Philosophy and Consumption: Discussions on Trust and Brands

ASSOCIATION FOR CONSUMER RESEARCH

Labovitz School of Business & Economics, University of Minnesota Duluth, 11 E. Superior Street, Suite 210, Duluth, MN 55802

Philosophy and Consumption: Discussions on Trust and Brands Shona Bettany, Bradford University, UK Susan Dobscha, Bentley College, USA Marcus Giesler, York University, Canada

Conversations continue around Philosophy and Consumption after last year’s roundtable attracted 45 participants, encouraging consumer researchers to reflect on philosophical concepts, frameworks, or philosophical figures that have impacted, and continue to impact, the way each of us asks questions, conceptualizes problems, and carries out our arguments and analysis. This session spurs interaction between scholars with a variety of interests and research paradigms who are engaged –whether implicitly or explicitly–in philosophically informed consumer research. Examples include research on tragedy; constructions of “consumers that matter”; and consumer subjectivities constituted through interaction with brands and marketing representations.

[to cite]: Shona Bettany, Susan Dobscha, Marcus Giesler, and Kent Grayson, Ashlee Humphreys, Krittinee Nuttavuthisit, Rob Kleine, Jonathan Schroeder, Alladi Venkatesh, Clara Gustafsson (2007) ,"Philosophy and Consumption: Discussions on Trust and Brands", in NA - Advances in Consumer Research Volume 34, eds. Gavan Fitzsimons and Vicki Morwitz, Duluth, MN : Association for Consumer Research, Pages: 703-749.

[url]: http://www.acrwebsite.org/volumes/12730/volumes/v34/NA-34

[copyright notice]: This work is copyrighted by The Association for Consumer Research. For permission to copy or use this work in whole or in part, please contact the Copyright Clearance Center at http://www.copyright.com/. Progressing a Taxonomy of Possible Reflexivities: Guidelines for Reflexive Practice in Consumer Research Shona Bettany, University of Bradford, UK Helen Woodruffe-Burton, University of Lancaster and University College of St. Martins, UK

EXTENDED ABSTRACT encounter; this position, we believe, reflects closely key issues in Reflexivity has been identified in consumer research contemporary consumer research. In terms of ontology, we re- (Wallendorf and Brucks, 1993, Hirschman 1993, Thompson 2002) flected upon reflexivity as it has been seen as a route to better and in the wider social sciences (e.g. Wasserfall, 1993, Mauthner knowledge and enhanced theory building through reflection on the and Doucet, 1998, 2003) as a way to address power and control in process of research. This to us suggested questions and challenges the research encounter, to attend to the researcher/researched dy- arising pertaining to the ‘reality’ of the research process, not just in namic and to give insightful commentaries into the research process terms of what ‘really’ happened, but the models of reality implied itself. The ‘reflexive turn’ (Weick 1999) has emerged in the social in different reflexive approaches. Different types of reflexivity, like sciences largely as a response to the notion that data collection and different research approaches, carry their own ontological assump- analysis is a neutral activity, a technology of knowledge making in tions about the nature of subjects (researcher and researched) and which the researcher can take a disinterested, objective position in what knowledge is and how it is constructed. These assumptions are relation to their research. Reflexivity instead is used to highlight the clearly crucial to the conceptualisation of multiple reflexivities and personal, interpersonal, emotional, institutional, disciplinary and render attempts to achieve closure around the notion of reflexivity pragmatic influences on the research, as well as the methodological, as a singular concept highly problematic. Attending to ontology as epistemological and ontological assumptions built in to specific one of the axes in our taxonomy allowed us to structure our approaches (Mauthner and Doucet 2003). As such, reflexivity has theorising to account for this multiplicity of reflexivities and the been seen a route to more accountable and responsible research, and constitutive and generated realities they imply. In terms of power, as a stimulus to better accounts of the social and cultural world. this pertains to reflexivity as a route to addressing the power Researcher reflexivity has been conceptualised in consumer imbalances and issues inherent in any research encounter, whether research discourse as a sub type of introspection. It is argued that that be conceptualised in terms of an overt remit for an emancipatory there is much to recommend reflexivity as a practice in consumer research politics, or as a way to direct the researcher encounter research as it will ‘improve our understanding of the actual research towards one with less of a dominating and instrumental dynamic. process’ and act as such to develop consumer research theory For us, reflexivity is inherently political; our exposure to it initially (Wallendorf and Brucks 1993:355). Additionally it has been seen as emerged from our own engagement with feminist theory and our a way to address the power issues in consumer research and identification as feminist researchers. Reflexivity has been seen by specifically the instrumentality of the researcher/respondent rela- feminist researchers (and other critical researchers) as a political tionship. For example, Hirschman (1993:551) advocates a more tool for critically engaging with power issues in research reflexive approach to research, and argues against the use of (Ramazanoglu and Holland 2002). Underpinning the different detached and ‘objective’ methods in consumer research. She ex- types of reflexivity in our taxonomy are assumptions relating to the horts consumer researchers to ‘choose to abandon any method or nature of power in the researcher/respondent relationship and these practice that is premised on power inequalities between researcher different reflexivities also generate different (and sometimes am- and subject’ she argues that research should be designed which bivalent) power relations in the research encounter. In terms of this recognises respondents as ‘equal sentient beings’. constitutive and generative relation to power we found that the Although it is clear that consumer researchers have found the taxonomy of reflexivities was a useful tool to reflect in a more idea of reflexivity attractive and recognised its potentials, there is structured manner upon the shifts in the researcher/respondent a need to develop the discourses already apparent in consumer power dynamic which occurred at different parts of our own research and formalise the conceptualisation and operation of this research. As Wolf (1996) has argued, power differentials are an research approach. Currently, despite significant discourse around inevitable part of research, and the necessity is to critically reflect the issue in the wider social sciences there seems to be a lack of upon these inevitable differentials and the way that these shift and discourse around reflexivity in consumer behaviour research and transform as the research progresses. Attending to power as one although reflexive practice is evident in some work, the research axis of our taxonomy allowed us to structure our thinking about processes around reflexive research are not articulated clearly, and reflexivity in terms of these shifting power relations. a plethora of different approaches are evident. The aim of this paper This paper progresses the authors’ recent work on researcher is to stimulate essential new discourse around the practice of reflexivity which has been developed and augmented with practical reflexivity. In order to do this it is necessary to formalise and guidelines for the different approaches to reflexivity. Guidelines for structure reflexive practices as a foundation and basis for discus- practical action and points for reflection are given which relate to sion. This work examines the many different reflexivities which each of the possible reflexivities outlined in the taxonomy. might be enacted within consumer research and their concomitant implications and foundations. To accomplish this it provides a taxonomy of ‘possible reflexivities’ to structure the debate around this element of consumer research and to guide the potentially reflexive consumer researcher. The possible reflexivities which emerge from this study are envisaged as ‘Objectivist Reflexivity’, Experiential Reflexivity’, ‘Perspectival Reflexivity’ and ‘Multi- plex Reflexivity’. The possibilities we have identified are distinguished by the dimensions of ontology and power in the research

703 Advances in Consumer Research Volume 34, © 2007 Auspiciousness: Coping with Kratophany Karen V. Fernandez, University of Auckland, New Zealand Ekant Veer, University of Auckland, New Zealand John L. Lastovicka, Arizona State University, USA

The sacred, described as “more significant, powerful and benevolent, and some may be largely seen as malevolent, all sacred extraordinary than the self” (Belk, Wallendorf and Sherry, 1989, entities have the potential to be both benevolent and malevolent, p.13) is distinguished from the merely special by the key character- provoking mixed feelings of awe and terror (Belk et al. 1989). We istic of efficacy. Belk et al. (1989) also note that the sacred should surmise from this that kratophany exists because the power of the be approached with care, because the sacred has kratophanous sacred does not operate on a simple continuum. Instead, sacred power. Kratophany—the simultaneous devotion to, and fear of, the power can be manifest along two orthogonal dimensions—a di- sacred (Pimentel and Reynolds 2004)—is experienced by consum- mension of good-sacredness and a dimension of bad-sacredness. It ers because the efficacious power of the sacred is unpredictably is the two-dimensional nature of sacred power, coupled with its multi-valenced: it can be manifest in a benevolent and/or malevo- unpredictability, which generates the tension inherent in kratophany. lent way (Belk and Wallendorf 1990). Accordingly, when consum- Furthermore, even when particular sacred entities are usually ers seek the benevolence of the good-sacred, they likely also expose perceived of as wholly benevolent, these entities tend to be ap- themselves to the malevolence of the bad-sacred (Belk and proached with care, because the power of the sacred, irrespective of Wallendorf 1990). Consequently, consumers can simultaneously its valence, is understood to be dangerous to ordinary beings. It experience strong positive feelings (e.g. devotion) and strong seems logical that consumers would employ coping mechanisms to negative feelings (e.g. fear) towards the sacred (Belk et al. 1989; deal the kratophanous power of the sacred. Accordingly, we ask Pimentel and Reynolds 2004). what these forms these coping mechanisms might take. Belk et al. (1989, p.8) suggest that ritual prepares profane Belk and Wallendorf (1990) examine one such kratophanous persons to approach the sacred, and surrounds the contact of these sacred entity—money—in depth, and suggest that money can be profane persons with the sacred to forestall the unleashing of the perceived as kratophanous by consumers because it is viewed as evil powers of the sacred. We believe this implies that consumers having both malevolent and benevolent powers. We surmise from require ways to cope, when faced with the kratophanous power of their work that if sacred power is NOT understood or perceived as the sacred. Consumer researchers have not yet devoted much multi-valenced (for example a child’s faith in a wholly benevolent attention to discovering these coping mechanisms. Consequently, God), then kratophany may be considerably lessened. Hence it is we begin to fill this gap in the consumer research literature by possible that consumers could cope with kratophany by choosing to examining how Hindu Indians use auspiciousness to cope when view the sacred entity is purely benevolent (leading to a pure faced with the tension inherent in their kratophanous reactions to approach orientation) or by viewing the sacred entity as purely the sacred. malevolent (leading to a pure avoidance orientation). To be precise, in these situations, choosing to perceive the sacred entity as wholly THEORETICAL FOUNDATIONS positive or wholly negative collapses the good-sacred bad-sacred dialectic to a simple continuum anchored on either end by the The Meaning of Kratophany benevolent good-sacred and the malevolent bad-sacred respec- The term kratophany was introduced into the English lan- tively. However, a consumer with an approach orientation to the guage by Mircea Eliade, as a technical term meaning “an appear- sacred would still need to undertake further preparatory rituals such ance of the sacred in which the experience of power dominates” as self-cleansing, because of the danger inherent in any contact with (Miller 2005). Belk et al. (1989) introduced kratophany to the sacred power. consumer literature as one of the twelve properties of the sacred, Two other consumer research studies have briefly discussed and explained kratophany as the ability of the sacred to elicit both kratophany using the illustration of consumers’ relationships with strong approach and strong avoidance tendencies. More recently, inherited family heirlooms (Curasi, Price and Arnould 2004) and Pimentel and Reynolds (2004) defined kratophany as the simulta- their college football teams (Pimentel and Reynolds 2004) respec- neous devotion to, and fear of, the sacred. This later definition, tively. In both cases, consumers highly value something (the which focuses on consumers’ reactions to the sacred, is consistent heirloom or the team’s victory) while simultaneously fearing the with the way the term has been used in more recent consumer loss of that thing. These two examples attest to the tension caused research (e.g. Curasi, Price and Arnould 2004). Our reading of the by the simultaneous antithetical reactions to the sacred, suggesting relevant extant literature suggests that consumer researchers’ that when consumers do perceive sacred power as valenced, it is this progress in understanding of kratophany may have been impeded uncertainty as to the outcome in a particular instance that causes the because kratophany has been used to describe both the sacred, and tension associated kratophany. For example, a confident custodian consumers reactions to the sacred. We adopt Pimentel and Reynold’s of family heirlooms or a devoted fan who is aware that loss can (2004) definition of kratophany, and carefully distinguish between occur but who is sure that loss will not occur, will likely not “the kratophanous power of the sacred” (a property of the sacred) experience kratophany. This suggests to us that a more sophisti- and kratophany (consumers’ strong, mixed reactions when faced cated coping mechanism might involve attempts to reduce uncer- with the sacred’s kratophanous power), throughout this paper. tainty. We believe that the Hindu Indian (“Hindu”) concern with auspiciousness may shed light on how consumers cope with Coping with Kratophany kratophany by reducing uncertainty. Although Belk et al. (1989) give different examples of sacred entities associated with the benevolent good-sacred (e.g. gods) and Auspiciousness malevolent bad-sacred (e.g. corpses) powers respectively, they Auspiciousness is a favorable state that bodes well for the then note that consumers fear malevolence, at the same time as they future (Inden 1985). According to Inden, the Roman Empire seek benevolence. Although some entities may be largely seen as practice of augury—interpreting signs from nature in order to 704 Advances in Consumer Research Volume 34, © 2007 Advances in Consumer Research (Volume 34) / 705 TABLE 1 Auspicious and Inauspicious Elements in Hindu Discourse1

Auspicious Inauspicious

Days Monday, Wednesday, Tuesday, Thursday

Colors Red Black

Numbers Odd (“Uneven”) Even

Directions East South

People Pregnant Woman Widow

Rituals Marriage Funeral

1Based on information in Das 1982; Srinivas 1952; and Yagi (1999)

determine the will of the gods—gave us the word auspicious, which similar vein, Inden and Nicholas 1977) define the Hindu term comes from the Latin noun avis (bird) and the Latin verb spicere (to mangala to mean luck or well-being. Srinivas, on the other hand, look at). Although we shall use the Hindu notion of auspiciousness writes that “mangala [an emic term referring to the marriage to illustrate our discussion, we note for the record that auspicious- ceremony which involves the tying of a sacred thread or mangala ness is also an important conceptual category in other Asian sutra around the bride’s neck] is an auspicious or good-sacred cultures. Furthermore, we note that in contrast to Western consum- ceremony” (1952, p.74), and later implies that inauspiciousness ers’ idiosyncratic use of rituals designed to generate favorable relates to bad-sacredness. Thus, we seek to clarify the relationship outcomes (such as personalized pre-game rituals undertaken by of auspiciousness to the Western concept of “luck” and its some professional sportspeople), auspiciousness systematically relationship(s) to the benevolent and the malevolent sacred. pervades Hindu Indians’ daily lives. Hindu Indians’ shared understanding of auspiciousness also suggests that such rituals are more METHOD likely to be publicly employed in India, as compared to the West. We utilize data from depth interviews conducted with twelve Almost every thing, creature, person, act, or event can be described Hindu Indian informants. Two of our informants were resident in in Hindu discourse as either auspicious or inauspicious. Anything India, two were resident in Australia, and eight were resident in that predisposes the gods to favor a human undertaking is auspi- New Zealand. The ten non-resident Hindu Indians we interviewed cious. People, events, objects, words, numbers and points in time had spent the majority of their adult lives in India, had married can be more or less auspicious. Hindu Indian spouses in India, and continued to maintain strong ties Table 1 draws on the work of Indian sociologists to present with India after immigrating to Australasia. Our informants, eight examples of auspicious and inauspicious elements in Hindu of whom were female, ranged in age from 24 to 77. In keeping with discourse. As indicated in the table, Das (1982) points out that interpretive research norms, we utilize the use of pseudonyms to among Hindus, marriage and other rituals pertaining to life are preserve our informants’ anonymity. We employed the constant regarded as auspicious while cremation and other rituals pertaining comparative method of analysis, engaging in open ended and axial to death are viewed as inauspicious. Consequently in her view, coding of interview transcripts and fieldnotes and developing auspiciousness is defined as pertaining to life and inauspiciousness interpretations after each wave of data collection. Our emergent as pertaining to death. In a similar vein, Marglin (1985) suggests insights then informed and directed the next phase of data collec- that auspiciousness is reflected in festivities and inauspiciousness tion. Our interpretations in each phase were informed by the in mourning. Narayanan’s (1985) more nuanced example of the relevant literature. auspicious death of an elderly man pre-deceased by his wife and survived by his progeny suggests that inauspiciousness is anything FINDINGS that prevents the extended family from carrying out its dharma or Hindus utilize as many auspicious elements and signs as destiny of biological immortality (Channa 2000). possible, and avoid anything remotely inauspicious. Auspicious Every married Hindu woman, during her husband’s lifetime, times, places, persons, colors, and objects seem to be valued by is perceived to be the concrete embodiment of temporal auspicious- Hindus because they attract the benevolence of cosmic powers. ness, as shown in her colorful attire, jewelry and the red dot on her forehead (Carman 1985). But, as soon as she becomes a widow, she Auspiciousness is a control mechanism is viewed as inauspicious, and generally avoided (Srinivas 1952). Our informants found it easy to list objects, times or places that Yagi explains the difference in auspiciousness between the two as are auspicious and those which are inauspicious. We spoke to a function of their ability to bear “fruit” (1999 p.275). Hence Yagi Sudha (IF 77) and her daughter-in-law Shaila (IF 39) about the (1999) views auspiciousness as conduciveness to future prosperity, meaning of auspiciousness. much like the concept of “lucky” numbers found elsewhere. In a 706 / Auspiciousness: Coping with Kratophany Sudha: auspicious events mean something good… the use of auspicious artifacts facilitates this movement between Shaila: good liminal and bounded states (Das 1982). Accordingly, several of our Sudha: ….inauspicious events are the opposite of that.” informants described preparations for a typical wedding. First, a Shaila: “if we want to do something, we note it from the priest is consulted to select the most auspicious date and time for the calendar–in India we get calendars that show certain wedding, with respect to the birthdates of the bride and groom. Prior dates that are auspicious...lucky days! It’s something to the wedding, the bride, groom, and any ritual objects to be used to do with lucky and unlucky. in the ceremony are purified while the selected venue is physically Int: lucky and unlucky? and ritually cleansed and decorated. Often a dais is erected in a hall Sudha: we may see from the calendar which days are good. or under an awning outside the home. The location in physical space Shaila: we do the same with numbers too. Some are lucky and that houses the birth, marriage or death acts as what Eliade (1959) some are unlucky. Nine is a lucky number… terms an imago mundi (a representation of the cosmos on the earth). Sudha: seven also. For example, during the ceremony, the bride and groom often sit on a mandap, a raised platform, under a canopy with four pillars. The According to these comments, auspicious bears some relation- material directly over the place where the bride and groom sit is ship to “lucky” and “good”. However, another informant was often an auspicious red color. The Hindu bride is most often dressed careful to distinguish the concept of auspiciousness from that of in a new, auspiciously red sari, and adorned with auspicious luck. To Geeta (IF 44) an auspicious object is more than a lucky jewelry. Other ritual attempts to generate auspiciousness are made. object, as luck implied (to her) something that cannot be controlled. Vijay, recollected his own wedding day in India: Instead, auspiciousness was explained as something that helped a person increase the chances of a favorable outcome by increasing “The hawan [sacred fire altar] was prepared in the center of the the probability the gods would approve of their undertaking. This largest room in my ancestral home. As the wedding began, seems to imply that Hindus attempt to exert control over their lives Anita [the bride] and I were seated in front of the hawan …and by using auspicious elements and signs. According to Geeta, we added ghee [clarified butter] to it when instructed by the pandit [priest]. He would sing mantras [invocations] and after “a Hindu believes that there are definitely some things that are each one we’d add some ghee and the guests would add better than others and some times that are far better than other sawdust [sweet smelling wood grains].” times, they are the auspicious times. There are definitely times that are bad or things that are bad that as a Hindu we must Just as the central mall in Heritage Village was designed to allow its avoid. These are the things that have not met with God’s visitors to experience a time and space separate from the profane approval. Who are we to decide for ourselves what we want to world without (O’Guinn and Belk 1989), the transcendence of do? Inauspicious times are the worst times or things possible– spatio-temporal boundaries is also experienced at the focal point of yes there are ‘neutral’ times or things but they are avoided for the Hindu wedding ceremony, the hawan. The perfumed smoke of the big decisions as they can be seen as just as bad as the the fire, the burning of incense and the chanting of mantras all serve inauspicious times. If we have three choices, have God on our to transport the participants to the liminal zone of the axis mundi, side; have him against us or have him not care, I would like where they may commune with the sacred. The hawan is sacralized God on our side all the time. If he does not care about our by the addition of ghee [clarified butter] which is considered the decision either way then it is not auspicious, he is not against most distilled essence of the cow which is revered as sacred by us so it is not inauspicious or dirty, but he is not in approval– Hindus (Korom 2000). therefore for the big decisions we cannot go forward.” Sudha explained to us that the date for a marriage is fixed after checking that it is an auspicious date for both groom and bride (as Geeta’s observation–that small tasks can be undertaken in neutral determined from their respective birthdates). However, she elabo- times, but important tasks cannot—implies to us that while it may rated “Fate picks the date of death but within that you can pick your be acceptable to risk failure with regard to trivial tasks, important timings [for the cremation]. But if possible they do the cremation on tasks need every assistance possible, so as to ensure success. As the day of death. But they can pick mornings or evenings depending Dinesh (IM 55) put it, auspicious objects are like “….runway lights on which is more auspicious..” We learnt from Sudha that since the for a plane–the better the light, the more chance there is that Ganesh date of births and deaths cannot usually be chosen, special care must [a Hindu god] will visit us and give us success.” We surmise from be taken to removed auspiciousness e.g. by engaging a religious Dinesh’s observation that auspiciousness can serve to beckon the leader to engage in special prayers in order to “stop bad things from benevolence of the sacred. happening”. Later Vijay (IM 25) elaborated, “…the gods are often called We surmise from Sudha’s comments that during dangerous upon for daily as well as life altering events. However, the big times of transition such as birth, marriage and death, auspicious decisions require far more effort and planning—the women are the symbols do more than just attract the benevolent attentions of the ‘prayer warriors’ for the family and are seen as responsible for good-sacred—they also serve to repel the malevolent attentions of the[ir] husbands’ health.” As we shall show, this is because auspi- the bad-sacred and/or protect from the danger inherent in sacred ciousness can be transferred from one entity to another, and hence power. allowing the task of generating auspiciousness to be sub-contracted to another. Auspiciousness can be sub-contracted Many Hindu marriages involve the tying of a sacred necklace Auspiciousness can protect made of gold and black beads, called a Mangal Sutra—literally In traditional Hindu belief, the bride and groom ascend to a “auspicious thread”—(in North India) or a tali (in South India) temporary elevated state, and are believed to be the embodiment of around the bride’s neck. Maya (IF 44), a medical professional now deities (in the case of Brahmins) or royalty (in the case of non- residing in Australasia, was asked about her Mangal Sutra. Maya Brahmins) during the marriage ceremony (Dumont 1972). Situating told the interviewer that she was planning to buy a bigger and more a ritual performance in an auspicious space and time while employing expensive one when she made her next trip to India. This more Advances in Consumer Research (Volume 34) / 707 elaborate Mangal Sutra was to be purchased to serve as evidence of 2001) of auspicious wedding jewelry from a widow, is an obvious her now improved financial status. When asked how she would reversal of the adorning of the bride, to prevent attracting the dispose of the old one, she reacted in surprise, “Oh you can NEVER attention of bad-sacred powers. [emphasis present] sell it, you have to melt the old one and use the The new bride is adorned by the groom’s family with auspicious gold to help make the new one. If not the new one will not be jewelry to anchor her to her new identity. Once the bride is absorbed genuine…it will not be auspicious”. It appears that the sacred into the groom’s self, she is does not need to transcend the powers contained in the Mangal Sutra reside in the gold and black boundaries of the groom’s extended family but enters it as part of beads it is made of, and can be transported by incorporating said the groom. Consequently, the extended family’s boundaries are gold and beads into another Mangal Sutra. maintained intact throughout the process—the groom ascends to a This ability to transfer auspiciousness from one element to higher cosmic plane temporarily and returns to the extended family another, coupled with Vijay’s remark that “women were the prayer augmented with the bride as his “half-body.”When the new wife warriors of the family” led us to ask whether auspiciousness be utilizes auspicious symbols to attract the benevolence of the gods transferred between individuals as well. Can auspiciousness be for the benefit of her husband and his extended family, she is also generated by one party, while its benefits are enjoyed by another? helping to ensure her own favorable destiny since she is now as one Further probing elicited that Hindu women’s responsibilities extend with them. Thus selection of auspicious dates, times, and spaces, well beyond the modern Western notion of women ensuring their and the use of auspicious symbols are motivated by the desire to husband’s good health by managing their nutrition and medical attract the benevolent attentions of the good-sacred while repelling care. Hindu women are also expected to pray, fast and wear the malevolent attentions of the bad-sacred. auspicious symbols in order to ensure their husbands’ longevity. This “sub-contracting” of auspiciousness-generation to the DISCUSSION women of the family begins at marriage. One informant, Tara (IF Hindus believe in karma—that one’s present social status is 42) explained that once married, a bride’s new focus was expected prescribed by one’s actions in a previous incarnation (Channa to be the well-being of her husband. According to Tara, 2000). Thus Hindus are commonly thought of as fatalists and perceived to be very different from North American societies where “These glass bangles …are rings of glass with a bit of silver. the majority of individuals are believed to seek to control their own They are a symbol for matrimony… after marriage, in India, destiny. Yet, Hindus’ concern with generating auspiciousness is glass bangles are a must for every married woman. And you second only to their concern with attaining and maintaining purity. must be very careful that you don’t break them…glass bangles When we recognize that purity is viewed as the absence of pollution, give long life to your husband…The toe rings as such, that is which in turn is viewed as “essentially that which cannot be also part of married…uh…uniform that you have to wear… controlled” (Hershman 1974, p.290), we begin to understand that this Mangal Sutra [pointing to the gold necklace she was the Hindu attempt to be pure and auspicious is an attempt to control wearing) is another thing that every Indian woman wears, one’s condition by avoiding or removing that which cannot be these black beads are what we call a symbol of marriage. Every controlled. married woman wears these… This ….is also for the longevity Hindus are concerned with using auspicious symbols to attract of your husband…. everything for the husband, so we… we’re the benevolence of the gods and avoiding inauspicious symbols to happy.” avoid attracting the attention of malevolent beings. We note that this concern with auspiciousness is heightened at times of birth, The tinkling sounds made by jewelry such as bangles and marriage and death—all dangerous transitions where we are forced anklets are believed to be pleasant to the gods. According to Shukla to face the kratophanous power of the sacred. We view this concern (2000), a bride not wearing the marriage jewelry of toe rings, with auspiciousness as an attempt to indirectly control (via the anklets, and most importantly, bangles, can become a bad omen for superior powers of the gods) what one cannot control directly. Thus, the husband. Wearing the jewelry does not just generate positive Hindus are, after all, concerned with control. The need to exert outcomes for the husband, it avoids the negative outcomes gener- control over one’s destiny is universal- and it will find expression ated by not wearing the jewelry. We surmise from this that auspi- in some way, despite the constraints of society and religion. ciousness simultaneously functions to approach the good-sacred Consequently, even those in the West who seek to exert personal and repel the bad-sacred. control over their lives, may seek to indirectly control those aspects The relationship between auspiciousness and sacredness can of their lives cannot be directly controlled. For example, the popular be further unpacked by looking at the kratophanous power gold media abounds with stories of people fighting a terminal disease jewelry such as the Mangal Sutra. These strong, ambivalent reactions who seek to dominate the uncontrollable by using science, litigation, inherent in kratophany, are evident when we consider the reversal religion or even, magic. Like purity and pollution, science and of the rituals involved in marriage when a woman becomes a litigation operate on a social plane. Like the sacred gods, religion widow. During the marriage, the adorning of the bride with auspicious and magic operate on a cosmic plane. Our Western equivalents of gold jewelry prepares the bride to join the groom in a temporary auspicious symbols, such as lucky numbers, prayerbooks, magic elevated state, as they are believed to be the embodiment of a god/ crystals, and inalienable jewelry, are used to attract benevolent goddess (in the case of Brahmins) or royalty (in the case of non- cosmic powers to control what we cannot control, on our behalf. Brahmins) during the ceremony (Dumont 1972). While the couple Thus marketers are increasingly faced with consumers who seek to is in this common state of elevation, the bride is incorporated into control every aspect of the consumption experience but then may be her husband by virtue of the marriage rites which usually include dissatisfied when the product, service, or experience consumed is the groom or the officiating priest tying a golden thread bearing the not exactly what they had anticipated. For example, while awaiting Mangal Sutra pendant around her neck. Since Hindus view the wife the discovery of a magical weight-loss pill or elixir of life, some as incorporated into her husband’s body, his death means that she consumers seek to transform their bodies or regain their youth with is permanently associated with death and bad-sacredness. Thus the the aid of a surgeon’s scalpel (Schouten 1991) in an attempt to voluntary removal (Srinivas 1952) or forcible stripping (Firth control their appearance, and perhaps, their destinies. 708 / Auspiciousness: Coping with Kratophany How does the Hindu conceptualization of auspiciousness Korom, Frank J. (2000), “Holy Cow! The Apotheosis of Zebu, inform consumer research on luck, chance, and negotiation? Our or Why the Cow is Sacred in Hinduism,” Asian Studies realization that consumers employ auspiciousness in attempt to Folklore Studies, 59 (2), 181-203. indirectly control their destiny offers insight into the krataphonous Marglin, Frédérique Apffel (1985), “Types of Oppositions in power of the sacred. Krataphony may explain the revulsion and Hindu Culture,” in Purity and Auspiciousness in Indian vindictiveness consumers experience when omnipotent surgeons Society, eds. John B. Carman and Frédérique Apffel Marglin, and revered ballplayers fail to provide them with the favorable Leiden: E. J. Brill, 65-83. outcomes they seek. We call for research that looks at if, and how, Miller, Alan L. (2005), “Power”, ed. Jones, Lindsay, Encyclope- science, religion and magic inter-relate in the lives of postmodern dia of Religion, 2nd Edition, Detroit, MI: MacMillan consumers actively seeking to participate in production and Reference, 11, 7346-7354. consumption processes (Firat and Venkatesh 1995). Narayanan, Vasudha (1985), “The Two Levels of Auspicious- Thus our explication of auspiciousness sheds light on the ness in Srivaisnava Ritual and Literature,” in Purity and distinction between the good-sacred and the bad-sacred. Sacred Auspiciousness in Indian Society, ed. John B. Carman and powers, whether benevolent or malevolent, cannot be controlled by Frédérique Apffel Marglin, Leiden: E. J. Brill, 55-64. mere humans. Hence the use of auspiciousness represents a way of O’Guinn, Thomas C. and Russell W. Belk (1989), “Heaven on manifesting attempts to attract benevolent powers and the removal Earth: Consumption at Heritage Village, USA,” Journal of of inauspiciousness is an attempt to forestall malevolent powers Consumer Research, 16 (September), 227-39. while also protecting oneself from the danger inherent in sacred Pimentel, Robert W. and Kristy E. Reynolds (2004), “A Model power. What other coping mechanisms do consumers employ, for Consumer when faced with the kratophany of the sacred? Research into these Devotion: Affective Commitment with Proactive Sustaining coping mechanisms may shed further light on the nature of the Behaviors.” Academy of Marketing Science Review [Online] kratophanous power of the sacred, and the impact kratophany on 2004 (5) Available: http://www.amsreview.org/articles/ consumers. pimentel05-2004.pdf Schouten, Jonathan W. (1991), “Selves in Transition: Symbolic REFERENCES Consumption in Personal Rites of Passage,” Journal of Belk, Russell W., Melanie Wallendorf, and John F. Sherry Jr. Consumer Research 17 (March), 412-425. (1989), “The Sacred and Profane in Consumer Behavior: Shukla, Pravina (2000), “A Body of Art,” Voices: The Journal of Theodicy on the Odyssey,” Journal of Consumer Research, New York Folklore 26 (Fall-Winter), 28-33. 16 (June), 1-38. Srinivas, Mysore Narasimhachar (1952), Religion and Society Belk, Russell W. and Melanie Wallendorf (1990), “The Sacred Among the Coorgs of South India, Oxford: Clarendon Press. Meanings of Money,” Journal of Economic Psychology, 11 Yagi, Yuko (1999), “Rituals, Service Castes, and Women: Rites (1), 35-67. of Passage and the Conception of Auspiciousness and Carman, John B. (1985), “Conclusion: Axes of Sacred Value in Inauspiciousness in Northern India,” in ed. Masakazu Tanaka Hindu Society,” in Purity and Auspiciousness in Indian and Musashi Tachikawa, Senri Ethnological Studies 50, 243- Society, ed. John B. Carman and Frédérique Apffel Marglin, 281. Leiden: E. J. Brill, p.108-20. Channa, V.C. (2000), “Dharma: Purity and Pollution,” Journal of Human Ecology 11(4), 275-280. Curasi, Carolyn F., Linda L. Price and Eric J. Arnold (2004), “How Individual’s Cherished Possessions Become Families’ Inalienable Wealth,” Journal of Consumer Research (December), 31 (3), 609-623. Das, Veena (1982), Structure and Cognition: Aspects of Hindu Caste and Ritual, 2nd ed., Delhi: Oxford University Press. Dumont, Louis (1972), Homo Hierarchicus: The Caste System and its Implications, London: Paladin. Eliade, Mircea (1959), The Sacred and the Profane: the Nature of Religion, trans. Willard. R. Trask, New York: Harcourt, Brace & World. Firth, Shirley (2001), “Hindu Death and Mourning Rituals: The Impact of Geographic Mobility,” in Grief, Mourning and Death Ritual, ed. J. Hockey, J. Katz and N. Small, Buckingham UK: Open University Press. Firat, A. Fuat. and Alladi Venkatesh (1995), “Liberatory Postmodernism and the Reenchantment of Consumption”, Journal of Consumer Research, 22 (3), 239-67. Hershman, Paul (1974), “Hair, Sex and Dirt,” Man, New Series, 9 (2), 274-298. Inden, Ronald (1985), “Kings and Omens,” in Purity and Auspiciousness in Indian Society, eds. John B. Carman and Frédérique Apffel Marglin, Leiden: E. J. Brill, 30-40. Inden, Ronald and Ralph W. Nicholas (1977), Kinship in Bengali Culture, Chicago: University of Chicago Press. One or Many?: Examining Consumption Experiences from the Lens of Multiple Selves Shalini Bahl, University of Utah, USA George Milne, University of Massachusetts Amherst, USA

EXTENDED ABSTRACT ing positive and negative consumption experiences from the lens of This paper explores multiplicity in consumers’ self-concepts multiple selves, this research contributes to current literature in post and their relationship with consumption experiences. It builds upon purchase evaluation. recent streams of consumer research that recognize multiplicity in Current research suggests that products that evoke primarily self concepts and examine multiple selves in different consumption positive emotions during the consumption experience are posi- contexts. It addresses the question, what is the meaning of con- tively evaluated (Oliver 1993; Stokmans1998; Westbrook and sumption experiences from the perspective of multiple selves? Oliver 1991), which was also found in our study. However, when More specifically, does a consumption experience entail one set of understood from perspective of certain selves, even positive expe- attitudes and feelings that are consistent across selves or does it riences could involve primarily negative emotions. More interest- involve many sets of attitudes and feelings that vary across selves? ingly, some products and experiences that were considered nega- This research draws from dialogical self theory, which views tive and evoked overall negative feelings were described as very multiple selves as narrative voices that occupy multiple self- positive experiences when understood from perspective of certain positions (Hermans and Kempen 1993; Raggatt 2000, 2002). This selves. These findings suggest that researchers would benefit from theoretical perspective relies on people’s life stories centered on examining consumption attitudes and feelings from perspective of important attachments to people, historic events, beliefs, and con- the consuming selves in order to understand complex consumer sumption activities as the basis for their self-positions. This theory processes such as post purchase emotions and evaluations. assumes a natural state of multiplicity and allows multiple selves to The findings in this study, consistent with dialogical theory, enter into dialogical relationships. By grounding our study in this suggest that decisions are not based on overall feelings or attitude theory we are able to gain a broader understanding of the consumer towards a product, but based on feelings of the self that is dominant selves, which then can be applied to better understand consumption at the time. The idea of dominant self is similar to that of identity experiences from the perspective of the different selves. salience (Arnett et al 2003; Dimofte, Forehand, and Deshpande In this paper, the term multiple selves is used to mean multiple 2003; Forehand, Deshpande and Reed 2002; Kleine et al 1993; voices that speak from different positions, which is very different Laverie et al 2002). Yet, an importance difference is that the sphere from schizophrenia or multiple personality disorder. Moreover, we of influence for the dominant self extends beyond that of salient choose to use the word “multiple selves” instead of “self positions,” identities, which are primarily concerned with identity-related “multiple voices” or “dialogical self” as it is a more generic term consumption behaviors. Further, we found that the dialogical self that is consistent with current marketing literature on self-concept. perspective allowed us to study products that are not clearly related The dialogical self differs from other more commonly used to any identity and may be outside the purview of identity studies. perspectives in consumer behavior. It is different from the situ- We also found that overall metaphorical feelings were differ- ational self (Schenk and Holman 1980) in that the focus is on the ent from feelings expressed by the different selves. This reveals position from which the person is narrating an experience, which is limitations of research methods that do not take into account not necessarily tied to a particular situation. Dialogical self is more feelings and attitudes of the different selves. By looking through the of an internal perspective based on affective attachments as op- lens of multiple selves as a dialog between multiple self positions, posed to a self image that is desirable in a particular situation. It also we can gain a richer understanding of consumption patterns in varies from identity theory, and social identity theory. The dialogi- consumers. Instead of relying on compensatory reasoning as is cal self is a broader construct that subsumes roles and social done in multi-attribute models, we see that consumption is often memberships. While social identities gain salience through positive based on dominant feelings of one self which may not be congruent affirmations, dialogical self emerges from positive and negative with other self perspectives. attachments. Dialogical self also differs from other The findings in this paper reiterate the importance of studying conceptualizations as it allows for dialogical relationships. Conse- consumer behavior from the perspective of multiple selves. Other quently, the consumption choices would be different from what research programs can adapt the methodology used in this study to they would be in the absence of dialogical relationships. examine multiple selves in other consumption contexts such as This study employed a mixed-method approach in order to brand loyalty. Further, this study demonstrates the use of dialogical understand consumers within a larger context over and above self theory, which can also be used to understand other consumer current explanations of multiple selves as role identities, social topics such as ambivalence, conflict resolution, addictions, and identities, and narrators of consumption stories. Raggatt’s (2002) self-control. personality web protocol, comprising narratives and multi-dimensional scaling, was adapted and used in combination with meta- BIBLIOGRAPHY phors in order to discern multiple selves. Metaphor analysis was Aaker, Jennifer (1999), “The Malleable Self: The Role of Self- used to get a further understanding of the feelings attached to the Expression in Persuasion,” Journal of Marketing Research, selves discerned and to understand informants’ consumption expe- 36 (February), 45-57. riences at the overall and self levels. Ahuvia, Aaron C. (2005), “Beyond the Extended Self: Loved There were 12 hours of data collection for each of the study’s Objects and Consumers’ Identity Narratives,” Journal of six informants. In this paper we report data from two individuals. Consumer Research, 32:1, 171-184. The findings illustrate that the same consumption object or Arnett, Dennis B., Steve D. German, and Shelby D. Hunt (2003), experience can evoke different sets of feelings for different selves. “The Identity Salience Model of Relationship Marketing This has significant implications on how we study consumer Success: The Case of Nonprofit Marketing,” Journal of behavior and also for market researchers in their endeavors to Marketing, Vol. 67 Issue 2, 89-105 understand consumers’ attitudes and opinions. Further, by examin- 709 Advances in Consumer Research Volume 34, © 2007 710 / One or Many?: Examining Consumption Experiences from the Lens of Multiple Selves Bahl, Shalini and George R. Milne (forthcoming), “Mixed Oliver, R.L. (1992 ), “An investigation of the attribute basis of Methods in Interpretive Research: An Application to the emotion and related affects in consumption: Suggestions for Study of Self-Concept,” Handbook of Qualitative Marketing a stage-specific satisfaction framework,” in Advances in Research, Russell W. Belk, ed. London: Sage. Consumer Research, Vol. 19, ed. J.F. Sherry, Jr. and B. Belch, George E. and E. Laird Landon, Jr. (1977), “Discriminant Sternhal, Provo UT: Association for Consumer Research, Validity Of A Product-Anchored Self-Concept Measure,” 237-244 Journal of Marketing Research, 14 (May), 252-256. Raggatt, Peter T.F. (2002), “The landscape of narrative and the Belk, Russell W. (1975), “Situational variables and consumer plural self: Exploring identity using the Personality Web behavior,” Journal of Consumer Research, 2 (December) Protocol,” Narrative Inquiry, 12, 290-318. 157-164 Raggatt, P. (2000). Mapping the dialogical self: Towards a Belk, Russell W. (1988), “Possessions and the extended self,” rationale and method of assessment. European Journal of Journal of Consumer Research, 15 (September), 139-167 Personality, 14, 65-90. Dimofte, Claudiu V., Mark R. Forehand, and Rohit Deshpande Schenk, T. C. and R. H. Holman (1980), “A Sociological (2003), “Ad schema incongruity as elicitor of ethnic self- Approach To Brand Choice: The Concept Of Situational Self awareness and differential advertising response, Journal of Image,” in Advances in Consumer Research, J Olson ed., Advertising, 32 i4, 7-11 Vol. 7, 610614. Firat, Fuat A. and Clifford J. Shultz (2001), “Preliminary Metric Schouten, John W. (1991), “Selves in Transition: Symbolic Investigations into the Postmodern Consumer,” Marketing Consumption in Personal Rites of Passage and Identity Letters, 12 (2), 189-203 Reconstruction,” Journal of Consumer Research, 17 Firat, Fuat A. and Clifford J. Shultz (1997), “From Segmentation (March), 412-425 to Fragmentation: Markets and Marketing Strategy in the Sirgy, Joseph M. (1983), “Self-Concept in Consumer Behavior: Postmodern Era,” European Journal of Marketing, 31, 3/4, A Critical Review,” Journal of Consumer Research 9: 287 183-207. Stets, J. E., and Burke, P. J. (2000), “Identity Theory and Social Firat, Fuat A. and Alladi Venkatesh (1995), “Liberatory Identity Theory,” Social Psychology Quarterly, 63, 224-237 postmodernism and the reenchamntment of consumption,” Stokemans, Mia (1998) “The Relation between Post Purchase Journal of Consumer Research, 22, 239-267 Evaluations and Consumption Experiences of Hedonic Forehand, Mark R., Rohit Deshpandé, and Americus Reed Products: A Case of Reading Fiction,” European Advances (2002), “Identity Salience and the Influence of Differential in Consumer Research, Vol. 3, 139-145. Activation of the Social Self-schema on Advertising Thompson, Craig J. (1997), “Interpreting consumers: A Response,” Journal of Applied Psychology, Vol. 87 (6) hermeneutical framework for deriving marketing insights December, 1086-1099 from the texts of consumers’ consumption stories,” Journal Fournier, Susan (1998), “Consumers and Their Brands: of Marketing Research, Vol. 34 (4), 438-55 Developing Relationship Theory in Consumer Research,” Thompson, Craig, Howard R. Pollio, William B. Locander Journal of Consumer Research, 24, 343-73 (1994), “The spoken and the unspoken: A hermeneutic Hermans, Hubert J. M (1996), “Voicing the Self: From Informa- approach to consumer meaning,” Journal of Consumer tion Processing to Dialogical Interchange,” Psychological Research, Vol. 21, 432-452. Bulletin, 119 (1) (January), 31-50 Tian, Kelly and Russell W. Belk (2005), “Extended Self and Hermans, Hubert J. M. and Harry J. G. Kempen (1993), The Possessions in the Workplace,” Journal of Consumer dialogical self: Meaning as movement. New York: Academic Research, 32 (September), 297-310. Press Westbrook, Robert A. and Richard P. Oliver (1991), “The Hogg, Margaret K. and Maria H. Savolainen (1998), “Symbolic Dimensionality of Consumption Emotion Patterns and Consumption And The Situational Self,” European Advances Consumer Satisfaction,” Journal of Consumer Research, 18 in Consumer Research, Vol. 3, 11-16 (June): 84-91. James, William, (1890), The Principles of Psychology, Vol. 1, New York: Holt. Kleine, Robert E., III, Susan Schultz Kleine, and Jerome B. Kernan (1993), “Mundane Consumption and the Self: A Social-Identity Perspective,” Journal of Consumer Psychol- ogy, 2 (3), 209 235 Laverie, Debra A., Robert E. Kleine III, and Susan Schultz Kleine (2002), “Reexamination and Extension of Kleine, Kleine, and Kernan’s Social Identity Model of Mundane Consumption: The Mediating Role of the Appraisal Process,” Journal of Consumer Research, Vol. 28, 659-669 Mandel, N. (2003), “Shifting selves and decision making: The effects of self-construal priming on consumer decision- making,” Journal of Consumer Research, Vol. 30, 30-40. Morgan, Amy J. (1993), “The Evolving Self In Consumer Behavior: Exploring Possible Selves in Leigh McAlister and Michael L. Rothschild (eds.), Advances in Consumer Research, Vol. XX. Provo, UT: Association for Consumer Research. Lines in the Sand: Using Category Widths to Define and Pursue Self-Control Goals Cait Poynor, University of South Carolina, USA Kelly Haws, University of South Carolina, USA

EXTENDED ABSTRACT REFERENCES Much of the recent media attention on issues such as obesity Baumeister, Roy F. (2002), “Yielding to Temptation: Self- in children and adults (Brownell and Yach 2005), rising bankrupt- Control Failure, Impulsive Purchasing, and Consumer cies caused by overextension of credit (Fetterman 2005), and the Behavior,” Journal of Consumer Research, 28 (March), 670- lack of savings by Americans (Colvin 2005), focuses on individu- 676. als’ inability to regulate their own behavior. That is, many people Brownell, Kelly D. and Derek Yach (2005), “The Battle of the seem to repeatedly give in to temptations in an impulsive manner Bulge,” Foreign Policy, 151 (November-December), 26-27. (Baumeister 2002). While this lack of control may well account for Colvin, Geoffrey (2005), “We’re a Nation Helpless to Save many societal ills, lack of control alone does not tell the full story. Ourselves,” Fortune, April 4, 2005, 151 (8), 52. Rather, failure to properly categorize alternatives is likely a contrib- Fetterman, Mindy (2005), “For Many Americans, It’s Time to go uting factor to the problems associated with such issues as obesity on Financial Diets,” USA Today, April 5, 2005, A1. and consumer debt. In fact, the road to addressing many of the dark Gollwitzer, Peter (2003), “Why We Thought That Action Mind- issues associated with today’s society begins with individual con- Sets Affect Illusions of Control,” Psychological Inquiry, 14 sumers setting goals and creating plans to achieve those goals. One (3-4), 261-269. of the first steps in creating such plans is to define the category of Huang, Ming Shiunn (1981), “Category Width and Individual goal-consistent options and behaviors. For example, a consumer Differences in Information Processing Strategies,” Journal of embarking on a dieting plan needs to determine if buying a Psychology, 108, 73-79. McDonald’s hamburger should be categorized as a goal-consistent Pettigrew, Thomas F. (1958), “The Measurement and Correlates alternative or not. Similarly, a consumer creating a plan to reduce of Category Width as a Cognitive Variable,” Journal of personal debt will need to determine if “buying groceries” using Personality, 26, 532-544. credit should be categorized as an appropriate expenditure that Ülkümen , Gulden, Vicki Morwitz and Amitav Chakravarti, constitutes an acceptable or unacceptable behavior. Ill-defined “Flexibilitiy of Categorization: Situational Determinants of plans in this early categorization phase can contribute to self- Category Width,” Working Paper, Stern School of Business, control failure in a variety of consumption domains. New York University, New York, NY 10012. Categorization theory provides insight into the ways that individuals define and group stimuli in their environment. How- ever, despite the realization that self-control goals are contingent upon delineation of target behaviors and outcomes (Gollwitzer 2003), little research has focused on the goal-consistent and goal- inconsistent option categorization process as it relates to goal pursuit. Specifically, category width has long been viewed as an individual difference factor (Huang 1981; Pettigrew 1958), such that individuals are chronically either broad or narrow categorizers of various stimuli in their environments. In this tradition, neither goals nor beliefs about one’s own self-control ability should norma- tively impact the way in which a given set of options is categorized. More recent research has demonstrated that category width can be altered by situational factors (Ulkümmen, Morwitz and Chakravarti 2006). In the present research, we are primarily interested in exploring the relationship between self-control goals, individuals’ perceptions of their ability to meet self-control goals, and the widths of their categories of goal-consistent and goal-inconsistent alternatives. Importantly, we are interested in the ways that categories of goal-consistent and goal-inconsistent options are subjectively defined by individuals. Options included in the goal-consistent category are, conceptually, those alternatives the individual feels will facilitate the accomplishment of her/his goal(s). On the contrary, goal-inconsistent options are potential choices that hamper efforts toward goal achievement, based on the individual’s interpretation. Our research illuminates issues involving category construction and self-control by showing that variation in the construction of goal-consistent and goal-inconsistent option categories can be explained by examining individuals’ trait level self-control as well as their particular consumption goals. Furthermore, we demonstrate that the construction of these categories can differentially impact subsequent choice behavior.

711 Advances in Consumer Research Volume 34, © 2007 Stress and Eating: Disentangling the Effects of Psychological, Autonomic and Endocrine Components of Stress Response Zhenfeng Ma, University of Toronto Institute of Technology, Canada Aida Faber, McGill University, Canada Laurette Dubé, McGill University, Canada

ABSTRACT Using a psychophysiological approach to stress, this study examines the joint effects of psychological, autonomic and endocrine reactivity on consumer food preference under stress through a randomized controlled study. Stress was manipulated using guided mental imagery technique. Consumer showed increased preference for high energy-density snack foods in the stress vs. control condition. The changes in food preference were jointly predicted by psychological, autonomic and endocrine responses. A model comparison approach shows that the inclusion of autonomic and endocrine responses contributed incrementally to the prediction of food preference. Theoretical and practical implications are discussed.

712 Advances in Consumer Research Volume 34, © 2007 Beating Back that Triple-Chocolate Cake: Mental Accounts as Instruments of Self-Regulation Parthasarathy Krishnamurthy, University of Houston, USA Sonja Prokopec, University of Houston, USA

EXTENDED ABSTRACT Two studies were conducted as part of a program of research One of the vexing conundrums in everyday life is our inability focused on the self-regulatory effectiveness of mental accounts. to say “no” to actions that we would ideally like to say “no” to. It is The first study manipulated presence of mental accounts and task common for people to profess ‘healthy eating’ goals and yet frame, and found that decision time was considerably faster when repeatedly succumb to temptations in the form of foods high in the decision-maker had a mental account. However, this study calories, fat, sugar etc. Such instances of goal-behavior inconsis- found only directionally consistent effects on self-regulation on tency represent failures of self-regulation (Heatherton and account of limited variability in the dependent variable. The second Baumeister, 1996). study involved the same manipulations and found that mental Research in consumer behavior has long been interested in accounts are successful in facilitating self-regulation only when the strategies aimed at reducing self-regulation failures (Hoch and task was compatible with the goal of the mental account. Thus, Loewenstein, 1991). In this research, we examine whether having having a mental account significantly reduced the likelihood of mental accounts (Thaler, 1980, Thaler and Johnson, 1985; Tversky engaging in indulgent behaviors, but only when the participant was and Kahneman, 1981), defined as self-specified allowances for in a reject task, not in a choose task. action, will lead to improved self-regulation. Why should mental The contribution of this research is threefold. First, we inves- account influence self-regulation? According to Baumeister (2002), tigate this phenomenon in a non-monetary context, in contrast with effective self-regulation depends on three principal factors which the vast majority of the research on mental accounting which are (a) clear standards for the behaviors to be regulated, (b) ability focuses on monetary decision contexts. Second, we let the partici- to monitor one’s status with regard to the standards, and (c) pants estimate the size of their mental account. Majority of the capacity, i.e., resources to alter one’s own behavior should it be research in mental accounting examines this phenomenon by pre- warranted. senting participants with scenarios that involve pre-set mental Mental accounts appear to satisfy all the three criteria outlined accounts (both in terms of size and context). Third, and most by Baumeister (2002) that are necessary for self-regulation. First, important, the connection with self-regulation situates mental ac- mental accounts are set around active goals (Brendl, Markman, and counting as a factor that shapes the cognitive process underlying Higgins, 1998), so they inherently provide the necessary standards/ decision-making rather than as an evaluation frame that affects goals for behavior. Second, mental accounts enable a person to economic outcomes. evaluate current choice in terms of previous choices (Kahneman In summary, mental accounts offer an ecologically relevant and Tversky, 1984); therefore, mental accounts enable one to way of exerting self-control when faced with tempting choices. monitor one’s own behavior. Finally, mental accounts have been However, they do so only when the task adopted by the decision- thought of as instances of categorization (Henderson and Peterson, maker is compatible with the goal of the account. We believe that 1992), which results in resource-efficient decision-making. Thus, one of the salient features of this research is that it bridges the having a mental account leaves the decision-maker with more knowledge from three well researched genres, mental accounting, resources to exert self-control. Although the connection between self-regulation and task framing, to indicate conditions under which mental accounts and categorization has been made, its connection people can bring their behavior in line with their own goals. to the amount of resources needed for decision-making has neither been demonstrated nor been made before in the literature. Since the REFERENCES capacity to change behavior, i.e., having enough resources to Baumeister, Roy F. (2002), “Yielding to temptation: Self-control change, is a critical element without which having standards or failure, impulsive purchasing, and consumer behavior,” monitoring is less likely to be useful (Baumeister, 2002), we believe Journal of Consumer Research, 28 (March), 670-676. it is important to assess whether having a mental account renders Brendl, C. Miguel, Arthur B. Markman, and E. Tory Higgins decision-making more efficient. Based on the notion that speed of (1998), “Mental Accounting as Self-Regulation: Representa- decision is an index of the efficiency, we hypothesize that people tiveness to goal-derived categories,” Zeitschrift fur with a mental account will make quicker decisions regarding Sozialpsychologie, 29, 89-104. actions relating to the mental account than people without a mental Cheema, Amar and Dilip Soman (2006), “Malleable Mental account. Accounting: The effects of flexibility on the justification of Does this mean that whenever people have mental accounts, attractive spending and consumption decisions,” Journal of they will be able to achieve improved self-regulation? We argue Consumer Psychology, 16 (1), 33-44. that mental accounts are likely to facilitate self-control only when Heath, Chip and Jack B. Soll (1996), “Mental Budgeting and the task frame adopted by the decision-maker, choose versus reject Consumer Decisions,” Journal of Consumer Research, 23 (Shafir, 1993), is compatible with the goal. Specifically, if the goal (June), 40-52. is to limit behavior, we argue that a reject task is more compatible Heatherton, Todd F. and Roy F. Baumeister (1996), “Self- with the goal, allowing the mental account to exert its influence. A regulation failure: Past, present, and future,” Psychological choose task, on the other hand, orients the person to look for reasons Inquiry, 7 (1), 90-98. to adopt the behavior, and therefore is less compatible with the goal Henderson, Pamela W. and Robert A. Peterson (1992), “Mental of limiting the behavior. Based on this reasoning, we suggest that Accounting and Categorization,” Organizational Behavior there is reason to expect that the effect of mental accounts in and Human Decision Processes, 51, 92-117. limiting indulgent behaviors will be more pronounced in a reject Hoch, Stephen J. and George F. Loewenstein (1991), “Time- task rather than in a choose task. inconsistent preferences and consumer self-control, “Journal of Consumer Research, 17 (March), 492-507.

713 Advances in Consumer Research Volume 34, © 2007 714 / Beating Back that Triple-Chocolate Cake: Mental Accounts as Instruments of Self-Regulation Kahneman, Daniel and Amos Tversky (1984), “Choices, values, and frames,” The American Psychologist, 39, 341-350. Maphet, Harold W. and Arthur L. Miller (1982), “Compliance, temptation, and conflicting instruction,” Journal of Personal- ity and Social Psychology, 42 (January), 137-144. Muraven, Mark, Dianne M. Tice, and Roy F. Baumeister (1998), “Self-control as limited resource: Regulatory depletion patterns,” Journal of Personality and Social Psychology, 74 (March), 774-789. Shafir, Eldar (1993), “Choosing versus rejecting: Why some options are both better and worse than others,” Memory and Cognition, 21 (4), 546-556. Thaler, Richard H. (1980), “Toward a positive theory of consumer choice,” Journal of Economic Behavior and Organization, 1 (1), 39-60. (1985), “Mental accounting and consumer choice,” Marketing Science, 4(3), 199-214. (1999), “Mental accounting matters,” Journal Behav- ioral Decision Making, 12 (3), 183-206. and E.J. Johnson (1990), “ Gambling with the house money and trying to break even: the effects of prior outcomes on risky choice,” Management Science, 36 (6), 643- 660. Tversky, Amos and Daniel Kahneman (1981), “The framing of decisions and the rationality of choice,” Science, 211, 453- 458. Vohs, Kathleen D. and Todd F. Heatherton (2000), “Self- regulation failure: A resource-depletion approach,” Psycho- logical Science, 11 (3), 249-254. Making Prudent vs. Impulsive Choices: The Role of Anticipated Shame and Guilt on Consumer Self-Control HaeEun Chun, University of Southern California, USA Vanessa M. Patrick, University of Georgia, USA Deborah J. MacInnis, University of Southern California, USA

ABSTRACT happy, proud) versus negative emotions from not achieving a goal We examine the differential effects of anticipating shame vs. (e.g., guilty, regretful, sad) on impulse control (Bagozzi et al. 2003; guilt on choice likelihood of a hedonic product. The results demon- Perugini and Bagozzi 2001; Giner-Sorolla 2001). strate that when offered a hedonic snack (chocolate cake) consum- An interesting and relevant extension to this literature con- ers who anticipate shame are significantly less likely to choose to cerns the impact of anticipated emotions of the same valence on consume it compared to those who anticipate guilt. Anticipating self-control. As Lerner and Keltner (2001) point out, little work has guilt also has a more circumscribed effect, impacting choice like- been done to differentiate emotions of the same valence–and this is lihood only for those consumers who are not attitudinally inclined particularly true in the context of affective forecasting. Relevant toward the hedonic product. The results also show that anticipating here are negative anticipated emotions presumed to arise from the guilt versus shame has different effects on anticipated happiness failure to control impulses and consume a hedonic product. Al- after lapses in self-control. though Bagozzi et al. (1998, 2001) found that anticipating negative emotions like shame and guilt can facilitate impulse control, they INTRODUCTION did not assess whether each emotion can itself impact self-control. Maria was dismayed at how much weight she had gained. It Because both are negative emotions induced from moral transgres- seemed that no matter how hard she tried, she just couldn’t resist sions, one might expect that either shame or guilt can impact self- indulging in high calorie desserts. Vowing to remember how bad control compared to not anticipating any emotions. her overeating made her feel, she put a note on the box of left-over cake from her daughter’s birthday party that reads “if you eat this, H1: Compared to consumers who do not anticipate the emo- you will feel bad.” tions presumed to arise from a hedonic choice, those who Two powerful negative emotions of self-condemnation are anticipate the (a) shame or (b) guilt from hedonic con- shame and guilt. While commonsense knowledge reminds us that sumption will be more likely to exert self-control (and these emotions are reactions to self-control failures, little is known less likely to choose (consume) the hedonic product). about whether anticipating these emotions as a consequence of consumption will impact self-control. Hence, one objective of this The Differential Impact of Shame vs. Guilt on Self-Control paper is to examine whether anticipating shame and guilt from Lack of understanding of the potentially differential impact of lapses in self-control influences actual self-control. Although a shame vs. guilt on self-control is perhaps understandable because limited body of literature has examined the impact of guilt on self- the two emotions are sometimes used interchangeably (Tangney control, knowledge of the impact of anticipated shame is lacking. and Dearing 2002). This lack of distinction is likely due to the fact Thus, a second objective of this research is to assess whether shame that both are negative self-conscious emotions of self-condemna- and guilt exert differential effects on consumer self-control. Fi- tion that result in response to a moral transgression. nally, while many consumers are beset by problems of self-regula- Notably though, cumulative empirical studies reveal that these tion as evidenced by overeating, overspending, compulsive shop- two emotions are distinct. Shame is evoked from a perceived ping, gambling, smoking and drug use, not all consumers experi- transgression of the ‘self ’ (e.g., I am a horrible person), whereas ence problems with consumer self-control equally. Thus, the third guilt is evoked from a perceived transgression of one’s ‘behavior’ and final objective of this research is to examine whether shame and (e.g., I did that horrible thing) (Lewis 1971). Shame evokes guilt have a differential impact on consumers whose personality or counterfactuals of the self (“If only I weren’t”), whereas guilt preferences make them differentially vulnerable to self-control evokes counterfactuals for actions (“If only I hadn’t” for guilt) lapses. (Niedenthal et al. 1994). Finally, shame induces a behavioral motivation to hide, escape, sink or disappear from others and the THEORY AND HYPOTHESES situation (Lewis 1971). Guilt, on the other hand, induces a behavioral motivation to restore or make amends or atone for transgres- Anticipated Affect and Self-Control sions (Wicker et al. 1983; Tangney et al. 1996). Recent work in psychology and marketing has begun to Notably these differences between shame and guilt are with examine anticipated or forecasted emotions and their impact on regard to an experienced emotion. We theorize that these differ- choice. Anticipated emotions (also called affective forecasts) refer ences between shame and guilt manifest themselves even in antici- to the prediction of the emotional consequences of decision out- pation, differentially influencing self-control. Past research finds comes (Gilbert et al. 1998; MacInnis et al. 2005), in this case, to the that shame is a more powerful and intense emotional experience affective consequences of giving in or not giving in to an impulse. than guilt (e.g., Tangney et al. 1996; Roseman et al. 1994). Shame Though research on anticipated affect and self-control is is often associated with a stronger physiological response and is limited, the little that exists is intriguing. Bagozzi et al. (1998) more powerful by its implication of the self as the reason for the found that anticipating positive emotions (including pride, joy and transgression. Guilt on the other hand involves a less potent satisfaction) from self-regulatory success and negative emotions physiological response and is a result of an appraisal that implicates (including guilt, regret and sadness) from self-regulatory failure one’s behavior in the transgression. Furthermore, while guilt evokes predicted dieting intentions, plans, and dieting-related actions. the possibility of atonement for one’s wrongdoing, shame evokes Other studies have addressed the relative impact of anticipated no such potential, and indeed further implicates the self for one’s positive emotions associated with achieving a goal (e.g., delighted, weakness. Hence, we predict that anticipating the accusation of the 715 Advances in Consumer Research Volume 34, © 2007 716 / Making Prudent vs. Impulsive Choices: The Role of Anticipated Shame and Guilt on Consumer Self-Control self is more likely to keep people from transgressing than anticipat- in hedonic consumption. Once consumers decide to commit them- ing the blame of a specific behavior. Shame may also induce more selves to indulge even after anticipating guilt associated with self-control by virtue of the fact that it is viewed as a more public consumption, it is more likely that they justify their decision to emotion, whereas guilt has been viewed as a more private affair consume with inflated anticipatory happiness. The fact that guilt (i.e., reaction of one’s internalized conscience to a breach of one’s implicates the transgression of the specific behavior also implies personal standards) (e.g., Ausubel 1955; Benedict 1946; Gehm and the motivational readiness to discount their transgression as a one- Scherer 1988). By virtue of its power, lack of atonement potential, time, transient incident. As the term, guilty pleasure (Giner-Sorolla and public element, we hypothesize that: 2001), signifies, anticipating guilt, once committed to indulge, may inflate anticipated happiness about consuming the impulse-laden H2: Consumers who anticipate shame from consuming a product as a (consequence of) post-decision justification. hedonic product will be more likely to engage in self- On the other hand, since shame is a more intense emotion that control (e.g., choose to forego the consumption of a arises from the appraisal of the transgression of the self (Lewis hedonic product) than will consumers who anticipate 1971), it is likely that one’s decision to indulge cannot be easily guilt from consuming the hedonic product. justified. Moreover, the transgression is less easily discounted as a one-time, fleeting misbehavior since it implicates the self, tarnish- Attitude toward Consuming the Hedonic Stimulus ing one’s self-image. Therefore, we predict that consumers who Notably, exerting self-control may be more or less important anticipated shame, but still consume the hedonic product, are likely depending on consumers’ apriori predisposition to engage in he- to anticipate less happiness from consumption than are consumers donic choice. The same hedonic choice (e.g., eating chocolate cake) who anticipate guilt. may be extremely appealing to some consumers and less appealing to others given apriori attitudes toward the hedonic product. Hence, H5: Consumers who anticipate guilt from consuming a he- an interesting extension to H1 and H2 concerns whether shame and donic product will be more likely to anticipate happiness guilt exert an equivalent impact on self-control for consumers with following hedonic consumption compared to consumers positive vs. neutral attitudes toward the hedonic product. who anticipate shame. We predict that for consumers for who have neutral attitudes toward the hedonic product anticipating either shame or guilt will Individual Differences in Impulsivity impact self-control. Because the consumption item induces limited Self-control is more difficult for consumers with chronic attraction, anticipating either negative emotion should induce greater tendencies toward impulsive behavior. Consumer impulsivity has self-control than not anticipating these negative emotions. been widely regarded as an important individual difference variable in self-control (Puri 1996). The impact of shame and guilt on the H3: For consumers with neutral attitudes toward consuming effects noted above would be noteworthy if they remained robust the hedonic product, anticipating either shame or guilt even after controlling for individual differences in impulsivity. from consumption will have a greater impact on self- control than will not anticipating either emotion. EMPIRICAL INVESTIGATION

However, for consumers who are most vulnerable to self- Method control lapses—those with a positive attitude toward consuming H1-H5 were examined in a 3 (anticipated emotion: shame, the hedonic product— shame may exert a greater impact on self- guilt, no emotion control) X 2 (attitude toward the hedonic product: control than will guilt. Because attitudes are more positive, the positive vs. neutral) experiment. Ninety-one participants were allure of the product is greater, requiring a strong emotion like randomly assigned to one of the three anticipated emotion condi- shame to induce self control. We thus hypothesize that: tions. Attitude toward the hedonic product served as a measured variable. H4: For consumers who have a positive attitude toward Respondents were shown a delicious-looking chocolate cake consuming a hedonic product, anticipating shame from at the start of a computer-based questionnaire and were asked to consumption will have a greater impact on self-control imagine that it was placed in front of them. To manipulate antici- than will anticipating guilt. pated emotion, respondents were then told that although they wanted to reach out and take a bite of the cake, they stopped for a Anticipated Happiness from Self-Control or Lack Thereof- moment to anticipate how guilty (vs. ashamed) eating the cake Self-control in the context of hedonic choice involves a trade- would make them feel. The control condition was not given any off between short-term vs. long-term happiness. This notion is instructions to anticipate emotions. consistent with the idea that preferences can be time inconsistent Respondents then completed a set of questions designed to (Hoch and Lowenstein 1991; Metcalf and Mischel 1999; Giner- measure how likely they were to eat the cake (1=not at all likely; Sorolla 2001). Lapses in self-control brings about short-term hap- 9=very likely), their anticipated happiness from eating the cake piness as impulses are indulged—though longer-term consequences (1=not at all; 9=very much) and manipulation checks for antici- like shame and guilt may be subsequently experienced. On the other pated shame and guilt (1=not at all; 9=very much). Individual hand, engaging in self-control often sacrifices short-term pleasures differences in attitudes toward eating chocolate cake were also for longer-term happiness. Delaying gratification in the present can measured using a 3-item 9-point scale (favorable, positive, good; induce a larger and bigger reward—which presumably will induce Cronbach’s α=.92). The three items were averaged to yield a scale greater happiness than would succumbing to short-term gratifica- of attitude toward eating cake. A median split divided participants tions. into positive vs. neutral attitude conditions (Median=6.33; M=7.99 An interesting question regarding anticipated shame and guilt for positive and M=4.38 for neutral). To control for individual is whether they evoke similar effects on anticipated happiness in the differences in consumer impulsivity, a scale adapted from Puri short term as a consequence of giving in to impulses and engaging (1996) was used as a covariate in the results presented below. Advances in Consumer Research (Volume 34) / 717 FIGURE 1 The effect of anticipated emotion and attitude toward consuming the hedonic product on choice likelihood

a,b,c Means with different superscripts are different at p <.05.

Results Predicted Happiness. Respondents who experienced a lapse Manipulation Check. The results confirmed that participants in self-control and decided to choose the cake were asked to predict in the experimental conditions anticipated the emotions they were how happy they would be after eating the cake. A 3 (anticipated instructed to anticipate (M’s=4.58 and 4.61 for anticipated guilt and emotion) X 2 (attitude toward eating chocolate cake) ANCOVA shame respectively) in the context of eating the cake. with consumer impulsivity as a covariate revealed a significant Choice Likelihood. A 3 (anticipated emotion) X 2 (attitude) main effect for anticipated emotions (F (2, 48)=4.29, p<.05) and a ANCOVA on the choice likelihood as the dependent variable and significant main effect for prior attitudes toward cake (F (1, 84)=7.55, consumer impulsivity as a covariate revealed a significant main p<.01). The results were replicated in a regression analysis which effect for anticipated emotion (F (2, 84)=4.77, p<.05) and a signifi- used a continuous measure of attitudes (vs. a median split). cant main effect for prior attitudes toward cake (F (1, 84)=18.50, The main effect of anticipated emotions showed that consum- p<.001). As expected, the main effect of attitudes showed that ers who anticipated guilt anticipated feelings significantly happier consumers who had positive attitudes toward eating cake were (M=7.34) after consuming the cake than consumers who antici- more likely to choose it (and forego impulse control; M=6.93) pated shame (M=5.86) or control consumers (M=6.36). These compared to consumers who had a neutral attitude toward consum- effects support H5. The main effect of prior attitudes toward cake ing the hedonic product (M=4.65). The main effect for anticipated showed that consumers who held a positive attitude toward cake emotion showed that anticipating shame (M= 4.66) resulted in (M=7.13) anticipated feeling happier following hedonic consump- greater self-control than anticipating either guilt (M= 6.04) or the tion than consumers who held a neutral attitude toward cake control condition (M= 6.68). These results support H1a and H2. (M=5.91). However, we observed no support for H1b. Consumers who Post-hoc contrasts shown in Figure 2 revealed several interest- anticipated guilt did not show greater self-control (M= 6.04) than ing effects. First, consumers who anticipated guilt from consump- consumers in the no emotion control condition (M= 6.68). tion also anticipated greater happiness from consumption than did Post-hoc contrasts showed that for consumers with neutral consumers who anticipated shame—regardless of their prior atti- attitudes toward the hedonic product showed reduced choice like- tudes toward cake. Second, for consumers with neutral attitudes lihood when they anticipated guilt (M=4.55) or shame (M=3.62) toward cake anticipating guilt led to higher anticipated happiness compared to consumers in the no emotion control condition (M= from consumption than did not anticipating any emotion. Third, and 5.78). These results support H3. most interestingly, even consumers who were not favorably dis- The results also support H4. For consumers with positive posed toward the hedonic product but who anticipated guilt showed attitudes toward consuming the hedonic product, anticipating shame just as much anticipated happiness from consumption as those who had a greater impact on self-control (M= 5.7) than did anticipating were favorably disposed toward the product. The fact that both guilt (M= 7.53) or not anticipating emotion (M= 7.57). consumers with favorable and neutral attitudes toward cake antici- Collectively, these results suggest that while anticipating pated equivalent levels of post-consumption happiness from he- shame has a consistently positive effect on self-control, the antici- donic consumption suggests a potential association between guilt pation of guilt impacts self-control only for consumers who have a and pleasure (Giner-Sorolla 2001). The results are also consistent neutral attitude toward the hedonic product and are thus less with the recent finding that showed the association between the predisposed to making an impulsive choice. concepts of “unhealthy” and “tasty” operates on an implicit level (Raghunathan et al. 2006). 718 / Making Prudent vs. Impulsive Choices: The Role of Anticipated Shame and Guilt on Consumer Self-Control FIGURE 2 Anticipated Happiness for Consumers Who Decided to Consume the Hedonic Product

a,b,c,d Means with different superscripts are different at p <.05.

DISCUSSION to feel in the future. A rather constructive consequence of experi- Although prior research suggests that anticipating the emo- enced guilt (e.g., motivation to undo the error) may exert less tions that might result from consumption impacts choice of hedonic influence when anticipated. In the same way, shame can exert a products, our results extend this research by suggesting that two more powerful influence when anticipated than when experienced negative emotions of the same valence (shame vs. guilt) have in the context of self-control. Anticipating negative consequences different effects on both self-control and anticipated happiness of shame such as passivity and desire to escape and hide can serve from self-control failures. as a more powerful means to regulate impulses. We weigh the We find that compared to conditions where consumers do not emotional consequences of the events and direct our motivation and anticipate emotions from choice, those who anticipate shame are behavior in line with our affective prediction. significantly less likely to make a hedonic choice (and hence are Interestingly, while shame is a far more intense negative significantly more likely to engage in self-control). The impact of emotion than guilt, anticipating future feelings of shame could anticipated guilt on self-control, however, appears to depend on potentially decrease the likelihood of experiencing future negative whether consumers have positive or neutral attitudes toward the feelings by increasing self-control. Notably though, one wonders hedonic product. Anticipating guilt seems to facilitate self-control, about the effects of anticipating shame on outcomes like depression but only for those consumers who need it least—those who have or self-hatred when hedonic choice is rendered despite anticipated neutral attitudes toward the hedonic product. For those who are pro- shame. In this sense, anticipating shame from consumption may attitudinally inclined toward the hedonic product, anticipating guilt make consumers more vulnerable to negative feelings and self- has no effect on self-control. hatred. From the standpoint of happiness from consumption, we find Another limitation concerns the fact that we did not assess how that for consumers who chose to engage in hedonic consumption much guilt or shame consumers in the control condition anticipated. anticipated happiness with this decision varied as a function of both It is possible that consumers in this condition naturally anticipated anticipated shame/guilt and attitudes. Consumers felt least happy guilt (or shame). This might explain lack of difference in results with their decision to engage in hedonic consumption when they between the control and the anticipated guilt conditions. had a neutral attitude toward the hedonic product and anticipated The results of our study provide several opportunities for feeling shame. Consumers who anticipated feeling guilty antici- extension. Anthropologist Ruth Benedict (1946) classifies cultures pated they would feel just as happy as consumers who did not according to whether they use shame or guilt to regulate the social anticipate any negative emotions when their attitudes toward the activities of their members. She posits that some Asian cultures are hedonic product were positive, and anticipated feeling even happier considered shame cultures whereas European and modern Ameri- than those who did not anticipate any negative emotions when their can cultures are considered to be guilt cultures. Future research attitudes were neutral. Therefore, anticipating guilt does not appear might investigate cross-cultural differences anticipating shame vs. to deter anticipated happiness from choice for those who chose to guilt has on self-control. engage in consumption. The results of our study, while intriguing, are limited by virtue The results provide an opportunity to revisit the conceptual of the fact that self-control was assessed in terms of self-reports of distinction between experienced and anticipated emotions. We choice likelihood. A stronger test would examine whether the should note that affective forecasting is distinct from experienced effects we observe here are observed in the context of actual choice. affect. Anticipated emotions are cognitions about how one is likely On the other hand, these results were obtained even though a Advances in Consumer Research (Volume 34) / 719 relatively non-powerful inducement— a photo of a chocolate cake Mauraven, Mark and Roy F. Baumeister (2000). “Self-regulation (as opposed to an actual real-life temptation) was used as the and Depletion of Limited Resources: Does Self-Control stimulus (see Shiv and Fedorikhin 1999). Future research might Resemble a Muscle?” Psychological Bulletin, 74, 774-789. examine the impact of anticipated shame and guilt in the context of Niedenthal, Paula M., June Price Tangney, and Igor Gavanski real stimuli and actual choices. By comparing respondents’ (1994). “If Only I Weren’t Versus If Only I Hadn’t: behavioral responses of the current study with those using actual Distinguishing Shame and Guilt in Counterfactual Thinking,” stimuli, we can tell how far off people can predict the behavior as Journal of Personality and Social Psychology, 67 (4), 585- a result of anticipated shame vs. guilt. Another interesting extension 95. concerns the long-term impact of anticipated shame vs. guilt on Perugini, Marco and Richard P. Bagozzi (2001). “The Role of self-control. Prior studies showed that self-regulatory strength can Desires and Anticipated Emotions in Goal-directed Behav- in fact be “worn out” like a muscle when mental efforts are iors: Broadening and Deepening the Theory of Planned prolonged (e.g., Muraven and Baumeister 2000; Schmeichel and Behavior,” British Journal of Social Psychology, 40, 79-98. Baumeister 2004). Considering that shame is a more powerful and Puri, Radhika (1996). “Measuring and Modifying Consumer intense emotion than guilt, anticipating shame over time may wear Impulsiveness: A Cost-Benefit Accessibility Framework,” consumer’s self-regulatory resources out more easily than Journal of Consumer Psychology, 5 (2), 87-113. anticipating guilt. One wonders whether a greater self-control Raghunathan, Rajagopal, Naylor, Rebecda W., and Wayne D. ability induced by anticipating shame in the short-term can be Hoyer (2006). “The Unhealthy=Tasty Intuition and Its extended to the long-term. Anticipating shame may make one more Effects on Taste Inferences, Enjoyment, and Choice of Food vulnerable to relapse over the long term due to its intense mental Products,” Forthcoming, Journal of Marketing. efforts. The future research should examine the long-term impact of Roseman, Ira J., Cynthia Wiest, and Tamara S. Schwartz (1994). anticipated shame and guilt. “Phenomenology, Behaviors, and Goals Differentiate Discrete Emotions,” Journal of Personality and Social REFERENCES Psychology, 67 (2), 206-21. Ausubel, David P. (1995). “Relationships between Shame and Schemeichel, Brandon J. and Roy F. Baumeister (2004). Self- Guilt in the Socialization Process,” Psychological Review, regulatory Strength, In R. F. Baumeister and K. D. Vohs 67, 378-390. (eds), Handbook of self-regulation: Research, theory, and Bagozzi, Richard P., Hans Baumgartner, and Rik Pieters (1998). applications (pp. 84-98). New York: Guildford. “Goal-directed Emotions,” Cognition and Emotion, 12 (1), 1- Shiv, Baba and Alexander Fedorikhin (1999). “Heart and Mind 26. in Conflict: The Interplay of Affect and Cognition in Bagozzi, Richard P, Utpal M. Dholakia, and Suman Basuroy Consumer Decision Making,” Journal of Consumer (2003). How Effortful Decisions Get Enacted: The Motivat- Research, 26 (December), 278-92. ing Role of Decision Processes, Desires and Anticipated Tangney, June Price and Ronda. L. Dearing (2002). Shame and Emotions,” Journal of Behavioral Decision Making, 16, 273- Guilt, New York: Guilford. 95. Tangney, June Price, Rowland S. Miller, Laura Flicker, and Benedict, Ruth. (1946). The Chrysanthemum and the Sword: Deborah H. Barlow (1996). “Are Shame, Guilt, and Patterns of Japanese Culture. Boston: Houghton Mifflin. Embarrassment Distinct Emotions?” Journal of Personality Gehm, T.L. and Klaus P. Scherer (1988). Relating Situation and Social Psychology, 70 (6), 1256-69. Evaluation to Emotion Differentiation: Nonmetric Analysis Wicker, Frank W., Glen C. Payne, and Randall. D. Morgan of Cross-Cultural Questionnaire Data, In K. R. Scherer (Ed.), (1983). “Participant Description of Guilt and Shame,” Facets of Emotion: Recent Research (pp. 61-77). Hillsdale, Motivation and Emotion, 7 (1), 25-3. NJ: Erlbaum. Zemack-Rugar, Yael, James R. Bettman, and Gavan J. Gilbert, Daniel T., Elizabeth C. Pinel, Timothy D. Wilson, Fitzsimons (2005). “Effects of Specific, Nonconscious Stephen J. Blumberg, and Thalia P. Wheatley (1998). Emotions on Self-Control Behavior,” working paper, Duke “Immune Neglect: A Source of Durability Bias in Affective University. Forecasting,” Journal of Personality and Social Psychology, 75 (3), 617-38. Giner-Sorolla, Roger (2001). “Guilty Pleasures and Grim Necessities: Affective Attitudes in Dilemmas of Self- Control,” Journal of Personality and Social Psychology, 20 (2), 206-221. Hoch, Stephen J. and George F. Loewenstein (1991). “Time- Inconsistent Preferences and Consumer Self-Control,” Journal of Consumer Research, 17(4), 492–507. Lerner, Jennifer S. and Dacher Keltner (2001). “Fear, Anger, and Risk,” Journal of Personality and Social Psychology, 81 (1), 146-59. Lewis, Helen Block (1971). Shame and Guilt in Neurosis, New York: International University. MacInnis, Deborah J., Vanessa M. Patrick, and C. Whan Park (2005). “Looking through the Crystal Ball: The Role of Affective Forecasting and Misforecasting in Consumer Behavior,” Review of Marketing Research, Vol 2, forthcoming. Building Customer Relationships: A Comparison Across Multiple Service Encounters Hung-Chang Chiu, National Chung Hsing University, Taiwan Monle Lee, Indiana University, USA Yi-Ching Hsieh, National Central University, Taiwan Li-Ya Chen, Ming Chuan University, Taiwan

ABSTRACT may also give customers greater satisfaction. Accordingly, to The purpose of this study is to investigate the impacts of produce results that can be interpreted at a theoretical level, inves- relational bonds on customer loyalty in various service encounters. tigations must encompass a range of interfaces and consider the Data were obtained from a sample of 621 Taiwanese customers in effectiveness of customer relationships in different encounters. In the securities industry. The findings are threefold. First, financial this study, we attempt to determine whether the impacts of rela- bonds have little impact on customer loyalty, social bonds create an tional bonds on customer loyalty differ for customers in person-to- intermediate impact on customer loyalty, and structural bonds have person, interactive voice response system, and Internet encounters. the greatest impact on customer loyalty. Second, social bonds influence customer loyalty more in person-to-person encounters THEORETICAL ISSUES than in interactive voice response systems or Internet encounters. Third, structural bonds affect customer loyalty more on the Internet Customer Loyalty than in person-to-person or interactive voice response system In Oliver’s (1999, p. 34) study, loyalty is defined as “a deeply situations. held commitment to rebuy or repatronize a preferred product/ service consistently in the future.” Customer loyalty is considered INTRODUCTION one of the several means by which a firm can build a sustainable Relationship marketing has emerged a critical tool in the competitive advantage. Loyalty plays an important role in influencing current marketing climate, where managers must seek new ways to consumers’ repeat purchases (Jacoby and Kyner 1973) and creating create sustainable competitive advantage (Dibb and Meadows substantial income for companies (Jones and Farquhar 2003). From 2001). By emphasizing a focus on the ways that companies can a business perspective, structuring consumer retention mechanisms build, develop, and maintain successful relational exchanges, rela- to enhance profits has become more important than ever. tionship marketing represents an important means to build customer loyalty to an organization (Morris et al. 1999). Recently, a Relational Bonds body of research surfaced that attempts to discern the types of bonds Previous literature has indicated that a customer retention that enhance relationship marketing (e.g., Berry 1995; Peltier and strategy can focus on three types of bonds: financial, social, and Westfall 2000; Williams et al. 1998). These bonds have been structural (Berry 1995). Subsequent studies, including Berry (1995), categorized as financial (e.g., price incentives), social (e.g., inter- Peltier and Westfall (2000), and Williams et al. (1998), have personal relationships), and structural (e.g., services that are de- adopted this typology to discuss the effects of such bonds. signed into the system). However, despite the increasing impor- Researchers have argued that one motivation for customers to tance of relationship marketing, the effects of these three relational engage in relational exchanges is money savings (e.g., Berry 1995; bonds on customer loyalty have not been sufficiently studied Gwinner et al. 1998). However, although financial bonds seem empirically. useful in securing customer loyalty, some researchers have suggested Service encounters, defined as the moment of interaction that these bonds belong at the lowest level of the relationship between a customer and a firm, are increasingly critical in all hierarchy because price is the most easily imitated element (e.g., industries and may take place in an actual service setting, over the Peltier and Westfall 2000; Zeithaml and Bitner 2003). telephone, through the mail, or over the Internet (Bitner et al. 2000). Social bonds are defined as personal ties pertaining to service Prior studies have recognized the importance of service encounters dimensions that include interpersonal interaction, relationship build- in creating and maintaining good relationships between service ing (Wilson 1995), and identifications (Turner 1970). From the firms and their customers (e.g., Haring and Mattsson 1999). As customer perspective, a social bonding strategy benefits them customers and businesses interact over time, the opportunity rises significantly through strong service relationships (Gwinner et al. for this aggregation of encounters to transform into an exchange 1998). These social bonds positively influence customers’ emo- relationship (Morgan and Chadha 1993). tions or feelings associated with the service experience and contrib- Although it is increasingly evident that technological ad- ute to the formation of the affective component of attitude, all of vances will continue to be critical components of customer–firm which are salient aspects of intimate, high-quality relationships interactions, little is known about how interactions with such (Thorbjørnsen et al. 2002). Therefore, though social bonds alone technological options affect customer evaluations and behavior may not bind a customer permanently to a company, they will (Meuter et al. 2000). In different service encounters, the ways that ultimately prove more difficult for competitors to imitate (Zeithaml customers interact with a business may also differ. For example, and Bitner 2003). Some researchers (e.g., Berry 1995; Peltier and customers must personally visit some service facilities and engage Westfal, 2000) in turn suggest that social bonds sit at the interme- in interactions with employees for person-to-person encounters. diate level in the relationship hierarchy and help companies protect Therefore, developing buyer–seller relationships through interper- against competitive pricing pressures. sonal interactions or friendships between customers and employees Finally, Berry (1995, p. 241) suggests that marketers practice may be more useful for these customers. However, for Internet relationship marketing, which depends on structural bonds when encounters, customers can easily access high-quality information “the solution to the customer’s problem is designed into the service- from the Web site (Detlor 2003). Using a service-delivery system, delivery system.” The structural bonds “offer target customers rather than relying on the relationship-building skills of individual value-adding benefits that are difficult or expensive for customers service providers, allows firms to customize their offerings and to provide and that are not readily available elsewhere” (Berry 720 Advances in Consumer Research Volume 34, © 2007 Advances in Consumer Research (Volume 34) / 721 1995, p. 240). From the industrial marketing perspective, Turnbull personal interaction or friendship elements to customers. Similarly, and Wilson (1989, p. 233) argue that effective structural bonds in a virtual environment, though the interactivity of the Internet can “create value to the customer and either require investment by the initiate a two-way conversation, online shoppers are hampered by buyer that cannot be salvaged if the relationship ends or may be the lack of personal contact inherent in remote transactions (Croft expensive if the buyer must supply this service themselves if they 1998). In summary, though e-retailers may build social bonds with change sources.” Accordingly, structural bonds refer to the value- individual customers through personalized e-mails or customer-to- adding services that are contained in the service-delivery system, customer interactions and friendships (Zeithaml and Bitner 2003), including knowledge and information about the industry and prod- many consumers may still prefer the social component of a service uct customization; that are not readily available elsewhere; and that encounter. are expensive for customers to supply if they terminate the relation- Because repeated interactions between customers and service ship. suppliers help customers assess the service firm’s credibility and In addition, because customers regard the provision of infor- benevolence (Donney and Cannon 1997), more interactions may mation, knowledge, and goods/services by the business and their lead to an emotional attachment through the customer’s developing participation in this provision as a mutual relationship investment a sense of mutual trustworthiness. Therefore, social bonds that and adaptation, these valuable services forge customer loyalty by stress interpersonal interactions may be more useful for customers creating a psychological barrier to terminating the relationship in person-to-person encounters than in person-to-machine (Wilson 1995). In turn, some studies suggest that structural bonds encounters. In turn, we posit that belong in the highest level of the hierarchy of relational bonds (Peltier and Westfall 2000). Accordingly, we propose that: H2: The effect of social bonds on customer loyalty is moderated by the types of customers in the service encounter, H1: The impacts of financial, social, and structural bonds on such that the effect is greater for customers of person-to- customer loyalty differ. person encounters than for those of Internet and interac- H1a: The positive impact of social bonds on customer loyalty tive voice response encounters. will be greater than that of financial bonds. H1b: The positive impact of structural bonds on customer Finally, structural bonds also may affect customers of the three loyalty will be greater than that of social bonds. types of encounters differently. The development of electronic technologies has dramatically reduced the time and cost of The Role of the Type of Service Encounter information exchange, effectively linked processes, and enabled Prior studies have demonstrated the importance of service close integration of different parties in the value chain (Tang et al. encounters to businesses. For example, encounters affect critical 2001). Through the Internet, customers may receive personalized outcomes such as customers’ intention to repurchase, relationship services (Wilson 1995). By building fast, well-structured hyperlinks quality, and loyalty (Bitner et al. 2000). A service encounter, with Web sites of strategic partners, businesses can also provide defined as a period of time during which customers interact directly more integrated services to their customers, which may lead to with a service (Lovelock 2001), may take place in an actual service improved customer loyalty. setting, over the telephone, through the mail, or even over the In addition, online customers generally want to access product Internet (Bitner et al. 2000). specifications, usage instructions, warranty information, and lists Prior research about service encounters focused on person-to- of products on sale (Burke 2002). They also want to be able to person encounters or interpersonal interactions (Morgan and Chadha receive an e-mail message confirming that the order was received 1993). However, to reduce costs and improve customers’ conve- and shipped, and then track their shipment on the Web. nience, many companies have redesigned their customer flows. Compared to an Internet encounter, salespersons offer some Instead of encouraging situations where customers deal with the similar benefits as they visit customers regularly to provide product company representative in a person-to-person setting, these compa- information and resolve customer concerns, particularly in industries nies encourage lower contact levels where customers have remote like financial services. From the customer’s perspective, however, contact with company representatives (Lovelock 2001). Accord- structural bonds developed through salespeople are probably less ingly, recent research has investigated person-to-machine encoun- timely and less convenient than Internet encounters. Similarly, ters (e.g., Bitner et al. 2000; Meuter et al. 2000). though interactive voice response systems are easily accessible and In all of these settings, the same three types of bonds may apply provide timely information, they are not customized for each to the company’s attempts to retain its long-term customer relation- customer and are therefore likely to offer a weaker structural bond. ships. Because financial bonds are easily imitated and do not offer Accordingly, we propose that the impact of structural bonds on any differences with competitors in the long run (Peltier and customer loyalty is lower for customers in person-to-person Westfall 2000), we omit any discussion of their effects on customer encounters and interactive voice response systems than for those in loyalty across service encounters. Internet encounters. In turn, we posit that: Social bonds, however, may affect customers in these types of encounters (i.e., person-to-person, interactive voice response sys- H3: The effect of structural bonds on customer loyalty is tem, and Internet) differently. In person-to-machine encounters, moderated by the type of customers in service encounter, customers obtain services primarily through self-service technolo- such that the effect is greater for customers of the Internet gies, which, according to Sauer and Burton (1999), are hard-to- encounter than those of person-to-person and interactive reproduce social elements. In Donthu and Garcia’s (1999) study, voice response encounters. they reveal that Internet shoppers’ socioeconomic, motivational, and attitudinal characteristics differ from those of non-Internet METHODOLOGY shoppers; furthermore, decreased pressure from salespeople contributes to why consumers make online purchases (GVU 1997). In Sample this context, interactive voice response systems are easily acces- To examine H1–H3, we conducted a field study in the securi- sible and provide timely information, but they seldom offer inter- ties industry in Taiwan. Questionnaires were distributed to one 722 / Building Customer Relationships: A Comparison Across Multiple Service Encounters thousand securities customers who experienced person-to-person, 8.54. Although the χ2 values of the CFA models of the three interactive voice response systems, and Internet encounters. To relational bonds were 408 (d.f.=167, p<.05), the χ2 / d.f., CFI, GFI, recruit respondents we used customer lists obtained from sponsor SRMR, and RMSEA were 2.4, .87, .86,. 05, and .06, respectively. securities companies and employees of high-tech or service indus- These latter values suggest an adequate fit of the model to the data. tries with experience in stock exchanges. A total of 621 responses Furthermore, convergent and discriminant validity should be were gathered, which represents a response rate of 62.1%. T-test examined for construct validity, which is supported when the results reveal there are no significant differences across these average variance extracted (AVE) between the constructs and their demographics between these two samples. measures is greater than .50 (Fornell and Larcker 1981) and the Each respondent was asked to select one of the three encounters loading on the hypothesized construct is significant. Here, the he or she confronted most frequently. Then the respondent selected AVEs for the financial, social, and structural factors were .59, .64, one securities company in the context of this type of encounter that and .68, respectively, and the factor loadings were all significant at he or she had patronized previously to purchase the good/service. p<.05. Thus, convergent validity is supported. The AVE value also Finally, the respondent indicated his or her perceptions of the can be used to assess discriminant validity (Fornell and Larcker company. 1981); discriminant validity is supported if the AVE for each factor According to the received questionnaires, the sample of re- and its measures is greater than the square of the estimated correlation spondents consists of approximately the same percentage of women between factors. Discrimination was evident in this study because and men (50.1% and 49.9%, respectively), who range in age from the largest shared variance among the three factors (.46) was less 15 to 60 years. The sample sizes for the three types of encounters than the lowest AVE value for each factor and its measures (.59). were as follows: 387 person-to-person encounters, 116 interactive voice response system encounters, and 118 Internet encounters. Relationship between Relational Bonds and Customer Loyalty To investigate the impacts of the three relational bonds on Measures customer loyalty, we calculated composite scores for each bond by We conducted three separate in-depth interviews with two summing its measurement items. These composite (or simple entry-level managers in the securities industry to determine how factor) scores often are highly correlated with factor scores ob- they implemented their relationship strategies. From these investi- tained by the more complex least squares and regression methods gations, we identified 20 measurement items that we categorized as (Johnson and Wichern 1992). In addition, the variance inflation financial (4 items such as “The company provides cumulative factors (VIF) values were much below the threshold of 10. There- points programs” and “The company offers more rebates if I trade fore, the effects of multicollinearity can be ignored in this study. We beyond a certain amount.”), social (7 items such as “The company then conducted a regression using customer loyalty as a dependent is concerned with my needs,” “Service providers help resolve my variable and the three relational bonds as predictors. The results problem even if it is not related to the business,” and “The company indicate that the structural bond and the social bond significantly sends me greeting cards or gifts on special days,”), or structural (9 influence customer loyalty (β=.50 and .18, p<.05), whereas the items such as “The company suggests suitable investment options financial bond has an insignificant impact on customer loyalty (β=- according to my records and assets,” “The company integrates .05, p>.05). Therefore, H1a is supported. services from other sources to solve my problem,” and “The To further investigate whether the unstandardized regression company provides professional knowledge in the financial area.”). coefficients of structural bonds and social bonds (which were Each item was scored on a seven-point Likert scale (1=strongly significantly greater than 0) can be treated as equal, we conducted disagree; 7=strongly agree). a t-test. The results indicate that the coefficient of the structural We refined the instrument through a pilot test. We applied an bonds is significantly larger than that of the social bonds (t=3.81, item-to-total correlation analysis to find any items that needed to be p<.05), in support of H1b. As we have already discussed, the dropped at this stage, but because we could not improve any alpha insignificant financial bond effect may reflect the ease with which coefficients in the financial, social, or structural dimensions by competitors can imitate a financial bond strategy (Berry 1995). deleting any items, we did not remove any. Virtually every securities company provides financial bonds to its The loyalty construct often is gauged by the relative attitude customers, which decreases their effectiveness. In addition, we find that a consumer has toward a company and his or her intention to that structural bonds are the most effective in enhancing customer purchase again from the same company (e.g., Jones and Farquhar loyalty, in line with Berry’s (1995) suggestion that structural bonds 2003; Peltier and Westfall 2000). To measure the dependent variable create strong foundations for customer relationships. of customer loyalty, we adopted three indicators in accordance with Furthermore, to examine the moderating effects of the service the literature: “I am willing to repurchase” (Peltier and Westfall encounters, we tested H1–H3 through hierarchical regression analy- 2000), “I am willing to purchase other products or services from this ses, with customer loyalty as the dependent variable and a specific firm” (Sirohi et al. 1998), and “I am willing to recommend this firm relational bond as the predictor. We employed hierarchical regres- to others” (Peltier and Westfall 2000). Again, respondents used a sion analyses in which we mean centered all the variables to reduce seven-point scale to respond to these items. the risk of multicollinearity. If there is a significant change in R2 when the interaction between a relational bond and the encounter is DATA ANALYSIS added, moderation is supported (Arnold 1982). In this study, because our service encounter variable has three levels (i.e., person- Reliability and Construct Validity to-person, interactive voice response system, and Internet), we code To investigate the reliability of the scale, we computed the two dummy variables that provide qualitative information. Cronbach’s alphas for the financial, social, and structural bonds as In table 1, we report a series of hierarchical models for H2–H3. .85, .93, and .94, respectively. For the dependent variable of Moderation is supported for social (∆R2=.012, p<.05) and struc- customer loyalty, the coefficient alpha was .83. These values tural (∆R2=.01, p<.05) bonds; moderation is not supported for suggest reasonable internal consistency. financial bonds (∆R2=.002, p>.05). To test the construct validity of the three relational bonds, we To further investigate whether a specific relational bond has a conducted a confirmatory factor analysis (CFA) using LISREL differential impact on loyalty for a specific encounter, we tested Advances in Consumer Research (Volume 34) / 723 TABLE 1 RESULTS OF HIERARCHICAL REGRESSION ANALYSES

Setep IRndependent Variabl Total 2 ∆R2 1 Financial bond and service encounters .109 .109* 2 With financial bond x service encounters .111 .002 1 Social bond and service encounters .287 .287* 2 With social bond x service encounters .299 .012* 1 Structural bond and service encounters .380 .380* 2 With structural bond x service encounters .390 .010* *p<.05.

TABLE 2 RESULTS OF STANDARDIZED REGRESSION COEFFICIENTS Types of Encounters Onverall Sample Person-to-Perso Interactive Voice Internet Encounter Variables Encounter Response System Independent variables Financial bonds - .05 - .06 - .12 - .02 Social bonds .18** .26** .14 .04 Structural bonds .50** .47** .44** .63** R2 .40 .45 .24 .46 F value 82.34** 61.65** 7.01** 19.42** **p<.05.

three separate regression models with customer loyalty as the acting with self-service technologies rather than employees to dependent variable and each type of relational bond as the predictor create service outcomes, Meuter et al. (2000) suggested that re- for three different encounters. To avoid the issues related to high search should compare the underlying differences in how custom- levels of multicollinearity among independent variables, which ers respond to telephone- versus Internet-based self-service tech- make it difficult to draw inferences about regression estimates, we nologies. We responded to this suggestion by investigating the determined whether the data fit the condition of multicollinearity impacts of the three relational bonds on loyalty for customers in prior to the regression. The variance inflation factors values ranged different types of encounters. from 1.52 to 2.29 and thus are below the suggested threshold of 10. The empirical results indicate that financial bonds have no Therefore, the effects of multicollinearity can be ignored in this significant impact on loyalty for customers in any type of encoun- study. ter. Social bonds, however, have a significantly positive impact on Table 2 also indicates the relationships between the three customer loyalty, but only in person-to-person encounters. Finally, relational bonds and customer loyalty during different encounters. structural bonds are effective in gaining customer loyalty in every The social bond has a significantly positive impact on customer type of encounter, though they are more useful in Internet situations loyalty only in person-to-person encounters (β=.26, p<.05), in than in other types of encounters. This finding suggests that e-tailers support of H2. The structural bond has the greatest impact on should leverage structural bonds to strengthen their customers’ customer loyalty in Internet encounters (β=.63, p<.05). To investi- loyalty. gate the significance of H3, we also compare the unstandardized regression coefficients between structural bonds and customer Managerial Implications loyalty across service encounters (Arnold 1982). The results indi- Loyal customers buy more, are willing to spend more, are cate that the impact of structural bonds on loyalty is significantly easier to reach, and act as enthusiastic advocates for firms (Harris higher in Internet than in person-to-person (p<.05) or interactive and Goode 2004). The results of this study therefore provide some voice response system (p<.05) encounters in one-tailed tests. H3 is key strategic implications for managers, especially those in the therefore supported. Because the Internet provides an inexpensive securities industry, who seek to build customer loyalty. means to search, organize, and disseminate complete, relevant, Identify the nature of the bonds between consumers and timely, up-to-date, and accessible information, it may be more businesses. When there is effectively none or little difference effective for companies to deliver structural bonds to customers between suppliers, customer switching is very possible because one through Internet encounters. service or supplier is essentially equivalent to another (Turnbull and Wilson 1989). If, as recent research argued, the main purpose of DISCUSSION managing customers is to improve long-term profitability, businesses Service encounters constitute critical moments of truth where must identify their most profitable customers and strengthen those a firm leaves an indelible impression on the customer (Bitner et al. relationships through relational bonds. Therefore, understanding 2000). Recently, because growing numbers of customers are inter- the key factors that bond customers to firms is important for 724 / Building Customer Relationships: A Comparison Across Multiple Service Encounters customer loyalty efforts and, in turn, for businesses’ profit. Managers of Chinese culture appears to be a more collectivistic orientation, and employees must be aware of the three types of relationship whereas North American culture is typically characterized as strategies to enhance customer loyalty. According to the results of individualistic (Hofstede, 1980; Williams et al. 1998). According to this study, structural bonds are the most effective and financial Williams et al. (1998), structural bonds have a greater influence on bonds are the least effective ways to enhance customer loyalty customer commitment in individualistic countries than in overall. collectivistic countries, whereas social bonds are more significant Differentiate relational bonds in service encounters. for collectivistic countries than for individualistic countries. Traditionally, research about service encounters has focused on Although this article sheds light on how different types of relational person-to-person encounters. However, to reduce costs and achieve bonds affect customer loyalty in the Taiwan securities industry, greater customer convenience, many companies have turned to additional research should also examine the generalizability of our person-to-machine encounters. The results of our study indicate results to other countries—possibly those considered more that customers in person-to-person encounters find both social and individualistic—and compare the results with this study. structural bonds useful; however, the structural bond is the most effective for those in the Internet encounters. According to Garvin REFERENCES (1987), managers tend to employ only a few strategies in their drive Arnold, Hugh J. (1982), “Moderator Variables: A Clarification to compete with other companies, so these findings should help of Conceptual, Analytic, and Psychometric Issues,” Organi- companies identify attributes that they should prioritize in their zational Behavior and Human Performance, 29(2), 143-174. encounters with their customers. Berry, Leonard L. (1995), “Relationship Marketing of Services: Develop marketing investment and training programs that Growing Interest, Emerging Perspective,” Journal of the lead to customer loyalty in different encounters. The careful Academy of Marketing Science, 23(4), 236-245. development of a relationship investment strategy makes a company Bitner, Mary J., Stephen W. Brown, and Matthew L. Meuter more venerable to attack from competitors (Turnbull and Wilson (2000), “Technology Infusion in Service Encounters,” 1989), but careful allocation of the marketing budget can be Journal of the Academy of Marketing Science, 28(1), 138- planned according to the results of this study. For example, a 149. business might use the measurement items developed in this research Burke, Raymond R. (2002), “Technology and the Customer to benchmark its relationship activities with those of competitors Interface: What Consumers Want in the Physical and Virtual and identify comparative strengths and weaknesses from a customer Store,” Journal of the Academy of Marketing Science, 30(4), standpoint. The company then could develop a bonding strategy 411-432. based on the customers that appear in each type of encounter. For Croft, Martin (1998), “Shopping at Your Convenience,” example, customers in person-to-person encounters place a great Marketing Week, 21(18), 36-37. deal of importance on personal ties, so the company should invest Detlor, Brian (2003), “ Internet-Based Information Systems Use more to develop personal ties with this group. However, for in Organizations: An Information Studies Perspective,” customers in an Internet situation, value-added services should Information Systems Journal, 13(2), 113-132. constitute the primary investments the company uses to manage its Dibb, Sally and Maureen Meadows (2001), “The Application of customer relationships. The business might focus on building a Relationship Marketing Perspective in Retail Banking,” technologies that record customer preferences and then provide The Service Industries Journal, 21(1), 169-194. detailed information and customized services to them. Donney, Patricia M. and Joseph P. Cannon (1997), “An Examination of the Nature of Trust in Buyer-Seller Future Research Directions Relationships,” Journal of Marketing, 61(2), 35-51. Additional research might take some of the following direc- Donthu, Naveen and Adriana Garcia (1999), “The Internet tions. First, further research might study the problems of cross- Shopper,” Journal of Advertising Research, 39(3), 52-58. channels in different buying stages. To enhance customer loyalty, Fornell, Claes and David F. Larcker (1981), “Evaluating firms likely use a combination of relational bonds at different Structural Equation Models with Unobservable Variables customer decision stages. For example, customers may use the and Measurement Error,” Journal of Marketing Research, Internet to search for information but eventually make the purchase 18 (1), 39-50. from local brick-and-mortar stores. Alternatively, a salesperson Garbarino, Ellen and Mark S. Johnson (1999), “The Different might send an e-mail to a customer to provide relevant information Roles of Satisfaction, Trust, and Commitment in Customer before his or her visit to the store, or Web sites might offer contact Relationships,” Journal of Marketing, 63(2), 70-87. information for service/sales personnel, which interested custom- Garvin, David A. (1987), “Competing on the Eight Dimensions ers can use. Therefore, research should investigate the effects of a of Quality,” Harvard Business Review, 65(6), 101-109. combination of relational bonds and service encounters at different GVU (1997), “GVU’s WWW User’s Survey,” http:// buying stages. www.gvu.gatech.edu. Second, some studies have suggested that organizations should Gwinner, Kevin P., Dwayne D. Gremler, and Mary Jo Bitner analyze the position of their customers according to a continuum of (1998), “Relational Benefits in Service Industries: The transactional to collaborative exchanges and then apply transactional Customer’s Perspective,” Journal of the Academy of or relational marketing to those customers on the basis of their Marketing Science, 26 (2), 101-114. orientation (e.g., Garbarino and Johnson 1999). Therefore, further Haring, Marten Den and Jan Mattsson (1999), “A Linguistic research might examine the relationship between the three relational Approach to Studying Quality of Face-to-Face Communica- bonds and customer loyalty for low versus high relational customers. tion,” The Service Industries Journal, 19(2), 28-48. For example, might social bonds be more effective in developing Harris, Lloyd C. and Goode, Mark M. H. (2004), “The Four customer loyalty than using structural bonds for newer customers? Levels of Loyalty and the Pivotal Role of Trust: A Study of Third and finally, Hofstede (1980) proposes four dimensions Online Service Dynamics,” Journal of Retailing, 80(2), 139- of culture: power distance, uncertainty avoidance, individualism/ 158. collectivism, and masculinity/femininity. The primary characteristic Advances in Consumer Research (Volume 34) / 725 Hofstede, Geert (1980). Culture’s Consequences. Beverly Hills, CA: Sage. Jacoby, Jacob and Kyner, David B. (1973), “Brand Loyalty vs. Repeat Purchasing Behavior,” Journal of Marketing Research, 10(1), 1-9. Johnson, Richard A. and Dean W. Wichern (1992), Applied Multivariate Statistical Analysis, 3rd ed., Englewood Cliffs, NJ: Prentice-Hall. Jones, Henry and Farquhar, Jillian Dawes (2003), “Contact Management and Customer Loyalty,” Journal of Financial Services Marketing, 8(1), 71-78. Lovelock, Christopher H. (2001), Services Marketing: People, Technology, Strategy, 4th ed. Upper Saddle River, NJ: Prentice Hall. Meuter, Matthew L., Amy L. Ostrom, Robert I. Roundtree, and Mary Jo Bitner (2000), “Self-Service Technologies: Understanding Customer Satisfaction With Technology- Based Service Encounters,” Journal of Marketing, 64(3), 50- 64. Morgan, Robert E. and Sanjay Chadha (1993), “Relationship Marketing at the Service Encounter: The Case of Life Insurance,” The Service Industries Journal, 13(1), 112-125. Morris, D. S., Barnes, Bradley R. Barnes, and James E. Lynch (1999), “Relationship Marketing Needs Total Quality Management,” Total Quality Management, 10(4/5), S659- 665. Oliver, Richard L. (1999), “Whence Consumer Loyalty?” Journal of Marketing, 63 (Special issue), 33-44. Peltier, James W. and John Westfall (2000), “Dissecting the HMO-Benefits Managers Relationship: What to Measure and Why,” Marketing Health Service, 20(2), 4-13. Sauer, Chris and Suzan Burton (1999), “Is There a Place for Department Stores on the Internet? Lessons From an Abandoned Pilot,” Journal of Information Technology, 14(4), 387-398. Sirohi, Niren, Edward W. McLaughlin, and Dick R. Wittink (1998), “A Model of Consumer Perceptions and Store Loyalty Intentions for a Supermarket Retailer,” Journal of Retailing, 74(2), 223-245. Tang, Jeung-Tai Eddie, Daniel Y. Shee, and Tzung-I Tang (2001), “A Conceptual Model for Interactive Buyer-Seller Relationship in Electronic Commerce,” International Journal of Information Management, 21(1), 49-68. Thorbjørnsen, Helge, Magne Supphellen, Herbjorn Nysveen, and Per Egil Pedersen (2002), “Building Brand Relationships Online: A Comparison of Two Interactive Applications,” Journal of Interactive Marketing, 16(3), 17-34. Turnbull, Peter W. and David T. Wilson (1989), “Developing and Protecting Profitable Customer Relationships,” Indus- trial Marketing Management, 18(3), 233-238. Turner, Ralph H. (1970), Family Interaction, New York: John Wiley & Sons. Williams, Jerome D., Sang-line Han, and William J. Qualls (1998), “A Conceptual Model and Study of Cross-Cultural Business Relationships,” Journal of Business Research, 42(2), 135-143. Wilson, David T. (1995), “An Integrated Model of Buyer-Seller Relationships,” Journal of the Academy of Marketing Science, 23(4), 335-345. Zeithaml, Valarie A. and Mary Jo Bitner (2003), Services Marketing: Integrating Customer Focus across the Firm, 3rd ed. Boston, MA: Irwin/McGraw-Hill. Switching Barriers in the Four-Stage Loyalty Model Markus Blut, University of Muenster, Germany Heiner Evanschitzky, University of Muenster, Germany Verena Vogel, University of Muenster, Germany Dieter Ahlert, University of Muenster, Germany

ABSTRACT CONCEPTUAL FRAMEWORK Oliver (1997) suggests a four-stage loyalty model proposing Until the 1970’s, loyalty was understood as repeat purchase that loyalty consists of belief, affect, intentions, and action. Al- behavior, primarily considering repeat purchase cycles (Bass 1974). though this model has recently been subject to empirical examina- Following that, a behavioral approach toward explaining purchase tion, the issue of moderator variables has been largely neglected. patterns emerged. Among the first proponents of such a behavioral This article fills that void by analyzing the moderating effects of approach was Jacoby (1973, 1978). Loyalty was defined as a biased switching barriers, using a sample of 589 customers of a large do- (nonrandom) repeat purchase of a specific brand (from a set of it-yourself (DIY) retailer. The results suggest that these moderators alternatives) over time by a consumer, using a deliberate evaluation exert an influence on the development of the different stages of the process (Jacoby and Kyner 1973). Later, Jacoby and Chestnut loyalty sequence. Specifically, switching costs, social benefits, and (1978) note that the belief, affect, and intention structure of a the attractiveness of alternatives are found to be important modera- consumer must be examined in order to analyze loyalty. tors of the links in the four-stage loyalty model. Despite these seminal works, there is still no universal agree- ment on the definition of loyalty (Dick and Basu 1994; Jacoby and INTRODUCTION Chestnut 1978; Oliver 1999; Uncles, Dowling, and Hammond Numerous studies have linked customer satisfaction to finan- 2003). According to Uncles et al. (2003), three popular cial outcomes (Anderson, Fornell, and Lehmann 1994; Anderson, conceptualizations of loyalty exist: loyalty as an attitude that leads Fornell, and Rust 1997; Bernhardt, Donthu, and Kennett 2000; to a relationship with the brand; loyalty expressed mainly in terms Ittner and Larcker 1998; Keiningham et al. 1999). However, in of revealed behavior; and buying moderated by the individual’s moving from a transaction orientation to a relationship orientation characteristics, circumstances, and/or the purchase situation. (Berry 1995; Grönroos 1995; Morgan and Hunt 1994), contempo- We use Oliver’s (1997) definition, because it includes both rary marketing thought acknowledges that gaining and sustaining attitudinal and behavioral aspects of loyalty. Oliver (1997) defines customer loyalty as the ultimate goal may be more important than loyalty as a deeply held commitment to rebuy or repatronize a achieving customer satisfaction (Agustin and Singh 2005). preferred product or service consistently in the future, thereby Obviously, the link between customer satisfaction, customer causing repetitive same-brand or same brand-set purchasing, de- loyalty, and financial outcome is not as straightforward as it may spite situational influences and marketing efforts that have the seem (Carroll 1991; Carroll and Rose 1993; Reinartz and Kumar potential to cause switching behavior. He introduces a four-stage 2000). Yet researchers and managers acknowledge that small loyalty model, implying that different aspects of loyalty do not changes in loyalty and retention can yield disproportionately large emerge simultaneously, but rather consecutively over time (Oliver changes in profitability (Reichheld 1993; Reichheld, Markey, and 1999). More than a clarification, this model extends the loyalty Hopton 2000; Reichheld and Teal 1996). sequence “cognitive-affective-conative” by including an observ- Despite this obvious managerial relevance, earlier research able behavior, for example actual purchase behavior. At each primarily analyzed the link between satisfaction ratings and repur- loyalty stage, different factors influencing loyalty can be detected. chase intention. Few studies have examined the link between satisfaction ratings and repurchase behavior (Mittal and Kamakura Cognitive Loyalty 2001; Zeithaml 2000). Adding to that stream of research, Seiders et At this stage, consumer loyalty is determined by information al. (2005) summarize and extend the literature by proposing that the relating to the offering, such as price, quality, and so forth. It is the relationship between satisfaction and repurchase behavior is mod- weakest type of loyalty, since it is directed at costs and benefits of erated by customer, relational, and marketplace characteristics. an offering and not at the brand itself. Therefore, consumers are Only recently, has Oliver’s (1997) four-stage loyalty model been likely to switch once they perceive alternative offerings as being subject to more extensive empirical testing (Evanschitzky and superior with respect to the cost-benefit ratio (Kalyanaram and Wunderlich 2006; Harris and Goode 2004; Olsen 2002). Our study Little 1994; Sivakumar and Raj 1997). Cognitive loyalty is influ- adds to that stream of research by (1) discussing the links between enced largely by the consumer’s evaluative response to an experi- the four loyalty stages and (2) identifying switching barriers as ence, in particular to the perceived performance of an offering potential moderators of these links. relative to price (= value). In the following article, we briefly discuss earlier conceptualizations of the loyalty construct. Oliver’s (1997, 1999) Affective Loyalty framework of the four-stages of loyalty is then introduced. The Affective loyalty relates to a favorable attitude towards a focus of analysis is on switching barriers as moderator variables specific brand. Attitude itself is a function of cognition (e.g., which affect the links between the adjacent loyalty stages. A set of expectation). Satisfaction is a global affect evaluation or feeling hypotheses is derived and subsequently tested, using a sample of state which can be predicted from perceived performance as the 589 customers of a large DIY retailer. In the final section of the cognitive component of the evaluation (Oliver 1993; Phillips and article, we discuss implications of our findings, with a focus on Baumgartner 2002; Westbrook and Oliver 1991). Expectancy generating a deeper understanding of loyalty. confirmation leads to satisfaction, which in turn effectuates affec-

726 Advances in Consumer Research Volume 34, © 2007 Advances in Consumer Research (Volume 34) / 727 tive loyalty (Bitner 1990). Oliver (1997) defines satisfaction as “the Therefore the customers will not seek alternatives to their current consumer’s fulfillment response, the degree to which the level of provider, even if quality varies. Hence, consumers with higher fulfillment is pleasant or unpleasant.” Affective loyalty is also social bonds rely more on past experiences with a particular retailer subject to deterioration, caused primarily by an increased attrac- as a key information cue. Since it is riskier to switch a provider and tiveness of competitive offerings (Sambandam and Lord 1995) and to try something new, customers with lower social bonds should be an enhanced liking for competitive brands. This can be, for in- less likely to remain affectively loyal if their satisfaction levels stance, conveyed through imagery and association used in competi- change. Empirical evidence in the context of loyalty shows that tive communications (Oliver 1999). social benefits moderate the relationship between various aspects of satisfaction and selected measures of loyalty. For instance, Jones Conative Loyalty et al. (2000) found that the relationship between satisfaction and Conative loyalty implies that attitudinal loyalty must be ac- repurchase behavior is moderated by social benefits. Similarly, companied by a desire to intend an action, for example repurchase Crosby, Evans, and Cowles (1990) offer some support for such a a particular brand. It is stronger than affective loyalty, but has moderating effect. They were able to demonstrate that the quality vulnerabilities as well. Repeated delivery failures are a particularly of the relationship between salesperson and the customer deter- strong factor in diminishing conative loyalty. Consumers are more mines the probability of continued interchange. Holloway (2003) likely to try alternative offerings if they experience frequent service discusses social bonds as switching barriers, which especially failures. Even though the consumer is conatively loyal, he has not influence the relationship between the perception of service quality developed the resolve to avoid considering alternative offerings and satisfaction. Based on these findings, a moderating effect of (Oliver 1999). social benefits on the four-stage loyalty model can be assumed.

Action Loyalty H1: As perceived social benefits increase, the link between Action control studies imply that not all intentions are trans- cognitive and affective loyalty will become stronger. formed into action (Kuhl and Beckmann 1985). The three previous loyalty states may result in a readiness to act (in this case, to buy). Attractiveness of Alternatives This readiness is accompanied by the consumer’s willingness to Depending on the quality of competing alternatives, the cus- search for the favorite offering despite considerable effort neces- tomer perceives a benefit in changing the provider (Oliver 1997). sary to do so. Competitive offerings are not considered as alterna- The more attractive the alternatives are, the higher the perceived tives. benefits when switching (Jones et al. 2000). Therefore, consumers Despite the many attempts to consider selected links between are likely to switch once they perceive alternative offerings as being different loyalty phases, relatively little empirical research has been superior with respect to the cost-benefit ratio (Kalyanaram and conducted on testing the total four-stage loyalty model. Our study Little 1994; Sivakumar and Raj 1997). There is also empirical tries to fill that void by empirically testing Oliver’s (1997) loyalty evidence from Rusbult, Zembrodt, and Gunn (1982), reporting that model and possible moderators affecting the links between the the quality of alternatives is associated positively with exiting and loyalty phases in a retail setting. negatively with loyalty. In addition to possible direct effects of attractiveness of alternatives on loyalty, moderating effects can be SWITCHING BARRIERS AND CUSTOMER anticipated: In line with Oliver (1997), it can be expected that the LOYALTY link between cognitive and affective loyalty will be stronger under In the last decade, a substantial body of research has been the condition of unattractive alternatives. Hence: conducted to investigate moderating variables influencing the formation of customer loyalty (Bloemer and Kasper 1995; Hom- H2a: As the attractiveness of competing alternatives de- burg and Giering 2001; Homburg, Giering, and Menon 2003). In creases, the link between cognitive and affective loyalty this context, some studies focus explicitly on the moderating effects will become stronger. of switching barriers (Bell, Auh, and Smalley 2005; Jones, Mothersbaugh, and Beatty 2000; Lam et al. 2004). Switching Furthermore, the deterioration of loyalty at the conative stage barriers are defined as “any factor, which makes it more difficult or of Oliver’s (1997) model is caused primarily by an increased costly for consumers to change providers” (Jones et al. 2000). In attractiveness of competitive offerings (Sambandam and Lord line with the literature (Caruana 2003; Jones et al. 2000), we 1995). Conversely, the less attractive the alternatives, the more examine the following switching barriers: social benefits, attrac- favorably customers perceive the offers of its current provider tiveness of alternatives, and perceived switching costs. (Jacoby, Speller, and Kohn 1974). Consumers who generally prefer shopping at a certain retailer might therefore differ in their intention Social Benefits to repurchase, depending on the perception of the attractiveness of Customers build interpersonal relationships with service per- alternatives. In line with Oliver (1997), we conclude: sonnel. These bonds between the customers and the firm result in the former receiving social benefits (Berry and Parasuraman 1991). H2b: As the attractiveness of competing alternatives de- The same interactions can lead consumers to develop strong per- creases, the link between affective and conative loyalty sonal relationships with the company (Grönroos 1990; Parasuraman, will become stronger. Zeithaml, and Berry 1985) and bind customers (Bateson and Hoffman 1999). As interactions between provider employees and Perceived Switching Costs customers are repeated over time, the motivation for the develop- In many instances, customers are loyal to a provider, because ment of a social aspect to the relationship necessarily increases of the difficulty of changing to a different firm. In accordance with (Czepiel, Solomon, and Suprenant 1985). In addition, the more Jones et al. (2000), switching is likely to involve various behavioral customers rely on the personnel, and the more trust the customers and psychological aspects, given that switching costs include the develop, the less risky customers perceive variations in quality. time, money and effort the customer perceives, when changing 728 / Switching Barriers in the Four-Stage Loyalty Model FIGURE 1 CONCEPTUAL MODEL AND HYPOTHESES

from one provider to another, more specifically, they entail search respect, switching costs might serve as an aid to transform intention and learning costs (Jones, Mothersbaugh, and Beatty 2002). The into action. Therefore, we expect: consumers already know the routines of their current provider, acting as a kind of specific investment, whereas these investments H3: As perceived switching costs increase, the link between were lost when changing to another provider. Switching costs can conative and action loyalty will become stronger. affect loyalty, such as with increasing perceived costs of an activity, the probability of a consumer acting that way diminishes. Figure 1 summarizes the conceptual model of our study and The impact of switching costs on loyalty has received rela- the hypotheses we derived. tively little attention in the literature (Burnham, Frels, and Mahajan 2003; Dick and Basu 1994; Guiltinan 1989). Aaker (1991) suggests METHOD that the analysis of switching costs can provide a basis for brand loyalty. There is empirical evidence that higher switching costs Data Collection Procedure and Sample positively influence customer loyalty (Burnham et al. 2003). Fur- We drew our sample for a retailer in the German DIY market thermore, switching costs are believed to moderate the link between in the summer of 2005. This retail market is highly competitive, satisfaction and repurchase intention (Jones et al. 2002). Therefore, with the top ten retailers accounting for roughly 80% of total sales. in addition to the possible direct effects, we believe that switching The retailer in our study is quite representative for this market in costs moderate the link between conative loyalty and action loyalty. terms of size and success. We randomly selected 2,500 customers Comparing two consumers who intend to purchase at a certain of that retailer and mailed them questionnaires with pre-paid return retailer, with one perceiving high switching costs while the other envelops. To avoid any bias, the return envelops were addressed to does not, the “locked-in” customer is much more likely to purchase the researchers’ university. A total of 589 respondents send back at that retailer, since the consumer is faced with additional time and usable questionnaires, resulting in a satisfactory response rate of effort associated with a change. Switching costs explicitly foster 23.6%. Comparing early and late responses, as suggested by transfering intentions into action–as suggested by the theory of Armstrong and Overton (1981), no signs of non-response bias were planned behavior–instead of influencing earlier stages of the four- found. stage loyalty model (Bansal and Taylor 2002). Hence, the link Since the data for dependent and independent variables were between conative and action loyalty is stronger for customers obtained from the same respondents, there is a possibility of perceiving higher switching costs. This reasoning is again consis- common method bias (CMB). We applied the methods suggested tent with that of Oliver (1997), who noted that key sustainers of by Podsakoff et al. (2003) to test for CMB, particularly the “single- action loyalty are, in fact, sunk costs, and actual purchase will be method-factor approach.” We estimated the model with a single- more likely for consumers faced with sunk costs, as opposed to method first-order factor added to the indicators of the constructs. those with no sunk costs which are associated with switching. In this The results reveal that even with common-method variance con- Advances in Consumer Research (Volume 34) / 729 trolled, fit-indices and the proportion of variance explained were Switching Costs. Finally, the switching costs measure, adapted almost unchanged in both models, and all path coefficients remain from Jones et al. (2000), captures costs across a variety of dimen- significant. Therefore, we conclude that CMB is not a significant sions and focuses on the overall perceptions of time, money, and issue in our study. effort associated with changing providers. Measurement reliability was examined through confirmatory Analysis factor analysis and Cronbach’s alpha coefficient. We use multi-group structural equation modeling to test our It can be noted that the coefficient alpha exceeds .7, the hypotheses. Median-splits based on the values of the moderator threshold generally proposed in the literature (Nunnally 1978). variables were used to create the groups. Testing for moderation, we Also, composite reliabilities exceed .6 for all constructs (Bagozzi first looked at a non-restricted model and then restricted three paths and Yi 1988). The discriminant validity of the constructs was in the four stage loyalty model so that they are equal across assessed using the criterion proposed by Fornell and Larcker (1981) subgroups. Chi-square differences with three degrees of freedom and this criterion was met. Therefore, reliability and validity of the (critical chi-square value (df=3; p=.05): 7.81) were assessed. After constructs in this study are acceptable. confirming a general moderating effect, we compared two models that only differ in one effect of one loyalty stage to the next, as RESULTS suggested by our hypotheses. One model restricts the parameter so Firstly, it is evident that the links between the various stages of as to be equal across groups, while the second model allows the loyalty model are all significant (p<.01) and the model displays variation in one of these parameters across groups. The restricted a good overall fit (χ2/df=4.312; CFI=.941; TLI=.926; SRMR=.073; model has one more degree of freedom than the general model. A RMESEA=.078). Therefore, we can continue analyzing the moder- moderating effect would be present when the improvement in chi- ating effects. square, moving for the restricted to the non-restricted model is After confirming the influence of the main effects in the significant, meaning the chi-square difference between the two model, we tested for moderator effects. A chi-square difference test models (and one degree of freedom) is larger than 3.84 (p=.05). was conducted for the three possible moderator effects, comparing a restricted and a non-restricted model. As can be seen, “attractive- Measures ness of alternatives” (at .01-level) and “switching costs” (at .05- A conceptualization and items for measuring the constructs level) have a significant, general moderating effect on all links in were developed, drawing on prior research in the loyalty literature. the loyalty model, whereas “social benefits” displays only a week Except for attractiveness of alternatives, multi-item seven-point moderating effect at .1-level. Nevertheless, we continue analyzing Likert scales (anchored at 1=strongly agree, 7=strongly disagree) the specific moderator effects suggested by our hypotheses. adapted from previous studies were used. As can be seen from table 4, each moderator only moderates Cognitive Loyalty. Cognitive loyalty was defined and mea- one link in the loyalty model. “Attractiveness of alternatives” sured as evaluation of perceived value associated with the retail moderates the link between affective and conative loyalty, “social outlet. In accordance with the operationalization of Verhoef, benefits” moderates the link between cognitive and affective loy- Langerak, and Donkers (2004), Sirdeshmukh, Singh, and Sabol alty, and “switching costs” moderates the link between conative (2002) and Baker et al. (2002), the respondents were asked to rate and action loyalty. the perceived value. Hence, three of four hypothesized moderating effects in the Affective Loyalty. Affective loyalty was defined as a person’s four-stage loyalty model were supported by these findings. It only global affect evaluation or feeling state. As proposed by Oliver seems that only “attractiveness of alternatives” fails to moderate the (1997), we used “Overall Satisfaction” and “Liking” to cover link between cognitive and affective loyalty as hypothesized in affective loyalty. H2a. Furthermore, a smaller random sample of 250 respondents Conative Loyalty. Conative loyalty was defined as a customer’s confirmed these results.1 behavioral intention to continue buying at a retail store in the future, accompanied by a deep commitment to that store. The scale used DISCUSSION, LIMITATION, FURTHER was adapted from the behavioral intention battery developed by RESEARCH Zeithaml, Berry, and Parasuraman (1996), and included the items: Gaining customer loyalty is less straightforward and more willingness to recommend, and repurchase intention. complex than several previous studies have suggested. Using Action Loyalty. Action loyalty focuses on (purchase) behav- Oliver’s (1997) model of loyalty, it is not until a customer shows ior. Hence, we do not focus on attitudes such as an intention to high consistency through the four distinct stages that loyalty is purchase or intention to overcome an obstacle. In accordance with achieved. In such case, customers possess not only a stable dispo- Harris and Goode’s (2004), we operationalize “action loyalty” as sition but also a stable, routine purchase pattern, making them displayed choice preference, and not as “overcoming obstacle.” relatively resistant to competitive offers. Therefore, examining The most common assessments of behavioral loyalty are repurchas- moderator variables of the links in the four-stage loyalty model ing patterns or behavioral frequency like word-of-mouth behavior, leads to a clearer understanding of how to build loyalty. purchase frequency, and actual money spent per year, as used here. The results of our empirical study suggest that there are in Attractiveness of Alternatives. The attractiveness of alterna- indeed moderator variables that display conditions under which a tives was measured with one item, accounting for the evaluation of customer moves from one loyalty stage to the next. In particular, existing alternatives. Research in the field of neuron science has high perceived social benefits from a relationship with a retailer are shown that the first choice is a good indicator for evaluating the likely to turn cognitively loyal customers into affectively loyal attractiveness of alternative brands (Deppe et al. 2005). customers. If the attractiveness of alternatives is relatively low, a Social Benefit. The social benefits measure was adapted from customers liking (affective loyalty) of a particular retailer can be Henning-Thurau, Gwinner, and Gremler (2002), measuring the social benefits resulting from interpersonal relationships with the service personnel. 1We thank one anonymous reviewer for suggesting this test. 730 / Switching Barriers in the Four-Stage Loyalty Model TABLE 1 DESCRIPTION OF ITEMS USED TO MEASURE THE CONSTRUCTS

Coefficient Composite Variance Scale/Item Alpha Reliability Extracted

Cognitive Loyalty (Baker et al. 2002; Sirdeshmukh et al. 2002; .862 .869 .571 Verhoef et al. 2004)* 1. How would you rate your overall shopping experience at this store? 2. The quality/price ratio with the dealer is good (merchandise). 3. The quality/price ratio with dealer is good (service). 4. For the time you spent in order to shop at this store, would you say shopping at this store is highly reasonably/highly unreasonable? 5. For the effort involved in shopping at this store, would you say shopping at this store is not at all worthwile/very worthwile? 6. The store is attractive. 7. For the prices you pay for DIY-items at this store, would you say shopping at this store is a very poor deal/very good deal?

Affective Loyalty (Bettencourt 1997; Oliver 1997)* .883 .885 .720 1. Based on all my experience with this store, I am very satisfied. 2. My shopping experiences at this store have always been pleasant. 3. Compared to other stores, I am very satisfied with this store.

Conative Loyalty (Zeithaml et al. 1996)* .804 .802 .670 1. Repurchase intention 2. Word-of-mouth intention

Action Loyalty (De Wulf, Odekerken-Schröder, and Iacobucci 2001) .850 .858 .673 1. How often do you buy DIY things in this store compared to other stores where you buy DIY things?* 2. What percentage of your total expenditures for DIY do you spend in this store?** 3. Of the 10 times you select a store to buy DIY at, how many times do you select this store?***

MODERATORS Switching Costs (Jones et al. 2000)* .863 .932 .870 1. In general, it would be a hassle changing DIY-retailers. 2. It takes me a great deal of time and effort to get used to a new company.

Social Benefits (Henning-Thurau et al. 2002)* .848 .958 .635 1. This store offers high-quality service. 2. Customers could expect to be treated well in this shop. 3. Employees of this store could be expected to give customers personal attention. 4. This store’s employees would be willing to help customers. 5. Employees of this store would not be too busy to respond to customers’ requests promptly.

Attractiveness of Alternatives* n.a. n.a. n.a. 1. This store will be my first choice for my DIY needs.

* Measured using seven-point Likert scales anchored by 1=strongly agree; 7=strongly disagree. ** Measured in percentage of total expenditures. *** Measured in times of selection. Advances in Consumer Research (Volume 34) / 731 TABLE 2 CORRELATION MATRIX

Cognitive Affective Conative Action Loyalty Loyalty Loyalty Loyalty

Cognitive Loyalty 1 Affective Loyalty .692 1 Conative Loyalty .622 .707 1 Action Loyalty .455 .526 .505 1

Average Variance .571 .720 .670 .673 Extracted

TABLE 3 PATH COEFFICIENTS (“FOUR-STAGES OF LOYALTY”)

Four-Stages of Loyalty

Coefficient R2

Cognitive Loyalty Affective Loyalty .786* .618 Affective Loyalty Conative Loyalty .799* .639 Conative Loyalty Action Loyalty .604* .365

*=.01-level.

transformed into a strong intention to repurchase. If retailers are Anderson, Eugene W., Claes Fornell, and Donald R. Lehmann able to create switching costs, those customers intended to pur- (1994), “Customer Satisfaction, Market Share, and Profit- chase, are very likely to do their actual purchase at that retailer. Our ability: Findings from Sweden,” Journal of Marketing, 58 results did not, however, find a moderating effect of the attractive- (3), 53–66. ness of alternatives on the link between cognitive and affective Anderson, Eugene W., Claes Fornell, and Roland T. Rust (1997), loyalty. A possible explanation could be that both satisfaction and “Customer Satisfaction, Productivity, and Profitability: liking are still just attitudes and consumers may not consider an Differences between Goods and Service,” Marketing actual purchase situation. However, we acknowledge that more Science, 16 (2), 129–45. research is needed to test this link in particular. Anderson, Eugene. W. and Vikas Mittal (2000), “Strengthening As with all empirical studies, our study suffers from limita- the Satisfaction-Profit Chain,” Journal of Service Research, tions. First, our object of analysis was a particular retailer from one 3 (2), 107–21. industry. Second, we analyzed data from one point in time only. By Armstrong, J. Scott and Terry S. Overton (1977), “Estimating doing so, we assume there is no time-lag between a customer’s Nonresponse Bias in Mail Surveys,” Journal of Marketing feeling of being satisfied and the attitudinal and behavioral conse- Research, 14 (3), 396–402. quences. Thirdly, we considered only linear relations between the Bagozzi, Richard. P. and Youjae Yi (1988), “On Evaluation of four loyalty stages. However, as indicated in the literature (Ander- Structural Equation Models,” Journal of the Academy of son and Mittal 2000; Keiningham, Perkins-Munn, and Evans 2003), Marketing Science, 16 (1), 74–94. these relationships might in fact be curvilinear. Furthermore, some Baker, Julie, A. Parasuraman, Dhruv Grewal, and Glenn B. Voss scholars are skeptical about the use of median split, due to the loss (2002), “The Influence of Multiple Store Environment Cues of information. Further research might consider alternative ways to on Perceived Merchandise Value and Patronage Intentions,” split the sample in appropriate sub-samples. Moreover, we only use Journal of Marketing, 66 (2), 120–41. self-reported measures for the action loyalty. Furthermore, longitu- Bansal, Harvir S. and Shirley F. Taylor (2002), “Investigating dinal data would further improve our understanding of the mecha- Interactive Effects in the Theory of Planned Behavior in a nisms influencing the link between satisfaction and loyalty in Service-Provider Switching Context,” Psychology & general, and the effect of certain moderators in particular. Marketing, 19 (5), 407–25. Bass, Frank M. (1974), “The Theory of Stochastic Preference REFERENCES and Brand Switching,” Journal of Marketing Research, 11 Aaker, David A. (1991), Managing Brand Equity: Capitalizing (1), 1–20. on the Value of a Brand Name, New York, NY: Free Press. Bateson, John E. G. and Douglas K. Hoffman (1999), “Manag- Agustin, Clara and Jagdip Singh (2005), “Curvilinear Effects of ing Services Marketing Text and Readings” (4th ed.), Fort Consumer Loyalty Determinants in Relational Exchanges,” Worth, TX: Dryden Press. Journal of Marketing Research, 42 (1), 96–108. 732 / Switching Barriers in the Four-Stage Loyalty Model TABLE 4 RESULTS OF MULTI-GROUP ANALYSIS

Attractiveness of Alternatives

Low High χ2 ∆ χ2 (df=1) Cognitive Loyalty Affective Loyalty .745 .618 413.476 1.006 Affective Loyalty Conative Loyalty .808 .589 404.281 10.261*** Conative Loyalty Action Loyalty .380 .422 413.489 1.053 ∆ χ2 (df=3): 12.654***

Social Benefits

Low High χ 2 ∆ χ2 (df=1) Cognitive Loyalty Affective Loyalty .644 .798 707.977 5.085** Affective Loyalty Conative Loyalty .767 .767 711.271 1.791 Conative Loyalty Action Loyalty .563 .569 713.062 .001 ∆ χ2 (df=3): 6.437*

Switching Costs

Low High χ 2 ∆ χ2 (df=1) Cognitive Loyalty Affective Loyalty .777 .702 445.766 .131 Affective Loyalty Conative Loyalty .829 .716 445.617 .280 Conative Loyalty Action Loyalty .365 .581 439.056 6.841*** ∆ χ2 (df=3): 7.836**

*=.1-level. **=.05-level. ***=.01-level.

Bell, Simon. J., Seigyoung Auh, and Karen Smalley (2005), Carroll, Peter (1991), “Rethinking the Basics of Retail Banking,” “Customer Relationship Dynamics: Service Quality and Journal of Retail Banking, 13 (3), 5–18. Customer Loyalty in the Context of Varying Levels of Carroll, Peter and Sanford Rose (1993), “Revisiting Customer Customer Expertise and Switching Costs,” Journal of the Retention,” Journal of Retail Banking, 15 (1), 5–13. Academy of Marketing Science, 33 (2), 169–83. Caruana, Albert (2003), “The Impact of Switching Costs on Bernhardt, Kenneth. L., Naveen Donthu, and Pamela A. Kennett Customer Loyalty: A Study among Corporate Customers of (2000), “A Longitudinal Analysis of Satisfaction and Mobile Telephony,” Journal of Targeting, Measurement and Profitability,” Journal of Business Research, 47 (2), 161–71. Analysis for Marketing, 12 (3), 256–68. Berry, Leonard L. (1995), “Relationship Marketing of Services: Crosby, Lawrence A., Kenneth R. Evans, and Deborah Cowles Growing Interest, Emerging Perspectives,” Journal of the (1990), “Relationship Quality in Services Selling: An Academy of Marketing Science, 23 (4), 236–46. Interpersonal Influence Perspective,” Journal of Marketing, Berry, Leonard L. and A. Parasuraman (1991), Marketing 54 (3), 68–81. Services: Competing through Quality, New York, NY: Free Czepiel, John A., Michael R. Solomon, and Carol F. Suprenant Press. (1985), The Service Encounter: Managing Employee/ Bettencourt, Lance A. (1997), “Customer Voluntary Perfor- Customer Interaction in Service Businesses, Lexington, MA: mance: Customers as Partners in Service Delivery,” Journal DC Heath. of Retailing, 73 (3), 383–406. Deppe, Michael, Wolfram Schwindt, Harald Kugel, Hilke Bitner, Mary J. (1990), “Evaluating Service Encounters: The Plassmann, and Peter Kenning (2005), “Non-Linear Effects of Physical Surroundings and Employee Responses,” Responses within the Medial Prefrontal Cortex Reveal when Journal of Marketing, 54 (2), 69–82. Specific Implicit Information Influences Economic Decision- Bloemer, José and Hans D. P. Kasper (1995), “The Complex Making,” Journal of Neuroimaging, 15 (2), 171–82. Relationship Between Consumer Satisfaction and Brand De Wulf, Kristof, Gaby J. Odekerken-Schröder, and Dawn Loyalty,” Journal of Economic Psychology, 16 (2), 311–29. Iacobucci (2001), “Investments in Consumer Relationships: Burnham, Thomas A., Judy K. Frels, and Vijay Mahajan (2003), A Cross-Country and Cross-Industry Exploration,” Journal “Consumer Switching Costs: A Typology, Antecedents, and of Marketing, 65 (4), 33–50. Consequences,” Journal of the Academy of Marketing Dick, Alan S. and Kunal Basu (1994), “Customer Loyalty: Science, 31 (2), 109–27. Toward an Integrated Conceptual Framework,” Journal of the Academy of Marketing Science, 22 (2), 99–113. Advances in Consumer Research (Volume 34) / 733 Evanschitzky, Heiner and Maren Wunderlich (2006), “An Keiningham, Timothy L., Melinda K. M. Goddard, Terry G. Examination of Moderator Variables in the Four-Stage Vavra, and Andrew J. Laci (1999) “Customer Delight and Loyalty Model,” Journal of Service Research, 8 (4), 330–45. the Bottom Line,” Marketing Management, 8 (3), 57–63. Fornell, Claes and David F. Larcker (1981), “Evaluating Keiningham, Timothy L., Tiffany Perkins-Munn, and Heather Structural Equation Models with Unobservable Variables Evans (2003), “The Impact of Customer Satisfaction on and Measurement Error,” Journal of Marketing Research, 18 Share-of-Wallet in a Business-to-Business Environment,” (1), 39–50. Journal of Service Research, 6 (1), 37–50. Grönroos, Christian (1990), “Relationship Approach to Market- Kuhl, Julius and Jürgen Beckmann (1985), “Historical Perspec- ing in Service Contexts: The Marketing and Behavior tives in the Study of Action Control,” in Action Control: Interface,” Journal of Business Research, 20 (1), 3–11. From Cognition to Behavior, eds. Julius Kuhl and Jürgen (1995), “Relationship Marketing: The Strategy Con- Beckmann, Berlin: Springer, 89–100. tinuum,” Journal of the Academy of Marketing Science, 23 Lam, Sun. Y., Venkatesh Shankar, M. Krishna Erramilli, and (4), 252–54. Bvsan Murthy (2004), “Customer Value, Satisfaction, Guiltinan, Joseph P. (1989). “A Classification of Switching Loyalty, and Switching Costs: An Illustration From a Costs With Implications for Relationship Marketing,” AMA Business-to-Business Service Context,” Journal of the Winter Educators’ Conference: Marketing Theory and Academy of Marketing Science, 32 (3), 293–311. Practice, eds. Terry L. Childers, Richard P. Bagozzi, and J. Mittal, Vikas and Wagner A. Kamakura (2001), “Satisfaction, Paul Peter, Chicago, IL: American Marketing Association, Repurchase Intent, and Repurchase Behavior: Investigating 216–20. the Moderating Effect of Customer Characteristics,” Journal Harris, Lloyd C. and Mark H. Goode (2004), “The Four Stages of Marketing Research, 38 (1), 131–42. of Loyalty and the Pivotal Role of Trust: A Study of Online Morgan, Robert M. and Shelby D. Hunt (1994), “The Commit- Service Dynamics,” Journal of Retailing, 80 (2), 139–58. ment-Trust Theory of Relationship Marketing,” Journal of Hennig-Thurau, Thorsten, Kevin P. Gwinner, and Dwayne D. Marketing, 58 (3), 20–38. Gremler (2002), “Understanding Relationship Marketing Nunnally, Jum C. (1978), Psychometric Theory (2nd ed.), New Outcomes: An Integration of Relational Benefits and York, NY: McGraw-Hill. Relationship Quality,” Journal of Service Research, 4 (3), Oliver, Richard L. (1993), “Cognitive, Affective, and Attribute 230–47. Bases of the Satisfaction Response,” Journal of Consumer Holloway, Betsy Bugg (2003), “The Role of Switching Barriers Research, 20 (3), 418–30. in the Online Service Recovery Process,” unpublished (1997), Satisfaction: A Behavioral Perspective on the dissertation, Department of Management and Marketing, Consumer, New York, NY: McGraw-Hill. University of Alabama, Tuscaloosa, AL. (1999), “Whence Consumer Loyalty?,” Journal of Homburg, Christian and Annette Giering (2001), “Personal Marketing, 63 (4), 33–44. Characteristics as Moderators of the Relationship between Olsen, Svein O. (2002), “Comparative Evaluation and the Customer Satisfaction and Loyalty: An Empirical Analysis,” Relationship between Quality, Satisfaction, and Repurchase Psychology & Marketing, 18 (1), 43–66. Loyalty,” Journal of the Academy of Marketing Science, 30 Homburg, Christian, Annette Giering, and Ajay Menon (2003), (3), 240–49. “Relationship Characteristics as Moderators of the Satisfac- Parasuraman, A., Valarie A. Zeithaml, and Leonard L. Berry tion-Loyalty Link: Findings in a Business-to-Business (1985), “A Conceptual Model of Service Quality and its Context,” Journal of Business-to-Business Marketing, 10 (3), Implications for Future Research,” Journal of Marketing, 49 35–62. (4), 41–50. Ittner, Christopher D. and David F. Larcker (1998), “Are Phillips, Diane M. and Hans Baumgartner (2002), “The Role of Nonfinancial Measures Leading Indicators of Financial Consumption Emotions in the Satisfaction Response,” Performance? An Analysis of Customer Satisfaction,” Journal of Consumer Psychology, 12 (3), 243–52. Journal of Accounting Research, 36 (3), 1–35. Podsakoff, Philip M., Scott B. MacKenzie, Lee Jeong-Yeon, and Jacoby, Jacob and Robert W. Chestnut (1978), Brand Loyalty: Nathan P. Podsakoff (2003), “Common Method Biases in Measurement and Management, New York, NY: Wiley. Behavioral Research: A Critical Review of the Literature and Jacoby, Jacob and David B. Kyner (1973), “Brand Loyalty vs. Recommended Remedies,” Journal of Applied Psychology, Repeat Purchasing Behavior,” Journal of Marketing 88 (5), 879–903. Research, 10 (1), 1–9. Reichheld, Frederick F. (1993), “Loyalty-Based Management,” Jacoby, Jacob, Donald Speller, and Carol Kohn (1974) “Brand Harvard Business Review, 71 (2), 64–73. Choice Behavior as a Function of Information Load,” Reichheld, Frederick F., Robert G. Markey Jr., and Christopher Journal of Marketing Research, 11 (1), 63–69. Hopton (2000), “E-Customer Loyalty–Applying the Jones, Michael. A., David L. Mothersbaugh, and Sharon E. Traditional Rules of Business for Online Success,” European Beatty (2000), “Switching Barriers and Repurchase Inten- Business Journal, 12 (4), 173–79. tions in Services,” Journal of Retailing, 76 (2), 259–74. Reichheld, Frederick F. and Thomas Teal (1996), The Loyalty (2002), “Why Customers Stay: Measuring the Underly- Effect, Boston, MA: Harvard Business Press. ing Dimensions of Services Switching Costs and Managing Reinartz, Werner J. and Vikas Kumar (2000), “On the Profitabil- Their Differential Strategic Outcomes,” Journal of Business ity of Long-Life Customers in a Noncontractual Setting: An Research, 55 (6), 441–50. Empirical Investigation and Implications for Marketing,” Kalyanaram, Gurumurthy and John D. C. Little (1994), “An Journal of Marketing, 64 (4), 17–35. Empirical Analysis of Latitude of Price Acceptance in Rusbult, Caryl E., Isabella M. Zembrodt, and Lawanna K. Gunn Consumer Package Goods,” Journal of Consumer Research, (1982), “Exit, Voice, Loyalty and Neglect: Responses to 21 (3), 408–19. Dissatisfaction to Romantic Involvement,” Journal of Personality and Social Psychology, 43 (6), 1230–42. 734 / Switching Barriers in the Four-Stage Loyalty Model Sambandam, Rajan and Kenneth A. Lord (1995), “Switching Behavior in Automobile Markets: A Consideration-Sets Model,” Journal of the Academy of Marketing Science, 23 (1), 57–65. Scott, William A., D. Wayne Osgood, and Christopher Peterson (1979), Cognitive Structure: Theory and Measurement of Individual Differences, Washington, DC: Winston. Seiders, Kathleen, Glenn B. Voss, Dhruv Grewal, and Andrea L. Godfrey (2005), “Do Satisfied Customers Buy More? Examining Moderating Influences in a Retailing Context,” Journal of Marketing, 69 (4), 26–43. Sirdeshmukh, Deepak, Jagdip Singh, and Barry Sabol (2002), “Consumer Trust, Value, and Loyalty in Relational Ex- change,” Journal of Marketing, 66 (1), 15–37. Sivakumar, K. and S. P. Raj (1997), “Quality Tier Competition: How Price Change Influences Brand Choice and Category Choice,” Journal of Marketing, 61 (3), 71–85. Uncles, Mark D., Grahame D. Dowling, and Kathy Hammond (2003), “Customer Loyalty and Customer Loyalty Pro- grams,” Journal of Consumer Marketing, 20 (4), 294–317. Verhoef, Peter C., Fred Langerak, and Bas Donkers (2004), “Understanding Brand and Dealer Retention in the New Car Market: The Moderating Role of Brand Type,” Working Paper No. ERS-2004-084-MKT, Erasmus Research Institute of Management, Rotterdam. Westbrook, Robert A. and Richard L. Oliver (1991), “The Dimensionality of Consumption Emotion Patterns and Consumer Satisfactions,” Journal of Consumer Research, 18 (1), 84–91. Zeithaml, Valarie A. (2000), “Service Quality, Profitability, and the Economic Worth of Customers: What We Know and What We Need to Learn,” Journal of the Academy of Marketing Science, 28 (1), 67–85. Zeithaml, Valarie A., Leonard L. Berry, and A. Parasuraman (1996), “The Behavioral Consequences of Service Quality,” Journal of Marketing, 60 (2), 31–46. Why Companies Should Make Their Customers Happy: The Neural Correlates of Customer Loyalty Hilke Plassman, Stanford University, USA and University of Muensterm, Germany Peter Kenning, University of Muenster, Germany Dieter Ahlert, University of Muenster, Germany

INTRODUCTION repatronize a preferred product/service consistently in the future.” The relevance of customer loyalty for company success is The repetitive same-brand or same brand-set purchasing would widely acknowledged in the literature as well as in practice occur among loyal customers, despite situational influences and (Chaudhuri and Holbrook 2001; Dick and Basu 1994; Gwinner, marketing efforts that could potentially enable switching behavior. Gremler, and Bitner 1998; Oliver 1999; Reichheld and Sasser Accordingly, commitment, i.e. the extent to which the customer 1990). While relationship marketing approaches have emphasized desires to maintain a continued relationship with the firm or brand, the importance of building personal relationships with their cus- is understood as the underlying psychological mechanism of loy- tomers, the affective foundations of customer loyalty has received alty (Allen and Meyer 1990; Morgan and Hunt 1994). Recently, less attention and is less understood than the cognitive or rational customer commitment was found to be composed of different basis of customer loyalty (Chaudhuri and Holbrook 2001; Fullerton dimensions (Bansal, Irving, and Taylor 2004; Fullerton 2003). On 2003). However, initial studies could show that for customer the one hand, these are more rational aspects such as a lack of loyalty the relative strength of affective loyalty antecedents is more alternatives and institutional or technical requirements to use a significant as compared to normative or “rational” ones (Fullerton specific product (Bendapudi and Berry 1997). On the other hand, 2003, Evanschitzky and Plassmann 2005). Given the homogeneity these are affective aspects understood as an emotional attachment of most of today’s product and service markets and other character- customers built with companies and brands (Fournier 1997; Fuller- istics that may constrict the customer’s use of “rational” criteria for ton 2003; Thomson, MacInnis, and Park 2005). product evaluation, brand-induced affect may represent a global In contrast to traditional studies we attempted here to gain evaluation measure more readily used by customers as empirical complementing insights by separating behavioral and psychologi- evidences from neighboring disciplines suggest (e.g. Bechara et al. cal aspects through the use of two different data sources. Firstly, we 1997; Bechara and Damasio 2005 and also Slovic, Finucane, Peters extracted “real” purchasing behavior from a club card database of and MacGregor 2002, 2004; Yeung and Wyer 2005). a department store including customers’ monthly spending and In this paper, we attempt to expand our understanding of the number of purchase days within a period of 6 months. Secondly, we basis of emotional bonds customers built with companies and investigated the underlying psychological states by means of func- brands. Past research could show that people sometimes use affec- tional brain imaging. In the neuropsychological literature a great tive reactions to a stimulus as information about its reward value body of studies is devoted to the impact of emotions on behavior and (Schwarz and Clore 1988, Yeung and Wyer 2004). As recent decision-making (for reviews, see for example O’Doherty 2004; studies on consumer decision-making propose that most informa- Schultz 2006). In the following we attempt to give an introductory tion processing and, in particular, the processing of emotions is overview about these studies. subconscious (see for reviews Bargh, 2002; Fitzsimons et al., 2002; Woodside, 2004; Zaltman, 2003) we followed the approaches from Neuropsychological Background Ambler et al., Deppe et al., McClure et al. and Yoon et al. and One initial common conclusion relating to the neuropsycho- employed functional brain imaging techniques to investigate the logical foundations of decision-making refers to the role of emotion neural correlates of customer loyalty (Ambler, Braeutigam, Stins, in human decision-making (e. g. Bechara 2004; Bechara and Rose, & Swithenby, 2004; Deppe, Schwindt, Kugel, Plassmann & Damasio 2005; Bechara, Damasio, and Damasio 2004; Paulus Kenning, 2005; McClure et al., 2004; Yoon, Gutchess, Feinberg, & 2005). It is suggested that “rational” or advantageous decision- Polk, 2006). making depends on prior emotional processing related to the The paper is organized as follows. First, we review concepts specific decision parameters (Bechara et al. 1997). This notion is of customer loyalty and the role of emotions within these concepts. based on neurobiological evidence that seeking rewards and avoid- Then, we provide a review of neuropsychological literature on the ing punishments guides the behavior of both, humans and animals neural foundations of emotion-based decision-making and offer (for reviews, see for example O’Doherty 2004; Schultz 2006). In hypotheses on their implications for customer loyalty. In the constantly changing environments, it is necessary to adapt deci- following, we describe our empirical study and present our results. sion-making strategies by updating changes of reward contingen- We conclude with implications for managers and suggestions for cies. Thus, central to the organization of such behavior is the ability future research. to estimate the reward value of stimuli, predict when rewards will occur and integrate these predictions into decision-making strate- CONCEPTUAL FRAMEWORK gies. Two areas found to be involved in the representation of Loyalty Concept rewards and punishments are the orbitofrontal cortex (OFC) and the Historically, loyalty was understood as observable repurchase amygdala. Human (O’Doherty, Kringelbach, Rolls, Hornak, behavior. It was defined as a biased (nonrandom) repeat purchase Andrews 2001) and animal (Rolls, Sienkiewiczs, Yaxley 1989) of a specific brand (from a set of alternatives) over time by a studies have confirmed the role of OFC in coding for stimulus consumer using a deliberate evaluation process (Jacoby and Kyner reward from a variety of sensory modalities, such as taste 1973). Jacoby and Chestnut (1978) later asserted that also psycho- (Kringelbach, O’Doherty, Rolls, and Andrews 2003), olfaction logical states of a consumer had to be examined to analyze loyalty. (Gottfried, Deichmann, Winston, Dolan 2002), somatosensory In accordance with this reasoning, customer loyalty has been (Rolls, O’Doherty, Kringelbach, Francis, Bowtell McGlone 2003), defined by Oliver (1999) as “a deeply held commitment to rebuy or auditory (Blood, Zatorre, Bermudez, Evans 1999), and vision 735 Advances in Consumer Research Volume 34, © 2007 736 / Why Companies Should Make Their Customers Happy: The Neural Correlates of Customer Loyalty (Aharon, Etcoff, Ariely, Chabris, O’Connor, Breiter 2001) as well guidelines of the University Medical School Ethical Committee. as for more abstract rewards such as money (Elliott, Newman, The sample size is smaller than in typical in behavioral studies, Longe, Deakin 2003) or sports cars (Erk, Spitzer, Wunderlich, however, it is representative of fMRI studies (see e.g. Desmond and Galley, Walter 2002). Furthermore, various studies have revealed Glover 2002; Murphy and Garavan 2004). that the amygdala is involved in the processing of aversive (Morris et al. 1996) and pleasant (Canli et al. 2002) stimuli. Experimental Procedure and Design In most consumption decision situations, however, rewards Participants were imaged during one functional run while are not received immediately, but at a later point in time and can be making choices between two out of four department stores where understood as a consequence of the decision. Thus, the ability to they would buy different types of clothing. The experimental predict future rewards is crucial for consumer decision-making. paradigm was designed to investigate systematic differences of Neuroimaging studies have found brain regions such as the amygdala, neural processing during purchase decisions in the presence or OFC, and striatum to be correlated with reward predictions (Gottfried, absence of the specific target department store (T, the department O’Doherty, and Dolan 2003; Knutson and Cooper 2005; O’Doherty store from which we extracted the subject pool), in contrast to three et al. 2004; Schultz, Dayan, and Montague, 1997). other department stores, which were classified as diverse (D1, D2, Then, the learning of stimulus-reward-response associations D3). Nine different pieces of clothing (P1, P2;…, P9) were chosen, is required to integrate anticipated and experienced values into the that could be bought in all four department stores. They were neutral decision-making process. Neuroscientific studies in this context in the sense that no brand name or other criteria that allow an have revealed that the dorsal striatum might play a crucial role in identification of a special brand were visible. The stimuli were processing contingencies between responses and rewards presented as follows. In middle of each stimulation, a piece of (O’Doherty 2004; Zink et al. 2005). Additionally, areas of emo- clothing (P1, P2,…, P9) was shown. Slightly above that piece of tional memory such as the ventromedial prefrontal cortex (VMPFC) clothing, on the left and on the right side, the logos of two need to be involved in order to retrieve rewarding memories and department stores were presented. These were either the target store relate them to the stimulus (Bechara and Damasio 2005; Paulus brand (T) and one of the diverse store brands (D1, D2, or D3), or two 2005). In the context of consumption decisions an initial study by of the diverse store brands, resulting in sequences of TDPz and Deppe at al. could show that an activation network linking the DDPz stimuli (or simplified TD and DD). The order of the stimuli ventromedial prefrontal cortex, the striatum and anterior cingulate was randomized referring to brands as well as to pieces of clothing. cortex (ACC) is involved in brand preference judgments (Deppe, Each subject had to make a total of 120 decisions (60 TD and 60 Schwindt, Kugel, Plassmann, Kenning 2005). Deppe et al. (2005) DD). described the network as an affective and self-referencing path We employed a one-factorial event related design (Bandettini during brand choice. Against this background our hypothesis was & Cox, 2000; Buckner, 2003; Dale, 1999; Friston, Zarahn, Josephs, that the underlying neural activation patterns of reward attribution Henson, & Dale, 1999) with an inter-stimulus-interval (ISI) of 5 to stimuli customers feel emotionally attached to would be similar seconds each and a randomized stimulus presentation. In order to to Deppe et al.’s findings. avoid response-related activations and movement artifacts we did Thus, we hypothesize: not request for any feedback during scanning (Deppe et al. 2005). The subjects were instructed to choose between two stores, H: During brand choice, loyal customers as compared to according to where they would prefer to buy the piece of clothing disloyal customer will show a greater activity in the shown on the screen. Thus, the store brand itself functions as the striatum, VMPC and ACC. only selection criterion. After the fMRI scan, participants were asked to provide general personal information. Then, participants In the following, we describe our empirical study to test this were compensated EUR 25 per hour, thanked and debriefed. hypothesis. Data Acquisition EMPIRICAL STUDY All fMRI data were acquired from a 3.0 Tesla whole body The empirical study was designed to investigate differences in scanner (Intera T30, Philips, Best, NL). Prior to the functional data brain activity of loyal customers as compared to disloyal customers acquisition structural T1 weighted datasets were acquired for of a department store. anatomical identification. The functional T2* images were acquired using single-shot gradient echo-planar imaging (EPI) se- Participants and Study Design quence that covered nearly the whole brain. The data set consisted A total of 300 customers of a local department store were of 36 transversal slices of 3.6 mm thickness without gap. Images chosen randomly from a department store’s club card database of were acquired every three seconds (TR). which 22 (9 females, 13 males, median age 46) were recruited, after In combination with the fMRI measurement, data about the several screening procedures, to participate in an fMRI study. subjects’ monthly spending at the target store and the number of Exclusion criteria were relating to fMRI safety (e.g., metal or shopping days per month in the fashion segment, were extracted implanted devices in the body, claustrophobia), factors known to from the department’s store club card database. affect brain organization, function, or blood flow (such as handed- ness, pharmaceutical / drug use, psychiatric, neuropsychological Data Analysis and medical disorders). Furthermore, participants were selected to All acquired fMRI data were preprocessed using Statistical proceed to the fMRI task only if they were either classified for the Parametric Mapping (SPM2; Wellcome Department of Cognitive last six month as “A” customer (monthly spending of 250 EUR or Neurology, London, UK). For the statistical analysis, we specified more for clothing and 5 or more monthly shopping days) or “C” regressors of a General Linear Model (GLM) according to our customers (monthly spending of 50 EUR or less for clothing and 1 experimental design at the single-subject level. We then calculated or less monthly shopping days). contrast images for the events when T was present (TD) compared All participants were healthy, right-handed and, gave their to when the T was not present (DD). These single-subject contrast written consent to participate in the study in accordance with the images were used for a random effects analysis at the group level. Advances in Consumer Research (Volume 34) / 737 FIGURE 1 Results from random effects group analysis

The sample was divided into two groups (A- and C-customers) in the present study, both the latter and areas involved in semanti- based on a customer value index (monthly spending and number of cally processing such as reading and speaking were found to be monthly shopping days). A two-sample t-test was calculated on the activated increasingly. group level to compare neural activations patterns of the two groups when T was present. DISCUSSION In this paper, we attempt to expand our understanding of the Results basis of emotional bonds customers built with companies and The most pronounced increased activation in the presence of brands. We conducted an fMRI study to assess differences in neural T in the group of A-customers compared to C-customers was found activations between loyal and disloyal customers of a department in the striatum (see Figure 1A). This area was revealed to be store. Based on prior findings in the literature, we assumed that for involved in the representation of reward prediction (Gottfried, loyal customers the store brand would modulate the underlying O’Doherty, and Dolan 2003; Knutson and Cooper 2005; O’Doherty information processing via an emotion-based decision path, as et al. 2004; Schultz, Dayan, and Montague, 1997). Zink, Pagnoni, compared to disloyal of the respective store. Martin, Dhamala, & Berns (2003) found increased activity in the As a result, we were able to replicate Deppe et al.’s neural particular part of the striatum that we are reporting here, in response correlate of brand preference for the particular case of store brands to the prediction of behaviorally relevant rewards, as compared to (Deppe et al., 2005). In addition to Deppe et al.’s findings (Deppe “only” unexpected rewards which are represented in other parts of et al., 2005), we linked the neural activation patterns to the actual the striatum. buying behavior of the subjects. We showed that, for customers Furthermore, other areas involved in the TD decision of A- with a high monetary customer value (A-customers) the presence of customers as compared to C-customers, confirm to Deppe et al.’s the specific store brand acts as a rewarding signal during the choice First-Choice-Brand activation pattern (Deppe et al. 2005, see task, whereas customers with a low monetary customer value (C- Figure 1 B). customers) do not reveal such a rewarding activation pattern. The brain areas shown in Figure 1 are the activation network linking VMPFC, striatum, and ACC. Deppe et al. (2005) described Implications for relationship management the network as an emotion-based and self-referencing information Our findings offer some important insights for management. processing mode during brand choice. This network was found to It can be noted that for a company, it is not sufficient to concentrate be involved in decisions when the subject’s first choice brand was on ‘bounding’ customers for instance with technical requirements present. Conversely, where only second or lower-ranked brands (e.g. hard- and software which is only compatible with specific were available, the underlying neural network represents an ana- operating systems) or general price promotions, but encourage lytical weighing of alternatives, mathematical calculation, strategic customers in creating affective bonds to the company or its brands. thinking and object recognition (Deppe et al., 2005). In particular This could be achieved by for example personalizing the commu- 738 / Why Companies Should Make Their Customers Happy: The Neural Correlates of Customer Loyalty nication with the customer and provide the basis for emotional Bendapudi, Neeli and Leonard Berry (1997), “Customers’ events the customers links to the company or brand. Motivations for Maintaining Relationships with Service Providers,” Journal of Retailing, 73 (1), 15-37. Limitations and future research Blood, Anne J., Robert J. Zatorre, Patrick Bermudez, and Alan The interpretation of our empirical findings are based on C. Evans (1999), “Emotional responses to pleasant and binary decision making tasks conducted in a controlled laboratory unpleasant music correlate with activity in paralimbic brain environment and thus, do not reflect the richness of brand choices regions,” Nature Neuroscience, 2 (1), 382–387. in a real-world shopping situation. Hence, further refinement of the Canli, Turhan, Heidi Sivers, Susan L. Whitfield, Ian H. Gotlib, experimental design, such as integrating more mobile brain imag- and John D. Gabrieli (2002), “Amygdala response to happy ing methods, will be needed in the future. faces as a function of extraversion, “ Science, 296, 2191. Furthermore, we focused only on the link between one depart- Chaudhuri, Arjun and Morris B. Holbrook (2001), “The Chain of ment store and their customers’ loyalty. Future studies should Effects from Brand Trust and Brand Affect to Brand broaden our findings by investigating multi-brand-loyalty link- Performance: The Role of Brand Loyalty,” Journal of ages. Marketing, 65 (2), 81-93. In addition, we linked the neural activation patterns to past Dale, A. M. (1999). Optimal Experimental Design for Event- buying behavior. In following studies it might be useful to investi- Related fMRI, Human Brain Mapping, 8, 109-114. gate whether the rewarding store brand effects can also predict Deppe, Michael, Schwindt, Wolfram, Kugel, Harald, Plassmann, future purchasing behavior. This could be implemented either by a Hilke and Peter Kenning (2005), “Nonlinear responses panel data analysis over a longer period of time before and after the within the medial prefrontal cortex reveal when specific fMRI scanning or by including customers’ purchasing intentions implicit information influences economic decision making”, (i.e. attitudinal loyalty). Journal of Neuroimaging, Vol. 15(2), 171–82. In conclusion, we could deepen the understanding of the Desmond, John E. and Gary H. Glover (2002), “Estimating psychological states underlying customer loyalty by providing Sample Size in Functional MRI (fMRI) Neuroimaging initial neurobiological evidences of “the tie that binds” customers Studies: Statistical Power Analyses,” Journal of Neuro- to companies and brands. science Methods, 118 (August), 115-28. Dick, Alan S. and Kunal Basu (1994), “Customer Loyalty: REFERENCES Toward an Integrated Conceptual Framework,” Journal of Allen, Natalie J. and John P. Meyer (1990), “The Measurement the Academy of Marketing Science, 22 (Winter), 99-113. and Antecedents of Affective, Continuance and Normative Elliott, Rebecca, Jana L. Newman, Olivia A. Longe and William Commitment to the Organization,” Journal of Occupational J.F. Deakin (2003) ”Differential response patterns in the Psychological Society, 63 (1), 1-18. striatum and orbitofrontal cortex to financial reward in Aharon, Itzhak, Nancy Etcoff, Dan Ariely, Christopher F. humans: a parametric functional magnetic resonance imaging Chabris, Ethan O’Connor, and Hans C. Breiter (2001), study,” Journal of Neuroscience 23: 303–307. “Beautiful faces have variable reward value: fMRI and Erk, Susanne, Manfred Spitzer, Arthur Wunderlich, Lars Galley, behavioral evidence,” Neuron, 32 (1), 537–551. and Henrik Walter (2002), “Cultural objects modulate reward Ambler, Tim, Sven Braeutigam, John F. Stins, Steven P. Rose circuitry, “ Neuroreport, 13 (18), 2499-2503. and Stephen Swithenby (2004), “Salience and Choice: Evanschitzky, Heiner and Hilke Plassmann (2005), “ Affective Neural Correlates of Shopping Decisions,” Psychology & and Calculative Commitment as Antecedents of Customer Marketing, 21 (4), 247-66. Loyalty,” Winter AMA 2006 Proceedings, 17 (1), 63-64. Bandettini, P. A., & Cox, R. W. (2000). Event-Related fMRI Fitzsimons, Gavan J., J. Wesley Hutchinson, Patti Williams, Contrast Using Constant Interstimulus Interval: Theory and Joseph W. Alba, Tanya L. Chartrand, Joel Huber, Frank R. Experiment, Magnetic Resonance in Medicine, 43, 540-548. Kardes, Geeta Menon, Priya Raghubir, J. Edward Russo, Bansal, Harvir S., P. Gregory Irving, and Shirley F. Taylor Baba Shiv and Nader T. Tavassoli (2002), “Non-Conscious (2004), “A Three-Component Model of Customer Commit- Influences on Consumer Choice,” Marketing Letters, 13 (3), ment to Service Providers,“ Journal of the Academy of 269-80. Marketing Science, 32 (3), 109-250. Fournier, Susan (1998), “Consumers and their Brands: Develop- Bargh, John A. (2002), “Losing Consciousness: Automatic ing Relationship Theory in Consumer Research,” Journal of Influences on Consumer Judgment, Behavior, and Motiva- Consumer Research, 24 (4), 343-74. tion,” Journal of Consumer Research, 29 (2), 280-85. Fullerton, Gordon (2003), “When Does Commitment Lead to Bechara, A. (2004), “The Role of Emotion in Decision-Making: Loyalty?” Journal of Service Research, 5(4), 333-344. Evidence from Neurological Patients with Orbitofrontal Gottfried, Jay A., Ralf Deichmann, Joel S. Winston, and Damage,” Brain and Cognition, 55 (1), 30-40. Raymond J. Dolan (2002), “Functional heterogeneity in Bechara, Antoine, Hanna Damasio and Antonio R. Damasio human olfactory cortex: an event-related functional magnetic (2000), “Emotion, Decision Making and the Orbitofrontal resonance imaging study,” Journal of Neuroscience, 22 (1), Cortex,” Cerebral Cortex, 10 (3), 295-307. 10819–10828. Bechara, Antoine, Hanna Damasio, Daniel Tranel and Antonio Gottfried, Jay A., John O’Doherty, and Raymond Dolan (2003), R. Damasio (1997), “Deciding Advantageously before “Encoding Predictive Reward Value in Human Amygdala Knowing the Advantageous Strategy,” Science, 275 (5304), and Orbitofrontal Cortex”, Science, 301(5636), 1104-1107. 1293-95. Gwinner, Kevin, Dwayne Gremler and Mary Jo Bitner (1998), Bechara, Antoine, and Antonio R. Damasio (2005), “The “Relational Benefits in Service Industries: The Customer’s Somatic Marker Hypothesis: A Neural Theory of Economic Perspective,” Journal of the Academy of Marketing Science, Decision,” Games and Economic Behavior, 52 (2), 336-372. 26 (2), 101-114. Advances in Consumer Research (Volume 34) / 739 Jacoby, Jacob and Robert W. Chestnut (1978), Brand Loyalty: Slovic, Paul, Melissa L. Finucane, Ellen Peters and Donald G. Measurement and Management, New York, NY: Wiley. MacGregor (2002), “The Affect Heuristic,” in Heuristics and Jacoby, Jacob and David B. Kyner (1973), “Brand Loyalty vs. Biases: The Psychology of Intuitive Judgments, ed. T. Repeat Purchasing Behavior,” Journal of Marketing Gilovich, D. Griffin and D. Kahneman, New York: 397-421. Research, 10 (February), 1-9. Slovic, Paul, Melissa L. Finucane, Ellen Peters and Donald G. Knutson, B., & Cooper, J. C. (2005), “Functional Magnetic MacGregor (2004), “Risk as Analysis and Risk as Feelings: Resonance Imaging of Reward Prediction,” Current Opinion Some Thoughts about Affect, Reason, Risk, and Rationality,” in Neurology, 18 (4), 411-417. Risk Analysis, 24 (2), 311-22. Kringelbach, Morton, John O’Doherty, Edmund T. Rolls, and Thomson, Matthew, Deborah J. MacInnis and C. Whan Park Christian Andrews (2003), “Activation of the Human (2005), “The Ties That Bind: Measuring the Strength of Orbitofrontal Cortex to a Liquid Food Stimulus is Correlated Consumers’ Emotional Attachments to Brands,” Journal of with its Subjective Pleasantness, Cerebral Cortex, 13 (10), Consumer Psychology, 15 (1), 77-92. 1064-1071. Westbrook, Robert A. and Richard L. Oliver (1991), “The McClure, Samuel, Li, Jian, Tomlin, Damon, Cypert, Kim S., Dimensions of Consumption Emotion Patterns and Con- Montague, Latané M. and P. Read Montague (2004), “Neural sumer Satisfaction,” Journal of Consumer Research, 18 (1), correlates of behavioral preference for culturally familiar 84-92. drinks”, Neuron, Vol. 44, 379-87. Woodside, Arch G. (2004), “Advancing from Subjective to Morgan, Robert M., and Shelby D. Hunt (1994), “The Commit- Confirmatory Personal Introspection in Consumer Research,” ment-Trust Theory of Relationship Management,” Journal of Psychology & Marketing, 21 (12), 987-1010. Marketing, 58 (July), 20-38. Yeung, Catherine W. M. and Robert S. Wyer (2004), “Affect, Morris, Stephen, Chris D. Frith, David I. Perrett, Daniel Appraisal, and Consumer Judgment,” Journal of Consumer Rowland, Andrew W. Young, Andrew J. Calder, and Research, 31 (2), 412-25. Raymond J. Dolan (1996), “A differential neural response in Yeung, Catherine W. M. amd Robert S. Wyer (2005), “Does the human amygdala to fearful and happy facial expres- Loving a Brand Mean Loving its Products? The Role of sions,” Nature, 383, 812–815. Brand-Elicited Affect in Brand Extension Evaluations”, Murphy, Kevin and Hugh Garavan (2004), “An Empirical Journal of Marketing Research, Vol. XLII (November 2005), Investigation into the Number of Subjects Required for an 495-506. Event-Related fMRI Study,” NeuroImage, 22 (June), 879-85. Yoon, Carolyn, Angela H. Gutchess, Fred Feinberg, and Thad A. O’Doherty, John (2004) “ Reward representations and reward Polk (2006), “Comparing Brand and Human Personality via related learning in the human brain: Insights from Functional Magnetic Resonance Imaging,” Journal of neuroimaging,” Current opinions in neurobiology, 14 (6): Consumer Research, 33 (June), 31-40. 769-776, Zaltman, G. (2003). How Customers Think. Boston: Harvard O’Doherty, John, Morton L. Kringelbach, Edmund T. Rolls, Business School Press. John Hornak and Christian Andrews (2001), “Abstract Zink, Carolin F., Guiseppe Pagnoni, Megan E. Martin, Reward and Punishment Representations in the Human Mukeswhar Dhamala, and Gregory S. Berns (2003), “Human Orbitofrontal Cortex,” Nature Neuroscience, 4 (1), 95-102. Striatal Response to Salient nonrewarding Stimuli”, Journal Oliver, Richard L. (1999), “Whence Consumer Loyalty?” of Neuroscience, 23(22), 8092-8097. Journal of Marketing, 63 (Special Issue), 33-44. Paulus, Martin (2005), “Neurobiology of decision-making: Quo vadis?” Cognitive Brain Research, 23(1), 2-10. Reichheld, Frederick F., W. Earl Sasser Jr. (1990), “Zero Defections: Quality Comes To Services,” Harvard Business Review, 68 (5), 105. Rolls, Edmund T., John O’Doherty, Morton L. Kringelbach, Susan Francis, Robert Bowtell, and F. McGlone (2003), “Representations of pleasant and painful touch in the human orbitofrontal and cingulate cortices,” Cerebral Cortex, 13 (1), 308–317. Rolls, Edmund, T., Sienkiewiczs, Zonon J., Yaxley Simon (1989), “ Hunger modulates the response to gustatory stimuli of single neurons in the caudolatoral orbitofrontal cortex of the macaque monkey, “ European Journal of Neuroscience, 1, 53-60. Schultz, Wolfram (2006), “Behavioral theories and the neuro- physiology of reward,” Annual Reviews in Psychology, 57 (1), 87-115. Schultz, W., Dayan, P., & Montague, R. P. (1997), “A Neural Substrate of Prediction and Reward,” Science, 275, 1593- 1599. Schwarz, Norbert and Gerald L. Clore (1988), “How Do I Feel about It? The Informative Function of Affective States,” in Affect, Cognition, and Social Behavior, ed. Klaus Fiedler and Joseph P. Forgas, Toronto: Hogrefe, 44-62. Optimizing Variety in Mass Customization: A Theoretical Perspective Sanjay Puligadda, Pennsylvania State University, USA Rajdeep Grewal, Pennsylvania State University, USA Arvind Rangaswamy, Pennsylvania State University, USA Frank R. Kardes, University of Cincinnati, USA

EXTENDED ABSTRACT knowledge of the product category would influence how they Markets today are fragmenting with individual needs and search and utilize attribute information, consumer knowledge would wants increasingly prone to shifts and changes (Hart 1995) neces- affect the processing of information related to variable attributes. In sitating products that are mass customized and customerized particular, we examine the moderating effects of two constructs (Simonson 2005; Wind and Rangaswamy 2001). Although consid- related to consumer knowledge, i.e., (1) objective knowledge, erable literature has examined mass customization from a produc- (Alba and Hutchinson 1987), and (2) knowledge function of atti- tion perspective (e.g., Papathanassiou 2004, Jiao, Ma, and Tseng tudes, (Katz 1960, Grewal, Mehta, and Kardes 2004), for the 2003), it is only recently that marketing scholars have begun to influence of the levels of universal and variable attributes on examine this phenomenon (e.g., Murthi and Sarkar 2003; Simonson consumer satisfaction with mass customization platforms. Objec- 2005; Wind and Rangaswamy 2001). tive knowledge refers to how knowledge with respect to the object Typically, in mass customization, customers are provided (e.g. cell phones) may influence attribute processing. The knowl- varying options of a product’s attributes to select and thus “build” edge function of attitudes pertains to the most basic function of their product. A major managerial challenge in such cases is to attitudes that enables individuals to make better sense of their world determine the optimal variety of options to offer (Simonson 2005). (Shavitt 1989). While the former refers to the use of knowledge with Too many options would be expensive, and confusing to customers, respect to the object (e.g. cell phone), the latter refers to the use of while too few options may annoy the customers who may be unable the object (cell phone) as a means of knowledge. Thus these two to configure the product based on their true preferences (e.g., knowledge related constructs represent two sides of a coin concern- Huffman and Kahn 1998). ing consumer knowledge. We suggest that, since a variable attribute We build on the work of Sherman et al. (1984) on human is more personal and more likely to serve as a peripheral cue than qualities to classify product attributes as universally or variably a universal attribute (Huffman and Houston 1993; Park and Lessig evaluated. According to Sherman et al (1984), for universally- 1981), variable attributes would be more important for novices than evaluated qualities, there is consensus in a population about whether for experts such that the positive effect of increasing variable a particular level is good or bad, regardless of individuals’ prefer- attribute options on satisfaction with the customization platform ence for the level. For example, people would agree that being would be stronger for novices than for experts. Further, we posit honest is good and being dishonest is bad whether they themselves that if a product serves the knowledge function of attitudes (Katz are honest or not. In contrast, there is more variability and person- 1960; Locander and Spivey 1978), consumers are likely to have a specificity in variably-evaluated qualities. For example, opponents stronger personal relationship with the product, thereby increasing of abortion may think of a pro-choice political candidate as unde- the importance of variable attributes such that the positive effect of sirable (“low quality”) whereas pro-choice individuals may find increasing variable attribute options on satisfaction with the that candidate very appealing. Thus, for universally-evaluated customization platform would be stronger when the product serves qualities there exists a widely accepted evaluation basis, whereas the product knowledge than when it does not. for variably-evaluated qualities, the evaluation schemes are more We test these propositions with two experiments set in the individual-specific. context of customizing cell phones. In the first experiment, we Similarly, in a consumption context, universally-evaluated examine the moderating role of objective knowledge and find that product attributes (universal attributes), such as battery life in cell variable attributes are more important for novices as opposed to phones, would have a common and widely accepted evaluation, experts . In the second experiment, we study the moderating role of whereas the evaluation schemes for variable attributes, such as the knowledge function of attitudes towards cell phones and find exterior color of cell phones, would be more consumer-specific. In that variable attributes are likely to generate higher satisfaction with laptops, consumers would generally agree that 40 Giga Bytes (GB) the mass customization platform if cell phones perform the knowl- of hard disk space (a universal attribute) is better than 20GB of hard edge function of attitudes than if the cell phones do not perform the disk space. In contrast, the color kit options in laptops (a variable knowledge function of attitudes. The results of the two experiments attribute) are likely to be evaluated based on personal preferences– also suggest, as hypothesized, that objective knowledge and knowl- ‘Charcoal Leather’ maybe evaluated positively by one while an- edge function of attitudes do not influence the evaluation of other consumer may prefer ‘Mediterranean Blue.’ universal attributes. We suggest that this difference in evaluation schemes would There are several important theoretical implications of this result in consumers using different strategies for evaluating the two research. We distinguish between universal and variable attributes types of attributes.1 Specifically, we suggest that, since consumers’ and show their differential effects on consumer evaluation of a mass customization platform. Future research might study the criticality of this categorization in other consumption contexts, such as 1Customization platform represents an interface where the product evaluation of competing offerings of durable goods. It would be options are presented as attributes and options of those attributes also worthwhile to examine how this attribute classification influ- are also listed. For example, a hypothetical customization plat- ences satisfaction with the product or service. Additionally, we form can have two attributes–color and quality, with three options have identified important moderators (objective knowledge and for color–red, green, and blue, and two options for quality–high knowledge function of attitudes) in consumers’ assessment of and medium. universal and variable attributes. From a managerial standpoint, our 740 Advances in Consumer Research Volume 34, © 2007 Advances in Consumer Research (Volume 34) / 741 results should help marketers in deciding the number of attribute Grewal, Rajdeep, James M. Comer, and Raj Mehta (2001), “An options that would provide the optimal variety in customization Investigation into the Antecedents of Organizational platforms. Also, the moderators we have identified, (namely, objec- Participation in Business-to-Business Electronic Markets,” tive knowledge and knowledge function of attitudes), offer insights Journal of Marketing, 65 (July), 17-33. for segmenting the customers for design of mass customization Grewal, Rajdeep, Raj Mehta, and Frank R. Kardes (2004), “The platforms. Timing of Repeat Purchases of Consumer Durable Goods: The Role of Functional Bases of Consumer Attitudes,” REFERENCES Journal of Marketing Research, 41 (1), 101-15. Ajzen, I. and Martin Fishbein (1969), “The Prediction of Hart, Christopher W. L. (1995), “Mass Customization: Concep- Behavioral Intentions in a Choice Situation,” Journal of tual Underpinnings, Opportunities and Limits,” International Experimental Social Psychology, 5, 400-16. Journal of Service Industry Management, 6 (2), 36-45. Alba, Joseph W. and J. Wesley Hutchinson (1987), “Dimensions Herek, G.M. (1987), “Can Functions be Measured ? A New of Consumer Expertise,” Journal of Consumer Research, 13 Perspective on the Functional Approach to Attitudes,” Social (4), 411-54. Psychology Quarterly, 50, 285-303. (2000), “Knowledge Calibration: What Consumers Hsee, Christopher K. (1998), “Less Is Better: When Low-value Know and What They Think They Know,” Journal of Options are Valued More Highly Than High-value Options,” Consumer Research, 27 (2), 123-56. Journal of Behavioral Decision Making, 11 (2), 107-21. Arnold, Stephen J., Tae H. Oum, and Douglas J. Tigert (1983), Hsee, Christopher K. and Yuval Rottenstreich (2004), “Music, “Determinant Attributes in Retail Patronage: Seasonal, Pandas, and Muggers: On the Affective Psychology of Temporal, Regional, and International Comparisons,” Value,” Journal of Experimental Psychology: General, 133 Journal of Marketing Research, 20 (2), 149-57. (1), 23-30. Bettman, James R. and C. Whan Park (1980), “Effects of Prior Hsee, Christopher, Yuval Rottenstreich, and Zhixing Xiao Knowledge and Experience and Phase of the Choice Process (2005), “When Is More Better? On the Relationship Between on Consumer Decision Processes: A Protocol Analysis,” Magnitude and Subjective Value,” Current Directions in Journal of Consumer Research, 7 (3), 234-48. Psychological Sciences, 14 (5), 234-37. Brown, Christina L. and Gregory S. Carpenter (2000), “Why is Huffman, Cynthia and Michael J. Houston (1993), “Goal- The Trivial Important? A Reasons-based Account for The oriented experiences and the development of knowledge,” Effects of Trivial Attributes on Choice,” Journal of Con- Journal of Consumer Research, 20 (2), 190-207. sumer Research, 26 (4), 372-85. Huffman, Cynthia and Barbara E. Kahn (1998), “Variety for Brucks, Merrie (1985), “The Effects of Product Class Knowl- Sale: Mass Customization or Mass Confusion?,” Journal of edge on Information Search Behavior,” Journal of Consumer Retailing, 74 (4), 491-513. Research, 12 (1), 1-16. Hutchinson, J. Wesley and and Joseph W. Alba (1991), Carpenter, Gregory S., Rashi Glazer, and Kent Nakamoto “Ignoring Irrelevant Information: Situational Determinants of (1997), Readings on Market-Driving Strategies: Towards a Consumer Learning,” Journal of Consumer Research, 18 (3), New Theory of Competitive Advantage. New York, NY: 325-45. Addison Wesley Longman. Inman, J. Jeffrey (2001), “The Role of Sensory-specific Satiety Celsi, Richard L. and Jerry C. Olson (1988), “The Role of in Attribute-level Variety Seeking,” Journal of Consumer Involvement in Attention and Comprehension Proce,” Research, 28 (1), 105-20. Journal of Consumer Research, 15 (2), 210-24. Iyengar, Sheena S. and Mark R. Lepper (2000), “When choice is Chandrashekaran, Murali, Kristin Rotte, and Rajdeep Grewal demotivating: Can one desire too much of a good thing?,” (2005), “Knowledge in Error: Decoding Consumer Judg- Journal of Personality and Social Psychology, 79 (6), 995. ments with the JUMP Model,” in Applying Social Cognition Jacoby, Jacob and Naresh K. Malhotra (1984), “Perspectives on to Consumer-Focused Strategy, Frank R. Kardes, Paul M. Information Overload/Reflections on the Information Herr ,and Jacques Nantel, Eds. Mahwah, NJ: Lawrence Overload Paradigm in Consumer Decision Making,” Journal Erlbaum Associates. of Consumer Research, 10 (4), 432-40. Chernev, Alex (1997), “The Effect of Common Features on Jiao, Jianxin, Qinhai Ma, and Mitchell M. Tseng (2003), Brand Choice: Moderating Role of Attribute Importance,” “Towards High Value-added Products and Services: Mass Journal of Consumer Research, 23 (4), 304-11. Customization and Beyond,” Technovation, 23 (10), 809-21. Coffey, Peter (1958), Epistemology, or the Theory of Knowledge Johnson, Eric J. and J. Edward Russo (1984), “Product Familiar- : An Introduction to General Metaphysics. Gloucester: ity and Learning New Information,” Journal of Consumer P.Smith. Research (pre-1986), 11 (1), 542-50. Coulter, Robin A., Linda L. Price, and Lawrence Feick (2003), Kahn, Barbara E. (1998), “Dynamic Relationships With “Rethinking the origins of involvement and brand commit- Customers: High-variety Strategies,” Academy of Marketing ment: Insights from postsocialist Central Europe,” Journal of Science. Journal, 26 (1), 45-53. Consumer Research, 30 (2), 151-69. Katz, Daniel (1960), “The Functional Approach to the Study of Dickson, Peter R. (1982), “Person-Situation: Segmentation’s Attitudes,” Public Opinion Quarterly, 24 (2), 163-204. Missing Link,” Journal of Marketing, 46 (4), 56-64. Katz, Daniel and E. Stotland (1959), A Preliminary Statement to Duray, Rebecca and Glenn W Milligan (1999), “Improving a Theory of Attitude Structure and Change. New York: Customer Satisfaction Through Mass Customization,” McGraw Hill. Quality Progress, 32 (8), 60-66. Kelman, H.C. (1961), “Processes of Opinion Change,” Public Fishbein, Martin (1967), Readings in Attitude Theory and Opinion Quarterly, 25, 57-78. Measurement. New York: Wiley. Kivetz, Ran and Itamar Simonson (2000), “The Effects of Incomplete Information on Consumer Choice,” Journal of Marketing Research, 37 (4), 427-248. 742 / Optimizing Variety in Mass Customization: A Theoretical Perspective Lampel, Joseph and Henry Mintzberg (1996), “Customizing Rosenthal, Robert and Ralph L. Rosnow (1985), Contrast Customization,” Sloan Management Review, 38 (1), 21-30. Analysis. Cambridge: Cambridge University Press. Lau, Ronald S. M. (1995), “Mass Customization: The Next Sarnoff, I. and Daniel Katz (1954), “The Motivational Bases of Industrial Revolution,” Industrial Management, 37 (5), 18- Attitude Change,” Journal of Abnormal and Social Psychol- 19. ogy, 49, 115-24. Lee, Byung-Kwan and Wei-Na Lee (2004), “The Effect of Schwartz, Barry , Andrew Ward, John Monterosso, Sonja Information Overload on Consumer Choice Quality in an Lyubomirsky, Katherine White, and Darrin Lehman (2002), On-Line Environment,” Psychology & Marketing, 21 (3), “Maximizing Versus Satisficing: Happiness Is a Matter of 159-83. Choice,” Journal of Personality and Social Psychology, 83 Locander, William B. and W. Austin Spivey (1978), “A (5), 1178-97. Functional Approach to Attitude Measurement,” Journal of Selladurai, R. S. (2004), “Mass Customization in Operations Marketing Research, 15 (4), 576-87. Management: Oxymoron or Reality?,” Omega, 32 (4), 295- MacDuffie, John Paul, Kannan Sethuraman, and Marshall L. 300. Fisher (1996), “Product Variety and Manufacturing Perfor- Shavitt, Sharon (1989), Operationalizing Functional Theories of mance: Evidence From the International Automotive Attitude. Hillsdale, NJ: Lawrence Erlbaum Associates. Assembly Plant Study,” Management Science, 42 (3), 350- Sherman, Steven J., Laurie Chassin, Clark C. Presson, and Gina 69. Agostinelli (1984), “The Role of the Evaluation and MacInnis, Deborah J., Christine Moorman, and Bernard J. Similarity Principles in the False Consensus Effect,” Journal Jaworski (1991), “Enhancing and Measuring Consumers’ of Personality and Social Psychology, 47 (6), 1244-62. Motivation, Opportunity,” Journal of Marketing, 55 (4), 32- Simonson, Itamar (2005), “Determinants of Customers’ 53. Responses to Customized Offers: Conceptual Framework McGill, Ann L. and Punam Anand (1989), “The Effect Of Vivid and Research Propositions,” Journal of Marketing, 69 (1), Attributes On The Evaluation Of Alternatives,” Journal of 32-45. Consumer Research, 16 (2), 188-96. Smith, M.B., J.S. Bruner, and R.W. White (1956), Opinions and Medin, Douglas L. and Edward E. Smith (1984), “Concepts and Personality. New York: Wiley. Concept Formation,” in Annual Review of Psychology, Mark Sujan, Mita (1985), “Consumer Knowledge: Effects on Evalua- R. Rozenzweig and Lyman W. Porter, Eds. Vol. 35. Palo tion Strategies Mediating Consumer Judgments,” Journal of Alto, CA: Annual Reviews, Inc. Consumer Research (pre-1986), 12 (1), 31-46. Merton, Robert K. (1957), Social Theory and Social Structure. Tu, Qiang, Mark A. Vondermbse, and T.S. Ragu-Nathan (2001), Glencoe, IL: The Free Press. “The Impact of Time-based Manufacturing Practices on Murthi, B.P.S. and Sumit Sarkar (2003), “The Role of Manage- Mass Customization and Value to Customer,” Journal of ment Sciences in Research on Personalization,” Management Operations Management, 19 (2), 201-17. Science, 49 (10), 1344-62. Wind, Jerry and Arvind Rangaswamy (2001), “Customerization: Olkinuora, Erkki (1974), Norm Socialization: The Formation of The Next Revolution in Mass Customization,” Journal of Personal Norms: University of Jyvaskyla. Interactive Marketing, 15 (1), 13-32. Papathanassiou, Eleutherios A. (2004), “Mass Customisation: Management Approaches and Internet Opportunities in the Financial Sector in the UK,” International Journal of Information Management, 24 (5), 387-89. Park, C. Whan and V. Parker Lessig (1981), “Familarity and Its Impact on Consumer Biases and Heuristics,” Journal of Consumer Research, 8 (September), 223-30. Petty, Richard E., John T. Cacioppo, and David Schumann (1983), “Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement,” Journal of Consumer Research, 10 (2), 135-46. Pine, B. Joseph and James H. Gilmore (2000), “Satisfaction, Sacrifice, Surprise: Three Small Steps Create One Giant Leap Into the Experience Economy,” Strategy & Leadership, 28 (1), 18-23. Pine, B. Joseph II (1993), “Mass customizing products and services,” Planning Review, 21 (4), 6-14. Pratkanis, Anthony R., Steven J. Breckler, and Anthony G. Greenwald (1989), Attitude Structure and Function. Hillsdale, NJ: Lawrence Erlbaum. Rao, Akshay R. and Kent B. Monroe (1988), “The Moderating Effect of Prior Knowledge on Cue Utilization,” Journal of Consumer Research, 15 (2), 253-64. Rao, Srikumar and John U. Farley (1987), “Effects of Environ- mental Perceptions and Cognitive Complexity on Search and Information Processing,” Psychology & Marketing, 4 (4), 287-302. Temporal Sequence Effects: A Memory Framework Nicole Votolato Montgomery, Ohio State University, USA H. Rao Unnava, Ohio State University, USA

EXTENDED ABSTRACT immediately, replicating the findings from Redelmeier and While much attention has been given recently to studying Kahneman (1996). However, the initial events are heavily weighted temporal sequences of events, few have offered explanations to when the experience is evaluated following a delay, resulting in account for how people integrate these events into a holistic higher evaluations for the improving trend immediately but a evaluation of the experience, and virtually no attention has been preference for the declining trend following a delay, a finding given to the underlying mechanism involved. A temporal sequence which is predicted by proposed framework but contrary to previous refers to an experience that consists of a series of outcomes spaced findings in the temporal sequence literature (e.g., Ariely 1998; over time. The findings from the temporal sequence literature Redelmeier and Kahneman 1996). indicate that people form global retrospective ratings of an experi- Studies 3 and 4 further confirm the role of memory by ence using salient characteristics of the experience, such as the peak manipulating the uniqueness of one event in a vacation experience intensity, the end intensity, and the trend (Ariely, Kahneman, and to test whether a memory-based structure can account for the impact Loewenstein 2000). The current explanations that have been pro- of the peak of an experience on retrospective evaluations of the posed to account for temporal sequence findings (e.g., Ariely and experience (e.g., Kahneman et al. 1993). Uniqueness is examined Carmon 2000; Baumgartner, Sujan, and Padgett 1997) tend to be ad because past research using diary studies suggests that intense hoc in nature and are limited to explaining reported findings rather emotional experiences are unique (Bower 1981), leading to the than offering a comprehensive account of the findings in the expectation that peak intensity incidents are more accessible than literature. We propose that adopting a memory-based framework other incidents because they are uncommon, consistent with the can provide a parsimonious explanation for the findings on retro- von Restorff effect in memory (Wallace 1965). We argue that if a spective evaluations of temporal sequence. peak intensity event occurs in a temporal location of an experience Retrospective evaluations inherently require that people recall that impedes memory, such as the center or end of an experience that aspects of the experience. Past research suggests that information is evaluated following a delay (Ashcraft 2002), then making that that is easily accessible in memory will impact judgments the most event unique will facilitate recall of the event and result in that event (e.g., Hamilton and Gifford 1976; Tversky and Kahneman 1973). being heavily weighted in retrospective evaluations of the experi- Accessibility of information has been studied in the memory ence. Consistent with expectations, study 3 shows that making the literature using a free recall task in which people are presented with peak intensity final event in an improving vacation experience a list of words that they may recall in any order (Ashcraft 2002). atypical enhances the recall of this event following a delay relative Findings from this research show that respondents have higher to a common event with the same affective intensity, demonstrating recall performance on early items on a list (primacy) because these a recovery of the recency effect previously shown to diminish after items receive sufficient rehearsals to transfer them to long-term a delay. Additionally, study 3 shows that evaluations are based on memory and final items on a list (recency) because these items are the peak intensity final event when the event is unique, consistent temporarily available in short-term memory. Though a temporal with Kahneman et al. (1993), while evaluations are based on the sequence is affective in nature and a free recall task is cognitive in initial events when the peak intensity final event is common, nature, we suggest that there are parallels between an affective contrary to previous findings (e.g., Fredrickson and Kahneman experience and the experience of learning a list. Thus, memory 1993). Study 4 further confirms these findings by replicating the findings on free recall are expected to explain current findings in the results with the peak incident in the center of an experience and by temporal sequence literature and predict findings that would other- demonstrating the robustness of the effects for both positively and wise not be predicted by the temporal sequence literature. negatively-valenced peak incidents. In studies 1 and 2, we test whether a memory-based frame- In sum, the findings from this research demonstrate that a work, utilizing primacy and recency effects, can account for the conventional memory-based framework appears to provide a co- impact of the beginning, end, and trend of an experience on herent structure to account for consumers’ use of specific aspects of retrospective evaluations of the experience (e.g., Ariely 1998; a temporal sequence when forming retrospective evaluations of an Redelmeier and Kahneman 1996). Specifically, we argue that experience. In addition to accounting for past findings such as a assuming that a memory-based structure is responsible for temporal preference for improving over declining temporal sequences and sequence findings, then introducing a delay between an affective the important role of peak (both high intensity and unique) experi- experience and evaluations of the experience should diminish the ences, our findings demonstrate that imposing a delay prior to impact of the end of an experience and enhance the impact of the retrospective evaluations can create a preference reversal due to the beginning of an experience on retrospective evaluations, reversing reduced accessibility of final or common instances. consumer preferences for improving versus declining trends. Con- sistent with expectations, in study 1 we show that an improving REFERENCES vacation experience is evaluated more favorably immediately than Ariely, Dan (1998), “Combining Experiences Over Time: The after a delay. Consistent with primacy and recency effects, recall of Effects of Duration, Intensity Changes, and On-line Measure- the early instances was not affected by the delay; however, recall of ments on Retrospective Pain Evaluations,” Journal of the final incidents declined following a delay. Study 2 replicates the Behavioral Decision Making, 11 (March), 19-45. findings from study 1 using both an improving and declining Ariely, Dan and Ziv Carmon (2000), “Gestalt Characteristics of vacation trend and demonstrates that the differential memory for Experiences: The Defining Features of Summarized Events,” events is based on their temporal location, not their valence. Journal of Behavioral Decision Making, 13 (April/June), Consistent with a memory-based explanation, the final events in the 191-201. vacation are heavily weighted when the experience is evaluated 743 Advances in Consumer Research Volume 34, © 2007 744 / Temporal Sequence Effects: A Memory Framework Ariely, Dan, Daniel Kahneman, and George F. Loewenstein (2000), “Joint Comment on ‘When Does Duration Matter in Judgment and Decision Making?’” Journal of Experimental Psychology: General, 129 (December), 524-29. Ashcraft, Mark H. (2002), Cognition, Upper Saddle River, NJ: Prentice Hall. Baumgartner, Hans, Mita Sujan, and Dan Padgett (1997), “Patterns of Affective Reactions to Advertisements: The Integration of Moment-to-Moment Responses into Overall Judgments,” Journal of Marketing Research, 34 (May), 219- 32. Bower, Gordon H. (1981), “Mood and Memory,” American Psychologist, 36 (February), 129-48. Fredrickson, Barbara L. and Daniel Kahneman (1993), “Dura- tion Neglect in Retrospective Evaluations of Affective Episodes,” Journal of Personality and Social Psychology, 65 (July), 44-55. Hamilton, D.L. and Gifford, R.K. (1976), “Illusionary Correla- tion in Interpersonal Perception: A Cognitive Basis of Stereotypic Judgments,” Journal of Experimental Social Psychology, 12 (July), 392-407. Kahneman, Daniel, Barbara L. Fredrickson, Charles A. Schreiber, and Donald A. Redelmeier (1993), “When More Pain is Preferred to Less: Adding a Better End,” Psychologi- cal Science, 4 (6), 401-05. Redelmeier, Donald A. and Daniel Kahneman (1996), “Patients’ Memories of Painful Medical Treatments: Real-Time and Retrospective Evaluations of Two Minimally Invasive Procedures,” Pain, 66 (1), 3-8. Tversky, Amos and Daniel Kahneman (1973), “Availability: A Heuristic for Judging Frequency and Probability,” Cognitive Psychology, 5 (September), 207-32. Wallace, W. (1965), “Review of the Historical, Empirical, and Theoretical Status of the von Restorff Phenomenon,” Psychological Bulletin, 63 (6), 410-24. Mood and Comparative Judgment Cheng Qiu, National University of Singapore, Singapore Catherine W. M. Yeung, National University of Singapore, Singapore

EXTENDED ABSTRACT Experiment 1. We examined the influence of mood on choices Previous research has shown that consumers’ affective expe- among options that only differed in global, aesthetic aspects. We rience, regardless of whether it is a genuine integral affective induced participants’ mood by asking them to write a piece of happy response to a target, or a feeling arising incidentally from a transient or unhappy personal experience. Then they moved on to the second mood state, can influence their evaluations of products (e.g., (ostensibly unrelated) task where they saw pictures of three mango- Adaval 2001; Pham 1998; Schwarz and Clore 1996). While we flavored desserts sequentially. After that, they indicated their know mood can influence consumers’ absolute judgments of prod- choices and absolute evaluations of the options. As predicted, the ucts, we are less clear about how it would influence consumers’ choice share of the first option was higher when participants were comparative judgments and choices among products. Consider a happy than when they were unhappy. Absolute evaluation data consumer who is choosing among a number of options. Suppose s/ confirmed that happy participants liked the first option more than he happens to be experiencing a mood state when s/he receives the unhappy participants did. information on each of these options one by one, would mood Experiment 2. In experiment 2, the timing of the first evalua- influence his/her evaluations of each of the options to a similar tion was delayed to the last option due to changes in choice set extent, and hence have no net effect on choice? Alternatively, characteristics. The procedure was essentially the same as experi- would mood influence evaluations of only one of the options, and ment 1. However, participants chose among three desserts that lead to changes of his/her choice? If so, which particular option differed in flavor (which is presumably an important descriptive would be influenced? Our research provides answers to this series feature for desserts). Consistent with our prediction, choice share of of questions. the last option was higher when participants were happy than when We suggest that mood can have a differential effect on com- they were unhappy. Happy participants also liked the last option parative judgment because consumers are likely to incorporate their more than their unhappy counterparts. mood into the first evaluation being constructed during the com- Experiment 3. In experiments 1 and 2, we assumed the timing parison process. Moreover, once incorporated, mood will cease to of the first evaluation would change as a function of choice set influence evaluations of the latter options because its informational characteristics. While the literature provides theoretical supports value for these options will be discounted. Note, however, consum- for the assumptions, we validated these assumptions empirically in ers may form the first evaluation at different points in time during experiment 3 by manipulating the timing of the first evaluation the comparison process. When the timing of the first evaluation through experimental instructions. This experiment had a 2 (in- varies, the influence of mood on comparison may vary accordingly. duced mood: positive vs. negative) by 2 (flavor: same-flavor vs. Suppose consumers evaluate the very first option from a set (e.g., different-flavor) by 3 (instruction on the timing of the first evalua- the item shown on the first page of a catalog) right away when they tion: no-instruction vs. immediate-judgment vs. deferred-judg- encounter it, before they see the other options. Their mood is likely ment) factorial design. Participants either received no instruction or to be incorporated into the evaluation of this first option rather than an instruction as to when to start making their evaluations before into those of the subsequent ones. Alternatively, suppose consum- they saw pictures of two desserts. After seeing the two options one ers deliberately wait until they encounter the last option before by one, they indicated their relative preferences. As predicted, beginning their evaluation and take a perspective focusing on the happy participants in the immediate-judgment group of conditions last encountered option. In this case, the evaluation into which (i.e., the “no-instruction, same-flavor” condition and the two “im- mood is incorporated may “shift” to the last option. As such, we mediate-judgment” conditions) had a stronger preference for the expect mood to influence the evaluation of the last option, rather first option, whereas their unhappy counterparts had a weaker than those of the preceding ones. preference for the first option. A reversed pattern was observed in We further show that the timing of constructing the first the deferred-judgment group of conditions (i.e., the “no-instruc- evaluation may vary systematically as a function of choice set tion, different-flavor” condition and the two “deferred-judgment” characteristics. For example, it may depend on whether the alterna- conditions). tives are differentiated by descriptive features or by their global In combination, findings from experiments 1-3 support the appearance. When consumers choose among options that have notion that mood is likely to be incorporated into the first evaluation similar descriptive features but differ in external appearance, they consumers make during a comparison process, and this evaluation are likely to form an evaluation of the first option very quickly once could be the evaluation of the initial option or the final option of a they come across it, without waiting to see the comparison options. choice set. Consequently, their mood is likely to influence their evaluations of the first encountered option, and hence their choices of this option. REFERENCES In contrast, when consumers choose among options that are differ- Adaval, Rashmi (2001), “Sometimes It Just Feels Right: The entiated by descriptive features, they tend to wait until they have Differential Weighting of Affect-Consistent and Affect- encountered all the options and to know about their features before Inconsistent Product Information,” Journal of Consumer making any comparison. Consequently, their mood is likely to Research, 28 (June), 1-17. influence their evaluations of the last encountered option, and Agresti, Alan (1996), An Introduction to Categorical Data hence their choices of this option. Analysis, New York: John Wiley & Sons, Inc. Three experiments were conducted to test the above predic- Bettman, James R., Eric J. Johnson, and John Payne (1991), tions regarding (a) how the influence of mood on comparison may “Consumer Decision Making,” in Thomas S. Robertson and vary as a function of the timing of the first evaluation and choice set Harold H. Kassarjian, eds., Handbook of Consumer Behav- characteristics and (b) the underlying mechanism through which ior, Englewood Cliffs, NJ: Prentice-Hall. the influence occurs. 745 Advances in Consumer Research Volume 34, © 2007 746 / Mood and Comparative Judgment Dumaine, Brian (1991), “Designs that Sells and Sells and. . .,” Strack, Fritz (1992), “The Different Routes to Social Judgments: Fortune, 11 March, 86-94. Experiential versus Informational Strategies,” in Construc- Houston, David A., Steven J. Sherman, and Sara M. Baker tion of Social Judgments, ed. Leonard L. Martin and (1989), “The Influence of Unique Features and Direction of Abraham Tesser, Hillsdale, NJ: Lawrence Erlbaum Associ- Comparison on Preferences,” Journal of Experimental Social ates, Inc., 249-75. Psychology, 25 (March), 121-41. Wang, Jing and Robert S. Wyer (2002), “Comparative Judgment Gorn, Gerald J., Marvin E. Goldberg, and Kunal Basu (1993), Processes: The Effects of Task Objectives and Time Delay “Mood, Awareness and Product Evaluation,” Journal of on Product Evaluations,” Journal of Consumer Psychology, Consumer Psychology, 2 (3), 237-56. 12 (4), 327-40. Johnson (1984), “Consumer Choice Strategies for Comparing Yeung, Catherine W. M. and Robert S. Wyer (2004), “Affect, Noncomparable Options,” Journal of Consumer Research, Appraisal, and Consumer Judgment,” Journal of Consumer 11 (December), 741-53. Research, 31 (September), 412-24. Lazarus, Richard S. (1982), “Thoughts on the Relations between Yeung, Catherine W. M. and Robert S. Wyer (2005), “Does Emotion and Cognition,” American Psychologist, 37 Loving a Brand Mean Loving Its Products? The Role of (September), 1019-24. Brand-Elicited Affect in Brand Extension Evaluations”, (1991), Emotion and Adaptation, New York: Oxford Journal of Marketing Research, 42 (November), 495-506. University Press. Pham, Michel T. (1998), “Representativeness, Relevance, and the Use of Feelings in Decision Making,” Journal of Consumer Research, 25 (September), 144-59. (2004), “The Logic of Feeling,” Journal of Consumer Psychology, 14 (4), 360-69. Pham, Michel T., Joel B. Cohen, John W. Pracejus, and G. David Hughes (2001), “Affect Monitoring and the Primacy of Feelings in Judgment,” Journal of Consumer Research, 28 (September), 167-88. Pham, Michel Tuan and Tamar Avnet (2004), “Ideals and Oughts and the Reliance on Affect versus Substance in Persuasion,” Journal of Consumer Research, 30 (March), 503-18. Schmitt, Bernd H. and Alex Simonson (1997), Marketing Aesthetics: The Strategic Management of Brands, Identity and Image, New York: The Free Press. Schwarz, Norbert (1990), “Feelings as Information: Informa- tional and Motivational Functions of Affective States,” in Handbook of Motivation and Cognition, Vol. 2, ed. Richard. M. Sorrentino and E. Tory Higgins, New York: Guilford, 521-61. (2001), “Feelings as Information: Implications for Affective Influences on Information Processing,” in Theories of Mood and Cognition: A User’s Guidebook, ed. Leonard L. Martin and Gerald L. Clore, Mahwah, NJ: Erlbaum, 159-76. Schwarz, Norbert and Gerald L. Clore (1983), “Mood, Misattribution, and Judgments of Well-Being: Informative and Directive Functions of Affective States”, Journal of Personality and Social Psychology, 45 (September), 513-23. (1996), “Feelings and Phenomenal Experiences,” in Social Psychology: Handbook of Basic Principles, ed. E. Tory Higgins and Arie W. Kruglanski, New York: Guilford, 433-65. Shiv, Baba and Alexander Fedorikhin (1999), “Heart and Mind in Conflict: The Interplay of Affect and Cognition in Consumer Decision Making,” Journal of Consumer Research, 26 (December), 278-92. (2002), “Spontaneous versus Controlled Influences of Stimulus-Based Affect on Choice Behavior,” Organizational Behavior and Human Decision Processes, 87 (March), 342- 70. Srull, Thomas K. (1987), “Memory, Mood, and Consumer Judgment,” Advances in Consumer Research, 14, 404-07. Complexity Effects on Consumer Choice Consistency Benedict Dellaert, Erasmus University Rotterdam, The Netherlands Bas Donkers, Erasmus University Rotterdam, The Netherlands Bernard Conlon, TNS Australia, Sydney, Australia Arthur Van Soest, Tilburg University, The Netherlands

EXTENDED ABSTRACT rate choices. An opposing effect of increases in VAR may be that The consistency with which consumers make product choices the differences between the products become larger, increasing the has important consequences for organizations interacting with incentives to perform well in the choice task. Hence, VAR could consumers. If consumer choices are not very consistent (i.e., if also increase choice consistency as more effort is exerted. With consumers do not make the same choices in otherwise identical respect to task complexity, Caussade et al (2005) investigate the choice situations), the impact of changes in marketing mix vari- effects of the number of alternatives and the number of attributes, ables on choice will be lowered. This situation may be harmful to finding a positive effect of the number of alternatives and a negative for example innovative producers that develop products that are effect on the number of attributes, implying a mixed result for the superior to existing alternatives. The reason is that if consumers do effect of task complexity (TASK). We are not aware of previous not respond consistently to the new product’s strengths in their research on the effect of DIF, but greater differences in utility make decisions, market share of the new product will be lower than it the best alternative stand out from the rest, we expect it to facilitate could otherwise have been. For consumers also it may be undesir- consistency in choosing the best alternative from a set. able to make inconsistent choices, for example if it leads them to To investigate the proposed effects of complexity on consis- ignore beneficial product attributes in making product purchase tency, data was collected through a stated choice experiment where decisions. each one of the 1114 respondents was randomly assigned to two of In this study we focus on choice set complexity as a potentially twelve possible choice complexity conditions. These twelve choice important driver of consumer choice consistency. Recent empirical conditions were composed of different numbers of alternatives, work on the relationship between effort and consistency (Haaijer et different numbers of attributes, and different attribute levels, result- al. 2000), and on the effects of choice set composition on the level ing in substantial variation in task and context complexity across of error in consumer choice models (Caussade et al. 2005, Swait and conditions. Adamowicz 2001) suggests that consumer choice consistency may To quantify context-based complexity and its effects on choice vary with complexity. We add to this research an analysis of consistency we first estimate a heteroskedastic mixed logit model. observed consistency in consumers’ repeated choices and how it is The variance of the error component in the mixed logit model is affected by task-based and context-based complexity. allowed to vary across choice sets, permitting the level of error in Task-based complexity refers to the number of cognitive steps the choice model estimates to vary freely on choice set composition. a consumer needs to choose an optimal product (Johnson and In the next step, we obtain individual level preference parameters. Payne, 1985). It is expressed as the combined effect of the number These parameters are used to compute attribute utilities for each of attributes and the number of alternatives in the choice set. alternative for each respondent in each condition. Based on these Context-based complexity refers to the difficulty of the trade offs utilities we calculate measures VAR, COV and DIF for each choice that consumers have to make. We express this effect using three set. TASK is a function of the number of attributes and alternatives variables based on Shugan (1980): the variability of the attribute only, and varies only across choice sets. utilities of the products in the choice set (VAR), the covariance In our main analysis, we empirically test the effects of choice between the attribute utilities of these products (COV), and the set complexity on choice consistency using a Logit model. The difference in total utility between these products (DIF). dependent variable is whether the repeated choice task is performed The effects of choice set complexity on choice consistency are consistently, i.e. the same alternative is chosen twice or not, and the not straightforward. The simulation analysis of Johnson and Payne independent variables are the complexity measures introduced (1985) shows that with equal effort, consumer choice consistency above. The results show that TASK complexity decreases consis- is inversely related to choice complexity. The effect of complexity tency and we conclude that increases in task complexity are not on consistency then depends on how consumers adapt their choice sufficiently compensated by increased effort to maintain the same strategy and their effort level. In particular, if consumers respond to level of accuracy. For context complexity, we find that differences increased complexity by increasing their effort, the consistency of in utility between the alternatives in the choice set increased choice their decisions may be stable (or even improve) if complexity consistency, Variance of the attribute utilities in the choice set increases. alternatives also increased choice consistency, and the effect of Results of Haaijer et al. (2000) suggest that in general covariance between the attribute utilities in the choice set alterna- consumers’ effort responses to increases in choice set complexity tives was not significant. These results suggest that increased are not sufficient to maintain equal choice consistency. Fischer et differences in utility between alternatives and greater variability in al. (2000) analyze consumer preference judgments. They find that, the attribute levels raise the stakes to the decision maker, resulting if judgment tasks become more complex in terms of variance, in more effort being exerted in the choice task and higher choice responses take more effort and become less accurate. The latter consistency. finding suggests that increases in VAR in a choice context may also lead to less consistent choice behavior. Dellaert et al. (1999) find REFERENCES that logit model error increases when price based utility trade-offs Bettman, James R., Eric J. Johnson and John W. Payne, (1990), increase. This effect also suggests that increases in VAR (i.e., “A Computational Analysis of Cognitive Effort in Choice,” higher price variance), and possibly decreases in COV (i.e., lower Organizational Behavior and Human Decision Processes 45, correlations between price and other attributes), lead to less accu- 111-139.

747 Advances in Consumer Research Volume 34, © 2007 748 / Complexity Effects on Consumer Choice Consistency , Eric J. Johnson, Mary Frances Luce, and John W. Payne, (1993), “Correlation, Conflict and Choice,” Journal of Experimental Psychological: Learning, Memory and Cognition 19(4), 931-950. Caussade, Sebastian, Juan de Dios Ortuzar, Luis I. Rizzi and David A. Hensher (2005), “Assessing the Influence of Design Dimensions on Stated Choice Experiment Estimates,” Transportation Research Part B 39(7), 621-640. Dellaert, Benedict G.C., Jeff D. Brazell and Jordan J. Louviere, (1999), “The Effect of Attribute Variation on Consumer Choice Consistency,” Marketing Letters 10(2), 139-147. Fischer, Gregory W., Mary Frances Luce and Jianmin Jia, (2000), “Attribute Conflict and Preference Uncertainty: Effects of Judgment Time and Error,” Management Science 46(1), 88-103. Haaijer, Rinus, Wagner Kamakura and Michel Wedel, (2000), “Response Latencies in the Analysis of Conjoint Choice Experiments,” Journal of Marketing Research 37(August), 376-382. Huber, Oswald, (1980), “The Influence of Some Task Variables in an Information-Processing Decision Model,” Acta Psychologica 45, 187-196. Johnson, Eric J. and John W. Payne, (1985), “Effort and Consistency in Choice,” Management Science 31(4), 395- 414. Louviere, J.J. (2001) “What If Consumer Experiments Impact Variances As Well As Means: Response Variability As A Behavioral Phenomenon,” Journal of Consumer Research, 28 (3), 506-511. McFadden, Daniel and Kenneth Train, (2000), “Mixed MNL Models for Discrete Response,” Journal of Applied Econo- metrics 15(5), 447-470. Newell, A. and Simon, H. (1972), Human Problem Solving, Prentice-Hall, Englewood Cliffs, N. J. Shugan, Steven M. (1980), “The Cost of Thinking,” Journal of Consumer Research 7, 99-111. Swait, Joffre and Wiktor Adamowicz (2001), “Choice Environ- ment, Market Complexity, and Consumer Behavior: A Theoretical and Empirical Approach for Incorporating Decision Complexity into Models of Consumer Choice,” Organizational Behavior and Human Decision Processes, 86(2), 141-167. Recollections of Pleasure and Pain: When Losers are Motivated to Edit their Retrospective Evaluations Elizabeth Cowley, University of Sydney, Australia Colin Farrell, University of Sydney, Australia

EXTENDED ABSTRACT won on the peak win, the amount of money lost on the peak loss, the People have unrealistically positive views of self and are trend of outcomes (winning, losing), the final cash position, and unrealistically optimistic about the future (Taylor and Brown their score on the illusion of control scale. The retrospective 1988). For instance, they believe they will be more likely than their evaluation of the experience for winners was determined by the peers to like their first job, earn a good salary, and have a ‘gifted’ final cash position only, a larger win resulted in a more positive child (Weinstein 1980). On the other hand, people believe that they retrospective evaluation. The retrospective evaluation for losers are much less likely than their peers to be a victim of crime (Perloff was determined by the peak win. Losers appear to look for the silver and Fetzer 1986) or contract the HIV virus (Raghubir and Menon lining (i.e. something positive) in an otherwise negative situation. 1998). All of these examples show that people are optimistic when The illusion of control score was also significant: the higher the they predict the outcome of future events. Will the same biases illusion of control, the more positive the retrospective evaluation. occur when people reflect back on events that have already oc- We also ran an ANOVA on the retrospective evaluation with curred? Are people motivated to make the past seem better than it final outcome (winner, loser) and the illusion of control score (high, was? Autobiographical memory research shows that the intensity low). The ANOVA revealed a significant interaction. It is the losers of negative affect associated with memories fades faster over time with a high score on the illusion of control scale that edit the than positive affect (Walker, Vogl, and Thompson 1997). Addition- retrospective evaluation to be more positive. The losers with a low ally, hedonic editing suggests that people focus on the ‘silver lining’ illusion of control score were significantly less positive than the in a loss frame (Thaler 1985) to maximize their happiness. How- winners with a low illusion of control score. The losers with a high ever, to date there is no evidence that people in a loss frame edit their illusion of control were not significantly different in their evalua- retrospective evaluations of an experience to be more positive. tion of the experience. We propose that gamblers in a loss frame are motivated to Previous research has found that the peak moment and the final hedonically edit their evaluation of the experience such that the moments of an experience play an important role in the construction gambling experience will be remembered more positively. We of retrospective evaluations of purely painful experiences. An show that losers use the peak win to improve the retrospective opportunity to use an optimizing strategy in the construction of a evaluation of a gambling session. In particular, gamblers that retrospective evaluation is present when reflecting back on a mixed believe that winning at gambling is a skill they possess are moti- experience that is not available when remembering a purely painful vated to edit their retrospective evaluation of a losing gambling experience. We found that when people are asked about a past episode. We also propose that gamblers are not motivated to gambling experience, the peak win is important in the memory of hedonically edit their evaluation of a winning gambling episode. losers and the final cash outcome is important in the memory of Instead, we show that the retrospective evaluation of the winners is winners: facilitating the construction of a most positive retrospec- determined by the amount of cash they won, regardless of their tive evaluation. Interestingly, we found that the illusion of control belief that gambling is a skill that they possess. did not affect the retrospective evaluation of the winners, but it did In study 1, participants were given 30 opportunities to bet on affect the losers. It was the losers with a high illusion of control the future outcome of a coin toss. Once each outcome was revealed, score that edited their retrospective evaluation. the study administrator distributed winnings and collected losses based on the outcome of the coin toss. Participants were asked to REFERENCES report how the outcome of the bet made them feel by marking an ‘X’ Perloff, Linda S. and Barbara K. Fetzer (1986), “Self-Other on a 125 mm continuous scale anchored with a happy face and a sad Judgments and Perceived Vulnerability to Victimization,” face. A time series cross sectional regression analysis was run on Journal of Personality and Social Psychology, 50(3), 502- moment-by-moment evaluations with the outcome for each bet as 510. the independent variable as an additional test of whether the current Raghubir, Priya and Geeta Menon (1998), “AIDS and Me, Never outcome explained the current moment-by-moment evaluation. To the Twain Shall Meet: The Effects of Information Accessibil- check for the independence of moment-by-moment evaluations, a ity on Judgments of Risk and Advertising Effectiveness,” lagged outcome variable and a lagged evaluation variable were Journal of Consumer Research, 25(1), 52-63. included to test whether current moment-by moment evaluations Taylor, Shelley E. and Jonathon D. Brown (1988), “Illusion and were affected by the previous outcome or the evaluation associated Well-Being: A Social Psychological Perspective on Mental with the previous outcome. The results indicate that the current Health,” Psychological Bulletin, 103(2), 193-210. outcome explained the moment-by-moment evaluation, that there Thaler, Richard H. (1985), “Mental Accounting and Consumer were no remnants of the previous outcome, and that there were only Choice,” Marketing Science, 4 (3),199-214. slight remnants of previous evaluations. This is an important Walker, W. Richard, Rodney J. Vogl, and Charles P. Thompson finding because study 2 is based on an assumption that the outcome (1997). Autobiographical Memory: Unpleasantness Fades of each gamble (amount won or lost) determines the moment-by- Faster than Pleasantness Over Time,” Applied Cognitive moment evaluations. Psychology, 11, 399-413. In Study 2, 102 gamblers bet on 20 coin toss games. After the Weinstein, Neil D. (1980), “Unrealistic Optimism about Future gambling experience, gamblers were asked to report their evalua- Life Events,” Journal of Personality and Social Psychology, tions of the experience and their intention to replay the game, and 39(5), 806-820. the illusion of control scale. The retrospective evaluations of winners and losers were tested with separate regressions including the amount of money 749 Advances in Consumer Research Volume 34, © 2007