. , 1,

. .

igéfk‘.
2:-

.
x. «Y;
A.“
”4.. .
, .

$7. . ,.

.. {k , WI)
“31’ . #21394
1 3% “£351

.'
553k 2» ‘
‘zzwﬁiv’a, if?"
3; ; 3,
| J

.
"ca-«‘3

.-.. 4 -r.

‘ ‘-.‘."

-‘ ‘79:;

'3“.— — z a‘ t'n'v‘.‘

2‘..

4

- .~
1. ‘-

 

L. . '
r; {‘2‘ 32’" I 5" >45 ~ 1
YE”? , ’1 - ‘1" ‘. -
as 33“» ""

35a

xmgn

.4-

I u - 'i
,?-‘d'*S‘-’n
3“...”

‘ 9:115,"
’ l

 

was lllllllililHllHllWlilillllllllllilllHHHHIHIUW '

3 1293 01018 4327

 

This is to certify that the

dissertation entitled

The Differential Effects of Value-Relevant
and Outcome-Relevant Involvement on Evoke Set Size

presented by

Richard Lawrence Divine

has been accepted towards fulﬁllment
of the requirements for

Doctor of Philosophy degreein Business Administration

 

 

”Major pré/ 5501' r .
Date%W./Zz /g7(/

 

MS U is an Afﬁrmative Action/Eq ual Opportunity Institution 0-12771

 

 

 

LIBRARY
Mlchlgan State
H Unlverslty

 

 

 

PLACE It RETURN BOXto monthl- chockomm your record.

TO AVOID FINES mum on or More duo duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU loAnNﬂrmatlvo Action/EN Opponunltylnotltwon
W

”39.1

 

 

THE DIFFERENTIAL EFFECTS OF VALUE-RELEVANT AND
OUTCOME-RELEVANT INVOLVEMENT ON EVOKED SET SIZE

BY

Richard Lawrence Divine

A DISSERTATION
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Marketing and Logistics

1994

ABSTRACT

THE DIFFERENTIAL EFFECTS OF VALUE-RELEVANT AND OUTCOME
RELEVANT INVOLVEMENT ON EVOKED SET SIZE

By

Richard L. Divine

Currently there has been strong debate regarding the issue of whether outcome-
relevant and value-relevant involvement produce differential effects, and hence merit
recognition as separate constructs. This study sought to address this issue in the
context of the effects these two involvement types have on evoked set size. Previous
research had only examined outcome-relevant involvement and had consistently found
a negative relationship. This research, citing motivational theory, hypothesized that
value-relevant involvement would have a positive effect on evoked set size. The
primary basis for this hypothesizes is that value-relevant involvement represents
intrinsic motivation and research suggests that intrinsic motivation enhances task
enjoyment and enthusiasm. As a result value-relevant involved consumers enjoy the
shopping task more and thus instead of being motivated to simplify the task by
limiting the size of evoked sets, they might actually seek to complicate it by

considering more brands than necessary.

For the study a theoretical path model of the evoked set formation process was
devised, and measurement scales for each variable in this process were developed. In
addition to the two involvement types and evoked set size, ﬁve mediating variables
were also measured. Four of these; product knowledge, awareness set size, lattitudes
of acceptance and number of salient attributes were identiﬁed from previous research

on evoked sets. The ﬁfth, shopping enthusiasm, was a new variable that had not been

studied before, and was introduced in this study to allow the effects of affect toward

the shopping process to be assessed.

In terms of results, sets of hypotheses regarding 21 different bivariate
relationships were tested, and 14 received statistically signiﬁcant support. While most
of the hypothesized relationships between the two involvement types and the
mediating variables were supported, shopping enthusiasm was the only one of the four
mediating variables hypothesized to have a direct effect on evoked set size, that was
actually found to have a signiﬁcant relationship. Consequently, the results of the path

analysis indicated that the model did not provide a good ﬁt to the data.

Finally, as hypothesized, a signiﬁcant positive relationship between value-
relevant involvement and evoked set size was found, however the relationship
between outcome-relevant involvement and evoked set size was insigniﬁcant. This
still provides some evidence of differential effects since the signiﬁcant positive
relationship contradicts the accepted view of a negative involvement/evoked set size
relationship since this view was based on research ﬁndings in which only outcome-

relevant involvement was operationalized.

Copyright by
RICHARD LAWRENCE DIVINE
1 994

Dedicated
to the memory of
BARBARA R. DIVINE

ACKNOWLEDGMENTS

I would like to express my deep appreciation to Dr. Thomas J. Page, Jr.
chairman of the dissertation committee for the guidance, and encouragement, he gave
me during this project and the mentoring he provided throughout my doctoral
education. In addition I would like to thank the other members of the dissertation
committee: Dr. Roger J. Calantone, Dr. Richard A. Spreng, and Dr. Stan Kaplowitz

for their cooperation, effort and valuable contributions.

1 would also like to express my heartfelt gratitude to Colleen McIntyre for her
strong moral support. Finally I would like to gratefully acknowledge the support of
my father Dr. Robert A. Divine, whose exemplary career in academia inspired me to

follow in his footsteps.

vi

II.

III.

IV.

TABLE OF CONTENTS

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

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

LITERATURE REVIEW .....................................................................

Review Of Involvement Types ..............................................................
Social Psychology Literature .........................................................
Marketing Literature ......................................................................
Synthesis Of Involvement Types ....................................................

Evoked Sets ..........................................................................................

The Effects Of Involvement On Evoked Set Size ....................................

Hypothesized Effects Of Value-Relevant Involvement On
Evoked Set Size .............................................................................

Conclusion ............................................................................................

HYPOTHESES ....................................................................................

METHOD .............................................................................................

Overview ..............................................................................................
Subjects ................................................................................................
Experimental Design .............................................................................
Product Selection ..................................................................................
Test Procedure ......................................................................................
Pretest Questionnaire .....................................................................
Outcome-Relevant Involvement Manipulation ...............................
Product Evaluation Task Questionnaire ..........................................
Post Test ........................................................................................
Measurement Of Constructs ..................................................................
Value-Relevant Involvement ..........................................................

Page

xi

5

5
5
8
16
27
29

4]
49

50
69

69
69
7O
71
72
72
72
75
76
77
77

Page

Outcome-Relevant Involvement ..................................................... 78
Awareness Set Size ........................................................................ 79
Number of Salient Evaluative Attributes ........................................ 79
Lattitudes of Acceptance ................................................................ 80
Evoked Set Size ............................................................................. 81
Shopping Enthusiasm ..................................................................... 81
Product Knowledge ........................................................................ 82

V. DATA ANALYSIS AND FINDINGS .................................................. 84
Introduction .......................................................................................... 84
Reliability Analysis ............................................................................... 84
Development of Final Measures ..................................................... 84
Value-Relevant Involvement .......................................................... 86
Outcome-Relevant Involvement ..................................................... 94
Number of Salient Attributes ......................................................... 96
Product Knowledge ........................................................................ 97
Awareness Set Size ........................................................................ 97
Shopping Enthusiasm ......................... ' ............................................ 98
Lattitudes of Acceptance ................................................................ 98
Evoked Set Size ............................................................................. 99
Summary ....................................................................................... 99
Tests of the Hypotheses of the Study .................................................... 99
Summary of Results .............................................................................. 132
VI. SUMMARY AND DISCUSSION ........................................................ 134
Research Summary ............................................................................... 134
Research Objectives and Conclusions ................................................... 136
Research Contributions ......................................................................... 143
Managerial Implications ....................................................................... 146
Research Limitations ............................................................................ 149
Suggestions for Future Research ........................................................... 150

APPENDIX A: Instructions for High Outcome-Relevant
Involvement Condition .................................................................. 152

APPENDIX B: Instructions for Low Outcome-Relevant

Involvement Condition .................................................................. 154
APPENDIX C: Pretest Questionnaire .................................................. 155
APPENDIX D: Product and Attribute Information Booklet ...... . .......... 161

viii

Page

APPENDIX E: Evaluation Task Questionnaire .................................... 168
APPENDIX F: Post-Test Questionnaire ............................................... 177
BIBLIOGRAPHY ................................................................................. 180

LIST OF FIGURES

Page
Path Model of the Effects of Involvement on Evoked Set Size:

Hypothesized Relationships ........................................................... 51
Path Models of the Effects of Involvement on Evoked Set Size:

Current Literature .......................................................................... 52
Results of Path Analysis: Newly Hypothesized Model ......................... 130
Results of Path Analysis: Current Literature Model ............................. 131
Results of Path Analysis: Revised Newly Hypothesized Model ............ 140

1.1

1.2

1.3

1.4

1.5

5.1

5.2

5.3

5.4

LIST OF TABLES

Involvement Types Consistent with Value-Relevant Involvement .........
Involvement Types Consistent With Outcome-Relevant Involvement
Antecedents of Value-Relevant Involvement ........................................
Antecedents of Outcome-Relevant Involvement ....................................
Review of Evoked Set Studies ..............................................................
Item Analysis of Final Scales ................................................................
Pearson Correlations of All Variables ...................................................
Results of Hypothesis Tests: T-Tests ....................................................

Tests For Differences Between Correlations .........................................

xi

Page
19

21
23
24
30
87
101
102

105

CHAPTER ONE
INTRODUCTION

Involvement has emerged as one of the most prominent concepts in consumer
research (Sherrell and Shimp 1982). The primary reason for its importance is
because it is the main determinant of how much decision making effort an individual
will exert when making a purchase decision (Assael 1984). Consequently almost
every stage of the consumer buying process is affected by involvement. How many
stores one shops, the number of brands one considers, the amount of information
one collects and how deeply one processes purchase related information are all
greatly inﬂuenced by how involved a consumer is with the purchase (Howard and

Sheth 1969, Katonna and Mueller 1954).

One of the most critical issues currently being debated by researchers in this
area is whether involvement is actually a single concept or instead represents a
conglomeration of related, but conceptually distinct constructs (Johnson and Eagly
1989, 1990, and Petty and Cacioppo 1990). In the marketing literature,
involvement has traditionally been treated as a single unidimensional construct that
refers to the level of interest or concern a person has toward a purchase, product, or
advertisement (Day 1970, Mitchell 1979). This traditional viewpoint has been
challenged by two groups of researchers. One group has focused on the issue of
whether involvement is truly unidimensional. These researchers have analyzed
empirical data and have uncovered anywhere from two to ﬁve separate dimensions
or components of the construct (Kapferer and Laurent 1985, Lastovicka and Gardner
1979, Zaichkowsky 1985). Because of these ﬁndings, serious questions must be

raised about the validity of theories based on research where involvement was

operationalized as a unidimensional construct.

Perhaps because this evidence makes it fairly clear that involvement is not
unidimensional, the second group of researchers have pushed the issue one step
further by not just recognizing multiple dimensions, but actually proposing the
existence of several distinct types of involvement. These different involvement
types are based on the notion that the interest level one has toward a purchase can be
attributed to different sources of concern (Houston and Rothschild 1978, Park and

Mittal 1985).

What has resulted then from these alternative viewpoints is a debate over the
true conceptual nature of involvement. This debate has centered around one
fundamental question: if involvement is not unidimensional, then is it a single
construct with multiple dimensions or is it actually several separate constructs?
Resolving this debate is important because these alternative conceptual izations imply
different operationalizations of the construct. In particular the multiple dimension
view sees different involvement types as simply different antecedents to an overall
measure or level of involvement. In contrast, the multiple construct view sees these
different involvement types as separate constructs. The main implication of this
perspective is that it is inappropriate to pool the different involvement types together
‘ into an overall measure because it is believed that these types not only have different

antecedents, but also have different consequences as well.

Resolving this controversy surrounding the true conceptual nature of

involvement is critical given its prominent role in the established theories of

consumer decision making. In order to validate the need to recognize multiple types
of involvement, it must be shown that these different types are not only elicited by
different antecedents, but have different behavioral consequences as well. If these
different involvement types can be shown to have differential relationships with
other variables in the consumer decision making process, then it would be hard to
justify continuing to pool or integrate them together as a single construct. This will

be the primary focus of the present study.

While many different typologies have been proposed by those researchers
subscribing to a multiple construct perspective in the marketing literature, the
position taken here is all of these types can be condensed into the two basic forms of
involvement delineated by Johnson and Eagly (1989) in their meta—analysis of the
construct. The ﬁrst form, value—relevant involvement, is involvement induced by a
perceived linkage between a psychological object and one's important personal
values. Value-relevant involvement would include those conceptualizations that are
theoretically consistent with the original version of the concept, ego involvement
(Johnson and Eagly 1989). The second form, outcome-relevant involvement, is
involvement induced by a perception that a psychological object might have an effect
on one's ability to obtain desirable outcomes. Outcome-relevant involvement would
include those conceptualizations of the construct that are not consistent with value-

relevant or ego involvement.

The overall purpose of this study is to examine the issue of whether outcome-
relevant and value relevant involvement differentially effect consumers in terms of
how they form evoked sets. An evoked set is the subset of the brands on the market

a particular consumer will consider when making a purchase (Campbell 1969,

Howard and Sheth 1969). Currently the accepted view in the marketing literature is
that involvement has a negative effect on the size of one's evoked set (Gensch and
Javalgi 1987, Harrell 1986, Rothschild and Houston 1977). However, the
theoretical and empirical research that supports this ﬁnding has typically only dealt
with outcome-relevant forms of involvement (Belonax and Javalgi 1989, Houston
and Rothschild 1978). This research will attempt to make the case, both
conceptually and empirically, that outcome-relevant and value-relevant involvement
have the opposite effect on evoked set size. Speciﬁcally it is believed that while
outcome-relevant involvement has a negative relationship with evoked set size,
value-relevant involvement is actually positively related. If this hypothesis is
supported, it would demonstrate that the two involvement types do indeed have
different consequences. Such a ﬁnding would not only extend the literature on the
effects of involvement on evoked set formation, but in addition, give greater

credence for the need to recognize both types of involvement as separate constructs.

CHAPTER TWO
LITERATURE REVIEW

This chapter is divided into four main sections: the ﬁrst will
review the literature on involvement types, the second will synthesize
this literature and identify two primary underlying types of
involvement, the third will review the literature on the effects of
involvement on evoked set size, and the fourth will discuss how the
effects of value-relevant involvement on evoked set size might differ

from what is in the literature.

REVIEW OF INVOLVEMENT TYPES
Social Psychology Literature

All the various manifestations of involvement that have
appeared in the literature can be traced back to the initial concept of
ego involvement, which originated in the ﬁeld of social psychology.
The works of Allport (1943) and Sherif and Cantril (1947) are
generally credited with providing the seminal treatments of the
subject. Allport (1943) deﬁned ego involvement as "a condition of
total participation of the self“ (p.459) and described the ego involved
person as being "seriously committed to a task" (p. 459). Sherif and
Cantril (1947) deﬁned it as the extent to which an object or idea
touches upon one's ego or self concept and it was later equated with

the ”intrinsic importance of an object” (Sherif and Hovland I961).

Subsequently Zimbardo (1960) developed a different version
of involvement which was labeled response involvement. This was
deﬁned as ”an individual's concern with the consequences of his
response" (p.87). While deﬁnitionally this conceptualization is rather
broad, his operationalizations of response involvement dealt
exclusively with concern about the social acceptability of one's
position on an issue (Johnson and Eagly 1989). Involvement was
heightened in Zimbardo's experiments by forcing subjects to express
publicly their positions on certain issues that were otherwise
uninvolving. As a result, response involvement has subsequently
been associated with situations where a person has little inherent
interest or concern in the issue itself but is concerned about the
impression their opinion on the issue will have on others (Johnson and

Eagly 1989).

Petty and Cacioppo (1979) offered another variation of the
concept which they labeled issue involvement. This was deﬁned as
”the extent to which the attitudinal issue under consideration is of
personal importance” (p. 1915). While Petty and Cacioppo considered
issue involvement to be the same construct as ego involvement, it is
in fact a much broader conceptualization. Ego involvement only
occurs when an issue is intrinsically important or affects one's self
concept (Sherif and Hovland 1961). Issue involvement on the other
hand would include all issues considered important, regardless of

whether involvement was evoked by an intrinsic interest in the subject

matter or by utilitarian concerns about how the issue will affect one's

ability to obtain desirable outcomes in the external environment.

Johnson and Eagly (1989) performed a meta analysis on the
effect of involvement on persuasion and found different results
depending upon the type of involvement that was operationalized. In
ego involvement studies, involvement had a negative effect on
persuasion (signiﬁcant attitude change was obtained for low
involvement receivers, but not for those with high involvement),
while in studies that manipulated non-ego forms of involvement, a
signiﬁcant interaction between involvement and argument strength
was found (signiﬁcant attitude change for high involvement receivers
occurred when strong arguments were used). Johnson and Eagly felt
these differences in empirical results warranted a reclassiﬁcation of
Petty and Cacioppo's issue involvement into ego and non-ego forms
of involvement. Johnson and Eagly's (1989) value relevant
involvement is essentially a relabeling of ego involvement, and is said
to occur when an issue activates one's enduring values. Their other
type was labeled outcome-relevant involvement and is based on the
operationalizations of Petty and Cacioppo. This is said to occur when
individuals believe an issue will have a signiﬁcant effect on their

ability to obtain desirable outcomes.

Petty and Cacioppo (1989) strongly challenged Johnson and
Eagly's assertion that issue involvement actually represents two

conceptually distinct types of involvement (value and outcome

relevant). They contend for the sake of parsimony the two should
remain integrated because in both types what is critical is that the
issue is perceived as being important to the self, regardless of whether
this importance is due to a linkage with one's values or one's goals.
While they acknowledge that value-relevant issues are generally more
important to a person than outcome-relevant issues, this does not
mean that they are operating on different theoretical dimensions.
Rather they assert that any differences observed between the two are
ones of magnitude or position along the same personal importance
continuum. In terms of the discrepancy in empirical ﬁndings they
point out that value-relevant involvement, as operationalized in
previous studies, was confounded with a host of other variables
(knowledge, commitment, conﬁdence) that all acted to make an ego

involved person more resistant to counterattitudinal messages.

Marketing Literature

While the marketing literature does not feature the same kind
of formal debate over the existence of multiple types of involvement
as appears in social psychology, it is readily acknowledged that there
is little agreement among marketers about what exactly involvement is
(e.g. Gensch and Javalgi 1987, Muncy and Hunt 1984, Stone 1984,
Tybejee 1979). This literature contains such a wide variety of
approaches toward the concept that it has been frequently criticized
for not developing a " generally accepted" view on how involvement
should be conceptualized and deﬁned (e. g. Houston and Rothschild
1978, Muncy and Hunt 1984, Park and Mittal 1985, Stone 1984).

After analyzing this diverse and unstructured literature, it is believed
that three distinct perspectives of involvement can be identiﬁed. The
traditional viewpoint sees involvement as a simple unidimensional
construct and emphasizes the differences between consumers with
high and low levels. The multiple dimension perspective sees a single
underlying type of involvement but acknowledges that it is
characterized by much greater complexity than is depicted in the
traditional view. This complexity is reﬂected by the identiﬁcation of
multiple dimensions or components within the involvement construct
and the recognition that there are several conceptually distinct
categories of antecedents which can elicit it. The multiple type
perspective sees several conceptually distinct types of involvement
and emphasizes the characteristics that differentiate them. In relation
to the social psychology debate over involvement types, the ﬁrst two
perspectives share with Petty and Cacioppo (1990) the belief that
involvement is a single uniﬁed construct, while the third expresses a
theoretical position consistent with Johnson and Eagly's multiple

involvement typology.

The traditional conceptualization of involvement as a single
unidimensional construct can be traced back to Krugman's (1965)
seminal work on the subject. Krugman's characterization of a low
involvement consumer profoundly inﬂuenced subsequent marketing
thought. According to Lastovicka and Gardner (1977), his research
forced consumer researchers to realize that the high involvement

attitudinal theories they had borrowed from social psychology were

10

not applicable in explaining the mundane and trivia] decision making
that seems to typify consumer behavior. As a result, the primary
focus of researchers taking this perspective has been in the
development of theories that explain the process of low involvement
decision making (e. g. DeBruicker 1972, Ray 1973, Robertson 1976).
Because this research tended to focus on uninvolved consumers, the
issue of whether there might be multiple types of involvement (and
therefore multiple types of high involvement consumers) was
generally not addressed. As a result, the broad unidimensional view
of involvement that appeared in this research was not speciﬁcally
advocated over others, but rather was just implicitly assumed to be a

valid representation of the concept.

The second perspective is the multiple dimension view of
involvement. This arose out of attempts by involvement researchers
to fully explicate what exactly involvement is so it could be
operational ized in a more consistent manner across research studies.
Problems with the concept developed because there has never been a
universally accepted deﬁnition of involvement, even within social
psychology, and because the concept has been used in such a wide
variety of contexts (e. g. attention to advertising, information
processing, learning, purchasing behavior). As a result many
researchers have decried the confusing number of deﬁnitions and
operationalizations of involvement which have resulted (Antil 1984,
Cohen 1983, Houston and Rothschild 1978, Muncy and Hunt 1984,
Sherrell and Shimp 1982, Stone 1984, Zaichkowsky 1985). Many of

11

those subscribing to this viewpoint were interested in actually
measuring involvement rather than just manipulating it and thus
sought to develop deﬁnitions of the concept that could capture its full
meaning and be valid across many contexts (Antil 1984, Sherrell and
Shimp 1982, Zaichkowsky 1985). These researchers focused their
attention on discerning what the underlying dimensions of
involvement were rather than just examining its consequences (e.g.
more attention, deeper processing, more deliberation). Once the
dimensions or components of involvement were identiﬁed, the
concept could then be properly measured through the development of

scales to capture each of its dimensions.

Lastovicka and Gardner (1972) analyzed empirical data using
multidimensional scaling and conﬁrmatory factor analysis and
uncovered two primary components of involvement: commitment
(brand loyalty) and normative importance (connections to one's
values). Zaichkowsky (1985) examined the literature and identiﬁed
three areas that affect one's involvement with an object: personal
factors (goals and values of the person that relate to the object),
physical factors (characteristics of the object that increase interest)
situational factors (temporary circumstances that increase interest in
the object). Laurent and Kapferer (1985) used factor analysis to
identify ﬁve facets of involvement: interest in the product, pleasure
derived from the product, sign value of the product, perceived
negative consequences of a mispurchase, and the probability of a

mispurchase. Celsi and Olson (1988) identiﬁed two antecedents of

12

involvement: intrinsic sources of personal relevance and situational

sources of personal relevance.

While there are some differences in the ﬁndings of these
researchers, they do nevertheless provide rather convincing evidence
that involvement is multidimensional and thus raise serious questions
about the validity of the traditional perspective. This second
perspective explicitly recognizes that because involvement has
multiple dimensions, two people who are equally involved with a
purchase could be quite different in terms of the source of their

interest or concern.

The ﬁnal perspective sees involvement not as a single concept
but rather as a family of related but conceptually distinct types. This
multiple type perspective developed as an attempt by researchers to
impose some structure on the concept given the wide diversity in
deﬁnitions and approaches that have been taken (Houston and
Rothschild 1978, Muncy and Hunt 1982). The creation of such a
structure allows different conceptualizations of involvement to be
classiﬁed into their appropriate type instead of continuing to lump

them all together as a single construct.

Perhaps the most inﬂuential of these involvement typologies
was created by Houston and Rothschild (1978). They recognized that
involvement may result as a consequence of both the purchase

circumstances (situational involvement) as well as the idiosyncratic

13

concerns and interests of the individual (enduring involvement).
Situational involvement is external to the consumer and refers to the
ability of the situation to elicit from consumers, concern for their
behavior. This is said to occur when consumers perceive adverse
consequences will result if their behavior in a situation is sub-optimal.
This type of involvement is the one most frequently addressed in
marketing studies and thus basically reﬂects how the concept is
viewed in the traditional perspective. It is determined jointly by
product characteristics that may raise concern (e. g. perceived
heterogeneity of price, complexity, quality among purchase
alternatives) as well as the individual's concern about the social
acceptability of the purchase. It is considered to be extrinsically
oriented because the involvement is induced by the situation the
person is in (i.e., the need to make a purchase that has some level of
perceived risk) and not by an intrinsic interest in the product (Arora

1982, Muncy and Hunt 1984).

Houston and Rothschild‘s second type, enduring involvement,
is internal to the consumer, and refers to the strength of the pre-
existing relationship between the individual and the product.
Enduring involvement is closely related to value-relevant involvement
and is based on two elements, one is the person's previous
experiences with the product or purchase situation and the second is
linkage between the product and the individual's value system. The
primary distinguishing feature of enduring involvement is .that it is

elicited by intrinsic interest in the product and not by situational

14

concerns regarding the product's purchase. Thus, unlike situational
involvement, it tends to be present even during those times in which

the product is not being considered for purchase.

Park and Mittal (1985) provide an alternative framework that
recognizes two other forms of involvement: cognitive and affective.
In their typology, involvement is categorized based on the nature of
the underlying motive for one's concerns. Cognitive involvement is
motivated by the utilitarian motive (concern about the cost/beneﬁts of
the purchase) while affective involvement is motivated by the value—
expressive motive (concern about the effect of the purchase on one's
self-concept image). This distinction is important because Park and
Mittal believe that an overemphasis on cognitive involvement in past
research has resulted in the misclassiﬁcation of many consumers with
high affective involvement as being uninvolved, and has led to the
widely held but possibly overstated view that consumers are
uninvolved with most of their purchase decisions (Hupfer and

Gardner 1971, Kassarjian 1978).

Product vs. purchase involvement (Hawkins, Best and Coney
1983) and product vs. task involvement (Clarke and Belk 1979) are
two other typologies appearing in the marketing literature. These two
are closely related to the Houston and Rothschild framework and
differentiate between one's involvement with a particular product and
one's involvement with the task of purchasing that product. They

recognize people may be highly involved in the purchase of a product

15

that is not involving to them (e. g., a dishwasher) or may be
uninvolved in the purchase of a product that is very involving to them

(e. g., the repurchase of their favorite brand of cigarettes).

In terms of evaluating the three perspectives of involvement in
the marketing literature, it seems clear the traditional unidimensional
perspective has little to offer. It is contradicted both by the empirical
ﬁndings of the multiple dimension perspective (which provided
evidence that involvement is not unidimensional) and the qualitative
distinctions emphasized in the multiple type perspective. The
question of whether involvement should be considered as several
separate constructs or rather as a single multidimensional one is
conceptually equivalent to the Johnson and Eagly/Petty and Cacioppo

debate in social psychology, and has likewise yet to be resolved.

The multiple type conceptualization of involvement differs
from the multiple dimension perspective in that the former recognizes
that the nature of the involvement elicited by different types of
antecedents may be conceptually distinct. As a result these different
types of involvement may produce disparate consequences. Thus
while the multiple dimension view established that involvement may
have different causes, the multiple type perspective takes it a step
farther by recognizing that the involvement elicited by different

causes may produce differential effects.

16

Since it is more parsimonious to conceptualize involvement as
a single construct, it is incumbent upon those advocating the multiple
type perspective to clearly demonstrate that several conceptually
distinct types of involvement do in fact exist. It is believed that the.
clearest means of resolving this issue is to see if the different types of
involvement identiﬁed in the literature actually generate conceptually
different types of consumer responses. In other words if it can be
shown that the consequences of one type of involvement are different
from those of another, then it would seem to invalidate the viewpoint
that these involvement types should be pooled together as a single
variable. If however the consequences are the same or similar, then it
would be difﬁcult to justify the additional complexity of the multiple

involvement perspective.

Synthesis of Involvement Types

Because the concept of involvement has spawned such a
diverse and extensive literature, it is not surprising that there really is
no single universally accepted framework for delineating involvement
into types. As has been detailed previously, numerous typologies
have been suggested by researchers from within and across different
disciplines, but a consensus framework has yet to emerge. As a
result, before the validity of a multiple involvement perspective can
be empirically investigated, a synthesis of the literature is required in
order to reveal the fundamental types of involvement that underlie the
entire body of work on the concept. The goal of such a synthesis is

not to add yet another new perspective on involvement types to the

17

literature, but rather to facilitate empirical testing by integrating the

existing typologies into a more cohesive and organized structure.

Since involvement is considered by many researchers to be a
motivational state (Celsi and Olson 1988, Johnson and Eagly 1989,
Park and Mittal 1985, Petty and Cacioppo 1986), then a valid
criterion that can be used for differentiating and integrating
involvement conceptualizations is the type of motivational orientation
they represent. According to Reeve (1992), motivation is deﬁned as
the antecedent conditions that energize and direct behavior. Behavior
can be energized and directed through two primary types of
motivation: intrinsic and extrinsic. lntrinsicly motivated behavior is
deﬁned as "behavior done solely for the interest and enjoyment
inherent in performing a given activity (p. 141). " Extrinsically
motivated behavior is that which is performed in order to obtain some
reward or avoid some punishment (Reeve 1992). Thus using
motivational orientation as a classiﬁcation criterion, involvement will

be divided into intrinsically and extrinsically motivated types.

In order to avoid the confusion that might result from
introducing a new involvement typology to the literature, this research
will identify these two main types of involvement using the
terminology developed by Johnson and Eagly (1989). Value-relevant
involvement will be used as the designation for those types of
involvement that are intrinsically motivated. This label signiﬁes that

such people ﬁnd an issue to be intrinsically involving due to a

18

perceived linkage between that issue and their personal values. This
would include those conceptualizations in which involvement is

elicited by an inherent interest in or concern about the issue. Value—
relevant involvement motivates people to perform issue-related tasks

because such activities are found to be self-rewarding.

Outcome-relevant involvement will be used to designate those
types of involvement that are extrinsically motivated. This label
signiﬁes that such people become involved, not because they ﬁnd the
issue itself to be of inherent interest, but rather because they perceive
the issue might have an effect on their outcomes. Outcome-relevant
involvement would include those conceptualizations in which
involvement is elicited by concerns about the consequences an issue
will have on people's ability to achieve their goals. This type of
involvement motivates people to perform issue-related tasks because
such activities are seen as instrumental to the attainment of rewards or

the avoidance of punishments.

Table 1.1 summarizes those involvement types discussed in the
literature review that are consistent with value-relevant involvement.
Table 1.2 summarizes those involvement types discussed in the
literature review that are consistent with outcome-relevant
involvement. Table 1.3 summarizes the antecedents of value-relevant
forms of involvement discussed in the literature review. Table 1.4
summarizes the antecedents of outcome-relevant forms of involvement

discussed in the literature review.

 

19

TABLE 1.1

INVOLVEMENT TYPES CONSISTENT WITH VALUE-

 

 

 

 

RELEVANT INVOLVEMENT
Type Study Description Operationalization
value- Johnson and motivational state identiﬁed from the
relevant Eagly (1990) induced by an results of a meta
involvement association between an analysis
activated attitude and
one's enduringv_alues
ego Sherif and extent to which and an sampled members of
involvement Cantril object touches upon groups that were
(1947) one's self concept actively taking a stand
on an issue.
affective Park and goal direCted arousal manipulated by
involvement Mittal elicited by interest in instructing subjects that
(1985) enhancing one's self

 

concept and projecting

a desired self image.

 

brands they were to
evaluate varied greatly

in terms of their brand

 

 

image and personality

 

20

TABLE 1.1 (Continued)
INVOLVEMENT TYPES CONSISTENT WITH VALUE-

 

 

 

 

RELEVANT INVOLVEMENT

Type Study Description Opmtlonalization

product Clarke and involvement elicited by manipulated by
involvement Belk (1979) the perceived assigning subjects to

importance of the high and low
product involvement product
categories.

enduring Houston and strength of the measured through
involvement Rothschild preexisting relationship scales which assessed
(197 8) between an individual ongoing interest in a

and the product product category

 

 

 

 

 

21

TABLE 1.2

INVOLVEMENT TYPES CONSISTENT WITH OUTCOME-

 

 

 

 

 

 

 

 

RELEVANT INVOLVEMENT
Type Study Description Operationalization
outcome- Johnson and motivational state identiﬁed from results
relevant Eagly (1990) induced by an of a meta analysis
involvement association between an
activated attitude and
one's ability to obtain
desirable outcomes
response Zimbardo an individual's concern manipulated by forcing
involvement (1960) with the consequences subjects to publicly
of his response express their position
an issue
issue Petty and the extent to which an manipulated by telling
involvement Cacioppo attitudinal issue is of subjects an issue will
(197 9) personal importance have a personal effect
on them
cognitive Park and goal directed arousal manipulated by telling
involvement Mittal elicited by concerns subjects there were
(1985) about the functional major functional
costs and beneﬁts of a differences between
product brands.

 

 

 

22

TABLE 1.2 (continued)
INVOLVEMENT TYPES CONSISTENT WITH VALUE-
RELEVANT INVOLVEMENT

 

Type Study Description Operationalizatlon

 

purchase Clarke and involvement elicited by manipulated by telling
involvement Belk (1979) the importance of the subjects they had to
purchase situation make a gift purchase

for a friend

 

 

situational Houston and the ability of a situation manipulated by telling
involvement Rothschild to elicit from subjects an issue would
(1978) individuals concern for personally affect them

their behavior

 

 

 

 

 

ANTECEDENTS OF VALUE-RELEVANT INVOLVEMENT

23

TABLE 1.3

 

 

 

 

 

 

Antecedent Study Description Operationalization
intrinsic Celsi and characteristics within measured with
sources of Olson (1988) the individual that Zaichkowski's PII
personal cause one to be scale
relevance involved
personal Zaichkowsky goals and values of measured with items
factors (1985) the person that relate from PII scale
to the object
hedonic Laurent and the ability of an measured with 3 item
value Kapferer object to provide scale
(1985) pleasure and affect
sign value Laurent and perceived expressive measured with 3 item
Kapferer value of a product scale

 

(1985)

 

 

 

 

24

TABLE 1.4

ANTECEDENTS OF OUTCOME-RELEVANT INVOLVEMENT

 

 

 

 

 

 

 

Antecedent Study Description Operationalization
situational Celsi and physical and social manipulated by
sources of Olson (1988) aspects of the having a lottery

personal environment that giving away the
relevance cause one to be product subjects were
involved to evaluate
product factors Zaichkowsky characteristics of the measured with items
(1985) product that increase from PII scale.
interest
situational Zaichkowsky temporary measured with items
factors (1985) circumstances that from the P11 scale
increase interest in
the product
risk Laurent and the perceived measured with 3 item
importance Kapferer importance of the scale
(1985) consequences of a
purchase
risk Laurent and the subjective measured with 2 item
probability Kapferer probability of a scale
(1985) mispurchase

 

 

 

 

 

25

As stated, it is believed the most important difference between
value-relevant and outcome-relevant involvement lies in the area of
motivational orientation. Empirical research on the consequences of
intrinsic and extrinsic motivation has uncovered numerous differences
in response tendencies. These ﬁndings have shown that an extrinsic
motivational orientation leads to a reduction in the enjoyment of the
activity (Condry 1977, Lepper, Greene, and Nisbitt 1973), a
reduction in learning and task mastery (Condry 1977, 1987), more
short cuts in the performance of the activity (Pittman, Boggiono, and
Rubble 1983), more frustration with the activity (Garbarino 1975), a
preference for simpler tasks (Pittman, Emery, and Boggiono 1982),
and a faster termination of the activity when a satisfactory outcome is
achieved (Kruglanski, Stein & Riter 1977). Essentially these ﬁndings
show extrinsic motivation lessens the enjoyment one receives from
performing an activity. Intrinsic motivation on the other hand
increases enjoyment of the activity since by deﬁnition it is brought
about by the inherent pleasure one feels when performing the activity
(Reeve 1992). In a marketing context then, it is hypothesized that .
value—relevant involvement is positively related and outcome-relevant
involvement is negatively related to a consumer's enjoyment of the

purchasing task.

Besides motivational orientation, the two types of involvement
can also be differentiated by the following conceptual distinctions:
- With value-relevant involvement the individual's interest lies in the

product itself, while with outcome-relevant involvement their interest

26

is focused on the outcome of the purchase. (Hawkins, Best, and
Coney 1992)

- With value-relevant involvement, interest in the product is ongoing,
while with outcome-relevant the interest occurs only at the time of
purchase. (Houston and Rothschild 1978)

- Value-relevant involvement is based on an inherent liking of the
product, while outcome-relevant involvement is based on a desire to
avoid the consequences of a mispurchase (Bloch and Bruce 1984,
Laurent and Kapferer 1985)

- Value-relevant involvement is based on value expressive motives
while outcome-relevant involvement is based on utilitarian motives

(Park and Mittal 1985)

It should be noted that while these two types of involvement
are considered to be separate and distinct they are not thought to be
mutually exclusive. There is no evidence to suggest that a consumer
cannot simultaneously experience both value-relevant and outcome-
relevant forms of involvement. For example, in regard to the
purchase of an automobile, consumers may be both intrinsically
interested in the different cars they are considering while at the same
time fee] concerned about making the correct decision regarding
which of the cars to purchase. Thus studies of these two should not
he attempts to compare consumers who have been categorized as
being either value-relevant or outcome-relevant, but rather should

seek to identify the differential effects of both within consumers.

27

EVOKED SETS

An evoked set (also called a consideration set) is the subset of
the brands in a product category that a particular individual considers
when making a purchase (Campbell 1968, Howard and Sheth 1969).
Since each product category contains a vast number of brands from
which to choose, consumers use evoked sets to simplify their decision
process. The evoked set is culled from a consumer's awareness set,
which are those brands in the product category of which the consumer
is aware. Brands in the consumer's awareness set but not included in
the evoked set are considered to be in either the inept set or the inert
set (Narayana and Markin 1975). The inept set is brands rejected
from consideration because the consumer evaluates them negatively,
while the inert set is brands the consumer evaluates positively or
neutrally but are rejected because they are not rated positively enough
to merit inclusion in the evoked set or are thought to be unavailable

for purchase.

The concept of the evoked set, with its view that consumers
only seriously consider a small number of the available brands when
making a purchase decision, has received wide spread acceptance and
solid empirical validation from within the marketing literature (e. g.
Campbell 1969, Parkinson and Reilly 1979). The only really notable
area of ambiguity in the literature revolves around the
operationalization of evoked sets. Some studies (e.g. Jarvis and
Wilcox 1973, Belonax and Mittelstaedt 1978, Belonax and Javalgi
1989) have operational ized evoked sets by asking respondents to

28

indicate which brands they ﬁnd acceptable for purchase. Others (e. g.
Ostlund 1973, Parkinson and Reilly 1980, Eroglu et. al. 1983) have
operationalized it by asking respondents to indicate which brands they
would consider for purchase. This difference could affect the
measurement of evoked sets since brands may be seen as acceptable
for purchase and yet still not be considered during a particular
purchase situation (perhaps because a sufﬁcient number of other

brands are more acceptable).

The ”considered brands" operationalization is the one that
seems more consistent with the generally accepted deﬁnition of the
construct, however there are certain situations in which the
”acceptable brands” operationalization may be more appropriate. C The
key to this issue is whether one views the evoked set as a relatively
stable entity that the consumer uses to guide purchase decisions over
time, or rather as the speciﬁc brands the consumer considers during
one particular purchase occasion: If one takes the former viewpoint,
then acceptable brands may be a more valid operationalization, since
situational characteristics (e. g. stockouts, sales promotions) may
affect the composition of the evoked sets on speciﬁc occasions but not
reﬂect a consumer's general consideration tendencies over time. This
would be particularly relevant for measuring evoked sets for products
with short purchasing cycles (e.g. grocery and drug store items),
since their frequent purchase would likely motivate consumers to
maintain a stable and enduring evoked set to help guide their frequent

purchase decisions. If the product has a long purchasing cycle then it

29

is more likely that the evoked set would be formed anew on each
purchase occasion, and the "considered brands” operationalization
would probably be more appropriate. A review of

some of the different studies of evoked sets and their

operationalizations of the construct appear in Table 5.

In terms of the types of research that have been done on
evoked set size, it has largely been directed toward two areas:
examinations of the determinants of evoked set size and examinations
of the processes used by consumers to select brands for their evoked
sets. This paper will focus on the former. Empirical research on
evoked set size has found positive correlations between it and
awareness set size (Brown and Wildt 1987) education (Maddox,
Gronhaug, Homans and May 1978) extent of information search
(Stewart and Punj 1982, Ostlund 1973), temporal proximity to
purchase (Eroglu et. al. 1983), and venturesomeness (Gronhaug
1974). Evoked set size has been found to be negatively correlated
with the number of evaluative criteria used, information variability
(Belonax and Mittlestaedt 1978) brand loyalty (Ostlund 1973,
Campbell 1969), variability of product class quality (Belonax and
Javalgi 1989) and age (Maddox, Gronhaug, Homans and May 1978).

THE EFFECTS OF INVOLVEMENT ON EVOKED SET SIZE
Involvement has also been hypothesized to be a prominent
determinant of evoked set size. While the accepted view in the

marketing literature is that there is a negative relationship between the

30

TABLE 1.5
REVIEW OF EVOKED SET STUDIES

 

 

 

 

Study Design Products Operationalization
Campbell survey of toothpaste, respondents asked how
(I968) _ buyers detergent many brands they
comider before making
a purchase
Ostlund survey of automobiles respondents asked to
(1973) current buyers name (unaided) the
brands they are
actually considering__
Jarvis and survey of coffee, respondents asked to
Wilcox (1973) buyers napkins identify acceptable

detergent alternatives from a list

of available brands

 

 

Belonax and experiment micro-wave subjects asked to
Mittelstaedt with students ovens indicate which of six
(197 8) unnamed brands are
acceptable for

 

 

 

 

purchase

 

31

TABLE 1.5 (CONTINUED)

REVIEW OF EVOKED SET STUDIES

 

 

 

 

 

 

 

Study Design Products Operationalization
Parkinson and experiment toothpaste and subjects asked to
Reilly (1980) with students deodorant indicate which brands

they would consider if
faced with an
immediate purchase
decision
Eroglu, survey of automobiles respondents were asked
Omura and buyers to identify all models
Machleit they would seriously
(1983) consider if they were
to buy one today
Stewart and survey of automobiles respondents asked to
Punj (1983) buyers indicate the price
range, sizes and
makes they are willing
to consider.
Nedungadi experiment fast food, subjects asked to make
(1990) with students mixers. a choice and then
condiments indicate which other

 

 

brands they had
comidered

 

 

32

two (Belonax and Javalgi 1989, Lastovicka and Gardner 1978,
Rothschild and Houston 1977), there has been some evidence to
suggest that the correlation is actually positive. Gronhaug (1974)
found that buyers who indicated that they were interested in cars (an
antecedent of value-relevant involvement) considered signiﬁcantly
more car models than those with moderate or little interest in cars.
Campbell (1968) meanwhile, found a strong positive relationship
between perceived price importance (an antecedent of outcome-
relevant involvement) and evoked set size. However due to two
methodological problems the validity of Campbell's ﬁndings are open
to question. First, Campbell only tested low involvement products
(toothpaste and detergent) and second evoked set size was measured
through self reports of the number of brands respondents claimed they
usually considered instead of recording which speciﬁc brands
respondents actually did consider in a real or simulated buying

situation.

Despite these ﬁndings, the predominant view in the literature
is that involvement has a negative effect on evoked set size and this
view is supported by the ﬁndings of a number of empirical research
studies. Jarvis and Wilcox (1973) found larger mean evoked sets for
product categories rated low in importance than for those rated more
highly. Rothschild and Houston (1977) found evidence to suggest
that the more important a consumer feels an attribute is, the more
intolerant they will be of brands with less than ideal performance on

that attribute (i.e. the more narrow their latitudes of acceptance).

33

While they did not attempt to correlate involvement with evoked set
size, a negative relationship was inferred since it can be assumed the
more involved one is with a purchase the more importance will be
placed on the attributes. Since importance was found to cause
narrower latitudes of acceptance, the end result of higher involvement
would be fewer brands being found acceptable, and hence smaller

evoked sets.

In the Belonax and Javalgi (1989) study, involvement was
operationalized through the use of the Consumer Involvement Matrix
(CIM), a formula that divides the number of attributes considered by
the average width of the latitudes of acceptance for those attributes.
With the CIM, involvement is deﬁned not by the level of concern
about the purchase but rather by the nature of consumer responses.
For example, a high involvement consumer was deﬁned as one who
evaluated brands on many attributes with narrow latitudes of
acceptability. Thus while a negative relationship was found between
involvement and evoked set size, there was considerable bias in these
results since involvement was operationalized so as to be synonymous
with the stringency of the criteria used to select the evoked set.
Nevertheless while individually these studies have some limitations,
together they provide fairly strong support for the theory that there is
a negative linkage between involvement and evoked set size.
Rothschild and Houston (1977) showed that involved consumers tend

to have narrower latitudes of acceptance while Belonax and Javalgi

34

(1989) established that consumers with narrower latitudes of

acceptance tend to have smaller evoked sets.

It should be noted however that in all the research studies that
found support for a negative involvement/evoked set size relationship,
involvement was operational ized in a manner much more consistent
with outcome-relevant than value-relevant involvement. Both
Rothschild and Houston (1977) and Belonax and Javalgi (1989)
measured involvement using the CIM, which measures cognitive
responses, not intrinsic interest in the product category. This type of
measure equates involvement with the amount of cognitive activity
elicited, and thus totally ignores the affective evaluative aspects
typically associated with value-relevant involvement (Park and Mittal

1985).

Additionally, it is also worth noting that in one of the most
prominent instances where the empirical results supported a positive
relationship between involvement and evoked set size, the measure
used was much more consistent with value-relevant than outcome-
relevant involvement (Gronhaug 1974). Gronhaug measured how
interested consumers were in the product category of cars rather than
using a more outcome-relevant involving measure such as how
important or risky they thought a car purchase would be for them.
While this measure is not isomorphic with value-relevant
involvement, it comes the closest of any operationalization in the

evoked set/ involvement literature. Thus it is possible that the

35

difference in empirical results regarding the involvement/evoked set
relationship may be largely a consequence of different types of

involvement being addressed in different studies.

As previously stated, the accepted view in the literature is that
involvement has a negative effect on evoked set size. As a result,
little has been written in the way of theoretical support for a positive
involvement/evoked set size relationship. However it is believed that
not only can a positive relationship be conceptually justiﬁed, but that
the explanation is actually more logical and intuitive than those given
for a negative relationship. The reasoning behind a positive
relationship is that since involved consumers are more concerned
about the purchase decision than those who are uninvolved, they
would be more motivated to consider a greater number of purchase
alternatives. The fact that this logical explanation is inconsistent with
the bulk of the empirical research on the involvement/evoked set size
relationship could be a result of the conceptual and measurement

problems that have afﬂicted the involvement concept.

Despite the intuitiveness of the reasoning behind a positive
relationship, the vast majority of the literature lines up in support of a
negative involvement/evoked set size relationship. Three different
theoretical frameworks, social judgment dreary, information
processing theory, and economics of information theory, have all
been invoked to provide conceptual justiﬁcation for this belief. All

three hypothesize that involvement indirectly affects evoked set size

36

by working through the number of attributes considered. They all
agree that as involvement increases, a consumer will evaluate brands
on more attributes, which then results in fewer brands being
considered (Belonax and Mittelstaedt 1978, Hauser and Wernerfelt
1990, Lastovicka and Gardner 1978, Rothschild and Houston 1977).
There are two main reasons why highly involved consumers evaluate
objects using more attributes: the ﬁrst is that they have greater
motivation to process and think about purchase related information
(Chaiken 1980, Johnson and Eagly 1989, Petty and Cacioppo 1986 ).
As a result they tend to have greater product knowledge and thus
greater awarness and understanding of the attributes. The second is
that their more heightened concern gives them the motivation to
evaluate alternatives on more of these attributes (Gensch and Javalgi
1987, Lastovicka and Gardner 1978). The conceptual explanation as
to why evaluating more attributes leads to smaller evoked sets is

where these three perspectives differ.

The social judgment perspective on the relationship between
involvement and evoked sets was ﬁrst advocated by Jarvis and Wilcox
(1973) and later elaborated on by Rothschild and Houston (1977).
Social judgment theory was developed by Sherif and Hovland (1961),
and its main proposition is that when a person has formed an attitude
on an issue, they will only consider credible those messages
advocating positions located within some proximity to the position
they hold. This range of credible positions surrounding asperson's

own position on an issue is called the latitudes of acceptance, and

37

only messages perceived as falling within these latitudes will have any
impact on the receiver's attitude. Messages falling outside these
latitudes will be rejected and thus will not elicit attitude change.
Social judgment hypothesizes involvement to have a negative effect
on latitude width. Thus a highly involved person will tend to have
very narrow latitudes of acceptance and thus will be difﬁcult to
persuade, while the uninvolved will tend to have wide latitudes and

thus be open to a much broader range of messages.

In using social judgment theory to explain evoked set
formation, Jarvis and Wilcox hypothesized that the high involvement
consumer would have narrower latitudes of acceptance and thus
would ﬁnd fewer brands to be acceptable enough to include in their
evoked set. Rothschild and Houston (1977) developed a more
detailed interpretation based on two propositions: the ﬁrst
hypothesizes that the high involvement consumer will evaluate brands
on more attributes than the low involvement consumer, the second
hypothesizes that because of their narrower latitudes of acceptance the
high involvement consumer is less willing to tolerate substandard
performance on any of those attributes considered. Thus since the
high involvement consumer evaluates brands using more stringent
criteria across a greater number of attributes they are not likely to
ﬁnd many brands acceptable enough to be included in their evoked

SCI.

38

The information processing explanation of why involvement is
negatively related to evoked set size was put forth by Belonax and
Mittelstaedt (1978) and is based on the recognition that an
individual's ability to process information is limited. As such,
restrictions must be placed on the number of brands one considers for
purchase and/or the number of attributes one uses to evaluate these
brands. Involvement is hypothesized to cause a reduction in evoked
set size because it motivates the consumer to use a greater number of
attributes to evaluate brands. Since processing capacity is limited,
this increase in attributes must be offset by a reduction in the number
of brands considered. The high involvement consumer is said to
simplify the processing task by reducing the number of brands, not
attributes, because doing so reduces the number of comparisons that
must be made and eliminating attributes would force consumers to

forego consumption goals (Belonax and Mittelstaedt 1978).

The economics of information perspective is based on the
work of Stigler (1961) who introduced the concept of search costs to
consumer decision making. His basic proposition is that consumers
will search for information to assist their decision making only up to
the point that its expected beneﬁts offset its costs. An equilibrium
point is eventually reached since the marginal utility gained by
acquiring additional information decreases while the costs of
gathering additional information remain constant. Hauser and
Wernerfelt (1990) applied this theory to evoked sets and proposed that

the marginal beneﬁt obtained from adding brands to an evoked set

39

decreases for each subsequent brand included, while the searching
costs associated with adding brands remains the same. The ﬁnal
evoked set is formed when there is no net utility to be gained by

adding an additional brand.

Involvement is hypothesized to have a negative effect on
evoked set size by increasing the evaluation costs associated with
adding a brand to an evoked set. Involvement increases these costs
by motivating the consumer to evaluate selected brands on more
attributes, thereby increasing evaluation costs on a per brand basis.
Since the low involvement consumer is evaluating brands on much
fewer attributes, the evaluation costs of adding an additional brand to

the evoked set is much lower.

To summarize, all three of the theoretical perspectives
discussed above agree that as involvement increases, a consumer will
evaluate brands on more attributes, which causes fewer brands to be
considered. Social judgment says evaluating on more attributes gives
consumers more opportunities to exclude brands from consideration;
information processing says more attributes require greater processing
capacity on a per brand basis, thus limiting the number a consumer is
capable of considering, while economics of information says more
attributes increase the evaluation cost on a per brand basis thus

reducing the number that would be rational to consider.

40

The main area of difference between these three perspectives
is in regard to the nature of the direct effect involvement has on
evoked set size, holding constant the number of attributes. According
to social judgment theory there is a negative direct effect because high
involvement consumers have narrower latitudes of acceptance for
each attribute. Thus even if the number of attributes is held constant
for both high and low involvement consumers, the high involvement
consumer would still tend to have smaller evoked sets, because they
are less accepting of brands not possessing the desired rating on each

attribute considered.

Both the information processing and economics of information
perspectives imply a positive direct effect. In the information
processing case, if the number of attributes is held constant for both
high and low involvement consumers, then the amount of processing
effort required to evaluate a brand in one's evoked set would be the
same for people in both groups. Since high involvement consumers
are more concerned about making a correct decision they are likely to
be willing to devote more processing capacity to evaluating brands
and thus employ a bigger evoked set. Similarly according to
economics of information if the number of attributes is held constant,
then the evaluation cost of adding brands to an evoked set would be
the same for both high and low involvement consumers. Since high
involvement consumers are more concerned about the consequences
of a mispurchase they are likely to place greater value on the beneﬁts

resulting from the addition of new brands to their evoked set. Thus,

41

given equal costs but greater beneﬁts, high involvement consumers

would be hypothesized to have bigger evoked sets.

HYPOTHESIZED EFFECTS OF VALUE-RELEVANT
INVOLVEMENT ON EVOKED SET SIZE

While a linkage has been established between involvement and
evoked set size, it is not particularly clear whether this negative
relationship is valid for both types of involvement. As previously
discussed, this relationship has been explained from three different
theoretical perspectives: social judgment, information processing and
economics of information. While involvement was treated in the
traditional manner as a single uniﬁed construct in all these
explanations, the conceptual justiﬁcation behind them is much more
appropriate for outcome-relevant forms of involvement. If
involvement were to be conceptualized in a manner more consistent
with value-relevant involvement, then the validity of these three

perspectives becomes highly suspect.

In the social judgment case, a small evoked set for high
involvement consumers is hypothesized because their heightened
concern makes them less accepting of alternatives that don't meet all
their criteria. However this rationale does not seem as applicable to
value-relevant involvement for several reasons. First, value-relevant
involvement is associated with a greater general interest in the product
category, more extensive product knowledge, and ongoing

information search (Bloch and Bruce 1984). This broader base of

42

knowledge seems to indicate that value-relevant involved consumers
would generally have a larger awareness set than those with outcome-
relevant forms of involvement. Since value-relevant involved
consumers are aware of more brands, then, other things equal, they
would likely know more brands acceptable enough to be included in

their evoked set.

Second, since those with value-relevant involvement have
shopping goals besides the determination of the best purchase
alternative, the criteria for inclusion in the evoked set are likely to be
less intensely focused on the sufﬁciencyof functional attributes. For
people who are value-relevant involved, shopping is also a leisure
activity, a way to satisfy curiosity about various brands, a way to
acquire more product knowledge, and a way to socially interact with
knowledgeable sales personnel (Bloch and Bruce 1984, Tauber 1972).
All of these goals can work to encourage such consumers to shop for
and evaluate a greater number of alternatives, even those they suspect
are likely to be found unacceptable in the ﬁnal analysis. For
example, a car enthusiast may include a sports car in his evoked set
that he knows does not offer sufﬁcient room for his family, because
of his admiration of its performance characteristics and his desire to

test drive it.

Third, value-relevant involved consumers are more concerned
about the aesthetic and experiential characteristics of the product

(Bloch and Bruce 1984). Since evaluations on these characteristics

43

are very subjective, personal inspection of brands are 'often required
to determine their acceptability. For example, characteristics such as
the handling of a high performance sports car or the sound of a hand
crafted piano need to be personally experienced in order to be
adequately evaluated. As a result it is more difﬁcult for the value-
relevant involved consumer to eliminate alternatives from
consideration without a personal inspection than it is for the outcome-
relevant involved consumer who can base such decisions on more
objective functional criteria. For example the outcome-relevant
involved automobile consumer is more likely to automatically
eliminate car models from consideration that do not meet their
standards on such objective criteria as E.P.A. ratings, horsepower,

trunk space, and airbag availability.

The information processing perspective hypothesizes an
involved consumer will consider fewer alternatives because they
evaluate brands on so many more attributes. This restricts the
number of brands they are capable of evaluating due to limits in their
processing capacity. However, value-relevant involved consumers
have better product knowledge than the outcome-relevant involved
(Bloch and Bruce 1984) and thus likely have a more developed
schema of the product category. This is reﬂected by research that has
shown product knowledge enhances one's ability to process new
information (Johnson and Russo 1984, Punj and Staelin 1983). In
addition to being able to process information more efﬁciently, their

greater pre-information search familiarity with brands and salient

44

attributes frees up processing capacity that can be directed toward the
consideration of lesser known brands in the product category. As a
result they would be less restricted by information processing
limitations than those who are outcome-relevant involved, and thus

are better able to handle larger evoked sets.

Finally the economics of information perspective hypothesizes
involved consumers consider fewer alternatives because they evaluate
brands on more attributes which results in higher evaluation costs
thereby reducing the number of brands it is cost effective to consider.
However while value-relevant involved consumers are just as likely to
incur these high evaluation costs, they are better able to offset them
since they are likely to receive greater perceived beneﬁts from the
addition of brands to their evoked sets. These beneﬁts result from the
intrinsic value consumers receive from shopping for goods with which
they have value-relevant involvement. Since shopping for these
goods is somewhat of a leisure activity (Bloch and Bruce 1984), the
value-relevant involved consumer will be willing to incur higher
evaluation costs than those consumers with outcome-relevant

involvement only.

To summarize, the conceptual justiﬁcations for the negative
relationship between evoked set size and involvement provided by the
literature do not seem to have much validity for value-relevant
involvement. The reason for this is that all of these perspectives

ignore the effects of an important factor, how much the consumer

45

enjoys shopping for the product. The level of enthusiasm the
consumer has toward the purchasing task is important not only
because it has a direct impact on evoked set size but because the two
different types of involvement may affect it in completely opposite

ways.

It is hypothesized that value-relevant involvement has a
positive relationship with shopping enthusiasm while outcome-
relevant involvement is related negatively. The key distinction
between the two is that with value-relevant involvement, an act is
performed because the individual ﬁnds performance of the act to be
self gratifying, while with outcome-relevant involvement acts are
performed because the individual believes these acts will bring about
a rewarding outcome. In the marketing context then, consumers with
value-relevant involvement are motivated to shop by a hobby like
interest in the product category and thus view the process as
something closer to a leisure activity than a task. Since by deﬁnition
value-relevant involved consumers shop because they enjoy doing so,
it stands to reason that value—relevant involvement has a positive

relationship with shopping enthusiasm.

With outcome-relevant involvement, shopping is assumed to
be a task performed in order to avoid an undesirable purchase
outcome. Since shopping effort is seen only as a means to an end,
the consumer is unlikely to receive much enjoyment from the activity.

While it is fairly clear that outcome-relevant involvement will not

46

enhance shopping enthusiasm like value-relevant involvement does, it
is proposed here that in actuality, it works to reduce the level of
shopping enthusiasm a consumer might have. There are two primary
reasons for this negative relationship. First, the primary antecedent
of outcome-relevant involvement is perceived risk, which motivates
consumers to engage in extended shopping effort because they
perceive a negative purchase outcome is too costly and/or too
probable to allow them to gamble on making an uninformed choice.
The greater the perceived risk, the more stressful the purchase task
becomes for consumers thereby decreasing their enjoyment of the
activity. Second, research has shown that an extrinsic motivational
orientation toward an activity will cause a loss of interest in, and
enjoyment of that activity (Condry 1977, Lepper, Greene and Nisbett
1973). Since outcome-relevant involvement is an extrinsically
oriented motivator of shopping behavior, these ﬁndings all give
support to the hypothesis that outcome-relevant involvement reduces

consumer enthusiasm for the shopping task.

Thus in terms of a causal chain, the two types of involvement '
are hypothesized to have differential effects on consumer enthusiasm
toward the shopping task. Shopping enthusiasm, in turn is
hypothesized to have a positive effect on evoked set size. The
reasoning behind the latter relationship is that consumers who are
enthusiastic about shopping for a particular product will be less
motivated to simplify the task by eliminating a lot of brands from

their evoked set. Conversely to the extent that consumers ﬁnd the

47

purchasing process for a product to be unpleasant, the more motivated
they will be to speed up this process by eliminating as many brands as

possible from consideration.

Thus although it has been ignored by the involvement
literature, shopping enthusiasm is considered an important mediator
of the relationship between involvement and evoked set size. A
reason for its neglect could be that shopping enthusiasm is a variable
that does not ﬁt well into the long dominant information processing
paradigm of consumer behavior. This paradigm models the consumer
as a rational problem solver who makes decisions based on objective
evaluations of the costs and beneﬁts of various alternatives. Recently
however, researchers dissatisﬁed with the limitations of the
information processing approach have attempted to expand the
boundaries of consumer research by emphasizing the importance of
the experiential aspects of consumption (Bloch and Bruce 1984,
Hirshman and Holbrook 1982, Holbrook and Hirshman 1982, Unger
and Kernan 1983). Instead of focusing on rational problem solving,
this experiential approach emphasizes the sensory pleasures and
emotional responses the consumer receives from consumption

experiences.

In terms of the need to purchase a particular product, the
information processing approach is more likely to view the shopping
activity as nothing more than a task that must be performed, i.e.

simply as the means to an end. The experiential viewpoint however

48

gives greater consideration to the notion that for some a shopping trip
may be an enjoyable self rewarding leisure activity, i.e. an end unto
itself. As a result, consideration of the level of enthusiasm toward the
shopping activity may help account for some important differences in
the ways in which consumers engage in the purchasing process. For
example how much a car shopper enjoys (or detests) dealing with car
salespeople may have as much impact on how many dealerships that
consumer visits than the perceived functional importance of the car
purchase. Thus, shopping enthusiasm is a variable that could
potentially help bridge the gap between the information processing
and experiential approaches by giving researchers a means through
which they can account for the effect of some of the experiential
aspects of consumption within the context of the traditional

information processing model.

It should be noted however that this variable is not meant to be
an indicator of how much one enjoys shopping in general but rather is
speciﬁc to a particular product category. This recognizes that even
the most reluctant shoppers probably have at least one product
category with which they maintain an ongoing interest and in which
they enjoy shopping. For example a sportsman may not enjoy
shopping in general, but could still be a very enthusiastic shopper of

guns and ﬁshing equipment.

49

CONCLUSION

It has become widely accepted that involvement has a negative
effect on evoked set size. While this appears to be valid for outcome-
relevant forms of involvement, the theoretical reasoning used to
justify this relationship seems to be inapplicable for value—relevant
involvement. Additionally, in terms of empirical support, the one
study in which involvement was operationalized in a manner more
consistent with value-relevant than outcome-relevant involvement (i.e.
as interest in the product category), a positive relationship was found
(Gronhaug 1974). Thus it is felt that there is considerable doubt
about whether the two types of involvement have the same effect on
evoked set size. This is an important issue because if their effects can
be shown to be truly different, it would indicate that value-relevant
and outcome-relevant involvement are conceptually distinct constructs

and not just different variations of the same base construct.

CHAPTER THREE
HYPOTHESES

The hypotheses explicated below are all derived from the
conceptual path model represented in Figure 1. This model lays out
the hypothesized effects of value-relevant and outcome-relevant
involvement on evoked set size, working indirectly through a number
of mediating variables. This model of the overall process differs
from what is currently accepted in the marketing literature in two
main areas: it treats involvement as two separate constructs instead of
as a single uniﬁed one, and it considers the potential mediating effects
that shopping enthusiasm has on the involvement/evoked set size
relationship. The numbers next to each path give the number of the

hypothesis where the support for these relationships is provided.

While a path model of the currently accepted view of the
effects of involvement on evoked set size has not been presented in
the literature, the model in Figure 2 represents a conceptualization of
this perspective. This path model is based on the research cited in the .
literature review and the numbers next to each path give the number
of the hypothesis where the support for these relationships is

provided.

50

51

FIGURE 1

PATH MODEL OF THE EFFECTS OF INVOLVEMENT ON EVOKED SET
SIZE:
HYPOTHESIZED RELATIONSHIPS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VALUE-RELEVANT OUTCOME-RELEVANT
INVOLVEMENT INVOLVEMENT
1 3 2 9
4
AWARENESS SHOPPING WIDTH 01’ a or SALIENT
SET SIZE ENTHUSIASM LAMBS 01“ ATTRIBUTES
, ACCEPTANCE
1 6
19 13 -17
EVOKED
SET SIZE

 

Number indicates hypothesis; (-) indicates an hypothesized negative relationship.

52

FIGURE 2

PATH MODEL OF THE EFFECTS OF INVOLVEMENT ON EVOKED SET
SIZE:
CURRENT LITERATURE

 

OUTCOME-
RELEVANT
INVOLVEMENT

 

 

14 1

 

 

 

 

 

 

 

 

 

 

 

4
AWARENESS “IDTH 0F g OF SALIENT
SET SIZE LATITUDES OF ATTRIBUTES
ACCEPTANCE
1 9 18 -1 7

 

l

EVOKED
SET SIZE

 

Number indicates hypothesis; (-) indicates hypothesized negative relationship.

53

The hypotheses discussed below will be of three types,
hypothesized relationships between constructs, hypothesized
differences between high and low independent variable groups, and an
hypothesized ﬁt of an overall path model relating the constructs to

each other.

Hypothesis 1a: Outcome-relevant involvement has a positive effect

on product knowledge.

Hypothesis 1b: Subjects with high outcome-relevant involvement
have signiﬁcantly greater product knowledge than subjects with low

outcome-relevant involvement

Outcome-relevant involvement has a positive effect on product
knowledge, because it provides motivation for more extensive
information gathering (e.g., Bettman 1979, Engel, Kollat and
Blackwell 1968) processing and purchase relevant thinking (Chaiken
1980, Johnson and Eagly 1989, Petty and Cacioppo 1986). Thus
more extensive information processing should result in greater

product knowledge than for those who are uninvolved.

Hypothesis 2a: Value-relevant involvement has a positive effect on

product knowledge.

54

Hypothesis 2b: Subjects with high value relevant involvement have
signiﬁcantly greater product knowledge than subjects with low value-

relevant involvement

Value-relevant involved consumers are characterized as having
a strong hobby-like interest in a particular product category and thus
often serve as opinion leaders for purchases in that product category
(Bloch 1982). These people maintain an ongoing relationship with
their favorite product categories (Houston and Rothschild 1978) and
thus keep up with new developments through continuous information
search (Richins and Bloch 1986). As such it is hypothesized that
value-relevant involvement has a positive effect on product

knowledge.

Hypothesis 3: Value-relevant involvement has a stronger relationship

with product knowledge than outcome-relevant involvement.

Value-relevant involvement is associated with ongoing interest
in the product category, while with outcome-relevant involvement
such interest generally only occurs at the time of purchase (Houston
and Rothschild 1978). Richins and Bloch (1986) found that
information search activities after a purchase (e.g. reading of
specialty magazines, acquiring information from interpersonal
sources) were stable over time for those with ego (enduring)

involvement but decreased in frequency /across time for those with

55

outcome-relevant involvement. As a result it is inferred that
information search under value-relevant involvement is more of a
continuous process in which the consumer strives to keep current with
new developments in the product category, while with outcome-
relevant involvement active search only likely occurs when it is seen
as necessary to aid consumer decision making. As a result it is
hypothesized that the ongoing search associated with value-relevant
involvement will result in more extensive and deeper product

knowledge than that resulting from outcome-relevant involvement.

Hypothesis 4a: Product knowledge has a positive effect on awareness

set size.

Hypothesis 4b: Subjects with high product knowledge have larger

awareness sets than subjects with low product knowledge.

The awareness set is deﬁned as the set brands in a given
product category of which the consumer is aware (Narayana and
Markin 1975). The awareness set is thus generated from an internal
information search (Hawkins, Best and Coney 1992) and hence its
size is considered to be a function of the extensiveness of the

consumer's product knowledge.

56

Hypothesis 5a: Outcome-relevant involvement has a positive effect

on awareness set size.

Hypothesis 5b: Subjects with high outcome-relevant involvement
have Signiﬁcantly larger awareness sets than subjects with low

outcome-relevant involvement

Since outcome-relevant involvement is hypothesized to have
a positive effect on product knowledge (see Hypothesis 1a), and
product knowledge is hypothesized to have a positive effect on
awareness set size (see Hypothesis 4a), it is therefore hypothesized
that outcome-relevant involvement has an indirect positive effect on

awareness set size.

Hypothesis 6a: Value-relevant involvement has a positive effect on

awareness set size.

Hypothesis 6b: Subjects with high value relevant involvement have
signiﬁcantly larger awareness sets than subjects with low value-

relevant involvement

Since value-relevant involvement is also thought to have a
strong positive effect on product knowledge (see Hypothesis 2a), and
product knowledge is hypothesized to have a positive effect on

awareness set size (see Hypothesis 4a), it is therefore hypothesized

57

that value-relevant involvement has an indirect positive effect on

awareness set size.

Hypothesis 7: Value-relevant involvement has a stronger relationship

with awareness set Size than outcome-relevant involvement.

Since value-relevant involvement is hypothesized to have a
Stronger positive effect on product knowledge than outcome-relevant
involvement (see Hypothesis 3a), it is therefore hypothesized to have

a stronger indirect effect on awareness set size as well.

Hypothesis 8a: Product knowledge has a positive effect on the

number of salient attributes used to evaluate brands.

Hypothesis 8b: Subjects with high product knowledge have more

salient attributes than subjects with low product knowledge.

Product knowledge leads not only to a greater awareness of
the attributes, but also to better understanding of their importance as
well (Brucks 1985). As such, consumers with high product

knowledge have the ability to evaluate alternatives on more attributes.

58

Hypothesis 9a: Outcome-relevant involvement has a positive effect

on the number of salient attributes.

Hypothesis 9b: Subjects with high outcome-relevant involvement
have signiﬁcantly more salient attributes than subjects with low

outcome-relevant involvement

This relationship has received empirical support from a
number of researchers (Lastovicka and Gardner 1978, Rothschild and
Houston 1978). There are two primary reasons why outcome-
relevant involvement is believed to have a positive effect on the
number of salient attributes. The ﬁrst is that outcome-relevant
involvement is thought to have an indirect positive effect on the
number of salient attributes, working through product knowledge.
Outcome-relevant involvement is hypothesized to have a positive
effect on product knowledge (see Hypothesis 1a) which is
hypothesized too have a positive effect on the number of salient
attributes (see Hypothesis 8). Thus because of their product
knowledge, high outcome-relevant consumers have the ability to
evaluate using more attributes. Second, outcome-relevant
involvement is also thought to have a positive direct effect on the
number of salient attributes. This is because outcome involved
consumers are believed to have a greater concern about the outcome
of the purchase (than those low levels of outcome-relevant
involvement) and thus would have a greater motivation t0‘evaluate on

more of these attributes (Gensch and Javalgi 1987, Lastovicka and

59

Gardner 1978). The reasoning behind this being that consumers
perceive that the more attributes they consider the more thorough
their evaluations of alternatives will be and hence the better able they
will be to discriminate between acceptable and unacceptable

alternatives.

Hypothesis 10a: Value-relevant involvement has a positive effect on

the number of salient attributes.

Hypothesis 10b: Subjects with high value-relevant involvement have
signiﬁcantly larger awareness sets than those subjects with low value-

relevant involvement

High value-relevant involved consumers evaluate on more
attributes those with low value-relevant involvement because they
tend to have superior product knowledge (see Hypothesis 2b) and
hence are aware of and understand the importance of more attributes

(see Hypothesis 8).

Hypothesis ll: Outcome-relevant involvement has a stronger
relationship with the number of salient evaluative attributes than

value-relevant involvement.

60

Outcome-relevant involvement is believed to have the stronger
relationship with the number of salient evaluative attributes because it
provides both the ability (through product knowledge) and the
motivation to evaluate on multiple attributes (see Hypotheses 9a and
9b). Evaluating on multiple attributes is seen as a means by which
consumers can increase the likelihood they will make a correct
purchase decision. Value-relevant involvement however only
increases the ability to evaluate on multiple attributes (through
product knowledge) but there is no evidence to indicate that it also
provides the motivation to do so as well (see Hypotheses 10 a and
10b). Thus outcome-relevant involvement is hypothesized to have a
stronger relationship with the number of salient attributes than value-
relevant involvement because it provides stronger motivation for
using more attributes, so as to produce more thorough product

evaluations.

Hypothesis 12a: Outcome-relevant involvement has a negative effect

on shopping enthusiasm.

Hypothesis 12b: Subjects with high outcome-relevant involvement
‘ have signiﬁcantly less shopping enthusiasm than subjects with low

outcome-relevant involvement

As stated in the literatue review, outcome-relevant.

involvement is generally caused by concerns about making a

61

mispurchase and thus consumers with this involvement are motivated
to engage in extended shopping effort in order to reduce the
likelihood of a negative purchase outcome (Assael 1984, Richins and
Bloch 1988). Because of these concerns about the consequences of
making a mispurchase, the shopping task is likely to be much more
Stressful for them than it is for those who do not have such concerns.
As a result of this high stress level, outcome-relevant involvement is

hypothesized to reduce consumer enthusiasm for the shopping task.

Additionally, with outcome-relevant involvement, interest in
the purchasing task is based on concerns about the consequences of
the activity and not on an inherent interest in the activity itself. As a
result it is considered to be an extrinsically oriented motivator of
shopping behavior (Reeve 1992). Since research on motivation has
shown that an extrinsic motivational orientation toward an activity
will cause a loss of interest in and enjoyment of that activity (Condry
1977, Lepper, Greene and Nisbett 1973), it is believed that outcome-

relevant involvement is inversely related to shopping enthusiasm.

Hypothesis 13a: Value-relevant involvement has a positive effect on

shopping enthusiasm.

Hypothesis 13b: Subjects with high value-relevant involvement have
signiﬁcantly greater shopping enthusiasm than subjects with low

value-relevant involvement

62

Value-relevant involved consumers are considered to be
intrinsically motivated shoppers since their behavior is driven by a
hobby like interest in the product category (Bloch 1982) and not by
their concerns about the consequences of the purchase.~ Thus value-
relevant involvement is believed to be related positively to shopping
enthusiasm since by deﬁnition intrinsically motivated individuals
engage in such behavior because of the inherent enjoyment they get
from performing the activity (Reeve 1992). Additionally there seems
to be some logic to the argument that other things equal, consumers
will be more enthusiastic shoppers of products they have an intrinsic

interest in, than for those they are not interested in.

Hypothesis 14a: Outcome-relevant involvement has a negative effect
on the width of the latitudes of acceptance for the evaluative

attributes.

Hypothesis 14b: Subjects with high outcome-relevant involvement
have narrower latitudes of acceptance than subjects with low

outcome-relevant involvement

Latitudes of acceptance will be operationalized in this study by
presenting subjects with a seven point rating scale (from below
average to above average) for each evaluative attribute, arid having

them indicate what is the lowest rating on that attribute a brand could

63

possess and still be considered for purchase. The hypothesized
negative relationship has already received empirical support from
Houston and Rothschild (1978). Their interpretation of social
judgment theory states that as involvement increases, people become
more willing to automatically reject purchase alternatives that do not
meet or come close to their ideal standards on any attribute they

consider salient.

Hypothesis 15: Outcome-relevant involvement has a stronger
negative effect on the width of the latitudes of acceptance than value-

relevant involvement.

It is not clear whether value-relevant involvement has a
positive or negative effect on the width of the latitudes of acceptance.
On the one hand the heightened interest in the product category could
result in strong commitments to beliefs in the necessity of certain
attributes. However it could also be that value-relevant involvement
might spawn a broad based appreciation for the merits of various
noncompatible attributes. For example the superior product
knowledge of the wine connoisseur could lead to an appreciation for a
wide variety of different types of vintage wines and not just for those
from a certain region and/or of a certain type or color. Given this
uncertainty it is believed that at minimum the relationship is not as
negative as that between outcome-relevant involvement and latitudes

of acceptance and in actuality it could be neutral or positive.

Hypothesis 16a: Shopping enthusiasm has a positive effect on evoked

set size.

Hypothesis 16b: Subjects with high shopping enthusiasm have larger

evoked set sizes than subjects with low shopping enthusiasm.

According to information processing theory, evoked sets are
used by consumers to simplify the purchasing task (Brisoux and
Iarouche 1981, Parkinson and Reilly 1979). However research in the
area of motivation shows that complex activities are more intrinsically
rewarding than simple ones, provided they are not beyond the
individual's skill level (Berlyne 1960, Reeve 1992). As a result
enthusiastic shoppers of the product have less motivation to simplify
the purchasing task by eliminating brands from their evoked set.
These consumers may actually seek to increase the number of brands
considered in order to complicate the shopping task and thus make it
more rewarding and last longer. Consumers who do not enjoy
shopping for such a product will be more motivated to simplify and
thus speed up the process by limiting the number of brands they will

consider.

Hypothesis 17a: The number of salient evaluative attributes has a

negative effect on evoked set size.

65

Hypothesis 17b: Subjects with a high number of evaluative attributes
have smaller evoked set sizes than subjects with a low number of

evaluative attributes.

The number of salient attributes is hypothesized to have a
negative effect on evoked set Size for three reasons: 1) it increases the
number of hurdles a brand must surpass in order to merit inclusion in
the evoked set (Rothschild and Houston 1977); 2) Since evaluating
on many attributes uses up a lot of processing capacity, it limits the
number of brands consumers are capable of including in their evoked
sets (Belonax and Mittelstaedt 1978); and 3) it increases the
evaluation costs for each brand in the evoked set thereby reducing the
number of brands it is cost effective to seriously consider (Hauser and

Wernerfelt 1990).

Hypothesis 18a: The widths of the latitudes of acceptance have a

positive effect on evoked set size.

Hypothesis 18b: Subjects with wide latitudes of acceptance have
larger evoked set Sizes than subjects with narrow latitudes of

acceptance.

This relationship has been empirically validated by Belonax

and Javalgi (1989). The wider one's latitudes of acceptance are, the

 

66

more brands will be considered acceptable enough to merit inclusion

in the evoked set.

Hypothesis 19a: Awareness set size is positively related to evoked set

size.

Hypothesis 19b: Subjects with large awareness sets have larger

evoked sets than subjects with small awareness sets.

Awareness set size is the number of brands in a product
category a consumer is aware of. The larger the size of ones
awareness set, the larger the pool of potential brands one has to select
from, the more brands one is likely to ﬁnd acceptable. This
relationship has already been empirically validated by Brown and

Wildt (1987).

Hypothesis 20a: Outcome-relevant involvement is negatively related

to evoked set size.

Hypothesis 20b: Subjects with high outcome-relevant involvement
have Signiﬁcantly smaller evoked sets than subjects with low

outcome-relevant involvement.

A negative indirect relationship is hypothesized because

outcome-relevant involvement is believed to have large negative

67

effects on the widths of the latitudes of acceptance (see Hypothesis
14a) and Shopping enthusiasm (see Hypothesis 13a) both of which
positively effect evoked set size. Additionally it is believed to have a
positive effect on the number of salient attributes (see Hypothesis 9a),
which negatively effect evoked set size. Thus, even though it might
have a slight positive effect on awareness set size (see Hypothesis 5a),
the overall effect of outcome-relevant involvement on evoked set size

is believed to be strongly negative.

Hypothesis 21a: Value-relevant involvement is positively related to

evoked set size.

Hypothesis 21b: Subjects with high value-relevant involvement have
signiﬁcantly larger evoked sets than subjects with low value-relevant

outcome-relevant involvement.

A positive indirect relationship is hypothesized because value-
relevant involvement is believed to have positive effects on Shopping
enthusiasm (see Hypothesis 13a) and awareness set size (see
Hypothesis 6a), both of which positively affect evoked set size.
Thus, while it is believed to have a slight positive effect on the
number of salient attributes (see Hypothesis 10a), which negatively
affects evoked set size, overall its effect on evoked set size is thought

to be positive.

 

68

Hypothesis 22: The hypothesized path model explicated in Figure 1,
will provide a better ﬁt to the data than the path model derived from

the current literature (Figure 2).

The new model will provide a better ﬁt because it separates
out the differential effects of outcome-relevant and value relevant
involvement on evoked set size (see Hypotheses 20a and 21a) and
includes the mediating effects of shopping enthusiasm on the
involvement/evoked set size relationship (see Hypotheses 12a, 13a,

and 16a) .

 

fr .1!

CHAPTER FOUR
METHOD

OVERVIEW

The methodological goal of this study was to set up an
experimental procedure that allowed statistical tests to be made of
both the hypothesized relationships between individual constructs as
well as assessments of the ﬁts of the overall path models (see ﬁgures
1 and 2). In order to perform these tests, a regression/correlational
approach was undertaken. The subjects were placed in an
experimentally created situation and asked to perform a product
evaluation task. In the course of this task, quantitative measures were
taken of each of the constructs in Figure l. The individual
relationships between constructs were tested through simple
correlational analysis and means tests while the ﬁts of the overall

models were assessed through path analysis.

SUBJECTS

The subjects for this study were 182 undergraduate students in
upper division marketing and management classes. All questionnaires
were administered to students during regular class time. Students in
these classes were offered an incentive in exchange for their
participation in this research. This incentive was the opportunity to
participate in a drawing for gift certiﬁcates and/or cash prizes.

Additionally some instructors provided additional incentives by either

69

 

70

mandating participation or offering extra credit for doing so. All
subjects were told this research was part of a market survey on the
buying intentions of college students, and were thus encouraged to
respond as realistically as possible to the experimental buying

situation.

EXPERIMENTAL DESIGN

Because there was a concern about a potential lack of variation
on one of the exogenous variables, outcome-relevant involvement, a
manipulation was used in order to introduce additional variation on
this variable into the experiment. High and low conditions of
outcome-relevant involvement were established by creating two
differently worded sets of instructions for the questionnaires
distributed to the subjects. Assignment of subjects to the two
treatment conditions was done on a class by class basis, by randomly
assigning each class to one of the two conditions. The reason the
other exogenous variable, value-relevant involvement, was not also
treated as a factor was that it is an individual difference variable and

hence cannot be manipulated (Houston and Rothschild 1977).

In order to guard against the disproportionate allocation of
value-relevant involved subjects to one of the two outcome-relevant
involvement treatment conditions, the mean value-relevant
involvement scores of each class were continually monitored. In the
end, the mean value-relevant involvement scores for subjects in the

high and low outcome-relevant involvement conditions were 76.33

 

 

71

and 75.25 respectively. A t-test on the difference between the two

groups yielded a t-value of .36, and a p-value of .717, thus indicating
no signiﬁcant difference. As a result, there is no evidence that value-
relevant involved subjects were disproportionately allocated to the two

experimental conditions.

PRODUCT SELECTION

Since value-relevant involvement cannot be manipulated
(Rothschild and Houston 1977), it was considered critical that a
product category be used in which the level of value-relevant
involvement varied naturally and widely across subjects. If such a
product category was not selected, then there might have been a
failure to introduce a sufﬁcient amount of variation on this factor into
the study. Since the majority of consumers have a low level of value-
relevant involvement with most products (Bloch 1981) the category
selected for this study needed to be one with a demonstrated ability to
elicit high levels of value-relevant involvement from a sizable

segment of the market.

The product category chosen for this study was car Stereos.
This category was selected because it is known to have a well
established-in constituency of high value-relevant involvement
consumers, namely audiophiles. Evidence of the ability of car stereo
equipment to elicit high value-relevant involvement is provided by the
proliferation of specialty magazines such as Stereo Review, High

Fidelity, and Audio magazine, which provide detailed and complex

 

72

information on the latest developments in stereo equipment to those
with an enduring interest in the product category. In addition, this
product category elicited the highest level of interest from a
convenience sample of college students, who were asked to rate their
level of interest in seven product categories (stereo receivers,

walkmans, cameras, VCRS, bicycles, and computers).

TEST PROCEDURE
Pretest Questionnaire

Research questionnaires were administered to all the
participating students in a particular class during the last 30 minutes
of the class period. The questionnaire was divided into three separate
parts; the pretest, the evaluation task and the post-test. The pretest
was administered ﬁrst without any verbal explanation of the purpose
of the research. The pretest featured a series of questions that
measured awareness set size, value-relevant involvement, car stereo
product knowledge, and the subjective likelihood that they will
purchase a car stereo within the next three years. A copy of the

complete pretest form appears in Appendix 23.

Outcome-Relevant Involvement Manipulation

Once all the subjects had ﬁnished the pretest, the forms were
collected and the evaluation task questionnaire was distributed.
Subjects were then read a series of instructions explaining the purpose
of the research and what procedures they should follow in‘ order to

complete the evaluation task. In order to create the high and low

 

73

outcome-relevant involvement conditions, two differently worded sets

of instructions were used.

The instructions were written so as to heighten outcome-
relevant involvement in one condition, but not in the other.
According to Sherrell and Shimp (1982), one of the most accepted
means of manipulating involvement in an experimental situation is to
personalize the evaluation task to subjects in the high involvement
condition but not to personalize it to those in the low involvement
group. This is seen as an appropriate manipulation procedure since
involvement is generally deﬁned as the personal relevance of a
situation (Johnson and Eagly 1989, 1990, Petty and Cacioppo 1990).
This approach has previously been used by involvement researchers
such as Apsler and Sears (1968), and Petty and Cacidppo (1979)
among others. According to Sherrell and Shimp (1982) the
evaluation task can be made personally relevant by communicating to
subjects that their performance of the task will be instrumental in
helping them achieve some goal, avoid some sanction, or that their
input is important and will have major implications for someone else.
All of these manipulations have the underlying goal of convincing
subjects in the high involvement condition that they should engage in

extensive cognitive activity when performing the evaluation task.

For this study the evaluation task was made more personally
relevant to the subjects in the high outcome-relevant involvement

condition by exposing these subjects to two involvement inducing

 

74

manipulation procedures suggested by Sherrell and Shimp (1982). In
the ﬁrst, performance of the evaluation task was made more
instrumental by linking it to the incentive drawing. This was done by
informing these subjects that the prizes offered in the drawings were
gift certiﬁcates for the purchase of a car Stereo at a local car stereo
specialty store. They were told in the instructions that the group of
car stereos they were to evaluate represented the entire selection of
models carried by this store, and thus if they did win the drawing,
they would only be able to redeem their prize by selecting one of
these brands for purchase. This was intended to give these subjects
practical motivation for realistically evaluating the car stereo brands
since they may have wanted to use these evaluations as the basis for
selecting a car stereo in the event they did win the drawing.
Additionally, the instructions explicitly asked subjects in this
treatment condition to assume they had won one of these gift
certiﬁcates when performing the evaluation task, in order to give
them a context in which it would be realistic for them to purchase a

stereo.

In addition to this ﬁrst procedure, involvement was also
heightened for subjects in the high outcome relevant condition by
communicating to these subjects that their responses were considered
very important and would have a signiﬁcant impact on the purchasing
and inventory decisions of the car stereo specialty store. Subjects
were told that the store was vitally interested in their perceptions of

the attractiveness of different car stereo models and would be using

“I.

 

75

these results to decide which models they would carry. It was
stressed to subjects that the only way the store could obtain accurate
information was for subjects to treat the simulated purchase situation
as realistically as possible, and to give responses that reﬂected the
decisions they would have made if they were actually shopping for a

031' stereo.

Subjects in the low outcome-relevant involvement group were
not exposed to any of these involvement inducing manipulations.
These subjects were simply informed that the incentive drawing was
for cash prizes and thus no attempt was made to link the evaluation
task to the incentive drawing. In addition, no mention was made of
the importance of the research. They were simply asked to indicate
what purchase decisions they would make if they had the motivation
and the resources necessary to buy the product now. This treatment
condition was designed to make the purchase Situation less involving
than the other condition since the purchase context would only be
relevant to those who actually did have the motivation and the
resources necessary to make a car stereo purchase. The wordings for

both sets of instructions appear in Appendices A and B.

Product Evaluation Task Questionnaire

After the instructions were given, the subjects were given a
questionnaire and a supplementary booklet giving them information
they could use to help them perform the product evaluation task.

This booklet contained two types of information: descriptions of

76

important car stereo attributes and product proﬁles showing how each
of the car stereos they were to consider rated on each of the
attributes. The questionnaire ﬁrst asked subjects to indicate which
evaluative attributes they would use to evaluate car stereos. Next
they were then asked to indicate their latitudes of acceptance for each
of the attributes they selected. The next two series of questions
sought to measure the amount of outcome-relevant involvement and
shopping enthusiasm subjects had regarding the purchase of a car

stereo.

Next, the subjects were asked to select their evoked set by
looking over the product proﬁles of the different car stereo brands
and indicating which of the various models presented they would
consider for purchase. Finally the subjects were asked to indicate
what strategy they used to process the product related information and
formulate their evoked sets. The supplementary information booklet

and the evaluation task forms appear in Appendices C and D.

Post-test

Upon completion of the evaluation task, subjects were asked to
turn in their questionnaire, pick up the post-test form, and then
answer a short series of questions designed to assess the amount of
product knowledge they had regarding car stereos, and a group of self
report manipulation check items designed to assess the amount of
outcome relevant involvement elicited by the treatment conditions.

When ﬁnished with the post-test, subjects handed in their

77

questionnaire and were then given an entry blank to ﬁll out for the
incentive drawing. A copy of the complete posttest form appears in

Appendix F.

MEASUREMENT OF CONSTRUCTS
Value-Relevant Involvement

According to Bloch (1981) value-relevant (product)
involvement is conceptualized as the amount of interest, arousal or
emotional attachment a product elicits from a person. To measure
this construct, Zaichkowski's (1988) revised personal involvement
inventory (RPII) was used (see Appendix C, items 29—38). This scale
was designed to provide a context free measure of a person's
involvement with any of three different psychological objects:
products, advertisements, and purchase decisions. To measure value-
relevant involvement this scale was used to assess involvement with

the product itself and not with the purchase decision.

In addition to the RPII, items from other involvement scales in
the literature were used in order to improve measurement reliability.
These include Six items from Lastovicka and Gardner's (1978)
involvement scale that relate to interest and emotional attachment to
the product and nine items from Bloch's (1981) product involvement
scale (see Appendix C, items 39-55). Since the Bloch scale was
developed to measure involvement with the product category of cars,
the items borrowed were modiﬁed to assess involvement with the car

SICI'COS .

78

Outcome-Relevant Involvement

Outcome-relevant involvement is conceptualized to be the
amount of concern an individual has about the consequences of a
purchase decision. To assess this construct, Jacoby and Kaplan's (1972)
perceived risk scale (Appendix E, items 37-45) and 10 items from Slama
and Tashchian's (1985) 33 item purchasing involvement scale were used
(Appendix E, items 46-55). The entire scale was not used since much
of it pertains to measuring an individual's involvement with purchasing
activities in general rather than involvement of a speciﬁc purchase. The
items that were selected were ones that could be adapted to measuring
involvement with the speciﬁc task of purchasing a car stereo. For
example Slama and Tashchian's (1985) item "On most purchase
decisions the choice I make is of little consequence (p.79)," has been
changed to, "In regards to the purchase decision of which car stereo to

buy, the choice I make is of little consequence."

In addition several subjective-State items were also used to
assess how important subjects thought the evaluation task was and how
much cognitive activity they devoted to this task (Appendix F, items 42-
45). The main purpose of these items was to check the eﬁ‘ectiveness of

the outcome-relevant involvement manipulations.

Awareness Set Size
Awareness set size was deﬁned as those brands in a product

category with which a person has some degree of familiarity.

79

Awareness set size was measured by presenting subjects with a list of
actual and ﬁctitious stereo manufacturers and having them self report
their degree of familiarity with the brand name on a three point scale.
The three categories were unfamiliar, somewhat familiar, and deﬁnitely
familiar. A trichotomous measure was used instead of a dichotomous
familiar-unfamiliar format, because the two familiar categories allowed
more information to be obtained about the degree of a subject's
familiarity. Only one unfamiliar category was used since logically there
are no degrees of unfamiliarity. Fictitious manufacturers were included
as a way to account for false self reports of awareness (see Appendix C,

items 1-28).

For the analysis discussed in this paper, awareness set size was
measured by assigning subjects two points for being deﬁnitely familiar,
one point for being somewhat familiar, and no points for being
unfamiliar with actual manufacturers. For ﬁctitious manufacturers the
scoring was -1 points for deﬁnitely familiar, no points for somewhat

familiar, and two points for unfamiliar.

Number Of Salient Evaluative Attributes

Evaluative attributes are deﬁned as those attributes a person
considers when evaluating products for purchase. The number of
evaluative attributes was measured by presenting subjects with a list of

attributes and having them indicate which ones they would use to

80

evaluate the adequacy of different brands of stereos (see Appendix E,
items 1-16). These attributes were derived from product reviews in

the 1992 Crutchﬁeld Catalog (Crutchﬁeld Corporation 1992).

Latitudes Of Acceptance

The latitudes of acceptance are deﬁned as the range of
acceptable positions or levels of an attribute a person will consider
acceptable when evaluating products for purchase. The measurement
procedures used to assess latitudes of acceptance were different for
the two different types of attributes used in this study. The ﬁrst type
of attribute is a performance rating. This is a one (above average) to
seven (below average) rating that indicates how well a car stereo
performs on an audio speciﬁcation in comparison to other models on
the market. Latitudes of acceptance were assessed for this type of
attribute by having subjects indicate what was the lowest rating a
model of car stereo could have on each of these attributes, without
automatically being rejected from consideration for purchase (see

Appendix E, items 28-36).

The other type of attribute is a product feature. This is a car
stereo feature that is either present or absent from a particular model.
Latitudes of acceptance were assessed for this type of attribute by
having subjects indicate on a seven point scale how important it was
to them that a model of car stereo possess or not possess the feature
(see Appendix E, items 17-27). To the extent that a feature is either

required (or forbidden) the narrower the latitudes were. To the extent

81

that feature is not required (or forbidden) the wider the latitudes will

be.

Evoked Set Size

Evoked sets are the brands or models in a product category
that a person will consider for purchase. This was measured by
having the subjects indicate from the list of brands which ones they

would seriously consider for purchase (see Appendix E, part 4 items

1-28).

Shopping Enthusiasm

Shopping enthusiasm is deﬁned as the anticipated enjoyment
an individual feels with respect to a speciﬁc Shopping task. Since this
is a new construct there were no existing instruments in the marketing
literature to measure this trait. As a result, scales were borrowed
from the leisure discipline and then adapted to a consumption context.
Leisure scales were used because the measurement goal was to assess
the level of affect one has for a particular activity, and this is the type
of measurement task that is commonly performed in leisure research.
The ﬁrst scale was adapted from Unger and Domoff (1979) and is a
10 point scale that asks subjects to rate their preference for shopping
for stereos in comparison to all other products. The second scale was
adapted from Iso-Ahola (1979) and asked subjects to assess to what

extent they feel shopping for a stereo would be a leisure activity.

82

The ﬁnal leisure scale consisted of seven items adapted from
Unger and Kernan (1983). The full scale features 26 items and seeks
to assess the amount of subjective leisure one gets out of an activity.
All the items give a statement and ask subjects to indicate whether

they agree or disagree using a ﬁve point Likert format.

In addition to the scales borrowed from the leisure literature,
ﬁve additional seven point agree-disagree Likert items were
developed to directly assess shopping enthusiasm. See Appendix E,

part 3, items 56-69, for all of the shopping enthusiasm items.

Product Knowledge

Product knowledge was assessed three different ways in the
study. First objective product knowledge was measured in the pretest
before subjects had performed all of their decision making tasks (see
Appendix C, items 56-65). This was designed to assess the amount
of internal or pre-existing product knowledge a subject brought with
them into the experiment. In addition, objective product knowledge
was also measured in the post-test as well, using the same nine pretest
questions plus two additional items (see Appendix F, items 30-41).
Measuring product knowledge after the decision making task was
performed allowed for the asSessment of product knowledge that
subjects acquired in the course of performing their product
evaluations. In both cases objective product knowledge was assessed
by asking subjects questions about what certain car stereo attributes

were and what their signiﬁcance was.

83

The third means by which product knowledge was measured
was through the administration of a two item subjective product
knowledge scale during the pretest (see Appendix C, items 66-67).
This scale was designed to assess how much a person thought they
knew about car stereos. A measure of subjective product knowledge
was taken for two reasons: ﬁrst, subjective knowledge has been
recognized as being distinct from objective knowledge because it
indicates self conﬁdence levels as well as knowledge levels (Brucks
1985). As a result it was felt that a more complete assessment could
be provided by measuring both subjective and objective knowledge.
Second the subjective scale provided a backup measure for product
knowledge in case the objective measures were plagued by a lack of

reliability.

CHAPTER FIVE
DATA ANALYSIS AND FINDINGS

INTRODUCTION

The preceding chapter discussed the experimental design and
data collection procedures. This chapter presents the analysis of the
data. This discussion will be divided into two main sections; the ﬁrst
will present the reliability analysis used to develop ﬁnal measures for
each variable in the study. The second will present the results of the
statistical analyses performed in order to test the hypotheses outlined

in Chapter 3.

RELIABILITY ANALYSIS
Development of Final Measures

As discussed in the previous chapter, each variable in the
study was measured using at least one multi-item scale. Multiple
indicators were used in order to help insure that reliable measures
could be obtained for each variable. In order to determine which
combination of items provides the most reliable measure for each
variable, item analysis was performed. The goal of item analysis is to
produce ﬁnal measures that are more reliable by eliminating from the
measurement scales those items identiﬁed as being weak or

ambiguous.

84

85

The reliability of the overall measurement scales were assessed
with measures of internal consistency. Cronbach's alpha was
calculated for all of the scales except for product knowledge, evoked
set Size and number of salient attributes. KR20 was calculated for
these three scales Since they were composed of dichotomous rather
than multichotomous scale items (Nunnally 1978). Internal
consistency was used to assess reliability because Nunnally (1978)
states that it provides the best estimate when ”the major source of
measurement error is because of the sampling of content" (p.230).
Since each scale is composed of only a limited sampling of items
from the domain of all items that potentially might have been used, it
is felt that the sampling error associated with the selection of scale
items represented the greatest source of measurement error. Internal
consistency was considered appropriate even for the evaluation task
variables (product knowledge, lattitudes of acceptance, awareness set
size and evoked set size) because each of these scales was designed to
capture the latent decision making characteristic underlying these
variables (such as familiarity with a product category or stringency in

requirements) and thus none were considered to be multi-dimensional.

Judgments as to the reliability of individual scale items were
based upon two criteria: the size of the corrected item-to-total
correlation and the direction of the change in alpha if the item were
deleted from the scale. The corrected item-to-total correlation is the
correlation between a scale item and the sum of the raw score for the

other items in the scale. This is used as an indicator of how much an

86

item correlates with the other items in the scale. The alpha if the item
is deleted indicates what the Cronbach's alpha (KR20) is for the other
items in the scale. By comparing this to the alpha (KR20) for the
overall scale it can be determined whether the alpha of the scale will
be increased or decreased if the item is eliminated. This is used as an
indicator of the relative contribution a particular item makes to the

overall internal consistency of the scale.

These two criteria were used to purify the measurement scales
by eliminating items that had both low item-to-total correlations
(below .30) and whose deletion led to an improvement in the alpha 4
(KR20) of the scale. These eliminations were made in a stepwise
manner, such that the most unreliable item that met the two criteria
was eliminated ﬁrst and then the reliability analysis was recomputed.
The process was repeated until no other scale items met the criteria
for elimination. Results of the item analyses for each of the eight

variables in the study are presented in Table 5.1.

Value-Relevant Involvement

Two scales were used to measure value-relevant involvement:
the 10 item Revised Personal Involvement Inventory (Zaichkowsky
1988), and a combination of 17 items taken from scales developed by
Bloch (1981) and Lastovicka and Gardner (1978). Separate item
analyses of the two scales were performed with the RPII yielding a
Cronbach's alpha of .9059, and the Bloch-Lastovicka generating an

alpha of .8896. When the two scales were combined the alpha was

87

TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES

1. VALUE-RELEVANT INVOLVEMENT

SCALE- CORRECTED ALPHA IF
ITEM ITEM-TO-TOTAL ITEM
CORRELATION DELETED

RPII-l .7140 .9341
RPII-2 .7672 .9332
RPll-3 .6429 .9356
RPIl-4 .6499 .9353
RPII-5 .6890 .9346
RPll-6 .6926 .9346
RPII-7 .6953 .9345
RPII-8 .6243 .9356
RPIl-9 .6863 .9346
RPlI- 1 O .4605 .9386
BL-l .6835 .9348
BL-2 .6874 .935 1
BL-4 .6432 . 9355
BL—5 .5904 .9361
BL-6 .5755 . 9363
BL-7 .4681 . 9378
BL-8 .5256 .9370
BL-10 .5472 .9368
BL-12 .5831 .9362
BL— 14 .6294 . 9357
BL- 15 .5400 .9368
BL— 16 .5028 .9373
BL-17 .6263 .9359

 

Cronbach's a = .9384
SCALES: RPII - Revised Personal Involvement Inventory,
Zaichkowsky 1988
BL - Bloch 1981(BL-1 to BL-6), Lastovicka and
Gardner 1978 (BL-7 to BL—l7)

88

. TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES (CONTINUED)

2. OUTCOME-RELEVANT INVOLVEMENT

SCALE- CORRECTED ALPHA IF
ITEM ITEM-TO-TOTAL ITEM
CORRELATION DELETED
ST-l .3158 .7570
ST-2 .6267 .7057
ST-3 .6037 .7126
ST-3 .4447 .7376
ST-S .4208 .7426
ST-6 .2972 .7580
ST-7 .4342 .7393
ST-8 .2971 .7584
ST-9 .5295 .7249

 

Cronbach's at = .7604 .
SCALE: ST - Slama and Tashchian (1985)

3. SALIENT ATTRIBUTES

ATTRIBUTE CORRECTED KR20 IF ITEM
. ITEM-TO-TOTAL DELETED
CORRELATION
POWER .2347 .7301
FM SENSITIVITY .2805 .7256
FM SEPARATION .2770 .7280
FM QUIETING .3623 .7157
T. FREQ. RESPONSE .5414 .6919
T. WOW & FLUTTER .4811 .6969
T. SIGNAL/ NOISE .4880 .6965
SEP. BASS & TREBLE .3040 .7228
CD CHANGER .2780 .7292
BIAS CONTROL .4282 .7071
DOLBY N. R. .4932 .6966

KR20 = .7326

89

TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES (CONTINUED)

4. SHOPPING ENTHUSIASM

SCALE- CORRECTED ALPHA IF
ITEM ITEM-TO-TOTAL ITEM
CORRELATION DELETED
CL—l .7805 .8924
CL-2 .7193 .8977
SL-l .7299 .8946
SL-2 .7434 .8932
SL-3 .5952 .9006 '
SL-4 .5805 .9016
SL—S .4205 .9082
SL-6 .6516 .8981
SL-8 .8208 .8890
SL-9 .4483 .9072
SL—lO .8259 .8900

 

if"

Cronbach's or = .9063
SCALES: CL - Comparative Leisure: Unger and Dornoff 1986,
Iso-Ahola 1979 (CLSZ)
SL - Subjective Leisure: Unger and Kernan 1983

90

TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES (CONTINUED)

5. PRODUCT KNOWLEDGE

 

SCALE- CORRECTED KR20 IF
ITEM ITEM-TO-TOTAL ITEM
CORRELATION DELETED
PRE-1 .2964 .7053
PRE-2 .2607 .7078
PRE-3 .1685 .7139
PRE-4 .2450 .7088
PRE-6 .2273 .7101
PRE-7 .3736 .7003
PRE-8 .2505 .7107
PRE-9 .2398 .7098
PRE-10 .1960 .7125
SUB-1 .5841 .6703
SUB-2 .5237 .6853
POST-1 .1929 .7123
POST-2 .2819 . .7064
POST-3 .2538 .7082
POST-4 .2507 .7087
POST-5 .1501 .7148
POST-6 .2087 .7125
POST-7 .4121 .6991
POST-8 .2746 .7068
POST-9 .2734 .7088
POST-10 .3009 .7057
POST-11 .1353 .7051
POST-12 .1793 .7136
KR20 = . 7164
SCALES: PRE - Pretest

SUB - Subjective Evaluation
POST - Post-test

91

TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES (CONTINUED)

6. AWARENESS SET

MANUFACTURER CORRECTED ALPHA IF
ITEM-TO-TOTAL ITEM
CORRELATION DELETED

CLARION .6426 .8792
IV C .6517 .8791
BLAUPLNUKT .4241 . 8870
KENWOOD .5509 .8834
PIONEER .5424 .8854
SANYO .5100 .8841
SHERWOOD .4312 .8873
SONY .4644 .8865
ALPINE .5665 . 8822
JENSEN .6560 .8785
PANASONIC .3683 . 8882
AUDIOVOX .6209 .8800
CRAIG .6052 .8806
KRACO .6465 .8790
AIWA .5173 .8841
DENON .5155 .8846
TARGA .4457 .8861
NAKAMICHI .3902 . 8883

Cronbach's a = .8894

 

92

TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES (CONTINUED)

7. LATTITUDES OF ACCEPTANCE

ATTRIBUTE CORRECTED ALPHA 1F
ITEM-TO—TOTAL ITEM
CORRELATION DELETED
POWER .3808 .8030
T. FREQ. RESPONSE .6307 .7841
FM SENSITIVITY .5123 .7933
FM SEPARATION .5183 .7928
FM QUIETING .6201 .7851
T. WOW & FLUTTER .4865 .7959
T. SIGNAL/NOISE .4623 .7980
ANTI-THEFT .3260 .8056
CD CHANGER .2847 .8076
T. MUSIC SEARCH .3353 .8052
RADIO/T APE SWITCH .2825 .8076
BIAS CONTROL .3652 .8037
POWER OFF EJECT .2598 .8091
DOLBY NR. .3850 .8025
SECURITY CODE .2949 .8071
REMOVABLE CHASSIS .3236 .8057
DETACHABLE FACE .2764 .8079

Cronbach's a = .8107

93

KR20 = .8807

TABLE 5.1
ITEM ANALYSIS OF FINAL SCALES (CONTINUED)
8. EVOKED SET
CAR STEREO MODEL CORRECTED ALPHA IF
ITEM-TO-TOTAL ITEM
CORRELATION DELETED
PIONEER KEHMSSOO .4632 .8769
KENWOOD KRC540 .3853 .8809
SONY XRU220 .4133 .8798
CLARION 3670RC .3190 .8850
PIONEER KEHM7200 .6217 .8692
IV C KSRG8 .5474 .8729
NAKAMICHI TD3PPI .5052 .8751
KENWOOD KRC740 .6982 .8647
TARGA Q7000 .4502 .8777
SONY XRU550 .7466 .8617
PIONEER KEHM8200 .6995 .8645
SONY XRU770 $7581 .8610
KENWOOD KRC940 .7163 .863 5

 

94

raised to .9357. Thus it was decided to utilize a combined scale.
This combined scale was then puriﬁed by eliminating four items (all
from the Bloch-Lastovicka scale). This puriﬁed value-relevant
involvement scale features a Cronbach's alpha of .9384, and item-to-

total correlations that range from .4605-.7672.

Outcome Relevant Involvement

Two scales were used to measure outcome-relevant
involvement: Jacoby and Kaplan's 9 item perceived risk scale, and
10 items adapted from a scale used by Slama and Tashchian (1985).
Since four pairs of the perceived risk questions were multiplied
together (probability times cost for the four different types of risk),
the perceived risk scale is mathematically reduced to ﬁve items for
reliability purposes. Separate item analyses of the two scales were
performed with the perceived risk scale generating an alpha of .387,
and the Slama-Tashchian scale obtaining a reliability of .6555. A
combination of the two scales produced an alpha of only .426. As a
result it was decided that using the Slama-Tashchian scale alone
provided the most reliable means of measuring outcome-relevant
involvement. The reliability of this scale was then further increased
by eliminating an item with a negative item-to-total correlation. This
ﬁnal puriﬁed measure of outcome-relevant involvement features an
alpha of .7604, and item-to-total correlations ranging from .2972 to
.6276.

I:

95

Although outcome-relevant involvement was the variable that
was manipulated in this study, no signiﬁcant differences were found
between the two treatment groups in terms of their mean scores on the
ﬁnal puriﬁed measure of outcome relevant involvement (p=.25).
Additionally no signiﬁcant differences were found between the two
experimental groups on either of the two manipulation check
questions. However while it is apparent that the manipulations were
not effective, it should be noted that their purpose was to introduce
variation in outcome-relevant involvement into the experiment. An
examination of respondent scores on this variable indicate that despite
the ineffectiveness of the manipulations, an adequate amount of
variation was obtained. This is evidenced by the fact that the
standard deviation for outcome-relevant involvement was 4.6 with
total raw scores on the ﬁnal nine itemoutcome—relevant involvement

scale ranging from 9 to 33.

Thus, while sufﬁcient variation in outcome-relevant
involvement was obtained, it is was possible, given the ineffectiveness
of the manipulation, that the variation in this variable was not
conceptually distinct from that found in value-relevant involvement.
In order to test whether there was discriminant validity between the
measures of the two types of involvement, two conﬁrmatory factor
analyses were run on the items in both scales. The ﬁrst C.F.A. tested
a one factor model in which all the items from both scales were
combined into a single involvement factor. The one factor

(involvement) model produced a x2 of 1654.27 at 560 degrees of

96

freedom (p. < 01). The second conﬁrmatory factor analysis tested
the two factor model in which scale items were constrained a priori as
belonging to either outcome or value relevant involvement. The two
factor solution produced a x2 of 1470.38 at 559 degrees of freedom (p
< .01). The difference in x2 values between the one and two factor
models is 183.89 at 1 degree of freedom, which is signiﬁcant at the
.01 level. Additionally, a 95% conﬁdence interval was constructed
around 4). Unity was not conatined in this interval (6 = .525,
standard error = .064), thus providing further evidence that there is
discriminant validity between the outcome-relevant and value-relevant

involvement scales.

Number of Salient Attributes

The number of salient attributes was measured by asking
subjects to indicate whether or not they would use a particular
attribute to evaluate car stereos. The original scale asked them to
make this judgment in regard to 16 different attributes, and the score
for this variable was obtained by summing up for each subject the
number of attributes they indicated that they would use. The KR20
for the original scale was .6744. After 5 items were eliminated in the
item analysis, the KR20 for the scale was increased to .7326. The
item-to—total correlations for the puriﬁed measure range from .2347 to

.5414.

97

Product Knowledge

Product knowledge was assessed with three different scales: a
10 question pretest of objective knowledge, a two item assessment of
subjective knowledge, and a 12 question post-test of objective
knowledge. The pretest was designed to measure preexisting
knowledge, the post-test was designed to capture both preexisting
knowledge and knowledge acquired during the experimental task, and
the subjective Scale was designed to assess how much a subject
thought they knew about the product category. The pretest scale had
an KR20 of .3591 (.5935 if one unreliable question is removed), the
two item subjective scale had an KR20 of 1.0, and the post-test had
an KR20 of .5303 (.6020 if one unreliable question is removed).
When the three scales were combined the KR20 increased to .7145.
Thus it was decided to use a combined scale to measure product
knowledge. The removal of one unreliable item increased the KR20
of the ﬁnal measure to .7164, with the item to total correlations

ranging from .1353 to .5841.

Awareness Set Size

Awareness set size was assessed by asking subjects to rate
their degree of familiarity (deﬁnitely familiar, somewhat familiar,
unfamiliar) with 20 real and 7 ﬁctitious manufacturers of car stereo.
The reliability analysis for the entire 27 item scale yielded an alpha
.8728. Item analysis led to the elimination of eight items, including

all seven ﬁctitious manufacturers. The ﬁnal puriﬁed measure for

98

awareness set size features a Cronbach's alpha of .8894, with item-to-

total correlations ranging from .3683 to .6465.

Shopping Enthusiasm

Shopping enthusiasm was measured with a scale consisting of
9 items adapted from research in the leisure literature (Unger and
Dornoff 1979, Iso-Ahola 1979, Unger and Kernan 1983) and 5 items
that were newly created. The complete 14 item scale yielded an alpha
of .8861. After the elimination of two unreliable items, the ﬁnal
measurement scale produced an alpha of .9101, with item-to-total

correlations ranging from .5588 to .8224.

Lattitudes of Acceptance

Lattitudes of acceptance were assessed using two scales: one
measuring the minimum level of performance one was willing to
tolerate on nine different performance speciﬁcations, the other
measuring how necessary it was that the car stereo have or not have
11 different product features. Since conceptually it is believed that
both feature and performance requirements are equally valid aspects
of the lattitudes of acceptance, a combined scale was used. The
overall combined scale had an alpha of .7623. After the measure was
puriﬁed by eliminating three unreliable scale items, the alpha
increased to .8107. The item-to—total correlations for the ﬁnal

measure range from .2598 to .6341.

99

Evoked Set Size

Evoked set Size was measured by asking subjects to indicate
whether or not they would seriously consider for purchase a particular
model of car stereo. The original scale asked them to make this
judgment in regard to 28 different models of car stereo, and the score
for this variable was obtained by summing up for each subject the
number of models they indicated that they would consider. The
KR20 for the original scale was .8319. After 15 items were
eliminated in the item analysis, the KR20 for the scale was increased
to .8807. The item-to-total correlations for the puriﬁed measure

range from .3190 to .7466.

Summary

The reliability coefﬁcients that were computed for the ﬁnal
scales used to measure the eight variables in the study range from
.7164 to .9384. Since these all greatly exceed Nunnally's (1967)
criterion for acceptable internal consistency of .50 to .60, the

variables are considered acceptable for further analysis

TESTS OF THE HYPOTHESES OF THE STUDY

The results of the tests of the research hypotheses outlined in
Chapter 3 are presented below. For t-tests, the high and low groups
were created by using median splits. Subjects with scores above the
median were put in the high group, subjects with below median
scores were put in the low group. Subjects with the median score

were put in the group that best equalized the size of the two groups.

100

The median was used as the dividing point for the two groups instead
of the mean because it better allowed for the creation of groups with

near equal sample sizes.

Hypotheses that predicted a difference in the strength of two
different correlations were tested using a formula developed by
Steiger (1980) to test for differences between correlations from the

same sample (see Table 5.4 for complete explication of the formula).

All signiﬁcance tests were two-tailed and used a .05 level of
signiﬁcance. The statistical results of the simple correlational tests
are presented in Table 5.2, the t-test results are shown in Table 5.3,
and the test for differences between correlations are presented in

Table 5 .4.

101

TABLE 5:2
PEARSON CORRELATIONS OF ALL VARIABLES

V O P A S # L E
R R K S E of of S
I I S S A S
13
value-
relevant
involvement
outcome- .40
relevant **
involvement
product .39 .26*
knowledge ** *
1 a
awareness .46 .27 .48
SCI SIZC * t t t t *

shopping .65 .29* .41 * .43*
enthusiasm ** * * *
13a 12a

number of .25 .09 .10 .23 .23
salient ** ** **
attributes 10a 9a , 8a

lattitudes 0f -.33 -.21 -.08 -. 16 -20 -.48
acceptance ** ** * ** **
15 14a

evoked set .16 .09 -.01 .03 .16 .14 .08

Size *
21a 203 19a 16a 17a 18a

Numbers indicate hypotheses **p<.01, *p<.05 (2 tailed)

102

- TABLE 5.3:
RESULTS OF HYPOTHESIS TESTS : T-TESTS

hypoth dependent independent variable # of mean t- p-
# variable groups cases value value
1b product high outcome- 88 21.72 4.41 .001

knowledge relevant involvement.
low outcome relevant 95 18.55
involvement
2b product high value-relevant 91 21.62 4.10 .00]
knowledge involvement.
low value relevant 92 18.62
involvement
4b awareness low product 98 16.33 -6.95 .001
set size knowledge
high product 85 23.25
knowledge
5b awareness high outcome 88 21.61 3.70 .001
set size relevant involvement
low outcome-relevant 95 17.62
involvement
6b awareness high value relevant 91 22.34 5.36 .001
set size involvement
low value-relevant 92 16.77
involvement
8b # of salient low product 98 8.65 -2.08 .039
attributes knowledge
high product 85 9.32
knowledge
9b # of high outcome- 91 9.30 ' 2.02 .045
attributes relevant involvement
high outcome- 92 8.65
relevant involvement
10b # of salient high value-relevant 91 9.30 2.09 .038
attributes involvement
low value-relevant 92 8.63

involvement

103

TABLE 5.3 (continued)
RESULTS OF HYPOTHESIS TESTS: T-TESTS

hypoth dependent groups # of mean t- p—
# variable cases value value
12b shopping high outcome- 88 30.55 -3.29 .001

enthusiasm relevant involvement
low outcome-relevant 92 34.90
involvement

13b shopping high value-relevant 91 28.14 -7.92 .001

enthusiasm involvement
low value-relevant 92 37.41
involvement
14b lattitudes high outcome- 91 48.65 -3.51 .001
of relevant involvement
acceptance
low outcome-relevant 92 54.80
involvement
16b evoked set high shopping 88 7.99 1.23 .220
size enthusiasm
low Shopping 95 7.27
enthusiasm

17b evoked set low number of salient 88 7.26 -1.18 .239
size attributes
high number of 95 7.95
salient attributes

18b evoked set narrow lattitudes of 88 7.78 .54 .589

Size acceptance
wide lattitudes of 95 7.46
acceptance
19b evoked set low awareness 88 7.28 -1.16 .247
Size set size
high awareness 95 7.96

set size

104

TABLE 5.3 (continued)
RESULTS OF HYPOTHESIS TESTS: T-TESTS

hypoth dependent groups # of
# variable cases
20b evoked set high outcome- 91
size relevant involvement
low outcome-relevant 92
involvement

21b evoked set high value-relevant 91
size involvement

low value-relevant 92
involvement

mean

7.99

7.27

7.95

7.29

t- p-
value value
1.23 .220
1. 12 .263

105
TABLE 5.4: TESTS FOR DIFFERENCES BETWEEN

CORRELATIONS
Hypoth Dependent Independent rs *t- p-Value
esis Variable Variables Value
#
3 product outcome-relevant -.2651 1.996 .0456
knowledge involvement
value-relevant -.3945
involvement
7 awareness outcome-relevant -.2703 2.878 .004
set size involvement
value-relevant -.4630
involvement.
11 # of outcome-relevant -.0897 2.024 .043
attributes involvement
value-relevant -.2491
involvement
15 lattitudes outcome-relevant . 1737 2.49 .014
of involvement
acceptance
value-relevant .3969
involvement

*Steiger's (1980) t-test for difference between dependent r's
(Ixy - Ivy) 4(11 -l)(1+ I'xv)

 

 

t =
I 2 (n- 1)/(n-3) | RI + r2 (14...):
l‘xy + Ivy
{2 =
2

I RI = l - Ixy2 - rvy2 - I'xv2 + 2(Ixyl'vyi'xv)

106

Hypothesis 1a: Outcome-relevant involvement has a positive effect

on product knowledge.

Hypothesis 1b: Subjects with high outcome-relevant involvement
have signiﬁcantly greater product knowledge than subjects with low

outcome-relevant involvement.

Both hypotheses are supported. In 1a, a signiﬁcant positive
relationship was found between outcome-relevant involvement and
product knowledge. The correlation is .256, R2 =.066, and the p-

value is < .001.

In lb, a signiﬁcant difference in mean product knowledge was
found between the high and low outcome-relevant involvement
groups. The mean of the high group was 21.7, the mean of the low
group was 18.5. The t-value for the means test was 4.41 with a p-

value < .001, and the difference was in the predicted direction.

The ﬁndings of both 1a and 1b, provide strong support for the
existence of a positive relationship between outcome-relevant
involvement and product knowledge. This supports the reasoning that
consumers who are concerned about the outcome of the purchasing
task will be more motivated to improve their expertise by actively

seeking out information to help them make their decision.

107

Hypothesis 2a: Value-relevant involVement has a positive effect on

product knowledge.

Hypothesis 2b: Subjects with high value relevant involvement have
signiﬁcantly greater product knowledge than subjects with low value-

relevant involvement

Both hypotheses are supported. For 2a, a signiﬁcant positive
relationship was found between value-relevant involvement and
product knowledge. The correlation is .3935, R2 =.155, and the p-

value is < .001.

For 2b, a signiﬁcant difference in mean product knowledge
was found between the high and low value-relevant involvement
groups. The mean of the high group was 21.6, the mean of the low
group was 18.6. The t-value for the means test was 4.10 with a p-

value < .001, and the difference was in the hypothesized direction.

The ﬁndings of both 2a and 2b, provide support for the
existence of a positive relationship between value-relevant
involvement and product knowledge. These ﬁndings support the
notion that value-involved consumers are likely to have superior
product knowledge due to their maintenance of an ongoing

relationship with the product category.

 

108

Hypothesis 3: Value-relevant involvement has a Stronger relationship

with product knowledge than outcome-relevant involvement.

Hypothesis 3 is supported. A signiﬁcant difference was found
between the correlations of the two different involvement types and
product knowledge. The correlation between outcome relevant
involvement and product knowledge was -.27, the correlation between
value—relevant involvement and product knowledge was -.39. The
Steiger (1980) signiﬁcance test yielded a t-value of 1.996, with a p-
value of .046, and the difference was in the predicted direction. The
results provide support for the hypothesis that value-relevant
involvement has a stronger relationship with product knowledge than
outcome-relevant involvement. The reasoning for this difference is
that value-involved consumers are thought to have maintained a
longer term relationship with the product category, and thus should

have developed a deeper set of product knowledge.

Hypothesis 4a: Product knowledge has a positive effect on awareness

set size.

Hypothesis 4b: Subjects with high product knowledge have larger

awareness sets than subjects with low product knowledge.

Both hypotheses are supported. For 4a, a signiﬁcant positive

relationship was found between product knowledge and awareness set

‘

109

size. The correlation between the two variables is .477, R2 =.228,

and the p-value is < .001.

For 4b, a signiﬁcant mean awareness set size difference was
found between the high and low product knowledge groups. The
mean of the high group was 23.2, the mean of the low group was
16.3. The t-value for the means test was -6.95 with a p-value <
.001, and the differences are in the predicted direction. Thus the
ﬁndings of 4a and 4b, provide strong support for the existence of a
positive relationship between product knowledge and awareness set
size. This was expected since awareness set size is thought to be a

type of product knowledge.

Hypothesis 5a: Outcome-relevant involvement has a positive effect

on awareness set size.

Hypothesis 5b: Subjects with high outcome-relevant involvement
have signiﬁcantly larger awareness sets than subjects with low

outcome-relevant involvement.

Both hypotheses are supported. For 5a, a signiﬁcant positive
relationship was found between outcome relevant involvement and
awareness set size, in the hypothesized direction. The correlation is

.2703, R2 =.0731, and the p—value is < .001.

110

For 5b, a signiﬁcant mean awareness set size difference was
found between the high and low outcome-relevant groups. The mean
of the high group was 21.6, the mean of the low group was 17.6.
The t-value for the means test was 3.70 with a p-value < .001, and
the difference was in the predicted direction. Thus both results
provide strong support for the existence of a positive outcome-

relevant involvement - awareness set size relationship.

Hypothesis 6a: Value-relevant involvement has a positive effect on

awareness set size.

Hypothesis 6b: Subjects with high value relevant involvement have
signiﬁcantly larger awareness sets than subjects with low value-

relevant involvement

Both hypotheses are supported. For 6a, a signiﬁcant positive
relationship was found between value-relevant involvement and
awareness set Size. The correlation between the two is .463, R2

=.078, and the p-value is < .001.

For 6b, a signiﬁcant mean awareness set size difference was
found between the high and low value-relevant groups. The mean of
the high group was 22.3, the mean of the low group was 16.8. The t-
value for the test was 5.36 with a p-value < .001, and the difference

occurs in the predicted direction.

111

The ﬁndings of both 6a and 6b, provide strong support for the
existence of a positive relationship between value-relevant
involvement and awareness set size. These results support the
reasoning that value involved subjects are likely to be familiar with
even the lesser known brands due to their maintenance of an ongoing

relationship with the product category.

Hypothesis 7: Value-relevant involvement has a stronger relationship

with awareness set size than outcome-relevant involvement.

This hypothesis is supported. A signiﬁcant difference was
found between the two types of involvement in terms of the strength
of their correlations with awareness set size. The correlation between
outcome relevant involvement and awareness set size was -.27, the
correlation between value-relevant involvement and product
knowledge was -.46. The Steiger (1980) formula yielded a t-value of
2.878, and a p—value of .004, and the difference occurs in the
predicted direction. Thus the results of Hypothesis 7 indicate that
value-relevant involvement is more strongly related to awareness set
size than outcome-relevant involvement. The primary reason for this
difference is believed to be the signiﬁcant difference in the strength of
their relationships with product knowledge (see Hypothesis 3) which
has been shown to be a signiﬁcant antecedent of awareness set size

(see Hypotheses 4a and 4b).

112

Hypothesis 8a: Product knowledge has a positive effect on the

number of salient attributes.

Hypothesis 8b: Subjects with high product knowledge have more

salient attributes than subjects with low product knowledge.

Hypothesis 8a is not supported. While a positive correlation
of .1012 was found between the two variables (R2 =.01), the p-value

of .17 indicates that the relationship is not signiﬁcant.

Hypothesis 8b is supported. At the .05 level, a signiﬁcant
difference in the mean number of attributes was found between the
high and low product knowledge groups. The mean of the high group
was 9.3, the mean of the low group was 8.7. The t-value for the test
was
-2.08 with a p-value of .039, and the difference was in the predicted

direction.

These ﬁndings provide some support for the existence of a
positive relationship between product knowledge and the number of
salient attributes. Even though the sample correlation in 8a was
insigniﬁcant, it was in the hypothesized direction. More convincing
were the results of the t-test, which although less powerful than the

correlational test, did produce signiﬁcant results.

I

113

The inconsistency in the results of the two tests is believed to be
due to possible non linearity in the relationship between product
knowledge and the number of attributes. This possibility was examined
by breaking down product knowledge into quartiles, and looking at the
pattern of the means to see if they deviated from linearity. The results
of this breakdown are shown below:

P. K. Quartile Mean # of salient attributes

1 (low) 8.89
2 8.43
3 9.48
4 (high) 9.08

The breakdown does appear to show some evidence of non-
linearity in the relationship between product knowledge and the number
of salient attributes (e. g.; Q3>Q4, Q1>Q2). This pattern of results is
similar to the inverted-U relationship found between product knowledge
and the number of salient attributes inprevious research (Johnson and
Russo 1984). Using this quartile breakdown an Eta-square - R-square
test was conducted to see if there is a signiﬁcant non-linear relationship
between product knowlwdge and the number of salient attributes. The
signiﬁcance test produced an F-ratio of 2.41 which is insifgniﬁcant at
the .05 level. Thus it is concluded that the t-test does provide limited
evidence of a weak positive relationship, but that the results of the linear
and non-linear association tests indicate that the relationsip is not very

strong or consistent across different levels of product knolwedge.

 

114

Hypothesis 9a: Outcome-relevant involvement has a positive effect

on the number of salient evaluative attributes.

Hypothesis 9b: Subjects with high outcome-relevant involvement
have signiﬁcantly more salient attributes than subjects with low

outcome-relevant involvement

Hypothesis 9a is not supported. Although the relationship
between the variables is in the predicted direction, the correlation is

only .09, R2 =.01, and the p-value is .227, and thus is insigniﬁcant.

Hypothesis 9b is supported. At the .05 level, a signiﬁcant
difference in the mean number of attributes was found between the
high and low outcome relevant involvement groups. The mean of
the high group was 9.3, the mean of the low group was 8.7. The t-
value for the means test is 2.02 with a p-value of .045, and the

difference is in the hypothesized direction.

The results of the statistical tests above mirror the results of
Hypotheses 8a and 8b; an insigniﬁcant correlation in the
hypothesized direction and a signiﬁcant difference between the high
and low groups. Thus, like in the previous set of hypotheses, it was
thought that the inconsistency in the results of the two tests could also
be due to non linearity. This possibility was examined by breaking

down outcome-relevant involvement into quartiles, and looking at the

 

115

means to see if they deviated ﬁ'om linearity. The results of this
breakdown are shown below:

0. R. Involvement Quartile Mean # of Salient Attributes

1 (high) 9.22
2 9.38
3 8.51
4 (low) 8.79

The breakdown does appear to show some evidence of non-
linearity in the relationship between outcome-relevant involvement and
the number of salient attributes (Q2 > Q1, Q4 > Q3). Once again an
Eta-square - R-Square test of nonlinearity was performed and once
again the results were insiginicant (F = 1.91, p> .05). Thus it is
concluded that limited evidence of a weak positive relationship is
provided by the t-test, but that the results of the linear and non-linear
association tests indicate that the relationsip is not very strong or

consistent across different levels of outcome-relevant involvement.

Hypothesis 10a: Value-relevant involvement has a positive effect on the

number of salient attributes.

Hypothesis 10b: Subjects with high value-relevant involvement have
signiﬁcantly more salient attributes than those subjects with low value-

relevant involvement

 

116

Both hypotheses are supported. For 10a, a signiﬁcant positive
relationship was found between value-relevant involvement and the
number of salient attributes. The correlation between the two is .25,

R2 is .063, and the p-value is < .001.

For 10b, a signiﬁcant difference in the mean number of
attributes was found between the high and low value-relevant
involvement groups. The mean of the high group was 8.6, the mean
of the low group was 9.3. The t-value for the means test is 2.09 with
a p-value of .038, and the difference is in the hypothesized direction.
Thus the ﬁndings of both tests provide clear support for the existence
of a positive relationship between value-relevant involvement and the

number of salient attributes.

Hypothesis 11: Outcome-relevant involvement has a stronger
relationship with the number of salient evaluative attributes than

value-relevant involvement.

The hypothesis is not supported. While a signiﬁcant
difference was found between the two types of involvement in terms
of the number of salient attributes, the direction of the difference was
opposite of what was hypothesized. The correlation between outcome
relevant involvement and the number of salient attributes was -.09,
the correlation between value-relevant involvement and the number of

attributes was -.25. The Steiger (1980) formula yielded a t-value of

117

2.02, and a p-value of .043. Thus value relevant involvement was
actually found to have a signiﬁcantly stronger relationship with the

number of salient attributes than outcome-relevant involvement.

It had been thought that outcome-relevant involvement would
relate more strongly to the number of salient attributes because it
would provide more motivation for making thorough evaluations of
alternatives. The fact that the results indicate value-relevant
involvement relates more strongly could be due to greater awareness
and understanding of the attributes resulting from superior product

knowledge (see Hypothesis 3).

Hypothesis 12a: Value-relevant involvement has a positive effect on

shopping enthusiasm.

Hypothesis 12b: Subjects with high value-relevant involvement have
signiﬁcantly greater shopping enthusiasm than subjects with low

value-relevant involvement

Both hypotheses are supported. In 12a, a signiﬁcant positive
relationship was found between value-relevant involvement and
shopping enthusiasm. The correlation between the two is .65, R2

=.423, and the p-value is < .001.

 

118

In 12b a signiﬁcant mean difference in shopping enthusiasm
was found between the high and low value-relevant involvement
groups. The mean of the high group was 30.5, the mean of the low
group was 34.9. The t-value for the means test is -7.92 with a p—
value < .001, and the difference is in the hypothesized direction.
Thus very strong support for the hypothesized positive relationship
between value-relevant involvement and shopping enthusiasm was

found.

 

Hypothesis 13a: Outcome-relevant involvement has a negative effect

on shopping enthusiasm.

Hypothesis 13b: Subjects with high outcome-relevant involvement
have signiﬁcantly less shopping enthusiasm than subjects with low

outcome-relevant involvement

Neither hypothesis is supported. While there is a signiﬁcant
relationship between outcome-relevant involvement and shopping
enthusiasm, it is not in the hypothesized negative direction. The
correlation between the two variables is .29, R2 =.084, and the p—

value is < .001 thus indicating a strong positive relationship.

Similarly the t-test found a signiﬁcant mean difference in
shopping enthusiasm between the high and low outcome-relevant

involvement groups, however the difference was not in the

119

hypothesized direction. The mean of the high group was 28.1, the
mean of the low group was 37.4. The t-value for the means test is -

3.29 with a p-value of .001.

The results of these two hypothesis tests provide strong
evidence that a positive relationship exists between outcome-relevant
involvement and shopping enthusiasm. It had been hypothesized that
the relationship would be negative because outcome-relevant involved
people would be more driven by extrinsic motivation (achievement of
a satisfactory outcome) and thus ﬁnd the task itself to be less

enjoyable and more stressful.

Possible reasons why a signiﬁcant positive relationship was

found include the following:

1) The drawing used to increase outcome-relevant involvement
may have eased the stress of the purchase decision for the subjects by
creating a no—lose situation. Instead of being concerned about the
consequences of misspending their hard earned money, such subjects
may have been excited by the opportunity to win a free car stereo gift '
certiﬁcate and thus have been very enthusiastic about car stereo

shopping in that context.

2) A strong positive relationship between value-relevant and
outcome-relevant involvement (r = .3969) may indicate that value-

relevant involvement is an antecedent of outcome-relevant

120

involvement. This has been suggested by other researchers (Bloch
and Richins 1983), but not actually empirically tested. Since value-
relevant involvement has already been shown to have a strong positive
relationship with shOpping enthusiasm (see Hypotheses 12a and 12b),
the positive relationship found between outcome-relevant involvement
and Shopping enthusiasm may be largely the result of a spurious i

correlation mediated by value-relevant involvement. A test of this , F

 

explanation was made by calculating the partial correlation between 1.,
outcome-relevant involvement and shopping enthusiasm, controlling -.
for value-relevant involvement. The partial correlation between the

two was only .037 with a p-value of .55 thereby providing support

for the explanation that the relationship may be spurious. However

the sign of the partial correlation is positive which is contradictory to

the direction speciﬁed in the hypothesis.

3) Finally, giving subjects access to such extensive product
information (particularly product ratings) may have eased the
perceived difﬁculty of the shopping task for the those who initially
might have lacked conﬁdence and been most likely to ﬁnd the task
stressful. With such information they may have had conﬁdence that
they could make a satisfactory choice thereby increasing their

enthusiasm.

Hypothesis 14a: Outcome-relevant involvement has a negative effect

on the width of the latitudes of acceptance.

121

Hypothesis 14b: Subjects with high outcome-relevant involvement
have narrower latitudes of acceptance than subjects with low

outcome-relevant involvement.

Both hypotheses are supported. For 14a, a signiﬁcant
negative relationship was found between outcome-relevant
involvement and the width of the lattitudes of acceptance. The
correlation between the two is -.215, R2 =.046, and the p-value is

.004.

For Hypothesis 14b, a signiﬁcant mean difference in the
lattitudes of acceptance was found between the high and low outcome-
relevant involvement groups. The mean of the high group was 48.6,
the mean of the low group was 54.8. The t-value for the means test
is -3.51 with a p-value of .001, and the difference is in the
hypothesized direction. As a result both tests provide support for the
conclusion that outcome involved individuals are more demanding in

their attribute requirements.

Hypothesis 15: Outcome-relevant involvement has as stronger
negative effect on the width of the latitudes of acceptance than value-

relevant involvement.

122

This hypothesis is not supported. While a signiﬁcant
difference was found between the two types of involvement in terms .
of the strength of their negative relationships with lattitudes of
acceptance, the direction of the difference was not as hypothesized.
The results indicate that value relevant involvement has a signiﬁcantly
Stronger negative relationship with lattitudes of acceptance than
outcome relevant involvement. The correlation between outcome
relevant involvement and lattitudes of acceptance was -.17, the
correlation between value-relevant involvement and lattitudes of
acceptance was -.40. The Steiger (1980) formula produced a t-value

of 2.49, and a p-value of .014.

Thus the results indicate that value-relevant involved
consumers are actually more demanding than outcome relevant
involved consumers. One possible explanation for this might be that
the value-relevant involved consumers maintain an ongoing hobby
like interest in the product category. This ongoing interest could
likely result in strong commitments being made to beliefs about the
necessity of certain attributes. Outcome relevant involved consumers
do not necessarily have this background and thus may be less likely to

place too much importance on any one attribute.

Hypothesis 16a: Shopping enthusiasm has a positive effect on evoked

set size.

123

Hypothesis 16b: Subjects with high shopping enthusiasm have larger

evoked set Sizes than subjects with low shopping enthusiasm.

Hypothesis 16a is supported. A signiﬁcant positive
relationship was found between shopping enthusiasm and evoked set
size. The correlation between the two is .163, R2 =.027, and the p-

value is .028.

However hypothesis 16b is not supported. While the
differences are in the hypothesized direction, a signiﬁcant difference
does not exist between the mean evoked set Sizes of the low and high
shopping enthusiasm groups. The mean of the high group was 8.0,
the mean of the low group was 7.3. The t-value for this means test is

1.23, with a p-value of .220.

Despite the insigniﬁcant t-test, the data provides evidence that
a positive relationship exists between shopping enthusiasm and evoked
set size. It is believed that the discrepancy in the results of the two
tests is simply due to the fact that the relationship between the two
variables is not particularly strong, and that the t—test may have lacked

the power to detect it.

Hypothesis 17a: The number of salient evaluative attributes has a

negative effect on evoked set size.

124

Hypothesis 17b: Subjects with a high number of evaluative attributes
have smaller evoked set sizes than subjects with a low number of

evaluative attributes.

Neither hypothesis is supported. For 17a, a marginally
signiﬁcant positive relationship was found between the number of
salient attributes and evoked set size. The correlation between the

two variables is .144, (R2 =.021), and the p-value = .052.

For 17b, a signiﬁcant difference in mean evoked set size was
not found between the low and high number of salient attributes
groups. The mean of the high group was 7.9, the mean of the low
group was 7.3. The t-value for the means test is -1.18, the p-value
is .239, and the differences are not in the hypothesized direction.
Thus given the results of the two tests above do not provide sufﬁcient
evidence that a relationship exists between the number of salient

attributes and evoked set size.

Hypothesis 18a: The widths of the latitudes of acceptance have a

positive effect on evoked set size.

Hypothesis 18b: Subjects with wide latitudes of acceptance have
larger evoked set sizes than subjects with narrow latitudes of

acceptance.

125

Neither hypothesis is supported. In 18a, an insigniﬁcant
positive relationship was found between latitude of acceptance width
and evoked set size. The correlation between the two is .083, R2

=.007, and the p-value is .259.

In 18b, a signiﬁcant difference was not found between the
mean evoked set sizes of the wide and narrow lattitudes of acceptance
groups. The mean of the wide group was 7.5, the mean of the low
group was 7.8. The t—value for the means test is .54, the p-value is

.588, and the differences are not in the hypothesized direction.

Thus surprisingly, based on the results above, no evidence was
found that lattitudes of acceptance have a signiﬁcant effect on evoked
set size. The Simplest explanation for this lack of a relationship might
be that consumers do not adhere very closely to their own stated
selection criteria. Another, perhaps more plausible explanation, is
that differences in the price ranges subjects were willing to consider
may have worked to equalize the evoked sets of demanding (narrow
lattitudes) and non demanding (wide lattitudes) consumers. The non
demanding consumer may have found only a limited number of the
models to be practical due to a rather limited price range, while the
demanding consumer might have been more willing to consider the
higher priced models which offered the kind of performance he/she
demanded. Thus perhaps non demanding consumers were limited in
their choice alternatives by price, and demanding consumers were

limited by their more stringent performance requirements.

 

126

Hypothesis 19a: Awareness set size is positively related to evoked set

size.

Hypothesis 19b: Subjects with large awareness sets have larger

evoked sets than subjects with small awareness sets.

Neither hypothesis is supported. In 19a, an insigniﬁcant
positive relationship was found between awareness set size and
evoked set size. The correlation between the two is .026, R2 =.001,

and the p-value = .724.

In 19b, a signiﬁcant difference was not found between the
mean evoked set sizes of the low and high awareness set size groups.
The mean of the high group was 8.0, the mean of the low group was
7.3. The t-value for this means test was -1.16, the p-value is .247,
and the differences are in the hypothesized direction. Thus no
support is found for a relationship between awareness set size and
evoked set size. One obvious explanation for the lack of a signiﬁcant
relationship is that the checklist format of the evoked set measure
eliminated the necessity for subjects to make unaided recalls of which
brands to consider. Another possible explanation could be that the
”Consumer Reports style" product ratings provided to the subjects
may have given them enough conﬁdence to consider unfamiliar
brands. These attribute by attribute ratings of each model may have

convinced subjects that high rated, but unfamiliar brands, were

 

127

worthy of consideration. Without such ratings, unknowldgeable

people may have been more wary of considering unfamiliar brands.

Hypothesis 20a: Outcome-relevant involvement is negatively related

to evoked set Size.

Hypothesis 20b: Subjects with high outcome-relevant involvement
have Signiﬁcantly smaller evoked sets than subjects with low

outcome-relevant involvement.

Neither hypothesis is supported. For 20a, an insigniﬁcant
positive relationship was found between outcome-relevant
involvement and evoked set size. The correlation between the two is

.094, R2 =.009, and the p-value = .205.

For 20b, a signiﬁcant difference was not found between the
mean evoked set sizes of the low and high outcome-relevant
involvement groups. The mean of the high group was 8.0, the mean
of the low group was 7 .3. The t-value for this test is 1.23, the p-
value is .220, and the differences are not in the hypothesized

direction.

While this relationship was not signiﬁcant as hypothesized, the
results are not surprising given the results of the Hypotheses 17a

through 18b. Outcome-relevant involvement had been hypothesized

128

to affect evoked set size primarily by working through lattitudes of
acceptance and the number of salient attributes. However since both
of these variables had insigniﬁcant relationships with evoked set size,

so too did outcome-relevant involvement.

Hypothesis 21a: Value-relevant involvement has a positive effect on

evoked set size.

Hypothesis 21b: Subjects with high value-relevant involvement have
signiﬁcantly larger evoked sets than subjects with low value-relevant-

relevant involvement.

Hypothesis 21a is supported. A signiﬁcant positive
relationship was found between value-relevant involvement and
evoked set size. The correlation between the two is .162, R2 =.026,

and the p-value = .028.

Hypothesis 21b is not supported. A signiﬁcant difference was
not found between the mean evoked set sizes of the low and high
value-relevant groups. The mean of the high group was 8.0, the
mean of the low group was 7.3. The t-value for this test is 1.12, the

p-value is .263, and the differences are in the hypothesized direction.

Despite the insigniﬁcant t-test, the data does support the

conclusion that value-relevant involvement has a positive effect on

129

evoked set size. Similar to the interpretation of the ﬁndings in
Hypotheses 17a and 17b, it is believed that the discrepancy in the
results of the two tests is simply due to the fact that the relationship
between the two variables is not particularly strong, and that the t-test

may have lacked the power to detect it.

The positive relationship is supported despite the ﬁndings in

Hypotheses 14a, 14b , and 15 which indicate that value-involved

 

consumers are more demanding (narrow lattitudes of acceptance).
One explanation that might reconcile these two seemingly
contradictory results could be that value-relevant involvement may
cause consumers to be less price conscious and thus more likely to
consider the high priced/high performance alternatives that may be

out of the price range of less demanding consumers.

Hypothesis 22: The hypothesized path model explicated in Figure 1,
will provide a good ﬁt to the data (p> .1).

The hypothesis, as stated above is not supported. The path
model explicated in ﬁgure 1, produced a 12 value of 41.14 with 14
degrees of freedom. This x2 corresponds to a p-value < .005, thus
indicating the model fails to provide a good ﬁt to the data. The

results of the path analysis for this model are shown in Figure 3.

 

130
FIGURE 3

RESULTS OF PATH ANALYSIS:
NEWLY HYPOTHESIZED MODEL

 

 

 

\ /

 

 

OUTCOME-
RELEVANT
INVOLVEMENT

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PRODUCT
KNOWLEDGE
.08
AWARENESS SHOPPING WIDTH 01“ 4 OF SALIENT
SET SIZE ENTHUSIASM MWUDES 01“ ATTRIBUTES
- ACCEPTANCE
.07 .1 6 .03 .1 0
EVOKED
SET SIZE
x2 41.14
df 15

All ll

.01

131

FIGURE 4

RESULTS OF PATH ANALYSIS:
CURRENT LITERATURE MODEL

 

RELEVANT

OUTCOWﬂb
INVOLVEMENT

 

.26 .21

 

Y

PRODUCT
KNOWTEDGE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.02
AWARENESS WIDTH 0F c or SALIENT
SET SIZE LATITUDES OF A'I'I‘RIBUTES
ACCEPI‘ANCE
.01 .1 2 .04
Y
EVOKED
SET SIZE
x2 = 17.47
df = 7
p < .025

132

Hypothesis 23: The hypothesized path model explicated in Figure 1,
will provide a better ﬁt to the data than the path model derived from

the current literature (Figure 2).

This hypothesis is not supported. While the model derived
from the current literature also fails to provide a good ﬁt (x2: 17.47,
df =7 , p < .025) it does actually have a higher p-value than that for
the hypothesized path model. Thus no evidence is provided that the
hypothesized path model ﬁts the data better than the one derived from
the current literature. The results of this path analysis are shown in

Figure 5.

SUMMARY OF RESULTS

In all, 23 sets of hypotheses were tested that examined the
effects of two different types of involvement (outcome and value-
relevant) on the size of consumer evoked sets, working through a
number of mediating variables (product knowledge, lattitudes of
acceptance, number of salient attributes, awareness set size, and
shopping enthusiasm). Support was found for all ﬁve hypothesized
bivariate relationships that involved value-relevant involvement as an
antecedent. Three of the ﬁve hypothesized bivariate relationships
involving outcome-relevant involvement as an antecedent also
received support. Of the Six variables that were hypothesized to be
antecedents of evoked set size, only two, shopping enthusiasm and

value-relevant involvement, had signiﬁcant relationships. -

 

 

133

The results of the path analysis did not Show that the
hypothesized path model ﬁt the data better than the model derived
from the current literature. However, it is worth noting that the only
two variables in the study that were found to have a signiﬁcant effect
on the dependent variable (evoked set size), appeared only in the new
model. Thus while the current literature model was able to provide
accurate speciﬁcations of the effects of involvement on the mediating
variables, it failed to Specify any mediating variables which could
explain signiﬁcant variance in evoked set size. Thus, despite the fact
that the new model did not provide a better overall ﬁt to the data, the
results of the study provide justiﬁcation for the necessity of
considering the effects of both shopping enthusiasm and value-
relevant involvement, in the study of the antecedents of evoked set

size.

CHAPTER SIX
SUMMARY AND DISCUSSION

This chapter gives a summary of the research and the
conclusions that were reached. This chapter also discusses the
contributions of the study, its managerial implications, and its

limitations. In addition, future research directions are suggested.

 

RESEARCH SUMMARY

The current research study examined the effects that value-
relevant and outcome-relevant involvement have on evoked set size,
working through a number of mediating variables. A theoretical path
model of the hypothesized evoked set formation process was devised,
and measurement scales for each variable in this process were
developed from measures used in previous research. These scales
were put together in a questionnaire that speciﬁcally applied this

model toward the formation of evoked sets in the product category of

 

car SICICOS .

Data were collected, by administering the resulting

 

questionnaire to 182 upper division undergraduate college students.
Subjects were asked to assume they were currently in the market for a
car stereo, and to evaluate the acceptability of 28 different car stereo
models. Outcome relevant involvement was manipulated by tying the

evaluation of the car stereo models to an incentive drawing. This was

134

135

done by telling the subjects in the high outcome-relevant involvement
condition that the prize for the incentive drawing was a car stereo gift
certiﬁcate that could only be redeemed on one of the car stereos they

were asked to evaluate.

After the data were collected, path analyses were performed
on both the newly hypothesized model of the evoked set formation
process as well as the competing model derived from the current
literature. The results of the path analysis indicated that neither
provided a good ﬁt to the data. In addition individual tests were
made of the relationships between key variables in the process. Sets
of hypotheses regarding 21 different bivariate relationships were
tested, and 14 received statistically signiﬁcant support. While most
of the hypothesized relationships between the two involvement types
and the mediating variables were supported, shopping enthusiasm was
the only one of the four mediating variables hypothesized to have an
effect on evoked set size, that was actually found to have a signiﬁcant

relationship.

In summary, the major conceptual focus of the research was
an examination of the effects that different types of involvement have
on the evoked set formation process. Previous research had tended to
focus only on outcome-relevant involvement, however this study also
considered the potentially differential effects of value-relevant
involvement as well. The results indicated that several previously

neglected variables (value-relevant involvement, shopping

136

enthusiasm) had a much stronger effect on evoked set size than those
more frequently studied in previous research (outcome-relevant
involvement, awareness set size, number of attributes, lattitudes of

acceptance).

RESEARCH OBJECTIVES AND CONCLUSIONS

For the purpose of summary and conclusions, the 40 separate
hypotheses that were tested in the research Study can be condensed
into three major research objectives. These objectives were: 1) To
provide evidence showing that the two types of involvement have
differential effects on the process of forming evoked sets. 2) To
develop a good ﬁtting path model of the evoked set formation
process. 3) To provide justiﬁcation for the inclusion of two variables
previously neglected in studies of evoked sets, value-relevant

involvement and Shopping enthusiasm.

In terms of the ﬁrst objective, the research results do not
provide evidence that the two types of involvement have differential
effects on evoked set siZe nor for that matter any of the other
variables in the study. As a result, these ﬁndings alone will not
support the conclusion that the two types of involvement are
conceptually distinct. Instead the evidence more strongly supports the
perspective that value-relevant involvement is an antecedent of
Outcome-relevant involvement. The conceptual justiﬁcation for this
relationship is that one's interest in a product category might cause

them to become more concerned about the outcome of a purchase

 

 

137

decision regarding that product category. This relationship is
supported by the relatively high correlation of .39 between the two
types of involvement. In addition the two variables had parallel
correlations with all the other variables in the study, with value-
relevant involvement always having the stronger correlation (see table

5.2).

While it is still believed that the two types of involvement are
conceptually distinct, the study results do provide strong evidence that
value-relevant involvement is an antecedent of outcome-relevant
involvement. The parallel correlations seem to be a result of a
combination of this relationship and a failure to create a signiﬁcant
amount variation in outcome-relevant involvement with the treatment
manipulation. Comparison of mean manipulation check and outcome-
relevant involvement scale scores for the two treatment groups
indicate that relatively high levels of involvement were induced in
both groups. Thus it appears the main problem with the manipulation
was that the use of a fairly high involvement product, and a cash
drawing, created too much involvement in the low involvement
condition. As a result the only strong source of variation on
outcome-relevant involvement was the level of value-relevant
involvement one had. Thus the outcome-relevant involvement
measure in this study appears to be little more than an attenuated

version of value-relevant involvement.

 

 

138

While it is unfortunate that the manipulation was not more
effective, it does not totally nullify the usefulness Of the research in
regard to providing support for this objective. This is particularly
true when the results of this study are compared to the ﬁndings of
previous studies. One of the most notable ﬁndings of this study
actually tends to support the ﬁrst objective. This ﬁnding is that, as
hypothesized, value-relevant involvement had a signiﬁcant positive
effect on evoked set size. This is important because the accepted
view in the literature is that involvement has a negative effect on
evoked set size (Gensch and Javalgi 1987, Harrell 1986, Rothschild
and Houston 1977). Given that these previous studies only
operationalized outcome-relevant involvement, the difference in
results could be attributable to the hypothesized differential effects
that outcome and value relevant involvement have on evoked set size.
Thus while the ineffectiveness of the manipulation may explain why
some of the hypothesized results were not Obtained for outcome-
relevant involvement, it does not diminish the validity of the
hypothesized results that were obtained with regard to value-relevant
involvement. Thus it is believed that limited support for this
objective was found. This support is based on the differential effects
value relevant involvement had on evoked set size in comparison to
the effects that outcome relevant involvement had on evoked set size

in previous studies.

The second objective, to develop a good-ﬁtting model of the

evoked set process, was not met. Path analysis of the hypothesized

 

 

139

path model produced a )8- that was signiﬁcant at the .01 level.
However, it should be noted that if the model is revised by adding
three relatively rather minor linkages (minor in terms of the
conceptual focus of the paper), the x2 can be dropped to 6.10, at 12
degrees of freedom. This revised model produces a p-value < .25,
thereby indicating a good ﬁt. The three additions that were made to
the revised model are; a link from lattitudes of acceptance to the
number salient of attributes, a link from value-relevant involvement to
awareness set size and a link from value-relevant involvement to
lattitudes of acceptance. The results of the revised model are Shown

in Figure 5.

The link between lattitudes of acceptance and the number
salient of attributes is considered minor because it simply links two
mediating variables and hence does not involve either of the two
involvements nor evoked set size (the primary antecedent and
consequent variables in the model). The latter two additions are new
linkages between value-relevant involvement and two of the mediating
variables. They are considered minor since their inclusion simply
indicates that value-relevant involvement has greater inﬂuence over

the evoked set formation process than was initially proposed.

Conceptual justiﬁcation for the linkage between lattitudes of
acceptance and number of salient attributes can be justiﬁed by the fact
that previous research by Belonax and Javalgi (1989), and' Rothschild

and Houston (1977) has indicated that two variables are highly

 

 

 

 

 

 

140

FIGURE 5

RESULTS OF PATH ANALYSIS:
REVISED NEWLY HYPOTHESIZED MODEL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VALUE- OUTCOM-
RELEVANT RELEVANT
INVOLVEMENT INVOLVEMENT
. 32 .03
.64
PRODUCT
KNOWLEDGE
.48 .07
AWARENESS SHOPPING WIDTH OF -048 a or SALIENT .
SET SIZE ENTHUSIASM LATITUDES 0F ATTRIBUTES
, ACCEPTANCE
.07 .16 .03 10
EVOKED
SET SIZE

 

x2 6.10
df 12
.25

VII II

 

141

interrelated. Both of these studies combined the two variables
together into a surrogate measure of involvement (e. g. the Consumer
Involvement Matrix), and thus it is felt that the omission of this

linkage from the original model was simply an oversight.

The proposed direct positive link between value-relevant
involvement and awareness set size is thought to occur in addition to
the indirect positive effect mediated through product knowledge.

This indicates that value-involved consumers would have awareness
sets larger than expected even when their superior product knowledge
is taken into consideration (controlled for). Possible explanations for
this positive direct effect might be that the strong hobby like interest
of value involved consumers might result in greater memory
accessibility (car stereos might occupy a more central position in their
memory schemas) and/or it might give them stronger motivation to

retrieve previously stored information regarding car stereos.

The third revision was the link added between value-relevant
involvement and lattitudes of acceptance. This addition to the model
was inspired by the results of the test of Hypothesis 15. These
ﬁndings showed that value-relevant involvement has a signiﬁcantly
stronger negative effect on lattitudes of acceptance than outcome-
relevant involvement. This is noteworthy since Hypotheses 14a and
14b, had already established support for the existence of a Strong
negative relationship between outcome-relevant involvement and

lattitudes of acceptance. Thus, given the uncovering of this

 

EL

 

142

unexpectedly strong relationship, it was felt that the model should be
revised in order to better reﬂect the true nature of the relationships

between the variables.

While these revisions do alter the originally conceived path
model it Should be noted that none of these three additional links
contradict nor detract from the central focus of the research, that was
outlined in Chapter 2. AS a matter of fact, the addition of two new
links with value-relevant involvement as an antecedent actually
provides validation for one of the major research propositions: that
value-relevant involvement, although neglected by previous
researchers, plays a major role in the evoked set formation process.
Thus while the second objective was not met, the results of the path
analysis point to several revisions that could be made to the model

and investigated further in future research.

The third objective, establishing that value-relevant
involvement and shopping enthusiasm play a signiﬁcant role in the
evoked set formation process, was also achieved. Despite the fact
these two variables had been neglected in previous studies of the
evoked set process, they were the only two in the study to have a
signiﬁcant effect on evoked set size. In addition, value-relevant
involvement had a signiﬁcant effect on every other variable in the
study (see table 5.1) thereby indicating its critical role in driving the
process. Thus the results indicate that evoked sets are not‘ selected

just on the basis of a rational need for task simpliﬁcation, but rather

 

 

143

selection is also affected by one's interest in the product and affect for

the shopping process.

RESEARCH CONTRIBUTIONS

Conceptually this research has made several contributions.
First, in Chapter 2 a theoretical basis for synthesizing the many
different kinds of involvement in the literature into two basic forms of
the construct was presented. The classiﬁcation criterion that was
developed is based on an application of motivational theory to the
study of involvement. Under this classiﬁcation criterion involvement
types are categorized on the basis of whether they represent an

intrinsic or extrinsic motivational state.

A second conceptual contribution of the research was the
introduction of a new variable, shopping enthusiasm, to consumer
choice research. While the view that shopping can be a form of
leisure behavior has long been recognized and written about, the
inclusion of this new variable provides a means by which the extent
of one's affect toward the shopping process and its effect on
consumer choice decisions can be properly assessed. The
introduction of this variable is important because it provides a means
by which the experiential aspects of consumer behavior can be

considered in an information processing based study.

A third conceptual contribution of the research, was the

development of a theoretical path model of the evoked set process.

 

 

 

144

While a great deal of research has been done on various evoked set
antecedents and selection strategies, this research was the ﬁrst to

devise a comprehensive model of the overall process.

In terms of methodology, the ﬁrst contribution was the
development of a reliable measurement scale (alpha = .91) for the
new variable, shopping enthusiasm. Second, a measurement method
was developed for measuring lattitudes of acceptance for product
features. Previous studies had only developed measures for assessing
the lattitudes of acceptance for performance standards, but this
research extended this work by developing a scaling approach for
attributes which are characterized by being either present or absent

from a product.

A third methodological contribution was made by establishing
an approach for addressing one of the most hotly debated conceptual
issues in social psychology; the issue of whether value-relevant and
outcome-relevant involvement are actually separate constructs. While
the results of the current research were inconclusive, the approach
taken to resolve this issue, testing for differential effects in a
correlational experiment, establishes a methodology by which this

issue can be more conclusively resolved in future studies.

A ﬁnal contribution of the current research was it was the ﬁrst
to examine the involvement-evoked set size relationship uSing a

product category that features wide variations in the price and quality

 

 

 

145

of the purchase alternatives . Previously, Campbell (1968) had used
the product categories of toothpaste and detergent, while Jarvis and
Wilcox (1973) used coffee, napkins and detergent. All of these
categories are characterized by relative homogeneity in terms of the
price and quality of alternatives. While Belonax and Javalgi (1989)
actually used a fairly heterogeneous product category, microwave
ovens, they homogenized it by limiting the choice alternatives to six

and by not giving subjects price information.

Wide variability in price is considered to be an important
factor for several reasons. First, most durables are characterized by
large differences in price and quality, and thus this is important in
terms of extending the generalizability of existing theory. Second,
price/quality variability is thought to be a potential moderator of the
involvement-evoked set size relationship. This is reasoned as follows;
when high involvement subjects are allowed to select an evoked set
from a choice domain that includes a number of higher priced/higher
quality alternatives, their evoked sets are likely to be larger than
average. This is because their high interest in the product is likely to
make them less price sensitive, and more willing to consider paying a
higher price to get higher quality. If the choice domain is fairly
homogeneous in terms of price and quality, then high involvement
consumers are likely to have smaller than average evoked sets since

few alternatives will be able to meet their more stringent Standards .

 

 

 

146

MANAGERIAL IMPLICATIONS

The main managerial implications to come out of the current
research are centered around the ﬁndings regarding the effects of
value-relevant involvement on shopping enthusiasm, lattitudes of
acceptance and evoked set size. These ﬁndings indicate that
marketers may need to develop special strategies to sell to
enthusiastic/value—relevant involved shoppers, because they may
approach the shopping process much differently than other types of

consumers .

The first implication deals with the ﬁnding that value-relevant
involvement has a positive effect on evoked set size. This ﬁnding
contradicts the previous research which had indicated that
involvement negatively effects evoked set size. The negative
relationship had been explained by reasoning that the highly involved
customer is more demanding (narrower lattitudes of acceptance), and
thus tends to ﬁnd fewer brands worthy of consideration. The problem
with this reasoning is that it indicates small evoked sets result from an
inadequate retail selection, not because the high involvement

consumer actually wishes to reduce his or her choice options.

In the current research, high value-relevant involvement
consumers were given the opportunity to consider a large number of
models including many higher priced/higher performance brands.
The result was that the value-relevant involved subjects were more

likely to consider these high end models and thus tended to have

 

147

larger overall evoked sets. Thus the results of the current research
indicate that even though consumers with value-relevant involvement,
tend to be more selective, they still tend to have larger evoked sets

because they are more willing to consider higher priced brands.

The managerial implication of this ﬁnding for retailers is that
specialty retailers who cater to the value-relevant involved customer
might do better by concentrating on carrying a good selection of high
performance models from a variety of different manufacturers, rather
than carrying the full product lines of a limited number of
manufacturers. For manufacturers the implication is that for value-
relevant involved customers, performance/quality is a more important
determinant of evoked set composition than price (although price may
play a role in deciding which model in the evoked set eventually gets
purchased). For consumers with low value-relevant involvement
price is more likely to be an important determinant of evoked set
composition and thus identifying the appropriate reference prices and
price thresholds is critical if the manufacturer is going to get this

group to seriously consider its products.

A second managerial implication to arise from the research is
the need for marketers to consider the effects of shopping enthusiasm .
on the shopping process. Enthusiastic shoppers are unlikely to want
the process to be simpliﬁed for them, and thus would likely favor
stores with a large number of alternatives available for their

consideration. large superstores would likely appeal to this group

 

148

provided they carry a sufﬁcient amount of high quality merchandise.
For small retailers, attracting such consumers might require a strategy
of concentrated depth rather than superﬁcial breadth. In other words
they would likely be better off concentrating their product offerings in
speciﬁc sub-categories of the market rather than providing minimal
coverage of the entire product category. For example a small record
shop would probably do better if they had an extensive inventory of
jazz records than if they only had the top sellers across several

musical genres.

Unenthusiastic shoppers on the other hand are more likely to
want to want the process simpliﬁed for them. As a result, retailers
who make things easy for these customers are more likely to get their
business. For example many car dealers are simplifying the process
of car shopping by going to a one price system. This has great appeal
to the unenthusiastic car shoppers who do not enjoy negotiating
prices. Home shopping networks on cable TV have made the process
very easy for those who do not like to shop for products like clothing,
jewelry and home electronics. Selection is very limited, but this
actually makes the decision process much easier (particularly since
most have liberal return policies). Thus selection seems to be less
important to these Shoppers than the promise of getting a satisfactory

purchase outcome, with a minimum of effort.

Personal selling is one way larger retailers can Simplify the

process for unenthusiastic shoppers. These customers may be

 

 

149

intimidated by the large selections of superstores, and thus would
likely welcome the advice of a salesperson to help them select an
evoked set. The role of a salesperson in this context would be to
determine the customer's needs and then ﬁnd a set of alternatives that
meet their requirements and from which they can make a satisfactory

ﬁnal purchase choice.

RESEARCH LIMITATIONS

The ﬁrst limitation in the research was that only one product
class was examined, therefore it is possible that the results found are
idiosyncratic to car stereos. There is no guarantee that the
relationships between variables that were observed in this Study would

hold true for a different product category.

A second limitation was the use of students as subjects. Such
subjects were used because their high degree of homogeneity was
considered to be desirable due to its positive effects on statistical
power and internal validity. However by relying totally on the
responses of student subjects the external validity of the ﬁndings are

limited.

A third limitation of the current research is that subjects were
not actual buyers of car stereos and thus the purchase scenario may
have been too contrived and artiﬁcial. It was hoped however that the
use of an incentive drawing may have increased the realness of the

experimental situation by providing subjects with a context in which

 

 

 

150

they might actually become a buyer. However it is recognized that a
paper and pencil simulation of a purchase situation will not be

isomorphic with actual purchasing behavior.

A ﬁnal limitation was the weakness of the outcome-relevant
involvement manipulation. While variation was still achieved on this
variable, it is felt that much of this came from a suspected antecedent,
value-relevant involvement. As a result the effects that outcome-
relevant involvement has on the showing process that are independent
of value-relevant involvement may not have been adequately

introduced into the study.

SUGGESTIONS FOR FUTURE RESEARCH

Based on the last limitation listed above, it is felt that the best
way to extend the current research would be to replicate the
experiment with treatments that would more effectively manipulate
outcome-relevant involvement. It is felt that the best way to do this
would be to put subjects in high and low involvement purchase
situations rather than trying to manipulate their involvement by using
incentives. One way the outcome-relevant involvement of a purchase
situation could be varied is by manipulating the perceived visibility of

the purchase outcome.

Another area for future research might be directly testing to
see if price variability has a moderating effect on evoked Set Size.

One of the ﬁndings of the current research was that value—involved

 

 

151

customers appeared to be less price conscious in the formation of their
evoked sets. This could be more directly tested by making the price
variability of the choice domain one of the factors in a similar

experiment.

Another suggestion for future research would be to compare
the relative effects of the two types of involvement on purchase
decisions in different kinds of product categories. For example it
might be hypothesized that value-relevant involvement has a greater
inﬂuence on the decision making process for purchases of highly
experiential products like movies, music, and computer games, than
outcome-relevant involvement. On the other hand, outcome-relevant
involvement might have more inﬂuence on purchases of instrumental
products like tools, appliances and computer hardware. By
comparing their relative effects across product categories the potential
moderating effect of product type on the involvement-evoked set

relationship can be addressed.

 

 

 

 

APPENDICES

 

 

 

APPENDIX A:
INSTRUCTIONS FOR HIGH OUTCOME RELEVANT
INVOLVEMENT CONDITION

Thank you for your participation in this research study. The
purpose of this study is to collect data on consumer preferences
regarding car stereos. This data will be used by a Michigan based
home electronics retailer to help them decide what models of car
stereo they Should carry in their stores. Because this data is being
used to help make important business decisions, it is critical that you
answer all questions as realistically and truthfully as possible. You
have been selected as part of a limited sample and thus your opinions
are likely to have a great deal of inﬂuence on these decisions.

The primary task you will be asked to perform as part of your
participation in this research will be to evaluate the acceptability, to
you, of 28 different models of car stereo. To assist you in making
these determinations you have been proVided with the accompanying
Product and Attribute Information Booklet. This booklet consistsof
two types of information: descriptions of different attributes that you
might use to evaluate car stereos, and ratings showing how each of
the 28 models of car stereo you are to consider rates on each of these
attributes. After looking over this information booklet, you will be
asked to identify which of the 28 different models you would consider
for purchase. You may select as many or as few as you like.

In exchange for your participation in this research your name
will be placed, along with the names of the other participants, into a
drawing for three $100 car Stereo gift certiﬁcates. These gift
certiﬁcates may be used toward the purchase a new car stereo at the
home electronic stores that is sponsoring this research. The models of
car stereo you will be asked to evaluate represent all the models that
are carried by this retailer, and thus represent the only models you
can redeem your gift certiﬁcate on should you win. You may redeem
your certiﬁcate on any of the 28 models of car stereo carried by the
retailer however you will be responsible for paying (or ﬁnancing) any
amount over $100.

One of the primary purposes of the incentive drawing is to
motivate you to perform the product evaluations in a manner similar
to what you would do in an actual purchase situation. You are
encouraged to give serious and thoughtful consideration to your
evaluations of the 28 models of car stereo since you might want to

152

 

 

 

 

153

actually use these evaluations to help you select a car stereo in the
event you do win one of the gift certiﬁcates.

Because of the importance of this research it is hoped that you
will evaluate the different models of car stereo as realistically and as
carefully as you would if you were actually making such a purchase
decision. The only way this retailer can obtain accurate information
on your preferences is for you to treat the simulated purchase
situation as being real as possible and to give responses that reﬂect the
decisions you would make if you were actually shopping for a car
stereo.

 

 

154

APPENDIX B:
INSTRUCTIONS FOR LOW OUTCOME RELEVANT
INVOLVEMENT CONDITION

Thank you for your participation in this research study. The
purpose of this study is to collect data on consumer preferences
regarding car stereos for a Michigan based home electronics retailer.
In particular we are interested in what type of car stereo college
students are interested in buying.

The primary task you will be asked to perform as part of your
participation in this research will be to evaluate the acceptability, to
you, of 28 different models of car stereo. To assist you in making
these determinations you have been provided with the accompanying
Product and Attribute Information Booklet. This booklet consists of
two types of information: descriptions of different attributes that you
might use to evaluate car stereos, and ratings showing how each of
the 28 models of car stereo you are to consider rates on each of these
attributes. After looking over this information booklet, you will be
asked to identify which of the 28 different models you would consider
for purchase. You may select as many or as few as you like.

In exchange for your participation in this research your name
will be placed, along with the names of the other participants, into a
drawing for three $100 cash prizes.

Please evaluate the different models of car stereo as
realistically and as carefully as you would if you were actually
making such a purchase decision since this is the only way we can
obtain accurate information on your preferences.

 

 

155

APPENDIX C:
PRETEST QUESTIONNAIRE

QUESTIONNAIRE 1

CAR STEREO RESEARCH STUDY

Department Of Marketing And Hospitality Services Administration
Central Michigan University

1.AreYou: 1. male 2. female

I. For each car stereo manufacturer listed below please circle the response that best represents your
degree of familiarity with that manufacturer.

DEFINITELY SOMEWHAT
FAMILIAR FAMILIAR UNFAMILIAR
2. CLARION 1 2 3 1
3. JVC 1 2 3
4. FUTURA 1 2 3
5. BLAUPUNKT 1 2 3
6. KENWOOD 1 2 3
7. PIONEER 1 2 3
8. SANYO 1 2 3
9. SHERWOOD 1 2 3
10. PM 1 2 3
11. SONY 1 2 3
12. ALPINE 1 2 3
13. ACUVOLT 1 2 3
14. JENSEN 1 2 3
15. PANASONIC 1 2 3
16. TANAKA 1 2 3
17. AUDIOVOX 1 2 3
18. CRAIG 1 2 3
19. MOWEN 1 2 3
20. KRACO 1 2 3
21. AIWA 1 2 3
22. WOLFE 1 2 3
23. DENON 1 2 3
24. TARGA 1 2 3
25. FULTRON 1 2 3

 

 

156

26. OMEGA 1 2 3
27. NAKAMICHI I 2 3
28. SENTREK 1 2 3

II. The purpose of this next set of questions is to measure your interest or involvement with the
product category of car stereos. To take this measure we need you to judge car stereos against a
series of descriptive items according to how YOU perceive car stereos. Please circle the response that
best represents your true feelings about car stereos.

Important
1. Be sure that you respond to every item, please do not omit any.
2. Please do not circle more than one response on any single item..

Make each item a separate and independent judgment. Work at a fairly high Speed through this set of
questions. Do not worry or puzzle over individual items. It is your ﬁrst impressions, the immediate
feelings that we want. On the other hand, please do not be careless, because we want your true
impressions.

CAR STEREOS...
29. Boring To Me 1 2 3 4 5 6 7 Interesting To Me
30. Exciting To Me 1 2 3 4 5 6 7 Unexciting To Me
31. Appealing To Me I 2 3 4 5 6 7 Unappealing To Me
32. Fascinating To Me 1 2 3 4 5 6 7 Mundane To Me
33. Involving To Me 1 2 3 4 5 6 7 Uninvolving To Me
34. Important To Me 1 2 3 4 5 6 7 Unimportant To Me
35. Relevant To Me 1 2 3 4 5 6 7 Irrelevant To Me
36. Worthless To Me 1 2 3 4 5 6 7 Valuable To Me
37. Means Nothing To Me 1 2 3 4 5 6 7 Means A Lot To Me
38. Not Needed By Me 1 2 3 4 5 6 7 Needed By Me

111. Following are a series of items regarding your attitudes toward car stereos. For each item please
indicate whether you strongly agree, agree, neither agree nor disagree, disagree, or strongly disagree
with the statement. There are no right or wrong answers: we only want to know what you think.
neither
strongly agree nor strongly
agree agree disagree disagree disagree
39. I could talk about car stereos
for a long time. 1 2 3 4 5

40. I am interested in car stereos. 1 2 3 4 5

41. I deﬁnitely have a ”wanting”
for a new car stereo. 1 2 3 4 5

 

 

 

42.

43.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54

I rate car stereos as being of
the highest importance to me
personally.

A new car stereo would help
to deﬁne and express the “I"
and "me" within myself.

Because of my personal values,
I feel that car Stereos is a
product that ought to be
important to me.

Car Stereos offer me relaxation
and fun when life's pressures
build up.

Car stereos are nothing
more than appliances. '

Listening to music is one
way I often use to relieve
daily pressures .

I get bored when other people
talk to me about their car '
stereos.

I have little or no interest in
music.

I enjoy talking about car stereo
equipment with my friends.

It is natural that young people
become interested in car stereos.

I do not pay much attention to
advertisements for car stereo
equipment.

It is worth the extra cost to
own a car stereo that is
attention getting.

. I would not like to think of

my car stereo equipment as
being ordinary.

157

strongly
agree

agree

neither
agree nor strongly
disagree disagree disagree

3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5
3 4 5

 

 

158

neither
strongly agree nor strongly
agree agree disagree disagree disagree

55. I prefer to own a car stereo

with a strong personality
of its own. 1 2 3 4 5

IV. The following questions are designed to assess how much you know about car stereos. For each
question circle the response that you think is correct.

56.

a. 9.9.9:.»

999‘.”

LII

999'?“

999'?

.8

9957'!”

O\
r—s

e-sssrss

7.

8.

Which of the following speciﬁcations indicates the ability of a tape deck to reproduce the
audible range.

tape deck quieting

tape deck frequency response

tape deck signal to noise ratio

tape deck sensitivity

Which of the following speciﬁcations indicates the ability of an FM receiver to pick up weak
broadcast signals and stations from far away markets.

FM quieting

FM frequency response

FM signal to noise ratio

FM sensitivity

Which of the following speciﬁcations indicates the amount of background noise being produced
by a car Stereo's tape deck?

tape deck wow and ﬂutter

tape deck frequency response

tape deck signal to noise ratio

tape deck quieting

Which of the following speciﬁcations indicates the accuracy of a tape deck's playback speed and
indicate the consistency of the playback mechanism.

tape deck wow and ﬂutter

tape deck frequency response

tape deck signal to noise ratio

tape deck quieting

Which of the following speciﬁcations is used to indicate how strong an FM signal must be to
provide noise free reception.

FM quieting

FM frequency response

FM signal to noise ratio

FM sensitivity

 

 

. Which of the following speciﬁcations indicates how loud a car Stereo can be played.

power
wow and ﬂutter
sensitivity
frequency response

159

62. Which feature gives you the ability to select the correct setting for playing back metal or
chromium dioxide tapes.

music search

adjustable bias control

Dolby noise reduction

radio/tape switch

e-sssrs»

05
U)

. Which feature gives you the ability to virtually eliminate background noise during playback of
tapes.
music search
adjustable bias control
Dolby noise reduction
separate bass and treble controls
radio/ tape switch

9.9-9.7!”

.8

Which car stereo feature gives the listener the ability to listen to the radio while rewinding or
fast forwarding a tape?

music search

adjustable bias control

separate bass and treble controls

radio/tape switch

 

9953'.”

65. Which car stereo feature gives the listener the ability automatically fast
forward or rewind to the beginning of the next song on a tape, at the touch of

a button.
a. music search
b. adjustable bias control
c. separate bass and treble controls
d. radio/tape switch

V. For the next two questions please circle the response that best represents your opinion regarding
the amount of knowledge you feel you have regarding car stereos. There are no right or wrong
answers: we only want you to answer the questions as truthfully as you can.

66. Rate your knowledge of car stereos as compared to the average person. I am

one of the LEAST 1 2 3 4 5 6 7 one Ofthe MOST
knowledgeable knowledgeable

people people

67. Circle one of the numbers below to describe your familiarity with car stereos.

not atall familiar 1 2 3 4 5 6 7 extremely familiar

 

160

IV. For the next two questions please circle the response that best represents your opinion. There are
no right or wrong answers: we only want you to answer the questions as truthfully as you can.

very
likely likely
68. How likely do you think it is
that you will purchase a car stereo
within the next three years? 1 2
69. Assuming you could afford very
to buy a new car stereo now, likely likely

how likely do you think it is
that you would purchase one. 1 2

unlikely

unlikely

very
unlikely

very
unlikely

 

 

 

161

APPENDIX D:
PRODUCT AND A'I'I'RIBUTE INFORMATION BOOKLET

PART 1: ATTRIBUTE DESCRIPTIONS

There are two types of attributes listed below: speciﬁcations and product features. Speciﬁcations are
statistics or measurements that indicate how good or bad a particular car stereo performs in speciﬁc
areas. Taken together these Speciﬁcations indicate the overall level of sound quality a car stereo is
capable of providing. Product features are car stereo functions or options which are either present or
absent from a particular model of car stereo. These features may enhance audio performance, make
the unit easier to use, or provide additional beneﬁts.

A. SPECIFICATIONS - indicators of car stereo performance

1. POWER - Power gives you the ability to play music at a higher volume level without
unpleasant distortion. For a basic system, a low power (3-4 watts) car stereo unit will produce good
results, if efﬁcient Speakers are used. A high power (10-15 watts) unit may be needed if you want to
play music very loud, want extra bass, or want to drive larger or inefﬁcient speakers. If you select a
low power car Stereo unit the power can always be upgraded by hooking it up to an external
ampliﬁer.

2. AM SENSITIVITY - AM sensitivity indicates the ability of a receiver to pick up AM stations
with weak signals. Thus this is the primary indicator of the strength of the AM receiver and is most
important to people who want to pick up weak stations or who live in fringe areas away from
broadcast signals. .

3. FM SENSITIVITY - FM sensitivity indicates the ability of a receiver to pick up FM stations
with weak signals. Thus this is the primary indicator of the strength of the FM receiver and is most
important to people who want to pick up weak Stations or who live in fringe areas away from
broadcast signals.

4. FM QUIETING - FM quieting indicates how strong the FM Signal has to be for the receiver to
provide clear noise-ﬂee reception. Thus this is the primary indicator of the clarity of the FM

reception.

5. FM STEREO SEPARATION - Stereo separation indicates the ability of a receiver to separate
the left and right channels of a stereo broadcast. Thus this is the primary indicator of the
quality (or ﬁdelity) of the FM stereo sound.

6. TAPE DECK FREQUENCY RESPONSE - The frequency response indicates the ability of a
tape (led: to reproduce the audible range (from 20 to 20,000 hertz for the average human ear). The
wider the range of sound a tape deck is able to reproduce the greater the sound ﬁdelity. Thus this is
theprimaryindicatorofthecapability ofatapedeckto reproducehigh quality sound.

7. TAPE DECK WOW AND FLUTTER - Wow and ﬂutter is the accuracy of a tape deck's
playback speed. Wow is distortion caused by low-frequency speed ﬂuctuation. Flutter is caused by
high-frequency speed changes. The less wow and ﬂutter the more precise the playback mechanism is.
Thus this is the primary indicator of the consistency of the sound produced-by the tape deck.

8. TAPE DECK SIGNAL TO NOISE RATIO - The signal to noise ratio tells you how much
background noise is being produced by comparing the strength of the desired signal (music) to the

 

162

level of the background noise. The bigger this ratio the less background noise is audible. Thus this
is the primary indicator of the amount of audible distortion produced during tape playback.

B. PRODUCT FEATURES - optional functions that add to the utility of a car stereo

1. SEPARATE BASS AND TREBLE CONTROLS - separate controls for bass and treble provide
greater control over the sound than can be obtained with a single tone control.

2. ANTI-THEFT PROTECTION - this helps to prevent your car stereo from being stolen from an
unattended car. There are three types of anti theft protection currently available:

a. Security Code - This requires the correct entry of a code number into the car stereo unit to enable
it to play. This also features a ﬂashing LED message to warn thieves that the receiver is code
protected.

 

b. Removable Chassis - This allows you to remove the entire car stereo unit and carry it off with you
when you leave the car.

c. Detachable Face - This feature allows you to remove the face or the control panel of the car stereo
unit and carry it off with you when you leave the car.

The rest of the unit is rendered worthless without the control face, and thus not of any value to
thieves.

Comparative Rankings Of The Three Theft Protection Systems
l=best to 3=worst

 

 

 

Theft Protection System Cost Level Of Theft Ease Of
- Protection Use
security code 1* 3 I
removable chassis 2 1 3
detachable face 3 2 2

 

 

 

 

 

 

 

" 1 = cheapest system

 

3. CD CHANGER CONTROLS - this offers dual function controls which will allow one to directly
hook up and operate a compatible multiple-CD changer through the base unit. While CD changers
can still be hooked up to car stereo units without this feature, doing so requires the purchase of
additional equipment (such as a CD/FM tuner interface, a CD remote control pad, and a CD antenna)
to make it compatible and the frequency response is not as high.

4. AUTO MUSIC SEARCH - the ability to automatically fast forward or rewind the tape to the next
song, at the touch of a button. This is done through a mechanism that locates the silent gap between
songs.

5. RADIO/TAPE SWITCH - this allows you to change from playing the tuner to playing a tape with
the touch of a button. The primary advantage of this feature is that enables you to listen to the radio
while rewinding or fast-forwarding a tape.

6. POWER OF EJECT - this automatically ejects a tape from the car stereo unit whenever power is
shutoff to the unit. This protects both the tape and the mechanism by releasing tension.

 

163

7. ADJUSTABLE BIAS CONTROL - this gives you the ability to playback high bias tapes. High
bias tapes (Type II and IV ) have the potential to deliver a wider dynamic range with less distortion
than standard Type I tapes, if proper techniques were used to record the music.

8. DOLBY NOISE REDUCTION - this system decodes Dolby encoded tapes during playback to
virtually eliminate background noise (i.e. tape hiss). The noise reduction will only work on tapes

encoded with Dolby during recording. Dolby B is the most widely used system, but the newer Dolby
C is generally considered to be more effective.

 

 

164

PART 2: PRODUCT RATINGS

Listed below are proﬁles of the 28 different brands of car stereo you are to consider. The brands are
listed in order of price. The prices reﬂect typical selling prices at the sponsor's retail outlets. The
speciﬁcations are presented in the form of ratings. These ratings are on a l to 7 scale, with 1
indicating well above average performance on that speciﬁcation, 4 indicating average performance,
and 7 indicating well below average performance. In terms of features ”y ”indicates that a particular
model of car stereo has a feature, 'n" indicates the feature is not present. In terms of anti-theft
protection, code signiﬁes security code, chassis signiﬁes remOvable chassis, and face signiﬁes
detachable face.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l. Clarion 2. WC 3. Sanyo 4. Pioneer
$5200 KSR125 MARIOOO KE2828
price 99 99 99 129
SPECIFICATION
RATINGS
l-well above average
7-well below average
power - 7 7 7 7
AM sensitivity 4 5 7 4
FM sensitivity 4 6 7 3
FM separation 4 6 7 1
FM quieting 4 6 6 3
tape frequency response 4 7 3 6
tape wow & ﬂutter 3 4 4 5
tape signal/noise ratio 6 7 5 7
PRODUCT FEATURES
separate treble and bass n n n n
anti-theft n n n n
CD-changer controls n n n n
tape music search n n n n
radio/tape switch n n n n
adjustable bias control n n n n
dolby noise reduction n n n n

 

 

 

 

 

I l

165

PRODUCT RATINGS (CONTINUED)

5. JVC 6. Kenwood
KSR47O KRC1005
149 149
SPECIFICATION
RATINGS
l-well above average
7-well below

AM

FM

FM
FM

wow & ﬂutter
ratio
PRODUCT FEATURES
treble and bass
anti-theft
controls
music search
switch n
bias control n
noise reduction n n

\I-hQQOKO‘Aw
G$¥NNUI§O~

9. Audiovox 10. Blauplunkt
AV238
159 169
SPECIFICATION
RATINGS
l-well above average
7-well below

AM

FM

FM
FM

wow & ﬂutter
ratio

UtUIUtUJ-h-b-hN
(It-hANO’sAAO‘

PR UCT FEA
treble and bass
anti-theft“
controls
music search

:3

control
noise reduction

7. Sanyo
MAR120
149

th‘GO‘OUIA

NNMOMOGN

8. Sherwood
XR4127

159

O‘NNND—on—e

II

12.

Pioneer

229

mmAu—‘mwh

 

166
PRODUCT RATINGS (CONTINUED)

13. Clarion 14. Blauplunkt
9772RT Boston
229 249
SPECIFICATION
RATINGS
l-well above average
7-well below

AM

FM

FM
PM

wow & ﬂutter

wwNAAAUJJL
wthGhWU-i

PRODUCT FEATURES
treble and bass
and-theft“
controls
music search
switch
bias control
noise reduction

17. Kenwood 18. Sony
KRC-540 XRU-220
299 299
SPECIFICATION
RATINGS
l-well above average
7-well below

AM
FM sensi
PM

PM

wow & ﬂutter
' ratio
PRODUCT FEATURES

OAmNi—bN—t
O‘MNWhﬁmN

treble and bass

controls
music search
switch
bias control
noise

15. Blauplunkt
Louisville
269

wwhnoaww

l9. Clarion
3670RC
299

NwNnbbbNN

16. Pioneer
KEHMSSOO
289

qumw—wmm

20. Pioneer
KEHM7200
349

WMWWHWNW

E

 

167
PRODUCT RATINGS (CONTINUED)

21. JVC 22. 23. Kenwood 24. Targa
KSR470 XRU550 KRC-740
349 369 389 399
SPECIFICATION
RATINGS
l—well above average
7-well below

AM

FM

FM
FM

wow & ﬂutter
' ratio
PRODUCT FEATURES
treble and bass
anti—theft“
controls
music search
switch
bias control
noise reduction

#NNw-hﬁNN
t-‘NNUJA-hwt-t
negro—AN...
NNN~4§H4§H

 

25. Sony 26. 27 . Sony 28. Kenwood
XRU-SSO KEHM8200 XRU770 KRC940
449 449 499 529

2
3
4
4
3
2
2
l

SPECIFICATION
RATINGS
l-well above average

7-well below

AM

FM

FM
FM

 

wow & ﬂutter

HN—‘WAI—INt—I

PRODUCT FEATURES
treble and bass
anti-theft"
controls
music search

S
g

bias control
noise reduction

 

168

APPENDIX E:
EVALUATION TASK QUESTIONNAIRE

QUESTIONNAIRE 2

CAR STEREO RESEARCH STUDY

Deparnnent Of Marketing And Hospitality Services Administration
Central Michigan University

PART 1

Assume you are in the market for a car stereo and have to decide which models of car stereo you
would like to give serious consideration for purchase. In order to make such determinations the ﬁrst
thing you will need to decide is what exactly it is you want from a car stereo. This entails
determining which product attributes are relevant to your needs in a car stereo.

In Part 1 of the Product and Attribute Information Booklet, there are detailed descriptions of the
different attributes. Please look over this information and use it to help answer the following sets of
questions (A, B, and C).

A. DETERMINATION OF RELEVANT AT'I'RIBUTES

For each car stereo attribute below please indicate whether or not it is something you would use to
evaluate the acceptability of different models of car stereo. If you are not sure what an attribute is
refer back to the descriptions in Part 1 of the Product and Attribute Information Booldet.

Iwillusethis Iwillnotusethis
attribute to attribute to

evaluate car stereos evaluate car stereos
SPECIFICATIONS

Price 1

Power 1

H

Am Reception

Fm Sensitivity

Fm Stereo Separation

Fm Quieting

Tape Frequency Response
Tape Wow And Flutter
Tape Signal To Noise Ratio
FEATURES

10. Separate Treble & Bass
11. Anti-Theft

12. Cd-Changer Controls

13. Tape Music Search

14. Radio/Tape Switch

15. Adjustable Bias Control

PFSP‘PPP’P!‘
NNNNNNNNN

ﬂuﬂﬂﬂﬂ

I—ﬁI-II—tt—It-‘H
NNNNNN

169

16. Dolby Noise Reduction 1 2

B. PRODUCT FEATURE REQUIREMENTS
For each car stereo feature listed below indicate which of the seven responses most closely
corresponds to your view of the role that feature will play in your evaluations of different
models of car stereos. Keep in mind that each of these features adds to the cost of a car stereo
and thus should be evaluated in terms of your willingness to pay a little extra to get the
feature. If you indicated above that you are not going to consider a feature, please circle
response 4. Please circle only one response.

\IQQQNIQQQ

RESPONSES
1. I will not consider car stereo models that do not have this feature.
2. I will be unlikely to consider car stereo models that do not have this feature.
3. I will consider car stereo models that do not have this feature, although I prefer that they have it.
4. I will not consider this feature when making evaluations of car stereo models.
5. I will consider car stereos that have this feature, although I prefer that do not have it..
6. I will be unlikely to consider car stereos that have this feature.
7. I will only consider car stereos models that do not have this feature.
CAR STEREO FEATURES RESPONSES
17. Separate Treble & Bass 1 2 3 4 5 6
18. Anti-Theft l 2 3 4 5 6
l9. Cd-Changer Controls 1 2 3 4 5 6
20. Tape Music Search 1 2 3 4 5 6
21. Radio/'1‘ ape Switch 1 2 3 4 5 6
22. Adjustable Bias Control 1 2 3 4 5 6
23. Power Off Eject l 2 3 4 5 6
24. Dolby Noise Reduction 1 2 3 4 5 6
Anti-Theft Protection Systems
25. security code 1 2 3 5 6 7
26. removable chassis 1 2 3 4 5 6 7

27. detachable face 1 2 3 4 5 6 7

170

C. PRODUCT SPECIFICATION REQUIREMENTS
Please indicate what is the minimum performance rating a car stereo model could have on the
following speciﬁcations without automatically being rejected from consideration for purchase.
This minimum rating should indicate the lowest level of performance you would be willing to
tolerate on a particular attribute. This means that if a particular model of car stereo falls below
your minimum rating on this speciﬁcation it would automatically be eliminated from
consideration regardless of how well it performed on all other evaluative criteria. Please keep
in mind that higher ratings are generally associated with higher priced models which tend to
have more sophisticated technological designs. Thus if you demand higher performance
expect to pay for it. If you indicated in part A. that you are not going to consider a
speciﬁcation, please circle response 7. Only circle one response per speciﬁcation.

 

28. POWER 1
well above average 1 2 3 4 5 6 7 well below average '
29. AM RECEPTION 5‘1
well above average 1 2 3 4 5 6 7 well below average

30. TAPE FREQUENCY RESPONSE

well above average 1 2 3 4 5 6 7 well below average
31. FM SENSITIVITY

well above average 1 2 3 4 5 6 7 well below average
32. FM STEREO SEPARATION

well above average 1 2 3 4 5 6 7 well below average
33. FM QUIETING

well above average 1 2 3 4 5 6 7 well below average
34. TAPE WOW AND FLUTTER

well above average 1 2 3 4 5 6 7 well below average

35 . TAPE SIGNAL TO NOISE RATIO
well above average 1 2 3 4 5 6 7 well below average

 

171

For the attribute of price please circle the number that most accurately reﬂects or is the closest to the
maximum price you would pay for a car stereo. This price should be one such that no matter how
impressive a particular model of car stereo is, you would not even consider purchasing it if it has a
price above this upper limit.

36. PRICE 100 150 200 250 300 350 400 450

PART 2:

This section seeks to uncover your attitudes toward the process of shopping for a car stereo. Car
stereo shopping generally entails such activities as gathering car stereo information from magazines,
talking to knowledgeable friends, and visiting car stereo retailers so that you can listen to different
models, demonstrate their features and obtain pricing information. Such activities are generally
performed in order to give the buyer enough information to make a satisfactory purchase decision.

A. Following are a series of items which seek to assess your perceptions about the amount risk there
is in a car stereo purchase decision. For each item listed below please indicate which of the seven
responses most closely corresponds to your view. There are no right or wrong answers: we only
want to know what you think.

Financial risk

37. What are the chances that you would pay too much money for a car stereo, if you were to
make an uninformed purchase decision?
low ﬁnancial risk 1 2 3 4 5 6 7 high ﬁnancial risk

38. If this ﬁnancial loss happened to you, it would be:
unimportant 1 2 3 4 5 6 7 important

Performance risk

39. What are the chances that the car stereo model you select will not work properly if you were
to make an uninformed purchase decision?
low performance risk 1 2 3 4 5 6 7 high performance risk

40. If this performance loss happened to you, it would be:
unimportant1234567important

Psychological risk

41. What are the chances that the model of car stereo you select will not ﬁt in well with your
self-image (i.e. the way you think about yourself) if you make an uninformed purchase
decision.
low psychological risk 1 2 3 4 5 6 7 high psychological risk

42. If this psychological loss happemd to you, it would be:

unimportant1234567important

172

Social risk
43. What are the chances that the model of car stereo you select will have a negative effect on the
way others think about you if you make an uninformed purchase decision?

low social risk 1 2 3 4 5 6 7 high social risk
44. If this social loss happened to you, it would be
unimportant 1 2 3 4 5 6 7 important

Overall perceived risk
45. On the whole, considering all sorts of factors combined, about how risky would you say it is
to make an uninformed car stereo purchase:

not risky at all 1 2 3 4 5 6 7 Vextremely risky

B. Following are a series of items regarding how much involvement you would have with the task of
selecting and purchasing a car stereo. For each item please indicate whether you strongly agree,
agree, neither agree nor disagree, disagree, or strongly disagree with the statement.
neither
strongly agree nor strongly
agree agree disagree disagree disagree

46. In regards to the
purchase decision on which
car Stereo to buy, the choice I
make is of little consequence. 1 2 3 4 5

47. For a car stereo purchase, I
would probably spend a lot of
time and effort making my
purchase decision, since it is
important to get the best deal. 1 2 3 4 5

48. It would be important to me to
be aware of all the alternatives
before buying a car stereo. l 2 3 4 5

49. It is part of my value system to
shop around for the best price. 1 2 3 4 5

50. Inregardstoacarstereo
purchase, it wouldn't make
much difference which
brand I choose. 1 2 3 4 5

51. It is not worth it to read what
Comumer Reports says about

car stereos since most brands are
about the same. 1 2 3 4 5

 

 

 

173

neither
strongly agree nor strongly
agree agree disagree disagree disagree

52. I am willing to spend extra time
shopping in order to get the
cheapest possible price on the
car stereo model I decide I want. 1 2 3 4 5

53. I won't worry so much about get-
ting the best deal when I make
a car stereo purchase, I will just
spend the money as I please. 1 2 3 4 5

54. Because of my personal values
I feel that making a ”smart car
stereo purchase" ought to be
important to me. 1 2 3 4 5

55. I feel very conﬁdent in my
ability to make a good decision
regarding which model of car
stereo to buy. 1 2 3 4 5

PART 3

The next two series of questions are designed to assess your feelings about the process of shopping
for a car stereo. This process might entail visiting different car stereo retailers and doing such things
as listening to demonstration models, getting information from salesmen and obtaining price
information.

A. For the next two questions please circle the responses that best represent your opinions about the
process of shopping for a car stereo. There are no right or wrong answers: we only want you to
answer the questions as truthfully as you can.

56. In comparison to how much I like to shop for all other products, I anticipate that a car stereo
would be.. ..

one of the products I would one of the products I would
most enjoy shopping for 1 2 3 4 5 6 7 least enjoy shopping for

57. In comparison to all other tasks and activities that I perform, I anticipate that shopping for a car
stereo would be....

very much like a very much like a

leisure activity 1 2 3 4 5 6 7 task orachore

174

B. Below are a series of items asking you to give your opinions regarding how you would feel about
the process of shopping for a car stereo. For each item please indicate whether you strongly agree,
agree, neither agree nor disagree, disagree, or strongly disagree with the statement.
neither
strongly agree nor strongly
agree agree disagree disagree disagree

I would emjoy the process
of car stereo shopping for its
own sake, not just for any

beneﬁts it will get me. 1 2 3 4 5
59. Shopping for a car stereo

would be an activity I would

ﬁnd to be intrinsically rewarding. 1 2 3 4 5

60. "Not because I have to but I
want to” would characterize my
reasons for car stereo shopping. 1 2 3 4 5

61. The car stereo shopping
process would satisfy my sense
of curiosity. . 1 2 3 4 5

62. The car stereo shopping
process would offer novel

experiences. 1 2 3 4 5
63. Car stereo shopping would
be an activity that would totally
absorb me. 1 2 3 4 5

64. Car stereo shopping is an
activity I feel obligated to perform
before making a stereo purchase. 1 2 3 4 5

65. Car stereo shopping is an
activity I would be enthusiastic
about performing. 1 2 3 4 5

66. I think that car stereo shopping
would be a very stressful experience

for me. I 2 3 4 5
67. I think I would ﬁnd car stereo

shopping to be a very enjoyable

activity. 1 2 3 4 5

68. I would prefer to have a
knowledgeable person do my car
stereo shopping for me. I 2 3 4 5

175

neither
strongly agree nor strongly

agree agree disagree disagree disagree

69. I would rather just buy the

car stereo rated highest by Consumer

Reports, than try and make my own

decision about what model to buy. 1 2 3 4 5

PART 4:
Please look over the product ratings of car stereos in Part 2 of the Product and Attribute
Information booklet. Please use this information to help you decide which models of car stereo
might be appropriate for your needs. We realize that before you make a ﬁnal purchase decision you
would probably want to visit a stereo retailer and personally inspect models that seem appealing on
paper. Additionally before making a ﬁnal purchase decision you would need to ﬁnd out which of
these car stereos are compatible with your model of car. However with the information provided
you can take the preliminary step of identifying which alternatives are either unacceptable, or
unimpressive and which alternatives merit further consideration. Please keep in mind that the more
brands of car stereo you consider for purchase, the more effort you will have to exert in the shopping
process in order to make a ﬁnal decision. With this in mind please indicate which of the following
models of car stereo you would seriously consider for purchase based upon the product ratings
provided. -

I would I would NOT

seriously consider seriously consider

this model this model

2

 

Clarion 85200

JVC KSR125

Sanyo MARIOOO
Pioneer KE2828
JVC KSR470
Kenwood KRC1005
Sanyo MAR1200
Sherwood XR4127
Audiovox AV238
10. Blauplunkt Newport
11. Fultron 8000

12. Pioneer KE3700QR
l3. Clarion 9772RT

14. Blauplunkt Boston
15. Blauplunkt Louisville
16. Piomer KEHMSSOO
17. Kenwood KRC540

 

9939999.“?

“ﬂﬂﬂﬂﬂﬂﬂﬂﬂﬂbﬂﬂﬂﬂﬂ
NéNNNNNNNNNNNNNN

 

176

I would I would NOT
seriously consider seriously consider
. this model this model

18. Sony XRU220 1 2

l9. Clarion 3670RC l 2

20. Pioneer KEHM7200 l 2

21. JVC KSRG8 1 2

22. Nakamichi TD3/PP1 l 2

23. Kenwood KRC740 1 2

24. Targa Q7000 l 2

25. Sony XRU550 l 2

26. Pioneer KEHM8200 1 2

27 . Sony XRU770 1 2

28. Kenwood KRC940 1 2

PART 5:

29. There are a variety of different methods that people use to determine whether purchase
alternatives are worthy of serious consideration for purchase. In this research study you were asked
to indicate which models of car stereo you would seriously consider if you were to make such a
purchase. We are interested in knowing what method you used to make these determinations. Look
through the descriptions of different decision making methods listed below and indicate which comes
the closest to describing the method you used. '

a. All brands that had acceptable performance on all relevant attributes were selected for
consideration.

b. All brands that had outstanding performance on any relevant attribute were selected for
consideration.

c. Brands that performed the best on the most important attribute were selected for
consideration.

d. Brands that had the best average performance across all relevant attributes were selected for
consideration. Relevant attributes were considered to be roughly equal in terms of
importance.

e. Brands that had the best average performance across all relevant attributes were selected for
consideration. Relevant attributes were differentially weighted in terms of their importance.

f. Some other method of decision making was used. Please describe your decision making method
in the space below : .

 

 

177

APPENDIX F:
POST-TEST QUESTIONNAIRE

QUESTIONNAIRE 3

CAR STEREO RESEARCH STUDY

Department Of Marketing And Hospitality Services Administration
Central Michigan University

The following questions are designed to assess how much you know about car stereos. For each
question circle the response that you think is correct.

30.

999'.”

U.)

9.0.6!»

U.)

999'?"

l.

2.

Which of the following speciﬁcations indicates the ability of a tape deck to reproduce the
audible range.

tape deck quieting

tape deck frequency response

tape deck signal to noise ratio

tape deck sensitivity

Which of the following speciﬁcations indicates the ability of an FM receiver to pick up weak
broadcast signals and stations from far away markets.

FM quieting

FM frequency response

FM signal to noise ratio

FM sensitivity

Which of the following speciﬁcations indicates the amount of background noise being produced
by a car stereo's tape deck?

tape deck wow and ﬂutter

tape deck frequency response

tape deck signal to noise ratio

tape deck quieting

. Which of the following speciﬁcations indicates the accuracy of a tape deck's playback speed and

indicate the consistency of the playback mechanism.
tape deck wow and ﬂutter
tape deck frequency response
tape deck signal to noise ratio
tape deck quieting

Which of the following speciﬁcations is used to indicate how strong an FM signal must be to
provide noise free reception.

FM quieting

FM frequency response

FM signal to noise ratio

FM sensitivity

 

999‘.”

U.)

999‘!”

999‘?

6.

. Which car stereo feature gives the listener the ability automatically fast

178

. Which of the following speciﬁcations indicates how loud a car stereo can be played.

power

wow and ﬂutter
sensitivity
frequency response

Which feature gives you the ability to select the correct setting for playing back metal or
chromium dioxide tapes.

music search

adjustable bias control

Dolby noise reduction

radio/tape switch

. Which type of anti-theft protection offers the greatest ease of use but the lowest level of

protection?
security code
removable chassis 3....
detachable face

. Which feature gives you the ability to virtually eliminate background noise during playback of

tapes.
music search
adjustable bias control
Dolby noise reduction
separate bass and treble controls
radio/tape switch

Which type of car stereo anti-theft protection is generally the most expensive?
security code

removable chassis

detachable face

Which car stereo feature gives the listener the ability to listen to the radio while rewinding or
fast forwarding a tape?

music search

adjustable bias control

separate bass and treble controls

radio/tape switch

forward or rewind to the beginning of the next song on a tape, at the touch of
a button.

music search

adjustable bias control

separate bass and treble controls

radio/tape switch

 

179

Following are a series of items regarding the amount of thought and effort you feel you put into this
evaluation task. For each item please indicate whether you strongly agree, agree, neither agree nor
disagree, disagree, or strongly disagree with the statement. There are no right or wrong answers: we
only want you to answer the questions as truthfully as you can.

strongly neither agree strongly

agree agree nor disagree disagree disagree

42. I put a lot of thought
into my responses. 1 2 3 4 5

43. I took the evaluation task
as seriously as if I was making
a real car stereo purchase. 1 2 3 4 5

44. I was very involved with
the evaluation task. 1 2 3 4 5

45. The evaluation task was
important to me. 1 ‘ 2 3 4 5

 

BIBLIOGRAPHY

BIBLIOGRAPHY

Allport, G.W. (1943), "The Ego in Contemporary Psychology,"
Wm. 50. 451-478.

Antil, J. H. (1984), "Conceptualization and Operationalization of

Involvement" W. 11 T Kinnear
ed. , 203- 209.

Apsler, R. and D. Sears (1968), "Warning, Personal Involvement and

Attitude Change." W131
Bartram. 9. 162-166.

Arora, R. (1982), "Validation of the S-O—R Model for Situation,
Enduring, and Response Components of Involvement," Journal

of Mgketing Rosoanoh, 29, 505-516.
Assael, H. (1984), Consomor Bohavior and Mgkoting Aotion,

Boston, Kent.

Belonax, J .J. and RA. Mittelstaedt (1978), "Evoked Set Size as a
Function of Number of Choice Criteria and Information

Variability." WW. 5 H. Hunt ed..
48-51.

Belonax J. J. and R. G. Javalgi (1989), "The Inﬂuence of Involvement
and Product Class Quality on Consumer Choice Sets," Joannal

Wham, 17(3), 209 216.

Berlyne, DE. (1960). Wm New York.
McGraw-Hill.

Bettman 1A. (1979). Wham
29W, Reading Ma., Addison Wesley.

Bloch, P. H. (1982), "Involvement Beyond the Purchasing Process:
Conceptual Issues and Empirical Investigation, Admin

W, 9, A. Mitchelled” 413-417.
180

181

Bloch, RH. and G. D. Bruce (1984), "Product Involvement as

Leisure Behavior," Adyanceg in Consnmor Rosamh, 11, T.
Kinnear ed., 197-202.

Brisoux LE. and M. Larouche (1981), "Evoked Set Formation and
Composition: An Empirical Investigation Under Routinized

Response Behavior Situation, " Adyangsjnmonler
Mil. 8, K. Monroe ed., 357-361.

Brown, JJ. and A.R. Wildt (1987), "Factors Inﬂuencing Evoked Set
Size," Working Paper Series, University of Missouri-Columbia.

Brucks, M. (1985), "The Effects of Product Class Knowledge on
Information Search Behavior," urn l m r h,
12, 1-15.

Campbell, BM. (1969), "The Existence of Evoked Set and
Determinants of its Magnitude in Brand Choice Behavior,"
Dissertation, Columbia University.

Celsi, R.L. and J.C. Olson (1988), "The Role of Involvement in
Attention and Comprehension Processes, " JotnnaLoLConsme;
Re h, 15, 210-224.

Chaiken, S. (1980), "Heuristic Versus Systematic Information
Processing and the Use of Source Versus Message Cues in

Persuasion," r l' i P hi ,39,
752-766.

Clarke, K. and R. W. Belk (1979), "The Effects of Product
Involvement and Task Deﬁnition on Anticipated Consumer

Effort" AW 6 W Wilkieed
313-318.

Cobb C. J. and W. D. Hoyer (1985) "Direct Observation of Search
Behavior," W, Spring 1985, 191-179.

Cohen, J. B. (1983), "Involvement and You: 1000 Great Ideas"

W 10 R Bagozzi and A
Tybout eds., 325 328.

182

Condry, J. (1977), "Enemies of Exploration: Self-Initiated Versus
Other-Initiated Learning," r l P Ii 11 ial

Wm, 35, 459-477.

Condry, J. (1987), "Enhancing Motivation: A Social Development
Perspective," v 'n M i i 11 an hievem

Emanoing Motivation, 5, 23-49.
Crutchﬁeld Corporation (1992), WM
Day. G. S. (1970). W. New York.

Free Press.

DeBruicker, F. S. (1979), "An Appraisal of Low Involvement

Consumer Information Processing," Amtndejosoaronlﬂayaiot
High Stakes ,.J C. Maloney and B. Silverman, eds., 112-130.

Engel J.R., D.T. Kollat, and RD. Blackwell (1968), Consumer
Behavior, New York, Holt, Rinehart, and Winston, Inc.

Eroglu, S.A., G.S. Omura, and K.A. Machleit (1983), "Evoked Set
Size and Temporal Proximity to Purchase," AMA Edngtot's

Wigwam. P- Murphy ed., 97-101.

Garbarino, J. (1975), "The Impact of Anticipated Reward upon
Cross- Aged Tutoring," r i i l

Emmy, 32, 421-428.

Gensch, DO. and RC. Javalgi (1987), "The Inﬂuence of
Involvement on Disaggregate Attribute Choice Models," Jour_na1

Wench. 14. 71-82.

Gronhaug, K. (1974), "Some Factors Inﬂuencing the Size of the

Buyers Evoked Set." W" 7. (3).
232-241.

Harrell, GD. (1986), Consntnonﬁohaﬂot, New York, Harcourt
Brace J ovanovich.

Hauser, J .R. and B. Wernerfelt (1990), "An Evaluation Cost Model

of Consideration Sets." WWW. 16.
393-408.

183
Hawkins, D..I ,R..] Best, andK. A. Coney (1992) 99311—91
Bohavior; Jmolioati one for Maggotin ng Sttate te,gy Homewood, IL,
Irwin.

Hirschman, EC. and MB. Holbrook (1982), "Hedonic
Consumption: Emerging Concepts, Methods and Propositions,"

Jonrnal of Matkoting, 46, 92-101.

Holbrook, MB. and EC. Hirschman (1982), "The Experiential
Aspects of Consumption: Consumer Fantasies, Feelings, and

Fun." MW. 9. 132-140.
Howard, J.A. and J .N. Sheth (1969) :1an flfnooty of Bnyot Bonavior,

New York, Wiley.

Houston M.J. and M.L. Rothschild (1978), "Conceptual and
Methodological Perspectives of Involvement," Rom

Frontiorg in Mgketing: Dialognos and Dirgotors, S.C. Jain ed.,
184-187.

Hupfer, NT. and D.M. Gardner (1971), "Differential Involvement
With Products and Issues: An Exploratory Study," Proceedings

of the Annual Conference - Association for Consumer
Research, 262-269.

Iso-Ahola, S. (1976), "On the Theoretical Link Between Personality
and Leisure." WM. 39. 3- 9

Jacoby, J. (1972), "The Components of Perceived Risk," KW
f l' fern -A 'in mr h,
382-383.

Jarvis, L. P. and J. B. Wilcox (1973), "Evoked Set Size- Some
Theoretical Foundations and Empirical Evidence" Contbjm

 

Assooatton, 35, TV Greered 236-240. L _

Johnson B. T. and A. H. Eagly (1990), "Involvement and Persuasion.

Types, Traditions, and the Evidence" Esyoholog'oaiﬁnﬂetni,
107 (3), 375- 384.

Johnson B.T. and A.H. Eagly (1989), "Effects of Involvement on

Persuasion: A Meta-Analysis," W, 106 (2),
290-314.

184

Johnson, EL and J.E. Russo (1984.), "Product Familiarity and

Learning New Information." Wench.
1 1, 741-753.

Kapferer, J .N. and G. Laurent (1985), "Consumer Involvement
Proﬁles: A New Practical Approach to Consumer

Involvement." MW. 25 . 48-56.

Kassarjian, H. H. (1978), "Presidential Address, 1977:
Anthropomorphism and Parsimony," n i n r
R h, 5, H..K Hunted” xii-xiv.

Katonna G. and E. Mueller (1954), "A Study of Purchase Decisions,"
Consnmor Bohavior, J. Clark ed., Vol. 1, New York, NYU
Press.

Kruglanski, A.W., C. Stein and A. Riter (1977), "Contingencies of
Exogenous Reward and Task Performance: On the "Minimax"
Strategy in Instrumental Behavior," r f A li i l

ErichoIan. 7. 141-148.

Krugman, HE. (1965), "The Impact of Television Advertising:

Learning Without Involvement," Enolic Qoinion Qnattorly, 29,
349-356.

LastovickaJ. L. and D. M. Gardner (1978), "Low Involvement Versus
High Involvement Cognitive Structures" Ad_ant‘&_s_in
W. 5. H Hunted. 87-92

Lastovicka J .L. and D.M. Gardner (1979), "Components of
Involvement" WWW. J.C.

Maloney and B. Silverman, eds., American Marketing
Association Proceedings Series, 87-92.

Laurent, G. and]. N. Kapferer (1985), "Measuring Consumer

Involvement Proﬁles" MW, 22, 41-
53.

Lepper, M. R., D. Greene and R. E. Nisbett (1973), "Undermining
Children's Intrinsic Interest With Extrinsic Rewards. A Test of
the "Overjustiﬁcation" Hypothesis, "JonmaLofPogtpnaiityand
SociaLEsxchom 28 129-137

 

185

Petty RE. and J. T. Cacioppo (1979), "Issue Involvement as a Moderator of the
Effects on Attitude of Advertising Content and Context, " Adanoosjn

W, 8, K Monroe ed., 20-24.

Petty RE. and J. T. Cacioppo (1986) "The Elaboration Likelihood Model of
Persuasion." Adrmeunﬁanenmentalﬁonaiﬁmholm 19 123-205

Petty, RE. and J .T. Cacioppo (1990), "Involvement and Persuasion: Tradition Versus
Integration, WW, 107(3), 367-374.

Pittman, T.S., A.K. Boggiono, and D.N. Ruble (1983), "Intrinsic and Extrinsic
Motivational Orientations: Limiting Conditions on the Undermining and
Enhancing Effects of Reward on Intrinsic Motivation, " IeaohonangLSntm

ﬁercenuonstnlicationafoLLeamina J Levine and M Wang eds Hillsdale
N.J., Erlbaum.

Pittman, T.S., J. Emery, and AK. Boggiano (1982), "Intrinsic and Extrinsic
Motivational Orientations: Reward-Induced Changesin Preference for

Complexity," 1 P 11 i 1 ,42, 789-797.

Punj, 6N. and R. Staelin (1983), "A Model of Consumer Information Search for a
New Automobiles," MW, 9, 366-380.

Ray. M. L. (1973), "Marketing Communications and the Hierarchy of Effects, " Now
NEW, P. C. Clarke, ed., 147- 176.

Reeve. J. (1992). MW. Fort Worth. TX. Harcourt

Brace Jovanovich.

Richins, M.L. and RH. Bloch (1986), "After the New Wears Off: The Temporal
Context of Product Involvement," MW, 13, 280-285.

Robertson, T. S. (1976), "Low-Commitment Consumer Behavior," mm
W. 16 (2) 19-23

Rothschild, M. L. and M. J. Houston (1977), "The Consumer Involvement Matrix:

Some Preliminary Findin88" Wineries. B A
Greenberg and D. N. Bellinger 95-98.

Sherif. M. and H. Cantril (1947). Iheﬁsxcholomfﬁamlnrolmem. New York.
Wiley.

186

Sherif, M. andC..L Hovland(l96l), ' t ' t' ’ t
W New Haven; Conn Yale

University Press.

Sherrell, D. L. and T. A. Shimpt (1982), "Consumer Involvement 1n a Laboratory
Setting," t Pr ' 2 A

W B Walker et al eds 104-108

Slama, ME. and A. Tashchian (1985), "Selected Socioeconomic and Demographic
Characteristics Associated with Purchasing Involvement," WW,
Vol. 49 (Winter), 72-82.

Steiger, J. H. (1980), "Tests for Comparing Elements of a Correlation Matrix,"

PwhologicoLBullctin. Vol 87. 245-251

Stone, RN. (1984), "The Marketing Characteristics of Involvement," AdyanooLin
Waggon, II, T. Kinnear, ed., 210-215.

Tauber, EM. (1972), "Why Do People Shop," JottmaJQLMarkgtjng, 36, 46-52.

Unger, LS. and RI. Dornoff (1979), "A New Framework for Analyzing Leisure
Pursuits," paper presented at the National Recreation and Park Association
Research Symposium , New Orleans, LA.

Unger, LS. and J .B. Kernan (1983), "On the Meaning of Leisure: An Investigation of
Some Determinants of the Subjective Experience," Maloﬂgononmgt
Resoamh, 9, 381-392.

Zaichkowsky, J .L. (1985), "Measuring the Involvement Construct," JottmaLof
ConsnmcLRcscarch. 12. 341-352.

Zaichkowsky, J.L. (1988), "The Personal Involvement Inventory: Reduction,
Revision, and Application to Advertising," Working Paper, Simon Fraser
University. ~

Zimbardo, P. G. (1960), "Involvement and Communication Discrepancy as
Determinants of Opinion Conformity." lonmsloﬂAhnomaLaniSocial
Psychology 60 (1) 86-94

"

1p-

 

"I111111111111111111111111115