Decision ambiguity and consumer brand loyalty

MISSING IMAGE

Material Information

Title:
Decision ambiguity and consumer brand loyalty
Physical Description:
vi, 87 leaves : ; 29 cm.
Language:
English
Creator:
Muthukrishnan, A. V
Publication Date:

Subjects

Subjects / Keywords:
Brand choice -- Research   ( lcsh )
Consumers' preferences   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1993.
Bibliography:
Includes bibliographical references (leaves 82-86).
Statement of Responsibility:
by A.V. Muthukrishnan.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001897767
notis - AJX3063
oclc - 29860613
System ID:
AA00003250:00001

Full Text











DECISION AMBIGUITY


AND CONSUMER BRAND LOYALTY


By

A.V. MUTHUKRISHNAN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF
THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF T-IF RFfITIPPRMF1ThT5 PYP TI-P 1lrlRF lUK















ACKNOWLEDGEMENTS

wish to thank some of those who made my doctoral education at Florida such a


rewarding experience.


An environment of wonderful intellectual intensity was provided by the


marketing faculty and by my fellow doctoral students.


My cochairmen, Joseph Alba and John


Lynch, gave me great inspiration, and remain important role models.


guidance that helped me define my interests.

support throughout this project. I wish to tha


I greatly appreciate their


I am indebted to Rich Lutz and Bart Weitz for their


nk my colleagues, Jackie Marks, S. Ramaswami, and


especially Susan Foumier and Luk Warlop for their help at various stages of this project.















TABLE OF CONTENTS
oage

ACKNOWLEDGEMENTS ......................................... ....... ................ ..... ...... .. ............. ii

A B STR A CT ......... .. .. .. .......... .... ..... ....... ..... ..... ....... ..... ..... ....... .... ....................... .. v

CHAPTERS

I INTRO DU CT IO N .......................... .... ............. ........... .............. ....... ....... 1

II DECISION AMBIGUITY, BRAND KNOWLEDGE AND
CONSUMER BRAND LOYALTY ....... .... ................... ........................ 3
O overview ........ ............... .............. ........... ... ... ........... ....... ..................... 3
Loyalty: Persistence of Choice in Comparative Judgments.................... 4
Decision Am biguity ........... ...... .. ....... ...... ...... ... ..... ....... ... ......... ......... .... 4
Consumer Knowledge........ ................................. ............. .. .................. 6
Other Processes Leading to Loyalty ... ............. .................. ...... ... .. 12

III EXPERIMENT 1: BRAND KNOWLEDGE AND LOYALTY............. 15
Pretests.. ................................................. ....... ...... ........ ........ 15
M ethod ..................... .................. ........ ........ ..... ............................... .... .. 20
R results ..... ............ ............. ................ ..... ... ........................... ............ 23

IV EXPERIMENT 2: LOYALTY AND AMBIGUITY DUE TO
MEMORY CON STRAINTS ......... ................ ............................ .......... 36

V EXPERIMENT 3: LOYALTY AND AMBIGUITY DUE TO
NON-OVERLAPPING ATTRIBUTES ....... ....... ............................... 40
Pretest..... ......... ........mt em......... ................................................................. 40
Design and Procedures............................................................................ 41
R results ........................ ...... ........ ................... ............. ......................... 42

VI EXPERIMENT 4: CHOICE CONTEXT AND LOYALTY ................. 58
Design and Procedure.. .............................................. .......... ... .................... 58
R results ........................................... ................................................ 60

VII EXPERIMENT 5: OPPORTUNITY FOR TRIAL AND CHOICE
BETWEEN THE INCUMBENT AND THE ATTACK BRAND......... 62

















GENERAL DISCUSSION AND CONCLUSION.................................


APPENDICES .................................................................. .............. ......... .................

REFERENCES.........c. c ................ ...................... ................. ................................ ..... .....

BIOGRAPHICAL SKETCH.....................................c..cc ...... ..... ........ .....................














Abstract of the Dissertation Presented to the University of
Florida in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy


DECISION AMBIGUITY


AND CONSUMER BRAND LOYALTY


Muthukrishnan


Chair: John G. Lynch
Cochair: Joseph W. Alba
Major Department: Marketing

Past conceptualizations of brand loyalty have stressed emotional commitment or habit


formation.


In contrast,


I conceptualize brand loyalty in terms of confidence in one's judgment


that the brand one has chosen previously (the incumbent brand) is superior to all the other brands


in the choice set.


Confidence of this nature causes resistance to any influence attempts by


competitors, even when the competitor's brand (the attack brand) is objectively superior to the


incumbent brand.


I further hypothesize that ambiguity in the choice context is a key variable that


can affect confidence in judgment and thereby resistance.


Ambiguity in the choice context will


be high when attribute-to-attribute comparison between two or more brands is difficult.


occurs when (1) brands share the same set of attributes but attribute information about the

incumbent brand is not available externally when a person encounters the new brand (i.e., the

consumer needs to rely on memory) and (2) the incumbent and attack brands possess non-


overlapping attributes.


In five experiments I examined the conditions under which decision


ambiguity leads to resistance to competitor's attack.













knowledge structures--experience


and thinking


in terms of product benefits--when


decision


ambiguity is high due to memory constraints.


and low).


Experience was manipulated at two levels (high


I found that the likelihood of choosing the incumbent brand over a new, objectively


superior brand was high when either experience or availability of brand beliefs was high.


In the second experiment, I varied decision ambiguity via a memory manipulation.


Apart


from replicating the results of experiment 1, I also found that when decision ambiguity was low

(i.e., attribute information about the incumbent brand was externally available when a person

encountered the new, objectively superior brand), resistance to competitor's attack was low despite


extensive experience with the incumbent brand and high availability of brand beliefs.


In this


condition more subjects chose the attack brand.

In the third experiment, decision ambiguity was created by describing the incumbent brand


and the new, attack brand in terms of non-overlapping attributes.


I examined the effects of


experience


and external


availability


of attribute


information


about the


incumbent brand on


resistance.


Unlike in Experiment


external availability of specific attribute information about


the previously chosen brand did not have any effect on resistance.


Thus, ambiguity created by


non-overlapping attributes can enhance the diagnosticity of even limited experience.

An analysis of the processes in the first and the third experiments partially supported that


loyalty could be explained in terms of overconfidence in judgment.


In the fourth experiment, I


established


a boundary


condition


for the effects of knowledge structure and


ambiguity


consumer brand loyalty.


In the fifth experiment, it was found that an opportunity to try the attack















CHAPTER I
INTRODUCTION


Brand loyalty is an important marketing construct from a strategy viewpoint.

degree of consumer brand loyalty is one of the greatest assets marketers can possess.


A high

Thus,


creating and sustaining consumer loyalty should be a central goal of the marketing function of a

firm. Loyal customers may resist offers from competitors (Reibstein 1985) and thus may represent


a sustainable competitive advantage to the firm.


For example,


when deregulation opened the


doors for new companies to enter the long distance field, it was quickly discovered that AT&T


could not be dislodged easily.


Many long distance customers refused to consider switching


despite the significant price savings that were offered (Engel, Blackwell and Miniard


However, not every company succeeds in establishing such loyalty.


1989).


For example, in the personal


computers market, IBM's margins and market share eroded when its competitors offered "clones"

(compatible personal computers) at lower prices.


Brand


loyalty


is also an interesting


topic


a psychological


point


view.


Surprisingly, however, prior research has not shed much light on what causes loyalty.


During the


1980s, the trend was to use "loyalty" as an explanatory variable in multinomial logit models of


choice to predict brand choice and market share.


For example, Guadagni and Little (1983)


operationalized "loyalty" as the last brand bought and found that this variable explained a large


amount of variance in choice.

variable or a given state. Th,


In these studies, loyalty seems to be treated as an exogenous


e utility of the construct of loyalty to marketing actions will be








2

competitor's (attack) brand is objectively superior to the currently used (incumbent) brand. In the

AT&T and IBM examples, the competitors' brands were objectively superior in as much as they


provided equivalent quality at a lower price. Ir

which resistance to such offerings will be high.


1 this dissertation, I examine the conditions under

Specifically, the goals of this dissertation are to


(1) show that loyalty can be manipulated, (2) identify a set of variables that can lead to loyal

behavior, and (3) propose a model of loyalty based on behavioral decision theory.


The core assumption of this dissertation is that loyalty vanes as a function of the

confidence one has in the superiority of one's incumbent brand. This assumption contrasts sharply


with prior research which has stressed emotional commitment or habit formation.

hypothesized that decision ambiguity is a key variable that can affect confidence.


Further, it is

It is proposed


that when decision ambiguity is high, experience with the incumbent brand or knowledge of the

benefits of the incumbent brand can affect confidence and thus resistance to competitor's attack.

In the next chapter, a conceptualization of loyalty is presented following a brief review of


prior research.


Chapter III presents an experiment designed to test the effects of experience and


belief crystallization on loyalty. Chapter IV discusses an experiment designed to test the effects


of memory-based ambiguity on loyalty.


Chapter V presents an experiment designed to test the


effects of ambiguity caused by nonoverlapping attributes.

designed to test the effect of context on loyalty formation.


Chapter VI discusses an experiment

Chapter VII examines the effects of


an opportunity to try the attack brand on loyalty to the incumbent brand.


The last chapter


discusses the theoretical and managerial implications of this research and proposes a set of themes

for future research.














CHAPTER II


DECISION AMBIGUITY


BRAND KNOWLEDGE AND CONSUMER


BRAND LOYALTY

Overview


Brand loyalty has been traditionally conceptualized in terms of emotional commitment and

ego involvement (e.g., Jacoby 1971; Jacoby and Chestnut 1978) or habitual purchase behavior


Howard and Sheth 1969;


loyalty).


see Jacoby and Chestnut 1978 for a review of early research on


The emotional commitment view argues that brand loyalty reflects a consumer's central


values, and it assumes a high degree of consumer involvement at the product category level and


at the brand level. With a few exceptions, high brand involvement does not characterize most

marketplace choices. Even in the few categories in which high levels of brand involvement are


likely, attempts to link brands to consumers' central values may be extremely difficult (Rokeach

1973).


In contrast, the habitual purchase behavior view argues that brand loyalty results from a


lack of cognitive activity.


That is, consumers repurchase the same brand out of habit and little


However, this view provides no insight into the initiation of loyalty.


In fact, it may be more


appropriate to consider routinized response behavior as a long term consequence of loyalty (Alba


and Hutchinson


1987).


If so, there exists a need to identify a set of factors that lead to loyal


behavior and to outline the underlying psychological mechanism.

Previous studies have identified resistance to competitor's attack as a key consequence of


loyalty (Ehrenberg and Charlton


1973; Jacoby and Kyner


1973; Newman and Werbal


1973).








4

comparative advertising, the focus of this dissertation is on attacks based on superior product


quality. The resistance mechanism proposed here, however, can be generalized to any other form

of attack. I rely on behavioral decision research to explain resistance to competitor's attack while


acknowledging that cognitive biases may lead to habitual purchase behavior or even to emotional

commitment.


Loyalty: Persistence of Confidence in Comparative Judgments


In this paper brand loyalty is conceptualized in terms of confidence that the brand one has


chosen previously (the incumbent brand) is superior to all other brands in the choice set.


More


specifically, loyalty is driven by persistence of such confidence, even when the competitor's brand


(attack brand) is objectively superior to the incumbent brand.


This view of loyalty is analogous


to overconfidence, wherein individuals overestimate their knowledge of past and future events


(Lichtenstein, Fischhoff and Philips 1982).


Unlike in previous conceptualizations, here it is easier


to identify a set of antecedents that can be manipulated by marketers.


It is also possible to isolate


the conditions that can weaken resistance.


To explain loyalty, the notion of overconfidence is extended to preference judgments.


a consumer judges an objectively inferior brand to be superior and expresses confidence in this


judgment, he or she exhibits overconfidence in preference judgments.


represents a primary route to persistence of preference.


It is proposed that this


The central premise of this dissertation


is that under ambiguous decision contexts, certain types of knowledge about the currently used

(incumbent) brand can lead to overconfidence in prior choice and thus resistance to competitor's

attack.


Decision Ambiguity










reasonable accuracy (see Camerer and Weber


1992 for a review of research).


Einhom and


Hogarth


(1985)


define


ambiguity


as the


uncertainty


about


uncertainty,


where


the latter


"uncertainty"


refers to outcome probability.


Most marketplace choices are characterized by


ambiguity.


Research on ambiguity in consumer choices, however, is sparse (Bettman, Johnson


and Payne 1991; but see Ha and Hoch 1988, Hoch and Ha 1986, and Kahn and Meyer 1991).

Ellsberg (1961) defined ambiguity as a "quality depending on the amount, type, reliability


and unanimity


of information"


(p 657).


Further, ambiguous situations


are those in


which


"available information is scanty


or highly conflicting; or where expressed expectations of


different individuals differ widely"


(p 660-661).


Based on this conceptualization,


decision


ambiguity in this paper refers to a set of conditions in which attribute-by-attribute comparisons


among brands are difficult.


Ambiguity can exist at different stages of brand evaluation including


identification of


relevant


attributes,


attribute


valuation,


information integration and


product


consumption (Hoch and Deighton


1989).


Ambiguity in consumer choices arises out of two


sources: incomplete information about the characteristics of one or more brands in the choice set

and unsure preference for the attributes of the competing brands.

Ambiguity due to incomplete information was the theme of classic research on uncertain


choices (Ellsberg 1961).


The conclusion of this stream of research was that people tend to avoid


ambiguous alternative for which the probability of a successful outcome is not known.


phenomenon of "ambiguity avoidance" has been also reported in consumer research involving


multiattribute judgments or choices (Biehal and Chakravarti


1983; Huber and McCann


1983;


Meyer 1981; and Simmons and Lynch 1991).

If ambiguity is due to unsure preference for attributes, the choice context itself can shape








6

A number of real-world causes can lead to incomplete knowledge or unsure preferences

and thus to difficulty in making attribute-by-attribute comparisons. First, the display environment

in retail settings often makes it difficult to judge the "objective superiority" of a brand (Russo


1977).


Second, most consumer choices involve a memory component (see Alba, Hutchinson, and


Lynch 1991), and imperfect memory can lead to decision ambiguity.


For example, consumers


may acquire information about brands at different points of time and may need to make decisions


based on what is recalled about each brand.


Third, due to low motivation, consumers may not


make


side-by-side


attribute comparisons even if information


about each


brand


is available


externally.


Fourth, in certain product categories, choice sets may have negative interattribute


correlations (Pareto optimal) and may require multiple trade-offs among attributes (Ha and Hoch


1989).


Finally, the competing brands may possess non-overlapping attributes. For example, two


brands of body


lotions may


emphasize


Vitamin


E and


Vera as


the key ingredients,


respectively, but may promise the same benefits of healing and moisturizing the skin.


This lack


of dimensional comparability increases decision difficulty (Slovic and Mcphallamy 1974; Tversky

1972).


Although


the decision


environment


not afford


conclusive


evidence


about


superiority of the incumbent brand, consumers may nonetheless have a feeling of control and


perceive the evidence as more diagnostic than it objectively is (Hoch and Deighton 1989).


Thus,


as suggested


earlier,


certain


types


of knowledge


about


the incumbent


brand


can induce


overconfidence, especially in ambiguous decision environments.

Consumer Knowledge


The other major determinant of loyalty in my model is consumer knowledge.


Knowledge








7

benefits of brand consumption, beliefs about the values and importance of the brand attributes,

and beliefs about the attribute-benefit association.

Such knowledge may be acquired through product trial or through advertising and a


variety of other marketer and nonmarketer dominated sources of information.


Hoch and Deigton


refer to the former as "knowledge by description" and the latter as "knowledge by acquaintance."

Prior research has shown that consumers attach higher validity to judgments acquired through


product acquaintance (Fazio and Zanna 1978; Smith and Swinyard 1983).


Thus, although beliefs


about


the benefits


of a brand


can be acquired


through


advertisements


and word-of-mouth


communications, self-generated cognitions, especially after trying the brand, may induce higher

confidence.

If a consumer has repeated experience with a brand, he or she will acquire a set of


confidently held beliefs about the abstract benefits of the brand.


The same effect can result from


limited experience if a person contemplates the putative benefits of the chosen brand.


Through


either means, a consumer's beliefs about the brand benefits and the overall quality of the brand


are strengthened.


These strong beliefs guide subsequent choices, especially in ambiguous choice


contexts in which discriminabiliity based on the specific attributes of the competing brands is low.

Experience and Belief Structure


Experience can influence knowledge structure in several ways.


additional information about the incumbent brand.


First experience provides


Additional information, regardless of its


relevance, can enhance confidence in beliefs without increasing accuracy in judgment (Oskamp


1965;


Peterson


and Pitz 1988).


Further,


the amount of information available can lead


judgmental stability (Davidson et al. 1985) and resistance to counterpersuasive influence attempts










Second,


experience


provides


qualitatively


different


types


of information


about


incumbent brand.


These include unique beliefs about the benefits of the incumbent brand and


knowledge of the consumption experience. Thus, the comparability between an experienced and


nonexperienced brand may be low.


For example, beliefs about the benefits of the incumbent


brand based on one's extensive experience may not be comparable to the specific attribute


information about a superior competitor.


In such noncomparable choice situations, self-generated


cognitions may be prominent and may be given higher weight in decisions.


Assuming that a


consumer has


had favorable


experience


a brand,


extensive


expenence


result


asymmetry in information to the advantage of the incumbent.


when the decision ambiguity is high.


This specific effect may hold only


When compatibility among choice inputs is increased by


describing all competing brands in terms of a set of common attributes, ambiguity is removed.

Even partial overlap among the attributes of the competing brands may reduce ambiguity, in as

much as common features are overweighted in decisions (Slovic and MacPhillamy 1974; Tversky,

Slovic and Sattath 1989).


Experience can also strengthen beliefs and lead to perceived knowledgeability.


literature on learning from experience concludes that, in some cases, experience may inhibit rather

than facilitating learning (Alba and Hutchinson 1987; Brehmer 1980; Einhom 1980; Einhom and


Hogarth 1978). According to Einhom and Hogarth (1978) experience results in "persistence of

the illusion of validity" due to the biases in hypothesis testing that accentuate the effect of


experience on unwarranted confidence in judgment.


The conditions


proposed


Einhom


and Hogarth


seem


to closely


correspond


ambiguous marketing choices.


First, in most marketplace choices, judgments are made for the










feedback on judgments.


Hence, the only source of feedback on judgment is the performance of


the chosen brand. In a number of categories, performance evaluation itself may be a difficult task.


Hence, unless there is a clear violation of expectations, outcomes are coded as positive.


likely that once people are satisfied with the quality of a brand, they rarely try other brands, even


if these brands are of equal quality.


The well documented biases in hypothesis testing are very


likely to occur in ambiguous marketing contexts (Fischhoff and Beyth-Marom 1983; Klayman and


Ha 1987;


Tschirgi 1980; see Hoch and Deighton 1989 for a review).


In Einhom and Hogarth's model, confidence


feedback.


is an increasing monotonic function of


Assuming that the positive aspects of a brand are greater than the negative aspects


(which is very likely with most brands in most product categories), feedback and hence confidence


becomes an increasing function of the number of trials or experience.


Thus, experience can


strengthen beliefs and lead to overconfidence in one's prior judgment and choice.

A set of marketing studies that examined the theme of advantages of the pioneering brands

offered conclusions that are consistent with my suggestions about the effect of experience with


a brand.


Examining a context in which the value of attributes, attribute weights and the ideal


attribute combinations are ambiguous, Carpenter and Nakamoto (1989) concluded that consumer

taste distribution and the attribute weighting shift toward the attribute combinations and the


relative attribute strengths of the pioneer brand.


The pioneer can frame expectations regarding


the category and can influence the formation of preference especially when product evaluation is

ambiguous (see also Carpenter and Nakamoto 1990).


In addition, Kardes and Gurumurthy (1992) found that


differential learning.


the order of entry results in


Specifically, the recall of unique and shared features of an early entrant was










Thinking About Benefits and Belief Structure


Experience is not the only route by which a brand may be judged superior to its


competitors.

several ways.


Mere mental elaboration about product attributes may produce similar effects in

First, thought increases the absolute amount of information available about the


incumbent brand and thus extremity and confidence in judgment.


Secondly, thought results in


information that is qualitatively different from the information available for the attack brand.


Finally, thought can increase the perceived knowledgeability.


For example, Koriat, Lichtenstein


and Fischhoff (1980) found that reasons for overconfidence include selectively focusing on the


evidence supporting choice and disregarding evidence contradicting it.


Further, they found that


although subjects could easily produce reasons supporting their choices, they had difficulty in

producing reasons against their choices.

Forcing consumers to think in terms of benefits may result in elaboration of the positive


aspects of the brand (Koriat, Fischhoff and Lichtenstein 1980).


may provide reasons to justify the choice.

salience of beliefs and confidence in beliefs i


Expression of these beliefs also


Thus, beliefs crystallize in the sense that both the

increase. Further, memory for these abstract beliefs


may decay less rapidly than specific attribute information (Chattophadhyay and Alba 1988), which


may then unduly influence future choices (Alba, Marmorstein and Chattophadhyay 1992).


Finally,


the act of articulating one's beliefs about the brand's benefits may strengthen these beliefs and


increase the confidence in judgment. Thus, thinking in terms of benefits after only limited brand

experience may result in loyal behavior. This scenario corresponds to real world advertisements


that remind consumers of brand benefits.


Recent decision research supports the argument that knowledge can influence


"loyal










knowledgeable about the likelihood of an outcome prefer betting on their judgments over an


equiprobable chance event, even though the former is more ambiguous than the latter.


In one of


their studies, subjects predicted the outcome of 14 football games each week for five consecutive


weeks.


For each game, subjects selected the team they thought would win and also judged the


probability of winning.


They then rated their knowledge about the game.


Afterward, subjects


were given the option of betting either on the team they chose or on a matched chance lottery.

For both high and low knowledge subjects, choice based on judgment was a function of the


judged probability.


However, when the judged probability was above 0.5, the high knowledge


subjects were more likely to choose a bet on the outcome of the football game that they had


earlier predicted than a matched chance lottery.


Heath and Tversky suggest that experience is one


of the factors that can lead to perceived competence.

To conclude, certain types of knowledge about the incumbent brand lead to loyal behavior


under


ambiguous


decision


conditions


rendering


the attack


brand


noncomparable


along


dimensions perceived to be diagnostic for choice.


This implies that experience will lead to loyal


behavior under ambiguous decision contexts, but there are certain conditions under which this


effect should not be observed.


First, when the competing brands have a set of non-overlapping


attributes, an opportunity to try the new brand may "disambiguate" the situation considerably and


tilt one's preference toward the new, superior brand.


Second, it was argued that resistance to


attack stems from an initial presumption that the incumbent brand was superior, coupled with an


inability to compare the incumbent and attacker on attribute dimensions.


Therefore,


when a


person has reasons to believe that the chosen brand is not necessarily the best brand, experience


may not produce overconfidence.


For example, if a person learns that the chosen brand is not the










overconfidence and resistance to a superior competitor's attack.



Other Processes Leading to Loyalty


Confidence acquired through knowledge structures may not be the only mechanism driving


loyalty under ambiguous decision conditions. Ir

quo options (Samuelson and Zeckhauser 1988).


i general, people prefer to stay with their status

Samuelson and Zeckhauser demonstrated the


status quo bias in a variety of decisions, including hypothetical choices about jobs, automobile


colors, and financial


investments.


Alternative versions of


each decision were presented


subjects.


In one of the versions an option was designated as the status quo.


In the other (neutral)


version, no option was singled out as the status quo.


The results offered strong support for the


existence of status-quo bias.

Samuelson and Zeckhauser (1988) invoked the status quo bias as an explanation of brand


loyalty and pioneer brand advantage.


They noted that rational economic models that ignore status


quo effects will exaggerate individual response to changing economic variables such as price and


predict greater instability than is observed in the real world.


However, their account of factors


that can lead to status quo bias makes it even more all-encompassing than commitment.


Their


list of factors include cost of thinking, transaction costs, psychological commitment and loss


aversion.


Of these,


cost of thinking and transaction costs are controlled in my empirical studies.


The disadvantages of using psychological commitment, without specifying its antecedents, as an


explanation of loyalty have been already discussed.


This leaves overconfidence (not suggested


by Samuelson and Zeckhauser) and loss aversion as the key factors explaining status quo bias or


loyalty in my framework.


The mechanisms accounting for overconfidence have been discussed








13

Tversky and Kahneman (1991) have discussed the role of loss aversion in riskless choices.

The status quo tendency, according to Tversky and Kahneman, may be driven by "loss aversion."

The basic intuition concerning loss aversion is that losses or outcomes below a reference state


loom larger than corresponding gains.


In other words, the loss of utility associated with giving


up a valued good is greater than the utility gain associated with receiving it.


Thaler (1985)


labeled


this phenomenon


the endowment


effect.


Kahneman,


Knetsch


Thaler


(1990)


demonstrated this phenomenon in a series of experiments.

In Tversky and Kahneman's framework of value analysis, reference levels play a large


role in determining preferences.


The same two objective states, depending on reference levels,


can be framed by sellers and evaluated by consumers as gains or losses.


Outcomes below the


reference state are considered to be losses; those above the reference state are gains.


While this


analysis focuses on how a reference state affects preference, it does not provide any answer for


the question what is or what determines a reference state.


A comprehensive answer for this


question is beyond the scope of this paper (see Kahneman and Millar 1986 and Simonson and


Tversky 1992 for a set of processes).


However, I will try to link the variables causing knowledge


structures and reference states specifically as they apply to brand choices.

The reference state itself is a knowledge structure and thus it is reasonable to expect the


factors influencing confidence to affect the reference state also.

different from the overconfidence phenomenon discussed earlier.


However, it is qualitatively

Compared to a set of clear


expectations about the incumbent brand, one is uncertain about the quality of the attack brand.

Thus, in brand evaluation, benefit beliefs and expectations derived from experience cause the

current state (incumbent brand) to be the reference point and giving up this brand constitutes loss.








14

adequate reasons to justify the choice of the attack brand, giving up the incumbent brand may no


longer be perceived as loss.


If the persistence of choice of the incumbent brand does not involve


attribute or benefit valuation, it amounts to the concept of sheer inertia in choice proposed by

Jeuland (1979).

The extent to which loyal behavior in our choice situations can be explained by loss


aversion and inertia is an empirical question.


In the experiments to be discussed in the subsequent


chapters


the overconfidence


explanation


was tested


against


the loss


aversion


and inertia


explanations through verbal protocols.


In the subsequent chapters the results of these analyses are


presented.














CHAPTER III
EXPERIMENT 1: BRAND KNOWLEDGE AND LOYALTY


The main objective of this experiment was to show that either extensive experience with

the incumbent brand or thinking in terms of the benefits of the incumbent brand can cause

resistance to attack by an "objectively" superior brand of a competitor when the decision situation


is ambiguous.


In the present experiment, ambiguity was created by making attribute information


about the incumbent brand unavailable when subjects encountered the attack brand.


The second


objective was to identify the processes that lead to loyal behavior.

Natural ingredients based hand and body lotions were used as stimuli in this experiment


as well as in the subsequent experiments.


The choice of product category was based on three


considerations.


First,


product evaluation through experience should be ambiguous.


In the


category of natural lotions, one needs a high degree of expertise to discriminate brands based on


trial alone.


Second, a category was needed in which subjects would examine attribute values and


then form beliefs and expectations based on these values.


reinforced by experience.


These initial beliefs needed to be


This required a product category that could be easily tried in an


experimental setting.


The category of hand and body lotions satisfied all these criteria.


Only


"natural" lotions were chosen to suppress the effects of brand familiarity.


A series of protests


were conducted to choose brand names and to check the manipulations of experience and benefit

belief formation.










Pretests


Pretest


The objective of this pretest was to select the two most liked brand names among a set


of eight.


These would be used later as the names of incumbent and attacking brands.


These eight


names were generated by the experimenter and two of his colleagues. The data were collected

from female students enrolled in an undergraduate marketing course. Thirty-six respondents

evaluated each of the eight fictitious brand names on a nine-point scale. The names (and mean


evaluations) were as follows: Botanical Spring (5.4), Nature's Nurture (4.7), Herbal Wonder (3.7),


Essence of Nature (5.7), Herbal Abundance (3.98), Natural Blossom


(5.5), and Garden Exotica


(4.6).


It was decided to assign names of equal likability to the "incumbent"


and the "attack"


brands. Accordingly the "incumbent" and the "attacker" were named Botanical Spring and Natural

Blossom, respectively.


Pretest


The objective of


pretest


was to calibrate


the number of trials


needed


for the


"extensive" and "limited" experience conditions.


four and six.


Number of trials was varied at three levels: one,


The dependent variable was "relative attitude," taken immediately after experience.


Relative attitude was defined as the difference in ratings between the most highly rated brand and


the second most highly rated brand. Twenty-one female students enrolled in a Principles of

Marketing course participated in this experiment. The choice set consisted of four brands of


natural lotions.

attribute matrix.


Subjects were asked to choose a brand based on information given in a brand x

After choice, they were asked to use the brand once, four times, or six times.


After the last trial, subjects rated their liking toward each brand on a nine-point scale.


Number










After the last trial, subjects rated their liking toward each brand on a nine-point scale.


Number


of trials explained 28% variance in relative attitude (F 2.20 =3.02, p


< 0.08).


Relative attitude was


higher in the six trials condition than in the one trial condition (1.5 versus -0.04, p<0.05).


other comparisons


yielded significant differences.


Hence,


was decided


to operationalize


extensive experience as six trials and limited experience as a single trial.

Pretest 3


The objectives of this pretest were to check the manipulation of benefit beliefs, to examine

the effects of experience on choice confidence, and to replicate the effects of experience on


relative attitude.


Forty female undergraduates enrolled


in a Principles of Marketing


course


participated in this experiment.


subjects in each of the four experimental conditions.


design was used in this experiment with ten

The two factors were experience (one trial


versus six trials) and benefit beliefs formation (rating the beliefs on a given scale versus rating

their overall evaluation).

Procedure


As in the second pretest, subjects chose a brand of lotion from a set of four based on the


attribute values given in a brand x attribute matrix. Subjects experienced the brand chosen one

time (limited experience) or six times (extensive experience). Afterward, subjects in the High


Benefit Beliefs Crystallization condition were asked to rate their beliefs on a rating scale (see

Appendix C). Subjects in the Low Benefit Beliefs Crystallization condition were asked to express


their overall evaluation of the four brands in the choice set.

rated beliefs would think about the benefits of the brand. It


It was expected that subjects who


was expected that this manipulation


would lead to the formation of benefit beliefs or crystallization of existing beliefs.


Throughout








18

Subjects in all conditions next were asked to express their confidence in the quality of the


brand chosen on a nine-point scale.


They then were asked to return after 48 hours.


In the second


session, subjects were asked to record whatever information they remembered about the brand they


had chosen and experienced.


This measure assessed the level of benefit belief formation.


It was


expected that belief formation would be high in all conditions except the Limited Experience -


Benefit Beliefs


Formation condition.


Further,


subjects'


responses to


the attitude


confidence scales during the first session were used to compute relative attitude and choice


confidence scores in each condition.


Although all the subjects responded to the confidence scale,


only 20 (who were in the Low Benefit Beliefs condition) responded to the attitude scale.

The results supported the finding of the earlier pretest that relative attitude was higher in


the six trials condition than in the one trial condition.


The mean in the six trials condition was


1.73 compared to the mean of 0.83 in the one trial condition (F


1.18 = 4.61, p<0.05).


Experience


explained 21% variance in relative attitude.

The information recalled on the second day was coded in terms of attribute-related


benefits, accurate attribute information and inaccurate information and/or irrelevant beliefs.


example,


"Botanical Spring can moisturize the skin" is a benefit belief formed on the basis of


attribute information.


On the other hand,


"Botanical Spring is pink in color" is an irrelevant


belief, as there was another lotion of the same color. The experimenter as well as a colleague

blind to the experimental conditions coded the responses. The coders agreed on about 90% of


brand beliefs initially.


Differences were resolved after discussion.


The following table presents


the number of benefit beliefs recalled in each condition.













TABLE III-1


NUMBER OF BENEFIT BELIEFS RECALLED IN DIFFERENT LEVELS OF
EXPERIENCE AND BELIEF CRYSTALLIZATION


The number of benefit beliefs recalled in the Low Experience


condition was lower than in any other condition.


- Low Belief Crystallization


Thus, either extensive experience or thinking


in terms of benefits can lead to higher degree of availability of benefit beliefs.


The effect of the


manipulation of belief rating on the number of benefit beliefs recalled was very strong (F


36=11.7


4.5, p


< 0.01).


< 0.05 ).


The effect of experience on the number of beliefs recalled was also strong (F


The effect of rating the benefits on beliefs recall was higher in the Limited


Experience condition (F

,,=2.67, p <0.12). Thu


, ig=10.14 p <0.01) than in the Extensive Experience condition (F


s, either experience or thinking in terms of benefits can account for


differences in knowledge structure.


The means for choice confidence are given in Table III-2.


Once again, choice confidence


in the Extensive Experience condition was higher than in the Limited Experience condition ( 6.7


versus


F 36=4.71, p<0.05).


The manipulation of thinking in terms of benefits also had a


high effect on confidence (5.4 in the Low condition versus 6.8 in the high condition; F


i.366.41,


Experience Belief Crystallization Low Belief Crystallization High


Limited 1.1 2.5


Extensive 2.1 2.8










TABLE III-2


CHOICE CONFIDENCE IN DIFFERENT LEVELS OF
EXPERIENCE AND BENEFIT BELIEFS FORMATION


As predicted, the effect of belief crystallization on confidence was higher in the limited

experience condition than in the extensive experience condition. Collectively these results support

the prediction that either experience or thinking in terms of benefits is sufficient to influence


consumers' knowledge structures.


The effects of these factors on resistance to competitor's attack


were tested in experiment 1.


Method


Subiects


Subjects for this experiment were volunteers recruited from a Principles of Marketing


class.


Because hand and body lotions are primarily directed toward women (Consumer Reports,


August 1977, p 449), only female subjects were included. As compensation for their participation,


the subjects earned extra credit in the class. Afte

below, responses were obtained for 88 subjects.


time.


:r excluding data for 14 subjects for reasons given

In most sessions, subjects participated one at a


Whenever there were two or more subjects in a session, each subject performed the choice


and experience tasks individually, while the others) waited in the adjacent room.


On average,


Experience Benefit Beliefs Low Benefit Beliefs High


Limited 4.6 6.4


Extensive 6.2 7.2










used them less frequently.


Only 15 subjects used natural lotions.


Procedure


Subjects were invited to participate in a lotion choice experiment. They were told

beforehand that they would be asked to use the brand(s) of lotions chosen by them. To prevent


random brand switching, they were also told that about 15% of the subjects participating in the

experiment would receive the brand chosen by them at certain critical rounds (known only to the

experimenter) as gift. After these instructions, subjects' were asked to read a story that provided


information about the key ingredients of the products.


This story familiarized subjects with the


importance of the key attributes of natural lotions and the benefits derived from these attributes.


To construct this story, six experts in the category of lotions were interviewed.


They included


professional beauticians managing skin care studios and managers of stores selling natural skin


care products.


Further, information about the herbs used in cosmetic products was collected


through pamphlets circulated by the distributors of natural beauty products.


Magazine articles on


skin care published in Glamour and Cosmopolitan served as additional sources of information (see

Appendix A for the story).

Subjects then saw a brand x attribute matrix that described four fictitious brands in terms


of seven attributes (see Appendix B).


maximum attribute value.


Subjects were asked to choose a brand that had the


One of these brands, Botanical Spring, was superior to all other brands


in the choice set.


Hence, it was expected that every subject would choose this brand.


All but


nine subjects chose Botanical Spring (hereafter referred to as the incumbent).


The data for those


who chose other brands were excluded from analysis.

Once they made their initial choice, subjects were asked to try all the four brands in the










the information given in the matrix.


This deception was necessary to minimize curiosity-related


brand switching in the latter half of the experiment.

Subjects in the Limited Experience condition were asked to try the chosen brand once,

while those in the Extensive Experience condition were asked to try the chosen brand six times.

The cover story given to those in the Extensive Experience condition was that previous research


had shown that people can evaluate lotions more accurately after a series of trials.


After every


trial, subjects in this condition were given paper towels to wipe off any lotion build-up.


Afterward, subjects in the High Benefit Beliefs


condition rated their beliefs about the


benefits of the chosen brand on an eight-item, nine-point scale while those in the Low Benefit


Beliefs Formation condition were assigned an irrelevant task.


These scales (see Appendix C) were


constructed after consultation with experts to ensure that the benefits listed mapped onto the


attributes given in the matrix.


Subjects in all conditions were then asked to rate their confidence


in the overall quality of the chosen brand on a single-item nine-point scale.

latency of evaluation was measured via computer for all subjects. Subjec


return after two days to complete the experiment.


Finally, the response


:ts were then asked to


Because there was no variation in latency


across conditions, this measure will not be discussed further.

On returning, subjects were asked to assume that they were going on a shopping trip to

buy lotions and that the market conditions had not changed since they made the original choice.


The task was to choose a brand among the four brands in the original choice set.


that everyone would choose the incumbent brand (Botanical Spring).


It was expected


Data from five subjects who


did not choose the incumbent brand were excluded.


Subiects were then given the critical choice task.


They were asked to assume that once








23

attribute brands for these brands were directly available, information about the incumbent brand


needed to be recalled to make a choice (see Appendix D).


One of the two new brands, Natural


Blossom, was objectively superior (dominating) to the other new brand, Botanical Glory, as well


as to


Botanical Spring.


However, decision ambiguity was created across all conditions via


memory


constraints.


At this critical


round subjects had the option of


choosing either the


incumbent brand or one of the two new brands. Botonical Glory was not chosen by any subject.

Thus, their choices were limited to Botanical Spring (the incumbent) and Natural Blossom (the


attack brand).


(Subjects were told to assume that going back to the store that stocks the original


set of brands would not involve any cost.)

the reasons for their choice. They were


After they chose a brand, they were asked to state


also asked to judge which of these two brands was


superior in terms of attribute values and how confident they were about this superiority judgment.

Analysis


A binomial logit approach was used to analyze choice data.

the proportion of subjects choosing the incumbent brand. Data


CATMOD procedure of SAS.


The dependent variable was


I were analyzed through the


The focus was on the main effect of experience and also on


comparing the simple effect of benefit beliefs formation in the Limited Experience condition with


the simple effect in the Extensive Experience condition.


Further,


choice proportion in each cell


was compared to 0.5 (random choice).


in the Limited Experience


It was predicted that the proportion should be below 0.5


- Low Benefit Beliefs Formation cell and above 0.5 in the other three


cells, despite the objective superiority of the attack brand.

I also analyzed two continuous dependent variables in this experiment: confidence about

the quality of the incumbent brand before the attack brand was introduced (a monodic judgment)










equated with loyalty and considered to be the direct antecedent of resistance.


The monodic


confidence judgment was measured to gain insights about the process that leads to loyalty.


two-factor, between-subjects ANOVA was used to test the effects of experience, benefit beliefs


and the interaction of these two variables on each of the dependent variables.


Furthermore, each


of these confidence judgments was used as a covariate in analyzing the effects of the independent


variables on choice.


This analysis was performed through the LOGIST procedure of SAS.


judgment of superiority


as well as the confidence in this


judgment may shed


light on the


processes) that might have led to choice of one brand over the other.



Results


Choice, Experience, and Benefit Beliefs


The proportion of subjects who chose the incumbent brand under different levels of


experience and benefit beliefs formation is given in Table III-3.


As expected, the proportion of


subjects choosing the incumbent brand was much less (14%) in the Limited Experience-Low


Belief crystallization condition than in the other three conditions (68% to 82%).


This cell


replicates the results of studies that report a pervasive tendency to avoid uncertainty (Biehal and


Chakravarti 1983, 1986).


However, extensive experience or thinking in terms of benefit beliefs


seem to provide diagnostic cues that counter the effect of uncertainty.


In each cell the proportion


was in the expected direction


That is, less than 0.5 in the Limited Experience


- Low Belief


crystallization


condition


more


0.5 in the other three


conditions.


The pair-wise


comparisons


indicated


that the


other significant


difference


was between


the Limited


Experience


- High Benefit Belief Crystallization condition and the Extensive Experience High










TABLE III-3


CHOICE PROBABILITIES FOR THE INCUMBENT BRAND UNDER DIFFERENT
LEVELS OF EXPERIENCE AND BELIEF CRYSTALLIZATION


Experience Low Belief Crystallization High Belief Crystallization


Limited 3/22 (0.14) 15/22 (0.68)

Z(p<0.5)=3.58** Z(p>0.5)= 1.79*


Extensive 16/22 (0.73) 18/22 (0.82)

Z(p>0.5)=2.28* Z(p>0.5)=3.18**

p < 0.05


< 0.01


The result of interest is that thinking in terms of benefit beliefs mattered in the limited

experience condition, but did not have any effect in the extensive experience condition (see Table


III-4).


These results combined with the results of pretest 3, offer support for the proposition that


either experience or availability of benefit beliefs is sufficient to cause resistance to competitor's

attack by making consumers acquire certain types of knowledge about the incumbent brand.

However, these results do not shed much light on the process by which these knowledge structures


influence loyalty.


In chapter II, it was


proposed that overconfidence in comparative judgment


TABLE 111-4

THE EFFECT OF BENEFIT BELIEFS FORMATION ON CHOICE
(LIMITED VERSUS EXTENSIVE EXPERIENCE)


Ynepripnen


rhi-.Srunrie (CWld Stntictic~


.-.t. II












leads to resistance.


Analysis of


confidence data along with the stated reasons for choice


constitutes a test of this proposition.


Table III-5 presents the results using monodic choice


confidence judgment as the dependent variable.

TABLE III-5


EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CONFIDENCE IN CHOICE
BEFORE THE INTRODUCTION OF THE ATTACK BRAND


As in the pretest, the effects of experience (F


(F 1, 4=20.29, p<0.01) were strong.


, 84=3.55, p<0.07).


i, 4=21.8, p<0.01) and belief crystallization


The interaction of these two factors was marginally significant


A comparison of simple effects revealed that the effect of thinking in terms


of benefits was stronger in the limited experience condition (F


1, 42=19.5, p<0.01) than was in the


extensive experience condition (F


, 42=3.6, p<0.07).


However, these effects did not lead to


resistance in all the cases.

An analysis of reasons for choosing the incumbent over the attack brand suggests that


resistance is not always due to confidence.


Subjects'


judgments regarding which of the two


brands was superior and confidence in these judgments were analyzed to understand the process

that leads to loyal behavior. If one judges the "objectively inferior" incumbent brand to be superior


Experience Benefit Beliefs Low Benefit Beliefs High


Limited 4.59 6.36


Extensive 6.41 7.14








27

incumbent brand but judges the attack brand as superior in terms of attribute values, then her


choice is driven by other considerations such as risk aversion or sheer inertia.


This may be also


true if she chooses the incumbent brand and judges the incumbent brand and the attack brand to

be of the same value.

Similarly, when a person chooses the attack brand and considers this brand to be superior

to the incumbent brand, she accurately judges the quality of brands and exhibits appropriate


confidence in this judgment.


On the other hand, if she chooses the attack brand and considers the


incumbent brand to be superior, she exhibits a variety seeking behavior.


Thus, there are at least


five different processes that can lead to choice: overconfidence, risk aversion, inertia, accurate


judgment of attribute values and variety seeking.


The first three of these explain choice of the


incumbent brand while the latter two explain the choice of the attack brand.


As it is extremely


difficult to disentangle the loss aversion explanation from the inertia explanation, these two

categories will be combined.


Of the


88 subjects


who participated


in this experiment,


54 (61


percent)


chose the


incumbent brand and 34 (39 percent) chose the attack brand.


incumbent brand,


Of the 54 subjects who chose the


40 judged it to be superior to the attack brand in terms of its attribute values.


Of these 40, 4 subjects expressed low confidence in their judgments (below the midpoint value


of 5).


In a sense these subjects also can be classified as risk aversive.


Thus, only 36 out of 88


subjects (about 40 percent) were loyal to the incumbent due to overconfidence in judgment.

However, in the conditions in which we would expect higher level of confidence (extensive

experience and / or thinking in terms of benefits), this proportion is higher 36 out of 66 or 55


percent.


Among those


----- -


who chose


Botanical


Snfing.


36 out of 54 (67


percent) exhibited


r-,










and choice.


The following table classifies subjects according to the reasons for their choices


inferred from their superiority judgments.



TABLE III-6


DIFFERENT CONSIDERATIONS FOR CHOICE


Confidence in Comparative Judgment as a Function of Experience and Benefit Beliefs Formation


In the earlier sections I presented results of the analyses using choice confidence measured

on the first day (immediately after the experience and the belief crystallization manipulations) as


the dependent variable.


In this section, I use confidence in superiority expressed on the second


day (after subjects chose between the incumbent brand and the attack brand).


The effect of experience was significant (F


Formation of benefit beliefs was


marginally significant (F


1. 8=2.8, p<0.1).


The interaction was nonsignificant.


The simple effect


contrast across levels of experience revealed that the effect of benefit beliefs was significant in


.I V 1 J ** *I ** r A n n- *.4 -


CHOICE


BOTANICAL SPRING






NATURAL BLOSSOM


REASONS FREQUENCY


Overconfidence 36

Risk Aversion and Sheer 18
Inertia

Accurate Judgment of
Superiority Based on 24
Attribute Values
10
Variety Seeking


, g4=23.3, p<0.01).










Once again the effect of experience was significant (F


, 6=-17.7, p< 0.01)


the effect of


benefit beliefs also was significant (F 1 60-4.86, p


< 0.04).


There was no interaction (F


1.63=1.86,


p <0.2).


The pairwise comparisons indicated that confidence was less in the Limited Experience-


Low Benefit Beliefs condition than in any other condition; none of the other three cells differed.

TABLE III-7


EFFECTS OF EXPERIENCE AND BELIEF CRYSTALLIZATION
ON CONFIDENCE IN COMPARATIVE JUDGMENT (ALL SUBJECTS)


TABLE III-8


EFFECTS OF EXPERIENCE AND BELIEF CRYSTALLIZATION ON CONFIDENCE
IN COMPARATIVE JUDGMENT FOR THOSE WHO EXHIBITED
JUDGMENT-CHOICE CONSISTENCY (N=64)


Experience Benefit Beliefs Low Benefit Beliefs High


Limited 4.05 (n=19) 5.07 (n=14)


Extensive 5.64 (n= 14) 5.88 (n= 17)

The results discussed thus tar demonstrate that (i) experience or belief crystallization is


a sufficient condition for loyalty, (ii) experience or belief crystallization is a sufficient condition


Experience Benefit Beliefs Low Benefit Beliefs High


Limited 3.95 4.68


Extensive 5.45 5.55








30

aversion and inertia, and (v) brand switching can be explained in terms of accurate superiority


judgments or variety seeking.


The first three conclusions imply that choice confidence and/or


confidence in comparative judgment mediate the relationship between the independent variables


and choice.


In the next section, a direct test of this mediational hypothesis is presented.


Mediational Tests


According to Baron and Kenny (1986), a proper test for the hypothesis that a set of

independent variables affect a dependent variable through a mediator involves three steps: (1) the

independent variables should directly affect the dependent variable, (2) the independent variables

should affect the mediator, and (3) when the dependent variable is regressed over the independent

variables and the mediator, the mediator should have an effect and the effect of the independent


variable should be weakened.


The following tables present these tests for the mediators of choice


confidence


and confidence


in comparative


judgment.


These


maximum


likelihood


logistic


regression analyses were done through the Logist procedure of SAS.


The results for all subjects


are presented followed by results for only those subjects whose superiority judgment and choice

were consistent.

Confidence in Overall Quality As the Mediator


Tables III-9, III-10, and III-11 present the analyses for all subjects.

TABLE III- 9


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CHOICE


Variable


Std.Error


Experience


Chi-Square


12.14


0.006










TABLE III- 10


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CONFIDENCE


Variable


Std.Error


Chi-Square


Experience


0.001


Benefit Beliefs


0.004


TABLE III- 11


THE EFFECTS OF EXPERIENCE, BENEFIT BELIEFS & CONFIDENCE
ON CHOICE


Variable


Std.Error


Chi-Square


Experience


Benefit Beliefs


0.17

0.015


Confidence


Tables III-12, III-13,


and III-14 present the analyses using only those subjects whose


superiority judgments were consistent with their choices.

TABLE III-12


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.002


Benefit Beliefs


0.002


TABLE III-13


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CONFIDENCE










TABLE III-14


THE EFFECTS OF EXPERIENCE, BENEFIT BELIEFS & CONFIDENCE
ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.217


Benefit Beliefs


0.216


Confidence


0.099


The first set of logistic regression results (Tables III-9 to III-11) indicate that there is no


support for the mediational hypothesis. In particular,


Table III-11 shows that


confidence does


not have any significant effect on choice.


In the second set of analyses that include only those


subjects whose superiority judgments were consistent with their choices (Tables III-12 to III-14),


there is very weak support for the proposition


that experience and


benefit beliefs lead


confidence and therefore to loyal behavior.


Overall, the monodic judgment of confidence in the


quality of the brand does not appear to have a mediational role that leads to loyal behavior.

next set of mediational tests pertain to the variable of confidence in comparative judgment.

Confidence in Comparative Judgment as the Mediator


Tables III-15 to III-17 present analyses that included all subjects.

experience and belief crystallization have strong effects on choice. In 1


In Table III-11 A, both


Fable III-16, experience


a strong


effect


on confidence


comparative


judgment,


while


the effect


of belief


crystallization is marginal.


Table III-17 gives the critical test for the mediational hypothesis.


Here, confidence has a significant effect on choice.


beliefs on choice are weakened.


Further, the effects of experience and benefit


Inclusion of confidence in comparative judgment in the model










TABLE III-15


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CHOICE


Variable


Std.Error


Experience


Chi-Square


12.14


0.001


Benefit Beliefs


0.002


TABLE III-16


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CONFIDENCE IN
COMPARATIVE JUDGMENT


Variable


Std.Error


Chi-Square


Experience


18.05


0.001


0.235


Benefit Beliefs


0.066


TABLE III-17


THE EFFECTS OF EXPERIENCE, BENEFIT BELIEFS &
CONFIDENCE IN COMPARATIVE JUDGMENT ON CHOICE


Variable


Std.Error


Chi-Square


Experience


Benefit Beliefs


0.005


0.225


Confidence


In this case, because the effect of benefit beliefs on confidence is not so strong, no firm

statement can be made about the mediational process. A subsequent analysis involving only those

subjects whose superiority judgments were consistent with their choices was performed (Tables

III-18 to III1-20).

The results in the preceding tables offer stronger support to the mediational hypothesis.










TABLE III-18


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.002


Benefit Beliefs


0.002


TABLE III-19


THE EFFECTS OF EXPERIENCE AND BENEFIT BELIEFS ON CONFIDENCE IN
COMPARATIVE JUDGMENT


Variable


Std.Error


Experience


Chi-Square


14.83


0.001


Benefit Beliefs


TABLE III-20


THE EFFECTS OF EXPERIENCE, BENEFIT BELIEFS &
CONFIDENCE IN COMPARATIVE JUDGMENT ON CHOICE


Variable


Std.Error


Chi-Square


Experience

Benefit Beliefs


Confidence


confidence as a predictor of


formation.


choice weakens the effects of


The beta value of experience goes down from


goes down from 2.12 to 1.81.


The final requirement pro;


experience and benefit beliefs


2.12 to 1.31 and that of benefit beliefs

posed by Baron and Kenny (1986) is


that the correlation between the mediator and the dependent variable as well as the correlation










variables and confidence are not high.


Thus, experience and benefit beliefs influence choice by


affecting confidence in comparative judgment.

Discussion


In this experiment, decision ambiguity was uniformly high across experimental conditions.

Ambiguity was created by withholding specific attribute information about the incumbent brand


when subjects encountered the attack brand.


It was predicted that experience and thinking in


terms of benefits lead to certain types of knowledge that induce confidence in comparative


judgment and thereby loyal behavior.


In pretest 3 there was evidence that either experience or


thinking in terms of benefits can cause a set of beliefs to be highly available.


The results of


experiment 1 suggest that either experience or belief crystallization can induce confidence about


the quality of the chosen brand. These variables can also cause overconfidence in comparative

judgment that leads to loyal behavior. Thus, under ambiguous decision conditions certain types


of knowledge induce confidence and thereby resistance to competitor's attack.


However, loyal


behavior under ambiguous decision conditions was not solely driven by overconfidence; there


were also other processes at work. Approximately one-third of the cases loyal behavior could be

explained in terms of loss aversion or inertia. In the cases in which judgment of superiority and


choice


were consistent,


overconfidence seems


to be the dominant process explaining loyal


behavior.

The choice results suggest that under extensive experience or high belief crystallization,


the incumbent


brand


comes


out as the


winner.


Does


experience


combined


belief


crystallization always lead to loyalty?


Are there any conditions under which the attack brand wins


share despite consumers' extensive experience with the incumbent and high belief crystallization?











CHAPTER IV


EXPERIMENT


LOYALTY AND AMBIGUITY DUE TO MEMORY CONSTRAINTS


The objective of this experiment was to examine the effect of ambiguity on resistance to


competitor's attack.


Ambiguity was manipulated by varying the accessibility of the attribute


values of the incumbent brand.


High ambiguity in Experiment


was created by imposing


imperfect memory for the attribute values of the incumbent brand.


It was argued that experience


or thinking in terms of benefit beliefs facilitated resistance to competitor's attack because of


decision ambiguity.


In one condition of Experiment


the decision context is disambiguated by


providing an opportunity to make side-by-side brand comparisons on specific attribute dimensions.

Method


The general procedure of this experiment was similar to that of Experiment 1.

Experience- High Benefit Beliefs condition of Experiment 1 was retained. Withi


there was a manipulation of ambiguity.


The Extensive


n this condition,


In the High Ambiguity condition, the tasks of the


Extensive Experience-High Benefit Beliefs cell of Experiment 1 were exactly replicated.


In the


Low Ambiguity condition, the only difference was that specific attribute information about the


incumbent brand was available when the attack brand was introduced.


It was hoped that this


would facilitate brand comparisons and would weaken the biases in choice due to extensive


expenence


and beliefs


crystallization.


Twenty


eight


female


students


participated


in this


experiment to earn credit in a Principles of Marketing course.

Results

The results given in Table IV-1 support the hypothesis that when decision ambiguity is

reduced, consumers choose a brand that is superior to others in terms of attribute values.










TABLE IV-1


THE EFFECT OF AMBIGUITY ON LOYALTY


High Ambiguity: Memory Constraints

(replicates a cell from experiment 1)


Chi-Square (Wald Statistic)= 3.42; Prob <0.07.


The results show that, as in Experiment

facilitate resistance under high ambiguity. Howe'


1, experience and benefit beliefs formation


ver, when ambiguity is removed by providing


attribute


information


about


the incumbent


brand,


a majority


subjects


(65%)


chose


"objectively superior"


attack brand.


It is surprising that even in this condition about 35% of


subjects exhibited loyal behavior.


The following tables provide subjects' reasons for choice.


logic of


classification is the same as in Experiment


The classifications were based on


superiority judgments and confidence in these judgments.


ambiguity.


The results are given for each level of


In the High Ambiguity condition, as in Experiment 1, a substantial proportion of


subjects chose Botanical Spring due to overconfidence in comparative judgment (8/14 or 57%).

About 13% subjects in this condition chose Botanical Spring to avoid risk or out of sheer inertia.


About


(3/14) in this condition chose Natural Blossom by judging superiority in terms of


attributes accurately.


One person chose this brand due to variety seeking.


In the Low Ambiguity


cell, risk aversion or inertia explained the choices of all those who chose Botanical Spring, and


accurate judgment explained the choices of all those who chose Natural Blossom.


Because


Low Ambiguity: All Information Externally

Available


0.7 (10/14) 0.35 (5/14)










day).


TABLE IV-2


REASONS FOR CHOICE IN HIGH AMBIGUITY CONDITION


Brand Chosen Reasons Frequency


Botanical Spring Overconfidence 8

Risk Aversion or Inertia 2


Natural Blossom Accurate Judgment of 3

Superiority

Variety Seeking 1




TABLE IV-3

REASONS FOR CHOICE IN LOW AMBIGUITY CONDITION


Brand Chosen Reasons Frequency


Botanical Spring Overconfidence 0

Risk Aversion or Inertia 5


Natural Blossom Accurate Judgment of 9

Superiority

Variety Seeking 0


- '*T> Ia:Cn t- rl t, n -^ C- ---^^-^ -










terms of attributes.


Does this imply that availability of attribute information for the incumbent


inevitably weakens the effects of experience on resistance?


Experiment 3 examines this question.














CHAPTER V
EXPERIMENT 3: LOYALTY AND AMBIGUITY DUE TO
NON-OVERLAPPING ATTRIBUTES

In the first two experiments, decision ambiguity was operationalized in terms of memory


constraints.


However, this may not be the only form of ambiguity. In a number of product


categories, a common form of ambiguity occurs when two brands have a set of non-overlapping


attributes but claim the same benefits.

same benefit of keeping skin healthy.


For example, Aloe-Vera and Vitamin E may produce the

If two competing brands emphasize different attributes,


choice difficulty increases, especially when a consumer has knowledge about the benefits these


attributes can provide.


Thus, there will be less comparability and high ambiguity in these choice


situations.


Further, unlike Experiment


availability of specific attribute information will not


reduce ambiguity or facilitate side-by-side brand comparisons.


In Experiment 3, high decision


ambiguity was created across conditions by describing the incumbent brand and the attack brand


in terms of a set of different attributes.


The objective of this experiment was to generalize the


findings of experiment


and also


to explore the effect of availability of specific attribute


information when ambiguity is caused by noncommon attributes.

Pretest4


The objective of this pretest was to identify sets of noncommon attributes of natural


lotions that do not vary in terms of importance and desirability.

marketing course volunteered to participate in this pretest. T


Eighteen students enrolled in a


bhey read a story that listed the


benefits derived from the key attributes of all natural hand and body lotions.


After two days they










importance and the desirability of each of these attributes on a nine-point scale.

familiarity ratings, some unfamiliar attributes were excluded (e.g., lemon grass).


and desirability ratings had a high correlation (r=0.89).


Based on the


The importance


I chose two sets of attributes on the basis


of these ratings so that both sets had attributes of almost equal importance and desirability.


of these sets contained the attributes of Aloe Vera, Cocoa Butter and Apricot Extracts and the


other set had the attributes of Vitamin E, Jojoba Extracts and Papaya Extracts.

desirability ratings of these sets were 5.62 and 5.38 respectively (t=1.45, p< 0.2).


The mean


The decision


was made to describe the incumbent brand in terms of the second set of attributes and the attack

brand by the first set of attributes, because the first set of attributes was rated slightly higher in

terms of desirability.


Design and Procedures


Subjects


One hundred and thirty two female subjects participated in this experiment.


Thirty were


enrolled in a General Psychology class and the remainder in a Principles of Marketing class.

compensation for participation, subjects earned credits in their respective courses.

Design


This experiment had a 3 x


fully crossed factorial design.


The factors were experience


and availability of specific attribute information.

limited (one trial) and extensive (six trials). Ava

two levels: available and not available. As in Ex


the incumbent brand was the key dependent variable.


Experience was varied at three levels: none,


inability of specific attribute information was at


periment 1, the proportion of subjects choosing


The other dependent variables were


confidence in initial choice measured immediately after subjects' experience with the incumbent










Procedure


As in Experiment


1, subjects first read a story that described the key attributes of all


natural lotions and the benefits derived of these attributes.


Then they saw a brand x attribute


matrix that described four brands of lotions in terms of seven attributes vitamin e and jojoba


extracts (see Appendix E). Subjects were asked to choose the best brand on the basis of attribute

information given in the matrix. Botanical Spring (the incumbent brand) was superior to other

brands in terms of overall value. As there was a No Experience condition, unlike in experiment


1, subjects were not given an opportunity to try all the four brands before deciding.


remainder of the procedure was identical to the first day procedure of experiment 1.

On returning, subjects saw the attack brand (Natural Blossom), which was described in

terms a different set of important attributes. For half of the subjects, specific attribute information


about the incumbent brand was not presented.


Others saw information about the incumbent and


the attack brand side-by-side (see Appendix F and Appendix G).


They were asked to choose


between these two brands, judge which of these brands was superior in terms of attribute contents,

and express confidence in this superiority judgment.

Results


Analysis followed the pattern of expe

binomial weighted least square logit models.

confidence data, and binomial maximum like


The choice data were analyzed using


Between-subjects ANOVA models were used for


blihood logistic regression models were used to


conduct mediational tests.


The choice proportions pointed to a main effect of experience.


Neither


the effect of availability nor the interaction between availability and experience was significant.


Table V-l presents the choice proportions in the various experimental conditions.


As expected.








43

condition choice proportion was unexpectedly high when specific attribute information was not


available.

was high.


However, this proportion did not differ very much from the cell in which availability

In the Limited Experience conditions, about one half of subjects chose the incumbent


brand.


Thus, the choice pattern differed at each level of experience (Z 0


vs 1 &


Z lvs6


< 0.05).


TABLE V-1


PROPORTION OF SUBJECTS CHOOSING THE INCUMBENT BRAND IN DIFFERENT
LEVELS OF EXPERIENCE AND AVAILABILITY


Experience


None


Availability of Specific

Attribute Information Low


9/22 (0.40)

Z (p<0.5)=1.01


Availability of Specific

Attribute Information High


6/22 (0.27)


Z (p<0.5)=


2.28 *


Limited 12/22 (0.545) 11/22 (0.50)


Extensive 16/22 (0.74) 17/22 (0.77)

Z (p>0.5)= 2.4 ** Z (p>.5)=2.71 **


= P<0.05


P<0.01


TABLE V-2


ANALYSIS OF VARIANCE OF THE EFFECTS OF EXPERIENCE AND AVAILABILITY


Source

Intercept

Availability


CHI-SQUARE (WALD STATISTIC)

0.85


PROB

0.36










Confidence in the Overall Quality


Because the availability of specific attributes for the incumbent brand was manipulated


after the confidence measure was taken, this analysis included only the effect of experience.


following table presents the mean confidence for different levels of experience.

TABLE V-3


CONFIDENCE IN THE QUALITY OF THE INCUMBENT IN DIFFERENT LEVELS OF

EXPERIENCE


Experience accounted for about 15


% of variance in confidence (F 2,29 = 10.5, p < 0.001).


There was no difference between the limited and the extensive experience conditions (F 1,86 =2.56,


p<0.15).


However, the no experience condition differed from the limited experience (F


p<0.01) and the extensive experience (F


1. 8623.2,


7.84,


p<0.001).


Test of Mediational Hvoothesis Using Confidence in Overall Ouality as the Mediator


As in the first experiment,


a set of maximum likelihood logistic regression analyses were


performed to test the mediational role of choice confidence.


As availability of specific attributes


did not have any effect on choice, the model had experience as the only independent variable.


Experience Mean Confidence


None 5.57


Limited 6.45


Extensive 6.95










as the independent variables and choice as the dependent variable.


Tables V-4


V-5 and V-6


present the results of these analyses.


TABLE V-4


THE EFFECT OF EXPERIENCE ON CHOICE


Variable


Std.Error


Experience


Chi-Square


12.65


0.001


TABLE V-5


THE EFFECT OF EXPERIENCE ON CONFIDENCE


Variable


Std.Error


Chi-Square


Experience


0.063


14.55


0.001


0.165


TABLE V-6


THE EFFECT OF EXPERIENCE AND CONFIDENCE
ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.226


0.077


0.005


0.190


Confidence


0.233


0.130


0.082


Evidence for the mediational hypothesis is very weak.


First, in Table V-4C confidence


has only


a marginal


effect on choice


(p<0.08).


Second,


the unstandardized beta


value of


experience did not change (0.24 vs. 0.23).


Thus, there is no clear evidence that confidence in


initial choice mediated the effect of experience on choice after the attack brand was introduced.

Confidence in Comparative Judgment


Table


V-7 gives the mean confidence in each experimental cell.


Both experience and










availability was estimated.


Availability had an effect on confidence under No Experience (5.68


vs 6.5; F


1 42=4.81, p<0.04) and Limited Experience (5.63 vs 6.64; F


1 42=8.53, p<0.01).


TABLE V-7


CONFIDENCE IN THE SUPERIORITY OF THE CHOSEN BRAND UNDER
DIFFERENT LEVELS OF EXPERIENCE AND AVAILABILITY


Experience


Specific Attribute


Information Not Available


Specific Attribute

Information Available


None 5.68 6.50


Limited 5.63 6.64


Extensive 6.86 7.31


However, in the Extensive Experience condition, availability did not have any effect on confidence


(6.86 vs


7.31; F


, 42=1.9, p<0.2).


Despite low availability of attribute information, extensive


experience resulted in high confidence in comparative


judgment.


Further, across levels of


availability, there was no difference between no experience and limited experience, but extensive


experience resulted in higher confidence.


It must be noted that when confidence pertained to the


quality of the chosen brand (a monodic judgment), even limited experience was sufficiently


diagnostic.


However, only extensive experience produced confidence in comparative judgment.


The next issue concerns whether confidence in comparative

relationship between choice and the independent variables. As avail


judgment mediated the


ability did not influence


choice, experience will be the only independent variable in this analysis.










experience as a predictor of choice and confidence in superiority judgment as the mediator.

TABLE V-8


THE EFFECT OF EXPERIENCE ON CHOICE


Variable


Std.Error


Chi-Square


Experience


12.65


0.001


TABLE V-9


THE EFFECT OF EXPERIENCE ON CONFIDENCE IN COMPARATIVE JUDGMENT


Variable


Std.Error


Chi-Square


Experience


0.064


17.36


0.0001


TABLE V-10


THE EFFECT OF EXPERIENCE AND CONFIDENCE IN COMPARATIVE JUDGMENT
ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.226


0.078


0.003


0.199


Confidence


0.169


0.155


0.0001


This set of analyses does not support the hypothesis that confidence in comparative

judgment acts as a mediator in the process by which experience leads to loyal behavior. As


discussed in


the previous chapters,


overconfidence may not be


the only process


by which


experience leads to loyal behavior.


Hence, it was decided to do the same set of analyses after


excluding subjects whose superiority judgments did not match their choices.

There were 114 subjects (84 percent) whose superiority judgments were consistent with


their choices.


The patterns of choice proportions and confidence in comparative judgment for










TABLE V-11


THE EFFECT OF EXPERIENCE ON CHOICE


Variable


Std.Error


Chi-Square


Experience


15.76


0.001


TABLE V-12


THE EFFECT OF EXPERIENCE ON CONFIDENCE IN COMPARATIVE JUDGMENT


Variable


Std.Error


Chi-Square


Experience


0.069


15.34


0.001


0.197


TABLE V-13


THE EFFECT OF EXPERIENCE AND CONFIDENCE IN COMPARATIVE JUDGMENT
ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.281


0.085


10.83


0.002


0.237


Confidence


0.185


0.0001


These analyses reject the hypothesis that the effect of experience on choice is mediated


confidence.


This may be due to the high number of subjects in the Low


Availability


conditions choosing the incumbent brand, in spite of limited or no experience.


Hence, it was


decided to partition the total sample into a No Availability Group and an Availability Group and

conduct the maximum likelihood logistic regression analysis for each group.

Test of Mediational Hvoothesis When Specific Attribute Information is Not Available


TABLE V-14


THE EFFECT OF EXPERIENCE ON CHOICE










TABLE V-15


THE EFFECT OF EXPERIENCE ON CONFIDENCE IN COMPARATIVE JUDGMENT


Variable


Std.Error


Chi-Square


Experience


0.288


0.095


0.003


TABLE V-16


THE EFFECT OF EXPERIENCE AND CONFIDENCE ON CHOICE


Variable


Std.Error


Chi-Square


Experience

Confidence


0.194

0.078


0.114


0.10

0.0001


0.237


In this set of analyses, although experience has an effect on choice and confidence

independently, confidence does not seem to mediate choice.

Test of Mediational Hvypothesis When Specific Attribute Information is Available


TABLE V-17


THE EFFECT OF EXPERIENCE ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.136


12.27


0.001


0.364


TABLE V-18


THE EFFECT OF EXPERIENCE ON CONFIDENCE IN COMPARATIVE JUDGMENT


Variable


Std.Error


Chi-Square


Experience


0.107


0.002


0.246










TABLE V-19


THE EFFECT OF EXPERIENCE AND CONFIDENCE IN COMPARATIVE JUDGMENT
ON CHOICE


Variable


Std.Error


Chi-Square


Experience


0.148


0.264


Confidence


0.460


0.240


This set of analyses supports the mediational hypothesis.


effects on choice and also on confidence.


Experience has significant


When confidence was included in the model, the effect


of experience on choice is weakened (the unstandardized beta value goes down from


0.47 to


0.40).


Further, in this model the effect of confidence on choice is significant.


The same set of


analyses was conducted (by partitioning subjects in to Low


Availability and High Availability


groups) without excluding subjects whose choice and superiority judgment were inconsistent


pattern of results was the same: confidence mediated the effect of experience on choice in the

High Availability group, but not in the Low Availability Group.

Differences in the Results of Mediational Tests Between High and Low Availability Grouos


The results indicate a difference between the group in which specific attribute information

was available for the incumbent as well as the attack brand and the group in which specific


attribute information was available only for the incumbent brand.


Because decision ambiguity


should be high in either group, this differential pattern of results was not predicted.


Experience


had stronger effects on both choice and confidence in the High Availability condition than in the


Low Availability condition (0.36 vs. 0.14 for choice and 0.25 vs. 0.19 for confidence).


stronger effects may explain the differential pattern of results.


These


The small effect in the Low










Reasons For Choice


Apart from superiority judgments, subjects also gave specific reasons for their choices.


The experimenter used a coding scheme to classify these attributions.


A colleagues


who was


blind to


conditions in this experiment also coded these reasons.


There was approximately 85


percent agreement


between


the coders.


reasons


for choosing


Botanical


Spring


were


categorized into three classes: satisfaction with the brand/risk aversion, higher attribute value

(more of the same attributes), and different, more preferred attributes.

The typical responses in the first category were "I tried it liked it, and don't want to


switch."


Although satisfaction with the brand and risk aversion are two different phenomena, it


is difficult to disentangle them through verbal protocols.

confirmatory bias or positive testing strategy. If a person


In the same category one can put


n says that she liked the brand (or is


satisfied with the brand) and does not want to switch, it may mean that she assumes that she will


not like the new brand although she has never tried it.


Thus, the first category of responses may


indicate genuine liking or satisfaction with the incumbent, risk aversion and confirmation bias.


The responses in the second category included "

had more aloe, it had more cocoa-butter." People eith


.. the old brand had more ingredients, it


ler thought that the incumbent was higher


in all ingredients or specified a particular attribute that they thought was more important and


thought that the incumbent had more of this attribute. It should be recalled that in certain

conditions they saw the attribute values of the attack brand only. In these conditions some


subjects thought that the incumbent brand had the same features as the attack brand but at a higher

level.


In the third category subjects attributed their choices to differences in attributes. For










benefits of the incumbent brand are attributed to the ingredient combination of this brand.


may be other ingredients leading to the same benefits.


There


However, due to a tendency to confirm


rather than disconfirm their hypothesis, our consumers did not want to try the attack brand that


has a different combination of attributes.


Some people also said that they avoided certain


attributes that only the new brand possessed (e.g., papaya extract).


Table


V-20 presents


subjects'


stated


reasons


for choosing


Botanical


Spring


across


experimental conditions.


This table shows that when availability is low, risk aversion/liking was


the dominant reason for choosing the incumbent in the No Experience and Limited Experience


conditions.


However, in the Extensive Experience condition, people attributed their choices to the


ingredients of the incumbent brand.


About 45% thought that the incumbent brand had more of


the ingredients of the attack brand (Aloe-Vera, Cocoa-Butter,etc.).


About 30% said that they


chose the incumbent brand because they liked the ingredients of the incumbent brand that were


different from those of the attack brand. Thus as t

offered coherent reasons that justified their choices.


ley gained experience with the brand, subjects

This pattern was observed irrespective of the


availability of attribute information about the incumbent brand. The attributions, however, differed


depending on availability.


When attribute information about the incumbent brand was available,


the difference in ingredients was salient and subjects attributed their choices to the incumbent


brand having a different set of attributes.


When attribute information was not available externally,


subjects wrongly attributed their choices to the incumbent brand having higher levels of the same

attributes that the attack brand had.










TABLE V-20


REASONS FOR CHOOSING BOTANICAL SPRING (INCUMBENT)


Experimental

Condition


No Experience-Low

Availability


Limited Experience-

Low Availability


Extensive


Experience


- Low


Availability


No experience-High

Availability


Limited Experience-

High Availability


Extensive

Experience-High

Availability


Risk Aversion/

Liking


6/9 (0.67)


8/12 (0.67)


4/16 (0.25)


3/6 (0.50)


(0.55)


7/17 (0.40)


Higher Value of

(same) Attributes


1/9 (0.11)


2/12 (0.165)


7/16 (0.44)


Different Attributes


2/9 (0.22)


2/12 (0.165)


5/16 (31)


3/6 (0.5)


5/11 (0.45)


10/17 (0.60)










TABLE V-21


REASONS FOR CHOOSING NATURAL BLOSSOM (ATTACKER)


Experimental


Condition


No Experience-Low


Availability


Limited Experience-


Low Availability


Extensive


Experience


- Low


Availability


No experience-High


Availability


Limited Experience-


High Availability


Extensive


Experience-High


Availability


Variety Seeking/


Dislike Incumbent


5/13 (0.386)


4/10 (0.40)


3/6 (0.50)


9/16 (0.56)


6/11 (0.55)


3/5 (0.60)


Higher Value of


(same) Attributes


6/13 (0.46)


4/10 (0.40)


3/6 (0.50)


* one person in this condition did not give any reasons


Different Attributes


2/13 (0.154)


2/10 (0.20)


0/6 (0)


6/16 (0.375)*


5/11 (0.45)


(0.40)










A similar analysis was conducted for those who chose the attack brand.

table classifies the reasons in different experimental conditions (see Table V-ll).


The following

This analysis


does not give any diagnostic information except that when attributes were available, more subjects


attributed their choices to differences in attributes.


When this information was not available, more


subjects attributed higher values on the same dimensions as the reason for their choice.


Although


data seem


to indicate


a difference


in the


proportion


of variety


seekers between


the Low


Availability and High Availability conditions, theoretically there is no reason for this difference.

Discussion


In this experiment, ambiguity was manipulated by describing the competing brands in


terms of a set of non-overlapping attributes.

also due to memory constraints. The main


Further, for one half of the subjects, ambiguity was


finding is that extensive experience with a brand led


to loyal behavior, irrespective of availability of specific attribute information about the incumbent


when the attack brand was introduced.


Experience also had an effect on confidence in choice


(measured before the introduction of the attack brand) and on confidence in comparative judgment


(measured after subjects chose between the incumbent and the attack brand).


Although the


Limited Experience and the Extensive Experience conditions did not differ with respect to choice

confidence, they differed with respect to confidence in comparative judgment. Although the effect

of availability of specific attribute information on choice was negligible, this factor influenced


confidence in comparative judgment.


Thus, there was clear support for overconfidence mediating


loyalty in the High Availability condition, but there is no evidence for this process in the Low


Availability condition.


After partitioning subjects in the High Availability condition into those


who chose the incumbent and those who chose the attack brand,


it was found that in the










chose the attack brand.


This pattern was not obtained in the Low Availability condition. This may


explain the difference in the mediational role of overconfidence in the High Availability condition.


An analysis of reasons for choice revealed two types of attributions based on availability.


In the


High Availability condition, the unique attributes of the incumbent were emphasized more, while


subjects in the Low


Availability condition subjects made normatively inappropriate inferences


about the ingredients of the incumbent based on what was available about the attack brand.

As in Experiment 1, the verbal protocols indicated that, along with overconfidence, other


processes such as risk aversion or the status quo bias may drive loyal behavior.

suggest two types of "confirmation bias" processes can lead to overconfidence.


These data also

When attribute


information was not available, the overall liking of the incumbent brand became salient, especially


after extensive experience.


These subjects had the tendency to stay with the incumbent brand


without venturing to test whether the attack brand also offered positive consumption experience.


In the High Availability conditions, this process was slightly different.


Subjects in the Extensive


Experience- High Availability condition had the tendency to overweigh the attribute combination


of the incumbent brand with reference to certain benefits these attributes can offer.


Subjects had


earlier learned from the "story" on natural lotions that several attributes can lead to the same


benefits.


However,


avoided


choosing


the attack


brand


that had a different attribute


combination but can lead to the same benefits.

Carpenter and Nakamoto (1989). Thus, percei


These results are consistent with the findings of


ived attribute importance that is hard to influence


(Lutz 1975) can be learned through normatively inappropriate means.

Thus, the key finding of the first three experiments is that certain types of knowledge

about the incumbent brand (acquired through extensive experience or thinking in terms of benefits)










overconfidence


in judgment,


received


partial


support.


However,


in all these


experiments,


Botanical Spring (the incumbent) was superior to other brands when subjects initially made their


choice.


This induced high confidence especially under extensive experience and/or thinking in


terms of benefits.


In the next experiment, I vary the superiority of the incumbent brand and see


if these effects can be sustained.














CHAPTER VI
EXPERIMENT 4: CHOICE CONTEXT AND LOYALTY

It has been argued thus far that perceptions of superiority of a brand are strengthened by


acquiring certain types of knowledge about this brand.


Experience and thinking in terms of


benefits can develop these knowledge structures and lead to loyal behavior.


The theme of this


chapter is that decision context also can determine perceptions regarding superiority of a brand,


especially when a person does not have an established brand preference.


The context interacts


with the effect of experience and influences beliefs regarding the brand benefits, confidence in


choice, and loyalty when a new, superior brand enters the market.


Specifically, if a brand is


framed as the best brand, loyal behavior will persist with extensive experience and high ambiguity


in the context in which the new brand is encountered.


On the other hand, if the same brand is


framed as the second best brand, next to a brand that is not currently available in the market, even

extensive experience and high decision ambiguity may not result in loyal behavior at the entry of


an objectively superior brand.


This proposition was tested in experiment 4.


Design and Procedure


This was a single factor study, with context being the factor manipulated at two levels.


Twenty eight female undergraduates participated in this experiment.


Experience and the level of


benefit beliefs formation were high across experimental conditions. Ambiguity created by memory


constraints was also high across the conditions.


Thus, except for the manipulation of context, this


experiment


replicated


the Extensive


Experience


- High


Benefit


Beliefs


Formation


cell of










were asked to examine the attribute values given in the matrix.


As they were about to announce


their choices, they were told as an afterthought that one of the brands given in the matrix, Herbal

Abundance, was not available in the market anymore and were asked to choose one of the other


three brands.


The key manipulation pertained to the attribute values of Herbal Abundance.


brand


will be called


as the


"context"


brand.


In one condition


the context brand


(Herbal


Abundance) was superior to the incumbent brand (Botanical Spring) in terms of attribute values;


in the other condition it was inferior.


The attribute values of the incumbent brand and the other


two brands in the choice set were constant across condition. Thus, in terms of superiority, the

order in the first condition was Herbal Abundance, Botanical Spring, and Garden Exotica or


Nature's Nurture depending on preference for apricot extracts over chamomile flower.


The order


in the second condition was Botanical Spring, Herbal Abundance, and Garden Exotica or Nature's


Nurture.


Appendix H and Appendix I depict these two conditions, hereafter called the "Second


Best" and the "Best" conditions, based on the position of the incumbent.


As mentioned earlier,


the values of Botanical Spring and the other two brands do not change across these conditions.

The changes pertain to the values of the context brand only.


In both conditions, Botanical Spring was chosen by all subjects. After choice, they were

asked to try the brand six times and then rate their beliefs on a given scale. The belief scale had


eight items, each ranging from 1 to 9.


each subject.


The mean of eight ratings constituted the belief score for


Next, subjects were asked to indicate their confidence in choice on a nine point


scale.


As in the previous experiments, they were asked to return after two days.


The tasks during


the second session were the same as in Experiment 1.


Decision ambiguity was created by not


presenting the specific attribute values of the incumbent brand when the attack brand (Natural








60

confidence in this judgment.

Results

Table VI-1 presents the proportion of subjects choosing the incumbent brand in the two

experimental conditions.


TABLE VI-1


THE EFFECT OF CONTEXT ON LOYAL BEHAVIOR


Context 1 (Incumbent Best)


Context


(Incumbent Second Best)


9/14 (0.64)


(3/14) (0.21)


TABLE VI-2


ANOVA FOR THE EFFECT OF CONTEXT


Source


Chi-Square (Wald Statistic)


Intercept

Context


As expected, more subjects chose the incumbent brand over the attack brand when the


incumbent brand was framed as the best.

on choice (e.g., Huber, Payne and Puto


Although researchers have reported contextual effects

1982), there is no research that has examined whether


these effects persist over time and carry over to a different situation.

preferences for brands learned in one context persists over time. The


Here, we find evidence that


Effects of context on beliefs


and choice confidence support these results.

Context had a significant effect on belief score (5.14 in the incumbent best condition


versus 4.78 in the incumbent second best condition; F


.26=5.96, p <0.03; r sqr-0.18).


Although


subjects in both conditions chose the same brand and had the same number of trials,


the beliefs










evidence for a biased appraisal of experience.


Context affected choice confidence also.


Compared to an average of 5.29 in the "second


best" condition, the average confidence score in the "best" condition was 6.07 (F


1, 26 =5.68, p


However, there was no difference among groups in terms of confidence in comparative


judgment although the means were in the predicted direction. The mean confidence in the "best"


and "second best" groups were 6.71 and 6.36 respectively (F


1 26=0.45, p<0.6).


Finally, in this experiment one third of the subjects (4/12) who chose the incumbent

brand thought that the attack brand was superior, exhibiting a tendency toward risk aversion.


Of those who chose the attack brand, 2 in the


"best"


condition and 2


in the


"second best"


condition (totally 4 out of 16 or about 25%) gave variety seeking as the reason for their choice.


Thus, as suggested by Kahneman and Miller (1986) and Simonson and Tversky


influenced the learning of expectations.


1992, context


Once learned, these expectations and preferences persist.














CHAPTER VII
OPPORTUNITY FOR TRIAL AND CHOICE BETWEEN THE INCUMBENT AND
THE ATTACK BRAND


In the previous experiment, the theme was how to build consumer loyalty.


In this


experiment we examine how a superior competitor can weaken consumer loyalty to an incumbent


brand.


Hoch and Deighton (1989) proposed a strategy for "underdog" brands especially when


ambiguity in the choice context is high.

can "disambiguate" the decisions. Thi,


attack brand.


According to them, an opportunity to try the new brand


s should increase the proportion of subjects choosing the


Experiment 5 was designed to test this assertion.


Method

The only factor in this study was opportunity to try the attack brand before choosing


between the incumbent and the attack brand.


of the Extensive Experience


The "No Trial" condition was an exact replication


- High Availability of Specific Attributes condition of experiment 3.


After excluding some for the reasons given below, fifty-two female students completed this


experiment.


Across conditions, subjects had extensive experience with the incumbent. Ambiguity


was also high across conditions due to competing brands having non-overlapping set of attributes.


As in other experiments, the initial tasks involved reading a "story" on natural lotions,


going


through the brand x attribute matrix, and choosing a brand based on information in the matrix.

Data from three subjects who did not choose Botanical Spring were excluded in the analysis.

In the protests and in the first four experiments, subjects evaluated the same set of four


real brands of lotions. In the protests as well as in experiments


1, 2 and 4, subjects had an














mentioned earlier, irrespective of what brand they chose, each subject was made to believe that

the brand chosen by her was Botanical Spring, the objectively superior brand in the initial choice


set of four.


It was observed in these experiments that two brands came out as the most preferred


brands.


Most subjects chose one of these two brands.


These two brands varied in terms of


fragrance.


It was decided to counterbalance these brands as the incumbent and attack brands.


After choice, subjects


were randomly assigned to two groups.


Subjects in


the first group


experienced one of these brands (brand A) six times while those in the second group tried the


other brand (brand B) six times.


Since subjects did not try all the brands before choice, it is


possible that some of them may not have liked the brand they tried and may have switched as


soon as an opportunity was given.


To reduce this possibility, each subject was asked to evaluate


the brand she tried on a dislike (1) to like (9) scale.


Data of five subjects whose evaluations were


less than 5 on this scale were excluded. Finally data of four more subjects were excluded because

they were mistakenly presented the wrong brand as the attack brand.


Subjects returned after two days.

the attack brand was introduced. The att


attributes.


The key factor in this experiment was manipulated when


iack brand in either condition had a set of nonoverlapping


One half of the subjects were asked to choose based only on the attribute information


given in the matrix.


The other half, apart from receiving information about the attributes of the


competing brands, also had an opportunity to try the attack brand once.


For each person in this


condition, the attack brand was the one that she did not try earlier out of the two most preferred


brands.


After choice, superiority judgments and confidence in these judgments were measured.










Results

TABLE VII-1


OPPORTUNITY TO TRY THE ATTACK BRAND AND LOYAL BEHAVIOR


Not tried the attack brand

17/26 (0.65)


Tried the attack brand

10/26 (0.385)


The above table shows that while 65 percent of those in the No


Trial of Attack Brand


condition chose the incumbent, only about 39 percent of those who had an opportunity to try the


attack brand chose the incumbent brand (Wald Statistics


= 3.68, p


< 0.06).


Although these two


proportions are different, the proportion in the Tried the Attack Brand condition is not different


from the chance proportion of 0.5.


There was no difference in choice proportions between


subjects who tried brand A as the attack brand and those who tried brand B as the attack brand

(6/13 versus 4/13, chi-square= 0.64, p <0.45).

The Process


Why should those who have an opportunity to try the new brand exhibit a less loyal


behavior?


One possibility is that it truly disambiguatess" choice. It clearly shows that the new


brand is superior to the old brand.


Second, due to recency effects, experience with the new brand


may be more salient than the one with the old brand.


Third, if overconfidence is due to


confirmation bias (i.e., someone attributes certain benefits to brand X or ingredient A


without


realizing that brand Y or ingredient B also may yield those benefits), experiencing the new brand


should reduce overconfidence.


A related explanation is that a limited experience with the new


brand may just induce interest in learning more about the new brand and hence confidence in


superiority of the new brand may not be high.


There may be multiple processes operating








65

An analysis of the superiority and confidence judgments may offer evidence for one or


more of these processes.


However, these tests are just exploratory.


In this experiment, the


superiority judgments were entirely consistent with the choices.


Thus, only the confidence data


will be discussed. If the process of disambiguation operates, then subjects in the


Trying the New


Brand condition should judge the attack brand to be superior and also should exhibit more


confidence in this judgment. Hence, there should be a crossover interaction between the effects

of trial of the attack brand and choice. In the no trial condition, those who judge the incumbent


brand superior should exhibit more confidence than those who judge the attack brand superior.

In the trial condition, those who judge the attack brand to be superior should exhibit more


confidence than those who consider the incumbent to be superior.


The process of salience of


experience with the new brand, although different conceptually, at least using our design, cannot


be disentangled from the disambiguation process.


The confirmation bias explanation requires trial


of new brand to have a main effect on confidence. In the trial condition confidence should be less


than in the no trial condition.


Table


VII-2 presents confidence in various levels of trial and


choice, followed by a table testing for various effects.

TABLE VII-2


TRIAL OF THE ATTACK BRAND, CHOICE AND CONFIDENCE IN COMPARATIVE
JUDGMENT


Choice Not Tried the Attack Brand Tried The Attack Brand


Natural Blossom (Attack 5.89 (n=9) 6.06 (n=16)

Brand)










TABLE VII-3


ANALYSIS OF VARIANCE TEST OF EFFECTS


Source


Trial


Choice


Trial*Choice


Choice


6.73

0.002


was the only factor that had any effect on confidence (6.7 for those who chose


the incumbent brand and 6.0 for those who chose the attack brand).


If the disambiguating process


or the salience of recent experience operated in the condition in which subjects had an opportunity

to experience the new brand, confidence of those who chose the incumbent brand should be lower


than that of those who chose the attack brand.


processes.


The results do not offer any evidence for these


The test of other processes was conducted by partitioning the set into those who had


an opportunity to try the new brand and those who did not have this opportunity.


did not try the new brand, choice had a moderate effect on confidence


For those who


(5.88 for the attack brand


versus 6.64 for the incumbent; F


1,24=3.57,


p <0.08; r sqr=0.13).


This finding is consistent with


those of the previous experiments.


Surprisingly, the effect was in the same magnitude and


direction when subjects tried the new brand (6.06 versus 6.8; F


S24=3.27, p<0.09; r sqr=0.12).


Combined with the difference in confidence between those who chose the incumbent brand in the

no trial condition and those who chose the attack brand in the trial condition (6.64 vs 6.06), this


suggests that multiple processes may be operating.


For those who chose the incumbent brand,


additional experience with the product category might have strengthened their perceived validity


in judgment.


For those who chose the incumbent brand, it could have been either a desire to learn











CHAPTER VIII
GENERAL DISCUSSION AND CONCLUSION


The central


theme of this dissertation is that certain types of knowledge about the


incumbent brand, combined with ambiguity in the decision environment, lead to overconfidence


in comparative judgment and thus to loyal behavior. The results of the first three experiments,

combined with those of a pretest, support these predictions. The results also suggested that in a


limited number of cases, loyalty was driven by loss aversion or sheer inertia.


In the


attributes.


first two experiments, ambiguity was due to memory constraints for specific


In this situation, either extensive experience with a brand or limited experience and a


focus on the benefits of the brand was sufficient to create loyal behavior.


When decision


environment


facilitated


side-by-side


comparison


of brands


on attributes,


despite


extensive


experience and formation of benefit beliefs, subjects preferred the attack brand which was superior


on attribute dimensions.


Thus, certain types of knowledge about the incumbent brand caused


overconfidence


and shielded the incumbent under ambiguous decision environment.


When


decision ambiguity was reduced, the overconfidence bias was rectified.

In the third experiment, ambiguity was created by describing the competing brands in

terms of a set of non-overlapping attributes that lead to the same benefits. Here, again, experience


with the incumbent brand influenced the degree of resistance to competitor's attack.


However,


in this situation providing


attribute information about the competing


brands did not reduce


ambiguity.


Instead, availability of specific attribute information enhanced overconfidence for


those who have had extensive experience.


Overconfidence caused by experience did not map on


to choice when attribute information about the incumbent brand was not available.











68

or order of entry effects (Carpenter and Nakamoto 1989; Kardes and Gurumurthy 1992).

In the first two experiments a variety of mechanisms including the set size effect (Kardes and


Gurumurthy) and the


focus of comparison (Dhar and Simonson


1992) can account for the


persistence of preference.


these processes may lead


to overconfidence


in preference


judgments in ambiguous decision contexts.


In the Kardes and Gurumurthy studies, as in the first


two experiments of this paper, ambiguity existed because of sequential rather than simultaneous


presentation of information.


This form of ambiguity existed also in the Dhar and Simonson


studies.


Carpenter and Nakamoto explicitly proposed ambiguity as a necessary condition for the


order of


entry


effects.


In their studies,


the value of attributes


as well


the ideal


attribute


combination were ambiguous.


The results of experiment 3 appear to be consistent with Carpenter


and Nakamoto's ideal point shift explanation of preference formation especially under ambiguous


decision conditions.


As in the Carpenter and Nakamoto studies, in the current experiments


preference updating involved heuristic


judgments and attributions


regarding the ingredients-


benefits mapping.


However, as shown in Experiment 4, preferences prior to trial are not so weak


as suggested by Carpenter and Nakamoto.


In this experiment, preference influenced by the


context and formed prior to trial had an overriding impact on beliefs about the benefits, and


subsequent choices.


framing.


The findings of Experiment 4 are consistent with consumer research on


For example, Hoch and Ha (1989) found that in an ambiguous context, advertisement


provided expectations and led to biased appraisal of product trial.


In Experiment 4, framing the


t I C I I I 1 I 1 I 1 1 )1 II I I 1 *










diagnosticity of information is our focus.


The knowledge structures induced by experience or


belief crystallization can overcome the effect of certain other forms of information and increase


confidence in comparative judgment as long as the decision context is ambiguous.


As the


Feldman and Lynch (1989) framework would predict, these diagnostic inputs have an important


influence on subsequent decisions (Lynch, Marmorstein and Weigold 1989).


The current studies


demonstrated that the perceived diagnosticity


itself can be


influenced even


by normatively


inappropriate factors.

The findings of the current experiments offer compelling evidence that at least for a subset

of consumers whose choices are consistent with the superiority judgments, overconfidence is the

primary mechanism driving loyalty, the other mechanisms being loss aversion and sheer inertia.


The overconfidence


mechanism


proposed


in this


paper


is not


inconsistent


commitment


view.


The origin


of overconfidence


can be motivational


as the emotional


commitment view would probably advocate or cognitive and contextual as demonstrated in the


current experiments.


However, the propositions pertaining to the role of decision ambiguity in


preference


reversal


not be


explained


by the emotional


commitment conceptualization.


Loyalty conceptualized in terms of emotional commitment may be irreversible. Such unassailable

loyalty may be rare in marketplace choices.

Conceptualizing loyalty in terms of confidence in comparative judgment helps identifying


a set of antecedents that can be controlled by managerial actions.


Specifically, experience and


benefit belief crystallization shielded the incumbent from a superior competitor when decision


ambiguity was high.


Both these variables can be controlled by managerial actions.


Our belief


crystallization manipulation corresponds closely to the "reminder" advertisements.


The process










consumers


saw ads before


product trial


and formed


expectations


were


confirmed


experience.


It is also possible that advertisements after trial (similar to our belief crystallization


manipulation) may help interpret one's experience with a brand and clarify preferences.


This lead


to the perceived


validity


of prior


choice


especially


if the


decision


context


is ambiguous.


Experiment 1 showed that this should be especially potent for consumers with limited experience.


Experience by itself also engendered loyalty.


Marketers can encourage the accumulation of


experience by price promotions that induce stockpiling of the brand.

Our conceptualization also facilitated isolating a set of conditions that can revert loyal


behavior.


The key variable of this paper, decision ambiguity is a necessary condition for the


effects of experience and crystallization of benefit beliefs on loyalty.


The results of experiment


1 and experiment 3 together show that in general it is better for an incumbent to avoid direct


attribute comparison.


However, from an attack brand point of view, it is possible to overcome


an entrenched competitor by encouraging side-by-side comparisons.


The results of experiment


3 are chastening, as they imply that unless comparison is possible on common dimensions,


experience may blind consumers to attackers'


superiority.


In these contexts, availability of


attribute information may possibly increase overconfidence and thereby loyalty.

Apart from decision ambiguity that was examined in all the experiments, context (in

experiment 4) and an opportunity to try the attack brand (in experiment 5) accounted for the


reversal of loyalty.


These variables can be easily manipulated by marketers of the attack brand


to overcome an entrenched incumbent. For example, marketers have control over the type of

brand information available at the point of purchase. They also have control over product design


and positioning.


Through these means they can manipulate ambiguity in the choice context and
















APPENDIX A
"STORY" ON NATURAL LOTIONS
BEAUTY AND FLOWER POWER


Botanically based beauty products are b(l)ooming!


to use.


These products are a sensual pleasure


Cosmetic companies through out the world are reformulating their skin care products and


are launching whole plant-based lines.


According to Susan Schiffman, Ph.D., director of the


Cosmetics Research Institute at Stanford University, the use of all natural skin products can


significantly lower your stress levels through the sensory pleasure they provide.


Kinder to the


environment than chemically-based beauty products, plant-based skin care products are for many

the ultimate "back to basics".


All natural hand and body lotions offer a lot of benefits to a young woman.


A young


woman can use all natural lotions to smooth out rough spots, soothe her legs after shave and to


tone and refresh her skin. All natural lotions help rejuvenate dry and damaged skin and also can

prevent premature aging of the skin. Comprised of plant extracts such as flowers and herbs, these


lotions help restore suppleness to the skin and also help to add vitality.


Moreover, they enrich


the skin's complexion by replenishing lost nutrients and providing a protective barrier from

external elements like air pollution, wind or sun.


How do you pick from the garden of botanical choices?


from those available and experiment.


Many people pick an ingredient


you can learn the effects the different ingredients


produce and choose the product that will give you the maximum benefits.


We have compiled a








72

The most important ingredients in natural hand and body lotions and their benefits are


given below in an alphabetical order (not according to their rank in terms of importance).


Please


study these descriptions to learn the benefits these ingredients will offer so that you can make an

efficient choice.


ALOE VERA: Aloe Vera is a plant that promotes healthy skin texture.


of anti-bacterial, astringent and coagulating agent.


It consists of a mixture


It is also a scar inhibitor and a skin tissue


growth stimulator.


It can soothe skin suffering from sunburns.


It also can remove blackheads


appearing on the skin.

CHAMOMILE FLOWERS: Chamomile is considered to be a soothing, calming herb. When used


on the skin, it restores lost moisture and provides nourishment.


Further, the exotic fragrance of


chamomile makes the experience of using the product particularly pleasurable.

COCOA-BUTTER: Cocoa-Butter is a necessary ingredient for women who desire smooth, soft


hands and legs.


Cocoa-butter contains extra rich emollients that naturally silken, soften and soothe


all skin types.


It protects the tenderness of the skin.


It can also minimize blackheads and can


disinfect skin.


JOJOBA


EXTRACTS:


Jojoba


Extracts


are exceptionally


effective


in general


firming


conditioning of the skin. Like cocoa butter, jojoba extracts contain emollients that silken, soften


soothe all


types. Jojoba


is highly recommended


for treating


wrinkles,


withering skin,


prematurely-aged skin and fine textured or thin skin.


It is very effective in protecting the


tenderness of skin.

TROPICAL FRUIT EXTRACTS: The tropical fruits used most frequently in skin lotions are

Apricot and Papaya.








73

Papaya: Papaya extracts contain important restorative enzymes and it is a natural softening agent

that adds smoothness to the skin.


VITAMIN E:


Vitamin E maintains the tenderness of the skin thus helping to keep the skin


younger-looking.


Although there are many sources of Vitamin E,


plants-based vitamin E is


particularly effective in promoting healthy new skin.


lasting softness.


It also quickly penetrates the skin for long-


It is an exclusive moisture-binding.


Apart from these important ingredients, certain other ingredients are used for specific


purposes.


Cucumber water and Mineral water are often used as cleansing agents.


However,


almost all the attributes we described


earlier can do


the jobs of


cleansing and minimizing


infection.


Hence, cucumber water and mineral water are clearly unimportant attributes in all


natural lotions.

Now, we're sure that you will appreciate the benefits of using all natural hand and body


lotions.


Consider each of these attributes very carefully before you choose a brand.


Remember


that a single brand of all natural hand and body lotion may not contain all the attributes.


learning the benefits of each of these ingredients becomes very important.


Hence,


We wish you all the


very best in your skin care.











APPENDIX B
INITIAL CHOICE SET IN EXPERIMENT 1


PRODUCT


ALL NATURAL HAND AND BODY LOTION


ATTRIBUTES BRANDS

GARDEN HERBAL BOTANICAL NATURE'S
EXOTICA ABUNDANCE SPRING NURTURE

ALOE VERA 18 MG 16 MG 19 MG 18 MG

COCOA BUTTER 10 MG 9 MG 12 MG 10 MG

APRICOT EXTRACT 6 MG 6 MG 7 MG 7 MG

CHAMOMILE FLOWER 5 MG 5 MG 6 MG 4.5 MG

CUCUMBER WATER 3.5 FL.OZ 3.0 FL.OZ 3.0 FL.OZ 3.0 FL.OZ

SIZE 12 FL.OZ 12 FL.OZ 12 FL.OZ 12 FL.OZ

PRICE $4.90 $4.90 $4.90 $4.90










APPENDIX C


BELIEF SCALE IN EXPERIMENT


BOTANICAL SPRING


(Stongly believe this
brand does not offer)


(Strongly believe this
brand does offer)


PROMOTES HEALTHY SKIN

TEXTURE


RESTORES LOST MOISTURE


SOFTENS AND PROTECTS

THE TENDERNESSS OF SKIN


FIRMS AND CONDITIONS SKIN


CLEANSES SKIN


DISINFECTS SKIN


CAN REMOVE BLACKHEADS


CAN PREVENT WRINKLES

THE LONG RUN


--- --- --- --- --- --- --- --- 9



--- --- --- --- --- --- --- --- 9






--- --- --- --- --- --- --- --- 9




--- --- --- -- --- --- --- ---9




---9
--- --- --- --- --- --- --- --- 9




--- --- --- --- --- --- --- --- 9






--- --- --- --- --- --- --- --- 9






- -- -- -- -- -- -- -- -- 9


Did You Choose This Brand ?
(BOTANICAL SPRING)


YES ---


NO ---- (Check One)


/rnTT 1 T Trnr" rfn rl A r'n r'T'TO 1n" A MT1 flkT IT V T VfT/T T -]f'i '' T'LT.TC DD A MTTh


T1 V lT T T|TT











APPENDIX D
CHOICE SET AFTER THE "ATTACK" BRAND'S ENTRY
(EXPERIMENT 1)


Attributes Brands

Garden Herbal Nature's Botanical Botanical Natural
Exotica Abundance Nurture Spring Glory Blossom*

Aloe Vera 18 Mg 19 Mg


Cocoa 11 Mg 14 Mg
Butter


Apricot 7 Mg 8 Mg
Extract

Chamomile 6 Mg 7 Mg
Flower

Cucumber 3.0 Fl.Oz. 3.0 Fl.Oz.
Water

Size 12 Fl.Oz. 12 Fl.Oz. 12 Fl.Oz.

Price $4.90 $4.90 $4.90


* = ATTACK BRAND











APPENDIX E


INITIAL CHOICE SET IN EXPERIMENT


PRODUCT


ALL NATURAL HAND AND BODY LOTION


ATTRIBUTES BRANDS
GARDEN HERBAL BOTANICAL NATURE'S
EXOTICA ABUNDANCE SPRING NURTURE

VITAMIN E 18 MG 16 MG 19 MG 18 MG

JOJOBA EXTRACTS 10 MG 9 MG 14 MG 10 MG

PAPAYA EXTRACT 6 MG 6 MG 8 MG 7 MG

CHAMOMILE FLOWER 5 MG 5 MG 7 MG 4.5 MG

CUCUMBER WATER 3.5 FL.OZ 3.0 FL.OZ 3.0 FL.OZ 3.0 FL.OZ

SIZE 12 FL.OZ 12 FL.OZ 12 FL.OZ 12 FL.OZ

PRICE $4.90 $4.90 $4.90 $4.90











APPENDIX F
CHOICE SETI AFTER THE ENTRY OF ATTACK BRAND
(EXPERIMENT 3)
PRODUCT = ALL NATURAL HAND AND BODY LOTION


Attributes Brands

Garden Herbal Nature's Botanical Botanical Natural
Exotica Abundance Nurture Spring Glory Blossom*

Vitamin E

Jojoba
Extracts

Papaya
Extracts

Aloe Vera 18 Mg 19 Mg


Cocoa 11 Mg 14 Mg
Butter


Apricot 7 Mg 8 Mg
Extract

Chamomile 6 Mg 7 Mg
Flower

Cucumber 3.0 Fl.Oz. 3.0 Fl.Oz.
Water

Size 12 Fl.Oz. 12 Fl.Oz. 12 Fl.Oz.

Price $4.90 $4.90 $4.90

* = ATTACK BRAND











APPENDIX G
CHOICE SET2 AFTER THE ENTRY OF ATTACK BRAND
(EXPERIMENT 3)


PRODUCT


ALL NATURAL HAND AND BODY LOTION


Attributes


Brands


Garden Herbal Nature's Botanical Botanical Natural
Exotica Abundance Nurture Spring Glory Blossom*

Vitamin E 19 Mg

Jojoba 14 Mg
Extracts

Papaya 8 Mg
Extracts

Aloe Vera 18 Mg 19 Mg


Cocoa 11 Mg 14 Mg
Butter


Apricot 7 Mg 8 Mg
Extract

Chamomile 7 Mg 6 Mg 7 Mg
Flower

Cucumber 3.0 Fl.Oz 3.0 Fl.Oz. 3.0 Fl.Oz.
Water

Size 12 Fl.Oz. 12 Fl.Oz. 12 Fl.Oz.

Price $4.90 $4.90 $4.90

= ATTACK BRAND











APPENDIX H
INITIAL CHOICE SET IN THE "BEST" CONDITION
(EXPERIMENT 4)


ATTRIBUTES BRANDS
GARDEN HERBAL BOTANICAL NATURE'S
EXOTICA ABUNDANCE* SPRING NURTURE

ALOE VERA 18 MG 19 MG 19 MG 18 MG

COCOA BUTTER 10 MG 11 MG 12 MG 10 MG

APRICOT EXTRACT 6 MG 7 MG 7 MG 7 MG

CHAMOMILE FLOWER 5 MG 6 MG 6 MG 4.5 MG

CUCUMBER WATER 3.5 FL.OZ 3.0 FL.OZ 3.0 FL.OZ 3.0 FL.OZ

SIZE 12 FL.OZ 12 FL.OZ 12 FL.OZ 12 FL.OZ

PRICE $4.90 $4.90 $4.90 $4.90
* = BRAND NOT AVAILABLE IN THE MARKET











APPENDIX


INITIAL CHOICE SET IN THE "SECOND BEST" CONDITION
(EXPERIMENT 4)


ATTRIBUTES


BRANDS


GARDEN HERBAL BOTANICAL NATURE'S
EXOTICA ABUNDANCE* SPRING NURTURE

ALOE VERA 18 MG 20 MG 19 MG 18 MG

COCOA BUTTER 10 MG 13 MG 12 MG 10 MG

APRICOT EXTRACT 6 MG 7 MG 7 MG 7 MG

CHAMOMILE FLOWER 5 MG 6 MG 6 MG 4.5 MG

CUCUMBER WATER 3.5 FL.OZ 3.0 FL.OZ 3.0 FL.OZ 3.0 FL.OZ
SIZE 12 FL.OZ 12 FL.OZ 12 FL.OZ 12 FL.OZ

PRICE $4.90 $4.90 $4.90 $4.90
S= BRAND NOT AVAILABLE IN THE MARKET












REFERENCES


Alba, Joseph W. & J.


Wesley Hutchinson (1987),


"Dimensions of Consumer Expertise," Journal


of Consumer Research, 13,


411-54.


Alba, Joseph W., J. Wesley Hutchinson &
Making," in H. H. Kassarjian and T.


John G. Lynch Jr. (1990),


"Memory and Decision


S. Robertson (eds.), Handbook of Consumer Theory


and Research, Englewood Cliffs: Prentice-Hall.


Alba, Joseph


Howard


Marmorstein


& Amitava


Chattophadhyay


(1992),


"Transition of


Preference over Time," Journal of Marketing Research (forthcoming).


Allison,


(1964),


"Influence


Beer


Brand


Identification


on Taste


Perception," Journal of Marketing Research, 1, 36-39.


Baron, Reuben M. & David A. Kenny (1986),


"The Moderator-Mediator Variable Distinction in


Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,"
Journal of Personality and Social Psychology, 51, 1173-82.


Bettman, James R., Eric J.


Johnson & John


W. Payne (1991)


"Consumer


Decision


Making," in


Robertson


& H.H.


Kassarjian


(eds),


Handbook


of Consumer


Behavior, Englewood Cliffs, NJ: Prentice-Hall.


Biehal, Gabriel & Dipanker Chakravarti, (1983),


Consumer Choice,"


"Information Accessibility as a Moderator of


Journal of Consumer Research, 10, 1-14.


Brehmer,


Berndt


(1980),


"In One


Word:


Not From


Experience,"


223-41.


Brucks,


Merrie


(1985),


" The Effects of Product


Class


Knowledge on


Information


Search


Behavior," Journal of Consumer Research, 12, 1-16.


Camerer,


Colin


Martin


Weber


(1992),


"Recent


Developments


Modeling


Preferences: Uncertainty and Ambiguity,"


Joumal of Risk and Uncertainty,


325-70.


Carpenter, Gregory & Kent Nakamoto (1989),


Advantage,"


"Consumer Preference Formation and Pioneering


Journal of Marketing Research, 26, 285-98.


Chattophadhyay, Amitava & Joseph W. Alba (1988:
Inference in Consumer Decision Making,"


), "The Situational Importance of Recall and
Journal of Consumer Research, 15, 1-12.


Psycholoeica,










Ehrenberg, Andrew


S. C. & P.Charlton (1973),


"An Analysis of Simulated Brand Choice," Journal


of Advertising Research, 13, 21-33.


Einhom,


Hillel


J. (1980),


"Leaming


From


Experience


and Suboptimal


Rules


in Decision


Making" in T.S.Wallsten (ed) Cognitive Processes in Choice and Decision Behavior,


Hillsdale


NJ: Lawrence Erlbaum.


Einhom, Hillel J. & Robin Hogarth (1978),


" Confidence in Judgment: Persistence of The Illusion


of Validity,"


Psychological Review


83, 37-64.


---------------------------------- (1


985),


" Ambiguity and


Uncertainty in Probabilistic Inference,"


Psychological Review, 92,


433-61.


Ellsberg, Daniel (1961),


"Risk, Ambiguity and Savage Axioms," Quarterly Journal of Economics,


75 643-669.


Engel, James F., Roger D. Blackwel


& Paul


W. Miniard (1989),


Consumer Behavior,


edition, Hindsdall, IL: Dryden Press.


Fazio, Russell H., & Mark P. Zanna (1978),


"Attitudinal Qualities Relating to the Strength of the


Attitude-Behavior Relationship," Journal of Experimental Social Psychology,


14, 398-


Feldman, Jack M., and John G. Lynch (1988),


"Self-Generated Validity and Other Effects of


Measurement of Belief, Attitude,


Intention and Behavior,"


Journal of Applied


Psychology, 73, 421-35.


Fischhoff,


Baruch


Perspective,"


Beyth-Marom


(1983


Psychological Review, 90,


I), "Hypothesis
239-60.


Evaluation


a Bayesian


Guadagni, Peter and John. D. C. Little (1983),


"A Logit Model of Brand Choice Calibrated on


Scanner Data,"


Marketing Science,


203-38.


Ha, Young Won and Stephen J. Hoch (1989), "Ambiguity, Processing Strategy, and Advertising
Evidence Interactions," Journal of Consumer Research, 16, 354-360.


Heath, Chip & Amos Tversky (1991),


"Preference and Belief: Ambiguity and Competence in


Choice Under Uncertainty," Journal of Risk and Uncertainty, 4,


Hoch, Stephen J. & John Deighton (1989) "Managing What Consumers Leam From Experience,"
Journal of Marketing, 53, 1-20.


Hoch, Stephen J., and Young


- Won Ha (


1986), "Consumer Learning: Advertising and the


5-28.










Howard,


& Jagdish


Sheth


(1969),


The Theory


of Buyer


Behavior,


York:


Wiley & sons.


Huber,


McCann


(1983),


Impact


of Inferential


Beliefs


Product


Evaluations,"


Journal of Marketing Research, 19, 324-33.


Jacoby, Jacob (1971),


"A Model of Multi-Brand Loyalty,"


Joumal of Advertising Research,


11,25-30.

Jacoby, Jacob & Robert W. Chestnut (1978), Brand Loyalty Measurement and Management, New
York: John Wiley and Sons.


Jacoby, Jacob & David B. Kyner (1973), "Brand Loyalty
Joumal of Marketing Research, 10, 1-9.


Vs. Repeat Purchasing Behavior,"


Jeuland


, Abel P. (1979),


"Brand Choice Inertia as One Aspect of the Notion of Brand Loyalty,"


Management Science,


671-82.


Kahn,


Barbara E. & Rakesh Sarin (1988), "Modeling Ambiguity
Uncertainty," Journal of Consumer Research, 15, 265-72.


in Decisions


Under


Kahneman,


Daniel


Miller


(1986),"Norm


Theory:


Comparing


Reality


Alternatives,"


Psychological Review, 93, 136-53.


Kahneman,


Daniel,


L. Knetsch


and Richard


Thaler


(1990),


"Experimental


Tests


Endowment Effect and Coase Theorem,"


Journal of Political Economy, 98, 1


Kardes, Frank R., and Kalyanaram Gurumurthy (1992),"Order-of-Entry Effects on Consumer


Memory and Judgment: An Information Integration Perspective,"
Research, 29, 343-57.


Journal of Marketing


Klayman,


Joshua and


Strategy,


Young


Structure


- Wan


and Contet


Ha (1989),
it," Journal


"Hypothesis
of Exoerin


1(


Testing in Rule
mental Psychology:


Discovery:
Learning,


Memory and Cognition,


594-604.


Koriat,


Asher,


Sarah


Lichtenstein


& Baruch


Fischhoff


(1980),"Reasons


For Confidence,"


Journal of Experimental Psychology: Human Leaming and Memory, 6, 107-118.


Lichtenstein,


Sarah,


Baruch


Fischhoff


Lawrence


Phillips


(1982),


"Calibration


Probabilities: The State of the Art to
(eds) Judgment under Uncertainty:
University Press, 306-34.


1980," in D.Kahneman, P.Slovic and A.


Heuristics


and Biases,


York:


Tversky


Cambridge










Lynch,


G. Jr.,


Howard


Marmorstein


Michael


Weigold


(1988),


"Choices


Sets Including Remembered Brands: Use
Evaluations," Journal of Consumer Research,


of Recalled


Attributes


and Prior Overall


169-185.


March, J.G. (1978),


" Bounded Rationality, Ambiguity and Engineering of Choice,


Bell Journal of Economics, 9, 587-608.


Meyer, Robert J. (1981),


"A Model of Multiattribute Judgments Under Attribute Uncertainty and


Information Constraint," Journal of Marketing Research, 18,


428-41.


Newman, Joseph W. & Richard A.


Werbel (1973),


"Multivariate Analysis of Brand Loyalty for


Major Household Appliances," Journal of Marketing Research, 10,


404-9.


Oskamp,


Scott


(1965),


"Overconfidence


in Case


Study


Judgments,"


Joumal


of Consulting


Psychology, 29, 261-5.

Park, C.Whan and V. P. Lessig (1981),"Familiarity and Its Impact on Consumer Heuristics and


Biases," Journal of Consumer Research,


Peterson, Dane K. & Gordon F. Pitz,


"Confidence, Uncertainty and the Use of Information,"


Journal of Experimental Psychology: Learming, Memory and Cognition, 14, 85-92.

Reibstein, David. J (1985), Marketing: Concepts, Strategies and Decisions, Englewood Cliffs,
NJ:Prentice-Hall.

Rokeach, Milton (1973) Nature of Human Values, New York, Free Press.


Russo, J.Edward (1977),


"The Value of Unit Price Information,"


Journal of Marketing


Research, 14, 192-201


Russo, J.


Edward,


& Eric.J. Johnson


(1980),


"What Do Consumers


Know


About Familiar


Products?" in J.C. Olson (ed), Advances in Consumer Research,7


417-423.


Samuelson,


William


& Richard Zeckhauser (1989),


"Status Quo Bias in Decision Making,"


Journal Of Risk and Uncertainty, 1, 7-59.


Shacklee, H., & B. Fischhoff(1982),


"Strategies of Information Search in Causal


Analysis,"


Memory and Cognition, 10, 520-530.


Simmons,


Carolyn


& John


Effects of Missing


Lynch


(1991),


Information on


"Inference
Discounting


Effects


Without


Inference


and Use of Presented


Making?


Information,"


Journal of Consumer Research, 18,


477-91.










Slovic,


and Douglas


MacPhillamy


(1974),


"Dimensional


Commensurability


and Cue


Utilization


Comparitive


Judgment,"


Organizational


Behavior


Human


Performance, 11, 172-194.


Smith,


Robert


E. & William


R. Swinyard


(1983),


"Attitude-Behavior


Consistency:


Impact of Product Trials
257-67.


Versus Advertising,"


Journal


of Marketing Research, 20,


Thaler, Richard H., (1985),


"Mental Accounting and Consumer Choice,"


Marketing Science, 4,


199-214.


Tschirgi,


Judith


(1980),


"Sensible


Reasoning:


Hypothesis


about


Hypothesis,"


Child


Development, 51,


Tversky, Amos and Daniel Kahneman (1991),


"Loss Aversion in Riskless Choice: A Reference


Dependent Model," Quartely Journal of Economics, 65, 1039-61.


Tversky, Amos, Shmuel Sattath and Paul Slovic (1988),


Choice,"


"Contingent Weighting in Judgement and


Psychological Review, 95, 371-84.


Wood,


Wendy (1982),


"Retrieval of Attitude-Relevant Information from Memory: Effects of


Susceptibility to Persuasion and in Intrinsic Motivation," Journal of Personality and Social
Psychology, 25, 379-400.











BIOGRAPHICAL SKETCH

A.V. Muthukrishnan attended Carmel Garden Matriculation School, Coimbatore, India.

Next, he pursued an undergraduate degree in liberal arts and a Master of Arts in industrial


relations at Loyola College, Madras, India.


He worked for about two years as a Junior Executive


in a textile research corporation in India before entering Indian Institute of Technology, Bombay,


as a doctoral student.


He received Ph.D in organizational behavior at the Center for Industrial


Management, Indian Institute of Technology.


His areas of concentration included expectancy


theories


of motivation


compensation


administration.


over


years,


he taught


management courses at the M.B.A level.


Before joining the department of marketing at the


University of Florida as a doctoral student, he worked as an Assistant Professor at the Indian


Institute of Management, Bangalore, India.


He entered the marketing doctoral program at the


University of Florida in Fall


1988.


In April


1990, he was sent to the Albert Haring annual


doctoral symposium in marketing as the University of Florida representative.


His current research


interests are in the fields of consumer decision making and branding.








I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation
for the degree of Doctor of Philosophy.


K-h


t -.'


Johil Lynch. Chair


Professor of Marketing


I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation
for the degree of Doctor of Philosophy.


eph Alba. Cochair


Associate Professor of Marketing


I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation
for the degree of Doctor of Philosophy.


/-^ /


Richard Lutz


Professor of Marketing


I certify that 1 have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation
for the degree of Doctor of Philosophy.

7, /
/z7r ,2K


Barton Weitz


Penny


Professor


Marketing


I certify that I have read this study and that in my opinion


confol


standards of scholarly presentation and is fully adequate, in scope and quality
fc..r .kL-^ ,-l1 c.- C- ^ A r^. .-. .-. ~.C flL 'i L -^ i -_


rms to acceptable
, as a dissertation


2
i-9









This dissertation was submitted to the Graduate Faculty of the Department of Marketing
in the College of Business Administration and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.

May 1993
Dean. Graduate School






































UNIVERSITY OF FLORIDA
Al 1111111lll II IIIW I 1 1111l lllN I