephone group. Only four of the

fifteen possible tests are significant at the .05 level.

75

Table 8. Summary of Mann4Whitney U and associated probabil-
ities with the null hypothesis that the differences
in attitude scores between pairs of groups come from
the same pOpulation

 

fi i r
—1 — T

Less Favorable More Favorable Level of
a b . . .
Group Group U P Sign1f1cance

 

Merchandise Suitability:

Control Letter 1402.0 0.4579 NS
Telephone Control 1184.5 0.0725 NS
Telephone Letter 754.0 0.0700 NS

Store Services:

Letter Control 1291.5 0.2133 NS
Telephone Control 1065.0 0.0138 .05
Telephone Letter 745.0 0.0597 NS

Sales Personnel

 

Control Letter 1226.5 0.1153 NS
Telephone Control 1232.5 0.1231 NS
Telephone Letter 687.0 0.0197 .05

Store Congeniality:

Control Letter 1347.5 0.3279 NS
Telephone Control 1016.0 0.0060 .01
Telephone Letter 618.5 0.0040 .01

Store Location:

Control Letter 1376.0 0.3943 NS
Telephone Control 1294.5 0.2192 NS
Telephone Letter 819.0 0.1800 NS

 

aValue of Mann4Whitney U with n greater than 20.

2
bProbability Of a true null hypothesis.

76

Predisposition to Purchase
and Dissonance Reduction

A second benefit posited to accrue to the retailer
for providing dissonance reducing information was that the
consumers who received assistance would be more predisposed
to purchase from the retailer than those wo did not receive
assistance (see Appendix B for the specific research hypoth-
eses). Table 9 presents the mean predisposition to purchase

for each group.

Table 9. Mean predisposition to purchase for each group

 

 

 

Predisposition
Group N to Purchasea
Control Group 66 2.92
Letter Group 43 2.70
Telephone Group 43 3.30

 

aThe arithmetic average of the scores Of all sub—
jects in each group.

Low scores are more favorable than high schores,
i.e., the subjects indicated that they intended to do a
higher percentage of their shopping at that store than they
did in the past.

The descriptive statistics would indicate that in
general the letter group had higher expectations of future

purchasing than had the control group. Also, the control

77

group had higher expectations of future purchases than did
the group which received the telephone call. The latter
result is not consistent with the research hypothesis but
is consistent with the findings of the dissonance section.

A Kruska14Wallis one way analysis Of variance by
ranks was conducted on the underlying data and the results
of the test for overall significance are summarized in
Table 10. The results indicate that the overall differences

among the three groups are not significant.

Table 10. Summary of Kruskal4Wallis H and associated
probabilities with the null hypothesis that the
predisposition to purchase scores Of the three
groups came from the same population

 

 

KruskalJWallis Ha Pb Level Of Significance

 

3.9676 0.1375 NS

 

aSee Appendix G for statistical procedures.

bProbability of a true null hypothesis.

Mann4Whitney U statistics were then computed for the
three groups taken two at a time. Table 11 summarizes the
results of the tests.

The findings corroborate the overall means in that
the telephone group did have the lowest expectations of

future purchases, whereas the letter group had the highest.

78

Table 11. Summary Of Mann4Whitney U and associated probabil-
ities with the null hypothesis that the differ-
ences in predisposition to purchase between pairs
Of groups come from the same population

 

 

Less Favorable More Favorable a b Level of
Group Group U P Significance
Control Letter 1323.5 0.2624 NS
Telephone Control 1197.5 0.0678 NS
Telephone Letter 719.5 0.0268 .05

 

3Value of Mann Whitney U with n2 greater than 20.

bProbability of a true null hypothesis.

However, only the differences between the telephone group
and the letter group are significant.
Consumer Chggactepistics
and Cognitive Dissonance
The final hypotheses of the study were concerned
with whether certain types of consumers were more likely to
experience dissonance than other types. Of particular inter-
est was whether consumers who expended more effort in shop-
ping for the applicance would subsequently eXperience less
dissonance, as had been claimed [13].
The consumers in each Of the three groups were
classified in the following 14 categories:
1. Shopping effort eXpended
2. Number of stores visited while shopping

3. Number of brands considered while shopping

79

4. Education level of consumer

5. Race of consumer

6. Age Of consumer

7. Family income of consumer

8. Brand of appliance purchased

9. Branch of store from which purchase was made

10. Consumer's previous eXperience with purchased brand
11. Stage in the family life cycle

12. Social class of consumer

13. Sex Of consumer

14. The interviewer who administered the questionnaire.

A KruskalAWallis one way analysis of variance by
ranks was conducted for each category for each group. The
test statistic was computed for each group individually,
instead of for all groups taken together, in an attempt to
prevent confounding Of the variables. The results are sum-
marized in Table 12, which, because of its length, has been
placed in Appendix A.

Without exception, Table 12 shows no significant
differences among the dissonance scores of the subjects when

they were classified by any of the categories investigated.

A Concluding Note

The findings Of the eXperiment have been presented in
Chapter IV and have followed the organization of the research
hypotheses. Since a summary Of the hypotheses and findings

is presented in Chapter V, no such review is made here.

CHAPTER V

SUMMARY AND CONCLUSIONS

Introduction

The purpose of Chapter V is to present a summary Of
the study and the major conclusions.

The chapter begins with a brief summary of the
purpose Of the research and the research design. The next
section reviews the research hypotheses and the findings.
The next section contains the major conclusions and implica-
tions for marketing. The last section suggests some direc-
tions for future research.

Brief Summary of Pgipose
and Research Desigm

The purpose of the research was to investigate some
of the action-oriented marketing implications of the theory
of cognitive dissonance. The study has investigated whether
certain post-transaction messages from a retailer to recent
purchasers would be effective in assisting the consumer in
reducing dissonance, and, if the messages were effective in
reducing dissonance, what benefits accrued to the retailer

for providing the messages.

80

81

An "after only-with control" design was used in the
experiment to test the above marketing implications Of cogni-
tive dissonance theory.

A sample population consisting of purchasers of
refrigerators from a major department store in the Detroit
metropolitan area was randomly divided into a control and
two experimental groups with a total N = 152. One eXperimen-
tal group received a letter from the department store giving
the consumers post-transaction reassurances (see Appendix E
for the message). The second experimental group received a
telephone call from the department store eXpressing the same
types Of post-transaction reassurances as were contained in
the letter (see Appendix F for the message).

Interviews were conducted with all subjects approx-
imately 7 days after the transaction took place.

To determine whether the messages were effective in
reducing dissonance, the extent Of post-transaction doubt
was used as an indicator Of the magnitude of dissonance the
groups were eXperiencing. To determine what benefits might
accrue to the retailer for providing the dissonance reducing
messages, the attitudes of the consumers toward the store,
as well as the cOnsumers' expectations of future shopping
behavior, were measured.

Demographic data were obtained to determine whether
certain types of people were more inclined to experience

dissonance than others. Of particular concern was the

82

relationship between shopping effort and dissonance, since
previous studies had shown an inverse relationship between
dissonance and shopping effort.

Overall tests of significance were conducted using
the KruskalAWallis one way analysis of variance. Pair com-
parisons were made using the MannAWhitney U test statistic

(see Appendix G for statistical procedures).

Review of Hypotheses and Findings

The following sections will review the findings of
the study with respect to each guiding hypothesis.
Guiding Hypothesis No. 1: The
Effectiveness Of Post-Transaction
Communications in Reducing
Cognitive Dissonance

The major guiding hypothesis of the study was that
post-transaction reassurances from a retailer would be
effective in assisting customers in reducing cognitive dis-
sonance. The study hypothesized that the dissonance scores
of the group which received the telephone call would be
lower than the group which received the letter, which would,

in turn, be lower than the control group. (See Appendix B

for the specific research hypotheses.)

Findings of the Study
There was a tendency for the dissonance scores of
the subjects in the letter group to be lower than the disso-

nance scores Of the control group, as hypothesized. Contrary

83

to the hypothesis, the dissonance scores Of the telephone
group were the highest of the three groups. Although the
overall differences among the groups were not significant at
the .05 level, the differences between the letter group and
the telephone group were significant.
Guiding Hypothesis NO. 2:
Attitude Change and
Dissonance Reduction

The second guiding hypothesis posited that the
groups which received the post—transaction messages would,
subsequently, have more favorable attitudes toward the store

than the control group which received no post-transaction

message.

Findings of the Study

There was a general tendency for the attitudes of
the letter group to be more favorable than the control group.
Contrary to the hypothesis, the attitudes of the telephone
group were the most unfavorable Of any of the three groups.
Only the differences among the groups for the attitude of
store congeniality were significant at the .05 level.
Guiding Hypothesis NO. 3:
Predisposition to Purchase
and Dissonance Reduction

The third guiding hypothesis posited that the groups
which received the post-transaction messages would be more

prediSposed to purchase from the store than the control

group which received no post-transaction message.

84

Findings of the Study
The subjects who received the letter tended to have
higher expectations of future purchases than did the sub-
jects in the control group. Contrary to the hypothesis, the
subjects in the control group tended to have higher expecta-
tions of future purchases than did the subjects in the tele-
phone group. Although the overall differences among the
three groups were not significant, the differences between
the telephone group and the letter group were significant at
the .05 level.
Guiding Hypothesis NO. 4:
Consumer Characteristics
and Cognitive Dissonance
The fourth guiding hypothesis concerned whether cer-

tain types of consumers would be more likely to experience
dissonance than other types. It was hypothesized that as
shopping effort increased, dissonance would decrease. It
was further hypothesized that there would be no relationship
between level of dissonance and each Of the following vari-
ableS:

1. Number of stores visited while shopping

2. Number of brands considered while shopping

3. Education level of consumer

4. Race of consumer

5. Age Of consumer

6. Family income of consumer

7. Brand of appliance purchased

85

8. Branch of store from which purchase was made

9. Consumer's previous eXperience with purchased brand
10. Stage in the family life cycle

11. Social class of consumer

12. Sex of consumer

13. The interviewer who administered the questionnaire.

Findings of the Study

Contrary to the hypothesis, there was no significant
relationship found between shopping effort and the level of
dissonance. Also, there was no significant relationship
between dissonance and any of the other demographic vari—
ables.

Conclusions and Implications
for Marketing

The major conclusions and implications for marketing
based on the author's interpretation of the results of the
study are presented in the following sections. Since most
Of the findings did not achieve the customary .05 level of
significance, the conclusions must be considered as tenta-
tive.

l. The major conclusion of the study concerns
whether post—transaction reassurances from a retailer to
recent purchasers would be effective in reducing dissonance.
The study yielded evidence to support the conclusion that at
least some types Of messages would be effective in reducing

dissonance, since the subjects in the letter group did tend

86

to eXperience less dissonance than did the subjects in the
control group. However, it would appear that good intentions
are no substitute for good information, since the subjects
who received the telephone call actually eXperienced more
dissonance than the subjects in the control group.

Neither theory nor the other findings of the study
explain why the results of the telephone group were contrary
to expectations. The author Offers the following conjectures
as possibly contributing to the finding:

1. The subjects who received the telephone call may
have suSpected some ulterior motive for its purpose.

2. The telephone call may have interrupted some of
the subjects at inopportune times.

3. Some Of the subjects may have been annoyed by firms
that sell over the telephone, and, therefore, inter-
preted the telephone call as a sales device.

The above conclusions would imply that retailers
and/or manufacturers should exercise great caution before
embarking on a program designed to help reduce the dissonance
of recent purchasers. Not only might the program be ineffec—
tive in helping the recent purchasers, but, also, the program
could aggravate the situation by increasing the dissonance of
recent purchasers. Certainly, extensive pre—testing of any
specific program would be advisable.

2. The second conclusion of the study concerns one
of the benefits posited to accrue to the retailer for pro-

viding post-transaction reassurances to recent purchasers.

Since the subjects who received the letter generally had

87

more favorable attitudes toward the store than did the con-
trol group, it would appear that a program of providing post-
transaction reassurances might be effective in improving the
attitudes toward the store of recent purchasers. However,
since the attitudes Of the telephone group were less favor-
able than the control group, it would be erroneous to con-
clude that all such programs would result in more favorable
attitudes.

The above conclusions would imply that if a retailer
and/or manufacturer desired to use post-transaction reassur—
ances as a vehicle to improve the attitudes of recent pur—
chasers, extensive pre-testing of such a program would be
recommended since the efforts could have either positive or
negative results.

3. The third conclusion concerns another benefit
posited to accrue to the retailer for providing post-trans-
action reassurances to recent purchasers. It would appear
that it is possible to increase the consumer's expectations
of future shopping activity by providing post—transaction
reassurances, since the letter group had higher "predispo-
sitions to purchase" than the control group. Once again,
all well-intentioned efforts cannot be expected to be
rewarded, since the telephone group actually had lower
eXpectations of future shopping activity than did the con-

trol group.

88

The above conclusions would not necessarily imply
that the subjects that received the post—transaction reassur-
ances would actually purchase more in the future than the
control group. Intentions are not necessarily the same as
behavior. Only a study with a longitudinal design could
yield information as to whether respondents actually behaved
according to expectations.

4. The fourth conclusion concerns whether certain
types of consumers are more likely to experience dissonance
than other types. The study did not yield evidence to sup-
port the assertion that any one of the 14 demographic vari—
ables was correlated with dissonance.

The above would imply that at the present time none
of the 14 demographic variables can be used with confidence
as a market segmentation tool for predicting propensity to
experience dissonance among consumers.

Cognitive Dissonance and
Marketing: A Commentapy

The study has attempted to test some of the implica—

tions of the theory of cognitive dissonance to marketing.
As with many research projects, the study appears to have
raised at least as many questions as it has answered:
1. Why did the telephone call actually increase post-
transaction anxiety in the subjects, instead of

reducing it?

89

2. Since the telephone call was not effective, would
such an impersonal, hit-or-miss technique such as
advertising be effective, as has been suggested in
the marketing literature [31:75]?

3. Would different copy in either the letter or tele-
phone call have changed the results?

4. The letter appeared to be effective in helping
consumers reduce anxiety. Would it be worthwhile
for a retailer to initiate such a program? Would
the benefits be worth the costs when compared with
other programs?

5. Would a larger survey-type design have been success-
ful in segmenting the customers along demographic
characteristics with respect to dissonance?

6. DO the absolute values of the dissonance scores have
any significance? Without a benchmark against which
to compare, it is difficult to make any statement as
to the absolute level of post-transaction anxiety
the subjects were eXperiencing. Only comparisons

between groups were possible.

In an age when words such as "interdisciplinary,"
"transdisciplinary," and "multidisciplinary" are the fad, it
is well to reflect on the dangers of borrowing theories from
other disciplines. Most modern marketing texts which have

sections on consumer behavior will discuss the theory of

9O

cognitive dissonance. Unfortunately, some texts present
the theory in much more positive terms than would appear
warranted.

The jury is still out on dissonance theory, and the
results of the present study would indicate that policy
recommendations based on dissonance theory are tenuous.
Such general recommendations as: "Manufacturers also can
run advertisements showing recent purchasers showing satis-
faction with their choice, and why," may be ill-advised
since there exists no eXperimental evidence to support the
assumption that the advertisements would be effective in
accomplishing the desired result. What is worse, since the
telephone call in the study actually aggravated the situa-
tion instead of improving it, such general policy recommenda-
tions might produce negative results instead of simply being
ineffective.

Marketing cannot afford to be, and should not be,
parochial, and yet, borrowing a theory from another disci-
pline (even under the halo of being "interdisciplinary")
must be done only after extensive investigation of the
theory in a marketing context. There would appear to be a
great deal of research to be done before definitive policy
recommendations can be made using dissonance theory as a

foundation.

91

gmggestions for Future Research

There are numerous directions that future research
could take in the general area of the marketing implications
of cognitive dissonance theory. The following represent a
sample of the author's suggestions:

1. The study was restricted to one type of purchase,
major appliances, and to one type Of product within this
group, refrigerators. Replication Of the study across
product categories would provide information on how "impor-
tant" the purchase has to be in order for dissonance to be
a factor. The study could provide information as to whether
price is a good indicator of "importance." Such studies
could seek to delineate which types of purchases are most
likely to be dissonance arousing.

2. The study was limited to one type of store, a
"conventional" department store. Replication of the study
for other types of establishments would be useful, e.g., for
"discount“ stores. Such studies could seek to delineate the
possible relationships among store prestige, service poli-
cies, merchandising policies and consumer dissonance.

3. The study used two types of post-transaction
messages, a letter and a telephone call. Other studies
could use different media and/or different copy. Newspapers
or television advertising might have completely different
results. The copy could go into the more technical details

of the product. Such studies would reveal whether the

92

negative results of the telephone call were unique to that
particular medium.

4. Finally, the research design did not permit the
inclusion of information concerning the differences in
shopping behavior Of the three groups after the receipt of
the post-transaction message. A longitudinal design which

included this important variable would be desirable.

APPENDIX A

TABLE 12

APPENDIX A

Table 12. Summary Of KruskaléWallis H and associated
probabilities with the null hypothesis that the
perceived dissonance scores of the subjects when
classified by the various consumer characteristics
came from the same population

 

 

Kruska14Wallis b Level of
Group Ha P Significance

 

Shopping effort expended:

Control 0.3388 0.8442 NS
Letter 2.7440 0.2541 NS
Telephone 0.9780 0.6132 NS

Numper of stores visited while shopping:

Control 1.4506 0.4842 NS
Letter 2.2788 0.3200 NS
Telephone 2.5512 0.2793 NS

Number of brands considered while shopping:

Control 3.8433 0.1464 NS
Letter 0.4391 0.8029 NS
Telephone 3.4074 0.1820 NS

Educational level of consumer:

Control 16.3885 0.0118 .05
Letter 1.5246 0.9102 NS
Telephone 2.7763 0.7344 NS

Race of consgmer:

Control 0.9121 0.6338 NS
Letter 2.5335 0.2817 NS
Telephone 0.3693 0.5434 NS

Age of consumer:

Control 3.9178 0.5613 NS
Letter 2.0662 0.7236 NS
Telephone 5.8286 0.2123 NS

Family income of consumer:

Control 7.2397 0.2992 NS
Letter 5.7553 0.4512 NS
Telephone 12.1455 0.0588 NS

93

94

Table 12--Continued

 

 

Kruska14Wallis Level of
Group Ha P Significance

 

Brand of appliance purchased:

Control 6.9864 0.1366 NS
Letter 2.5743 0.4620 NS
Telephone 3.9427 0.4138 NS

Branch of store from which purchase was made:

Control 6.1705 0.2900 NS
Letter 4.7791 0.3107 NS
Telephone 1.2320 0.8728 NS

Consumer's previous experience with purchased brand:

Control 0.6415 0.7256 NS
Letter 2.2370 0.3268 NS
Telephone 3.5776 0.1672 NS

Stagegin the family iife cycle:

 

Control 7.3137 0.1202 NS
Letter 1.5023 0.9128 NS
Telephone 5.1216 0.4012 NS
Sociai class of consumer:

Control 5.3734 0.0681 NS
Letter 0.4339 0.9332 NS
Telephone 0.4380 0.8033 NS

Sex of consumer:

Control 1.0285 0.3105 NS
Letter 1.9408 0.1638 NS
Telephone 0.0147 0.9036 NS

The interviewer who administered the questionnaire:

Control 1.2918 0.8628 NS
Letter 3.7157 0.4458 NS
Telephone 2.6310 0.7566 NS

 

aThe test statistic was computed for each group indi-
vidually, instead Of for all groups taken together, in an
attempt to prevent confounding of the variables.

bProbability of a true null hypothesis.

APPENDIX B

RESEARCH HYPOTHES ES

APPENDIX B

RESEARCH HYPOTHESES

A guiding hypothesis Of the proposed research is
that manufacturers and/or retailers can assist consumers in
their efforts to reduce cognitive dissonance by providing
post-transaction reassurances. Furthermore, not all types
of post-transaction messages will be equally effective.

The specific hypotheses are:

Al. The mean perceived dissonance scores of sub-
jects in Group III will be lower than the subjects in Group
I, and the differences will be significant.

A2. The mean perceived dissonance scores of the
subjects in Group II will be lower than the subjects in
Group I, and the differences will be significant.

A2.1. The mean perceived dissonance scores of the
subjects in Group III will be lower than the subjects in
Group II, and the differences will be significant.

A second guiding hypothesis is that one of the ben-
efits posited to accrue to the retailer for providing dis-
sonance reducing information is that the groups which
receive assistance will then view the store more favorably
than the groups which do not receive assistance. The

Specific hypotheses are:

95

96

B1. The mean scores of the subjects in Group III

for each of the five additional dimensions of store image
will be lower (more favorable) than the mean scores of
subjects in Group I, and the differences will be significant.

B2. The mean scores of the subjects in Group II for
each of the five additional dimensions Of store image will
be lower (more favorable) than the mean scores of subjects
in Group I, and the differences will be significant.

B2.1. The mean scores of the subjects in Group III
for each of the five additudinal dimensions of store image
will be lower (more favorable) than the mean scores of sub-
jects in Group II, and the differences will be significant.

A second benefit posited to accrue to the retailer
for providing dissonance reducing information is that the
assisted groups will be more predisposed to purchase at that
store than the group which does not receive assistance. The
specific hypotheses are:

C1. The mean scores of subjects in Group III for
prediSposition to buy will be lower (more favorable) than
the mean scores for subjects in Group I, and the differences
will be significant.

CZ. The mean scores of subjects in Group II for
predisposition to buy will be lower (more favorable) than
the mean scores for subjects in Group I, and the differences

will be significant.

97

C2.l. The mean scores of subjects in Group III for
prediSposition to buy will be lower (more favorable) than
the mean scores for subjects in Group II, and the differ-
ences will be significant.

Past research has indicated that people who expend
more effort in shOpping will eXperience less cognitive dis-
sonance [5]. The Specific hypotheses are:

D1. Within eggm_of the three groups of subjects,
those individuals who are classified as "Low Effort" will
have higher mean perceived dissonance scores than those who
are classified as "Medium Effort," and the differences will
be significant.

Dl.l. Within each Of the three groups of subjects,
those individuals who are classified as "Low Effort" will
have higher mean perceived dissonance scores than those who
are classified as "High Effort," and the differences will be
significant.

D2. Within each of the three groups of subjects,
those individuals who are classified as "Medium Effort" will
have higher mean perceived dissonance scores than those who
are classified as "High Effort," and the differences will be
significant.

Prior research has indicated that certain personal-
ity types are more likely to experience cognitive dissonance
than others [8:14-15]. However, there has been no research

which would indicate that the level of cognitive dissonance

98

should vary with any of the six demographic variables of age,
education, family income, stage in the family life cycle,
social class, and race. The Specific hypotheses are:

El. Within each of the three groups there will be
_g_signiticant differences in mean perceived dissonance
scores when subjects are classified by age.

E2. Within each of the three groups there will be
mg_significant differences in mean perceived dissonance
scores when subjects are classified by education.

E3. Within each of the three groups there will be
mg_significant differences in mean perceived dissonance
scores when subjects are classified by income.

E4. Within each of the three groups there will be
.mg significant differences in mean perceived dissonance
scores when subjects are classified by race.

E5. Within each Of the three groups there will be
mg_significant differences in mean perceived dissonance
scores when subjects are classified by stage in the family
life cycle.

E6. Within each Of the three groups there will be
m2 significant differences in mean perceived dissonance
scores when subjects are classified by social class.

E7. Within each of the three groups there will be
.mg significant differences in mean perceived dissonance

scores when subjects are classified by brand purchased.

99

E8. Within each of the three groups there will be
1mg significant differences in mean perceived dissonance
scores when subjects are classified by number of stores
visited.

E9. Within each Of the three groups there will be
‘_g significant differences in mean perceived dissonance
scores when subjects are classified by number of brands
considered.

E10. Within each of the three groups there will be
mg_significant differences in mean perceived dissonance
scores when subjects are classified on the basis Of whether
they are credit card purchasers or not.

Ell. Within each Of the three groups there will be
mg_significant differences in mean perceived dissonance
scores when subjects are classified on the basis of whether

the appliance has been delivered or not.

APPENDIX C

THE QUESTIONS USED

II.

APPENDIX C

THE QUESTIONS USED

Perceived Dissonance

l.

2.

6.

If a friend were going to buy a similar appliance,
I would (would not) recommend this store.

If a friend were going to buy a similar appliance,
I would (would not) recommend this brand.

There is no doubt (considerable doubt) in my mind as
to whether I should have made my purchase in this
store.

There is no doubt (considerable doubt) in my mind as
to whether I Should have selected this particular

brand of appliance.

I got a "good deal" ("bad deal") on this appliance.

I should have spent more (less) time in "shopping
around" for this appliance.

Store Image

Merchandipe Suitability

1.

They Show the newest styles sooner (later) than other
department stores.

Sells high (low) quality merchandise.

The merchandise they sell is (is not) the kind I
like to buy.

You can always (never) find a wide variety of things
to choose from.

A good (poor) place to get new ideas about home
furnishings.

Rarely (frequently) runs out of merchandise.

Prices are likely to be lower (higher) here than in
other department stores for the same merchandise.

100

101

8. When they have a sale, the sale merchandise is of
(below) regular quality.

9. I am always (never) satisfied with the goods I buy
there after I purchase them.

Store Services

1. They are liberal (strict) on exchanges and returns.
2. Their advertising is informative (not informative).
3. Charge account policies are good (poor).

4. Very liberal (strict) in cashing checks.

5. Repair services on appliances are excellent (poor).
6. A reliable (an unreliable) store.

Sales Personnei

1. Sales people make you feel that you are important
(unimportant).

2. Sales people know (don't know) their merchandise
very well.

3. Sales people are (are not) interested in being of
service to customers.

4. Sales people are courteous (discourteous) to custom-
ers.

5. Sales people make you feel comfortable (uncomfortable)
if you are just browsing.

6. Sales people are always (never) there when you need
them.

Congeniality
1. A warm, friendly (cold, unfriendly) store.

2. The store is never (Often) too crowded for enjoyable
shOpping.

3. I am likely (unlikely) to meet people like myself
there.

III.

IV.

102

4. The appearance of the store is attractive (not
attractive).

5. Shopping at the store is enjoyable (not enjoyable).
Locatipnal Convemiepce

1. Easy (difficult) for me to get there.

2. Parking facilities are excellent (poor).

3. I can (can't) save time by shopping there.

4. The way departments in the store are arranged makes
it easy (difficult) to find things.

5. Convenient (inconvenient) because it is near (not
near) other stores I want to shop at.

Predisppsition to Buy

1. In the future I plan to do a much higher (much lower)
percentage of my shopping in this store than I have
done in the past.

Shopping Effort

1. How many stores did you visit in shOpping for this
appliance (including the final choice)?

a. one
b. two
c. three or more

2. How many brands did you consider in shopping for
this appliance (including the final choice)?

a. one
b. two
c. three or more

Demographic Data

1. Please indicate the letter which corresponds to your
age category.

Less than 20 years

More than 20 but less than 30 years
More than 30 but less than 40 years
. More than 40 but less than 50 years
. More than 50 but less than 65 years
65 years or more

'IJtIZIUOUJIP'

103

Please indicate the letter which corresponds to the
last year of school which you completed.

Grade school or less

B Some high school

C Graduated from high school
D. Some college

E. Graduated from college

F. Graduate or advanced degree

>

Please indicate the letter which correSponds to your
total family income before taxes last year.

Less than $4,000
$4,000-$7,999
$8,000-$9,999
$10,000-$12,499
$12,500-$l4,999
$15,000-$24,999
$25,000 or more

Q'fldet3U1>

What is the race of the respondent?

A. White
B. Negro
C. Other (please specify)

 

Was your "old" appliance the same brand as your "new"
appliance?

 

a. Yes
b. No
c. Did not previously own this type of appliance

IF ANSWER TO 5 IS "b" OR "C" THEN ASK QUESTION 6.

6.

Do you at the present time own an other major
kitchen and/or laundry appliance [stove, refriger-
ator, dishwasher, clothes washer, dryer, or freezer]
which is the same brand as your new appliance?

a. Yes
b. No

Are you married?

A. Married

B. Single
C. Widowed
D. Divorced, Separated

10.

11.

12.

13.

14.

104

Do you have children living at home? [INCLUDE
CHILDREN SUPPORTED BY PARENTS WHO ARE AWAY AT SCHOOL].

A. Yes
B. No
C. Not applicable

 

What is the occupation of the male head of the house—
hold? [IF NO MALE HEAD THEN WHAT IS THE OCCUPATION
OF THE FEMALE HEAD?]

 

Where does he (she) work? [COMPANY NAME, ETC.]

 

Please indicate the letter which corresponds to the
greatest source of your family income.

. Inherited wealth

. Profits and dividends
. Earned salary

. Earned hourly wage

. Other, please specify

M!3(3U19

 

DO you own or rent this building?

A. Own
B. Rent

Would you please indicate which price category this
home would fall into if it were placed on the market
today?

More than $50,000

$35,000 to $49,999
$20,000 to $34,999
$15,000 to $19,999
$12,000 to $14,999
$10,000 to $11,999
. Less than $10,000

Q'flfilUCfiUfi»

 

Would you please indicate the category into which
your monthly rent falls?

. More than $500
$250 to $499
. $150 to $249
. $100 to $149

Utiu1>

15.

E.
F.

G.

105

$75 to $99
$50 to $74
Less than $50

Please Observe the respondent's neighborhood and
indicate into which Of the following classifications
it falls:

A.

Very high; North Shore, etc.

High; the better suburbs and apartment house
areas

Above average; areas all residential, larger
than average space around homes; apartment
areas in good condition

Average; residential neighborhoods, no
deterioration in area

Below average; area not quite holding its
own

Low; considerably deteriorated, run-down,
and semi-slum

Very low; slum.

APPENDIX D

INTERVIEWER INSTRUCTIONS AND QUESTIONNAIRE

APPENDIX D
INTERVIEWER INSTRUCTIONS

Interview Kip

Each interviewer will be provided with the following:

a. List of subjects.

b. Identification card and letter.

c. Questionnaires, instructions, and respondent cate-

gory cards.

Subject lists

All interviewers will be supplied with a list of sub-
jects at least one day prior to the day the interviewing
will take place. The list will contain the following for
each subject:

a. Name and address.

b. Telephone number.

c. Respondent number.

d. Brand name and type of appliance purchased.

e. Date on which interview will be conducted.
Who will be interviewed?

The female head of the household will be interviewed.
If there is no female head of the household, then the male
head Of the household can be interviewed. No one is to be
interviewed who is not on the subject list.
What about call-backg?

No interviews are to be made before the date on the
subject list. If the female head of the household is not at

home or cannot be interviewed at that time, try to arrange

an appointment for later that day or one of the two days

106

107

immediately following. If this is not possible, the subject
should be dropped from the active list and a notation made
to that effect on the subject list. Telephone calls to
set-up appointments are to be attempted. Subjects may be
dropped from the active list if there is no answer after
three telephone calls. If two calls are in the afternoon,
then the third must be in the evening, and vice-versa.
Thepguestionnaire

The questionnaire is in three parts: one part (pages
1-4) contains demographic questions to be asked of the
reSpondent. Part II of the questionnaire (pages 8-14),
which contain the rating scales, should then be self—
administered by the respondent. Upon the conclusion of the
rating scales, the demographic questions in Part III (pages
5-7) should be asked.

Be certain that the respondent's number appears both on
the front of the questionnaire and in the upper left corner

of page 8.

QUESTIONNAIRE

COGNITIVE DISSONANCE RESEARCH STUDY

DEPARTMENT OF MARKETING AND TRANSPORTATION ADMINISTRATION

MICHIGAN STATE UNIVERSITY

EAST LANSING, MICHIGAN

Respondent No.:

 

Name:

 

Street Address:

 

City, State:

 

Telephone:

 

Purchase Date:

 

Interview Date:

 

 

_. C7
Appliance Type and Brand: - C8
Interviewer Neme: C9

 

 

 

 

C1
Do Not write Below This Line C2
CB
No: . C“
Group: C5
Days: C6

 

108

-Lo-

109

Introduction

Good morning [AFTERNOON, EVENING], my name is

 

from the Market Opinion Research Company. ‘we are conducting a study
for Michigan State University concerning appliance purchasing habits.
Several major companies are cooperating on the study and our records
indicate that you recently purchased a refrigerator. Is that correct?
[PAUSE FOR REPLY].
May I come in to ask a few questions about your purchase experience?
[AFTER GAINING ENTRY, GET RESPONDENT TO SIT DOWN FOR FOLLOWING
QUESTIONS].
8.1 How many stores did you visit in shopping for this appliance
010 ....including the final choice?
_____ A. One
_____.8. TWO
C. Three or more
8.2 Which stores did you visit?

C11

 

 

8.3. In which store did you make your final purchase? [BE

C12 SPECIFIC, NORTHLAND, EASTLAND, ETC.]

 

8.5. Why did you decide to buy your refrigerator at that store?

C13 [PROBE]

 

 

 

9.1
C14
*015

9.2.

9.3
016

10.1

C17

10.2
C18

*C19

-3-
110
How many brands did you consider in shopping for this
appliance?
__ A. One or two
B. Three or four
C. Five or more

Which brands did you consider when you were shopping?

 

 

Why did you select your final brand? [PROBE]

 

 

was your "old" appliance the same brand as your "new"
appliance?

___.’_A. Yes

_______ B. NO

______C. Did not previously own this type of appliance.
Do you at the present time own any other major kitchen
and/or laundry appliances which are the same brand as
your new appliance?

A. Yes

B. No

[PUT RESPONDENT'S NUMBER IN THE UPPER LEFT CORNER OF THE FIRST

PAGE OF THE RATING SCALE PORTION OF THE QUESTIONNAIRE. HAND

RESPONDENT THE RATING SCALES AND INSTRUCT HER.AS POLLowS:]

Since you have recently purchased a major appliance, we are

interested in how you feel about your purchase. For all the questions,

keep in mind that we are interested in how you personally feel about

.11..
111
the 1_J_r_a_r_1_d_ of appliance you bought and the £9239. where you bought it.

The questions are similar to the sample question on the first
page. '

As you will notice, each question has two statements, one on the
left side of the page and one on the right. The statement on the
right says, "The store where I bought my appliance was gilty".

Whereas, the statement on the left says, "The store where I bought
my appliance was 91.3531".

First we want you to determine which of the two statements you
agree with. For example , if you agree with the statement on the right,
you will want to place an "X" in g_n_e_ of the three Spaces to the righi
of the center box. If you felt the store was m dirty, then you
would place your "X" in the space over the word "very" to the right
of the center box.

On the other hand, if you felt the store was "quite" sleep, then
you would want to put your "X" over the word "quite" to the is}: of the
center box.

If you don't-agree with giflie}: of the two statements, then you will
want to put your "X" in the center box over the word "neutral".

Please put an "X" in only one [space per question. However, be sure
to answer every question and remember that we're interested only in how
you feel about the RISES. of appliance you bought and the £53333 where
you bought it.

Are there any queStions? [IF "NO", ASK RESPONDENT T0 BEGIN
FILLING IN THE RATING SCALES].

[WHEN RESPONDENT IS FINISHED. TAKE BACK RATING SCALES AND

CONTINUE WITH THE QUESTIONNAIRE].

C20

C21

C22

*C23

*C24

C25

026

C27

11.2

11.3

- 12.1

12.2

12.3

12.4

12.5

112

Are you married?

A. Married*

B. Single
C. Widowed

m

D. Divorced , Separated

 

*[ IF MARRIED ASK 11.2, IF' NOT, SKIP T0 11.3]

Do you have children living at home? [INCLUDE CHILDREN
SUPPORTED BY PARENTS WHO ARE AWAY AT SCHOOL].

_____ A. Yes

_____ B. No

C. Not applicable
Please indicate the letter which corresponds to your age

category. [HAND RESPONDENT CARD # 1].
What is the occupation of the male head of the household.

[11* NO MALE HEAD, THEN WHAT IS OCCUPATION 0F FEMALE HEAD].

 

Where does he (she) work? [COMPANY NAME, ETC.].

 

Please indicate the letter which correSponds to the greatest
source of your family income. [HAND RESPONDENT CARD #2].
______ ______ ______ ______ ______ Other, please Specify.
Do you own or rent this building?.
__ A. 001*

_____ B. Rent

[11“ OWNED ASK 12.5.11? RENTED SKIP T0 12.6].

WCuld you please indicate which price category this home
would fall into if it were placed on the market today?

[HAND RESPONDENT CARD #3].

 

1 [I'm
. 111]
ll! I.||I.ll[\[1r[l 1 I'll

-6-
113
12.6 Would you please indicate the category which your monthly
C28 rent falls into? [HAND RESPONDENT CARD # h].
13.1 Please indicate the letter which corresponds to the last
C30 year of school you completed. [HAND RESPONDENT CARD #5].
13.2 Please indicate the letter which correSponds to your total
C31 family income before taxes. [HAND RESPONDENT CARD #6].
13.3 Do you have a Hudson"s credit card?
C32 Yes

No

 

13.4 Has your appliance been delivered yet?
C33 _________ A. Yes
8. No
[THE FOLLOWING QUESTIONS ARE TO BE ANSWERED BY THE INTERVIEWER
BY OBSERVATION].
1M1 What is the race of the respondent?
c3“ ______‘ A. White
__ B. Negro
________ C. Other, Specify -
10“? Please observe the respondent's neighborhood and indicate
C35 into which of the following classifications it falls.
A. Very high; North Shore, etc.
13. High; the better suburbs and apartment house areas.
C. Above average; areas all residential, larger than
average space around houses; apartment areas in
good condition.

D. Average: residential neighborhoods, no deterioration

114

E. Below average: area not quite holding its own.
F. Low; considerably deteriorated. run-down, and

semi-811m.

G. Very low, slum.

115

90.3.00 as:
350.3030 be 0.0900
H 0.02.? 0.030 00.0. .

huo>

 

3.00.6

 

 

00000000 00.5002 00000000 30.00

 

 

0 oz 8388 008000

 

hae>

 

 

.80.? 25 0020000090 he
pewson H p.803 oaovm one

.000wa wow
.00 08.000 vflsuwmmao

800000 000.0 00

Nmmm .00. 00000.0 000000.000
000 00000 000 :0
00005400900 00: 0:9

.OHOSa. ..

9H0mhs 0x00 0Hnooa

. 0005 00 Nfloxwaca 80 fl.

6
n 6000300000
0000 .8 080 080 0 0o0 0

 

0».
00:0 00 0:03 H 000000
00:00 000: 00: 00 wm
0000000 .ac00:0>:oonH

.H003.000>

0000:000003.000:¢ 30:3
0.:00 canoon uofium

 

 

.0smsw0 8000 0
1:00 0:4 00: 00 H000
000» 0000:0n0005 0&9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.80 8003 00000000 08.00.80 00000000 8005 0.80
a u u- _. 0 a

0.80 8008 00000000 00.00080 00000000 8005 0.80

0.80 8003 B00000...“ 00.000801 0.0000000. 8005 0.80
0 . w ._ w . . 0

0.80‘ 8003 00000000 r3080 00.000000. 8005 0.80

0.80 8008 00000000 00.00080 00000000 8008 0.80

0.80 8008 00000000 08080 00000000 8005 0.80
a a u- w a . 0

0000 00000 000>
.

8005 00000000 0900080 00000000
0 .

.0howv
00m 00. 06 .00.“ 00m

E

000% 0» vanvammfiu
00 00x05 vowcnaha
000 00000 000 :0
0¢aoaahunov_hnz 009

.00000
Haomza 000a 0amoom
0008 00 Nfloxda 30 H

0000000000 0000
:0 0000 000m 0 now H

O 8..“

00:0 00 0:03 H 00000»
00:00 000: ad 90
0000000 .0200000000

 

.0003 0000
00000030005 000:0
30:0 camoom noaum

 

.000 3. 2000
H 0200 0:0 00 HH00

00:0 0»00:u:0005.0ga

OH.”

0:00

on.“

.H.H

OmOo

0N0:

.n.~

.mchspoh can mowcunoxa
no HanopHH ohm mmna

 

.mnosovmao ow moH>hom

.Ho wfiwmufi EHmfioHsH
wan oHnooa onum

.ohopm
man vcosaoomh vHsoa H

Seaman .HmHHSHm a >45 op,

madam who: vzuwum a «H «

 

117

.mHOHn oHnuHHagzs <

.wcamsoun papa can

no» HH oHpuHuoHsco HooH
so» oxas oHaooa uoHum

 

.HnuaammaH

and 90% was» Hoom
99h oxus aHmooa uoHum

 

.oaoum was» 2H «unsungn
as ovus opus uHaonm
H uanvogs 0» ma wads
ha :a assoc o: nu «Mona

 

z§o>

huo>

 

hho>.

 

auo>H

 

.hfio>

 

muo>

 

hhp>

 

mHHpo HHHgmHHm ruanoz HHHgmHHm

oHHso HHHanHm HuaHnmz HHHsmHHm
«

 

.E

oHHso HHHsmHHm Huuaaoz HHHHMHHm

 

E

oHHuu HH+quHm Haquoz HHHHMHHm

 

E

oHHao HHHgmHHm Hugusoz HHstHHm

 

39¢
a

 

oHHao HHHanHm Huanoz HHHHMHHm
.

:E

.jfifiHHfifié fifima

.D

 

 

 

mHHsa Huo>
.

 

 

ouwzo hho>
H

 

 

oHHao Huo>
a

 

oHHaa
a

aho>

 

 

oHHao Hua>
.

oHHsa huo>
.

 

£6!
a

 

_huop

.mcnsvon and mowcdnoxu
cc vofinvm ohu_ho:a

 

.muosuvuso op oedphom

«0 wnHon 2H cavmohovcw
Ho: Qua onoom mOHum

 

 

.onopm was» unusadoou
no: uHaoz H HoocuaHmmu
huHHSHa «.599 on
madam 0&9: vacuum a «H

 

 

.ouovm OHDuHHoh 4

.mcdmwbun pmpn ¢nu :9»

HH oHnuphomabocu Haom
50% oxua mHmoom uoHum

 

.unsphmmaflca.
09d 50h wasp Hoom
sch oxua aHnoom uoHum

 

.onovn «an»

:H amazonsn as cums
aha: cHsonu H gunpomm
op ma vcHa.ha 2H 93:06

oHnnuouHmmwo «H.oaoga

.H.m

Ono‘:

.N.H

Omo¢fl

0H0:

.m.H

.mocuHHmaa

awn» AOH avnsouu
wcdmmonmz :« osdv ¢Aoa
‘ pecan 0>u£ UHsogu H

 

.oafivcunouoa can» 0:9

you m¢hopm unoavuuncv
heave :H zany whon_mmmmmm
on oa_nHoxHH mum moOHhA

.HJHHmsU.Aummmon Ho,

ma omfivcmncuos oHun asp
.oHam a m>us moan cogz

118

.wcaamoga oHnnHOnco

you vovsbho 00¢
pupae ma oaovm one

 

.ccmun «ng unussco¢h
Ho: vHsoa H HoocuHHmmu
HuHHEHo «.559 on
mfiom 98: 23¢ a .HH

 

.uhosamnao

. on cacophdoondv
and aHmoom uoHam

 

.Eugu woo: 59% can: onona .

 

hobb: aha oHnooA noHum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.oocaHHQmm
mafia how :v::ond

 

 

 

 

 

Hag... 3H8 HHHHHmHHm HEHaoz HHHHHmHHm 3H8 Hag wnHHHofig :H 25 33H
H H A w H H H 920mm abun szonm H
.omducdsouoe new»

H mag you manopm vacapuumom

HSH 3H3 HHHHHHmmI mHfiHsoz HHHnmmm Baa HHQH .850 :MHHHSHH 8% .833
H Hm- .H—H H H H on o» 39H PH.» «005
.HHHHSHH Hagmoum HOHB mH

H3, 2:5 HHHHHHwHHm HfiHnoz HHHHHmHHm. 3H8 HHoH 032223 38 «5
H I H H .33» u 283.523 22.5
.mcdmmosm OHDIHonnu

HHoH 3H8 HHHEmHHm Hausoz HHHanHm oHHpo HHHHH .8.H 8188 8H
H H_ w H H H ha>on ma onovu one
.vnuun nagv vacasooon

. vHsb3_H Hoonudema

H»; 330 HH HgmHHm Hfifioz HHHHHmHHmH 3H3 Hams .HuHHaH» a HS 3
H H_ _H H H madam oh»: vandhm a HH

hump ovfisa H_hH+nwfiHm Hahpzoz HHpanHmH manna Huo>_ .uhosovnao op msoophdoo
H H_ _H . H . H aha onoom moHum
Hag 0mg .HwHuzmHHm EH82 mHmmmm 3H3 . Hams .85 v8: 8H 8.? 823
H ¢I|H H—----. H H H gunman can oHaoom uoHum

OdOH

.u.~

OwON

.~.n

OMOH

.30.:

00.:

60.3...» afloafifimv .350
53.3. .8qu 3.73%
9330: 05 30am hone

 

 

.mprqumcH Ho:
2” wchSLobvm .285.

.mwSEnHFHPH
0H5: paona mmoBH 30a
paw ca 3me .Hoa 4

m .oocwfldaa

H 1 .Ho 98.3 ngoflha
6H5 vouooHom 23: 38:»
H .852? on ma v56

He 5 a? «H 22%

 

 

.023: 0.35
no use «55 madm

.935 wcwmaonm H3
055. 2:3 0.58 H

 

.Mflflnfldfl 3..
chow» 2.3 an wfixfionm

hug

HASH

55>

53>

93>

 

 

 

 

 

3H3 HHHHHwHHm H.532 HHHHHmHHm 3H8 HHoH
3H5 @wame .fiHsoz HHHmmHHm 3H8 Hnobl
3H3 H HsmHHm £282 HHHHHmHHm 3H3 HHoH
IIIIIIHIHIIIIIIH lHnlllllHIIn-II. IIIIIII
3H8 HHHHHwHHm H.582 HHHHHwHHm 3H8 HHoH
3H5 HHHHHHHHH HEHHaz HHEmHHm 3H8 HHoH
3H3 HHHHHHHHm Hfipsoz HHHHHmHHm 3H5 HHSH
H H_ _H H ,
8.30 HHEHHHm -EHzoz HHEHHHm 3H8 . H3,
H H

6933

ugéfiov yoga
:33 .853 3.13m
Hugo: on... 32% hoes.

 

 

II .obnpushog
2.. $3393va .323.

.mmcHfiHHEa
28: “Bond 933 so:
new on 003% voom 4

.oocaHHHEu

H0 933 .HqHsoHtw
«Ha» uaHooHom

28: 3.2» H .852!)
omima as? He 5 3:8
oHnwhonHmsoo ad 935..

 

.ochcunvoa o
vac m5.» so:

.935 Mignon» .3
053. 25a wanna: H

 

.oHanHOHco #0: m.“
0.89» 05 an gmmonm

.H.N

.~.m

.m.~

OQOH

HMow

.n.n

 

 

 

.28 a .5» 3E aflmflwm H8 9% 3.233 330 Hump .23»
McNEOEHIIWE-Hgfid . u. H u I o . a n 352% ”H00 4 0H.“

.58.“ 30030 .3 .82.“ 38:0 on

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

$55. .3 53.23 035 33> 8.30 2.22% Hafioz- .3238 390 98>... “wads .Ho 3222. 03....
a USE a 53 50H H H H .— _H H H H a v5.“ .85.» :3 :0» .34...
_ _ 5590.33.»
330$»? Hon 3 83.... b.» 8.30 mafimflm afiusoz #238 390 20> 2.. 93% a5
23 Ho savanna 08H, H H H [H H H H Ho angina 05. 3:6
H ”£2323 fig 330 Hafiwfim Hanan—oz mafia 3.20 has .H .wmmmfifioa
3H3 s an HHHHom H H H . H H H H 2..an 30H «.58. .N.N
o H I
u
Jan _ 5.2a
23. 5. ocov 2:3 H :53 . on» 5 econ oban H :53.
0.33 ads. 5 93.323 gown m3» 5 gmonm
Ho: Ho vmopcoohoa . H? Ho mmflboonon
puma? fins a on 1 has 390 W338 $3.32 523%. 1390 HHHS, ti .3on sons a 8 3
5 «SE H 9:55 23 :H H H H H_ _H H H H HBHHH H 0.33% 0.5 :H .H.\.
32.55005
0.: moocdnHHEu H90> 330 thnwflHm 7.3902 .3333 topgo 53> 50mm and 3223qu
so mooflruom .3:an H H H H_ _H H H H no 30an Afloaom .m.m
.vHHwHHooxo HEMP: ovwsuol. magwmm «.3902 mvswmwm 3.30 FNMA; 33m 0.3
0.: undid—noun wfiuiam H H H H H H H H moHfiquoaH wfiufiam .N6

121

.Wmmm 98 33:9“ 98> 320 3238 afifiaz 223?. 320 fig .nomm 93 833a

9853.. 0358 . . . . l. . . . van—coon omhuno .m.m

 

 

 

 

 

 

.uxoono @938 E: 330 hpfiflm Hausoz 3238 330 98> .8323 mfifiuo

 

 

 

 

 

 

 

 

 

 

 

 

 

5 82?. 98> a n . .l, . . . 5 H832 to» $6
.305 3305.3 ‘ .823. 3205.5
H 90pm“ 0.35 H 90.83 0.35
as H 38m 0 5? he, 255 .2238 Hepscz » 9. mfim 330 is, 53 H $8» of 5...:
3.3333 Holbuomm an H . . . .I.IHIm.lllu . . . @033qu a as H .m.~

APPENDIX E

THE POST-TRANSACTION LETTER MESSAGE

88

APPENDIX E
TIIE J. L. HUDSON COMPANY DETROIT, MICHIGAN~48226.PHONE 223.5100

Dear Mrs. (Miss)

 

Thank you for purchasing your refrigerator here at Hudson's.
'We're naturally pleased that you chose Hudson's to help you

fill your needs.
I felt that you.would be interested in learning that our
customers' experience with the (name of brand refrigerator)

has been particularly favorable. Also, our service records

show it to be excellent in both performance and dependability.
Since we constantly seek to maintain our high standards in both
merchandise and service, please do not hesitate to contact me
if I can be of any further assistance.

Thanks again for your patronage.

Sincerely)

Appliance Department Manager

[Notes Each letter was individually typed for each persnn.]

122

APPENDIX F

THE POST-TRANSACTION TELEPHONE MESSAGE

E.

APPENDIX F

'0

Hello, Mrs.

 

Yes

My name is Shelby Hunt and I'm with the J. L. Hudson Co.
[No Pause]. I understand you just purchased a [Name of
Brand] refrigerator from us . . . is that correct?

Yes

The purpose of my call is to eXpress our sincere appre-
ciation for your purchase . . . we're naturally very
glad that you selected Hudson's to help fill your needs.
[PAUSE]

Was your "old" refrigerator a [Name of Brand] also?

No

Well, you might be interested to know that our customers'
experience with the [Name of Brand] unit has been partic-
ularly favorable. Our service records show it to be
exceptionally dependable with very little maintenance
required.

[PAUSE]

Thanks again for your patronage and please don't hesitate
to contact us if we can be of any further assistance.
[PAUSE]

Goodbye

[Note: The above message was delivered to the female head
of the household. If the female head of the household was
not at home, then a call-back was made at a later time.]

l23

APPENDIX G

STATISTICAL PROCEDURES

APPENDIX G

MannAWhitney U Test

The Mann4Whitney U statistic is a nonparametric test

to determine the prObability that two independent samples,

which may be of unequal size, come from identical popula—

tions [42:75-83].

Procedure

1.

Determine the value of X and Y. X = the number of
cases in the smaller group; Y = the number of cases
in the larger group.

Rank together the scores for both groups, assigning
the rank of l to the score which is algebraically
lowest. Ranks range from 1 to N = X + Y. Assign
tied observations the average of the tied ranks.

Determine the value of U by the formulae:

U=XY+)”2‘—5—fl)— --R1

or, equivalently:

The method for determining the significance of the
value of U depends on the size of Y:
a. If Y is 8 or less, the exact probability asso-

ciated with a value as small as the Observed

124

125

value of U is shown in [the table on (42:271-
273)]. If the observed value of U is not shown
in the table, it is U1 and should be transformed
to U by the formula:

U = XY - Ul
If Y is between 9 and 20, the significance of
any observed value of U may be determined by
reference to [table on (42:274-277)]. If the
observed value of U is larger than XY/2, it is
U1 and apply above formula for transformation.
If Y is larger than 20, the probability asso-
ciated with a value as extreme as the observed
value of U may be determined by computing the

value of Z by the formula:

 

 

Z =
fXY gx+Y+12
12

and testing this value by reference to [table on
(42:247)]. For a two tailed test, double the
value of p shown in that table. If the propor-
tion of ties is very large or if the obtained p
is very close to alpha, apply the connection for

ties, i.e., use formula:

5.

126

XY

2

 

XY N3—N

N(N-l) 12 ' T

If the observed value of U has an associated
probability equal to or less than alpha, reject

H0 in favor of H1.

127

The KruskalAWallis One Way Analysis
of Variance by Ranks
The Kruskal4Wallis one-way analysis of variance by
ranks is a test for determining whether "k" independent

samples have come from the same population [42:184-194].

Procedure
1. Rank all of the observations for the "k" groups in a
series, assigning ranks for l to N.
2. Determine the value of R (the sum of the ranks) for
each of the groups.
3. If there are not a large proportion of ties, compute

H by the following formula:

12 R:
— -
H N(N+l) :1 nj 3 (N + l)

4. The method for assessing the significance of the
observed value of H depends on the size of "k" and
the number of groups:

a. If k=3 and if n are less than or equal

1' “2' n3
to 5, then Table 0 in Siegal [42:282—283] may be
used to determine the associated probability
under Ho of an H as large as that observed.

b. In other cases, the significance of a value as

large as the observed value of H may be assessed

by reference to Table C in Siegal [42:249], with:

128

If the probability associated with the observed
value of H is equal to or less than the previously
set level of significance, Alpha, reject H0 in favor

of H1.

APPENDIX H

LIST OF SUBURBS INCLUDED IN THE SAMPLE

APPENDIX H

LIST OF SUBURBS INCLUDED IN THE SAMPLE

ALLEN PARK

BERKLEY

BEVERLY HILLS
BINGHAM FARMS
BIRMINGHAM
BLOOMFIELD HILLS
CENTERLINE

CLAWSON

DEARBORN

DETROIT

EAST DETROIT
ECORSE

FERNDALE

FRANKLIN

GARDEN CITY

GROSSE POINTE
GROSSE POINTE FARMS
GROSSE POINTE SHORES
GROSSE POINTE PARK
GROSSE POINTE WOODS
HAMTRAMCK

HARPER WOOD

HAZEL PARK

129

HIGHLAND PARK
HUNTINGTON WOODS
INKSTER

LATHROP VILLAGE
LINCOLN PARK
LIVONIA

MADISON HEIGHTS
MELVINDALE
NANKIN MILLS
OAK PARK
PLEASANT RIDGE
REDFORD TOWNSHIP
RIVER ROUGE
RIVERVIEW
ROSEVILLE

ROYAL OAK
SOUTHFIELD CITY
SOUTHGATE

ST. CLAIR SHORES
WAYNE

WYANDOTTE

WESTLAND

BIBLIOGRAPHY

BIBLIOGRAPHY

Adams, J. S. "Reduction of Cognitive Dissonance by
Seeking Consonant Information," Journal of Abnormal
Social Psychology, LXII (1961).

Adams, J. S., and Rosenbaum,‘W. B. "The Relationship of
Worker Productivity to Cognitive Dissonance,"
Journal of Applied Psychology, XLVI (1962).

Allyn, June, and Festinger, L. "The Effectiveness of
Unanticipated Persuasive Communications," Journal
of Abnormal and Social Psychology, LXII (1961).

Aronson, E. "The Effect of Effort on the Attractiveness
of Rewarded and Unrewarded Stimuli," Journal of
Abnormal and Social Psychology, LXIII (1964).

Aronson, E., and Mills, J. "The Severity of Initiation
on Liking for a Group," Journal of Abnormal and
Social Psychology, LIX (1959).

Banks, Seymour. Exgerimentation in Marketing. New York:
McGraw Hill Book Company, 1965.

Beijk, J. "Een onderjoek naar de reductie van het
subjectieve gevoel van vermoeidheid als gevolg van
het dissontieproces," Nederlands Tydschrift Voor De
Psychologie en haar Grensgebieden, XXII (1), (1967).

Bell, G. D. "The Automobile Buyer After the Purchase,"
Journal of Marketing, July, 1967.

Bramel, D. "A Dissonance Theory Approach to Defensive
Projection," Journal of Abnormal Social Psychology,
LXIV (1962).

Brehm, J. W. "Postdecision Changes in the Desirability
of Alternatives," Journal of Abnormal Social Psychol-
ng, LII (1956).

Brehm, J. W., and Cohen, A. R. Explorations in Cognitive
Dissonance. New York: John Wiley & Sons, 1962.

 

130

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

131

Brock, T. C. "Cognitive Restructuring and Attitude
Change," Journal of Abnormal and Social Psychology,
LXIV (1962).

Cardozo, R. N. "An EXperimental Study of Consumer
Effort, Expectation and Satisfaction." Unpublished
Ph.D. dissertation, Department of Marketing, Univer-
sity of Minnesota.

Chapanis, N. P., and Chapanis, A. "Cognitive Disso-
nance: Five Years Later," Psychological Bulletin,
January, 1964.

Cronkhite, G. L. "Autonomic Correlates of Dissonance
and Attitude Change," Speech Monographs, XXXIII
(1961).

Day, Robert C., and Hamblin, Robert L. "The Relation
Between Group Member's Self Confidence and Punitive
Styles of Persuasion," The American Journal of
Sociology, LXIX (March, 1964).

Dieker, Richard J. "The Effect of Freedom of Choice on
Group Attractiveness." Unpublished Ph.D. disserta-
tion, Michigan State University.

Ehrlich, D., Guttman, I., Schonbach, P., and Mills, J.
"PostéDecision Exposure to Relevant Information,"
Journal of Abnormal Social Psychology, LIV (1957).

Engel, James F. "Are Automobile Purchasers Dissonant
Consumers?" Journal of Marketing, April, 1963.

Engel, J. F., and Light, L. "The Role of Psychological
Commitment in Consumer Behavior: An Evaluation of
the Theory of Cognitive Dissonance," Applications
of the Sciences in Marketing Management. Edited by
Frank M. Bass, Charles W. King, and Edgar A.
Pessemier. New York: John Wiley and Sons, 1968.

Feather, N. T. "Cigarette Smoking and Lung Cancer: A
Study of Cognitive Dissonance," Journal of Abnormal
Social Psychology, XIV (1962).

Feldman, Shel. Cognitive Consistency. New York:
Academic Press, 1966.

Festinger, Leon. A Theory of Cognitive Dissonance.
Stanford, California: Stanford University Press,
1957.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

132

Festinger, Leon. Conflict, Decision and Dissonance.
Stanford, California: Stanford University Press,
1964.

Fett, John H. "The Role of Volition in Attitude Change
Resulting from Hearing an Attacking Speech."
Unpublished Ph.D. dissertation, The University of
Wisconsin.

Fillenbaum, S. "Dogmatism and Individual Differences
in Reduction of Dissonance," Psychologigal Reports,
XIV (February, 1964).

Fisk, G. "A Conceptual Model for Studying Customer
Image," Journal of Retailing. XXXVII, No. 4
(Winter, 1961—62).

Green, P. E., and Tull, D. S. Research For Marketing
Decisions. Englewood Cliffs, New Jersey: Prentice
Hall, Inc., 1966.

Holloway, Robert J. "An Experiment on Consumer Disso-
nance," Journal of Marketing, XXXI (January, 1967).

Kassarjian, Harold H., and Cohen, Joel B. "Cognitive
Dissonance and Consumer Behavior: Reactions to the
Surgeon General's Report on Smoking and Health,"
California Management Review, Fall, 1965.

Kotler, Philip. Marketing Management: Analysis,
Planning and Control, Englewood Cliffs, New Jersey:
Prentice Hall, Inc., 1967.

LoSciato, L. A. "The Influence of Product Preferences
Upon Dissonance Reduction." Unpublished Ph.D.
dissertation, Purdue University.

Lowin, Aaron. "Information Selectivity as a Function of
Message Agreement and Ease of Message Refutation."
Unpublished Ph.D. dissertation, The University of
North Carolina at Chapel Hill, 1965.

Mills, J., Aronson, E., and Robinson, J. "Selectivity
in Exposure to Information," Journal of Abnormal
Social Psychology, LIX (1959).

Mindak. W. W. "Fitting the Semantic Differential to
Marketing Problems," Journal of Marketing, XXV
(April, 1961).

36.

37.

38.

41.
42.

43.

(44 .

45.

46.

47.

133

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

Powell, F. A. "Source Credibility, Dissonance Theory,
and Attitude Change." Unpublished Ph.D. disserta-
tion, Department of Communications, Michigan State
University.

Raven, B. H., and Fishbein, M. "Acceptance of Punish-
ment and Change in Belief," Journal of Abnormal and
Social Psychology, LXIII (1961).

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

Rosen, S. "Post-Decision Affinity for Incompatible
Information," Journal of Abnormal Social Psychology,
LXIII (1961).

Sears, D. 0., and Freedman, J. L. "Commitment, Informa-
tion Utility and Selective Exposure," USN Tech. Rep.
(ONR) Nonr-233(54) NR 171-350, No. 12 (August, 1963).

Siegel, Sidney. Nonparametpic Statistics for the
Behavioral Sciences. New York: McGraw Hill Book
Company, Inc., 1956.

Silverman, Irwin. "In Defense of Dissonance Theory,"
Psychological Bulletin, LXII (September, 1964).

 

Smith, E. E. "The Power of Dissonance Techniques to
Change Attitudes," Public Opinion Quarterly, XXV
(1961).

Spring, F. E. "Determinants of Modes of Dissonance
Reduction." Unpublished Ph.D. dissertation,
Department of Psychology, University of North
Carolina.

Stephenson, P. R. "Post Purchase Economic Attitude
Change: A Dissonance Analysis." Unpublished Ph.D.
dissertation, Department of Economics, Ohio State
University.

Straita, Bruce C. "The Pursuit of the Dissonant
Consumer," Journal of Marketing, July, 1964.

134

48. Wyckham, R. G. "Aggregate Department Store Images:
Social and Experimental Factors." Unpublished
Ph.D. dissertation, Michigan State University,
Department of Marketing and Transportation
Administration.

49. Yaryan, Ruby, and Festinger, L. "Preparatory Action
and Belief in the Probable Occurrence of Future
Events," Journal of Abnormal Social Psychology,

LXIII (1961).

50. Zaltman, Gerald. Marketing: Contributions from the
Behavioral Sciences. New York: Harcourt Brace

and World, Inc., 1965.

HICHIGAN STRTE UNIV.

3ll |1l| WIIHI Wll 1mm ll llll Illll 11] ll lllll illHl