Cs1

CHEZY OFIR and ITAMAR SIMONSON* Customers’ expectations are key determinants of their consumption experiences, satisfaction, and loyalty. The authors examine alternative theoretical predictions about the impact of stating expectations before purchase on postpurchase perceptions of the shopping experience and the firm. In particular, the authors suggest that asking customers to articulate their expectations can backfire and lead to lower postpurchase evaluations of the shopping and consumption experience. A series of field experiments indicate that compared with a control group, stating prepurchase expectations leads customers to focus on negative aspects of the shopping experience and perceive the same performance more negatively. The tendency for consumers to rate their shopping experiences less favorably after stating prepurchase expectations is inconsistent with confirmation bias as well as assimilation, contrast, and positivity effects. The final study contrasts the impact of stating expectations about the next store shopping experience with the impact of evaluating the store’s past performance. The results show that though (prepurchase) expectations are indistinguishable from evaluations of the store’s past performance, the former leads to lower postpurchase than prepurchase evaluations, whereas the latter tends to generate higher postpurchase evaluations. The article concludes with a discussion of the theoretical and practical implications of this research.

The Effect of Stating Expectations on Customer Satisfaction and Shopping Experience It is now widely accepted that exceeding customer expectations is key to customer satisfaction, delight, and loyalty (e.g., Kotler 2000, p. 36). Accordingly, it is critical for marketers to find out about their customers’ expectations in advance because a failure to meet or exceed these expectations could lead to dissatisfaction and defection. In some instances, customers have well-formed expectations—for example, when they have a great deal of experience with a particular service or product. In other instances, expectations may be ill-defined, in which case asking customers to state expectations might help formulate or even create them (e.g., Bettman, Luce, and Payne 1998; Schwarz and Bohner

2001). Furthermore, regardless of whether the measurement of expectations affects their content and clarity, the need for customers to state expectations before a consumption experience is likely to make them more accessible during the experience (Feldman and Lynch 1988). As we discuss subsequently, these and related issues have been examined in the context of research on the effects of measuring intentions, judgments, and satisfaction (e.g., Dholakia and Morwitz 2002; Fitzsimons and Williams 2000; Kardes, Allen, and Pontes 1993; Morwitz and Fitzsimons 2004). In the current research, we examine the impact of measuring customers’ expectations just before a shopping or consumption experience (e.g., before entering a store or a bank) on their subsequent evaluations of that experience and the firm. Prior research has suggested that measuring customers’ satisfaction with the firm’s past performance induces more favorable evaluations of the firm and has a positive effect on purchase behavior and loyalty (Dholakia and Morwitz 2002). As Dholakia and Morwitz (2002) point out, this finding suggests the possibility that satisfaction surveys can be used strategically to strengthen customer relationships. Because evaluations of satisfaction with the

*Chezy Ofir is Academic Director of the Davidson Research Center, School of Business Administration, Hebrew University, Mount Scopus, Jerusalem (e-mail: [email protected]). Itamar Simonson is Sebastian S. Kresge Professor of Marketing, Graduate School of Business, Stanford University (e-mail: [email protected]). The authors acknowledge the support of the Davidson and Kmart Research Centers at the Hebrew University and the Stanford Graduate School of Business, as well as the assistance of students who participated in data collection and analysis.

© 2007, American Marketing Association ISSN: 0022-2437 (print), 1547-7193 (electronic)

164

Journal of Marketing Research Vol. XLIV (February 2007), 164–174

Stating Expectations Affects Customer Satisfaction firm’s past performance are likely to provide the basis for expectations about its future performance, measuring expectations might have a similar effect on subsequent evaluations and behavior as measuring satisfaction. However, research also indicates that customers who are forewarned before a service encounter that they will be asked subsequently to evaluate that service provide more negative postexperience evaluations than customers who are unaware of an upcoming service evaluation task (Ofir and Simonson 2001). If it is assumed that expecting to evaluate a service or product spontaneously generates expectations (for a related argument, see, e.g., Fitzsimons and Williams 2000), stating expectations before a consumption experience might actually generate more negative (i.e., lower) evaluations of the firm after the experience. In the next section, we explore the different predictions that can be derived from prior research about the effect of stating expectations on subsequent evaluations. We then show that measuring expectations tends to generate lower evaluations. We also present evidence regarding the evaluation processes triggered by stating expectations, which indicates that the measurement of expectations causes customers to pay more attention to negative aspects and to encode performance characteristics more negatively. In the final study, we directly contrast the effect of measuring prior satisfaction with the effect of stating expectations before purchase. We find that though stated expectations about future performance are indistinguishable from judgments of past performance, the measurement of expectations generates lower subsequent evaluations, whereas the assessment of past performance tends to produce more positive evaluations. We conclude with a discussion of the theoretical and practical implications of the results. INFLUENCES OF STATING EXPECTATIONS: CONFIRMATION, POSITIVITY, AND/OR NEGATIVITY ENHANCEMENT As is often the case in the analysis of the impact of a particular condition on response, the difficulty is not in identifying potentially relevant theories but rather in determining which of several candidate theories is most applicable. In particular, existing theories could support different predictions about the effect of stating expectations about a consumption experience. Specifically, (1) stating expectations might confirm and reinforce existing beliefs about the experience under consideration, which in turn affects postexperience evaluations in accordance with the stated expectations; (2) stating expectations might generate higher postexperience evaluations; and (3) stating expectations might generate lower postexperience evaluations. An assumption that we can make rather confidently is that the act of stating expectations for a service or product experience just before the service encounter or consumption experience makes the stated expectations more accessible and salient during that experience and possibly the subsequent experiences (for related arguments, see, e.g., Dholakia and Morwitz 2002; Feldman and Lynch 1988; Fitzsimons and Williams 2000; Kardes, Allen, and Pontes 1993; Morwitz and Fitzsimons 2004). It might also be expected that the mere act of stating expectations enhances involvement with the subsequent shopping experience, leading to more central-route processing of information and poten-

165 tially more accurate judgments (e.g., Petty and Cacioppo 1984). Furthermore, prior research indicates that though stated expectations may be influenced by preexisting expectations and other information stored in memory, they are often constructed and are clarified when the need to articulate them arises (e.g., Bettman, Luce, and Payne 1998; Schwarz and Bohner 2001). However, these prior findings regarding accessibility, involvement, and preference construction do not lead to directional predictions about the impact of stating expectations on the valence of postexperience evaluations. Assimilation and confirmation bias (e.g., Klayman and Ha 1987; Nisbett and Ross 1980) can possibly account for the effect of stating expectations on postexperience evaluations. Specifically, stating expectations may reinforce prior beliefs and lead to postpurchase evaluations that confirm and are even more polarized than prior expectations (e.g., Tesser 1978). Alternatively, we might observe a contrast effect or regression to the mean, such that people with positive prior expectations will subsequently have less favorable evaluations and those with lower expectations will subsequently evaluate the store more favorably. If we assume that the measurement of satisfaction with past performance is equivalent to the measurement of expectations about future performance, we might observe a positive effect of stating expectations on subsequent evaluations. Specifically, as Dholakia and Morwitz (2002) argue, users of a firm’s service (e.g., a store, a bank) are likely to include a relatively high proportion of satisfied customers, which is often the reason they are that firm’s customers in the first place. Accordingly, asking these customers to state their expectations for the firm causes them to articulate their favorable evaluations and thus further reinforces their positive attitude, leading to more positive postexperience evaluations. It is noteworthy that participants in Dholakia and Morwitz’s studies were aware that the satisfaction study was being conducted on behalf of the firm, which might have contributed to the positive impact of the survey on subsequent customer behavior (see also Dholakia, Morwitz, and Westbrook 2005). However, recent research by Williams, Fitzsimons, and Block (2004) suggests that such mere measurement effects also occur and could even be more pronounced when the organization that sponsors or conducts the research is not identified. Yet another possible effect of stating expectations is to generate lower postexperience evaluations. Ofir and Simonson (2001) show that customers who expect to evaluate a service before the actual experience subsequently tend to evaluate that service more negatively than customers who find out about the evaluation task after the experience. They offer two possible explanations for this effect: negativity enhancement and role expectations. The latter account is specific to the expected evaluation task and is based on the notion that consumers believe that their evaluation role calls for constructive criticism, which generates more negative evaluations. Conversely, the negativity enhancement account suggests that a general tendency to overweigh negative aspects (e.g., Fiske 1980; Kanouse 1984; Tversky and Kahneman 1991) becomes more pronounced when evaluations are constructed online. As a result, any task condition that stimulates more online than memory-based evaluations (e.g., Hastie and Park 1986) would be expected to promote more negative evaluations (e.g., of service

166

JOURNAL OF MARKETING RESEARCH, FEBRUARY 2007

providers). Consistent with this account, an explicit need to form and state performance expectations before a service encounter might have a similar negative effect on subsequent evaluations because it is likely to encourage comparisons between the stated expectations and the actual performance. Furthermore, Ofir and Simonson (2001) show that stating expectations about a service also has an independent negative effect on subsequent evaluations of the service. However, they do not pursue this finding but rather focus on the effect of expecting to evaluate. Thus, because negativity enhancement accounts for the negative effect of expecting to evaluate, it leads to the prediction that stating expectations before a service encounter generates more negative subsequent evaluations of the service experience. In summary, prior research can support different predictions about the impact of stating expectations before a service encounter on postexperience evaluations of that service. Specifically, (1) an assimilation effect and confirmation bias suggest that (positive or negative) expectations stated before the experience largely determine and make more extreme postexperience evaluations; (2) a contrast effect suggests that people who state positive (negative) expectations tend to evaluate the service more negatively (positively) after the experience; (3) if the measurement of expectations is equivalent to the measurement of satisfaction with past performance, stating expectations could lead to more positive evaluations; and (4) negativity enhancement suggests that stating expectations produces more negative subsequent evaluations. We examine these rival predictions in the following studies. STUDY 1 In Study 1, we contrast the postpurchase evaluations of two customer groups: shoppers who state their expectations for a shopping experience before entering a supermarket and shoppers in a control group who are interviewed only when they exit the supermarket. As we discussed previously, given conflicting theoretical accounts, we use this study as a preliminary investigation. Method The participants were 80 shoppers who were interviewed at the entrance to a supermarket and received no compensation. Those randomly assigned to the stated-expectations group (n = 40) were interviewed both before and after they shopped at the store, whereas the control group (n = 40) was interviewed only after they shopped. When respondents were first interviewed, they were informed that the survey was part of a university research project and were assured of confidentiality. Those in the stated-expectations group were asked to indicate their expectations about the service they would receive at the supermarket on that day. Specifically, they stated their expectations for the store on nine dimensions, using seven-point scales. Six items assessed specific dimensions of expected service quality, including politeness of store employees, employees’ willingness to assist shoppers, employees’ professionalism, length of wait at the checkout, convenience of product display, and level of service. Two items measured expected satisfaction directly: overall satisfaction with the service and overall satisfaction with the shopping visit. Finally, participants were asked about the likelihood that they would recommend the store to friends

(for an alternative scale, see Parasuraman, Zeithaml, and Berry 1988). When participants exited the store, those in the statedexpectations group and those assigned to the control group were unexpectedly asked to evaluate their experience and the store on the basis of the visit they just completed. They rated the store using the same items as the statedexpectations group used before entering the store. Respondents were also asked whether they paid more attention to “positive” (7) or “negative” (1) aspects while shopping at the store. Respondents were randomly assigned to conditions as they were about to enter the supermarket. Interviewers were able to track shoppers assigned to each group, who entered the supermarket through the same entrance/exit. A concern may arise about the condition assignment procedure and the possibility that participants in the stated-expectations and control groups differed in systematic ways. We addressed these concerns in four ways. First, interviewers were blind to the purpose of the study and the research hypotheses. Second, to disguise the fact that those in the statedexpectations group would be reinterviewed when they exited the store, the interviewer thanked them at the conclusion of the preshopping interview as if they completed their task. Third, we included the following question at the end of the postshopping interview: “To what extent did you expect that we would interview you about your shopping experience at the store?” (1 = “did not expect at all,” and 7 = “definitely expected”). Fourth, in Study 2, we replicated the test of the effect of stating expectations using a more traditional lab experiment in which random assignment of participants to conditions was straightforward. Results Neither the stated-expectations group (M = 2.0, on a seven-point scale) nor the control group (M = 1.7) expected to be interviewed after the shopping visit (F(1, 78) = 1.2, p > .1). The average previsit service expectations in the stated-expectations group across the six specific service items (α = .91) was 5.4 on a seven-point scale. The average of the two satisfaction items (α = .98) was 5.9, and the average rating of the likelihood of recommending the store was 5.9. After the shopping visit, participants in the statedexpectations group evaluated the supermarket more negatively than those in the control group: For the combined service measure, MAfter = 3.9 versus MControl = 4.5 (F(1, 78) = 6.2, p < .01); for the combined satisfaction measure, MAfter = 4.0 versus MControl = 4.8 (F(1, 78) = 6.2, p < .01); and for the recommendation measure, MAfter = 3.7 versus MControl = 4.7 (F(1, 78) = 5.3, p < .01). Moreover, the postpurchase evaluations of those in the statedexpectations group were lower than their preshopping expectations on all dimensions (p < .01).1 It is noteworthy that the prepurchase expectations in the stated-expectations group were higher than the postpurchase evaluations of the control group; however, we observed no such pattern (or

1A median-split based on stated expectations further shows that after the shopping experience, participants with higher prior expectations were as negative as those with more negative prior expectations. We observed a similar pattern in the subsequent studies.

Stating Expectations Affects Customer Satisfaction any other pattern) in subsequent studies. Finally, when participants were asked during the postpurchase survey whether they paid more attention to “negative” (1) or “positive” (7) aspects of their visit, the ratings of customers in the stated-expectations group indicated that they paid more attention than those in the control group to negative aspects of their shopping visit (MStated Exp = 4.3 versus MControl = 5.2; F(1, 78) = 9.9, p < .01). Discussion The main finding of Study 1 was that stating expectations just before a shopping experience led participants to have more negative evaluations of the service and the store after the experience compared with the control group (which evaluated the store only after the shopping experience) and relative to their preexperience expectations. We measured all expectations/evaluations, both before and after the experience, on the same dimensions. These findings are inconsistent with accounts based on confirmation bias or assimilation, which imply that stating positive expectations before entering a store makes subsequent evaluations more positive rather than more negative. Furthermore, the results of Study 1 indicate that unlike satisfaction surveys, stating expectations does not appear to create more favorable attitudes toward the firm or the store. The results of Study 1 do not enable us to distinguish between the negative effect of stating expectations, as negativity enhancement suggests (Ofir and Simonson 2001), and a contrast effect. That is, although a median-split analysis showed that both the relatively high- and the relatively low-stated-expectations groups expressed more negative evaluations after the shopping experience, a contrast effect could have still contributed to the results. We addressed this issue in a follow-up study with 90 consumers in which we tested the effect of stating expectations before entering a supermarket that had a rather poor reputation (due to prior service problems). Thus, given low expectations, a contrast effect would suggest that postpurchase evaluations would be more positive than stated expectations, whereas negativity enhancement predicts that the negative effect of stating expectations on postexperience evaluations would still be observed. In this follow-up study, we added a recall measure that was taken immediately after participants exited the store. Specifically, respondents were asked, “What do you remember from today’s visit to the supermarket?” Coders, who were blind to the hypotheses, classified the responses as “positive,” “negative,” or “neutral.” We also included a third condition in which respondents were interviewed only before they entered the store and were asked the following open-ended question: “What are your expectations for today’s shopping visit at this supermarket?” We designed this question so that the answers would provide insights into the spontaneous expectations of shoppers that were not influenced by provided closed-ended items. The results of this follow-up study showed that the postexperience evaluations in the stated-expectations group were again lower than in the control group on all measures (ps < .01). Furthermore, an analysis of the open-ended responses showed that participants in the statedexpectations group recalled more negative (M = 2.2 versus M = 1.1) and fewer positive (.8 versus 2.0) aspects of the shopping experience than those in the control group. These

167 results were consistent with the participants’ own ratings of their focus of attention; on a seven-point scale (1 = “paid attention more to negative aspects”), the average rating in the control group was 4.9, whereas in the statedexpectations group, it was 3.4 (F(1, 58) = 19.4, p < .01). These results support the negativity enhancement account and are inconsistent with the explanation based on the operation of a contrast effect. Finally, respondents in the (preshopping) expectations-only group stated more positive (1.6) than negative (.9) expectations. A possible concern with the studies described so far is that despite the attempt to assign respondents randomly to conditions and the reliance on interviewers who were blind to the research purpose and hypotheses, the participants in the control and stated-expectations groups differed on other dimensions (e.g., their prior evaluations of the store). Accordingly, in Study 2, we used a procedure that ensured that the assignment of respondents to conditions was random. We also attempted to gain tighter control over the actual content of the considered information. Furthermore, Study 2 examines another rival (or complementary) account, according to which stating expectations also changes how participants encode the same items of information. That is, it might be conjectured that in addition to any shift in the focus of attention as a result of stating expectations, the task may affect how customers interpret the observed performance of the store (see, e.g., Schwarz and Bohner 2001). STUDY 2 Method The respondents were 60 Israeli consumers who were interviewed at their homes or offices. They were asked to evaluate a new (fictitious) drugstore (referred to as “Pharm”) chain on the basis of a newspaper review of the chain’s first store. Respondents were randomly assigned to the stated-expectations (n = 30) or control (n = 30) group and were informed that the research was part of a university study. Participants in the stated-expectations group were told in advance that they would be shown a newspaper review of the first store of the new Pharm chain. They were also informed that they were randomly selected for the survey to gauge customers’ likely evaluations of the new chain. Before reading the article, these respondents stated their expectations with respect to service and overall expected evaluation of the store (both on a seven-point scale). Note that though respondents had no real basis for stating expectations, the use of the (abbreviated) English term “Pharm” and the existence of a popular drugstore chain called SuperPharm were likely to produce relatively high expectations (compared with typical Israeli pharmacies, which tend to offer less variety and poorer service). Respondents in the control group were given the same introduction but were not asked to state their expectations for the store. The translated text of the article appears in the Web Appendix (see http://www.marketingpower.com/content 84061.php). We designed the store review to contain 30 characteristics of a drugstore, with the same number of positive, negative, and neutral characteristics. Specifically, 18 pretest respondents from the same population that participated in Study 2 were asked to rate characteristics of a Pharm store on a seven-point scale (1 = “negative,” and 7 =

168

JOURNAL OF MARKETING RESEARCH, FEBRUARY 2007

“positive”). We classified items with an average rating higher than 4.75 as “positive” characteristics, items with ratings between 4.75 and 3.25 as “neutral,” and items with average ratings less than 3.25 as “negative.” After respondents finished reading the article, it was removed from sight. They were then given four minutes to list as many of the characteristics of the drugstore as they could remember from the article. Next, respondents rated the degree to which each of the 30 drugstore characteristics mentioned in the article was “negative” (1) or “positive” (7). The predesignated positive, negative, and neutral characteristics were intermixed on the list. Respondents then rated the service of the reviewed store on four (seven-point) items (cleanliness, professionalism of employees, product variety, and service quality), overall store evaluation on two items (negative/positive and bad/good), and whether they would recommend the store to friends. Finally, respondents indicated whether they had paid more attention to “negative” (1) or “positive” (7) aspects. Results As we expected, solely on the basis of the mention of a new Pharm chain, respondents in the stated-expectations group had relatively high expectations: 5.3 on service and 5.6 for the store’s overall score. We next examined respondents’ recall of store characteristics. Those in the control group remembered an average of 4.1 (predesignated) positive features, 2.9 negative features, and 2.9 neutral features. Conversely, the average number of positive, negative, and neutral recalled characteristics in the stated-expectations group was 1.7, 3.9, and 1.8, respectively. Thus, the control group remembered more positive (and neutral) features and fewer negative features (F(1, 58) = 41.8, p < .01, and F(1, 58) = 6.1, p < .01, respectively); negative features represented 54% of the features recalled in the statedexpectations group, compared with only 29% in the control group. These results were consistent with the participants’ own ratings regarding their focus of attention, which indicated that those in the stated-expectations group focused more on negative aspects (M = 3.3; 7 = “paid attention more to positive aspects”) than those in the control group (M = 5.6; F(1, 58) = 77.4, p < .01). Next, respondents rated each of the 30 store characteristics mentioned in the review on a seven-point scale (1 = “negative,” and 7 = “positive”). In all cases, the characteristics were rated more negatively in the stated-expectations group (in all cases, p < .01). For example, (1) the statement “A store membership card can be purchased for a small fee” received a rating of 2.6 in the stated-expectations group and 6.1 in the control group, (2) the statement “The store manager has an academic degree” received an average rating of 5.0 in the stated-expectations group and 6.6 in the control group, and (3) the statement “The store does not accept checks” received an average rating of 2.3 in the stated-expectations group and 4.2 in the control group. Thus, the same store features were rated systematically more negatively in the stated-expectations group. Furthermore, the average ratings on the provided service items (α = .95), store overall (α = .94), and recommendation items followed a similar pattern. The average service (M = 3.2 versus M = 5.7; F(1, 58) = 127.7, p < .01), store overall (M = 3.2 versus M = 5.9; F(1, 58) = 161.8, p < .01), and likelihood of recommendation (M = 3.1 versus M = 5.9;

F(1, 58) = 156.7, p < .01) ratings were all lower in the stated-expectations group. It is also noteworthy that we observed the lower evaluations produced by stating expectations among both participants with high expectations and participants with low expectations (based on a median split of prior expectations), which is inconsistent with a simple contrast effect of stated expectations on subsequent evaluations. Discussion The results of Study 2 provide further insights into the lower postpurchase evaluations of customers who first state their expectations. Specifically, the results show that not only do stated expectations lead to a focus on negative aspects of the available information, but stating expectations also affects the encoding and interpretation of the same items of information. That is, in addition to remembering more negative features mentioned in the store review, participants who stated expectations about the store subsequently rated the same store features lower (i.e., more negative) than the control group. Thus, stating expectations evidently triggers a critical, negatively skewed evaluation mode and focus of attention. A question that arises about the negative impact of stating expectations on subsequent evaluations is whether the mere act of stating expectations, even regarding an unrelated entity, triggers a negative bias (i.e., lower evaluations). Accordingly, in Study 3, we examined the impact of stating expectations about unrelated firms, one that was known to be associated with favorable expectations and one that was known to be associated with unfavorable expectations. STUDY 3 Method Study 3 included four experimental groups: a control and three different stated-expectations groups. The respondents were 160 supermarket shoppers who were randomly assigned to one of the four conditions (n = 40 per group). In addition to the standard stated-expectations group, the other two groups stated expectations (also before entering the supermarket) about either the duty-free shop in Israel’s main airport or a service center of the largest Israeli communication company (“Bezek”). The duty-free shop was known to have favorable associations even among those who had not shopped at the store, whereas the communication company had been involved in a severe labor dispute (related to the privatization of the company) and, in general, was associated with rather poor customer evaluations. We reasoned that if the mere act of stating expectations leads to lower evaluations even when the expectations refer to an unrelated target, all stated-expectations groups should evaluate the supermarket equally negatively after leaving. An alternative prediction is that stating expectations about the supermarket visit lowers subsequent evaluations, whereas stating expectations about other entities has no effect or leads to a corresponding contrast effect. According to the latter account, people who state expectations about the subsequently evaluated target will be more negative than the control group, whereas the sign of the difference between the other stated-expectations groups and the control will depend on whether the evaluated supermarket is perceived more or less favorably than the reference point evaluated earlier.

Stating Expectations Affects Customer Satisfaction

169

All three stated-expectations groups indicated their expectations (about the supermarket, the duty-free shop, or the communication company’s service center) using items similar to those we used in Study 1. There were only two slight changes with respect to the telephone company’s service center: We changed length of wait at the checkout to length of wait at the service desk, and we changed convenience of product display to product availability. After participants in all four groups completed their shopping, they answered an open-ended question about what they could recall from the just-completed shopping experience. Respondents then evaluated the store visit in terms of service, satisfaction, and likelihood of recommendation (similar to Study 1). Finally, respondents indicated whether they expected to be asked about their shopping experience at the supermarket and whether they paid more attention to “negative” (1) or “positive” (7) aspects while shopping.

Omnibus one-way analyses of variance applied to all post– supermarket visit responses (i.e., service, satisfaction, and recommendation measures) were significant at the p < .01 level. A set of contrasts supported the conclusions about the differences between groups. An analysis of the content of participants’ recall from the supermarket visit revealed a similar pattern. Those in the supermarket stated-expectations group remembered the highest number of negative aspects, whereas the control group recalled the highest number of positive aspects; the other two groups were either equal to the control group or between the control and the supermarket statedexpectations groups (see Table 1). Finally, participants’ own reports (on a seven-point scale) about their focus of attention while shopping showed that those in the supermarket stated-expectations group were most focused on negative aspects (M = 2.5), followed by the duty-free group (M = 3.1), the control group (M = 4.8), and the communication company group (M = 5.3).

Results All four groups did not expect to be questioned about their supermarket shopping experience (the average ratings in all groups were under 2.0 on a seven-point scale. Consistent with prior studies, in the standard stated-expectations group, expectations about the supermarket were higher than the same customers’ subsequent postpurchase ratings (for service, paired-t(39) = 12.2, p < .01; for satisfaction, pairedt(39) = 11.6, p < .01; and for recommendation, pairedt(39) = 14.0, p < .01). The postvisit evaluations were also lower than the control group’s evaluations (p < .01 for service, satisfaction, and recommendation; see Table 1). As we predicted, the stated expectations about the dutyfree shop were the highest (over 6 on a seven-point scale), followed by the supermarket (between 5.8 and 5.9); the communication company received substantially lower expectation ratings (2.0–2.3). The postvisit evaluations showed that participants in the telephone company group rated their supermarket visit most favorably, followed by the control group, the duty-free group, and, finally, the supermarket stated-expectations group, which was the most negative. These results suggest that both a contrast effect and a negative effect of stating expectations about the target (supermarket) contributed to the pattern of findings.

Discussion Consistent with the previous findings, the results of Study 3 indicate that though a contrast effect relative to a reference point can affect evaluations of the target experience, a contrast effect cannot account for the robust negative effect of stating expectations about the target on postexperience evaluations. For example, the relative proportion of recalled negative aspects was the highest in the supermarket stated-expectations group and was higher than in the duty-free group, despite the positive stated expectations for that shop. This finding is also consistent with the results of the follow-up study we reported previously, in which even a supermarket associated with relatively negative expectations was evaluated less favorably among shoppers who had stated their expectations. Furthermore, consistent with the prior studies, the results of Study 3 show that stating expectations does not produce a confirmation bias. The finding that stating expectations about a service provider before a service encounter generates more negative postexperience evaluations may appear to be inconsistent with prior findings, which indicate that asking customers to evaluate their satisfaction with a firm’s past performance

Table 1 STUDY 3: STATED-EXPECTATIONS FOCUS AND POSTEXPERIENCE EVALUATIONS Stated Expectations: Supermarket

Stated Expectations: Duty Free

Stated Expectations: Communication Company

5.8 5.9 5.8

6.1 6.5 6.5

2.3 2.0 2.0

5.5 5.6 5.6

4.3 3.7 3.5

4.5 4.1 4.0

5.8 6.0 6.1

1.8

1.0

1.2

1.8

0.6

2.4

1.8

0.6

Control Stated Expectations Expected service Expected satisfaction Recommendation Postvisit Supermarket Evaluations Service Overall satisfaction Recommendation Postvisit Recall Number of positive recalled aspects Number of negative recalled aspects

Notes: Entries are average ratings (on seven-point scales) or the average number of recalled aspects.

170

JOURNAL OF MARKETING RESEARCH, FEBRUARY 2007

has a positive effect on their relationship with the firm (Dholakia and Morwitz 2002; Dholakia, Morwitz, and Westbrook 2005). That is, if the assumption is made that expectations from a service encounter are based on prior experience with the service provider, a task that requires customers to state their expectations should have a similar effect to a task that involves evaluations of the firm’s past performance. However, whereas stating (future) expectations generates negativity enhancement, past-performance evaluations may trigger a process of confirming or seeking support for those evaluations. In the next study, we examine whether the seemingly conflicting effects of considering future versus past performance can occur within the same study. STUDY 4 Method The respondents in Study 4 were 201 supermarket shoppers, who received no compensation for their participation. They were told that the study was part of university graduate students’ research and were assured that their responses would remain anonymous. Participants were randomly assigned to one of four groups, including a control group (n = 52), a statedexpectations group (n = 49), and two groups in which respondents evaluated (before shopping) the supermarket’s past performance. Those in the stated-expectations group were asked to “indicate your expectations regarding the service that you will receive during your visit today at the supermarket.” They then provided (on a seven-point scale) ratings on nine dimensions, including (1) the politeness of the checkout employees, (2) the willingness of employees to assist shoppers, (3) the professionalism of employees, (4) the length of wait at the checkout, (5) the display arrangement, (6) the level of service, (7) overall satisfaction from the service, (8) overall satisfaction from the store visit, and (9) whether they would recommend to others to shop at the store after their visit. There were two groups in which respondents evaluated the store’s past performance before entering the store. Respondents in one of these two groups (completed past evaluation, n = 52) provided ratings on essentially the same nine items we used in the stated-expectations group. For each of the first seven (service-focused) items, these respondents were simply asked, “Please evaluate this supermarket with respect to.…” They were next asked, “What was your level of satisfaction in previous visits to this supermarket?” For the last item, respondents were asked, “Would you recommend to your friends to shop at this supermarket?” The final group (partial past evaluation, n = 48) evaluated the supermarket’s past performance (before they entered the store) on three general items: (1) the level of service, (2) satisfaction with service, and (3) overall satisfaction in previous visits to the store. Respondents in all four groups were asked the same questions after they exited the supermarket. They were first asked to indicate everything they could recall from their just-completed visit to the supermarket. Next, they evaluated the supermarket on the basis of the service they received during their shopping visit on that day, using the exact same items as those used before they entered the store

in the stated-expectations and complete-past-evaluation groups (including the three general items used in the partial-past-evaluation group). In addition, respondents indicated (on seven-point scales) (1) the degree to which they paid attention to positive or negative aspects while shopping and (2) the degree to which they expected to be interviewed after they completed the store visit. Results Again, respondents in all four conditions indicated that they did not expect to be interviewed after they completed their supermarket shopping (average ratings of 1.1–1.4 on a seven-point scale). Next, we compared the stated expectations and the (past) supermarket evaluations (in the complete-past-evaluations group) before they entered the store, which involved the same items and scales (see Figure 1). This comparison revealed that the expectations were statistically indistinguishable from the past evaluations. Specifically, for the combined service measure (for expected service, α = .95; for prior service, α = .89), the average ratings were MExpected = 5.2 and MPrior = 5.2 (F(1, 99) < 1, p > .1); for the combined satisfaction measure (for expected satisfaction, α = .86; for prior satisfaction, α = .89), the average ratings were MExpected = 5.2 and Figure 1 STUDY 4: THE EFFECTS OF STATING PREPURCHASE EXPECTATIONS AND EVALUATING PAST PERFORMANCE ON POSTPURCHASE EVALUATIONS

Stating Expectations Affects Customer Satisfaction MPrior = 5.3 (F(1, 99) < 1, p > .1); and for the recommendation measure, the average ratings were MExpected = 5.2 and MPrior = 5.1 (F(1, 99) < 1, p > .1). These findings are consistent with the assumption that a firm’s past performance provides the basis for expectations about future performance. In addition, on the comparable items, the complete and partial past evaluations of the store were statistically indistinguishable; that is, the service evaluations (t(98) = 1.1, p > .1) and the satisfaction evaluations (t(98) = 1.0, p > .1) in the two groups were similar. Furthermore, postpurchase evaluations of respondents in the complete- and partial-past-evaluation groups were similar (F(1, 197) < 1, p > .1). Consistent with the previous studies, the postpurchase evaluations in the stated-expectations group were lower than their expectations (see Figure 1): For the combined service measure, MBefore = 5.2 versus MAfter = 4.3 (pairedt(48) = 4.2, p < .01); for the combined satisfaction measure, MBefore = 5.2 versus MAfter = 4.2 (paired-t(48) = 3.7, p < .01); and for the recommendation measure, MBefore = 5.2 versus MAfter = 4.1 (paired-t(48) = 3.8, p < .01). In addition, the postvisit evaluations were lower than the corresponding evaluations in the control group: For the combined service measure, MStated = 4.3 versus MControl = 5.1 (F(1, 197) = 11.0, p < .01); for the combined satisfaction measure, MStated = 4.2 versus MControl = 5.1 (F(1, 197) = 10.9, p < .01); and for the recommendation measure, MStated = 4.1 versus MControl = 4.8 (F(1, 197) = 3.7, p < .06). Remarkably, however, although the measured (prepurchase) past evaluations were indistinguishable from the stated expectations, the former task had a different effect on postpurchase evaluations. Unlike the (negative) effect of stating expectations, providing past evaluations of the store tended to produce more positive postpurchase evaluations. In the partial-past-evaluation condition, both the service (MPartial past = 5.3) and the satisfaction (MPartial past = 5.6) postpurchase ratings were more positive than the corresponding prepurchase (past) evaluations (both Ms = 5.1; pairedt(47) = 2.0, p < .05, and paired-t(47) = 2.7, p < .05, respectively). The complete-past-evaluation condition rated postvisit service (M = 5.5) higher than (previsit) service past evaluations (M = 5.2; paired-t(51) = 1.9, p < .06). For the satisfaction (MPrevisist = 5.3 versus MPostvisist = 5.5) and recommendation (MPrevisist = 5.1 versus MPostvisist = 5.3) measures, the differences were in the same direction but did not reach statistical significance (paired-t(51) = 1.4 and 1.0, respectively, p > .1). Furthermore, postvisit evaluations in both the partial- and the complete-past-evaluation groups were higher than the postvisit evaluations of the statedexpectations group. The contrasts between the statedexpectations group and the average of the two pastevaluation groups were all statistically significant (for service, F(1, 197) = 25.1, p < .01; for satisfaction, F(1, 197) = 32.5, p < .01; and for recommendation, F(1, 197) = 12.4, p < .01). In addition, across both pastevaluation groups, satisfaction ratings were higher than in the control condition (F(1, 197) = 3.9, p < .05). For service and recommendation, these contrasts had the same sign but were not statistically significant (F(1, 197) = 1.5, p > .1, and F(1, 197) = 1.8, p > .1). Next, we examined the recall data from the shopping visit. The control and two past-performance-evaluation

171 groups were statistically indistinguishable and remembered an average of .2–.6 more positive than negative aspects, whereas the stated-expectations group remembered an average of 1.0 more negative than positive aspects. Responses to the closed-ended item about the focus of attention on “negative” (1) or “positive” (7) aspects during the visit led to a similar conclusion. The average rating in the statedexpectations group was 4.43, compared with 5.23 in the control, 5.21 in the complete-past-evaluation group, and 5.29 in the partial-past-performance evaluation group; a contrast between the average of the control and the two past-evaluation groups with the stated-expectations group was statistically significant (p < .01). Discussion There are cases in which results that confirm and replicate seemingly conflicting prior results are surprising nonetheless, and Study 4 illustrates such a case. Although the current research shows that stating expectations tends to produce lower postpurchase evaluations whereas Dholakia and Morwitz (2002) show that the evaluation of satisfaction has the opposite effect, we were skeptical whether both effects could be demonstrated in the same study. However, Study 4 shows this pattern, though it also supports the assumption that (prepurchase) past-performance evaluations are indistinguishable from the stated expectations. Process data confirm that the two tasks trigger different evaluation processes, perhaps reflecting the forward-looking perspective of stated expectations and the backward-looking focus of past-performance evaluations. Note that the robustness of the positive effect of past evaluations on subsequent evaluations needs further study, and we still do not fully understand the factors that underlie this effect. Still, the results of Study 4 suggest that evaluating past performance and stating expectations about future performance create different mind-sets. GENERAL DISCUSSION Customers’ expectations are key determinants of their consumption experiences, satisfaction, and loyalty. Therefore, knowing in advance what customers expect is critical for the success of marketing strategies. However, the current research suggests that measuring expectations can backfire, particularly if the stated expectations are still accessible during the consumption experience. Specifically, contrary to the standard assumption that expectations are either confirmed or disconfirmed on the basis of any discrepancy between expectations and actual experience, our research reveals a tendency for negative disconfirmation of stated expectations. Review of Findings and Theoretical Implications The conclusion that stating expectations before a consumption experience leads to lower evaluations of that experience appears to be robust. In the studies we described in this article, participants who first stated their expectations subsequently reported more negative evaluations of their experiences than a control group and relative to their own expectations. As the follow-up to Study 1 shows, this result occurred even in the case of a supermarket associated with negative expectations, which is an unusual situation because

172

JOURNAL OF MARKETING RESEARCH, FEBRUARY 2007

customers tend to shop at establishments they prefer. Thus, the findings of this research suggest that stated expectations tend not to be met, and explicitly stating expectations influences the likelihood that postexperience evaluations would be lower than they would have been had the participants not articulated their expectations. As we discussed previously, different theories could predict and explain the effect of stating expectations on subsequent evaluations. We rejected an account based on confirmation bias because positive stated expectations should have resulted in more positive postexperience evaluations, not more negative ones. Furthermore, there is no positivity effect associated with measuring expectations. In all likelihood, this conclusion is not the result of presenting our studies to respondents as part of a university study that is not affiliated with the store being evaluated. In particular, using the notion of persuasion knowledge (Friestad and Wright 1994), Williams, Fitzsimons, and Block (2004) show that the mere measurement effect of intent questions might actually be stronger without the mention of any sponsor. However, it is possible that consumers are less likely to perceive a persuasive intent when they are asked to state expectations (or evaluations of past performance) than when they are asked to indicate their intention to buy. These differences in perceptions of and responses to questions about intentions versus expectations can be examined in greater detail in further research. The theory that appears to account best for the observed findings, including the process measures, is based on the notion that stating expectations triggers an evaluation process in which negative aspects receive more attention and ambiguous dimensions are encoded more negatively than in the absence of salient expectations. This negative bias goes beyond the generalization that negative aspects receive more attention (e.g., Kanouse 1984) and losses loom larger than gains (Tversky and Kahneman 1991), which should apply regardless of whether expectations are stated before experience. Instead, stating expectations appears to generate a more active, online evaluation process, which exacerbates the negative bias of evaluations (e.g., Ofir and Simonson 2001). That is, after people state expectations, they are more likely to consider how actual performance compares with those expectations, and this process is characterized by an emphasis on negative aspects. The negativity enhancement account can readily explain the finding that stating expectations leads to better recall of negative aspects and poorer recall of positive aspects of the consumption experience, under the reasonable assumption that recall reflects the focus of attention while shopping. However, the finding of Study 2 that stating expectations causes customers to view the same features more negatively suggests that the impact of stating expectations goes beyond focus of attention. In particular, negativity enhancement might involve a motivational component that is triggered by stating expectations, as if customers seek negative disconfirmation of their expectations. Although the current research does not provide much insight into such a process, a possible mechanism that can be explored in further research is that customers spontaneously or unconsciously use negative disconfirmation as a means to improve marketers’ performance.

Although not inconsistent with prior research (Dholakia and Morwitz 2002), perhaps the most surprising finding we observed is that of Study 4, in which stating expectations about future performance and evaluating past performance had the opposite effects on subsequent evaluations of a shopping experience. What is particularly remarkable is that we observed these conflicting effects even though the stated expectations and past evaluations that were measured just before participants entered the supermarket were virtually identical. Still, caution must be used in drawing strong conclusions based on a single study, and, in particular, the positive effect of past-performance evaluations may not replicate. However, although the magnitude of the positive effect of evaluating past performance on subsequent evaluations was smaller than that of stating expectations and not statistically significant in all cases, it appears to be consistent with the prior findings of Dholakia and Morwitz (2002; see also Dholakia, Morwitz, and Westbrook 2005). Furthermore, we replicated the effect of past-performance evaluation in two conditions of Study 4; participants who evaluated past performance on all items and those who evaluated only three general dimensions subsequently provided more positive postpurchase evaluations than the control group. In addition, because past evaluations were approximately 5.2 on a seven-point scale, the impact of considering past performance might have been limited somewhat by a ceiling effect. That the stated expectations were numerically indistinguishable from the past-performance evaluations indicates that the valence of the accessible cognitions cannot account for the conflicting effects of these conditions on subsequent evaluations. The recall process measure we included in Study 4 shows that similar to participants in the control condition, those who evaluated past performance tended to focus more on positive aspects of their experience, whereas those who stated expectations focused on negative aspects. What underlies this divergence of evaluation processes and resulting evaluation? We have no definite answer to this question, which needs to be addressed in further research. We can speculate that the two tasks create different mindsets or primes—one that is forward looking on the to-beevaluated experience and one that is backward looking— and these conditions fundamentally change the shopping experience and/or the way consumers evaluate it. Alternatively, it might be conjectured that considering past performance promotes the assimilation of new information into the stated evaluation, whereas stating expectations triggers a process of disconfirmation that is more akin to a contrast effect. In any case, the negative effect of stating expectations compared with the positive effect of evaluating past performance indicates that mere measurement effects can have wide-ranging influences on judgments and behavior. They could reinforce ambiguous intentions (e.g., Morwitz, Johnson, and Schmittlein 1993; Sherman 1980), they could lead to more positive or more negative experience evaluations, and they might be spontaneous or involve conscious processing of intentions (e.g., Fitzsimons and Williams 2000; Williams, Fitzsimons, and Block 2004). At a more general level, similar to the construction of attitudes and preferences (e.g., Bettman, Luce, and Payne 1998; Feldman and Lynch 1988; Schwarz and Bohner 2001), the impact of

Stating Expectations Affects Customer Satisfaction mere measurement effects is contingent on the conscious and unconscious processes triggered by the act of responding to the measure at issue. The current research did not examine the long-term impact, if any, of stating expectations on subsequent behavior, such as the effect on customers’ loyalty to and frequency of purchase at the evaluated store (e.g., Chandon, Morwitz, and Reinartz 2004; Dholakia, Morwitz, and Westbrook 2005). Specifically, there are at least two issues that are of interest with respect to the long-term effects of measuring expectations and subsequent evaluations. First, it would be important to examine whether the negative effect of stating expectations on immediate experience evaluations lasts beyond a relatively short period following the experience. Second, it might be worthwhile to test whether the fact that the (negative) impact of stating expectations is measured after the consumption experience has a reinforcing effect, which strengthens and extends the long-term consequences of asking customers to state their expectations. Finally, although our focus in this article has been on the influence of stating expectations on subsequent consumer evaluations, the similarity of this effect to the impact of expecting to evaluate (Ofir and Simonson 2001) is noteworthy. Both consumers who a priori expect to provide service/ experience evaluations (e.g., of a repair service or a store) and consumers who state expectations about service/experience without expecting to offer subsequent evaluations tend to focus on negative aspects and subsequently offer more negative evaluations. The finding that these rather different tasks have a similar effect on evaluations provides tentative support to the general conclusion that any task that stimulates more online and less memory-based evaluations (e.g., of a third party, such as a service provider) promotes negative evaluations, consistent with the negativity enhancement hypothesis. However, much more research on the conditions that moderate negativity enhancement is needed, particularly given the evidence that the relationship between online and retrospective evaluations is rather complex and is influenced by various factors (see, e.g., Hastie and Park 1986; Kahneman et al. 1993). Practical Implications As articles that examine mere measurement effects have pointed out, the observed impact of measurement on subsequent behavior suggests that marketers can use market research surveys as a strategic tool to affect customer behavior. For example, Dholakia and Morwitz (2002) argue that satisfaction surveys can be used to enhance customer loyalty and purchases from the firm (e.g., opening additional bank accounts). Prior research has also demonstrated that consumption experiences are susceptible to influence by seemingly minor changes and distractions (e.g., Nowlis, Mandel, and McCabe 2004). The current research illustrates another effect of participation in surveys on customer experience; namely, the measurement of expectations can create more negative evaluations of the consumption experience and the firm about which expectations are elicited. Thus, although knowing what customers expect is important, marketers should be careful not to create bad experiences by measuring expectations in temporal proximity to the actual experience. Because many customer experiences

173 are ongoing, such as many continuous relationships between vendors and business customers or between a supermarket or a bank and its frequent customers, there may not be a good time for measuring expectations. In such situations, marketers may be better off avoiding measurement of customer expectations and relying on measures of satisfaction instead. At a more general level, customer relationship management calls for developing a more intimate knowledge of customers (e.g., Winer 2001). In turn, this prescription may suggest a more extensive use of customer surveys. However, although there are significant differences among specific types of measures used in customer research, marketers should beware of overreliance on customer surveys, such as measuring service expectations. In addition to the sometimes questionable data that such surveys produce when responses are constructed to comply with survey demands, the current research shows that stating expectations can have a detrimental impact on customers’ experiences and relationships with the firm. REFERENCES Bettman, James R., Mary Frances Luce, and John W. Payne (1998), “Constructive Consumer Choice Processes,” Journal of Consumer Research, 25 (December), 187–217. Chandon, Pierre, Vicki G. Morwitz, and Werner J. Reinartz (2004), “The Short- and Long-Term Effects of Measuring Intent to Repurchase,” Journal of Consumer Research, 31 (December), 566–72. Dholakia, Utpal M. and Vicki G. Morwitz (2002), “The Scope and Persistence of Mere-Measurement Effects: Evidence from a Field Study of Customer Satisfaction Measurement,” Journal of Consumer Research, 29 (September) 159–67. ———, ———, and Robert Westbrook (2005), “Why Does the Mere Measurement Effect Occur?” working paper, Jesse H. Jones Graduate School of Management, Rice University. Feldman, Jack M. and John G. Lynch (1988), “Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior,” Journal of Applied Psychology, 73 (3), 421–35. Fiske, Susan T. (1980), “Attention and Weight in Person Perception: The Impact of Negative and Extreme Behavior,” Journal of Personality and Social Psychology, 38 (June), 889–906. Fitzsimons, Gavan J. and Patti Williams (2000), “Asking Questions Can Change Choice Behavior: Does It Do So Automatically or Effortfully?” Journal of Experimental Psychology: Applied, 6 (September), 195–206. Friestad, Marian and Peter Wright (1994), “The Persuasion Knowledge Model: How People Cope with Persuasion Attempts,” Journal of Consumer Research, 21 (June), 1–31. Hastie, Reid and Bernadette Park (1986), “The Relationship Between Memory and Judgment Depends on Whether the Judgment Task Is Memory-Based or On-Line,” Psychological Review, 93 (3), 258–68. Kahneman, Daniel, Barbara Fredrickson, Charles Schreiber, and Donald Redelmeier (1993), “When More Pain Is Preferred to Less: Adding a Better End,” Psychological Science, 4 (6), 401–405. Kanouse, David E. (1984), “Explaining Negativity Biases in Evaluative and Choice Behavior: Theory and Research,” in Advances in Consumer Research, Vol. 11, Thomas Kinnear, ed. Provo, UT: Association for Consumer Research, 703–708. Kardes, Frank R., Chris T. Allen, and Manuel J. Pontes (1993), “Effects of Multiple Measurement Operations on Consumer Judgment: Measurement Reliability or Reactivity?” in Advances in Consumer Research, Vol. 20, Leigh McAlister and

174

JOURNAL OF MARKETING RESEARCH, FEBRUARY 2007

Michael L. Rothschild, eds. Provo, UT: Association for Consumer Research, 280–83. Klayman, Joshua and Young-won Ha (1987), “Confirmation, Disconfirmation, and Information in Hypothesis Testing,” Psychological Review, 94 (April), 211–28. Kotler, Philip (2000), Marketing Management: Analysis, Planning, Implementation, and Control, 10th ed. Upper Saddle River, NJ: Prentice Hall. Morwitz, Vicki G. and Gavan J. Fitzsimons (2004), “The MereMeasurement Effect: Why Does Measuring Intentions Change Actual Behavior?” Journal of Consumer Psychology, 14 (1–2), 64–74. ———, Eric Johnson, and David Schmittlein (1993), “Does Measuring Intent Change Behavior?” Journal of Consumer Research, 20 (June), 46–61. Nisbett, Richard E. and Lee Ross (1980), Human Inference: Strategies and Shortcomings of Social Judgment. Upper Saddle River, NJ: Prentice Hall. Nowlis, Stephan M., Naomi Mandel, and Deborah Brown McCabe (2004), “The Effect of a Delay Between Choice and Consumption on Consumption Enjoyment,” Journal of Consumer Research, 31 (3), 502–510. Ofir, Chezy and Itamar Simonson (2001), “In Search of Negative Customer Feedback: The Effect of Expecting to Evaluate on Satisfaction Evaluations,” Journal of Marketing Research, 37 (May), 170–82. Parasuraman, A., Valarie A. Zeithaml, and Leonard L. Berry (1988), “SERVQUAL: A Multiple-Item Scale for Measuring

Consumer Perceptions of Service Quality,” Journal of Retailing, 64 (1), 12–40. Petty, Richard E. and John T. Cacioppo (1984), “The Effects of Involvement on Responses to Argument Quantity and Quality: Central and Peripheral Routes to Persuasion,” Journal of Personality and Social Psychology, 46 (1), 69–81. Schwarz, Norbert and Gerd Bohner (2001), “The Construction of Attitudes,” in Blackwell Handbook of Social Psychology, Vol. 1, Abraham Tesser and Norbert Schwarz, eds. Oxford: IntraIndividual Processes, 436–57. Sherman, Steven J. (1980), “On the Self-Erasing Nature of Errors of Prediction,” Journal of Personality and Social Psychology,” 39 (August), 211–21. Tesser, Abraham (1978), “Self-Generated Attitude Change,” in Advances in Experimental Social Psychology, Vol. 11, L. Berkowitz, ed. New York: Academic Press, 290–339. Tversky, Amos and Daniel Kahneman (1991), “Loss Aversion in Risky and Riskless Choice: A Reference Dependent Model,” Quarterly Journal of Economics, 106 (November), 1039–1061. Williams, Patti, Gavan J. Fitzsimons, and Lauren G. Block (2004), “When Consumers Do Not Recognize ‘Benign’ Intention Questions as Persuasion Attempts,” Journal of Consumer Research, 31 (3), 540–50. Winer, Russell S. (2001), “A Framework for Customer Relationship Management,” California Management Review, 43 (4), 89–105.