Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Superfluous Choices and the Persistence of Preference A. V. MUTHUKRISHNAN LUC WATHIEU* * A. V. Muthukrishnan is Associate Professor of Marketing at Hong Kong University of Science and Technology, Clearwater Bay, Kowloon, Hong Kong ([email protected]). Luc Wathieu is Associate Professor at Harvard Business School, Morgan Hall, Soldiers Field, Boston MA 02163 ([email protected]). Superfluous choices are unnecessary choice steps that concern irrelevant dimensions or lead to totally predictable outcomes. They are introduced in this paper in order to study the mere effects of consumer participation. Superfluous choices have no immediate impact on the chosen option, but they strongly increase the propensity of consumers to persist with the same option in future choice occasions. Five experiments demonstrate this effect and establish an underlying mechanism that highlights the mediational role of perceived decision validity. 2 Why consumers persist (or do not) with the choices they made in the past is one of the most pressing issues in consumer research. Consumer persistence might sometimes reflect functional superiority of the object at hand, but consumers generally exhibit a systematic preference for the status quo that wants deeper understanding (Samuelson and Zeckhauser 1988). This paper introduces and isolates a new source of persistence of preference, one closely associated with the now ubiquitous interactivity that characterizes most real life choice situations. The hypothesis is that the mere fact of posing multiple intermediary acts of choice at an early stage will cause a perception of decision validity giving the chosen option a persistent advantage in subsequent choice situations. Of course, offering additional choice steps may help consumers pick a more preferred option (Simonson 2005), but it is possible to isolate a mere effect of choosing by focusing only on superfluous choice. We define superfluous choices as additional choice steps in which the decisions are inconsequential either because they concern irrelevant dimensions or because their outcomes are highly predictable. The implication is that these decisions could easily be made by the firm on behalf of the consumer. This research argues by means of a series of experiments that superfluous choice, even though it has no immediate impact on consumer decision-making, is critical for perceived decision validity and the persistence of preference. Known Sources of Persistence Prior research has identified a variety of mechanisms that can cause persistence of preference. These differ mostly in terms of the locus of the cause of persistence. To our knowledge, save for a suggestion in Wathieu et al. (2002), persistence that originates from consumer participation in a sequence of decision steps has thus far been overlooked. Persistence of preference has been accounted for primarily by consumers’ psychological propensities: loss aversion (Tversky and Kahneman 1991); regret avoidance (Bell 1982); 3 sensitivity to sunk costs (Arkes and Blumer 1985); drive for consistency (Akerlof and Dickens 1982); overconfidence (Muthukrishnan 1995); habituation (Wathieu 2004); and sheer inertia (Jeuland 1979). Persistence could also be attributed to factors that (despite their apparent irrelevance) cause consumers to develop a strong belief that the chosen brand is a dominating one. Prior research has identified factors such as the pioneering status of a brand (Carpenter and Nakamoto 1989), choice set characteristics (Huber, Payne, Puto 1982; Simonson and Tversky 1992), and unique nature of featured attributes (Carpenter, Glazer, and Nakamota 1994). Additionally, confidence in the belief that the chosen brand was an absolute best is affected by factors such as the amount of information available about the option (Oskamp 1965; Peterson and Pitz 1988), user experience (Muthukrishnan 1995; Muthukrishnan and Kardes 2001), and degree of conflict among alternatives in the choice set (Dhar 1997; Luce, Payne and Bettman 1999). Superfluous Choice, Perceived Decision Validity, and Choice Persistence Prior research has highlighted consumers’ preference for choice versus no choice. In particular, different streams of research have shown that a majority of people asked to choose between a single option and set of options that does not contain anything superior to the single option prefer the multiple options set (Beattie, Baron, Hershey, and Spranca 1994; Langer 1972; Suzuki 2000). Similar results were obtained even when the option in isolation absolutely dominated the multiple options set (Bown, Read, and Summers 2003). Beattie et al. (1994) suggested that a rationale to prefer more choices is to obtain additional information on each option. In addition, opting for more decision-making opportunities might delay the final choice somewhat and thereby defer the resolution of uncertainty, even when uncertainty is irrelevant for the decision (Bastardi and Shafir 1998). 4 The present research focuses not on whether consumers prefer choice over no-choice, but on the consequences of superfluous choices in terms of subsequent decisions. To be truly superfluous choice steps should have no bearing on choice and be able to be removed or added without immediately affecting choice behavior. What, then, might be the link between superfluous choice and future choice persistence? Our hypothesis rests on the assumption that irrelevant choice steps can affect decision makers’ cognitions and feelings independently of their effect on choice in the following ways. First, going through superfluous choice steps can lead consumers to believe that their decisions are carefully considered or thoughtful. Just as frequency heuristics have been shown to play an important role in probability learning (Estes 1976) and consumer choices (Alba and Marmorstein 1987), so, under superfluous choices, consumers might equate the number of decision steps with the degree of thought and consideration that went into the decisions. Second, superfluous choices can strengthen confidence in the belief that a carefully considered decision has been made. Prior research has shown that factors that enhance fluency with decision-making can boost confidence even when the accuracy of a decision is not improved (Gill, Swann, and Silvera 1998). In addition, superfluous choice steps might act as learning progress cues, even though objectively no learning occurs. Einhorn and Hogarth (1978) note that judgmental confidence builds over time in the presence of such uninformative learning cues and the resulting illusion of validity has been shown to persist even when its illusory character is recognized (Einhorn and Hogarth 1978; Kahneman and Tversky 1973). In our conceptualization confidence in a decision and the belief that the decision has been well thought out are both consequences of superfluous choice and together contribute to a construct of perceived choice validity (see Figure 1). 5 Perceived choice validity can, in turn, influence the extent to which a person persists with a previously chosen option. In addition, perceived validity may also influence the extent to which a person is satisfied with the decision process. Satisfaction with the decision and a feeling of rejoicing for having chosen the best alternative may serve as an anchor for subsequent decisions. Thus, we acknowledge that decision satisfaction might independently influence the extent to which a person persists with a chosen alternative. Consequently, in Figure 1, there is a separate link between satisfaction and choice persistence. We summarize the foregoing discussion in the following hypotheses. H1: The presence of superfluous choices increases the likelihood of choosing a previously chosen option. H2: The presence of superfluous choices enhances the degree of satisfaction with the decision process. H3: The effects of superfluous choices proposed in H1 and H2 are mediated by perceived validity in choice. These three hypotheses were tested in five experiments in which we employed different manipulations of superfluous choices and measured choice persistence via different means. Note that if persistence of preference is compared in terms of an option presented in isolation and the same option presented in a choice set with one or more inferior options (as in Suzuki 2000 and in some conditions of Bown, Read, and Summers 2003) an alternative explanation might be proposed based on cognitive dissonance theory (Aronson and Mills 1959), that is, the desirability of the alternative goes up because it has been chosen voluntarily. Although our experimental paradigm is quite different, as we explain below, we still measured liking for/desirability of the chosen option across conditions to rule out this alternative theoretical account. 6 EXPERIMENT 1 This experiment investigated the effects of adding a superfluous choice step before the selection of a preferred alternative. The only factor manipulated was whether participants were assigned or permitted to choose a choice set, the possible sets differing only in terms of a valueless attribute. In the category of rewritable compact disks the attribute described as “number of colors of CD boxes” (e.g., in a package of six CDs) was identified in a pretest as the least important (the most important attributes being disk capacity, price, and internal buffer). A separate pretest (n = 30) found no difference in the number of people who preferred four, five, or six colors of CD boxes. These attribute levels were consequently used to provide superfluous choice in the experiment. Two days before the experiment the respondents were asked (among other questions, as part of a class assignment) whether the number of different colors of CD boxes in a pack of CDs mattered. Two of the ninety-three participants responded that the number of colors mattered. Data obtained from these two respondents were excluded from the final analysis of experiment 1 even though the results were not affected by including them. Ninety-one respondents participated in the experiment in exchange for course credit. They were told that CD packages were classified into three sets, set A featuring four, set B five, and set C six different colors of CD boxes. Each set included a more expensive X brand (800 MB disk capacity and a 2 MB internal buffer) and less expensive Y brand (700 MB disk capacity and a 1.5 MB internal buffer). Dollar prices were provided. Respondents in the Choice condition were asked to first choose a set from among A, B, and C. In the other (Assigned) condition, respondents were told which set was assigned to them. In the latter condition, each set (A, B, and C) was assigned to one-third of the respondents. 7 Participants in both conditions were then asked to choose between the more expensive (relatively higher quality) and less expensive (relatively lesser quality) brand. After a brand was chosen, participants were asked four questions (each answered using an 11-point scale) to assess (1) how confident they were in the brand choice (Not at all confident – Extremely confident), (2) how considered or well thought out they felt that choice to be (Not at all considered/thought-out – Well considered/thought-out), (3) how satisfied they were with the decision making process (Not at all satisfied – Highly satisfied), and (4) how much they liked the chosen alternative (Dislike it a lot – Like it a lot). After attending a two-hour class in an unrelated discipline the participants were asked to assume that three days after they had bought it they had lost their CD pack and, needing to work with CDs for their class assignments, had visited the local bookstore and discovered a new brand Z (750 MB disk capacity and a 1.75 MB internal buffer) priced between the more expensive X and less expensive Y brands. They were then asked to recall the label of the brand they had chosen earlier and to choose between the new brand and the previously chosen brand. Results As expected, neither the Assigned nor Choice condition exhibited any difference in respondents’ choice patterns across the A, B, or C set. The data were consequently aggregated across these sets for each experimental group and submitted to a binary logit analysis. The likelihood of choosing the brand that had originally been chosen was the first dependent variable. The proportion of respondents who chose the previously chosen brand was higher in the Choice condition (34/46 or 73.9%) than in the Assigned condition (21/45 or 46.7 %; Wald χ2 = 6.83, p < .01), indicating that, as hypothesized, superfluous choice engendered greater persistence with the chosen brand. 8 The conditions also differed in terms of confidence in choice (6.13 in the Assigned condition vs. 7.22 in the Choice condition; F(1, 89) = 10.3, p < .01) and perceived consideration (5.69 in the Assigned condition and 6.52 in the Choice condition; F(1, 89) = 6.41, p < .02). These dimensions being correlated (ρ = 0.67), we took the average of the two scales as a measure of the construct of perceived validity, in conformity with the theoretical framework. The two conditions varied significantly with respect to this measure (MAssigned = 5.91 and MChoice = 6.87; F(1, 89) = 10.06, p < .01). The superfluous choice manipulation also had a significant effect on degree of satisfaction with the choice process [MAssigned = 5.69 and MChoice = 6.83; F(1, 89) = 5.66, p < .02]. As predicted, these effects were not related to variations in degree of liking the chosen brand, which was identical across conditions (7.42 in the Assigned condition, 7.23 in the Choice condition, F < 1). Further confirming the superfluous nature of the added choice step, the experimental manipulation had no significant impact on the proportion of respondents who chose the more expensive brand (initial share of X was 21/45 in the Assigned condition and 26/46 in the Choice condition). The initial selection (X vs. Y) had no significant impact on the likelihood to persist with the previously chosen option. Mediational Tests To test whether persistent preference for the chosen brand and satisfaction with the decision process are mediated by the perceived validity of the choice, as hypothesized, we examined the three criteria proposed by Baron and Kenny (1986). First, the independent variable should have a significant effect on the dependent variable. Second, the independent variable should have a significant effect on the presumed mediator. Third, when both the independent variable and the mediator are included in the model the mediator should have a significant effect on the dependent variable but the effect of the independent variable should 9 be considerably reduced. The results reported thus far are evidence that the first two conditions are met. To verify the third condition we performed the following analysis. The second choice occasion was coded 0 if the new brand was chosen and 1 if the previously chosen brand was selected and the scores were submitted to a logistic regression analysis that included, with the independent variable superfluous choice (Assigned vs. Choice condition), perceived validity as a covariate. Including the covariate caused the previously significant effect of the superfluous choice variable to be non-significant (Wald χ2 = 1.97, p > .15). The effect of the perceived validity covariate was significant (Wald χ2 = 17.47, p < 0.001). Similarly, including perceived validity reduced the previously significant effect of superfluous choice on satisfaction with the decision process to non-significance (F(1, 88) = 0.72, p > 0.39). The effect of the perceived validity covariate was significant (F(1, 88 = 30.17, p < 0.001). To further validate the foregoing interpretation a small test (n = 24, within-subject analysis) was conducted to establish that adding superfluous choice caused respondents to perceive their choices to be more considered. Each participant read three variations of the stimulus used in the experiment. All three variations involved all three sets—A, B, and C— each described as containing one more expensive and one less expensive brand of CD. In the first variation participants were told simply that they would be assigned set A, B, or C. In the second variation participants were asked to select a set, but no explanation of the letter classification was given. In the third variation participants were told that the A, B, C classification reflected the number of different colors of CD boxes (four, five, and six, respectively) in the packages and asked to select a set. After each variation subjects were asked to rate on an 11-point scale (Not at all – Extremely) “to what extent the decision process offered you an opportunity to choose the best CD available.” Each subject rated all 10 three variations. Not surprisingly, respondents rated the first variation (in which the choice set was assigned) as offering the least opportunity (2.58). The second variation, which accommodated choosing among A, B, and C without explanation of the basis for that classification, was perceived to enhance the level of opportunity (3.79; F(1, 23) = 67.4, p < .001) and the variation involving superfluous choice related to the CD box color attribute to offer the greatest level of perceived opportunity (5.17; F(1, 23) = 66.8). This finding reinforces the notion that consumers interpret superfluous choice (above and beyond a mere control opportunity) to be a source of perceived choice validity. Discussion The results of the first experiment suggest that introducing superfluous choices along unimportant dimensions can lead consumers to perceive their choices to hold greater validity (i.e., to have greater confidence in choices they perceive to be well thought out). This perception, in turn, influences both the experience of satisfaction with the decision process (even in the absence of an impact on liking the chosen alternative) and the degree to which respondents persist in their initial choice. These effects emphasize the role of superfluous participation, which otherwise has no detectable impact on what consumers choose in the first period or on the degree of satisfaction with what they have chosen (as opposed to the process by which they have chosen it). EXPERIMENT 2 This experiment demonstrates that confidence and brand persistence are enhanced when consumers are permitted to control, in a preliminary step, the size of the choice set they will face even when their response is entirely predictable (and thus superfluous). The experiment thus generalizes the findings of experiment 1 to contexts in which (1) superfluous choice is logically unrelated to the characteristics of individual choice alternatives (not simply 11 subjectively unrelated as in experiment 1), and (2) superfluous choice corresponds to a fully predictable response (not simply to a state of indifference across responses). Ninety student participants were told that the approximately 200 different brands of general-purpose scissors available on the market could be classified into two broad categories—high quality brands priced around US$20 and ordinary quality brands priced around US$8—with little within-category variation in price or quality. Various brands might, however, vary product color. Participants were told that for the sake of simplicity the market had been segmented into 100 sets, each containing one high quality and one ordinary scissors. Half of the participants (the Assigned condition) were told that they could access two sets (i.e., four scissors) before making their final choice, the other half (the Choice condition) given the choice of examining one set (i.e., two scissors) or two sets (i.e., four scissors). As expected, all participants in the latter condition chose to examine two sets. A superfluous choice step was thus added in the sense that all participants had access to four scissors whether or not they had exercised their free will. Participants in both conditions were subsequently asked whether they would choose a more expensive high quality, or a less expensive ordinary quality, scissors from among the brands offered independent of the color they would ultimately select. Participants’ confidence in their responses was obtained on an 11-point scale (Not at all confident – Extremely confident). Thirty minutes later (after performing a number of unrelated tasks) the respondents were asked to assume they had visited a neighborhood stationery store that stocked 20 of the 200 brands and indicate the likelihood that they would find a brand that they would prefer over the brand chosen earlier from the four-scissors set. The rating was obtained on an 11-point scale (Most likely – Least likely), a higher rating signaling greater choice persistence. 12 Results and Discussion The conditions differed with respect to confidence (6.19 in the Assigned condition vs. 7.07 in the Choice condition; F (1, 88) = 3.99, p < .05) and in terms of the estimated likelihood of finding a better brand in the stationery store (5.43 in the Assigned condition vs. 6.50 in the Choice condition; F (1, 88) = 5.94, p < .02).1 The experiment produced significant shifts in confidence (a key aspect of perceived validity) and preference persistence based solely on having selected a four-scissors over a two-scissors choice set instead of simply being assigned a four-scissors choice set. Because the experiment included no product-relevant basis for choice in the superfluous choice step the effects obtained can be attributed exclusively to superfluous determination of choice set size. EXPERIMENT 3 This experiment, as does experiment 2, manipulates the presence of superfluous choice by asking consumers to make a choice that (1) has a perfectly predictable outcome, and (2) leaves absolutely no detectable stigma on the alternative subsequently tested for persistence. The experiment also tests for mediational processes in the manner of experiment 1. The seventy-two participants to this study were told that a market research company that wants to understand consumer preferences for several attributes of MP3 players will ask them to make a series of choices among models and reward their participation with a free gift of a video album of singer Bruce Springsteen recorded live in concert. Half of the participants were told that they would receive the video album in a DVD format, the other half asked to choose between DVD and VHS cassette formats. Given that, as anticipated, all but one respondent opted for the DVD format the only difference across our two conditions was a 1 There was no significant impact of the experimental condition on the proportion of respondents choosing high quality scissors and the likelihood to persist with the initial choice appeared to be independent of the initially chosen quality. 13 superfluous choice step that had no bearing on the choice set and its components (MP3 players). Subjects were then asked to choose between two MP3 players (branded A and B) that varied in terms of playing time, juke box capacity, and price but not in terms of recording time. The enhanced brand and its label (A or B) were counterbalanced and its price was 20% greater than that of the other brand. As in experiment 1, after choosing between the two brands participants recorded on 11-point scales their confidence that the brand selected was the best alternative, belief about how considered or thoughtful the choice was, satisfaction with the decision process, and degree that they liked the brand they had selected. After attending an unrelated two-hour class the participants were given the following story: Assume that a month after you bought the MP3 player your uncle, aunt, and cousin visit your home. Your cousin likes the player very much. Generous person that you are, you give the MP3 player as a gift to your cousin. In return your uncle offers you a gift in the form of cash sufficient to buy a new player of the kind that you examined last time. Because you are used to spending hours listening to music you would like to buy a player soon. On reaching the store you realize that the player you bought last time and another player, C, are the only ones available. Information for C is provided below. If you want you can ask the experimenter for information about the attributes of the player you bought last time. In terms of the attributes of playing time, juke box capacity, and price brand C was portrayed as being between brands A and B. Subjects were asked to choose between brand C and the brand they had chosen earlier. The proportion of participants that chose the brand they had chosen earlier was the main dependent variable. Results As in experiment 1, we coded subjects’ responses as non-persistent and persistent depending on their choice in the presence of brand C. These data were submitted to a binary 14 logit model with the presence of superfluous choice regarding the Bruce Springsteen DVD (Assigned vs. Choice conditions) as the independent variable. The conditions varied significantly in terms of persistence. Whereas only 17 of 35 (48.6%) in the Assigned condition chose the same brand they had chosen earlier, 27 of 37 (73%) in the Choice condition exhibited persistent preference for the previously chosen brand. The conditions also varied in terms of confidence that the chosen brand was the best in the set (5.83 in the Assigned vs. 7.27 in the Choice condition; F(1, 70) = 14.2, p < .01) and in the belief that the choice reflected a considered decision (5.82 in the Assigned vs. 7.38 in the Choice condition; F(1, 70) = 15.61, p < .001). These variables, having a significant positive correlation (ρ= 0.64, p < .001), were averaged to obtain a composite score of perceived choice validity. The effect of superfluous choice was significant on perceived choice validity (F(1, 70) = 19.09, p < .001; MAssigned = 5.83 and MChoice = 7.32). The conditions also varied with respect to the dependent variable of satisfaction with the decision process (MAssigned = = 5.77 and MChoice = 6.84; F(1, 70) = 10.44, p < .01). There was no difference between the conditions with respect to liking the chosen option (7.89 in the Assigned, 7.78 in the Choice condition, F < 1) or in the propensity to choose the more expensive MP3 player (.63 in the Assigned condition, .59 in the Choice condition). Mediational Analysis We conducted Baron and Kenny tests to examine the mediational role of perceived choice validity in the effects of superfluous choice on persistence of preference and satisfaction with the decision process. When perceived choice validity was included as a covariate in a logistic regression model the previously significant effect of superfluous choice opportunity became non-significant (Wald χ2 = 1.05, p > .3). The effect of the covariate was significant (Wald chi-square = 4.05, p < .05). Similarly, including perceived validity as a covariate in an ANCOVA model caused the previously significant effect of superfluous choice on 15 satisfaction with the decision process to become non-significant (F(1, 69) = 1.34, p < .25). The effect of the perceived validity covariate was significant (F(1, 69) = 20.09, p < .001). Discussion Experiment 3 not only demonstrated the intriguing effects of superfluous choice, but also provided evidence of the underlying mechanism at play. The only manipulation was having participants first make a predictable choice about a premium unrelated to the main decision. This choice nevertheless had a strong effect on participants’ propensity to persist with the chosen alternative on a later choice occasion, despite the absence of a noticeable effect of superfluous choice on the first period decision or on the degree of liking the chosen alternative. The superfluous choice step gave rise to a perception of choice validity and resulting satisfaction with the decision process. This perception of choice validity (measured as a combination of the confidence and considered decision ratings) might thus explain the significant persistence of preference effect. EXPERIMENT 4 This experiment differed from experiment 3 only in terms of the dependent variable, namely, substituting for the new MP3 player on the second choice occasion a price cut for the non-chosen brand. One relevant manifestation of persistence of preference is a consumer’s repetition of an initial choice in the presence of incentives to switch posted by competing brands. As in experiment 3 a two-hour delay preceded the eighty-two respondents’ choices of MP3 players. Two versions of the stimuli were used on the second choice occasion. For subjects who chose brand A in the first round brand B was portrayed as being offered at a 20% price reduction and brand A at the original price. For subjects who chose brand B in the first round brand A was portrayed as being offered at a 20% price reduction and brand B at the original price. Depending on their second period choice respondents were classified as 16 loyal or non-loyal. In addition to the new measure of brand persistence, we also measured confidence, belief that the choice reflected a considered decision, satisfaction with the decision process, and liking the chosen alternative. Results The results obtained from a binary logit analysis revealed superfluous choices to have a significant impact on loyalty and persistence of preference. Whereas only 17 of 40 (48.8%) in the Assigned condition selected the same brand on both occasions, 30 of 42 (71.4%) in the Choice condition did so (Wald χ2 = 5.88, p < .01). This highly persistent pattern of preference in the presence of superfluous choice obtained despite the brand previously not chosen being offered at a 20% discount. The conditions differed significantly with respect to confidence that the best brand had been chosen (MAssigned = 5.45 and MChoice = 7.40; F(1, 80) = 26.99, p < .001) and the belief that the choices reflected considered decisions (MAssigned = 5.70 and MChoice = 7.62; F(1, 80) = 29.26, p < .001). The correlation between the variables being significant (ρ = 0.73, p < .001), we constructed perceived choice validity as a linear combination of these two variables. There was a significant effect of superfluous choice on perceived choice validity (5.58 in the Assigned condition vs. 7.51 in the Choice condition; F(1, 80) = 34.5, p < .001). Mediational Tests We found evidence for perceived choice validity mediating both choice persistence and satisfaction with the decision process. Including choice validity as a covariate in a logistic regression model reduced the previously significant effect of superfluous choice on choice persistence to non-significance (Wald χ2 = 0.37, p > .5). The effect of the mediator on choice persistence was significant (Wald χ2 = 9.09, p < .01). Likewise, including perceived choice validity in an ANCOVA model eliminated the previously significant effect of superfluous 17 choice on satisfaction with the decision process (F < 1). The effect of the covariate was significant (F(1, 79) = 9.93, p < .01). Discussion These results of experiment 4 extend and reinforce those of experiment 3, despite the introduction of a more stringent measure of persistence in terms of loyalty when confronted with a discount on the competing offering. As in the earlier experiments, superfluous choice generated persistence of preference and satisfaction with the decision process, and both effects were mediated by a perception of validity. EXPERIMENT 5 This experiment investigated the effects of superfluous choice when such choice is implicit in the sense that it is not explicitly presented to consumers as a separate choice step. One limitation of the previous experiments is that participants who received explicit additional instructions to make a (superfluous) choice might have interpreted the stimulus as a signal of caring or respect for them as free-willed decision makers. Such an interpretation would leave the door open for an alternative explanation of the results in terms of interpersonal comfort or trust (instead of a perception of validity as suggested here). Superfluous choice in this experiment took the form of an editing stage. Consumers in the superfluous choice condition were offered a choice set in which each alternative was coupled with a clearly dominated alternative. It is hypothesized that in the process of dismissing these alternatives consumers build confidence in the process of choice making, which should translate into persistence of preference on later choice occasions. Whereas the effect of dominated alternatives on choice is well documented, this experiment attempts to create a situation in which, absent such a direct effect, an effect in terms of perception of validity and persistence of preference can be isolated. 18 Ninety-eight participants were assigned to one of the following experimental conditions: three alternatives; six alternatives with superfluous options. Subjects in the first condition were exposed to three MP3 players described in terms of four attributes: playing time (hours); recording time (hours); juke box space (GB); and battery life (hours). The brands did not vary in terms of recording time. Brand A exceeded the other brands in terms of battery life, brand B in terms of juke box space, and brand C in terms of playing time. The choice set thus contained three alternatives on the efficient frontier. The choice set in the second condition included six brands. Brands A, B, and C were described in the same manner as in the first condition, brands D, E, and F as being dominated by brands A, B, and C, respectively. Participants first made their choices within one of these choice sets. Confidence in having chosen the best alternative was measured on an 11-point scale (Not at all confident – Extremely confident). After thirty minutes of performing unrelated tasks the participants were asked to assume that three days after they bought it they lost their MP3 player and to decide whether they would buy the same player they had bought before or buy another (either from among the ones they had examined earlier or a new player from another store). Results and Discussion Choice persistence was greater in the superfluous choice condition (31/48 or 0.67) than in the three-alternative condition (20/50 or 0.4; Wald χ2 = 5.80, p < .02). Similarly, confidence in the initial choice condition was greater when the choice set contained six alternatives (MSix = 8.31 vs. MThree = 7.16; F(1, 96) = 11.27, p <.01). The results of this experiment demonstrate that an opportunity to screen out dominated alternatives has the same general effect on persistence of preference as explicitly instructed superfluous choice steps. It should be noted that unlike in the asymmetric dominance effect paradigm (Huber, Payne, and Puto 1982; Simonson and Tversky 1992), there was absolutely no difference between the three-alternative condition and the six-alternative condition in 19 terms of the choice share of any single brand. However, an opportunity to screen out alternatives in the six-alternative condition enhances the perception of validity in choice and thus causes persistence with the previously chosen brand. GENERAL DISCUSSION In five experiments we demonstrated that superfluous choices influence persistent preference for a previously chosen option. Superfluous choices are choice steps that could be removed without affecting the set of available options and without noticeable impact on what consumers end up selecting. These choice steps nevertheless impart a perception of decision validity that increases the likelihood that a consumer will persist with the same choice on future choice occasions. Our experiments introduced various types of superfluous choice: a preliminary choice between two levels of an unimportant attribute (experiment 1); a totally predictable choice between two choice set size levels (experiment 2); a totally predictable choice between two premiums obtained independently of the focal choice made by the consumer (experiments 3 and 4); and a trivial task of editing out dominated alternatives (experiment 5). The main dependent variable was in each case the propensity to persist with the previously chosen alternative in the context of a later choice occasion, usually in the presence of an attractive new opportunity. Experiment 4, for example, included on the second choice occasion an incentive for brand switching in the form of a 20% price discount on the nonchosen option. We also measured in several of the experiments the effect of superfluous choice on a number of consumer perceptions, namely, confidence in the choice, how considered or well thought out the decision was, satisfaction with the decision process, and how much the consumer liked the chosen option. The first two measures were highly correlated and combined into a more general measure of perceived validity that was found to mediate the effect of superfluous choice on persistence of preference and satisfaction with the 20 decision process. That liking the chosen item did not vary with the superfluous choice manipulation reinforces our theory that superfluous choice affects future decisions only through perceived validity. We should like to consider whether mechanisms isolated by other research streams that compared experimental conditions in which an element of the choice situation was chosen versus assigned might be at work in our experiments. According to the cognitive dissonance hypothesis, for example, the desirability of an option increases when it is freely chosen versus imposed. This mechanism might explain persistence of preference in some settings, but is not applicable to our experiments because the option adopted was always chosen. Our manipulation of assigned versus chosen involved only an element of the decision context, not the selection of the preferred item from among other alternatives. Even though the cognitive dissonance explanation is not applicable to our experiments we nevertheless obtained the target brands’ desirability ratings from the participants. In none of the experiments did we find an effect of superfluous choice on liking, or the desirability of, the target brand. The illusion of control phenomenon proposed by Langer (1975) suggests that respondents who choose an option (compared to those who are assigned the same option) might exhibit a stronger preference for that option. Our experiments do not test the difference between the conditions under which the target brand is chosen versus assigned. Another key difference is that the illusion of control phenomenon was proposed for situations in which chance and skill are equated. There is no shift of control when superfluous choice is offered because by definition the chosen alternative remains predictably unchanged, that is, the heightened activity or participation generated by superfluous choice in the selection of an alternative does not in any way determine the outcome. The research stream that advanced aversion uncertainty resolution as a key explanation for the preference for choice in a choice versus no-choice task (Bown et al. 2003; Suzuki 21 2000; also see Bastardi and Shafir 1998 regarding a similar phenomenon) has no relevance in our context inasmuch as all five experiments involve choice. Iyengar and Lepper (2001) examined choice behavior and satisfaction as it relates to the impact of “more choice,” that is, to the size of the choice set that is encountered. In all but the last of our experiments the size of the choice set was common across experimental conditions. Moreover, we interpret “more choice” in terms of increasing the number of preliminary choice steps while maintaining everything else constant. The literatures on the asymmetric dominance effect (Huber, Payne, Puto 1982; Tversky and Shaffir 1992) and compromise effect (Simonson 1989) maintain that a change in the choice set affects preference because different contexts afford opportunities to use different coherent reasons for choosing one brand over another. No such reason-based mechanism can underlie our effects because in all but experiment 5 our experimental conditions do not affect the choice set. Even in experiment 5 our manipulation (featuring an additional dominated alternative coupled with each possible option) did not impact choice initially. Therefore, in spite of their apparent similarity to the asymmetric dominance effect, the conclusions offered by experiment 5 are unique and hitherto unreported. Equating superfluous choices with greater perceived validity is thus a more adequate and parsimonious explanation of our results than any advanced in the theories discussed above. Ultimately, this research highlights a mere effect of consumer participation, whereby participation entails beliefs of validity about the choice process that become persistently associated with the chosen alternative. Future research might identify new moderators that can accentuate or dilute the effects of superfluous choice on persistence of preference. For instance, consumers’ assumptions about the motives for superfluous choices or about the predictability of their own preferences relative to those of others might interfere with our effects. It might be useful as well to further 22 investigate the origin of the superfluous choice effect (e.g., whether it occurs at the point of encoding in the initial context or at the point of retrieval when a new choice opportunity arises). Real world implications should be far-reaching inasmuch as the superfluous choice effect calls into question the frequent assumption that suppliers should use what they already know about consumers to create finely targeted, conveniently channeled offers. If there is a mere effect of participation and if superfluous choice generates greater perceived validity, then interactions that require more choice steps than necessary might prove unexpectedly useful. 23 REFERENCES Akerlof, George A. and William T. Dickens (1982), “The Economic Consequences of Cognitive Dissonance,” American Economic Review, 72 (June), 307-19. Alba Joseph W. and Howard Marmorstein (1987), “The Effects of Frequency Knowledge of Consumer Decision Making,” Journal of Consumer Research, 14 (June), 14-25. Arkes, Hal R. and Catherine Blumer (1985), “The Psychology of Sunk Cost,” Organizational Behavior and Human Decision Processes, 35, 124-40. Aronson, Elliot and J. Mills (1959), “The Effect of Severity of Initiation on Liking for a Group,” Journal of Abnormal and Social Psychology, 59, 177-81. Baron, Reuben and David Kenny (1986), “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,” Journal of Personality and Social Psychology, 51 (December), 1173-82. Bastardi, Anthony and Eldar Shafir (1998), “On the Pursuit and Misuse of Useless Information,” Journal of Personality and Social Psychology, 75 (1), 19-32. Beattie, Jane, Jonathan Baron, John Hershey, and Mark Spranca (1994), “Psychological Determinants of Decision Attitude,” Journal of Behavioral Decision Making, 7 (2), 12429. Bell, David (1982), “Regret in Decision Making under Uncertainty,” Operations Research, 30 (5), 961-81. Bown, Nicola J., Daniel Read, and Barbara Summers (2003), “The Lure of Choice,” Journal of Behavioral Decision Making, 16, 297-308. Carpenter, Gregory and Kent Nakamoto (1989), “Consumer Preference Formation and Pioneering Advantage,” Journal of Marketing Research, 25 (August), 285-98. 24 Carpenter, Gregory, Rashi Glazer, and Kent Nakamoto (1994), “Meaningful Brands from Meaningless Differentiation,” Journal of Marketing Research, 31 (August), 339-51. Dhar, Ravi (1997), “Consumer Preference for a No-Choice Option,” Journal of Consumer Research, 24 (September), 215-31. Einhorn, Hillel J. and Robin Hogarth (1978), “Confidence in Judgment: Persistence of the Illusion of Validity,” Psychological Review, 85 (5), 395-416. Elster, Jon and George Lowenstein (1992), “Utility from Memory and Anticipation,” in Choice over Time, ed. G. Lowenstein and J. Elster, New York: Russell Sage Foundation, 213-34. Estes, W. K. (1976), “The Cognitive Side of Probability Learning,” Psychological Review, 83 (1), 37-64. Gill, Michael J., William B. Swann, Jr., and David H. Silvera (1998), "On the Genesis of Confidence," Journal of Personality and Social Psychology, 75 (5), 1101-14. Huber, Joel, John Payne, and Christopher Puto (1982), “Adding Asymmetrically Dominated Alternatives: Violations of Regularity and the Similarity Hypothesis,” Journal of Consumer Research, 10 (June), 31-44. Iyengar, Sheena S. and Mark R. Lepper (2000), “When Choice is Demotivating: Can One Desire Too Much of a Good Thing?” Journal of Personality and Social Psychology, 76, 9951006. Jeuland, Abel (1979), “Brand Choice Inertia as One Aspect of the Notion of BrandLoyalty,” Management Science, 25 (7), 671-82. Kahneman, Daniel and Amos Tversky (1973), “On the Psychology of Prediction,” Psychological Review, 80, 251-73. 25 Langer, Ellen J. (1975), “The Illusion of Control,” Journal of Personality and Social Psychology, 32 (2), 311-28. Luce, Mary Frances, John W. Payne, and James R. Bettman (1999), “Emotional Trade-off Difficulty and Choice,” Journal of Marketing Research, 36 (May), 143-59. Muthukrishnan, A. V. (1995), “Decision Ambiguity and Incumbent Brand Advantage,” Journal of Consumer Research, 22 (June), 98-109. Muthukrishnan, A. V. and Frank R. Kardes (2001), “Persistent Preferences for Product Attributes: The Effects of Initial Choice Context and Uninformative Experience,” Journal of Consumer Research, 28 (1), 89-104. Oskamp, Stuart (1965), "Overconfidence in Case Study Judgments," Journal of Consulting Psychology, 19 (3), 261-65. Peterson, Dane K. and Gordon F. Pitz (1988), "Confidence, Uncertainty and the Use of Information," Journal of Experimental Psychology: Learning, Memory and Cognition, 14 (January), 85-92. Samuelson, William and Richard Zeckhauser (1988), “Status Quo Bias in Decision Making,” Journal of Risk and Uncertainty, 1 (1), 7-59. Simonson, Itamar (1989), “Choice Based on Reasons: The Case of Attraction and Compromise Effects,” Journal of Consumer Research, 16 (September), 158-74. _________ (2005), “Determinants of Customers’ Responses to Customized Offers: Conceptual Framework and Research Propositions,” Journal of Marketing, 69 (January), 32-45. Simonson, Itamar and Amos Tversky (1992), “Choice in Context: Tradeoff Contrast and Extremeness Aversion,” Journal of Marketing Research, 29 (August), 281-95. 26 Suzuki, Shuji (2000), “Choice between Single-response and Multichoice Tasks in Humans,” The Psychological Record, 50 (1), 105-15. Tversky, Amos and Daniel Kahneman (1991), "Loss Aversion in Riskless Choice: A Reference Dependent Model," Quarterly Journal of Economics, 106 (November), 1039-61. Tversky, Amos and Eldar Shafir (1992), “Choice under Conflict,” Psychological Science, 6 (November), 358-61. Wathieu, Luc (2004), “Consumer Habituation,” Management Science, 50 (May), 587-96. Wathieu, Luc, Lyle Brenner, Ziv Carmon, Amitava Chattopadhyay, Klaus Wertenbroch, Aimee Drolet, John Gourville, A. V. Muthukrishnan, Nathan Novemsky, Rebecca K. Ratner, and George Wu (2002), “Consumer Control and Empowerment: A Primer,” Marketing Letters, 13 (3), 295-303. 27 FIGURE 1: THEORETICAL FRAMEWORK Belief about having made a considered decision Superfluous choice steps e.g., • concerning irrelevant dimensions • highly predictable preferences No impact on choice Confidence in the attractiveness of the chosen option Satisfaction with the decision process Perceived choice validity Preference persistence across choice occasions 28