AFFECTIVE AND COGNITIVE BRAND CONVICTION: MAKING LOYAL RELATIONSHIPS By JOOYOUNG KIM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003
Copyright 2003 by Jooyoung Kim
To my family
iv ACKNOWLEDGMENTS I am deeply indebted to many people in my completion of this dissertation and all the thanks I express herein are meager substi tutes for the gratitude and appreciation I owe. Heartfelt gratefulness is extended to Dr. Jon Morris, my academic advisor and the chairman of my dissertation committee, for hi s invaluable guidance, mentorship, and the unfathomable emotional support he provided th roughout my research and doctoral study. For his untiring guidance on the research desi gn and data analysis of this research, I also wish to thank Dr. Joffre Swait. In addi tion, this dissertation has benefited immensely from the constructive comments I have received from my other committee members: Dr. Marilyn Roberts, Dr. Dolores Albaracin, and Dr. Chang-Hoan Cho. Without their efforts this would not have been nearly as rewarding an accomplishment. I would also like to take th is opportunity to thank my friends and faculty at the University of Florida, especially Dr. John Sutherland for his mentorship, and Samsup Jo and Jaemin Jung for their sincere support and stimulating academic discussions throughout this Ph.D. experience. The love of my wife, Daehyun, enabled me to start and complete the otherwise impossible dream of attaining this degree, a nd my two sons added further joy and peace to the journey. Without the tireless encourag ement and unconditional love of my other family members, parents in law, sister, brother, and especially the truly sacrificing love of my father and mother, I could not have completed this study. I deeply admire them.
v Last, but certainly not leas t, I thank God for being with me always. He was an omnipresent and constant support.
vi TABLE OF CONTENTS Chapter Page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.............................................................................................................x ABSTRACT....................................................................................................................... xi CHAPTER 1 INTRODUCTION........................................................................................................1 Need for Present Research............................................................................................1 Role of Current Study...................................................................................................3 Research Objectives......................................................................................................5 2 LITERATURE REVIEW AND HYPOTHESES.........................................................6 Overview....................................................................................................................... 6 Brand Loyalty...............................................................................................................6 Brand Commitment....................................................................................................12 Attitude Strength.........................................................................................................14 Cognitive Brand Conviction.......................................................................................20 Affective Brand Conviction........................................................................................23 Mediating Role of Attitude Strength..........................................................................26 Brand Credibility........................................................................................................27 Relationships between Brand Convictions and Brand Credibility.............................30 Moderating Roles of Invol vement and Product Type.................................................32 Proposed Model..........................................................................................................34 3 METHODOLOGY.....................................................................................................36 Research Design Overview.........................................................................................36 Sampling.....................................................................................................................36 Pretests....................................................................................................................... .37 Pretest 1: Validation of Conviction Scale...........................................................37 Pretest 2: Product Type Manipulation Pre-check................................................38 Main Survey................................................................................................................39 Sample.................................................................................................................39
vii Measurement Instruments...................................................................................39 4 RESULTS...................................................................................................................46 Pretest Results.............................................................................................................46 Pretest 1...............................................................................................................46 Pretest 2...............................................................................................................48 Main Study: Model Estimation and Results...............................................................51 Descriptive Statistics of Measurements...............................................................51 Assumption Check...............................................................................................55 Reliability and Validity.......................................................................................55 Confirmatory Factor Analysis (CFA)..................................................................58 Structural Equation Modeling.............................................................................63 Overview of structural modeling procedure.................................................63 Model estimation..........................................................................................64 Mediating role of attitude strength...............................................................72 Examining research questions......................................................................73 5 SUMMARY AND DISCUSSION.............................................................................77 6 CONTRIBUTION......................................................................................................84 APPENDIX A DESCRIPTIVE STATISTICS OF MEASUREMENT ITEMS.................................86 B CORRELATION MATRIX OF MEASUREMENT ITEMS.....................................87 C QUESTIONNAIRES..................................................................................................91 Pretest 1...................................................................................................................... 91 Pretest 2...................................................................................................................... 95 Main Survey Sample (HI-H: Designer Sunglasses).................................................100 LIST OF REFERENCES.................................................................................................106 BIOGRAPHICAL SKETCH...........................................................................................122
viii LIST OF TABLES Table page 2-1. AbelsonÂ’s Conviction Items......................................................................................21 2-2. Mediating Role of Attitude Strength.........................................................................27 3-1. Involvement scale and Hedoni c/Functional perception question..............................39 3-2. True Brand Loyalty Scale..........................................................................................40 3-3. Brand Commitment Scales........................................................................................41 3-4. Attitude Strength Scale..............................................................................................42 3-5. Brand Credibility Scale.............................................................................................45 4-1. Descriptive Statistics of Pretest 1..............................................................................47 4-2. Purification of Brand Convi ction Items based on Pretest 1.......................................47 4-3. Descriptive Statistics of Hedonic/Functional Perception..........................................49 4-4. Descriptive Statistics of Involvement........................................................................49 4-5. Dunnett T3 Multiple Comparison Test Results (Hedonic vs. Functional)................50 4-6. Dunnett T3 Multiple Comparison Test Results (Purchase Decision Involvement)..50 4-7. Details of Scales Used for CFA.................................................................................52 4-8. Discriminant Validity Tests.......................................................................................58 4-9. Select Modification Indices and Respecification Results..........................................59 4-10. Path Diagrams of Competing Models.....................................................................66 4-11. Fit Indices of Competing Models............................................................................69 4-12. Standardized Path Coefficients in Competing Models............................................70 4-13. Path Coefficients under Specific Conditions...........................................................76
ix 4-14. Effects of Convicti ons on Attitude Strength............................................................76
x LIST OF FIGURES Figure page 2-1. Proposed Model.........................................................................................................35 3-1. SAM (Self-Assessment Mannequin).........................................................................45 4-1. Final Measurement Model.........................................................................................62 4-2. Standardized Path Co efficients of Mt model.............................................................69
xi Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy AFFECTIVE AND COGNITIVE BRAND CONVICTION: MAKING LOYAL RELATIONSHIPS By Jooyoung Kim August 2003 Chair: Jon Morris Major Department: Journalism and Communications Despite the recent appreciati on of the relationship noti on in consumer marketing strategies, theoretical constr ucts of consumer-brand relati onships and the process of brand loyalty formation have been understudi ed in past consumer psychology studies. In this research, using the recently developed not ion of Â“true brand loyalty,Â” which is brand sensitivity-based repeat purch asing behavior, the psychologi cal process of brand loyalty formation was studied utilizing Structural E quation Modeling with six latent constructs: Brand Credibility, Affective Brand Convi ction, Cognitive Brand Conviction, Attitude Strength, Brand Commitment, and True Brand Loyalty. With an observation sample size of 952, the study tested the proposed model a nd examined the final model under different conditions that varied the involvement and pr oduct type as being hedonic and functional. The results indicate that all six latent construc ts take important roles in the brand loyalty formation process. Among the constructs, affective conviction, which has been often ignored in past studies, showed that it can independently influence the brand loyalty
xii formation process. In addition, attitude st rength construct was a necessary mediating construct between convictions and brand co mmitment. This study further found stability of the roles of affective and cognitive conviction across several conditions, while showing that the indirect influences of brand credibility to attitude strength via affective or cognitive convictions varied in di fferent conditions. Also, the study found a relationship between affectiv e and cognitive conviction, where affective conviction often influences the formation of cognitive conviction.
1 CHAPTER 1 INTRODUCTION Need for Present Research Â“The unit of value in business today and the coming years is no longer products but relationshipsÂ” (Duncan and Moriarty 1997, p. 41). Â“The relationships are the key, the basis of customer choice and company ad aptationÂ” (McKenna 1991, p. 4). Â“Relationship principles have virtually repl aced short-term exchange notio ns in both marketing thought and practiceÂ” (Fournier 1998, p. 343) . These arguments clearly s how the critical role of consumer-brand relationships to business su ccess, and attest to why Deighton (1996) considered relationship-focused marketing a pa radigm shift for the marketing field as a whole. Despite this recent appreciation of the re lationship notion in consumer marketing strategies, the theoretical constructs of consumer-brand relationships appear greatly understudied in consumer psychology liter ature (Fournier 1998). Although it may be argued that many insightful works exist in the marketing literature, most are, in fact, conceptual works, without much scientific i nvestigation into the st ructures underlying the consumer-brand relationships (Fournier 1998; Sheth and Parvatiyar 1995). It should not be surprising to find a lack of theore tically sound constructs on consumer-brand relationships, as the field itself has only recently received academic attention. Some recent special issues of academic journals (e.g., Journal of Business Research 1999; Journal of Strategic Marketing 1998; Psychology and Marketing 1997) focus attention on the topic of relationship marketing and serve as an indication of recent academic interest.
2 Although the relationship research field is still in the early st ages of maturity (Bejou 1997), Â“brand loyalty,Â” a very comp arable marketing concept dealing with consumer-brand relationship, has been studied extensively. Fournier suggests that Â“the quality of brand relationship represents an a lternative to the constr uct of brand loyaltyÂ” (Fournier 1998, p. 367) In consideration of th e fact that the prior purpose of building a strong consumer-brand relationship was to retain customers or maintain good relationships (Morgan and Hunt 1994; Berry and Parasuraman 1992), brand loyalty has been examined in several relationship-mark eting studies (e.g., Reynol ds and Beatty 1999; Beatty et al. 1996; Donath 1994). The study of brand loyalty has been represented in the marketing literature for nearly eight decad es, since CopelandÂ’s introduction of Â“brand insistenceÂ” in 1923 (Jacoby and Chestnut 1978). Early research was primarily focused on the operational definition of beha vioral aspects (i.e., repeated purchase) of brand loyalty, but subsequent to Jacoby and ChestnutÂ’s ( 1978) work, brand loyalty has been studied in terms of both attitudinal and be havioral aspects, producing va luable contributions to the field. Despite such efforts, several research ers (Odin et al. 2001; Fournier 1998; Bloemer and Kasper 1995) have recently argued that the true meaning of relationship or attitudinal aspects (e.g., commitment, Bloemer and Kasper 1995) of brand loyalt y have vanished in traditional brand loyalty research due to the i ndifferent operationalizations of inertia and loyalty. Thus the basic questions of whether, why, and in what forms consumers seek and value ongoing relationships with brands remain in large part unanswered (Webster 1992; Fournier 1998). Stressing this lack of rela tionship-oriented attitudinal framework of brand loyalty, several researchers have recently suggeste d valuable conceptual frameworks (e.g.,
3 commitment: Odin et al. 2001; brand sensitiv ity: Bloemer and Kasper 1995; commitment and trust: Morgan and Hunt 1994) that distingu ish true brand loyalty from inertia. Others have used constructs from theories of inte rpersonal relationships, such as love (Shimp and Madden 1988), commitment (Dick and Basu 1994), and trust (Hess 1995) in explaining the underlying cons tructs of true consume r-brand relationships, although without explicit and comprehensive consid eration of other important relationship constructs (Fournier 1998). In order to address this incomplete comprehensive relationship-oriented view of consumer-brand interactions, Fournier (1998) qualitatively researched a consumer-brand relationship framework based on interpersonal/social relationship theories and suggested six multifacet ed nature of relationship strengths (i.e., love and passion, self-connection, interdep endence, commitment, intimacy, and brand partner quality). Although many researchers have studied the concept of brand loyalty, no agreement on the scientific structure of th e loyalty construct ha s yet been clearly established. Role of Current Study The present research attempts to investigat e the scientific struct ure of brand loyalty by synthesizing past studies with some critical variables that are th eorized to explain the brand loyalty formation process. Those importa nt variables included in this study are the constructs of attitude strength, conviction, brand credibility, and brand commitment. Attitude strength theories taken from ps ychology are capable of explaining the construct and process of brand loyalty formati on because the concept of attitude strength has two characteristics: durability and imp actfulness (Krosnick and Petty 1995). These two characteristics should be important in understanding brand loyalty since it impacts repeat purchasing behavior in terms of maki ng the behavior more persistent and durable.
4 Among many extant attitude st rength studies (e.g., Bizer a nd Krosnick 2001; Haddock et al. 1999; Pomerantz et al. 1995; Abels on 1988; Raden 1985, to name a few recent studies), AbelsonÂ’s (1988) research presented a remarkable study that focused on distinguishing Â“nonattitudeÂ” (C onverse 1970; Rosenberg 1968) fr om true attitudes, and suggested that conviction was a necessary condition of beha viorally predictable true attitude, and thus on the contrary that att itudes without conviction were unstable and unpredictable non-attitudes. The current study uses this conviction cons truct in explaining the core underlying structure of brand loyalty because, consistent with Abelson (1988), brand loyalty can be a behavioral response based on a strong loyal atti tude held with a strong conviction for the brand. The preceding considered, loyal attit ude without brand conviction would be a nonattitude (i.e., with a low att itude strength) and would result in either non-loyal behavior or simple inertial behaviors. Based on Jacoby and ChestnutÂ’s work (1978), the present research postulates that a consumerÂ’s conviction to a brand resides in two distinct places: cognitive and affective areas of conviction. Jacoby and Chestnut s uggested brand loyalty is based on brandrelated beliefs, states of a ffect, and behavioral inten tions; these can be related respectively to the cognitive ar ea of conviction, the affective area of conviction, and loyal intention. In addition, this study compares the differential roles of affective and cognitive conviction across several product categories, wh ich are characterized as either functional or hedonic, and either low or high involvement product categories. Rationales for these types of product sel ections are discusse d later in detail.
5 Furthermore, this study proposes and include s brand credibility as an antecedent to cognitive and affective convicti ons. Brand credibility, defined as Â“the believability of the product position information contained in a br and, which entails cons istently delivering what is promisedÂ” (Erdem et al. 2002, p. 3), ha s recently attracted research attention (e.g., Erdem and Swait 1998) and been shown as in fluencing various constructs such as perceived quality, perceived risk, perceived information costs saved, and relationship certainty, which can be considered as importa nt particular reasons for brand loyalty. Consistent with Fishbein and Ajzen ( 1975), the present study views behavioral intention as the most predictable of behavior s, and thus proposes a direct antecedent of loyal behavior. This construct is Â“brand comm itment,Â” which is viewed in the present study as behavioral intention held with c onviction (i.e., affective and cognitive). In psychology, the concept of commit ment is regarded as having intentional aspects, as evidenced by KieslerÂ’s (1971, p. 30) definiti on of commitment as Â“the pledging or binding of an individual to behavioral acts.Â” Although in th e previous literature brand commitment has been studied as an anteceden t of brand loyalty or the direct indicator (i.e., a scale item) of brand loyalty itself , the present study more actively views this construct as an intentional construct of brand loyalty. Research Objectives The objectives of the present research are (1) to explore the underlying psychological process of brand loyalty form ation, focusing on the constructs of both cognitive and affective brand conviction and att itude strength theories together with other important constructs such as brand credibili ty and commitment, and (2) to identify the mechanisms of the brand loyalty formation process.
6 CHAPTER 2 LITERATURE REVIEW AND HYPOTHESES Overview The current study views brand loyalty as repeated purchasing behavior resulting from loyal intention, based on affective and cognitive conviction, arising from the credibility that consumers have attributed to a brand from their past direct and indirect experiences. Based on this view, relevant hypotheses are drawn from the theories of brand loyalty, brand commitment, attitude strength, brand conviction, and brand credibility. Following reviews of relevant literature and th eories, a full path model, including all relevant co nstructs, are suggested. Brand Loyalty Marketing literature has approached br and loyalty with two quite divergent philosophies, which are characterized as th e stochastic and deterministic approaches (Jacoby and Chestnut 1978). Stochastic theories pertain to behavioral outcomes of brand loyalty (i.e., repeat purchase) at the macro level, and stochastic researchers argue that there are too many unknown/inexplicable complex micro-level re asons behind the beha vior (McAlister and Pessemier 1982; Bass 1974). Therefore, brand loyalty is only considered and studied from a behavioral perspective, and the r easons (e.g., attitude) be hind the behavior are considered impossible to invest igate. This approach assume s that brand loyalty can be theorized and measured under th e existence of some number of consumers. Reynolds and WellsÂ’s (1977) repeat-purchas e rate index is an example of the stochastic approach,
7 which inspects the entire sampleÂ’s total pur chasing of one brand over a series of twoweek intervals. In their study, no consideration was gi ven to any single purchase, but instead the aggregate purchasing data were regarded as represen ting the brand loyalty. This type of aggregate view first appeared in 1932 when the Psychological Corporation began conducting a survey of 1500 different br ands in order to pr ovide a statistical monitoring of market share. Th e survey question was rather simple, asking participants to recall the last brand purchased. Cunningham (1956) suggested a si milar concept, the Â“market share concept,Â” which views brand loya lty as Â“the proporti on of total purchases represented by the largest singl e brand usedÂ” (p. 118). This numerical proportion based on the market share data serves as an index of the strength of the loyalty in a certain group of consumers (e.g., a household). Chur chill (1942) suggeste d an alternative approach to the survey method. He argued th at brand loyalty coul d be measured only by observing all purchasing data from a fixed gr oup of consumers. According to Jacoby and Chestnut, this would allow marketers to i nvestigate several ques tions such as: Â“From which brand am I obtaining customers? Once I have obtained them, are they staying with me? Finally, if they are switching, which br ands are they swit ching to?Â” (1978, p. 11). However, these panel-data approaches have been criticized because arbitrary cutoff points had to be established in defining loya lty. Normally if 50% or more of household purchases were concentrated on a single brand, the household could be said to be loyal to the brand. But, as Jacoby and Chestnut argue d, it becomes difficult to define loyalty if equal amounts of purchases are co ncentrated on three or more brands from a larger set of available brands.
8 This all-or-nothing conceptu alization of brand loyalty was replaced by PessemierÂ’s (1959) Â“continuum.Â” Using a laboratory simu lation of shopping behavior, he examined how long loyal buyers would accept price increases on brands they were loyal to before ultimately deciding to switch to alternative brands. This research shifted previous brand loyalty concepts and measures by considering the degree to which brand loyalty exists, in contrast to the previous con cepts of whether brand loyalty simply exists or not (Jacoby and Chestnut 1978). PessemierÂ’s study was quite innovative for the time, and the majority of brand loyalty research continue d on the macro and panel-data-based behavior level. For example, according to LipsteinÂ’s ( 1959) address to the annual conference of the Advertising Research Foundation (ARF), the majority of rese archers were of the opinion that brand loyalty could be observed only in terms of an aggregation of panel data for consumer purchases. Although the stochastic approach is known to be accurate in predicting future loyal behaviors in the aggregate, the major draw back is that compan ies cannot know how to influence repeat purchase behavior, because th e reasons for the behavior are considered unknown and inexplicable (Jacoby and Chestnut 1978). Furthermore, stochastic models are generally considered to have measurem ent problems, such as in attempting to measure dichotomous loyalty judgment (i.e., lo yal vs. disloyal), which is Â“singularly short of nuance, and requires a very arbitrar y judgment for the allo cation of a consumer to one or the other of the two categoriesÂ” (Odin et al. 2001, p. 76). In contrast to the stochastic view, dete rminism approaches brand loyalty at the micro consumer level. It views the repeated purchase of the same brand by the same consumer as the direct consequence of so me limited number of explanatory factors
9 underlying the consumerÂ’s behavior. Since being introduced by Copeland (1923), who suggested an extremity of attitude would ha ve a special effect on brand insistence, attitudinal aspects (i.e., determinism) of br and loyalty have been investigated in many studies (e.g., Day 1969: true bra nd loyalty as one consistent in both behavior and attitude; Cunningham 1967: a measure of perceived brand commitment as an antecedent of brand loyalty; Guest 1944: empirical support for the status of attitudes as one of the prime determinants of brand loyalty; McGregor 1940: importance of a relationship between advertising and brand loyalty). The major advantage of the dete rministic (i.e., attitudinal) approach is that it enables marketers to isolate, understand, and manipulate the underlying causes of brand loya lty (Jacoby and Chestnut 1978). Day (1969) was the one of the first resear chers to challenge the macro behavioral measures of brand loyalty. He suggested a num erical measure of bra nd loyalty that is a combination of both behavior and att itude. Further, Jacoby (1971, 1970, 1969; Jacoby and Olson 1970) advanced to distinguish betw een brand-loyal behavior and brand-loyal attitudes, by saying that Â“Bra nd loyal behavior is defined as the overt act of selective repeat purchasing based on evaluative psycholog ical decision process, while brand loyal attitudes are the underlying predispositions to be have in such a selective fashionÂ” (Jacoby 1971, p.26). SpellerÂ’s (1973) study also suggest ed a strong correlation between brand loyal attitudes and behaviors. Jacoby and ChestnutÂ’s (1978) work was one of the first efforts to suggest the reconciliation of both stochast ic and deterministic approaches. They are known as the first researchers to propose a definition of br and loyalty that include s six core elements. According to Jacoby and Chestnut (1978, p. 80), brand loyalty is Â“(1) the biased (i.e.,
10 non-random) (2) behavioral response (i.e., purc hase) (3) expressed over time (4) by some decision-making units (5) with respect to one or more alternative bra nds out of a set of such brands, (6) which is a function of a psychological (decision making, evaluative) processesÂ” (numbers in parenthesis are adde d). As elements (2) and (6) indicate, they viewed brand loyalty as both behavioral and attitudinal responses and they further conceptualized the attitudina l or psychological structure of brand loyalty as being composed of beliefs, states of affect, and behavioral intentions. Though this tripartite structure would benefit loyalty studies in finding/understanding the underlying structures, only a few studies (e.g., Dick and Basu 1994) actually adopted this framework. However, subsequent to Jacoby and ChestnutÂ’s work (1978), brand loyalty has been primarily studied with a composite (i.e., attitudinal a nd behavioral) view, and several studies have generated insightful contributions (Sherry 1987) . Unfortunately, the research stream has stagnated of late (Fournier 1998; Lehma nn 1996) without any agreed construct or measurement (Chaudhuri 1999). In the hope of discovering an agreeable meas ure and construct, several studies (e.g., Odin et al. 2001; Fournier 1998; Bloemer a nd Kasper 1995) have recently suggested the need for understanding the difference between tr ue loyalty and spuri ous loyalty and argue that the true meaning of attit udinal aspects of brand loyalty has been lost in traditional brand loyalty research (Fournier 1998) becau se of indistinguishable operationalizations of inertia (or spurious loyalty) and true loya lty. For this reason, some recent studies have explored and suggested several distinguishers or moderators for true loyalty and inertia (e.g., relative attitude: Dick and Basu 1994; brand sensitivity: Odin et al. 2001, and Bloemer and Kasper 1995).
11 Dick and BasuÂ’s (1994) Â“rel ative attitudeÂ” is a cons truct that contains two dimensions: the degree of attitudinal extremity (to a specific brand) and the degree of attitudinal differentiation (to alternative bra nds). While they did us e Â“attitude strength,Â” they used it interchangeably with attitude extremity. Thus, their attitude strength was rather closer to attitude extremity than to th e full concept of attitude strength. The full conceptualization of attitude st rength is discussed later in detail. Dick and Basu argued that there could be four cond itions based on the cross-classifi cations of the two levels of each dimension (i.e., attitudinal extremity: strong/weak; attitudinal differentiation: No/Yes). For example, the combination of an absence of attitudinal differentiation and weak attitudinal extremity would result in the lowest relative attitude, while high attitudinal differentiation t ogether with strong attitude extremity would result in the highest relative attitude. After the classificati ons of relative attitude, they reclassified the level of relative attitude w ith repeat patronage, which fi nally resulted in the four conditions of brand loyalty: no-loyalty (low RA/low RP), s purious loyalty (low RA/high RP), latent loyalty (high RA/low RP), and (tru e) loyalty (high RA/hi gh RP), where RA is relative attitude and RP is repeat patronage. Though it di d not rigorously apply the attitude strength concept, this research is exceptional in that it conceptually and theoretically disti nguishes spurious loyalty from true loyalty by including, to some extent, the concept of attitude strength. Brand sensitivity, used by Odin et al. (2001) is another concept that is theorized to distinguish between true and spurious brand loyalty. Like Filser ( 1994), and Kapferer and Laurent (1983), Odin et al. assumed that the repurchase of the same brand under conditions of strong perceive d brand differences characte rizes brand loyalty. They
12 conceptualized this perceived brand differen ce as brand sensitivity, and conjectured that the level of brand sensitivity differentiates lo yalty from inertia (i.e., spurious loyalty). Following Odin et al, the present study employs brand sensitivity as a distinguisher (or moderator) of true brand loya lty from inertial purchases. Brand Commitment Fishbein and AjzenÂ’s expectancy value mode l implies that an intentional aspect of attitude (i.e., loyal intention) would reside between loyal at titude and loyal behavior. The present research considers this intentional as pect of loyalty as Â“brand commitmentÂ” since this construct has a behaviorally intentional aspect, in accordance with a social psychology definition of commitment, Â“the pl edging or binding of an individual to behavioral actsÂ” (Kiesler 1971, p. 30). Ea gly and Chaiken (1993) also argue a behaviorally intentional f eature of commitment, saying that commitment pertains specifically to behavioral me thods of strengthening attit udes. Among various definitions of commitment (see Young and Denize 1995) subs equent to Kiesler, a recent definition of commitment in the relations hip marketing literature provid es a better understanding of the construct, as it incor porates both a rational and a ffective bond to relationship. In relationship marketing, Hennig-Thurau and Klee (1997, p. 752) define commitment as Â“a customerÂ’s long-term ongoing orientation toward a relationship grounded on both an emotional bond to the relationshi p (affective aspect) and on th e conviction that remaining in the relationship will yield higher net benefits than terminating it (c ognitive aspect).Â” Being an affectively and c ognitively bonded intention to behavioral loyalty, brand commitment serves as a critical direct ante cedent of behavioral (t rue) brand loyalty. As Kiesler (1971) stated, the main effect of commitment is Â“to make an act more difficult to undo, deny, distort, or reinterpretÂ” (p. 72). Commitment ha s been regarded as
13 a condition of attitude polarization and the re sistance to attitude change (e.g., Eagly and Chaiken 1993; Brickman 1987). Eagly and Chaiken (1993) suggest that commitment takes the form of public advocacy of a pos ition, and that high commitment would be created when subsequent behavior is very explicit and unambiguous, important to the subject, irrevocable, re peated, and freely chosen. In addi tion to the resistance effect of commitment, Millar and Tesser (1986) suggest ed that commitment w ould induce attitude polarization, because it would lead to attemp ts of justifying the existing attitude, which this study considers being directly re lated to the brand loyal behaviors. In fact, past literature has viewed brand commitment as a necessary and sufficient condition of brand loyalty (e.g., Knox and Walker 2001; Bloemer and Kasper 1995). Though the studies show the positive relatio nship between brand commitment and brand loyalty, that evidence is rather correlational. They even us ed brand commitment as an item of brand loyalty measurement (e.g., in Bloemer and Kasper 1995), rather than an antecedent of brand loyalty, and thus faile d to uncover the causal relationship between brand commitment and brand loyalty. Cunningham (1967) also viewed brand commitment as an antecedent of brand l oyalty, but the concept of brand loyalty in CunninghamÂ’s study did not distinguish true loya lty from inertia. On the contrary, the present research studies th e relationship between brand commitment and Â“trueÂ” brand loyalty. The first hypothesis of the present resear ch is stated as follows, based on the preceding discussions. This hypothesis applie s the classic intention-behavior model within the domain of brand loyalty.
14 H1: Consumers will be more Â“trulyÂ” loyal to a brand when they have a higher level of commitment toward the brand. Attitude Strength About eight decades ago, Copeland (1923) suggested that an extreme attitude toward a particular brand might have a sp ecial effect on buyer be havior, especially on what he called Â“brand insisten ce.Â” In social psychology, stro ng resistance to attitude change is regarded as related to the strengt h of the existing attitude (Eagly and Chaiken 1993). In other words, people have a psychologica l process that facili tates resistance to persuasion and that can hamper even well designed persuasive efforts. Once attitudes have been established, people expend a certain effort towards maintaining them when they are pressured to change . Though attitude strength has often been discussed in the literature, it has been argued that the concep t of attitude strength has been more of a vague metaphor than a formally defined scie ntific construct (Krosnick and Petty 1995). Stressing this lack of scien tific definition of attitude strength, Krosnick and Petty proposed a definition of attitude strength ba sed on the Â“defining featuresÂ” of attitude strength. They define attitude strength as the extent to which attitudes manifest the qualities of durability and impact fulness. According to them, manifestations of attitudinal durability are considered as persistence and resistance, and the manifestations of the impactfulness are viewed as judgment-in fluencing and behavior-guiding. Treating attitude strength in this ma nner would allow incorporating the most common meaning of the construct and would be consistent w ith past work (Krosnick and Petty 1995). The question about whether attitude streng th is unitary or multifaceted variables has been a major concern of attitude strength studies. Studies that follow the associative networks approach (e.g., Judd and Kronsnick 1989; Fazio 1986) ar gue that attitude
15 strength is a single construct, while others consider it a multidimensional construct (e.g., Abelson 1988; Raden 1985). Scott (1968) is kno wn as one of the earliest scholars to propose multidimensional properties of attitude strength. He described 10 attitude strength properties: extremity, intensity, ambivalence, salience cognitive complexity, overtness, embeddedness, flexibility, and cons ciousness. Raden (1985) expanded ScottÂ’s attitude strength properties by examining acce ssibility, evaluative-cognitive consistency, certainty, behavioral experience, importance, latitudes of acceptance /rejection, and vested interest. Since then, many scholars have stud ied these attributes or proposed some new properties. Although no clear agreement yet exists in th e psychology literature about the latent structure of attitude strength (Bizer and Krosnick 2001), the literature, as with ScottÂ’s (1968) and RadenÂ’s (1985) works, has converged into a consensus that attitude strength is multidimensional. Because the past liter ature has suggested too many or overly complicated properties, researchers have felt th at those properties shou ld be reduced to a few condensed factors. In response to th is demand, several studies have recently attempted to factor analyze many attributes of attitude strength. However, the results have been somewhat inconsistent, such as with findings of a single f actor (e.g., involvement: Verplanken 1989), two latent structur es (e.g., commitment and embeddedness: Pomerantz et al. 1995), three constructs (e.g., emotional commitment, ego-preoccupation, and cognitive elaboration: Abelson 1988), or four constructs (e.g., conviction, extremity, belief homogeneity, media exposure, and aff ective-cognitive consiste ncy: Erber et al. 1995). Criticizing these inconsistencies, some researchers have argued that each attitude strength attribute should repr esent a unique construct, ra ther than a portion of a few
16 reduced factors (Lavine et al. 1996; Krosnick and Petty 1995; Krosnick et al. 1993). However, according to Krosnick and Petty (1995 ), all of those specific properties can be explained by the simple principal definition of attitude strength: durability and impactfulness. Krosnick and Petty propose that strength-related attr ibutes can fit into three categories: (1) aspects of the attitude itself, (2) aspects of the cognitive structure associated with the attitude and attitude object in memory, and (3) determinants of attitude strength. First, attitude strength as an aspect of at titude means that it represents the extremity of attitude. In the literature, attitude is viewed as having both valence (positive or negative) and extremity. Extremity is the exte nt to which the attitude deviates from neutrality. The more extreme an attitude is, the more an individual likes or dislikes the object. Also, the structural aspects of at titude strength include accessibility, knowledge, evaluative-cognitive consistency, evaluative-affective consistency, and ambivalence. Second, Fazio (1986) proposed that an att itude could be considered as a link between the attitude object and its evaluation in memory. Attitude accessibility is defined as the strength of this object-evaluation li nk, and it refers to the ease with which an attitude comes to mind in the course of social perceptions. Some attitudes are accompanied by a great deal of attitude-relevant knowledge, while others are formed with little knowledge (Wood 1982). Thus, attitudes ar e also considered to be linked to knowledge about the attitude object. In the literature, knowledge ha s been studied in terms of the breadth, subjective perception, a nd content of the information a person has about an object. In addition, the knowledge in formation can include past experience (e.g., emotional, behavioral, specific attributes) that are evoked by th e attitude object. Attitudes
17 may also vary in the degree to which there is consistency between evaluations of the object and the information associated with it. In the literature, consistency has been studied following two streams: evaluative-c ognitive and evaluative-affective consistency. This consistency theory will be further discusse d later, when the determinants of attitude strength are discussed. In cont rast to consistency theories, the literature also provides theories about inconsistency, or ambivalen ce. Ambivalence refers to the degree of conflict between positive and negative component s associated with the same object. However, each relationship among the dimensions (e.g., accessibility and consistency) is largely known to be weak ly correlated. For example, Raden (1983) found a low correlation between intensity and certainty. Furthermore, more statistically sophisticated studies have also found weak relationships between the dimensions (e.g., Wilson et al. 1990, using multiple measures of dimensions; Chuang 1988, Judd and Krosnick 1982, using multiple indicator multitrait-multimethod analysis). The literature has shown that attitude strength dimensions are generally separabl e from one another, thus representing distinct c onstructs. Some researchers have attempted to identify dimensions that can coexist. AbelsonÂ’s ( 1988) naming of three factors, consisting of cognitive elaboration, emotional commitment, and ego preoccupation, is one such effort. Krosnick and Abelson (1992) interpret cognit ive elaboration as a similar construct to knowledge, emotional commitment as a combin ation of intensity a nd certainty, and egopreoccupation as a similar dimension to importance. Third, two categories of attitude strength de terminants can be found in the literature (Eagly and Chaiken 1993): the motivationa l and the cognitive determinants. Specific determinants such as ego-involvement (or attitude intensity), attitude importance,
18 certainty, and commitment have been cons idered as motivational determinants, and cognitive consistency has been studied as a cognitive determinant of attitude strength. Sherif and Cantril (1947) defined ego-i nvolvement primarily in motivational and affective terms and regarded this variable as indexing the intensity of attitudes. They assumed that people who are highly ego-invo lved with a topic experience tension and discomfort when they encounter informati on that challenges their attitudes on the topic because it threatens their personal identity. Atti tude intensity has been also studied in a more emotional sense. Krosnick and Schuma n (1988) define attit ude intensity as the strength of feelings about an attitude object. Attitude impor tance refers to the degree to which a person is passionately concerned about , and personally invested in, an attitude that they care deeply about (Krosnick 1989). One of the most well known attitudeimportance-related constructs is conviction (Abelson 1988). Attitude certainty refers to the degree to which an individual is confident th at his or her attitude toward an object is correct. The last motivational determinant fr equently studied is commitment. In fact, commitment may represent the behavioral e xpression of attitude certainty toward an object. Kiesler et al. (1977) ar gue that commitment is the behavioral method of attitude strengthening. As a cognitive determinant of resistance, cognitive consistency theory, like the domino theory, suggests that attitude cha nge is resisted because of the cognitive disruption it can produce. Once an attitude is linked to another attitude or to some other cognition, change in the target attitude is s een as more difficult due to peopleÂ’s natural tendency to maintain cognitive-consiste ncy. Positive relations between cognitive
19 consistency and attitudinal resistance have been identified in several studies (e.g., Chaiken et al.1995). Since there are numerous dimensions of a ttitude strength in the literature, the present study has adopted five important dimensions recommended by Krosnick and Abelson (1992). The five dimensions are ex tremity, intensity, certa inty, importance, and knowledge. Based on their extensive literature review, Krosnick and Abelson suggested using these five dimensions in measuring a ttitude strength because they possess three great virtues: they are the easiest to measure in surveys, they are the easiest to comprehend conceptually, and they are the most extensively validated as measures of the fixedness and conceptuality of attitudes. First, as discussed, extremity is the de gree of favorableness or unfavorableness of an individualÂ’s evaluation of a given object. Th e more extreme an individualÂ’s attitude, the farther it is from neutrality. Theref ore, attitude extremity has often been operationalized as the deviation from the neutra l point of an attitude scale (Downing et al. 1992). Second, attitude intensity is , as discussed, the strength of an individualÂ’s feelings about an attitude object (Krosnick and Schuma n 1988). Third, attitude certainty, as also discussed, refers to the degree to which an i ndividual is certain about the correctness of his or her attitude. Because some researchers have regarded certainty as the strength of the belief system underlying the attitude, those researchers have measur ed the certainty of beliefs directly, rather than measuring indirect ly. Fourth, attitude importance is the degree to which an individual consid ers an attitude to be persona lly important. Lastly, attituderelevant knowledge refers to the breadt h of stored beliefs about the object.
20 Although brand commitment, as H1 states, is a necessary construct leading to brand loyalty, the present research ta kes the additional view that true brand loyalty should be also based on attitude strength, which ma kes brand commitment more durable and impactful (Krosnick and Abelson 1992), and thus leads to true brand loyalty. From this postulation, the second hypothesi s is generated as follows: H2: Consumers will show stronger brand commitment to a brand when they have a higher level of attitude stre ngth toward the brand. Cognitive Brand Conviction Regarding attitude strength toward a brand, studies conducted by Abelson (1988) and Erber et al. (1995) provide important implications for the existence of a general attitude strength construct to ward an object, because their attitude stimuli focused on social objects (i.e., president Ronald Reag an: Erber et al. 1995; God: Abelson 1988), which differed from other studies that focuse d only on various social issues (e.g., capital punishment: Pomerantz et al. 1995). Because a br and is a social object, rather than an issue, past studies of attitude-strength to ward social objects would be capable of providing a closer understanding of brand loyal attitude. As discussed, AbelsonÂ’s study was undertaken with a need to distinguish true attitudes from non-attitudes (Converse 1970). Abelson argued that a durabl e and behaviorally predicta ble attitude is one with conviction, whereas an attitude without conviction is unstabl e. The results of his study suggest that the construct of conviction is multidimensional and a good predictor of the durability of attitudes over time. He found the three robust dimensions of conviction to be emotional commitment, ego-preoccupation, a nd cognitive elaboration (Table 2-1).
21 In consensus with Abelson, this stud y adopts conviction, as a major direct antecedent to attitude strength. Its generalizab ility, predictability, and construct validity have been confirmed in seve ral subsequent studies (e.g., Er ber et al. 1995). Accordingly, this study proposes that brand commitment is a strong, stable, and favorable attitude toward a brand, held with conviction to the brand. Consistent with recent efforts of differentiating true brand loyalt y from inertia, the current st udy proposes that the degree of conviction differentiates true loyalty from inertia. Table 2-1. AbelsonÂ’s Conviction Items Emotional Commitment 1. My beliefs about X express the real me. 2. I can't imagine ever changing my mind. 3. My beliefs are based on the moral sense of the way things should be. 4. I would be willing to spend a day a month working for a group supporting my views. 5. I think my view is absolutely correct. Ego Preoccupation 1. I think about X often. 2. I hold my views very strongly. 3. My belief is important to me. 4. I am extremely concerned about the issue. 5. When I think about the issue, I feel fearful. Cognitive Elaboration 1. I've held my views a long time compared to most people. 2. Several other issues could come up in a conversation about it. 3. Several things could happen if my views were enacted. 4. I have more knowledge on the i ssue than the av erage person. 5. It's easy to explain my views. However, application of AbelsonÂ’s items is employed differently in the present study because of two reasons. First, some items do not seem to fit effectively into the context of brand loyalty. For example, the third item under the emotional commitment factor doesnÂ’t seem to be relevant to brand loyalty context (see Tabl e 2-1). Therefore, the current research pretests the appropriateness of all Abels onÂ’s (1988) items in the domain
22 of brand loyalty. Secondly, the current research c onsiders that AbelsonÂ’s conviction items represent cognitive, rather than emotiona l, items. Though Abelson named one of three conviction factors as Â“emotional commitment, Â” each item in the emotional commitment factor seems to actually measure the mani festations of cognition-based, not emotionbased, conviction. In addition, Abelson stat ed that the two factors, emotional commitment and ego preoccupation, were subs tantially correlated (.51), and therefore suggested that combining these two factors would be possible. This result might be inconsistent with former attitude studie s (e.g., tripartite attitude model (cognitive, affective, conative): e.g., Breckler 1984; Katz and Stotland 1959) that view cognition and emotion as distinct constructs (e.g., Allen et al. 1992). This conjecture will be pretested when the validity of items are confirmed under the domain of brand loyalty. After the completion of the pretests, the third and fourth hypotheses will be tested. H3: Consumers with higher cognitive bra nd conviction will have a higher level of attitude strength. H4: Consumers with higher cognitive bra nd conviction will have a higher level of brand commitment. H3 and H4 will investigate the direct and indirect effect of cognitive brand conviction on attitude strength and brand co mmitment. Since the conviction construct is seen as an underlying construct of attitude st rength, H3 will test if it also works for the brand loyalty formation process. In additi on, H4 is hypothesized since brand commitment is seen as brand loyal intention, which is based on a high level of attitude strength that is closely related to convictions. By investigati ng the direct and indirect effect of cognitive
23 conviction on brand commitment, the study will also investigate the necessity of the attitude strength construct in th e brand loyalty formation model. Affective Brand Conviction The current study adopts affect as one of the critical components of brand conviction since the literatu re clearly implies that th e affective conviction can independently influence the bra nd loyal attitude formation. A ffect would apparently be a key component of attitude formation, as the classic tripartite attitude model consists of cognitive, affective, and conative respons es. Though affect would be an equally important construct of attit ude and its consequences, ther e has been a tendency of focusing on cognition-based attitude in the literature, which argues that cognition predominates over affective processing and a ffective reactions are always mediated by cognition (Tsal 1985; Greenwald and Leavit t 1984). Fishbein (e.g., Fishbein and Middlestadt 1995) also con ceived the notion of the c ognition-based attitude by suggesting that a consumerÂ’s attitude is a f unction of (cognitive) beliefs and those beliefs predict intentions of behavior. Other researchers (e.g., Brown and Stayman 19 92) have argued that factors such as affect can directly influence attitude and that many cognition-oriented studies failed to properly measure feelings associated with s ources of information (Edell and Burke 1987). Failing to understand the role of emoti ons by focusing on cognitive process would impede understanding about various consum er behaviors (Allen et al. 1992). Recently, many studies (Edell and Burke 1987; Holbrook and Batra 1987) have examined the role of emotion in the process of advertising messages, and have found that cognition can drive affect and that affect can drive intentions. The Advertising Research Foundation copy testing project (Brown and Stayman 1992) has also found that emotions
24 can have a direct influence on behavior th at is not captured or summed up by attitude judgments (Allen et al. 1992). More recent studies have found that affect and attitude are separate constructs and that affect is related to attitude and not dependent on cognitive variables (Machleit and Wilson 1988). One of the most recent st udies conducted by Morris et al. (2002) found that affect dominates cogniti on in advertising message proc essing under several different conditions that varied the types of media, ad content, ad format, etc. Literature on judgment under emotional cert ainty also implies the importance of the affective conviction, by showing th at the certainty associated with an emotion can affect information processing (Tiedens and Linton 2001). As stated by Jacoby and Chestnut (1978), brand loyalty is a stru cture of brand-related belief s, states of affect, and behavioral intentions, so the affect-directed preference can be more enduring than beliefs in qualitatively different levels. Furtherm ore, they suggested the interdependency between belief and affect. Hennig-Thurau a nd KleeÂ’s (1997) definition of relationship commitment also indicates the importance of the affective aspect in the quality of committed relationships. As previously discussed, they viewed the commitment as an ongoing orientation toward a relationship grounded on both emotional and cognitive bonds. The psychology literature on interpersona l intimacy suggests that the affective aspect is a prime factor of intimate relati onship conceptions (Pragne r 1995). Studies that highlight the affective aspects of intimate re lationship include, among others, OÂ’ConnorÂ’s (1992), SternbergÂ’s (1988), Miller and LefcourtÂ’s (1982) , and Sexton and SextonÂ’s (1982). These studies emphasize warmth, affec tion, involvement, love, and deep feelings of acceptance between partners in intimate relationships (Pragner 1995; Gilbert 1976). A
25 recent study by Gonzaga et al. (2001) also sugge sts love as a primary commitment device in interpersonal relationships. The mood and social memory literature further suggests a critical role of emotion in brand loyalty formation since it proposes that the major forces in shaping our memory are emotion and motivation, meaning that events that elicit motivati onal significance and intense feelings are better remembered (Bow er and Forgas 2001). Studies of mental representation of social ep isodes conducted by several re searchers (e.g., Forgas 1981; Pervin 1976; Magnusson 1971) also found that peoplesÂ’ mental representations are largely dominated by the affective characteristic s of episode stimuli, rather than by their actual descriptive features (Bower and Fo rgas 2001). A recent study by Niedenthal and Halberstadt (2000) further argues that affect often determines the use and evaluation of categories of stimuli. These results are consis tent with Zajonc (1980), who stated that the affective quality of the original input is th e first element to emerge when people try to retrieve an object such as an episode, person, piece of music, story, or name. Furthermore, affective recall can occur w ithout any factual re call. According to ZajoncÂ’s (2000) literature review on aff ective memory, there are many studies showing Â“people tend to selectively remember their affective reac tions to, and evaluations of stimuli, even when they have no recollecti on of ever seeing that stimulus before, and have no memory for the reasons for their preferencesÂ” (Bower and Forgas 2001, p. 99). Therefore, consumers can recall how they felt about a brand without much recall or specifically supported factors to justify their feelings. This is a critical component that previous attitude strength theories have faile d to address by only studying conscious and cognitive aspects of attitude. Even the studies that did examine emotions (e.g., emotional
26 commitment: Abelson 1988, as discussed; a ffective-cognitive consis tency: Chaiken and Yates 1985) were lacking because their emotional frameworks and measures represented cognitively filtered emotions, and thus ma y have overlooked some important affective reactions which do not require the specific memories, as previously discussed (Bower and Forgas 2001). Because the fundamental assump tion of brand loyalty is that it is based on consumersÂ’ memory of direct and indirect experiences with the brand, emotion (with or without memory) would serve an essential role in shaping brand loyalty. Therefore, hypotheses 5 and 6 are stated as follows: H5: Consumers with higher affective bra nd conviction will have a higher level of attitude strength. H6: Consumers with higher affective bra nd conviction will have a higher level of brand commitment. Like H3 and H4, H5 and H6 will also invest igate the direct and indirect effect of affective brand conviction on attitude strength and brand commitment. For the investigation of a ffective conviction, the PAD th eory of affective response (Russell and Mehrabian l977), which is known to be capable of characterizing diverse emotional responses in consumption situati ons (Holbrook and Batra l988; Mehrabian and Russell 1974), is used in the present study (Details about th e PAD theory are discussed later). In addition, emotional certainty (Tiedens and Linton 2001) was also used for the investigation of affec tive conviction in order to reflect the Â“convictionalÂ” aspect of the construct. Mediating Role of Attitude Strength As previously noted, Abelson (1988) ar gued that a durable and behaviorally predictable attitude is one with conviction, whereas an attitude without conviction is
27 unstable. Attitude strength, th erefore, serves as a mediator between the relationship of convictions and behaviors. This assertion is examined in th e current study in order to ascertain if both cognitive and affective convictions are mediat ed by the attitude strength construct in influencing brand commitment formation. The hypotheses that examine the mediating effects of attitude st rength are listed in Table 2-2. Table 2-2. Mediating Role of Attitude Strength Level of Conviction Coefficient Size between Attitude Strength and Brand Commitment H7-a: High Cognitive Conviction Large H7-b: Low Cognitive Conviction Small H7-c: High Affective Conviction Large H7-d: Low Affective Conviction Small Brand Credibility This study investigates the process of br and loyalty formation, and has theorized that consumers become truly brand loyal when they have a high level of brand commitment, which is based on high attitude strength held with c onviction. This study expands its scope to an additional construct, which is considered as affecting the formation of brand convictions. The consumer behavior literature ha s shown that brand loyalty (or loyal relationships) can be formed based on severa l reasons, such as satisfaction (e.g., Bloemer and Kasper 1995), risk-reducti on (e.g., Assael 1995), or tr ust (Garbarino and Johnson 1999). Among these reasons, evidence about the im portance of trust in loyal relationships is paramount. For example, in a theater ti cket subscription stu dy Garbarino and Johnson (1999) show that trust and commitment are the most appropriate indicators for consistent subscribers, while satisfaction is the most a ppropriate indicator for occasional subscribers. Morgan and Hunt (1994) also indicate that trust is a strong pred ictor of relationship
28 commitment. Many other studies have shown th at trust is at the core of successful relationships (Berry 1995; Moorman et al .1993; Dwyer et al.1987) . Morgan and Hunt (1994, p. 23) define trust as the perception of Â“confidence in the part nerÂ’s reliability and integrity.Â” Moorman et al. ( 1993) argue that trustworthines s results from expertise, reliability, and intentionali ty. Subsequently, Gwinner et al. (1998) have found the psychological benefit of trust to be more im portant than special treatments in consumer relationships with service firms. A recent study by Erdem et al. (2002) presen ts an additional implication of trust as a core component of brand loyalty formation by showing a significant causal relationship between brand credibility and consumer price-sensitivity. Brand credibility has also been studied as an important antecedent of percei ved quality, perceived risk, and informationcost-saved (Erdem and Swait 1998). Erdem and Swait (1998) define br and credibility as the believability of the product position inform ation contained in a brand, which entails consistently delivering what is promised, and it represents the cumulative effect of the credibility of all previous ma rketing actions taken by that br andÂ” (Erdem et al. 2002, p. 3). Using signaling theory and the information ec onomics view, they also argue the brand loyalty is a consequence of brand equity, due to the expected utility that motivates consumers to repeatedly buy the same brands . They view brand equity as an added expected utility a brand gives a product, wh ich is a consequence of brand credibility. According to Erdem and Swait (1998), cred ibility is conceptualized as having two dimensions, namely trustworthiness and expe rtise. Trustworthiness means that a brand will deliver what it has promised, and expert ise implies that the brand is capable of delivering the promises (Erdem and Swait 1998). They found that brand credibility serves
29 as a strong signal of perceived quality, percei ved risk, and information-cost-saved, which affects the overall expected utility of the br and. In fact, the literature (e.g., Bloemer and Kasper 1994; Bloemer and Lemmink 1992; Ka sper 1988; Garfein 1987; LaBarbera and Mazursky 1983; Kraft et al. 1973; Newman a nd Werbel 1973) has also implied the role of brand credibility by arguing the significant relationship between brand satisfaction and loyalty. Given the definition of brand satisfa ction as Â“the outcom e of the subjective evaluation that the chosen alternative meet s or exceeds the expectationsÂ” (Engel et al. 1990, p. 481), brand credibility would be ab le to subsume the concept of brand satisfaction because brand credibility is seen as affecting the expected utility, which can be interpreted as satisf action (Oliver 1997, see p. 78). Accordingly, the current study considers br and credibility as a signaling construct, which can explain various reasons for brand lo yalty formation, and thus includes it in the model as an antecedent of both cognitive and affective brand conviction. In other words, this study proposes that consumers form a cognitive and affective conviction toward a brand based on the brand credibility, which signa ls various expectations of utility from the brand purchase. As previously noted, a preponderance of the consumer psychology literature views the consumer purchase decision as both a c ognitive and emotional process (e.g., Morris et al. 2002; Bodur et al. 2000; Allen et al. 1992; Brown and Stayman 1992; Edell and Burke 1987), and it can be also theorized that bra nd credibility would also affect, or be influenced by, these two dimensions. Current research has included these two dimensions in the stage of conviction, and theorized that they are influenced by brand credibility because credibility, by definition, embraces th e personal history of brand experiences,
30 and thus can be seen as the initiator of the brand loyalty formation process. For example, if the impetus for loyal purchasing were emoti onal (e.g., looking cool in a new sports car), the loyalty formation would be processed mo re according to emoti on. If the reason for loyal purchasing were more cognitive (e.g., fuel efficiency of a new economy car), then the brand loyalty would be more cognitively based. In many cases, brand loyalty can be formed both cognitively and emotionally. For instance, VolvoÂ’s brand credibility, assuming it is mostly based on the perception of safety and reliability, can influence brand loyalty formation either through cogni tive conviction (e.g., for consumers who care about performance) or affective conviction (e.g., for people who care about the safety of their family). The final two hypotheses are stated as follows, and are based on the previous discussions. H9: Increases in brand credibility lead to increased cognitive brand conviction. H10: Increases in brand cr edibility lead to increased affective brand conviction. Relationships between Brand Convictions and Brand Credibility Various models of emotional response propose different relationships between emotion and cognition. Holbrook and OÂ’Sh aughnessy (1984) espouse a model based on the traditional consumer behavior paradigm, C-A-B, in which cognition determines affect, which leads to behavior. They theorize that a cognitive appraisal occurs in response to a stimulus, which then leads to an evaluation of the stimulus. The evaluation is followed by physiological changes and, finally, to subjecti ve feelings. Finally, a cognitive label is attached to the physiological change. Theref ore, cognition plays two important roles in this model of emotions: appraisal and attr ibution (Holbrook and OÂ’Shaughnessy 1984). Zajonc (1980) contested this model, arguing that emotion may precede and be entirely separate from cognition. While he acknowledged the necessary role of cognition
31 as a simple recognition of the stimulus, he stressed the importance of emotion. In this vein, a model proposed by Edell and Burke (198 7) assigns feelings a central role in responses to advertising. Feelings were shown as contributin g to the effects of advertising, which dispelled the previously held noti on that response to ad vertising was purely cognitive, such as semantic judgment of th e ad or beliefs about the brand (Edell and Burke 1987). The cognitive and affective convictions included in the current research are somewhat different from the pure cognitive and affective Â“responsesÂ” to stimuli (e.g., ad or brand) discussed above. Although conviction constructs ha ve the same orientation of being cognitive or affective, they are not si mple (spurious) responses to stimuli but are much deeper and more certain responses based on past consumer experiences. When cognitive and affective convictions share th e same past experience with respect to a specific brand, they may relate to each ot her, as well as to brand credibility. For the initial part of the model, th e present research follows Holbrook and OÂ’Shaughnessy (1984) by linking brand credib ility and affective conviction, with the cognitive label (i.e., brand credibility) at tached to physiological change (affective conviction). A question then occu rs as to whether affective and cognitive convictions are independent of each other. Following previous ly discussed theories (Bower and Forgas 2001; Niedenthal and Halberstadt 2000; Zajonc 2000; Forgas 1981; Pervin 1976; Magnusson 1971) that assert th e possible independent and pr eceding role of emotion in overall attitude formation, the present study examined wh ether there was a significant linkage from affective convicti on to cognitive conviction, as well as to attitude strength. The path to cognitive conviction tests the pre ceding role of emotion, and the path to the
32 attitude strength examines the independent influence of emotion on attitude formation. Another hypothesis based on this is: H8: Increases in affective conviction l ead to increased c ognitive conviction. Moderating Roles of Involvement and Product Type Since the proposed model primarily focuses on cognitive and affective brand convictions, it is necessary to examine speci fic conditions in which consumers take one type of conviction more strongly than the other. This investig ation can give more generalizability and robustness to the current studyÂ’s results by providing inferences relating to specific situations in which the hypotheses might work differently. In order to precisely investigate and compare the differential roles of cognitive and affective conviction, this study compares the brand loya lty formation process across two types of products, one hedonic and one func tional in nature. Research into these two types of products has often been conducted because at titudes for different types of products (especially hedonic v. functional) are known to be processed differently by consumers (Kempf and Smith 1998; Hoch and Ha 1986) . Hedonic products are those consumed primarily for affective or sensory gratifica tion purposes, while functional products deliver more cognitively-oriented benefits (Woods 1960 ). In an experiment of product trial attitude formation, Kempf (1999) showed th at functional product trial evaluation is influenced by both emotional and cognitive responses, while hedonic product evaluation is influenced by only emotional response. Since only emotion was found to be influencing attitude formation in both product ty pes, this finding seems to be consistent with several recent studies which have s hown that, when compared with cognitions, emotions are more predictive of the number and valence of peopleÂ’s thoughts toward objects under most conditions (Pham et al. 2001), and that emotion has a direct,
33 independent, stronger and more significant e ffect on attitude than does cognition (e.g., Morris et al. 2002, Bodur et al. 2000, Edell and Burke 1987). In addition, other studies have suggested involvement as a mode rator of emotional and cognitive processing. Batr a and Stephens (1994) argue th at the significant role of emotional and cognitive responses play in shaping brand attitude s differ according to various involvement conditions. They sugge st that affective responses are more important as determinants of brand attitudes in low involvement situations than in high involvement situations. Greenwald and Leav itt (1984) have also argued that cognitiveresponse-based persuasion effects will dominat e the affective-res ponse-based persuasion in high involvement situations. Involvement has been a frequently studied concept in the consumer behavior literature because it can be an important moderator of consumer behavior, which can fundamentally influence the consumersÂ’ evaluation processes on certain objects (Vakratsas and Ambler 1999; Mitchell 1981). Objects studied frequently are, for example, message (Petty and Caciopp o 1981; Krugman 1966), product (Bloch 1981; Lastovicka 1979; Robertson 1976; Bowen and Chaffee 1974), and situation (Mitchell 1979). Among the several involvement objects studied, the behavior al or situational involvement such as purchase-decision invol vement (Mittal 1989; Slama and Tashchian 1985; Laurent and Kapferer 1985) explains re lationships between consumer involvement and actual behaviors. For the current rese arch, the purchase decision involvement (PDI) scale is selected for determining the level of consumer involvement with regards to a specific product purchase, because the nature of brand loyalty formation would be better related to, or more influenced by, the purch ase situation of the product than the product
34 itself. PDI is defined as the extent of interest and concern that a cons umer brings to bear upon a purchase-decision task (Mittal 1989). Mi ttal clearly differen tiates the PDI from other involvements, e.g. product-involvement. He argues that a product class (such as coffee, salt, or bread) can be important to a consumer (thus indicating high product involvement), but the consumer may be indi fferent to the choice of a brand, thus implying a low PDI. However, a routine deci sion process (such as cigarette purchase) does not always indicate a low PDI, particularly if the consumer is not indifferent as to which brand is purchased. The present research explores the diffe rential roles of cognitive and affective conviction in brand loyalty formation, across these two types of product (i.e., hedonic and functional products) and two distinct involve ments (i.e., high PDI products and low PDI products). Because there are conflicting theories purporti ng that (1) cognition mainly drives attitude over emotion (e.g., Fishbein and Middlestadt 1995) or (2) emotions are primary driver of attitude (e.g., Morris et al. 2002; Pham et al. 2001; Bodur et al. 2000; Edell and Burke 1987), the current study investig ates which stream of theory would be more suitable in explaining the brand loyalty fo rmation process. Based on this construct, two research questions ar e generated as follows: RQ1: How differently would the affectiv e and cognitive convictions influence brand loyalty formation in di fferent involvement conditions? RQ2: How differently would the affectiv e and cognitive convictions influence brand loyalty formation fo r different product types? Proposed Model Based on the previous discussions, a convict ion model of brand loyalty is proposed in Figure 2-1. The model theorizes that true br and loyalty, which is different from inertia,
35 is a consequence of brand commitment (i .e., a loyal intention), and that brand commitment is a consequence of a strongly he ld positive attitude toward a brand (i.e., attitude strength) together with the cognitive/affective br and convictions, which can be significantly explained by bra nd credibility. The ke y insight of the model is that loyal intention can be labeled as tr uly loyal only when held with a high degree of conviction and attitude strength. Figure 2-1. Proposed Model Attitude Strength Cog. Conviction Brand Commit. True Brand Loyalty Aff. Conviction H1 H6 H5 H4 H3 H2 Brand Credibility H10 H9 H8
36 CHAPTER 3 METHODOLOGY Research Design Overview AbelsonÂ’s conviction items were initia lly tested for the purpose of an appropriateness check under the context of br and loyalty. After the check, a confirmatory factor analysis was conducted to see if th e two-factor solution (i.e., ego preoccupation and cognitive elaboration) of cognitive convi ction construct would be valid. For the main study, two products for each survey cell (i.e., LI-H, HI-H, LI-F, and HI-F), totaling eight products were selected from RatchfordÂ’ s (1987) FCB-grid, which classifies many products in a two-dimensional space, i.e., Think/Feel and Involvement (See Ratchford 1987, for a detailed review of the model). The primary study then surveyed subject responses to each construct in two involve ment (high vs. low) and two product type (hedonic vs. functional) c onditions. Subjects responded to the questionnaire, giving consideration to their own loyalty to brands in the given produc t categories. Finally, Structural Equation Modeling was used for the hypotheses testing and model selection/validation. Sampling Sampling criteria was limited to that of being a college student. Though this may limit the generalizability of the study to the total populat ion, choosing a relatively homogeneous group allows for a more controlled research sample that is consistent from pretests to main study, and more generalizable results for a specific age group, one that is among the most important target gr oups for many product categories.
37 Pretests Pretest 1: Validation of Conviction Scale AbelsonÂ’s conviction scale was examined by investigating (1) the face/content validity and (2) convergent/discr iminant validity of the meas ures for the domain of the cognitive brand conviction construct. The f ace validity procedure purified the original conviction items in the context of brand conviction, and the convergent/discriminant validity procedure tested the feasibility of two-factor so lution (i.e., ego-preoccupation, cognitive elaboration) for the brand conviction construct. For face validity, this study investigated the thoroughness with which the construct to be scaled and the extent to which the s cale items represent the constructÂ’s domain as Parasuraman et al. (1988) suggested. Thus , the face validity survey contained two questions for each item: Â“Does this item re present the domain of brand conviction? (DoesnÂ’t represent Â– Represents)Â” and Â“How n ecessary is it to use this item to measure brand conviction? (Not really necessary Â– Highly necessary).Â” The examination of face validity was conducted in two st ages. First, five locally av ailable consumer research experts, university professors of marke ting, psychology, and advert ising related fields, assessed the initial face validity. Following this first purification procedure, a web-based questionnaire was distributed through ACRlistserv, a listserv at the Association for Consumer Research. In additi on to the validity questions about representativeness and necessity for each conviction item, the que stionnaire included a ringer question that should be assessed as a Â“not representingÂ” and Â“unnecessaryÂ” item. The ringer item was: Â“When I think about my favorite brand of __, I feel fearful of death.Â” Furthermore, the questionnaire surveyed the level of educati on, research expertise, occupation, and years in the discipline for each respondent. Answers to these series of ques tions were used to
38 assess the academic expertise of respondents. After the surve y, the inter-judge reliability (e.g., Holbrook and Batra 1987; Holbrook a nd Lehmann 1980) was examined. This procedure assessed the homogeneity of face validity agreement across judges, by regarding each judgeÂ’s response as an Â“ite mÂ” in a series of multi-judge scales. The coefficient alpha test was conducted acr oss judges for each itemÂ’s mean of representativeness and necessity to the brand convicti on construct. Upon confirming the reliability, non-representative items that had relatively low mean scores were deleted. Subsequent to the verification of face validity, the convergen t and discriminant validity of the cognitive brand conviction construct was examined in the main study. In addition, a confirmatory factor analysis was conduc ted to determine if AbelsonÂ’s emotional commitment items could be merged with the ego-preoccupation construct, thus testing whether the proposed two-factor soluti on (i.e., ego-preoccupation and cognitive elaboration) was valid. Pretest 2: Product Type Manipulation Pre-check Fifty-five college students, homogeneous in demographics with those from the main survey sample, participated in the pr oduct category differentiation task for several product categories (categorizati on is based on Ratchford 1987, considered as representing the four product charact eristics, i.e. HI-H, L I-H, HI-F, and LI-F). Th e products, selected based on pretest 2, were designer sunglasses and high fashion watch for the HI-H cell; donuts and soft drinks for the LI-H cell; au to-insurance and black-a nd-white laser printer for the HI-F cell; and non-rechargeable AAA batteries and paper towels for LI-F cell. Details about the product selection procedure ar e discussed in the re sults section. Product category differentiation items were adopted from MittalÂ’s (1989) Purchase Decision
39 Involvement scale and KempfÂ’s (1999) hedoni c/functional perception question (Table 31). Table 3-1. Involvement scale and He donic/Functional perception question Purchase Decision Involvement (Modified from Mittal 1989) I care a great deal in selecting this pr oduct from many other choices available in the market. (1-9 scale). It is important to me to make a ri ght choice of this product. (1-9 scale). I am concerned about the outcome of my choice in making my selection of this product. (1-9 scale). Hedonic/Functional percep tion question (Kemp 1999) Would you characterize this product as primarily a functional product or a product for pleasure? (1) Â“primarily for functional useÂ” Â– (2) Â“primarily for pleasureÂ” (1-9 scale). Main Survey Sample Four hundred and seventy-six college student s participated in the main survey. The main survey was Web-based, with subjects visiting a questionnaire site and completing one of four randomly allocated questionnaires . Each questionnaire covered two products within the same research c ondition (e.g., high involvement and hedonic). Therefore, the total number of observations was 952 (476 2), because each condition surveyed two products. A Random Link Generato r was used for the randomiza tion of research cell (i.e., questionnaire) distribution. For the purpos e of structural equation modeling, the observation size (i.e., 952) exceeds the minimum sample size of 100 to 150, for the Maximum Likelihood Estimation method (Ding et al. 1995). Measurement Instruments True brand loyalty was measured using th e brand loyalty scale suggested by Odin et al. (2001), which combines repeat purchas ing behavior and brand sensitivity (Bloemer and Kasper 1995; Kapferer and Laurent 1983) . Accordingly, repeat purchasing of the
40 same brand based on the perceived importance of brand choice are operationalized as truly brand loyal behavior. Odin et al. used four items of repeat purchasing behavior and one item (#1 in Table 3-2) of brand sensitiv ity as shown in Table 3-2. Although the four repeat purchasing items are plausible, the curr ent study did not use the first scale item of Odin et al. because it does not measure the act ual loyal behavior but the intention to be loyal, which is viewed as brand commitment in this study. In addition to the one-item brand sensitivity scale (first row in Table 3-2) by Odin et al., MittalÂ’s (1989) percei ved brand similarity item (s econd row in Table 3-2) is selected following Kapferer a nd LaurentÂ’s (1983) assertion that the beliefs in differences between brands is a major determinant of brand sensitivity. Table 3-2. True Brand Loyalty Scale * RP item #1 is excluded from the true loyalty scale. In order to reflect the true loyalty cons truct, the interaction terms of repeat purchasing behavior (RP) and brand sensitivity (BS) were used as observed variables. Based on this construct, six operationaliz ed observed variables are created: RP1 BS1, RP1 BS2, RP2 BS1, RP2 BS2, RP3 BS1, and RP3 BS2. Such multiplication reflects the moderating role of brand sensitivity in the formation of brand loyalty. Because of these multiplications, error variances of items that share the same component, either RP or BS items, were correlated. C onfirmatory specification yielded nine error Brand Sensitivity* 1. The brand name is the first thing I'm looking at for the purchase of this product category. 2. Various brand names of this product available in the market are: all very alike /all very different Repeat Purchasing Behavior (RP) 1. During my next purchase, I will buy the same brand of __ as the last time. 2. I am loyal to only one brand of __. 3. I always buy the same brand of __. 4. Usually, I buy the same brand of __.
41 covariances of items within the true brand loyalty construct. They included: three RPbased error covariances (i.e., RP1, RP2, and RP3) and six BS-based error covariances (i.e., (BS1 three RPÂ’s) + (BS2 three RPÂ’s)). Brand commitment was measured with thr ee items in total: the item #1 (repeat purchasing behavior item) by Odin et al . (2001), and Knox and WalkerÂ’s (2001) two items, which combine TraylorÂ’s (1981) perceived brand commitment item and CunninghamÂ’s (1967) connotative brand co mmitment item (Table 3-3). Knox and WalkerÂ’s scale is selected because it e xpresses the psychological construct of commitment set in a behavior ally intentional context. Table 3-3. Brand Commitment Scales During my next purchase, I will buy th e same brand of __ as the last time. When buying __, how committed are you to buying your most favorite brand, rather than an alternative brand? If you could not get your most favorite brand of __ at the store you had gone to for them, would you: (1) Happily buy a different brand, (2) Reluctantly buy a different brand, (3) Not buy the product unt il the next time you shopped, (4) Try a different shop, (5) Keep trying different shops until you got the brand you wanted. Attitude strength was measured through the five dimensions recommended by Krosnick and Abelson (1992), as previously discussed. First, the operationalization suggested by Downing et al . (1992) was adopted for the measure of extremity, by calculating an absolute deviation score from the midpoint of a 10-point attitude scale anchored at -5 (Â“Very negativeÂ”) and +5 (Â“Very positiveÂ”), without a neutral point. Although excluding the neutral point may cause different effects on the responses, a recent study (Nowlis et al. 2002) shows that ex clusion of a neutral point from attitude scale can best reflect the underlying attitudes when the goal of scale is trying to determine respondentsÂ’ committed (or forced) futu re decisions such as voting in the near future. In a similar vein, the current study ex cludes the neutral point from the extremity
42 measure because it is trying to determine th e attitudinal extremity based on respondentsÂ’ past decisions (repeated purchases). Second, an attitude intensity question aske d respondents how strong or intense their feelings are toward their loyal brand in a given product category (1-9 scale). Third, attitude certainty was assessed using Fazio a nd ZannaÂ’s (1978) scale. The question asked how certain the subjects feel about their attitude toward their loyal brand in a given product category (1-9 scale). Fourth, attitude importance was measured using the scale suggested by Krosnick et al. (1993) that asks how personally important an object is to the respondent. Finally, knowledge was measured with self-reporte d knowledgeability. The literature shows that knowledge has generally been measur ed in three ways: knowledge listing, quizzes, and self-perceptions. K nowledge listing is a method of asking respondents open-ended questions that ask for a list of characteristic s and facts relevant to the object (e.g., Wood 1982). Quizzes are al so used for testing the real amount and correctness of respondent knowledge (e.g., Wilson et al. 1989). The self-perception method (e.g., Davidson et al. 1985) asks responden ts to gage the level of knowledge they feel they have about an object. The current study adopted this self-reporting method due to its time and cost efficiency. The five que stions related to the attitude strength are presented in Table 3-4. Table 3-4. Attitude Strength Scale My attitude to my favorite brand is : -5 (very negative)/+5 (very positive) How strong or intense is your feeling to ward your favorite brand in this product category? 1 (Not very intense)/9 (Very intense) How certain do you feel about your attitude toward your favorite brand in this product category? 1 (Not very certain)/9 (Very certain) How important would you say your favorite br and is to you personally? 1 (Not very important)/9 (Very important) How knowledgeable do you feel you are a bout your favorite brand? 1 (Not very knowledgeable)/9 (Very knowledgeable)
43 The cognitive conviction scale was derived from Abelson (1987), but, as discussed, only the domain-validated (i.e., brand loyalty) items was used. Details about the selected items are discussed late r in results section. For measuring of affective brand convict ion, the pictorial measure AdSAM (Figure 3-1, Morris et al. 2002), de veloped from LangÂ’s (1980) Sel f-Assessment Manikin (SAM), is used in the present study. SAM is a graphic char acter that reflects the PAD theory of affective response. This theory describes the full spectrum of huma n emotions in three independent bipolar dimensi ons. In this process, a ll emotional responses are combinations, in varying degrees, of these th ree basic emotions (Russell and Mehrabian l977). Evidence shows that three indepe ndent, bipolar dimensions reliably and sufficiently define all emo tional states (Mehrabian and Russell 1974). These three dimensions are: P (pleasure/displeas ure), A (arousal/non-arousal), and D (dominance/submissiveness). Pleasure/displ easure ranges from extreme happiness to extreme unhappiness. Arousal/non-arousal constitutes a physiological continuum connoting a level of physical act ivity, mental alertness or frenzied excitement on one extreme; inactivity, mental lethargy, or sleep on the other. Dominance/submissiveness refers to a feeling of power, control or infl uence, versus the inability to influence a situation or a feeling of lack of control. Subjects use the PAD scal es to report how they feel (Mehrabian a nd Wetter l987). Because the three dimensional PAD approach is capable of characterizing diverse emotional responses in consumption situati ons (Holbrook and Batra l988; Mehrabian and Russell 1974) it was used in the present st udy. Verbal emotional response measures however, are difficult to employ in consumer research. When adjective checklists or
44 semantic differential scales are used to asse ss emotional response, the precise meaning of the emotional words may vary from person to person. For example, joy or anger may mean one emotion to one person, but something slightly different to someone else. This may vary the outcome of the subject's real em otional response. Also problematic are the use of open-ended questions that request res pondents describe their emotional responses to the advertisements (Stout and Leckenby l 986; Stout and Rust l986) . Both approaches require a significant amount of cognitive processing. In contrast, the non-verbal measure, SA M, eliminates the cognitive processing associated with verbal measures (Edell and Burke 1987) and is quick and simple to use (Morris and Waine l994; Lang l980). Correla tions of .937 for pleasure, .938 for arousal, and .660 for dominance were found between ratings generated by SAM and by the semantic differential scales used by Mehr abian and Russell (Morris and Waine, l994; Morris et al. l993; Morris et al. 1992). SA M uses a nine-point scale for each of the dimensions. On each of the three scales, respondents were required to mark the dot below the manikin or between the manikins that best represent their feelings after seeing the advertisement. (See Figure 3-1). In addition, since the AdSAM measures the a ffective responses (or state) to a brand, each P/A/D dimension was multiplied by its own emotional certainty levels (Tiedens and Linton 2001), rated by each subject, and measured for each dimension, in order to operationalize the Â“convictionalÂ” aspect of the construct.
45 Pleasure Arousal Dominance Figure 3-1. SAM (Self-Assessment Mannequin) Lastly, brand credibility was measured by Erdem and SwaitÂ’s (1998) scale, which examines trustworthiness and expertise of the brand. Specific items are listed in Table 3-5. Table 3-5. Brand Credibility Scale This brand delivers what it promises. This brandÂ’s product claims are believable. You just canÂ’t believe what the ads say about this brand. My experiences with this brand make me wary of their claims. This brand has a name you can trust. This brand is at the forefront of usi ng technology to deliver a better product. This brand reminds of someone whoÂ’s co mpetent and knows what he/she is doing.
46 CHAPTER 4 RESULTS Pretest Results Pretest 1 Scholars (N=5) specializing in consumer research fields at a large southern university initially evaluated th e face validity of AbelsonÂ’s conviction scale for domain of brand loyalty. They all agreed to exclude it em #3 from the emotional commitment items, item #5 from the ego preoccupation items, and item # 2 and #3 from the cognitive elaboration items (see Table 4-1). Subseque nt to the agreed upon exclusions, the remaining conviction items were evaluated via members of ACR-listserv. There were a total 22 respondents to the ACR-listserv surv ey. Utilizing adequate expertise selection criteria (i.e., correct answer for the ringer question, consumer beha vior related research expertise, doctoral degree, over 3 years in the occupation), answ ers from 19 of the 22 respondents were used in analyzing face vali dity of conviction items. All 19 respondents were university professors w ith over 3-years of experienced and a doctoral degree in consumer research related fields. Inter-judge reliability was tested using Cronbach alpha. The alpha coefficient from 19 items (i.e., judges) and 11 cases (i.e., scal e items) was .95 which exceeds NunnallyÂ’s (1978) suggestion of .80. Having attained in ter-judge reliability, descriptive statistics were then analyzed. The results (Table 4-1) showed considerably low mean scores for Q3 (item #4 of emotional commitment) and Q9 (the ringer question). Although Q8 (item #4 of ego preoccupation) was slightly below 5 ( on a 1 to 9 scale), the item was included in
47 the final set of select items because the mean score was not exceedingly low. Details concerning the included and excluded items are presented in Table 4-2. Table 4-1. Descriptive Statistics of Pretest 1 Items N Mean Std. Deviation Emotional Commitment item #1 18 5.19 3.18 Emotional Commitment item #2 19 6.55 2.47 Emotional Commitment item #4 19 2.76 2.13 Emotional Commitment item #5 19 6.36 2.47 Ego Preoccupation item #1 19 5.28 2.25 Ego Preoccupation item #2 19 7.78 1.41 Ego Preoccupation item #3 19 6.50 2.29 Ego Preoccupation item #4 19 4.60 2.67 Ringer question 19 1.28 .76 Cognitive Elaboration item #1 19 5.15 2.28 Cognitive Elaboration item #4 19 6.68 1.85 Cognitive Elaboration item #5 19 5.18 2.65 Table 4-2. Purification of Brand Conviction Items based on Pretest 1 Emotional Commitment 1. My beliefs about my favorite br and express the real me. (Included) 2. I can't imagine ever changing my mind to a different brand. (Included) 3. My beliefs are based on the moral sense of the way things s hould be. (Excluded) 4. I would be willing to spend a day a month working for a group supporting my views about my favorite brand. (Excluded) 5. I think my view about my favorite br and is absolutely correct. (Included) Ego Preoccupation 1. I think about my favorit e brand often. (Included) 2. I hold my views about my favorite brand very strongly. (Included) 3. My belief about my favorite br and is important to me. (Included) 4. I am extremely concerned about my favorite brand. (Included) 5. When I think about the issue, I feel fearful. (Excluded) Cognitive Elaboration 1. I've held my views about my favorite brand a long time compared to most people. (Included) 2. Several other issues could come up in a conversation about my favorite brand. (Excluded) 3. Several things could happen if my vi ews about my favorite brand were enacted. (Excluded) 4. I have more knowledge on my favorite brand than the average person. (Included) 5. It's easy to explain my views about my favorite brand. (Included)
48 Pretest 2 A total of forty-eight college students participated in the product category manipulation check. The initia l set of product categories in corporated from Ratchfold (1987) were designer sunglasses and high fash ion watch for cell HI-H (high involvement and hedonic), donuts and soft drinks for cell LI-H (low involvement and hedonic), autoinsurance and 35mm camera for cell HI-F (h igh involvement and functional), and regular hair shampoo and paper towels for cell LI-F (low involvement and functional). During the initial stage of the pretest, with 20 respondents, the survey showed a strong pattern of unexpected responses for 35mm camera a nd regular hair sh ampoo. Although 35mm camera and regular hair shampoo were expect ed to be functional and low-involvement respectively, they were oppos itely viewed by respondents: 35mm cameras were viewed as hedonic (M=7.5, N=12), and shampoo was vi ewed as a moderately high involving product (M=6.74, N=12). Although unexpected, th is result is not surprising, because 35mm camera can be perceived as hedonic as it is usually used to capture pleasurable (i.e., hedonic) moments. Also, a regular hair shampoo might not register as low involvement because respondents (female co llege students in particular) may give exceptional care to their hair. Personal intervie ws with some of the respondents provided such insights. After the unexpected results, alternatives for camera and shampoo were immediately explored. The black-and-white lase r printer was selected as an alternative product for 35mm camera, while bar soap, floppy diskette, and a non-rechargeable AAA battery were selected as a lternatives for hair shampoo. A priori pretest with these products reflected that only black-and-white laser printe r (M= 2.4 (functional), N=13) and battery (M=3.7 (low involvement), N=13) were perceived as expected (i.e., HI-F and LI-F respectively), but bar soap and floppy di skette were unsatisfactory. Pretest 2 was
49 then resumed with the laser printer and a battery as repl acements for 35mm camera and regular hair shampoo. An additional 23 respondents participated in the pretest. Analysis with the final sample size of 35 (i.e., initial 12 + later 23) produced successful manipulations. Since the homogeneity of variances assumption was not met (Levene = 8.28, P<.05), Dunnett T3 multiple comparison method was used for the co mparisons of each pair of products. Test results are presented in Tables 4-3 through 4-6. Table 4-3. Descriptive Statistic s of Hedonic/Functional Perception Products N Mean Std. Deviation Designer sunglasses (H) 35 5.43 2.02 Fashion watch (H) 35 5.40 2.10 Donut (H) 35 6.69 2.45 Soft drink (H) 35 5.51 2.34 Insurance (F) 35 1.29 1.20 B/W laser printer (F) 35 2.66 1.59 Paper towel (F) 35 1.54 1.40 Battery (F) 35 2.31 1.43 Table 4-4. Descriptive St atistics of Involvement Products N Mean Std. Deviation Designer sunglasses (H) 35 7.39 1.10 Fashion watch (H) 35 7.64 0.90 Insurance (H) 35 8.22 1.16 B/W laser printer (H) 35 7.45 1.46 Donut (L) 35 4.63 2.02 Soft drink (L) 35 5.46 1.84 Paper towel (L) 35 4.10 2.03 Battery (L) 35 4.17 1.84
50 Table 4-5. Dunnett T3 Multiple Comparison Test Results (Hedonic vs. Functional) Designer sunglasses Fashion watch InsuranceLaser Printer Donut Soft drink Paper towel Battery Designer Sunglasses 0 .03 (.49) -1.26 (.54) -.09 (.52) 4.14* (.40) 2.77* (.43) 3.89* (.42) 3.11* (.42) Fashion Watch -.03 (.49) 0 -1.29 (.55) -.11 (.53) 4.11* (.41) 2.74* (.45) 3.86* (.43) 3.09* (.43) Insurance 1.26 (.54) 1.29 (.55) 0 1.17 (.57) 5.40* (.46) 4.03* (.49) 5.14* (.48) 4.37* (.48) Laser Printer .09 (.52) .11 (.53) -1.17 (.57) 0 4.23* (.45) 2.86* (.48) 3.97* (.46) 3.20* (.46) Donut -4.14* (.40) -4.11* (.41) -5.40* (.46) -4.23* (.45) 0 -1.37 (.34) -.26 (.31) -1.03 (.32) Soft drink -2.77* (.43) -2.74* (.45) -4.03* (.49) -2.86* (.48) 1.37* (.34) 0 1.11 (.36) .34 (.36) Paper towel -3.89* (.42) -3.86* (.43) -5.14* (.48) -3.97* (.46) .26 (.31) -1.11 (.36) 0 -.77 (.34) Battery -3.11* (.42) -3.09* (.43) -4.37* (.48) -3.20* (.46) 1.03 (.32) -.34 (.36) .77 (.34) 0 Mean differences are shown. Standard errors are in parentheses. *: significant at .05 level. Table 4-6. Dunnett T3 Multiple Comparison Test Results (Purchase Decision Involvement) Designer sunglasses Fashion watch InsuranceLaser Printer Donut Soft drink Paper towel Battery Designer Sunglasses 0 -.25 (.24) -.83 (.27) -.06 (.31) 2.76* (.39) 1.93* (.36) 3.29* (.39) 3.22* (.36) Fashion Watch .25 (.24) 0 -.58 (.25) .19 (.29) 3.01* (.37) 2.18* (.35) 3.53* (.38) 3.47* (.35) Insurance .83 (.27) .58 (.25) 0 .77 (.31) 3.59* (.39) 2.76* (.37) 4.11* (.40) 4.05* (.37) Laser Printer .06 (.31) -.19 (.29) -77 (.31) 0 2.82* (.42) 1.99* (.40) 3.34* (.42) 3.28* (.40) Donut -2.76* (.39) -3.01* (.37) -3.59* (.39) -2.82* (.42) 0 -.83 (.46) .52 (.49) .46 (.46) Soft drink -1.93* (.36) -2.18* (.35) -2.76* (.37) -1.99* (.40) .83 (.46)0 1.35 (.46) 1.29 (.44) Paper towel -3.29* (.39) -3.53* (.38) -4.11* (.40) -3.34* (.42) -.52 (.49) -1.35 (.46) 0 -.07 (.46) Battery -3.22* (.36) -3.47* (.35) -4.05* (.37) -3.28* (.40) -.46 (.46) -1.29 (.44) .07 (.46) 0 Mean differences are shown. Standard errors are in parentheses. *: significant at .05 level.
51 Main Study: Model Estimation and Results A structural equation modeli ng procedure, AMOS 4 (e.g., Gruen et al. 2000; Hox 1995; Arbuckle 1989 and 1994), was used to test the proposed model. The model was examined in three stages. First, the reliab ility and validity of the constructs were examined. Second, the overall fit of the m odel to the data was tested. Third, the measurement and structural parameters were examined to determine if the data supported the proposed hypotheses. During the second and third stages, comparisons of alternative models were conducted. All da ta from four research cells was combined and used for most model development. Upon determining the best-fit model most suited to the combined data, the data was divided into ce rtain manipulation conditi ons for the research question test. Descriptive Statistics of Measurements The descriptive results (Appendix A), with scale items used in the subsequent CFA (Table 4-7), provide a summary of variables that are important in subsequent analyses. In addition, a correlation matrix table (Appendix B) shows that items measuring the same overall construct, e.g., affective conviction items, are more highly correlated with each other than with any of the ot her items. Some items (e.g., brand commitment items) that indicated possible confounding w ith other items (e.g., true brand loyalty) were later analyzed for their convergent and discriminant validity in a CFA. Results of the CFA are discussed later.
52 Table 4-7. Details of Scales Used for CFA Original Scale for Each Construct Origin Adoption for this study (Cronbach ) True Brand Loyalty (TBL) Production of each RP and BS item. ( = .95) Repeated Purchase (RP) Odin et al. (2001) Item#1: RP1*BS1 1. I am loyal to only one brand of __. Item#2: RP1*BS2 2. I always buy the same brand of __. Item#3: RP2*BS1 3. Usually, I buy the same brand of __. Item#4: RP2*BS2 Brand Sensitivity (BS) Item#5: RP3*BS1 1. The brand name is the first thing I'm looking at for the purchase of this product category. Odin et al. (2001) Item#6: RP3*BS2 2. Various brand names of this product available in the market are: all very alike /all very different Kapferer and Laurent (1983) Brand Commitment (BCM) Original scale is used. ( = .61) 1. During my next purchase, I will buy the same brand of __ as the last time. Odin et al. (2001) 2. When buying __, how committed are you to buying your most favorite brand, rather than an alternative brand? Knox and Walker (2001) 3. If you could not get your most favorite brand of __ at the store you had gone to for them, would you: (1) Happily buy a different brand, (2) Reluctantly buy a different brand, (3) Not buy the product until the next time you shopped, (4) Try a different shop, (5) Keep trying different shops until you got the brand you wanted. Knox and Walker (2001)
53 Table 4-7. Continued Attitude Strength (AST) Original scale is used. ( = .83) 1. My attitude to my favorite brand is: -5 (very negative)/+5 (very positive) Downing et al. (1992) 2. How strong or intense is your feeling toward your favorite brand in this product category? 1 (Not very intense)/9 (Very intense) Krosnick and Abelson (1992) 3. How certain do you feel about your attitude toward your favorite brand in this product category? 1 (Not very certain)/9 (Very certain) Fazio and Zanna (1978) 4. How important would you say your favorite brand is to you personally? 1 (Not very important)/9 (Very important) Krosnick et al. (1993) 5. How knowledgeable do you feel you are about your favorite brand? 1 (Not very knowledgeable)/9 (Very knowledgeable) Davidson et al. (1985) Affective Brand Conviction (ACV) Production of each AR and EC item. (EC is measured for each AR separately), ( = .66) Affective Responses (AR) Morris et al. (2002) Item#1: AR1*EC1 1. Ad SAM Pleasure Item#2: AR2*EC2 2. Ad SAM Arousal Item#3: AR3*EC3 3. Ad SAM Dominance Emotional Certainty (EC) Tiedens and Linton 2001 1. How certain are you about your feeling?
54 Table 4-7. Continued Cognitive Brand Conviction (CCV) Abelson (1987)(Validated) original scale is used. ( = .93) 1. My beliefs about my favorite brand express the real me. 2. I can't imagine ever changing my mind to a different brand. 3. I think my view about my favorite brand is absolutely correct. 4. I think about my favorite brand often. 5. I hold my views about my favorite brand very strongly. 6. My belief about my favorite brand is important to me. 7. I am extremely concerned about my favorite brand. 8. I've held my views about my favorite brand a long time compared to most people. 9. I have more knowledge on my favorite brand than the average person. 10. It's easy to explain my views about my favorite brand. Brand Credibility (BCR) Erdem and Swait (1998) Original scale is used. ( = .82) 1. This brand delivers what it promises. 2. This brandÂ’s product claims are believable. 3. You just canÂ’t believe what the ads say about this brand. 4. My experiences with this brand make me wary of their claims. 5. This brand has a name you can trust. 6. This brand is at the forefront of using technology to deliver a better product. 7. This brand reminds of someone whoÂ’s competent and knows what he/she is doing.
55 Assumption Check Prior to the main analysis, several unde rlying assumptions for structural equation modeling were checked. The underlying assu mptions for the SEM analysis are: an adequate variable-to-sample ratio, normality, linearity, no extreme multicollinearity, and sampling adequacy (Hair et al.1998). The vari able-to-sample ratio was 1 to 35.3, which satisfied the criteria sugges ted by Nunnally (1978). Kaiser -Meyer-OlkinÂ’s measure of sampling adequacy was well over .50 (i.e., .923), and BartlettÂ’s test of sphericity index showed significant p-value at p=.01 level. Normality assumption was considered satisfied because all Skewness and Kurtosis values as sociated with each item were within the range of Â±1.96 (-.86< all Skewness values<.91, -1.11
56 acceptable given NunnallyÂ’s (1978) minimum sugge stion of .60 being adequate for basic research. Before conducting validity tests on the full measurement model, a CFA of cognitive brand conviction was performed. Fi rst, discriminant validity was evaluated between each pair of emotional commitment (EC), ego-preoccupation (EP), and cognitive elaboration (CE). Discriminant validity was evaluated using an approach suggested by Joreskog (1971). The test assessed two estimated constructs by constraining the estimated correlation parameter between them to 1.0 a nd then performing a chi-square difference test on the values obtained for the constr ained and unconstrained models. Bagozzi and Phillips (1982) asserted that a significantly large chi-square value difference between the unconstrained and constrained correlation mode l indicates that the constructs are not perfectly correlated and that discriminant validity is achie ved. The significance of the chi-square statistic was asse ssed by comparison with a critic al chi-square value of 3.84 (df=1). The results supported the orig inal three factors of c onviction. The chi-square difference between EC and EP was 231.9, 382.1 for the EC-CE pair, and 212 for the EPCE pair. Because all difference values were well over 3.84, the results strongly implied the three-factor conviction scale remain ed valid in measuring brand conviction. Goodness-of-fit indices further uphel d the three-factor solution (Â² =291.1, df=32, GFI=.94, NFI=.95, CFI=.96, RMSEA=. 09, SRMR=.03). Although the proposed two factor solution (EP and CE only) also showed a very comparable result (Â² =353.6, df=34, GFI=.92, NFI=.94, CFI=.95, RMSEA=.09, SR MR=.03), the original three factor solution was used in the subsequent SEM analyses due to the observed di scriminant validity
57 between all pairs of factors. However, as previously discussed the label of emotional commitment was renamed due to the poten tial misrepresentation of emotion. In order to verify the misrepresentation, a principle component exploratory factor analysis with Varimax rotation method was conducted for emotional commitment items and PAD items of SAM. The analysis extracted a clear two-factor so lution for emotional commitment and PAD, thus implying that AbelsonÂ’s emotional commitment items were not really emotion related, and i ndicating the label is misnamed. After this result, careful thought was give n to the items of emotional commitment and they were given a new label, Â“self-br and identification.Â” Since the three items, originally from emotional commitment factor, measured the expression of real self (item #1), never changing the (self-constructed) mi nd (#2), and perceived absolute correctness of self view (#3), a new label that describe s consumersÂ’ perceived similarity between a brand and self would be plausible. Based on th is result, items of each construct (i.e., selfbrand identification, ego-preoccupation, and cognitive elaboration) were averaged and those averages were regarded as observed vari ables. As a result, the cognitive conviction construct had three observed indi cators (CronbachÂ’s alpha=.90). In the next step, discriminant and conv ergent validity were assessed for all constructs and items in the measurement mode l. The results for discriminant validity, which was measured by chi-square tests for one pair of constructs at a time, are displayed in Table 4-8. They indicate that all pairs have significant discriminant validity. Though the discriminant validity was not confirmed for the pair of cognitive conviction and brand credibility at .05 level, the chi-square differen ce value of 3.7 was very close to the critical value of 3.84 and it was significant at .10 level (critical value=2.7).
58 Convergent validity was assessed by Â“det ermining whether each indicatorÂ’s estimated pattern coefficient on its posited underlying construct f actor was significantÂ” (Anderson and Gerbing 1988, p. 416). The result s indicated that all items significantly loaded to the intended factors. All factor loadings between items and constructs were from .39 to .94 and significant at .01 level. The results showed that the items used a nd the constructs have both convergent and discriminant validity. Table 4-8. Discriminant Validity Tests Correlation Pairs of Latent Constructs Â²df=1 (critical value=3.84 at p=.05) Cog. Conviction Brand Credibility 3.7 Aff. Conviction Brand Credibility 332.4 Attitude Strength Brand Credibility 12.4 Brand Credibility Brand Commitment 18.7 True Brand Loyalty Brand Credibility 103.2 True Brand Loyalty Brand Commitment 413.9 Attitude Strength Brand Commitment 145.7 Attitude Strength Aff. Conviction 270.2 Aff. Conviction Cog. Conviction 108.6 Attitude Strength Cog. Conviction 155.2 Cog. Conviction Brand Commitment 65.3 True Brand Loyalty Cog. Conviction 318 Aff. Conviction Brand Commitment 115 True Brand Loyalty Aff. Conviction 82.4 Attitude Strength True Brand Loyalty 296.2 Confirmatory Factor Analysis (CFA) A confirmatory factor analys is was conducted on all items for all constructs with all combined data from all research cells. The results (Â² =2200.8, df=300, GFI=.82, NFI=.89, CFI=.91, RMSEA=.08, SRMR=.08) demonstrated marginally acceptable overall fit and indicated that the pr oposed measurement model might need respecifications for improvement. In order to investigate the possible respecifications, modification indices (MI) were examined . Modification indice s provided useful
59 information regarding the cova riance relationships between variables in the proposed model that could be altered, when theoretically justifiable, to improve the fit between the hypothesized model and the data. Select modi fication indices and their theoretically justifiable respecification results are presented in Table 4-9. Table 4-9. Select Modification I ndices and Respecification Results Modification Sequence (MS#) MI (Modification Indices) Â² after modification Â²df=1 after modification 1 BCR3 -BCR4123.39 2070.6 130.3 2 BCR6 -BCR782.75 1986.2 84.4 3 AST1 -AST359.44 1921.0 65.2 4 AST1 -AST455.20 1887.7 33.3 * AST= error variances of Attitude Strength items BCR= error variances of Brand Credibility items * Numbers after construct name are the item numbers. There are four basic ways to respecify indicators for a converged and proper solution with unacceptable overall fit (Anderson and Gerbing 1988): (1) relate the indicator to a different construc t, (2) delete the indicator from the model, (3) relate the indicator to multiple factors, or (4) use corr elated measurement errors. Initial observation of MIÂ’s suggests that the first and third opti ons were not applicable to the current model as there were no high MIÂ’s between indica tors and different c onstructs. Although the second option may be generally preferred because the fourth option can result in a loss of interpretability and theore tical meaningfulness (Bago zzi 1983; Fornell 1983), both options were considered because the sec ond option could also lead to a loss of contributable information in the model. In agreement with suggestions from several researchers (e.g., Anderson and Gerbing 1988; Bagozzi 1983; Forn ell 1983), theorybased justifications of resp ecification were sought for each MI. Consideration of theory in respecification greatly redu ces the number of alternate models to investigate (Young
60 1977) and reduces the possibility of taking advantage of sampling error to attain goodness of fit (Anderson and Gerbing 1988). The first noticeably high MI (MS#1) was between brand credibility item #3 (canÂ’t believe ads) and #4 (wary of claims). Becau se both measured trustworthiness of brand (Erdem and Swait 1998), error covariance fo r the two items was freed to estimate. Likewise, error covariance of brand credibility items #6 (u sing forefront technology) and #7 (competent and knows what it is doing) we re freed as both measured the perceived expertise of the brand (MS#2). The next high MIÂ’s were between items of attitude strength. Regarding a relatively high MI (59.44) between attitude strength items #1 (extremity) and #3 (certainty), literature indicates that they are independent but often conf ounded conceptually (Gross et al. 1995). Gross et al. observed that alt hough earlier researchers focused on the equivalence between extremity and certainty (e.g., Mehling 1959; Osgood et al. 1957; Suchman 1950; Cantril 1946), la ter researchers did not (e .g., Krosnick et al. 1993; Krosnick and Abelson 1992). A current view about the relationshi p between extremity and certainty is that certainty does not neces sarily imply extremity, while extremity can connote certainty (Gross et al . 1995). In following suggestio ns in the literature, both extremity and certainty items were kept (bec ause they are independent) in the model but their error covariance was freed to estimate (bec ause they often reflect each other in spite of difference) (MS#3). Another item that ha d a considerably high MI was importance. It implies that people may have a tendency to exhibit a more extreme attitude when a particular attitude is persona lly important to them . The literature suppor ts the theoretical link between extremity and importance. For ex ample, Cialdini et al. (1976) found that
61 important attitudes became more polarized when an individual discussed an issue with someone with whom they disagreed. Consiste nt with the literatur e, covariance between extremity and importance was then freed to estimate (MS#4). After the theory-based re specifications, the final measurement model revealed considerable model fit increases (Â² =1887.6, df=296, GFI=.86, NFI=.91, CFI=.93, RMSEA=.07, SRMR=.08). The final measurem ent model with factor loadings for each item and the construct correlations is presented in Figure 4-1.
62 Figure 4-1. Final Measurement Model Brand Cred. Cog. Conv. Aff. Conv. Attitude Str. Brand Com. T. B. Loyalty bcr4 bcr5 bcr6 bcr7 bcr1 bcr2 bcr3 ec1 ec2 ec3 ec4 ec5 ec6 ec7 pleasure arousal domin. ep ea ed cog. ela. ego-pre. s-b id intensity extremity certainty import. knwge bcm1 bcm2 tbl1 tbl2 tbl3 tbl4 tbl5 esa eep ece eas1 eas2 eas3 eas4 eas5 ebl1 ebl2 ebl3 ebl4 ebl5 ebc1 ebc2 tbl6 ebl6 .81 .61 .74 .74 .75 .90 .76 .90 .68 .85 .66 .85 .54 .53 .86 .90 .86 .85 .86 .49 .51 .52 .49 .85 .45 .70 .66 .63 .49 .92 .79 .70 .44 .87 .67 .66 .53 .36 .69 bcm3 ebc3.59 .71 .39 .27 .37 .36 .72 .73 .68 .51 .52 .47 .37 .30 .22 -.21
63 Structural Equation Modeling Overview of structur al modeling procedure The Two-Step Modeling approach re commended by Anderson and Gerbing (1988) was employed in order to determine whether the proposed structural model was the ideal model for use within the research framework of the present study. The first step is to compare the null mode l (Mn) to the saturated model (Ms; i.e., measurement model). This is normally done by a pseudo chi-square test (Bentler and Bonett 1980) in which a statistic is created from the chi-square value for Mn (the largest number of df for any structural model). A null model is defined as one in which all parameters of the constructs relating to one a nother are fixed at zero (i.e., no relationships among constructs). If significance were f ound, then it would mean a possibility of fundamental misspecification. The second step is to compare the theo retical model (Mt) to other, more constrained (Mc) or unconstrai ned (Mu), models. Thus Mc and Mu are respectively the alternative most likely constrained and unconstrained models from a theoretical perspective (Anderson and Gerb ing 1988). Therefore, the stru ctural sub-models to be compared are nested in a sequence such that Mn < Mc < Mt < Mu < Ms. In comparing those models, sequential chi-square differe nce tests (SCDTs) are usually employed if models are nested. Each set of SCDT is a test of a null hypothesis of no significance between two nested models. For example, an SCDT compares Mu-M s in order to assess the reasonableness of the structural cons traints imposed by Mu. If the null hypothesis associated with this test (i.e., Mu-Ms=0) we re supported, then the Mt-Mu would be tested. If this were also supported, the Mt-Mc w ould be examined. For each set of SCDT, the more constrained model would be tentatively accepted if there was no difference in chi-
64 square between the two models. A decisiontree framework of the SCDT provided by Anderson and Gerbing (1988) sugge sts testing the sets of mode ls in this sequence: Mt-Ms Mc-Mt Mc-Ms (when Mc-Mt is not signif icant) or Mt-Mu (when Mc-Mt is significant) (see Anderson and Gerbing 1988 for more detailed procedures). Using SCDT procedure, the researcher can find the best theoretically plausible and parsimonious model. The present research , however, examined other relevant goodness-of-fit indices when comparing alternative models si nce some models were not nested. Upon determining the best model through the SCDT procedure, proposed hypotheses were tested by examining the significance of coefficients for each hypothesized path. In order to test the conviction interact ion hypotheses from H7-a to H7-d, samples were divided into quartiles, based on the level of each conviction score. Thus, there were four groups from cognitiv e conviction and another four groups from affective conviction. Among the quartiles, only the high and low quartiles were used in the analysis, and the mid-ranged quartiles were excluded for the purpose of clear interaction tests. In addition, two research que stions for this study were investigated to determine whether cognitive and affective conviction work differently under specific conditions that differentiate the level of involvement a nd product types (i.e. hedonic and functional). Model estimation As previously introduced, the first step of the model validation procedure was a comparison of Ms and Mn that showed whet her or not the proposed measurement model was theoretically meaningful in evaluation. Because the significant chi-square value for the measurement model was possibly due to the large sample size, a chi-square difference test between Ms and Mn was conducted, rather than the pseudo chisquare test. Bentler
65 and Bonett (1980) and Tucker and Lewis ( 1973) suggested fitting the independence model (or some other very bad-fit model) to observe breadth of the discrepancy function. Discrepancy (i.e., Chi-square) for the independence model (i.e., Mn) was 21,606 (df=351) while that of Ms was 1,887 (df=296) . The chi-square difference, 19,719, was absolutely larger than the critical chi-square value (i.e., 93.17) with 55 degrees of freedom difference at the p-level of .001. Th e result thus showed that the proposed model was theoretically meaningful, and based on the result, the second step of the model estimation proceeded. In following the suggestion of Anders on and Gerbing (1988), six alternative models (from most constrained to fully unc onstrained) were compared. The competing models with simple path diagrams are presented in Table 4-10. Mc3 de-emphasizes the preceding role of emotion (H8) and attitude strength as a predecessor of brand commitment (H2, 3, and 5), while Mc2 de-emphasizes only attitude strength (H2, 3, and 5). Mc 1 excludes th e direct influence of conviction on brand commitment, as well as H8. Mt is the most theoretically plausible model proposed in the present study. Mu1 is an unconstrained mode l only without H8, and Mu2 is the most unconstrained model that includes every theoretically justifiable path.
66 Table 4-10. Path Diagrams of Competing Models Model Hypothesis paths excluded from the full model. Tested Model Mu2 None Mu1 8 Mt 4, 6 Mc1 4, 6, 8 AST CCV BCM TBL ACV BCR AST CCV BCM TBL ACV BCR AST CCV BCM TBL ACV BCR AST CCV BCM TBL ACV BCR
67 Table 4-10. Continued Mc2 2, 3, 5 Mc3 2, 3, 5, 8 Alternative models were compared with re spect to chi-square differences, as well as through examinations of associated fit indi ces. The fit indices used in the comparison procedure were GFI (Goodness of Fit Index) , NFI (Normed Fit Index), CFI (Comparative Fit Index), RMSEA (Root Mean Square Error of Approximation), and SRMR (Standardized Root Mean Square Residual). GFI was chosen due to its popularity and NFI is a good index from a practical sta ndpoint (Anderson and Ge rbing 1988). According to Hu and Bentler (1998), SRMR is the most sensitive index to models with misspecified factor covariances, and CFI and RMSEA are th e most sensitive indices to models with misspecified factor loadings. Compared to NF I, CFI is preferred when many parameters are estimated because it assigns a penalty of one for every estimated parameter. A comparison of the pairs of Mu1-Mt and Mc 1-Mc2 were undertaken by examining the fit indices since the absolute chi-square differen tial test was not appropriate because Mt is AST CCV BCM TBL ACV BCR AST CCV BCM TBL ACV BCR
68 not nested within Mu1, and also Mc2 is not nested within Mc2. For the comparison of each Mu1-Mt and Mc1-Mc2 pair, AIC (Akaik e Information Criterion: Akaike 1987) and BIC (Bayes Information Criterion: Schwarz 1978; Raftery 1993) we re used during the model comparison, in which smaller values re presented a better fit (Hu and Bentler 1995). Table 4-11 indicates that the proposed model (i.e., Mt) provides the best model fit (despite the significant chi-s quare value) when compared with other alternatives. Although Mu2 is very comparable to Mt, Anderson and GerbingÂ’s (1987) decision-tree framework of SCDTs suggests that Mt is a better model for th is case because Mt is more parsimonious than Mu2. In the comparison of Mt and Mu1, AIC for Mt was 2134.9 and 2147.2 for Mu1, indicating that Mt is a better model. BIC showed a similar result (BIC for Mt=2738.3, for Mu1=2758.8). AIC and BIC were also used for the comparison of Mc1 and Mc2. AIC for Mc1 was 2146.1 a nd 3319.2 for Mc2, and BIC was 2741.4 and 3914.4 for Mc1 and Mc2 respectively, indicating that Mc1 is a better model. While the chi-square value rejected all competing mode ls, including the Mt model, the literature suggests that the chi-square statistic tends to improperly reject correct models when sample sizes exceed 200 (e.g., Fujii and R yuichi 2000; Hair et al. 1992; Long 1983). In addition, Anderson and Gerbing (1988) stat ed that the measurement model could be judged to provide acceptable fit even though th e chi-square value is still significant under the condition of acceptable normed fit index and the other fit indices. Because the present study has a data set of 952 observations, and the other fit measures are congruent with good model fit, the model is considered to fit the data well. A lthough other fit indices provided almost identical results across the a lternative models, the chi-square statistic,
69 AIC, and BIC measures clearly confirm that the proposed model (Mt, Figure 4-2) was the one that best explained the overall brand loyalty formation process. Table 4-11. Fit Indices of Competing Models Model Â² df GFI NFI CFI RMSEA SRMR Mu2 1984.2 302 .84 .91 .92 .08 .08 Mu1 1997.2 303 .84 .91 .92 .08 .08 Mt 1986.9 304 .85 .91 .92 .08 .08 Mc1 2000.1 305 .85 .91 .92 .08 .08 Mc2 3173.2 305 .81 .85 .87 .10 .23 Mc3 3187.0 306 .81 .85 .86 .10 .23 * Note: Chi-square comparison of Mt-Mu1 and Mc1-Mc2 are not meaningful because they are not nested each other. Figure 4-2. Standardized Path Coefficients of Mt model Attitude Strength Cog. Conviction Brand Commit. True Brand Loyalty Aff. Conviction .91 .42 .68 .78 Brand Credibility .75 .28 .25
70 Table 4-12. Standardized Path Co efficients in Competing Models H# Exogenous Endogenous Mt Mc3Mc2Mc1 Mu1 Mu2 1 Brand Commitment Brand Loyalty .91 .92 .92 .91 .91 .91 2 Attitude Strength Brand Commitment .78 _ _ .77 .75 .77 3 Cognitive Conviction Attitude Strength .68 _ _ .72 .71 .68 4 Cognitive Conviction Brand Commitment _ .60 .58 _ .06* .05* 5 Affective Conviction Attitude Strength .42 _ _ .41 .43 .43 6 Affective Conviction Brand Commitment _ .26 .26 _ -.05* -.05* 8 Affective Conviction Cognitive Conviction .25 _ .24 _ _ .24 9 Brand Credibility Cognitive Conviction .28 .48 .29 .48 .48 .29 10 Brand Credibility Affective Conviction .75 .74 .73 .75 .75 .75 * Not significant at .05 level. From the results of these tests, the final model (i.e., Mt), with standardized path coefficients, is presented in Figure 4-2, while standardized coefficients of other alternative models are summarized in Table 4-12. In not only the proposed model (i.e., Mt), but also in other competing models, all coefficients were significant (all p<.01) with expected signs. Thus the proposed hypotheses are considered supported. However, the direct effect of convictions on brand commitment (i.e., H4 and H6) was not supported, because adding paths within the models (Mu1 and Mu2) didnÂ’t significantly im prove overall fit in comparison to the Mt model. Furthermore, both path coefficients of H4 and H6 were not even significant in Mu models, although they were significant in Mc models that constr ained the effect of attitude strength construct. This suggests that the attitude strength c onstruct is a necessary
71 element in mediating consumer convictions for behavioral intention (i.e., brand commitment). A simple comparison of the same models, but one with true brand loyalty (i.e., the proposed model) and one with only repeated purchasing behavior in stead of true brand loyalty, supported the strong relationship betw een commitment and true loyalty. The path coefficient for the model with repeated purch asing behavior items was .84, with an Rsquare of .71, while the path coefficient fo r the true loyalty items model (i.e., proposed model) was .91, with R-square of .83. Thes e findings suggest that true brand loyalty is better explained by brand commitment a nd that the relationship is strong. The final model in Figure 4-2 suggests that cognitive conviction strongly influences attitude strength over the aff ective conviction, although affec tive conviction considerably influences cognitive conviction. Standardized total effect of affective conviction on attitude strength (which shows the amount of attitude strength unit changes, per one unit change in affective conviction) was .59 ((.25.68) + .42), compared to the standardized direct effect of cognitive conviction (.68) . Although the total effect of affective conviction on attitude strength is still less than the direct effect of cognitive conviction, this effect comparison shows very comparable influences of both affective and cognitive conviction on attitude strength formation. Brand credibility significantly influenced both cognitive and affective convictions, but the influence was much stronger on aff ective conviction (.75) than on cognitive conviction (.28). The standardized indirect eff ect of brand credibility on attitude strength was also stronger for the affective route (BCR-ACV-AST = .32) than it is for the cognitive route (BCR-CCV-AST = .19). Adding the effect of combined ACV-CCV route
72 (BCR-ACV-CCV-AST = .13), brand credibility was processed through the ACV-driven route (.45) more than through the CCV-driven route (.19). Although th e model tested the simultaneous effect of brand credibility on cognitive and affective conviction, the considerable difference of effects may s uggest the following sequence of psychological processes: Consumers initially recall the feelin gs associated with a certain brand rather than think about it, then they cognitively elaborate the aff ective information to form a certain level of attitude strength toward the brand. Thus in this process, affective conviction seems to act as a re inforcer of cognitive conviction, while it also acts as a direct influencer on attitude strength. Mediating role of attitude strength Hypothesis 7 (-a, b, c, and d) was tested to examine whether c onviction constructs are mediated by attitude streng th in influencing brand commitment formation. In order to test the attitude stre ngth mediation hypotheses from H7-a to H7-d, samples were divided into quartiles based on the levels of each conviction score. Affective conviction was employed as a ba sis for quartile divisions, and because pleasure was the only affectiv e conviction measure that could be interpreted for directional evaluation (i.e., good to bad), onl y pleasure was used when dividing the quartiles. As previously presented, arousal and dominance are not directional measures, meaning that we cannot determine whether hi gh arousal or dominan ce scores are good or bad. Comparison of path coefficient from attitude strength to brand commitment between the models with low (N=213, 22.4%) and high (N=253, 26.6%) affective conviction respondents showed significant di fference in the coefficients. The coefficient from attitude strength (AST) to brand commitment (BCM) for low affective conviction respondent group was .64, while that for the high affective
73 conviction group was .79. Signi ficant difference in path coefficients was assessed by testing the chi-square difference between the models: one model set the path coefficient (AST BCM) of the two groups as the same, while the other freed the path coefficients. A significant chi-square differe nce (over 3.84 at .05 level) between two model settings would indicate that the path coefficient should be freed to estimate, thus implying a significant path coefficient difference. Compar ing the restricted coefficient model against the freed coefficient model yielded a chi-s quare value of 8.4 (=1410.0-1401.6) with one (=609-608) degree of freedom. This difference in dicates that the coefficients of the ASTBCM path for both groups are significantly different. Cognitive conviction was also used to te st H7. For cognitive conviction quartile dividing, all items were first averaged to represent overall cognitive conviction, then the average was divided into quartiles. Comparison of the path coefficient from attitude strength to brand commitment between the models with low (N=251, 26.4%) and high (N=241, 25.3%) cognitive conviction res pondents showed significant difference in coefficients (Â²=10.6 > 3.84 at p=.05). The coefficien t from attitude strength to brand commitment for the low cognitive conviction respondent group was .60, while that for the high cognitive conviction group was .78. Based on the results from the quartile-divide d tests, H7 was considered supported. Examining research questions Two research questions were investigat ed in this study to determine whether cognitive and affective conviction work di fferently under specific conditions that differentiate the level of involvement a nd product types (i.e. hedonic and functional). Table 4-13 shows path coefficients across cond itions. The same chi-square difference test
74 method previously used in the attitude stre ngth mediation study (i .e., H7) was performed to examine the path coefficient differences across groups (i.e., high vs. low involvement; functional vs. hedonic). Results showed no significant difference in path coefficients across comparison conditions, except for the coefficient of ACV-AST (Affective Conviction Â– Attitude Strength) between high and low involvement . Though significant, this coefficient difference was minimal (.02). Although this direct-effect-only comparison may seem to conclude that cognitive conviction more str ongly influences attitude strength than does affective conviction, the total effect comparis on (Table 4-14) provides more insight to the relationship between convictions and attitude strength. The total effect comparison in Table 4-14 indicates that affective conviction, in fact, has more influence on attitude strength under low-involvem ent condition (.75), when comp ared to high-involvement condition (.61). In addition, affective convi ction showed equivalent influence (.64) on attitude strength, compared to cognitive conviction under the hedonic product condition. On the other hand, cognitive convictions more strongly influenced attitude strength under both high-involvement and functional product conditions. Although most path coefficients were ve ry steady throughout th e conditions, ACVCCV (Affective Conviction Cognitive Conviction) and BCR-CCV (Brand Credibility Cognitive Conviction) paths changed acro ss conditions. ACV-CCV paths demonstrate that cognitive conviction is highly independent from af fective conviction under highinvolvement and for functional product type. However, under a low involvement condition, ACV-CCV path has a significantly strong coefficien t, indicating that cognitive conviction is highly dependent on affective conviction. Furthermore, BCR-CCV path
75 under low-involvement condition was not significant, whereas BCR-ACV (Brand Credibility Affective Conviction) path coefficien t was significantly strong. Therefore, the higher the involvement, the higher BCRCCV and the lower ACV-CCV, implying that cognitive conviction is based more on brand credibility (rather than affective conviction) under high involvement condition. In addition, th e lower the involvement, the lower BCR-CCV and the higher ACV-CCV, in dicating cognitive conviction is more dependent on affective conviction under low involvement condition. Compared to the unsteady BCR-CCV paths, BCR-ACV paths, wh ich are highly stable across conditions, indicate that affective conviction is primarily influenced by brand credibility at all times, while cognitive conviction, especially under low involvement condition, is not. The hedonic product condition models resembled the model (Figure 4-2) that utilized pooled data. As with othe r conditions, BCR-ACV (Brand Credibility Affective Conviction) was stronger than BCR-CCV (Brand Credibility Cognitive Conviction), while cognitive conviction influenc ed attitude strength to a greater degree than did affective conviction. Compared to functional product group, hedonic product group was not signifi cantly different (Â²df=1 < 3.84) in terms of path coefficients for CCV-AST and ACV-AST. Given the condition th at the products were hedonic oriented (they seemed to be more affectively processe d), this result may not appear correct at a first glance. However, when compared w ith the functional product group, the ACV-CCV (Affective Conviction Cognitive Conviction) path indicates significance for the hedonic product group only. Thus this suggest s that consumers may process similarly established convictions toward attitude strength formation, but affective conviction more
76 significantly influences the formation of cognitive conviction in the brand loyalty formation process of hedonic products. Table 4-13. Path Coefficien ts under Specific Conditions Exogenous Endogenous High Inv. Low Inv. Functional Hedonic BCM TBL .94 .89 .99 .86 AST BCM .78 .80 .81 .76 CCV AST .74 .61 .72 .64 ACV AST .42 .44 .41 .45 ACV CCV -.06* .51 .16* .29 BCR CCV .49 .15* .28 .30 BCR ACV .74 .75 .79 .76 * Not significant at .05 level. Table 4-14. Effects of Convi ctions on Attitude Strength Effects High Inv. Low Inv. Functional Hedonic CCV-AST (Direct) .74 .61 .72 .64 ACV-AST (Total) .42 .75 .41 .64
77 CHAPTER 5 SUMMARY AND DISCUSSION The purpose of the present research was to unveil the underlying process of brand loyalty formation by focusing on its important antecedents, such as brand commitment, attitude strength, affective/cognitive c onviction, and brand credibility. Utilizing Structural Equation Modeling in conjunction with a considerably large sample size, this study found that constructs that were theorized to be related to the process of true brand loyalty formation have significant roles. In th e following section, fi ndings related to the hypotheses are presented. Hypothesis 1 proposed that br and commitment influences true brand loyalty. This was supported, as the path coefficient from brand commitment to true brand loyalty was significant. The significant a nd high (i.e., .91 for the fully pooled-data model) path coefficient confirmed findings in the existing literature that argued that intention predicts behavior (e.g., Fishbein and Ajzen 1975). The close relationship between brand commitment and true brand loyalty was furthe r confirmed by the str onger path coefficient from brand commitment to brand loyalty wh en the Â“true brand loyaltyÂ” (instead of repeated purchasing behavior) was used as the indicator of the brand loyalty. Therefore, it is preferable that future studies use true brand loyalty in place of repeated purchasing behavior that may produce spurious loyalties. Hypothesis 2, concerning the relationsh ip between attitude strength and brand commitment, was also supported. This implies that consumers should have strong attitude strength for a brand before developing str ong intentions to purchas e. The necessity of
78 attitude strength was further confirmed by the comparison of models with attitude strength and others without it (Mc2 and Mc3). As evidence d by the fit in dices and chisquare differences, Mc2 and Mc3 had poor fits , caused by omitting th e attitude strength construct from the model. In addition, as ex amined in Mt and Mc1, direct links between convictions and commitment failed to exhibi t significance when th e attitude strength construct was present in the model. Therefore, H4 and H6 were rejected. This finding implies that attitude strength is an importa nt and necessary mediat or between conviction and commitment. Without it, the link from c onviction to commitment would be unstable. H3 and H5 were supported as well. H3 c onfirms AbelsonÂ’s (1988) argument that a durable and behaviorally predic table attitude is one with co nviction. Because the attitude strength construct in the current model is ch aracterized as being dur able and behaviorally predictable (indicated by the significant path s of H1 and H2), the significant link from cognitive conviction to attitude strength vali dates AbelsonÂ’s (1988) assertion, especially in the brand loyalty domain. Affective convictio n (H5) also confirms the predictability of affective conviction on attitude strength. Although often ig nored in past studies, the present study clearly shows that affective c onviction can independently influence attitude strength. Therefore, Jacoby and Chestnut (1978) and He nnig-Thurau and KleeÂ’s (1997) argument that brand loyalty and commitment are based on both affective and cognitive sources was confirmed. In addition, an importa nt finding regarding this dual source effect (i.e., affective and cognitive conviction) is that those sources require the attitude strength construct to establish stability and proceed to commitment and loyalty. Thus, neither commitment nor loyalty is directly connected to cognitive and aff ective conviction, but the mediator of these convictions is attitude strength. The roles of affective and cognitive
79 conviction over specific conditi ons were very steady as obser ved in the research question investigation. Under any condition, whether it was high-involvement, low-involvement, functional, or hedonic product, no considerable variation of the two path coefficient was found. In general, cognitive conviction most strongly influenced attitude strength over affective conviction. However, a more importa nt finding would be the stable role of the two constructs across different conditions. This implies that once certain convictions are implanted in the consumerÂ’s mind, the next process in the formation of brand loyalty would be essentially the same for any pr oduct or situational conditions. This postconviction-saturation phenomenon has great poten tial: future studies should examine this with an expanded set of condition variations. Paths that varied significantly across different conditions were BCR-CCV (H9: Brand Credibility Cognitive Conviction) and ACVCCV (H8: Affective Conviction Cognitive Conviction), while BCR-ACV path (H10: Brand Credibility Affective Conviction) was very steady and significant across all conditions. The stability of BCRACV path coefficient implies that consumer s may consistently us e a certain amount of brand-relevant memory (i.e., brand credibility) as a source of affective conviction. This is convincing considering Bower and ForgasÂ’s (2001) claim that consumers can recall feelings about a brand without much recall of sp ecifically justifiable factors. In addition, this reflects ZajoncÂ’s (2000) a ssertion that affective quality of the original input is the first element to emerge when people try to retrieve an object. Because consumers may tend to use global feelings about a brand when they form affective conviction, the link between past memory (i.e., brand credibil ity) and affective conviction would become stable and global for any specific conditions. In addition to this stability, the path
80 coefficient of BCR-ACV was much higher than that of BCR-CCV. This indicates that, when forming cognitive conviction, consum ers may not only use brand credibility, but also utilize other sources. This was eviden ced by the significant link between affective conviction and cognitive conviction, which was theorized in H8. In the tripartite relations hip between brand credibility , affective conviction, and cognitive conviction, consumers mainly use past experience (i.e., brand credibility) with a brand to determine feelings about a brand. In this process, past experience would give some cognitive information about the bra nd and would generate the formation of cognitive conviction. And this cognitive conviction is boosted and elaborated upon by affective conviction. Although this elabora tion effect of affective conviction was significant in the model with fully pooled data, some specific c onditions did not have similar effects. Among the four conditions (i.e., high/low involvement and functional/hedonic product), ACV CCV (Affective Conviction Cognitive Conviction) path was not significant unde r high involvement condition, or under functional product condition. This implie s that although consumers would use both cognitive and affective conviction, they may not desire affective c onviction to influence cognitive conviction under hi gh involvement condition, in all probability because they want to process the brand information indepe ndently and rationally due to the relatively high risk associated with the outco mes of their loyal behaviors. Similar results were observed for the functional product condition. This is convincing because consumers would not necessarily need affective conviction in order to elaborate the cognitive conviction for f unctional products. On the other hand, the ACV-CCV path was significantly strong, es pecially under low i nvolvement conditions,
81 whereas BCR-CCV (Brand Credibility Cognitive Conviction) was not significant. These two results indicate that consumers us e brand credibility info rmation directly and exclusively for affective conviction, and th at affective convicti on strongly elaborates cognitive conviction. In fact , under low involvement conditi on, cognitive conviction was highly dependent on affective conviction. Overall, all hypotheses except for H4 and H6 were supported. The primary finding is that brand credibility serves as a source of cognitive and affective conviction, and that affective conviction often take s an elaborating role in c ognitive conviction formation. Those convictions then steadily influence at titude strength, whic h helps develop brand commitment, leading to true brand loyalty. In addition, attitude strength was found to take a critical role in connecting convictions to brand commitment. This study is not without limitations. Alt hough the significant chi-square value for all tested models (including the proposed mode l) can be tolerated given the large sample size of this study, Anderson and Gerbing (1988) cautioned that large sample size can also produce a significant value for an SCDT, even when there are trivial discrepancies between models. For this reason, Anderson and Gerbing suggested using the normed fit index in conjunction with formal statistical tests. Since NFI was identical (.91) for Mu2, Mu1, Mt, and Mc1, this may suggest that t hose models fit the obser ved data equally well from a practical standpoint. Because Mc1 (w hich excluded the ACVCCV path from Mt model) is the most parsimonious model am ong those with same NFI, Mc1 could be another good model choice. Although Mt was selected in this study because of its theoretical parsimoniousness and soundness, read ers should note that the Mc1 could also be a good model.
82 This research does not claim the proposed model to be the true model. Bollen (1989, p. 68) advised that Â“If a model is consiste nt with reality, then the data should be consistent with the model. But, if the data is consistent with a model, this does not imply that the model corresponds to reality.Â” Therefore, although th e proposed model has acceptable goodness of fit with the data, there ma y be other models that would have equal or better fit with reality. In addition, Â“what is known as the nominalistic fallacy, naming something does not necessarily mean that one understands itÂ” (Anderson and Gerbing 1988, p. 421) may inherently exist in the current study as in a ny other SEM study. Because the scales used in this study are not perfectly valid and re liable, measurement errors (while they are statistically controlled in SEM) are inevitabl e. Such errors could induce unexpected and undetected errors in the current study. Additionally, this study does not argue that the proposed model is a causal model. Relationships between latent variables in th e proposed model are only causal inferences (based on theories), not perfect causalities. Fo r this reason, SEM assumes that causality is probabilistic rather than deterministic (Kline 1998). Furthermore, as with any other SEM model, changes in the values of exogenous variables (the first exogenous variable in particular, i.e., brand credib ility) cannot be explained by the proposed model. Rather, they are considered to be influenced by ot her factors external to the model. Background variables such as gender and age are examples of such external f actors. Because other sampling categories such as gender, age, or education can take important roles in the overall process of the proposed model, any future study should examine other specific sample groups beyond the college students.
83 Although sampling college students was cons idered appropriate for effectively controlling sample variation acr oss the pretests and main study, the exclusive use of college students could have produced unexp ected sampling errors. Furthermore, such limited sample characteristics are a possible impediment to the generalizability of the study results in regard to the broader cons umer population. A recent study based on the meta-analysis of student and non-student sa mpled studies (Peterson 2001) emphasizes the importance of replicating research based on college student subjec ts with non-student subjects before attempting any generalizations . Therefore, due caution should taken in the generalization of this resear ch. Future research should al so examine the appropriateness of the proposed model to the general population. A further limitation may be inherent in the limited number of product selections investigated during this study. Including a broader spectrum of products, in order to satisfy certain conditions not approached by the current study, would enhance any future research. Expanding the sample criteria would provide more variety of possible product selections.
84 CHAPTER 6 CONTRIBUTION This study provides various theoretical and managerial implications to marketing practitioners and researchers. In order to i nvestigate the underlying process of true brand loyalty formation, this study adopted the at titude strength theory (i.e., AbelsonÂ’s conviction theory) of social psychology, whic h has been rarely used for brand loyalty studies, in spite of its theore tical applicability. In addition, this study balanced both emotional and cognitive perspectives of the attitude strength antecedents (i.e., brand conviction). Since cognition and affect have been shown as distinct (though related) attitudinal constructs in many studies, inve stigating both constructs simultaneously provided more theoretical plausibility to the proposed model. In addition, examination of brand credibility as an antecedent of brand convictions offered more thorough theoretical implications. Lastly, the generalizability and robustness of the proposed model was examined via the use of multiple product cl asses representing hedonism/functionality and low/high involvement. Managerial implications are that the a dvertisers will need to identify whether emotional or cognitive brand conviction is th e major factor that le ads consumers into a high level of attitude strength. By understandi ng this, advertisers woul d be able to create either cognition or emotion based communicatio n strategies in order to ultimately make their consumers truly loyal. This investigation will be important to undertake before creating any marketing communi cation strategy, especially for a new brand. Because making consumers truly brand loyal would require a long time in many cases, having a
85 firm base of overall strategy, i.e., emoti on/cognition or both, will make the long-term strategic communication planning process ea sier and more efficient. The product category classification scheme (i.e., HI-H, LI-H , HI-F, and LI-F) studied in this research can be used as a base for strategy differentia tion. This strategic research process will become critical in light of the wide pract ice of integrated marketing communications, whose ultimate goal is to build brand valu e, which is based on strong and long-term consumer support, i.e., true brand loyalty.
86 APPENDIX A DESCRIPTIVE STATISTICS OF MEASUREMENT ITEMS Item Mean SD N Item MeanSDN ACV1 54.74 18.43 952 BCR16.82 1.49 952 ACV2 34.98 19.34 952 BCR26.68 1.49 952 ACV3 44.55 21.22 952 BCR36.41 1.83 952 TBL1 27.04 19.13 952 BCR46.74 1.73 952 TBL2 30.08 22.15 952 BCR57.02 1.61 952 TBL3 24.49 18.30 952 BCR66.14 1.72 952 TBL4 27.54 21.34 952 BCR75.59 2.10 952 TBL5 31.06 19.46 952 CCV14.67 2.28 952 TBL6 34.59 22.77 952 CCV23.99 2.36 952 BCM1 5.53 2.26 952 CCV35.79 1.94 952 BCM2 2.22 1.33 952 CCV43.22 2.29 952 BCM3 6.43 2.15 952 CCV54.38 2.35 952 AST1 2.68 1.34 952 CCV63.96 2.34 952 AST2 5.39 2.01 952 CCV73.41 2.27 952 AST3 6.52 1.72 952 CCV84.07 2.37 952 AST4 4.51 2.35 952 CCV93.89 2.30 952 AST5 5.14 2.15 952 CCV104.94 2.35 952 * Item descriptions ACV: Affective Conviction item TBL: True Brand Loyalty item BCM: Brand Commitment item AST: Attitude Strength item BCR: Brand Credibility item CCV: Cognitive Conviction item
87 APPENDIX B CORRELATION MATRIX OF MEASUREMENT ITEMS ACV1 ACV2 ACV3TBL1TBL2TBL3TBL4 TBL5 TBL6 ACV1 1.00 0.44 0.45 0.28 0.26 0.24 0.23 0.35 0.31 ACV2 0.44 1.00 0.31 0.23 0.22 0.21 0.19 0.24 0.21 ACV3 0.45 0.31 1.00 0.16 0.13 0.14 0.14 0.18 0.16 TBL1 0.28 0.23 0.16 1.00 0.75 0.89 0.67 0.85 0.60 TBL2 0.26 0.22 0.13 0.75 1.00 0.70 0.91 0.61 0.88 TBL3 0.24 0.21 0.14 0.89 0.70 1.00 0.79 0.84 0.62 TBL4 0.23 0.19 0.14 0.67 0.91 0.79 1.00 0.61 0.87 TBL5 0.35 0.24 0.18 0.85 0.61 0.84 0.61 1.00 0.70 TBL6 0.31 0.21 0.16 0.60 0.88 0.62 0.87 0.70 1.00 BCM1 0.36 0.26 0.16 0.57 0.71 0.56 0.70 0.60 0.75 BCM2 0.16 0.15 0.10 0.28 0.30 0.26 0.29 0.24 0.27 BCM3 0.39 0.18 0.19 0.42 0.48 0.40 0.45 0.44 0.50 AST1 0.60 0.37 0.34 0.39 0.41 0.36 0.39 0.44 0.46 AST2 0.46 0.39 0.23 0.50 0.51 0.48 0.48 0.49 0.49 AST3 0.50 0.30 0.32 0.32 0.35 0.30 0.33 0.36 0.38 AST4 0.30 0.28 0.11 0.46 0.46 0.45 0.43 0.43 0.41 AST5 0.35 0.30 0.24 0.43 0.44 0.42 0.43 0.39 0.40 BCR1 0.51 0.28 0.33 0.29 0.34 0.27 0.34 0.33 0.39 BCR2 0.49 0.25 0.30 0.32 0.35 0.30 0.34 0.37 0.40 BCR3 0.30 0.15 0.16 0.18 0.22 0.14 0.19 0.25 0.28 BCR4 0.37 0.14 0.20 0.17 0.20 0.14 0.17 0.21 0.23 BCR5 0.45 0.28 0.31 0.23 0.30 0.21 0.29 0.29 0.35 BCR6 0.28 0.24 0.18 0.31 0.38 0.27 0.35 0.28 0.37 BCR7 0.32 0.28 0.22 0.38 0.39 0.36 0.37 0.35 0.35 CCV1 0.32 0.25 0.19 0.40 0.39 0.39 0.38 0.38 0.37 CCV2 0.21 0.20 0.10 0.53 0.61 0.57 0.63 0.48 0.56 CCV3 0.42 0.29 0.28 0.37 0.38 0.35 0.37 0.37 0.39 CCV4 0.21 0.24 0.08 0.43 0.41 0.43 0.40 0.38 0.37 CCV5 0.32 0.31 0.15 0.45 0.49 0.44 0.48 0.41 0.46 CCV6 0.24 0.27 0.11 0.41 0.43 0.41 0.43 0.35 0.37 CCV7 0.15 0.22 0.06*0.41 0.42 0.41 0.42 0.34 0.34 CCV8 0.26 0.26 0.13 0.50 0.53 0.49 0.50 0.44 0.48 CCV9 0.21 0.27 0.11 0.43 0.43 0.41 0.40 0.36 0.37 CCV10 0.36 0.30 0.29 0.32 0.32 0.30 0.30 0.30 0.31 * Not Significant at p=.05 ** Not Significant at p=.01
88 Appendix B. Continued BCM1 BCM2 BCM3AST1AST2AST3AST4 AST5 ACV1 0.36 0.16 0.39 0.60 0.46 0.50 0.30 0.35 ACV2 0.26 0.15 0.18 0.37 0.39 0.30 0.28 0.30 ACV3 0.16 0.10 0.19 0.34 0.23 0.32 0.11 0.24 TBL1 0.57 0.28 0.42 0.39 0.50 0.32 0.46 0.43 TBL2 0.71 0.30 0.48 0.41 0.51 0.35 0.46 0.44 TBL3 0.56 0.26 0.40 0.36 0.48 0.30 0.45 0.42 TBL4 0.70 0.29 0.45 0.39 0.48 0.33 0.43 0.43 TBL5 0.60 0.24 0.44 0.44 0.49 0.36 0.43 0.39 TBL6 0.75 0.27 0.50 0.46 0.49 0.38 0.41 0.40 BCM1 1.00 0.35 0.51 0.47 0.57 0.42 0.48 0.44 BCM2 0.35 1.00 0.14 0.22 0.34 0.21 0.33 0.29 BCM3 0.51 0.14 1.00 0.39 0.35 0.34 0.24 0.30 AST1 0.47 0.22 0.39 1.00 0.52 0.55 0.35 0.43 AST2 0.57 0.34 0.35 0.52 1.00 0.50 0.65 0.60 AST3 0.42 0.21 0.34 0.55 0.50 1.00 0.33 0.43 AST4 0.48 0.33 0.24 0.35 0.65 0.33 1.00 0.57 AST5 0.44 0.29 0.30 0.43 0.60 0.43 0.57 1.00 BCR1 0.41 0.19 0.39 0.53 0.37 0.52 0.28 0.43 BCR2 0.42 0.20 0.40 0.51 0.40 0.52 0.28 0.43 BCR3 0.23 0.06 0.24 0.30 0.18 0.32 0.14 0.20 BCR4 0.19 0.07 0.29 0.32 0.17 0.31 0.10 0.16 BCR5 0.37 0.17 0.38 0.49 0.31 0.44 0.20 0.30 BCR6 0.33 0.23 0.22 0.30 0.33 0.31 0.32 0.38 BCR7 0.42 0.28 0.27 0.38 0.44 0.33 0.51 0.48 CCV1 0.42 0.29 0.24 0.37 0.51 0.33 0.62 0.52 CCV2 0.57 0.31 0.33 0.32 0.53 0.32 0.56 0.51 CCV3 0.41 0.18 0.36 0.43 0.43 0.48 0.36 0.50 CCV4 0.40 0.24 0.18 0.26 0.52 0.24 0.59 0.47 CCV5 0.48 0.33 0.29 0.42 0.61 0.40 0.58 0.58 CCV6 0.42 0.27 0.24 0.30 0.57 0.30 0.63 0.54 CCV7 0.39 0.28 0.21 0.22 0.49 0.23 0.60 0.49 CCV8 0.49 0.30 0.28 0.35 0.55 0.35 0.57 0.54 CCV9 0.37 0.31 0.24 0.29 0.49 0.29 0.53 0.62 CCV10 0.34 0.22 0.29 0.38 0.44 0.44 0.41 0.53 * Not Significant at p=.05 ** Not Significant at p=.01
89 Appendix B. Continued BCR1 BCR2 BCR3BCR4BCR5BCR6BCR7 CCV1 ACV1 0.51 0.49 0.30 0.37 0.45 0.28 0.32 0.32 ACV2 0.28 0.25 0.15 0.14 0.28 0.24 0.28 0.25 ACV3 0.33 0.30 0.16 0.20 0.31 0.18 0.22 0.19 TBL1 0.29 0.32 0.18 0.17 0.23 0.31 0.38 0.40 TBL2 0.34 0.35 0.22 0.20 0.30 0.38 0.39 0.39 TBL3 0.27 0.30 0.14 0.14 0.21 0.27 0.36 0.39 TBL4 0.34 0.34 0.19 0.17 0.29 0.35 0.37 0.38 TBL5 0.33 0.37 0.25 0.21 0.29 0.28 0.35 0.38 TBL6 0.39 0.40 0.28 0.23 0.35 0.37 0.35 0.37 BCM1 0.41 0.42 0.23 0.19 0.37 0.33 0.42 0.42 BCM2 0.19 0.20 0.06 0.07 0.17 0.23 0.28 0.29 BCM3 0.39 0.40 0.24 0.29 0.38 0.22 0.27 0.24 AST1 0.53 0.51 0.30 0.32 0.49 0.30 0.38 0.37 AST2 0.37 0.40 0.18 0.17 0.31 0.33 0.44 0.51 AST3 0.52 0.52 0.32 0.31 0.44 0.31 0.33 0.33 AST4 0.28 0.28 0.14 0.10 0.20 0.32 0.51 0.62 AST5 0.43 0.43 0.20 0.16 0.30 0.38 0.48 0.52 BCR1 1.00 0.75 0.41 0.44 0.58 0.40 0.38 0.31 BCR2 0.75 1.00 0.43 0.44 0.59 0.44 0.39 0.32 BCR3 0.41 0.43 1.00 0.52 0.35 0.26 0.18 0.18 BCR4 0.44 0.44 0.52 1.00 0.40 0.24 0.17 0.14 BCR5 0.58 0.59 0.35 0.40 1.00 0.43 0.37 0.23 BCR6 0.40 0.44 0.26 0.24 0.43 1.00 0.48 0.38 BCR7 0.38 0.39 0.18 0.17 0.37 0.48 1.00 0.65 CCV1 0.31 0.32 0.18 0.14 0.23 0.38 0.65 1.00 CCV2 0.28 0.28 0.15 0.09 0.19 0.32 0.47 0.58 CCV3 0.44 0.38 0.24 0.20 0.33 0.27 0.42 0.45 CCV4 0.16 0.16 0.03**0**0.06*0.27 0.41 0.57 CCV5 0.33 0.35 0.14 0.13 0.24 0.32 0.49 0.60 CCV6 0.21 0.23 0.07 0.05 0.16 0.31 0.50 0.62 CCV7 0.14 0.16 0.04**-0.01**0.08 0.26 0.46 0.57 CCV8 0.26 0.29 0.12 0.11 0.18 0.28 0.41 0.55 CCV9 0.23 0.27 0.11 0.06 0.11 0.33 0.45 0.52 CCV10 0.35 0.38 0.19 0.18 0.27 0.28 0.39 0.43 * Not Significant at p=.05 ** Not Significant at p=.01
90 Appendix B. Continued CCV2 CCV3 CCV4CCV5CCV6CCV7CCV8 CCV9 CCV10 ACV1 0.21 0.42 0.21 0.32 0.24 0.15 0.26 0.21 0.36 ACV2 0.20 0.29 0.24 0.31 0.27 0.22 0.26 0.27 0.30 ACV3 0.10 0.28 0.08 0.15 0.11 0.06*0.13 0.11 0.29 TBL1 0.53 0.37 0.43 0.45 0.41 0.41 0.50 0.43 0.32 TBL2 0.61 0.38 0.41 0.49 0.43 0.42 0.53 0.43 0.32 TBL3 0.57 0.35 0.43 0.44 0.41 0.41 0.49 0.41 0.30 TBL4 0.63 0.37 0.40 0.48 0.43 0.42 0.50 0.40 0.30 TBL5 0.48 0.37 0.38 0.41 0.35 0.34 0.44 0.36 0.30 TBL6 0.56 0.39 0.37 0.46 0.37 0.34 0.48 0.37 0.31 BCM1 0.57 0.41 0.40 0.48 0.42 0.39 0.49 0.37 0.34 BCM2 0.31 0.18 0.24 0.33 0.27 0.28 0.30 0.31 0.22 BCM3 0.33 0.36 0.18 0.29 0.24 0.21 0.28 0.24 0.29 AST1 0.32 0.43 0.26 0.42 0.30 0.22 0.35 0.29 0.38 AST2 0.53 0.43 0.52 0.61 0.57 0.49 0.55 0.49 0.44 AST3 0.32 0.48 0.24 0.40 0.30 0.23 0.35 0.29 0.44 AST4 0.56 0.36 0.59 0.58 0.63 0.60 0.57 0.53 0.41 AST5 0.51 0.50 0.47 0.58 0.54 0.49 0.54 0.62 0.53 BCR1 0.28 0.44 0.16 0.33 0.21 0.14 0.26 0.23 0.35 BCR2 0.28 0.38 0.16 0.35 0.23 0.16 0.29 0.27 0.38 BCR3 0.15 0.24 0.03**0.14 0.07 0.04**0.12 0.11 0.19 BCR4 0.09 0.20 0**0.13 0.05**-0.01**0.11 0.06* 0.18 BCR5 0.19 0.33 0.06*0.24 0.16 0.08 0.18 0.11 0.27 BCR6 0.32 0.27 0.27 0.32 0.31 0.26 0.28 0.33 0.28 BCR7 0.47 0.42 0.41 0.49 0.50 0.46 0.41 0.45 0.39 CCV1 0.58 0.45 0.57 0.60 0.62 0.57 0.55 0.52 0.43 CCV2 1.00 0.47 0.62 0.62 0.63 0.60 0.65 0.56 0.46 CCV3 0.47 1.00 0.39 0.55 0.45 0.37 0.47 0.41 0.49 CCV4 0.62 0.39 1.00 0.66 0.73 0.74 0.65 0.63 0.46 CCV5 0.62 0.55 0.66 1.00 0.78 0.68 0.70 0.65 0.58 CCV6 0.63 0.45 0.73 0.78 1.00 0.80 0.69 0.68 0.55 CCV7 0.60 0.37 0.74 0.68 0.80 1.00 0.67 0.67 0.50 CCV8 0.65 0.47 0.65 0.70 0.69 0.67 1.00 0.72 0.59 CCV9 0.56 0.41 0.63 0.65 0.68 0.67 0.72 1.00 0.62 CCV10 0.46 0.49 0.46 0.58 0.55 0.50 0.59 0.62 1.00 * Not Significant at p=.05 ** Not Significant at p=.01
91 APPENDIX C QUESTIONNAIRES Pretest 1 Background of Survey I am a doctoral candidate at the University of Florida. I have created a scale that I would like you to examine for face validity. The origin of my scale is Abelson's (1989) conviction scale, which was devel oped to measure the level of conviction toward social issues. My research studies brand loyalty, and I theorize that the cognitive conviction a consumer has toward his/her favorite br and affects brand loyalty formation. Because the original conviction scale is designed for use in social settings, validation of my scale for use as a measure of cognitive brand conviction is necessary. In my research, brand loyalty is defined as repeat purchasing behavior of the same brand based on a perc eived importance of brand choice. Please respond to the following questions. Each item has two questions as follow: * Does this item represent the domain of brand conviction? * How necessary is it to use this item to measure brand conviction? Note: In the questionnaire below, you will see a term "loyal brand." Loyal brand means a brand that you have rep eatedly purchased based on a perceived importance of brand choice. Thanks so much for your participation! 1. My beliefs about my favorit e brand express the real me. Doesn't represent Represent 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 2. I can't imagine ever changing my mind to a different brand. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9
92 3. I would be willing to spend a day a month working for a group supporting my views about my favorite brand. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 4. I think my view about my favor ite brand is absolutely correct. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 5. I think about my favorite brand often. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 6. I hold my views about my fa vorite brand very strongly. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 7. My beliefs about my favorite brand are important to me. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 8. I am extremely concerned about my favorite brand. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9
93 9. When I think about my favorite br and, I feel fearful of death. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 10. I've held my views about my favorite br and a long time compared to most people. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 11. I have more knowledge about my favorite brand than the average person. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 12. It's easy to explain my vi ews about my favorite brand. Doesn't represent Represents 1 2 3 4 5 6 7 8 9 Not really necessary Highly necessary 1 2 3 4 5 6 7 8 9 13. Please provide us any comments or suggestions. Please tell us about you. Your education Ph.D. Master's Bachelor's
94 Your Occupation University Professor Private Institute Researcher Other: What is your main research area? Years in your discipline. "1-2" "3-6" "7-10" "Over 10"
95 Pretest 2 Please answer to the following questions. 1. Assuming you are buying a pa ir of designer sunglasses in a store, how much would you care in selecting a product from many ot her choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 2. Assuming you are buying a pa ir of designer sunglasses in a store, how important would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 3. Assuming you are buying a pa ir of designer sunglasses in a store, how much would you be concerned about the outcome of your choice in making your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9 4. Would you characterize a pair of designer sunglasses as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 5. Assuming you are buying a high fashion watch in a store, how much would you care in selecting a product from many other choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 6. Assuming you are buying a high fashion watch in a store, how important would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 7. Assuming you are buying a high fashion watch in a store, how much would you be concerned about the outcome of your choice in making your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9
96 8. Would you characterize a high fashion watch as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 9. Assuming you are buying a donut in a store, how much wo uld you care in selecting a product from many other choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 10. Assuming you are buying a donut in a store, how important would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 11. Assuming you are buying a donut in a store, how much would you be concerned about the outcome of your choice in making your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9 12. Would you characterize a donut as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 13. Assuming you are buying a soft drink in a store, how much would you care in selecting a product from many other choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 14. Assuming you are buying a soft drink in a store, how importan t would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 15. Assuming you are buying a soft drink in a store, how much would you be concerned about the outcome of your choice in making your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9
97 16. Would you characterize a soft drink as primarily a function al product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 17. Assuming you are buying an auto insurance , how much would you care in selecting an insurance from many other ch oices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 18. Assuming you are buying an auto insurance , how important would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 19. Assuming you are buying an auto insurance , how much would you be concerned about the outcome of your choi ce in making your selection? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9 20. Would you characterize an auto insurance as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 21. Assuming you are buying a paper towel in a store, how much would you care in selecting a product from many other choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 22. Assuming you are buying a paper towel in a store, how importa nt would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 23. Assuming you are buying a paper towel in a store, how much would you be concerned about the outcome of your choice in making your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9
98 24. Would you characterize a paper towel as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 25. Assuming you are buying a black-and-white laser printer in a store, how much would you care in selecting a product from many ot her choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 26. Assuming you are buying a black-and-white laser printer in a store, how important would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9 27. Assuming you are buying a black-and-white laser printer in a store, how much would you be concerned about the outcome of your choice in making your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9 28. Would you characterize a bl ack-and-white laser printer as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 29. Assuming you are buying a nonrechargeable AAA size battery in a store, how much would you care in selecting a product from ma ny other choices available in the market? Would not care at all Would care a great deal 1 2 3 4 5 6 7 8 9 30. Assuming you are buying a nonrecharbeable AAA size battery in a store, how important would it be to you to make a right choice? Not important at all Extremely important 1 2 3 4 5 6 7 8 9
99 31. Assuming you are buying a nonrechargeable AAA size battery in a store, how much would you be concerned about the outcome of your choice in maki ng your selection of the product? Not concerned at all Extremely concerned 1 2 3 4 5 6 7 8 9 32. Would you characterize a non rechargeable AAA size battery as primarily a functional product or a product for pleasure? For functional use For pleasure 1 2 3 4 5 6 7 8 9 Please tell us about you. Your Age: Your gender Male Female
100 Main Survey Sample (HI-H: Designer Sunglasses) Have you ever purchased (or considered buyi ng) a pair of designer sunglasses before? Yes, I have purchased desi gner sunglasses in the past. Yes, I have considered pur chasing designer sunglasses. No, I have never considered pur chasing designer sunglasses. If you have no experience of purchasing any one of the above product, please click here and you will be directed to th e different product categories. If you have either purchased or considered buying the product above, please complete the survey below. Thanks! Survey starts here . Thinking of your experiences associated with the specific br and(s) you have bought, please answer the following questions. If you do not yet have a favorite brand but have considered maki ng a purchase: please select a brand that you would be most interested in buying in the future, and answer the questions. Please select your most favorite brand (that you have repeatedly bought or would want to buy repeatedly in the near fu ture) from the list below. Designer Sunglasses: Bulgari Calvin Klein Fendi Fossil Gianfranco Ferre Giorgio Armani /Emporio Armani GUCCI Guess Lacoste Luxottica Sergio Tacchini Other:
101 1. How do you FEEL about your favorit e brand of designer sunglasses? Please click here to see the instructio ns for the following pictorial questions. 1-a. How certain are you about your f eeling you indicated for the first row of pictorial question #1 above? Not very certain Very certain 1 2 3 4 5 6 7 8 9 1-b. How certain are you about your f eeling you indicated for the second row of pictorial question #1 above? Not very certain Very certain 1 2 3 4 5 6 7 8 9 1-c. How certain are you about your f eeling you indicated for the third row of pictorial question #1 above? Not very certain Very certain 1 2 3 4 5 6 7 8 9
102 2. During my next purchase, I would buy my fa vorite brand of desi gner sunglasses as the last time. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 3. I think I am (would be) loyal to on ly one brand of designer sunglasses. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 4. I would always buy the same brand of designer sunglasses. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 5. Usually, I would buy the same brand of designer sunglasses. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 6. Various brands of designer sunglas ses available in the market are: Very alike Very different 1 2 3 4 5 6 7 8 9 7. The brand name is the first thing I w ould be looking at when purchasing designer sunglasses. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 8. When buying a pair of designer sungla sses, how committed would you be to buying your favorite brand, rather th an an alternative brand? Not very committed Very committed 1 2 3 4 5 6 7 8 9 9. If you could not get your favorite brand of designer sunglasses at the store, you would: Happily buy a different brand Reluctantly buy a different brand Not buy the product until the next time you shopped Try a different shop Keep trying different shops until you got the brand you wanted.
103 10. My attitude to my favorite brand of designer sunglasses is: Very negative Very positive -5 -4 -3 -2 -1 1 2 3 4 5 11. How strong or intense is your feeli ng toward your favorite brand of designer sunglasses? Not very intense Very intense 1 2 3 4 5 6 7 8 9 12. How certain do you feel about your attitude toward your favorite brand of designer sunglasses? Not very certain Very certain 1 2 3 4 5 6 7 8 9 13. How important would you say your favorite brand of designer sunglasses would be to you personally? Not very important Very important 1 2 3 4 5 6 7 8 9 14. How knowledgeable do you feel about your favorite brand of designer sunglasses? Not very knowledgeable Very knowledgeable 1 2 3 4 5 6 7 8 9 15. My favorite brand of designer su nglasses delivers what it promises. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 16. Product claims from my favorite bra nd of designer sunglasses are believable. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 17. I just can't believe what the ads say a bout my favorite brand of designer sunglasses. Strongly Agree Strongly Disagree 1 2 3 4 5 6 7 8 9 18. My experiences with my favorite brand of designer sunglasses make me wary of their claims. Strongly Agree Strongly Disagree 1 2 3 4 5 6 7 8 9
104 19. My favorite brand of designer sunglasses has a name I can trust. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 20. My favorite brand of designer sunglasses is at the forefront of using technology to deliver a better product. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 21. Wearing my favorite brand of designer sunglasses makes me feel like someone who is competent and know what he/she is doing. Strongly disagree Strongly agree 1 2 3 4 5 6 7 8 9 22. My beliefs about my favorite brand of designer sunglasses express the real me. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 23. I can't imagine ever changing my mind to a different brand of designer sunglasses. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 24. I think my view about my favorite brand of designer sunglasses is absolutely correct. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 25. I think about my favorite bra nd of designer sunglasses often. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 26. I hold my views about my favorite bra nd of designer sunglasses very strongly. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 27. My beliefs about my favorite brand of designer sunglasses are important to me. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9
105 28. I am extremely concerned about my favorite brand of designer sunglasses. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 29. I've held my views about my favorit e brand of designer sunglasses a long time compared to most people. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 30. I have more knowledge about my favorite brand of designer sunglasses than the average person. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 31. It's easy to explain my views about my favorite brand of designer sunglasses. Strongly Disagree Strongly Agree 1 2 3 4 5 6 7 8 9 End of survey. Thanks for your participation.
106 LIST OF REFERENCES Abelson, R. P. (1988). Conviction. American Psychologist, 43, 267-275. Abelson, R. P. (1995). Attitude Extremity. In R. E. Petty & J. A. Krosnick (Eds.), Attitude Strength: Antecedents and Consequences (pp. 25-41).Mahwah, NJ: Lawrence Erlbaum Associates, Publishers. Akaike, H. (1987). Factor Analysis and AIC. Psychometrika, 52, 317-332. Allen, C. T., Machleit, K. A., & Kleine, S. (1992). A Comparison of Attitudes and Emotions as Predictors of Behavior at Diverse Levels of Behavioral Experience. Journal of Consumer Research, 18(4), 493-504. Anderson, J. C., & Gerbing, D. W. (1988). Structural Equation M odeling in Practice: A Review and Recommended Two-Step Approach. Psychological Bulletin, 103(3), 411-423. Arbuckle, J. L. (1989). AMOS: An alysis of Moment Structures. The American Statistician, 43, 66-67. Arbuckle, J. L. (1994). AMOS: An alysis of Moment Structures. Psychometrika, 59, 135137. Assael, H. (1995). Consumer Behavior and Marketing Action. Cincinnati: South-Western College Publishing. Bagozzi, R. P. (1983). Issues in the Appli cation of Covariance Structure Analysis: A Further Comment. Journal of Consumer Research, 9, 449-450. Bagozzi, R. P., & Phillips, L. W. (1982) . Representing and Testing Organizational Theories: A Holistic Construal. Administrative Science Quarterly, 27, 459-489. Bass, F. M. (1974). The Theory of Stoc hastic Preference and Brand Switching. Journal of Marketing, 11, 1-20. Batra, R., & Stephens, D. (1994). Attitude Effects of Ad-Evoked Moods and Emotions: The Moderating Role of Motivation. Psychology & Marketing, 11(3), 199-215. Beatty, S. E., Coleman, J. E., Ellis, K. L., Lee, J., & Mayer, M. (1996). Customer-Sales Associate Retail Relationships. Journal of Retailing, 72(Fall), 223-247.
107 Bejou, D. (1997). Relationship Marketing: Evolution, Present State, and Future. Psychology & Marketing, 14(December), 727-735. Bentler, P. M., & Bonett, D. G. (1980). Si gnificance Tests and Goodness of Fit in the Analysis of Covariance Structures. Psychological Bulletin, 88, 588-606. Berry, L. L. (1995). Relationship Marketing of Services: Growing Interest, Emerging Perspectives. Journal of the Academy of Marketing Science, 23(Fall), 236-245. Berry, L. L., & Parasuraman, A. (1992). Prescr iptions for a Service Quality Revolution in America. Organizational Dynamics, 20(Spring), 5-15. Bizer, G. Y., & Krosnick, J. A. (2001). E xploring the Structure of Strength-Related Attitude Features: The Relation between Attitude Importance and Attitude Accessibility. Journal of Personality and Social Psychology, 81(4), 566-586. Bloch, P. H. (1981). Involvement beyond the Pu rchase Process: Conceptual Issues and Empirical Investigation. Advances in Consumer Research, 8, 61-65 Bloemer, J. M. M., & Kasper, H. D. P. (1995). The Complex Relationship Between Consumer Satisfaction and Brand Loyalty. Journal of Economic Psychology, 16, 311-329. Bloemer, J. M. M., & Lemmink, J. G. A. M. (1992). The Importance of Customer Satisfaction in Explaining Brand and Dealer Loyalty. Journal of Marketing Management, 8, 351-364. Bodur, H. O., Brinberg, D., & Coupey, E. (2000) . Belief, Affect, and Attitude: Alternative Models of the Determinants of Attitude. Journal of Consumer Psychology, 9(1), 1728. Bollen, K. A. (1989). Structural Equations wi th Latent Variables. New York: John Wiley & Sons. Boninger, D. S., Krosnick, J. A., & Bere nt, M. K. (1995). The Causes of Attitude Importance: Self-interest, Soci al Identification, and Values. Journal of Personality and Social Psychology, 68, 61-80. Bowen, L., & Chaffee, S. H. (1974). Produc t Involvement and Pe rtinent Advertising Appeals. Journalism Quarterly, 51(Winter), 611-644. Bower, G. H., & Forgas, J. P. (2001). Mood a nd Social Memory. In J. P. Forgas (Ed.), The Handbook of Affect and Social Cognition (pp. 95-120). Mahwah, New Jersey: Lawrence Erlbaum.
108 Breckler, S. J. (1984). Empirical Validati on of Affect, Behavior, and Cognition as Distinct Components of Attitude. Journal of Personality and Social Psychology, 47, 1191-1205. Brickman, P. (1987). Commitment, Conflict, and Caring. Englewood Cliffs, NJ: PrenticeHall. Brown, D. W. (1974). Adolescent Attitudes and Lawful Behavior. Public Opinion Quarterly, 38, 98-106. Brown, S. P., & Stayman, D. M. (1992). An tecedents and Consequences of Attitude toward the Ad: A Meta-analysis. Journal of Consumer Research, 19, 34-51. Budd, R. J. (1986). Predicting Cigarette Use: The Need to Incorporate Measures of Salience in the Theory of Reasoned Action. Journal of Applied Social Psychology, 16, 633-685. Cantril, H. (1944). Gauging Public Opinion. Princeton, NJ: Princeton University Press. Cantril, H. (1946). The Intensity of an Attitude. Journal of Abnormal and Social Psychology, 41, 129-135. Chaiken, S., & Baldwin, M. W. (1981). Aff ective-Cognitive Consistency and the Effect of Salient Behavioral Information on the Self-perception of Attitudes. Journal of Personality and Social Psychology, 41, 1-12. Chaiken, S., Pomerantz, E. M., & Giner-Soro lla, R. (1995). Structural Consistency and Attitude Strength. In R. E. Petty & J. A. Krosnick (Eds.), Attitude Strength: Antecedents and Consequences (pp. 387-412). Mahwah, New Jersey: Lawrence Erlbaum Associates, Publishers. Chaiken, S., & Yates, S. (1985). Affectiv e-Cognitive Consistency and Thought-Induced Attitude Polarization. Journal of Personality and Social Psychology, 49(6), 14701481. Chaudhuri, A. (1999). Does Brand Loya lty Mediate Brand Equity Outcomes? Journal of Marketing Theory and Practice, 7(2), 136-146. Chuang, Y. C. (1988). The Structure of Attitude Strength. Unpublished Doctoral Dissertation, Ohio Stat e University, Columbus. Churchill, H. (1942). How to Measure Brand Loyalty. Advertising and Selling, 35, 24. Cialdini, R. B., Levi, A., Herman, C. P., Ko zlowski, L., & Petty, R. E. (1976). Elastic Shifts of Opinion: Determinants of Direction and Durability. Journal of Personality and Social Psychology, 34, 663-672.
109 Converse, P. E. (1970). Attitudes and Non-attitu des: Continuation of a Dialogue. In E. R. Tufte (Ed.), The Quantitative Analys is of Social Problems (pp. 168-189). Boston, MA: Addison Wesley. Copeland, M. T. (1923). Relation of Consumer 's Buying Habits to Marketing Methods. Harvard Business Review, 1(April), 282-289. Cunningham, R. M. (1956). Brand Loyalty-What, Where, How Much? Journal of Marketing, 21, 206. Cunningham, S. M. (1967). Perceived Risk a nd Brand Loyalty. In D. Cox (Ed.), Risk Taking and Information Handling in Consumer Behavior (pp. 132-138). Boston: Harvard University Press. Day, G. S. (1969). A Two-Dimensional Concept of Brand Loyalty. Journal of Advertising Research, 9, 29-35. Davidson, A. R., Yantis, S., Norwood, M., & Montano, D. E. (1985). Amount of Information about the Attitude Objects and Attitude-Behavior Consistency. Journal of Personality and Social Psychology, 49, 1184-1198. Davis, J. A., & Smith, T. W. (1985). General Social Surveys, 1972-1985: Cumulative Codebook. Chicago: National Opinion Research Center. Deighton, J. (1996). The Future of Interactive Marketing. Harvard Business Review, 74(November-December), 151-166. Dick, A. S., & Basu, K. (1994). Customer L oyalty: Toward an Integrated Conceptual Framework. Journal of the Academy of Marketing Science, 22, 99-113. Ding, L., Velicer, W. F., & Harlow, L. L. (1995). Effects of Estimation Methods, Number Indicators per Factor, and Improper Soluti ons on Structural Equation Modeling Fit Indices. Structural Equation Modeling: A Multidisciplinary Journal, 2, 119-144. Donath, B. (1994). Consumer and Biz Marketing Look More Alike. Marketing News, 28(13), 14-17. Downing, J. W., Judd, C. M., & Brauer, M. (1992). Effects of Repeated Expressions on Attitude Extremity. Journal of Personality and Social Psychology, 63, 17-29. Duncan, T., & Moriarty, S. (1997). Driving Brand Value: Using Integrated Marketing to Manage Profitable Stakeholder Relationships. New York: McGraw-Hill. Dwyer, F. R., Schurr, P. H., & Oh, S. ( 1987). Developing Buyer-Seller Relationships. Journal of Marketing, 51(April), 11-27.
110 Eagly, A. H., & Chaiken, S. (1993). The Psychology of Attitudes. Fort Worth, TX: Harcourt College Publishers. Edell, J. A., & Burke, M. C. (1987). The Powe r of Feelings in Unde rstanding Advertising Effects. Journal of Consumer Research, 14, 421-433. Engel, J. F., Blackwell, R. D., & Miniard, P. W. (1990). Consumer Behavior (6th Ed.). Chicago. IL: The Dryden Press. Erber, M. W., Hodges, S. D., & Wilson, T. D. (1995). Attitude St rength and Attitude Stability. In R. E. Petty & J. A. Krosnick (Eds.), Attitude Strength: Antecedents and Consequences (pp. 433-454). Hillsdale, NJ: Erlbaum. Erdem, T., & Swait, J. (1998). Bran d Equity as a Signaling Phenomenon. Journal of Consumer Psychology, 7(2), 131-157. Erdem, T., Swait, J., & Louviere, J. (2002). The Impact of Credibi lity on Consumer Price Sensitivity. International Journal of Research in Marketing, 19(1), 1-19. Fazio, R. H. (1986). How Do Attitudes Guide Behavior? In R. M. Sorrentino & E. T. Higgins (Eds.), The Handbook of Motivation and Cognition: Foundations of Social Behavior (Vol. 1, pp. 204-243). New York: Guilford Press. Fazio, R. H., & Zanna, M. P. (1978). Attitudina l Qualities Relating to the Strength of the Attitude-Behavior Relationship. Journal of Experimental Social Psychology, 14, 398-408. Filser, M. (1994). Le Compor tement du Consommateur, Collection Precis de Gestion. Paris: Dalloz. Fishbein, M., & Ajzen, I. (1975). Belief, Attitude, Inten tion, and Behavior: An Introduction to theory and Research. Boston, MA: Addison-Wesley. Fishbein, M., & Middlestadt, S. E. (1995). Noncognitive Effects on Attitude Formation and Change: Fact or Artifact? Journal of Consumer Psychology, 4, 107-115. Forgas, J. P. (1981). Affective and Emotional Influences on Episode Representations. In J. P. Forgas (Ed.), Social Episodes: The Study of Interaction Routines (pp. 165-180). London and New York: Academic Press. Fornell, C. (1983). Issues in the Applic ation of Covariance St ructure Analysis: A Comment. Journal of Consumer Research, 9, 443-448. Fournier, S. (1998). Consumers and Their Br ands: Developing Relationship Theory in Consumer Research. Journal of Consumer Research, 24(March), 343-373.
111 Fujii, S., & Ryuichi, K. (2000). Evaluating of Trip-Inducing Effects of New Freeways Using a Structural Equations Model System of Commuters' Time Use and Travel. Trasportation Research, Part B, 34, 339-354. Garbarino, E., & Johnson, M. S. (1999). The Di fferent Roles of Satisfaction, Trust, and Commitment in Customer Relationships. Journal of Marketing, 63(April), 70-87. Garfein, R. (1987). Evaluating the Impact of Customer Service Delivery Systems. Marketing Review, 11-15. Gilbert, S. J. (1976). Self-disclosure, Intimacy and Communication in Families, The Family Coordinator, 25, 221-231. Gonzaga, G. C., Keltner, K. D., Londahl, E. A., & Smith, M. D. (2001). Love and the Commitment Problem in Roman tic Relations and Friendship. Journal of Personality and Social Psychology, 81(2), 247-262. Greenwald, A. G., & Leavitt, C. (1984). A udience Involvement in Advertising: Four Levels. Journal of Consumer Research, 11, 581-592. Gross, S. R., Holtz, R., & Miller, N. (1995). At titude Certainty. In R. E. Petty & J. A. Krosnick (Eds.), Attitude Strength: Antecedents and Consequences (pp. 215-245). Mahwah, New Jersey: Lawrence Erlbaum Associates, Publishers. Gruen, T. W., Summers, J. O., & Acito, F. (2001). Relationship Marketing Activities, Commitment, and Membership Behavi ors in Professional Associations. Journal of Marketing, 64(3), 34-49. Grunig, J. E., & Huang, Y. (2000). From Organization Effectiveness to Relationship Indicators: Antecedents of Relationshi ps, Public relations Strategies, and Relationship Outcomes. In J. Le dingham & S. D. Bruning (Eds.), Public Relations as Relationship Management: A Relational Approach to the Study and Practice of Public Relations (pp. 23-54). Mahwah, NJ: Lawrence Erlbaum Associates. Guest, L. P. (1944). A Study of Brand Loyalty. Journal of Applied Psychology, 28, 16-27. Gwinner, K. P., Gremler, D. D., & Bitner, J. M. (1998). Relational Benefits in Service Industries: The Customer's Perspective. Journal of the Academy of Marketing Science, 26(Spring), 101-114. Haddock, G., Rothman, A. J. R., & Schwarz, N. (1999). Forming Judgments of Attitude Certainty, Intensity, and Importance: The Role of Subjective Experiences. Personality and Social Psychology Bulletin, 25, 771-782.
112 Hair, J., Joseph F., Anderson, R. E., Ta tham, R. L., & Black, W. C. (1998). Multivariate Data Analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall. Hennig-Thurau, T., & Klee, A. (1997). Th e Impact of Customer Satisfaction and Relationship Quality on Customer Retent ion: A Critical Reassessment and Model Development. Psychology & Marketing, 14(8), 737-764. Hess, J. S. (1995). Construction and Assessment of a Sc ale to Measure Consumer Trust. Paper presented at the American Market ing Association Educators' Conference, Chicago. Hoch, S. J., & Ha, Y. W. (1986). Consumer Learning: Advertising and the Ambiguity of Product Experience. Journal of Consumer Research, 13, 394-404. Holbrook, M. B., & Batra, R. (1987). Assessi ng the Role of Emotions as Mediators of Consumer Response to Advertising. Journal of Consumer Research, 14, 404-420. Holbrook, M. B., & Lehmann, D. R. (1980). Fo rm Versus Content in Predicting Starch Scores. Journal of Advertising Research, 20, 53-62. Holbrook, M. B., & O'Shaughnessy, J. (1984). The Role of Emoti on in Advertising. Psychology & Marketing, 1, 45-64. Holmes, J. G., & Rempel, J. K. (1989). Trus t in Close Relationshi ps. In C. Hendricks (Ed.), Review of Personality and Social Psychology: Close Relationships (Vol. 10, pp. 187-220). Newbury Park, California: Sage. Holtz, R., & Miller, N. (1985). Assumed Similarity and Opinion Certainty. Journal of Personality and Social Psychology, 48, 890-898. Hox, J. J. (1995). Amos, EQS, and LI SREL for Windows: A comparative review. Structural Equation Modeling: A Multidisciplinary Journal, 2, 79-91. Hu, L. T., & Bentler, P. (1995). Evaluati ng Model Fit. In R. H. Hoyle (Ed.), Structural Equation Modeling. Concepts, Issues, and Applications (pp. 76-99). London: Sage. Huang, Y. (2001). OPRA: A Cross-Cultural, Multiple-Item Scale for Measuring Organization-Public Relationships. Journal of Public Relations Research, 13(1), 61-90. Jacoby, J. (1969). Toward a Multi-Brand Model of Brand Loyalty. Purdue Papers in Consumer Psychology, Paper No. 105. Jacoby, J. (1971). A Model of Multi-Brand Loyalty. Journal of Advertising Research, 11, 25-30.
113 Jacoby, J. (1975). A Brand Loyalty Concept: Comments on a Comment. Journal of Marketing Research, 12(3), 484-487. Jacoby, J., & Chestnut, R. W. (1978). Brand Loyalty: Measurement and Management. New York: John Wiley & Sons. Jacoby, J., & Olson, J. C. (1970). An Attitu dinal Model of Brand Loyalty: Conceptual Underpinnings and Inst rumentation Research. Purdue Papers in Consumer Psychology, No. 159. Joreskog, K. G. (1971). Statistical Anal ysis of Sets of Congeneric Tests. Psychometrika, 36, 109-133. Judd, C. M., & Johnson, J. T. (1981). Attitudes, Polarization, and Diagnosticity: Exploring the Eff ects of Affect. Journal of Personality and Social Psychology, 41, 25-36. Judd, C. M., & Krosnick, J. A. (1982). A ttitude Centrality, Organization, and Measurement. Journal of Personality and Social Psychology, 42, 436-447. Judd, C. M., & Krosnick, J. A. (1989). The Structural Bases of Consistency among Political Attitudes: Effects of Political E xpertise and Attitude Importance. In A. R. Pratkanis & S. J. Breckler & A. G. Greenwald (Eds.), Attitude Structure and Function (pp. 99-128). Hillsdale, NJ: Erlbaum. Kaiser, H. F. (1974). An Index of Factorial Simplicity. Psychometrika, 39, 31-36. Kapferer, J. N., & Laurent, G. (1983). La Se nsibilite aux Marques: Un Nouveau Concept pour Gerer les Marques, Paris: Fondation Jour de France pour la Recherche en Publicite. Kasper, J. D. P. (1988). On Problem Per ception, Dissatisfaction and Brand Loyalty. Journal of Economic Psychology, 9, 387-397. Katz, D., & Stotland, E. (1959). A Prelimin ary Statement to a Theory of Attitude Structure and Change. In S. Koch (Ed.), Psychology: A Study of a Science (Vol. 3, pp. 423-475). New York: McGraw-Hill. Kempf, D. S. (1999). Attitude Formation from Product Trial: Distinct Roles of Cognition and Affect for Hedonic and Functional Products. Psychology & Marketing, 16(1), 35-50. Kempf, D. S., & Smith, R. E. (1998). Cons umer Processing of Product Trial and the Influence of Prior Advertising: A Structural Modeling Approach. Journal of Marketing Research, 35(325-338).
114 Kiesler, C. A. (1971). The Psychology of Commitment: Experiments Linking Behavior to Belief. New York: Academic Press. Kline, R. B. (1998). Principles and Practice of Structural Equation Modeling. New York: Guildford Press. Knox, S., & Walker, D. (2001). Measuring and Managing Brand Loyalty. Journal of Strategic Marketing, 9, 111-128. Kraft, F. B., Granbois, D. H., & Summers, J. O. (1973). Brand Evaluation and Brand Choice: A Longitudinal Study. Journal of Marketing Research, 10, 235-241. Krosnick, J. A. (1986). Policy Voting in American Presidential Elections: An Application of Psychological Theory to American Politics. Unpublished Doctoral Dissertation, University of Michigan, Ann Arbor, MI. Krosnick, J. A. (1989). Attitude Im portance and Attit ude Accessibility. Personality and Social Psychology Bulletin, 15, 297-308. Krosnick, J. A., & Abelson, R. P. (1992). Th e Case for Measuring Attitude Strength in Surveys. In J. M. Tanur (Ed.), Questions about Questions: Inquiries into the Cognitive Bases of Surveys (pp. 177-203). New York: Russell Sage Foundation. Krosnick, J. A., Boninger, D. S., Chuang, Y. C., Berent, M. K., & Carnot, C. G. (1993). Attitude Strength: One Construct or Many Related Constructs? Journal of Personality and Social Psychology, 65, 1132-1151. Krosnick, J. A., & Petty, R. E. (1995). Attitude Strength: An Ov erview. In R. E. Petty & J. A. Krosnick (Eds.), Attitude Strength: Antecedents and Consequences (pp. 1-24). Mahwah, NJ: Lawrence Erlbaum As sociates, Inc., Publishers. Krosnick, J. A., & Shuman, H. (1988). Attit ude Intensity, Importance, Certainty, and Susceptibility to Response Effects. Journal of Personality and Social Psychology, 54, 940-952. Krugman, H. (1966). The Measurem ent of Advertising Involvement. Public Opinion Quarterly, 30. LaBarbera, P. A., & Mazursky, D. (1983). A Longitudinal Assessment of Consumer Satisfaction/Dissatisfaction. Journal of Marketing Research, 20, 393-404. Lang, P. J. (1980). Behavioral Treatment and Bio-Behavioral Assessment: Computer Applications in Technology. In J. P. Sidowski & J. H. Johnson & T. A. Williams (Eds.), Technology in Mental Heal th Care Delivery Systems (pp. 119-139). Norwood, NJ: Albex Publishing Corporation.
115 Lastovicka, J. L. (1979). Questioning th e Concept of Involvement Defined Product Classes. Advances in Consumer Research, 6, 174-179. Laurent, G., & Kapferer, J. (1985). Meas uring Consumer Involvement Profiles. Journal of Marketing Research, 22, 41-53. Lavine, H., Sullivan, J. L., Borgida, E., & Thompson, C. J. (1996). The Relationship of National and Personal Issue Salience to Attitude Accessibility on Foreign and Domestic Policy Issues. Political Psychology, 17, 293-316. Lehmann, D. R. (1996). Another Cup of Coffee: The View from Different Frames. In K. Corfman & J. J. Lynch (Eds.), Advances in Consumer Research (Vol. 23, pp. 309). Provo, UT: Association for Consumer Research. Lipstein, B. (1959). The Dynamics of Brand loyalty and Brand Switching, Better Measurement of Advertising Effectiv eness: The Challenge of the 1960's. New York: Advertising Research Foundation. Long, J. S. (1983). Confirmatory Factor Analysis, Sage University Paper Series on Quantitative Applications in the Social Sciences (Vol. 33). Newbury Park, CA: Sage Publications. Lowenstein, M. W. (1997). The Customer Loyalty Pyramid. Westport, CT: Quorum Books. Machleit, K., & Wilson, R. D. (1988). Emo tional Feelings and Attitude toward the Advertisement: The Roles of Brand Familiarity and Repetition. Journal of Advertising, 17(3), 27-35 Macintosh, G., & Lockshin, L. S. (1997). Retail Relationships and Store Loyalty: A Multi-level Perspective. International Journal of Research in Marketing, 14, 487497. Magnusson, D. E. (1971). An Analys is of Situational Dimensions. Perceptual Motor Skills, 32, 851-867. McAlister, L., & Pessemier, E. A. (1982). Vari ety Seeking Behavior: An Interdisciplinary Review. Journal of Consumer Research, 9, 172-175. McGregor, D. (1940). Motives as a Tool of Market Research. Harvard Business Review, 29, 42-51. McKenna, R. (1991). Relationship Marketing: Successful Strategies for the Age of the Customer. Boston, MA: Addison-Wesley.
116 Mehling, R. (1959). A Simple Test fo r Measuring Intensity of Attitudes. Public Opinion Quarterly, 23, 576-578. Mehrabian, A., & Russell, J. A. (1974). An Approach to Environmental Psychology. Cambridge, Boston, MA: M.I.T. Press. Mehrabian, A., & Wetter, R. D. (1987). E xperimental Test of an Emotion-Based Approach to Fitting Brand Names to Products. Journal of Applied Psychology, 72, 125-130. Millar, M. G., & Tesser, A. (1986). Thought-I nduced Attitude Cha nge: The Effects of Schema Structure and Commitment. Journal of Personality and Social Psychology, 51, 259-269. Miller, R. S., & Lefcourt, H. M. (1982) . The Assessment of Social Intimacy. Journal of Personality Assessment, 46, 514-518. Mitchell, A. A. (1979). Involvement: A Pote ntially Important Mediator of Consumer Behavior. Advances in Consumer Research, 6, 191-196. Mitchell, A. A. (1981). The Dimensi ons of Advertising Involvement. Advances in Consumer Research, 8, 25-30. Mittal, B. (1989). Measuring Purchase-Decision Involvement. Psychology & Marketing, 6, 147-162. Morgan, R. M., & Hunt, S. D. (1994). The Commitment-Trust Theory of Relationship Marketing. Journal of Marketing, 58(July), 20-38. Morris, J. D., Bradley, M., Sutherland, J., & Wei, L. (1993). Assessing Cross Cultural Transferability of Standardized Global Advertising: An Emotional Response Approach. Paper presented at the Association for Education in Journalism and Mass Communications, Kansas City. Morris, J. D., Bradley, M., Waine, C. A., & Lang, J. B. (1992). Assessing Affective Reactions to Advertisements with the Self-assessment Manikin (SAM). Paper presented at the Southern Mark eting Association Conference. Morris, J. D., & Waine, C. (1994). Managing the Creative Effort: Pre-production and Post-production Measures of Emotional Response. Unpublished manuscript, College of Journalism and Communications, University of Florida, Gainesville. Morris, J. D., Woo, C., Geason, J., & Kim, J. (2002). The Power of Affect: Predicting Intention. Journal of Advertising Research, 42, 7-17.
117 Newman, J. W., & Werbel, R. A. (1973). Mult ivariate Analysis of Brand Loyalty for Major Household Appliances. Journal of Marketing Research, 10, 404-409. Niedenthal, P., & Halberstadt, J. (2000). Grounding Categories in Emotional Response. In J. P. Forgas (Ed.), Feeling and Thinking: The Role of Affect in Social Cognition (pp. 357-386). New York, NY: Cambridge University Press. Noorman, C., Deshpande, R., & Zaltman, G. (1992). Relationships between Providers and Users of Market Research: The Dynamics of Trust Within and Between Organizations. Journal of Marketing, 58(July), 20-38. Norman, R. (1975). Affective-Cognitive C onsistency, Attitude s, Conformity, and Behavior. Journal of Personality and Social Psychology, 32, 83-91. Nowlis, S. M., Kahn, B. E., & Dhar, R. (2002). Coping with Ambivalence: The Effect of Removing a Neutral Option on Consumer Attitude and Preference Judgments. Journal of Consumer Research, 29(3), 319-334. Nunnally, J. C. (1978). Psychometric Theory (2nd Ed.). New York: McGraw-Hill. O'Connor, P. (1992). Friendships between Women: A Critical Review. New York, NY: Guilford Press. Odin, Y., Odin, N., & Valette-Florence, P. ( 2001). Conceptual and Operational Aspects of Brand Loyalty: An Empirical Investigation. Journal of Business Research, 53(2), 75-84. Oliver, R. L. (1997). Satisfaction: A Behavioral Perspective on the Customer. Boston: McGraw-Hill. Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The Measurement of Meaning. Urbana: University of Illinois Press. Parasuraman, A., Zeithaml, V. A., & Be rry, L. L. (1988). SERVQUAL: A Multi-Item Scale for Measuring Consumer Perceptions of Service Quality. Journal of Retailing, 64(Spring), 12-40. Pervin, L. A. (1976). A Free Response Desc ription Approach to the Study of PersonSituation Interaction. Journal of Personality and Social Psychology, 35, 465-474. Pessemier, E. A. (1959). A New Wa y to Determine Buying Decisions. Journal of Marketing, 24, 41-46. Peterson, R. A. (2001). On the Use of Colle ge Students in Social Science Research: Insights from a Second-Order Meta-Analysis. Journal of Consumer Research, 28(3), 450-461.
118 Petty, R. E., & Cacioppo, J. T. (1981). Issue I nvolvement as a Moderator of the Effects on Attitude of Advertising Content and Context. Advances in Consumer Research, 8, 20-24. Pham, M. T., Cohen, J. B., Pracejus, J. W., & Hughes, G. D. (2001). Affect Monitoring and the Primacy of Feelings in Judgment. Journal of Consumer Research, 28(September), 167-188. Pomerantz, E. M., Chaiken, S., & Tordesi llas, R. S. (1995). Attitude Strength and Resistance Process. Journal of Personality and Social Psychology, 69(3), 408-419. Pragner, K. J. (1995). The Psychology of Intimacy. New York, NY: Guilford Press. Raden, D. (1985). Strength-Related Attitude Dimensions. Social Psychology Quarterly, 48, 312-330. Raftery, A. E. (1993). Bayesian Model Selectio n in Structural Equati on Models. In K. A. Bollen & J. S. Long (Eds.), Testing Structural Equation Models (pp. 163-180). Newbury Park, California: Sage. Ratchford, B. T. (1987). New Insights about the FCB Grid. Journal of Advertising Research, 27(4), 24-38. Reynolds, F. D., & Wells, W. D. (1977). Consumer Behavior. New York: McGraw-Hill. Reynolds, K. E., & Beatty, S. E. (1999). Cust omer Benefits and Company Consequences of Customer-Salesperson Re lationships in Retailing. Journal of Retailing, 75(1), 11-31. Robertson, T. S. (1976). Low Commitment Consumer Behavior. Journal of Advertising Research, 16(April), 19-27. Rosenberg, M. J. (1956). Cognitive St ructure and Attitudinal Affect. Journal of Abnormal and Social Psychology, 53, 367-372. Rosenberg, M. J. (1968). Hedonism, Inauthentic ity, and Other Goads toward Expansion of a Consistency Theory. In R. P. Abel son & E. Aronson & W. J. McGuire & T. M. Newcomb & M. J. Rosenberg & P. H. Tannenbaum (Eds.), Theories of Cognitive Consistency: A Sourcebook (pp. 73-111). Chicago: Rand McNally. Rusbult, C. E., Martz, J. M., & Agnew, C. R. (1998). The Investment Model Scale: Measuring Commitment Level, Satisfacti on Level, Quality of Alternatives, and Investment Size. Personal Relationships, 5, 357-391.
119 Russell, J. A., & Mehrabian, A. (1977). Eviden ce for a three-factor theory of emotions. Journal of Research in Personality, 11, 273-294. Schwarz, G. (1978). Estimating the Dimension of a Model. The Annals of Statistics, 6, 461-464. Scot, B., & Lichtenstein, D. R. (1988). The Effect of Ad Claims and Ad Content on Attitude toward the Advertisement. Journal of Advertising, 17(Spring), 3-11. Scott, W. A. (1968). Attitude Measuremen t. In G. Lindzey & E. Aronson (Eds.), Handbook of Social Psychology (Vol. 2, pp. 204-273). Boston, MA: AddisonWesley. Sexton, R. E., & Sexton, V. E. (1982) . Intimacy: A Historical Perspective, Intimacy. New York, NY: Plenum. Sherif, M., & Cantril, H. (1947). The Psychology of Ego-Involvement. New York: John Wiley. Sherry, J. (1987). Cereal Monogamy: Brand Loyalty as Secular Ritual in Consumer Culture. Paper presented at the Associati on for Consumer Research, Boston. Sheth, J., & Parvatiyar, A. (1995). Relati onship Marketing in Consumer Markets: Antecedents and Consequences. Journal of the Academy of Marketing Science, 23(Fall), 255-271. Shimp, T., & Madden, T. (1988). Consumer-O bject Relations: A Conceptual Framework Based Analogously on Sternberg's Triangular Theory of Love. In M. Houston (Ed.), Advances in Consumer Research (Vol. 15, pp. 163-168). Provo, UT: Association for Consumer Research. Slama, M. E., & Taschian, A. (1985). Se lected Socio-Economic and Demographic Characteristics Associated with Purchase Involvement. Journal of Marketing, 49(Winter), 72-82. Speller, D. E. (1973). Attitudes and Intentions as Pr edictors of Purchase: A Cross Validation. Paper presented at the Ameri can Psychological Association. Sternberg, R. J. (1988). Triangulating Love. In R. J. Sternberg & M. L. Barnes (Eds.), The Psychology of Love (pp. 120-138). New Haven: Ya le University Press. Stouffer, S. A., Guttman, L., Suchman, E. A., Lazarsfeld, P. F., Star, S. A., & Clausen, J. A. (1950). Measurement and Prediction. Princeton, NJ: Princeton University Press. Stout, P. A., & Leckenby, J. D. (1986). Meas uring Emotional Response to Advertising. Journal of Advertising, 15(4), 35-42.
120 Stout, P. A., & Rust, R. T. (1986). The Effect of Music on Emotional Response to Advertising. Paper presented at the Ameri can Academy of Advertising. Suchman, E. A. (1950). The Intensity Component in Attitude and Opinion Research. In S. A. Stouffer & L. Guttman & E. A. Suchman & P. F. Lazarsfeld & S. A. Star & J. A. Clausen (Eds.), Measurement and Prediction (pp. 213-276). Princeton, New Jersey: Princeton University Press. Tannenbaum, P. H. (1956). Initial Attitude toward Source and Concept as Factors in Attitude Change Through Communication. Public Opinion Quarterly, 20, 413-425. Tiedens, L. Z., & Linton, S. (2001). Judgment under Emotional Certainty and Uncertainty: The Effects of Specific Emotions on Information Processing. Journal of Personality and Social Psychology, 81(6), 973-988. Tourangeau, R., Rasinski, K. A., Bradbur n, N., & D'Andrade, R. (1989). Belief Accessibility and Context Effects in Attitude Measurement. Journal of Experimental Social Psychology, 25, 401-421. Traylor, M. B. (1981). Product Invo lvement and Brand Commitment. Journal of Advertising Research, 21, 51-56. Tsal, Y. (1985). On the Relationship be tween Cognitive and Affective Process: A Critique of Zajonc and Markus. Journal of Consumer Research, 12, 358-362. Tucker, L. R., & Lewis, C. (1973). A Relia bility Coefficient for Maximum Likelihood Factor Analysis. Psychometrika, 38, 1-10. Vakratsas, D., & Ambler, T. (1999). How A dvertising Works: What Do We Really Know? Journal of Marketing, 63(1), 26-43. Verplanken, B. (1989). Involvement and Need for Cognition as Moderators of BeliefsAttitude-Intention Consistency. British Journal of Social Psychology, 28, 115-122. Webster, F. E. (1992). The Changing Ro le of Marketing in the Corporation. Journal of Marketing, 56(October), 1-17. Wieselquist, J., Rusbult, E., Foster, C. A ., & Agnew, C. R. (1999). Commitment, ProRelationship Behavior, and Tr ust in Close Relationships. Journal of Personality and Social Psychology, 77(5), 942-966. Wilson, T. D., Hodges, S. D., & Pollack, S. E. (1990). Effects of Explai ning Attitudes on Survey Responses. Charlottesville: University of Virginia.
121 Wilson, T. D., Kraft, D., & Dunn, D. S. (1989). The Disruptive E ffects of Explaining Attitudes: The Moderating Effect of Knowledge about the Attitude Object. Journal of Experimental Social Psychology, 25, 379-400. Wood, W. (1982). Retrieval of Attitude-Relevant Informa tion from Memory: Effects on Susceptibility to Persuasion and on Intrinsic Motivation. Journal of Personality and Social Psychology, 42, 798-810. Woods, W. (1960). Psychological Dimen sions of Consumer Decision. Journal of Marketing, 24, 15-19. Young, J. W. (1977). The Function of Theo ry in a Dilemma of Path Analysis. Journal of Applied Psychology, 62, 108-110. Young, L. C., & Denize, S. (1995). A Concep t of Commitment: Alternative Views of Relational Continuity in Business Service relationships. Journal of Business and Industrial Marketing, 19(5), 47-62. Zajonc, R. (1980). Feeling and Thinking: Preferences Need No Inferences. American Psychologist, 35, 151-175. Zajonc, R. (2000). Feeling and Thinking: Closin g the Debate on the Primacy of Affect. In J. P. Forgas (Ed.), Feeling and Thinking: The Role of Affect in Social Cognition (pp. 31-58). New York, NY: Cambridge University.
122 BIOGRAPHICAL SKETCH Jooyoung Kim was born in South Korea on April 19, 1970. He received his Bachelor of Arts degree in economics at Hong-Ik University (Seoul, South Korea) in 1996. After two years of indus try experience at a large fa shion company (ShinWon Corp. Inc.) in Korea, he came to the United States and completed a masterÂ’s degree specializing in integrated marketing communi cations at the University of Colorado at Boulder in 2000. During his three years in the Ph.D. program at the University of Florida, he has been collaborating primarily with Jon Morris on several research papers. In 2002 and 2003, Jooyoung taught international advertising and advertising campaigns courses at the University of Florida. He has now accepted a position as an Assi stant Professor of Advertising at the Greenl ee School of Journalism and Communication, Iowa State University.