Now or Never: Effects of Limited Purchase Opportunities on Patterns of Regret over Time

LISA J. ABENDROTH KRISTIN DIEHL*

Prior research has demonstrated that actions are regretted more than inactions in the short term. We show that, in limited purchase opportunitiessituations where the purchase decision cannot be reversednot purchasing (inaction) is seen as a loss and is associated with greater short-term regret than purchasing, reversing the omission bias. With respect to long-term regret, we use coping and availability mechanisms to suggest that, contrary to prior ndings, inaction (nonpurchase) regrets decrease over time. We also argue that action (purchase) regrets should increase over time, but only when long-term utility is low. We support our predictions with a eld study and two laboratory experiments.

rior research in psychology has demonstrated that actions are regretted more than inactions in the short run (Kahneman and Tversky 1982; Landman 1987) and that, over time, action regrets tend to decrease while inaction regrets increase (Gilovich and Medvec 1995). In a consumer context, making a purchase involves action while deciding not to purchase constitutes inaction. However, this context may lead to different patterns of regret. We are interested in situations where not buying regrets are more intense initially and decrease over time while buying regrets are minimal initially and increase over time, reversing prior ndings. We predict that this pattern exists in limited purchase opportunitiesnow or never situations where, once the window of opportunity has passed, the purchase can no longer be made. Imagine driving several hours to a destination retailer such as Ikea. In addition to your intended sofa purchase, you consider buying some home accessories. These items represent limited purchase opportunities as you are un-

*Lisa J. Abendroth is assistant professor of marketing at Boston University, School of Management, Boston, MA 02215 (lisaa@bu.edu). Kristin Diehl is assistant professor of marketing at the University of Southern California, Marshall School of Business, Los Angeles, CA 90098-0443 (kdiehl@marshall.usc.edu). The authors thank the editor, the associate editor, the reviewers, James Bettman, Gal Zauberman, and participants in the Third Greater Boston Marketing Research Colloquium for their helpful comments and suggestions, Shuili Du for additional background research, and Shreya Doshi, Melanie Etchison, and Cait Poynor for their assistance in coding study 3 responses. Authors are listed alphabetically. Dawn Iacobucci served as editor and Barbara Kahn served as associate editor for this article. Electronically published October 9, 2006

likely to make another trip to reverse your purchase decision (i.e., to buy or return an item). Although perceptions of scarcity triggered by this situation should increase purchase incidence (Cialdini 1993), we are interested in how this situation affects postdecision regrets both in the short term and over time. We argue that, on the drive home, consumers may reect back on the rug they did not buy and feel a sense of loss. However, 3 mo. later they either will have forgotten about the rug or found a way to cope with their decision. Meanwhile, the lamp they did not regret buying on the way home may not t with their decor or not get used, causing this purchase decision to be regretted over time. We rst review prior research on regret, especially as it relates to action and inaction regrets and the temporal pattern of regrets. Next we introduce our situation of interest, limited purchase opportunities, and develop hypotheses that focus on the unique pattern of regrets that should emerge for such purchases and nonpurchases. In contrast to previous research that singled out regretted life events (Gilovich and Medvec 1994, 1995), we consider that not all purchase situations will lead to regret. As such, our investigation includes a fuller range of decisions and outcomesthose that trigger regret and those that do not and it uses long-term utility to qualify when the predicted pattern of regrets will occur. Finally, we examine coping mechanisms and their inuence on long-term regret. We test our framework across three studies, beginning, in study 1, with a eld study that tracked purchase decisions of tourists on a cruise. In study 2, we then replicate our ndings in a more controlled setting. Study 3 addresses alternative explanations and explores the thought processes driving long-term regrets.
342
2006 by JOURNAL OF CONSUMER RESEARCH, Inc. Vol. 33 December 2006 All rights reserved. 0093-5301/2006/3303-0007$10.00

REGRET IN LIMITED PURCHASE OPPORTUNITIES

343

THEORETICAL BACKGROUND AND HYPOTHESES DEVELOPMENT Regret

Regret occurs when someone compares their chosen outcome to a preferred forgone alternative and feels self-blame (Sugden 1985). Postpurchase regret decreases satisfaction (Inman, Dyer, and Jia 1997; Oliver 1997; Tsiros and Mittal 2000), and consumers actively try to avoid regret when choosing (Cooke, Meyvis, and Schwartz 2001). Consumer regret can be associated with choosing the wrong brand, quantity, price, or time to buy (Cooke et al. 2001; Hetts et al. 2000; Simonson 1992). In the present research, we investigate regret associated with buying or not buying a single item at a single price and point in time, a situation that most closely mirrors the action/inaction research in psychology. The omission bias asserts that, in the short term, actions are regretted more than inactions (Kahneman and Tversky 1982; Landman 1987). Inaction is generally seen as the norm (Kahneman and Miller 1986), while action is associated with more risk and responsibility and therefore greater regret. Gilovich and Medvec (1994, 1995) add a temporal dimension to this research, nding that action regrets decrease and inaction regrets increase over time. As a result, long-term regrets tend to be greater for inactions than actions, although in later work these authors suggest that short-term and long-term regrets may be qualitatively different in nature (Gilovich, Medvec, and Kahneman 1998).

lost, which is substantively different from missing the entire purchase opportunity. In limited purchase opportunities, nonpurchase denotes missing the chance to own something desirable, which consumers may perceive as a forfeiture situation. Due to heightened mental endowment in forfeiture situations (Carmon, Wertenbroch, and Zeelenberg 2003), inaction may be framed as a loss and not purchasing may constitute the realization of that loss. If this is the case, postchoice nonbuyers should experience greater regret than buyers, who, by purchasing, are able to avert the loss and any feelings of regret. This prediction is interesting not only because it presents limited purchase opportunities as a boundary condition to the omission bias, as inaction should be regretted more than action, but also because action involves greater responsibility and nancial loss, factors that generally heighten regret. That is, although buying costs more than not buying, the sheer loss of an ownership opportunity causes regret. H1: Following a limited purchase opportunity, shortterm regret associated with nonpurchase should be greater than that associated with purchase. In an unlimited purchase situation, we expect minimal regret for both nonpurchase and purchase. When the purchase decision can be delayed, there is no forfeiture situation, making consumers less likely to experience mental endowment. As a result, deciding not to purchase should be relatively painless. We examine this explicitly in study 3.

Limited Purchase Opportunities and Short-Term Regrets

In todays marketplace, consumers opportunities to buy specic items seem generally innite. However, there are noteworthy instances when distribution for an item is limited either to a certain window of time, such as online auctions or Girl Scout Cookies, or to a certain location. Once the time window passes, the decision to purchase cannot be reversed. Such limited purchase opportunities are different from time pressure (Payne, Bettman, and Johnson 1993) as consumers have ample time to evaluate information; however, the opportunities are a type of scarcity as they limit the time available to act on that information. While most research on scarcity focused on increased item valuation and subsequently purchase likelihood (Brannon and Brock 2001; Cialdini 1993), we focus on postdecision reactions in such situations. Research on temporary price discounts also focused on prepurchase processes, showing that consumers, in anticipation of regret associated with missing a savings opportunity, are more likely to make a purchase (Inman and McAlister 1994; Simonson 1992). Temporary discounts differ from limited purchase opportunities in two key ways. First, consumers have come to expect future discounts (Kalwani and Yim 1992), making any individual promotion less of a one-time offer. Second, after the discount period has passed, only the savings opportunity is

Regrets over Time

Previous research showed that action regrets generally decrease over time while inaction regrets often increase over time due to coping and availability mechanisms (Gilovich and Medvec 1994, 1995). We use a similar logic to reach different predictions for limited-purchase situations, identifying a boundary condition to the traditional pattern of regrets over time. In the omission bias, action regrets tend to be greater than inaction regrets initially. As a result, people are more likely to engage in behavioral and/or psychological repair work to reduce the heightened regret associated with action (Gilovich and Medvec 1994, 1995). In limited purchase situations, we only predict substantial short-term regret for inaction. If, as Gilovich and Medvec suggest, the initial intensity of regret determines the amount of coping, then coping should be greatest for nonbuyers who faced a limited purchase opportunity. As a result of this initial coping, inaction regrets associated with a forgone opportunity should decrease over time. Prior work also shows that preferred outcomes associated with forgone actions (e.g. benets of studying in college) as opposed to inactions are easier to imagine and more cognitively available over time (Gilovich and Medvec 1995; Kahneman and Miller 1986). Such differential availability of forgone alternatives causes regrets of inaction to increase over time. Contrary, in purchase situations, actions often

344

JOURNAL OF CONSUMER RESEARCH

involve buying a physical item that serves as a reminder of the purchase decision, keeping availability high over time. Meanwhile, nonpurchased items are not physically present and therefore are more likely to be forgotten, suggesting again that regrets of inaction should decrease over time. While prior research has singled out life situations that were regretted, not all purchases trigger regret. We use an items long-term utility, that is, the overall quality, usefulness, or pleasure derived from an item, to qualify when the predicted pattern of regrets will occur. Purchase regret should be greatest when an items utility is low, and nonpurchase regret should be greatest when an items utility is high. Combined with the arguments above, nonpurchase regrets should generally decrease over time but to a lesser degree when an items perceived utility is high. Purchase regrets will remain low over time but only if the items utility is high. If utility turns out to be low, consumers may come to regret their purchase decision over time. H2: Nonpurchase regrets will decrease over time, especially when long-term utility is low. H3: Purchase regrets will increase over time but only if long-term utility is low. In summary, we argue that consumers buy/no buy decisions are unlikely to follow the traditional pattern of regrets, where actions are regretted more initially and inactions are regretted more over time. Rather, we argue that, in a limited purchase opportunity, when not purchasing means forfeiting ownership of something desirable, inactions should be regretted more than actions, at least in the short term. Over time, nonpurchase regrets should decrease due to initial coping activity and their likelihood to be forgotten. Meanwhile, purchases that turn out to be low in utility can grow in regret over time, particularly because their physical presence keeps the decision salient. These predictions are tested with a eld study and two laboratory experiments.

vacation. The instructions asked them to, each day, record all items they purchased or considered but did not purchase and rate each item on several dimensions as described below. Respondents were instructed to only record items that they would keep for themselves or their household and not items that would be given as gifts. Approximately 3 mo. after the cruise, willing respondents received a follow-up survey that individually listed their previously reported items and asked them to rate each item on the same dimensions as before. In phase 1, 19 participants provided responses on 180 items: (165 purchases and 15 nonpurchases). Of these participants, 10 completed the phase 2 questionnaire (53% of original), reporting on 94 items from phase 1 (89 purchases and ve nonpurchases).

Dependent Measures. In both phases, participants rated their happiness, satisfaction, and regret with each item using seven-point scales, with higher scores indicating more intense emotions. Happiness and satisfaction ratings were reverse coded and combined with ratings of regret to form an overall regret measure, with higher values indicating greater regret (a ST p 0.86, aLT p 0.93). As an indicator of long-term utility, participants were asked, in phase 2, how frequently they had used each item, ranging from (1) rarely use to (7) frequently use, with the denition of use including wearing or displaying the item.

Results
Short-Term Regrets. In limited opportunities, shortterm regrets should be greater for nonpurchases than for purchases. Given the unequal number of purchases (n p 165) and nonpurchases (n p 15), this comparison should yield a conservative test of hypothesis 1. We analyze our data on the item level using ANOVA, with purchase decision as the sole factor and item nested in participant as the relevant error term. As predicted, participants regretted not buying an item (M p 2.8) more than they regretted buying an item (M p 1.8; F(1, 178) p 11.5, p ! .001). Temporal Changes in Regrets. Note that this analysis is based only on 94 items evaluated in both time periods. We use a mixed ANOVA with item as the unit of analysis, purchase decision as a between-item factor, and time as a repeated factor. As predicted, the interaction of purchase decision by time is signicant (F(1, 92) p 4.04, p ! .05), indicating that regrets change differently over time depending on whether items were or were not purchased. Consistent with hypothesis 2, regrets for nonpurchases trend downward, but, given the few nonpurchases reported (n p 5), we concentrate only on regrets for purchases. We analyze purchases by adding long-term utility as a between-item measure. For purchases, regrets increased over time (MT1 p 1.67, MT 2 p 2.06; F(1, 88) p 21.96, p ! .0001) and decreased with item utility (F(1,87) p 42.39, p ! .0001). As predicted by hypothesis 3, these main effects were qualied by the interaction of utility and time (F(1, 87) p 15.99, p ! .001). Following Aiken and

STUDY 1: PATTERNS OF REGRET FOR REAL LIMITED PURCHASE OPPORTUNITIES

Study 1 investigates regrets among tourists on a cruise around Africa, a situation in which each destination is visited for only one day, prohibiting later reversal of purchase decisions. In addition, there is great natural variance in the inherent long-term utility of items studiedfrom low (e.g., Masai headdress) to high (e.g., South African diamond anniversary ring).

Method
Participants and Procedure. Nineteen adult travelers on a 10-day cruise around Africa were given a Souvenir Diary at the beginning of their trip and told that a fellow traveler was studying satisfaction with items purchased on

REGRET IN LIMITED PURCHASE OPPORTUNITIES

345
FIGURE 1 PURCHASE REGRET AS A FUNCTION OF TIME AND LONG-TERM UTILITY

West (1991), we probed the interaction at three levels of utility ( 1 SD p 2.58, mean p 4.70, +1 SD p 6.81). Regrets of high-utility options remained low over time (F(1, 87) p 1.09, p 1 .3), while regrets for items of average (F(1, 87) p 6.42, p ! .05) and low utility (F(1, 87) p 21.35, p ! .0001) increased over time. See gure 1.

Discussion
Study 1 investigates regrets in a real-life setting where purchase opportunities were extremely limited. Contrary to the omission bias but consistent with our prediction, participants regretted not purchasing more than purchasing in the short term. However, 3 mo. later, purchase regrets for lowand average-utility items increased. This nding is interesting because prior research shows a general decline in action regrets over time (Gilovich and Medvec 1995). Although our study nds support for two of our three hypotheses under authentic conditions, the nature of the situation created some problems. First, the number of items reported on was much larger for purchases (n p 165) than for nonpurchases (n p 15). Nonpurchases may have been underreported due to differential forgetting, as they were not physically present when diary entries were initially completed. This problem was further exacerbated by decreased participation in phase 2, making it impossible to test hypothesis 2. A second issue was that respondents reported on many different types of items that varied on dimensions besides frequency of use. Therefore, our next study will replicate these results using a controlled setting.

STUDY 2: PATTERNS OF REGRET IN A SIMULATED LIMITED PURCHASE OPPORTUNITY

Study 2 uses a simulated concert experience to create a limited purchase situation for a concert shirt. Using the same item across participants provides an element of control but also reduces variance in utility. Thus, we manipulate perceived long-term utility via a framing task.

a limited purchase opportunity by reading that the shirt is ofcially licensed merchandise and is only available at the concert. Purchase decision was manipulated by asking participants to imagine that you decide to buy (not to buy) the shirt. Participants then rated how much they regret their decision the day after the concert (short term). To increase awareness of the passing of time and thus to enhance the validity of the scenario methodology, participants were asked to imagine and describe two occasions they would experience during winter break, and then they were asked to imagine a specic day next semester, identied by weekday and calendar date, and to describe their activities. Next, the products long-term utility was manipulated using a framing task. Participants were asked to elaborate on how the shirt they bought (did not buy) either t or did not t (would or would not have t ) into their daily life. Finally, participants again rated their regret over their purchase decision (long term).

Method
Participants and Design. Three hundred and thirteen undergraduates at Boston University participated in this study, fullling their research requirement in a statistics class. Two factors were manipulated between subjects. Purchase decision (buy, no buy) was manipulated rather than self-determined to ensure equal cell sizes across the buy/no buy decision. Long-term utility (high, low) was manipulated by asking participants to elaborate on how well the item t or did not t with their daily life. Procedure. Participants were asked to describe attending a concert by their favorite artist or band, including the friends who accompanied them and the shirt for sale on the way out. Participants were asked to sketch the shirt to ensure that it was personally desirable. All participants encountered

Dependent Variables. We measured how much regret participants felt and how upset they were about the purchase decision, using nine-point scales with higher ratings indicating greater feelings. Responses were highly correlated in the short term (r p 0.9) and the long term (r p 0.9) and were averaged accordingly to form a measure of regret.

Results
Short-Term Regrets. Purchase opportunity was always limited in this study. Immediately after purchase, participants felt more regret if they did not buy the shirt than if they bought the shirt (F(1, 311) p 17.11, p ! .0001), supporting hypothesis 1. See gure 2 for all means. Temporal Changes in Regrets. As predicted, the threeway interaction of time, purchase decision, and long-term utility was signicant (F(1, 307) p 10.73, p ! .001). For nonpurchases, regrets decreased over time (F(1, 154) p

346
FIGURE 2 REGRET AS A FUNCTION OF TIME AND LONG-TERM UTILITY

JOURNAL OF CONSUMER RESEARCH

1993). If limited purchase opportunities increase item valuation, then assigning some participants to a nonpurchase outcome in study 2 may have translated into a greater perceived loss and therefore greater regret. In our next experiment, we explicitly measure item valuation and allow participants to decide whether or not to buy. Limited purchase opportunities may also trigger reactance, a motivational drive that leads people to pursue the limited option with greater determination than they would have otherwise. Brehm shows that a necessary precondition of reactance is that the restrictions are imposed on previously free behavior, thus removing peoples ability to choose freely among options (Brehm 1966, 3). We argue that presenting a purchase opportunity of limited duration is qualitatively different from removing an existing opportunity and therefore should not lead to reactance. However, since consumers may generally feel entitled to make a purchase decision, assigning them to a buy/no buy scenario may have triggered reactance. Therefore, the next experiment allows participants to choose whether or not to purchase, and it measures feelings of reactance.

143.14, p ! .0001), qualied by a signicant interaction of time and utility (F(1, 154) p 4.96, p ! .03). Supporting hypothesis 2, regrets decreased more sharply over time if participants focused on how the shirt would not have t into ones life (F(1, 76) p 89.51, p ! .0001) than if they focused on how it would have t (F(1, 78) p 53.83, p ! .0001). For purchases, long-term utility moderates the effect of time on purchase regrets (F(1, 153) p 5.77, p ! .02). Supporting hypothesis 3, elaborating on the items lack of t increased regret slightly (F(1, 77) p 3.55, p ! .06), while elaborating on its t decreased regret (F(1,76) p 2.33, p ! .15), although not signicantly so.

Discussion
Using a controlled setting, study 2 replicates the pattern of regrets obtained in the eld study. We again nd that right after a limited purchase opportunity, nonbuyers experience greater regret than buyers, which is contrary to the omission bias, and that over time, buyers regret increased but only if they focused on low-utility aspects of the item. As before, nonpurchase regrets decreased over time, even when utility was seen to be high. These ndings suggest a boundary condition to the temporal pattern of regrets. However, we have not yet shown that this pattern of regrets is unique to limited purchase opportunities. Study 3 will thus compare regrets for limited purchase opportunities to those observed in unlimited situations. In addition, we address alternative mechanisms that may affect our pattern of results.

Hot and Wistful Regret. So far we have treated regret as a one-dimensional construct. However, the nature of regrets may differ in the short term compared to the long term (Gilovich et al. 1998; Kahneman 1995). Specically, shortterm regrets are found to manifest themselves in hot feelings related to anger and irritation, while long-term regrets are less intense feelings that are more longing and wistful in nature. Although evidence for these different regret types was found in the context of life decisions, it is unclear whether this same change in type will occur for purchase decisions, especially given the different pattern of regrets for action and inaction. Since our two-item measure would not have picked up these subtle differences, study 3 includes a number of items tapping into these different regret types. Coping with Regret. We discussed earlier how the coping mechanisms previously used by Gilovich and Medvec (1995) should operate in purchase situations to inuence long-term regret. In the next study, we use a regrettable outcome to investigate the extent to which consumers engage in psychological repair work, such as nding a silver lining or attempting to minimize negative aspects of the outcome (Taylor, Pham, and Rivkin 1998).

STUDY 3: COMPARING PATTERNS OF REGRET FOR SIMULATED LIMITED AND UNLIMITED PURCHASE OPPORTUNITIES
Study 3 examines both limited and unlimited purchase situations within a simulated concert experience. In study 2, sketching the concert shirt may have inadvertently created an endowment effect, thus heightening regret for respondents assigned to the no buy condition. Although endowment alone would not predict the temporal pattern of regrets observed, the current study avoids this problem by presenting all respondents with the same itema live recording

Item Valuation and Reactance. We suggest that the

observed pattern of regret is driven by the limited nature of the situation. However, for short-term regrets, there are two other mechanisms that could be inuencing our results, item valuation and reactance. Theories on scarcity suggest that temporal limitations increase the perceived value of a positively valenced item (Brannon and Brock 2001; Cialdini

REGRET IN LIMITED PURCHASE OPPORTUNITIES

347

of the concert they just attendedand allowing them to choose whether or not to buy the item.

Method
Participants and Design. One hundred and ve undergraduates at Boston University participated in this study, fullling their research requirement in a statistics class. The purchase situation (limited, unlimited) was manipulated between subjects while the purchase decision (buy, no buy) as determined by the participant was included as a second between-subjects factor (Limited: 33 buy, 21 no buy; Unlimited: 36 buy, 15 no buy). To investigate coping behavior, we created a situation where long-term utility was always low. Procedure. As in study 2, participants mentally simulated attending a concert but, in this study, they encountered a CD instead of a shirt on their way out. The CD was described as a live recording of the entire 90-min. concert they had just attended, with the CDs being created in a truck behind the stand. Participants in the limited condition read that the CD will only be available at the concert, while those in the unlimited condition read that the CD will also be available on the bands website and at the Virgin Megastore on Newbury Street. Next, participants indicated how much they thought the CD should cost and made their purchase decision. After imagining that it was the next day, participants rated their emotional reactions along the dimensions described below. Passage of time was again simulated by having participants imagine and describe different events over the next 3 mo. They then imagined a specic date at the beginning of the next semester on which they and their friends were listening to the concert CD that they or, in the case of a no purchase decision, one of their friends had bought. The sound quality of the CD was always described as not very clear (low utility). Participants wrote down their thoughts and feelings regarding their purchase decision, rated how they felt about their decision, and responded to manipulation checks and reactance measures. Dependent Variables. The same two items as in study 2 were used to measure regret and were averaged to form an overall measure of regret in the short term (r 1 0.7) as well as the long term (r 1 0.9). In addition, we used three items tapping into hot regret (irritated, disgusted, and angry) and three items reecting wistful regret (sentimental, wistful, and longing; Gilovich et al. 1998). All items were measured on nine-point scales, with higher values reecting more intense feelings. Items for each factor were highly correlated in the short term as well as the long term (all as 1 0.8) and were averaged accordingly. Participants indicated, in dollars and cents, how much they thought this CD should cost as a measure of perceived value. We measured situationally induced reactance by modifying three statements from Hongs Psychological Reactance trait scale (Hong and Faedda 1996; see app. A for these items). Participants rated

these items on a nine-point scale, with higher values indicating greater degrees of felt reactance. Items were averaged to create a single reactance measure (a 1 0.8). Each thought listed by participants at time 2 was coded as focusing either on the current or an alternative outcome and then, within this general focus, as emphasizing positive or negative aspects of that outcome, or attempting to minimize positive or negative aspects of that outcome. An other category was also used (see app. B for the coding scheme). Three independent judges coded responses, knowing only whether a respondent bought or did not buy (used to identify current vs. alternative outcome). Discrepancies between judges were resolved using a majority rule. Ninetytwo percent of participants thoughts were captured by three categories: emphasizing negative aspects of the current situation, minimizing negative aspects of the current situation, and emphasizing positive aspects of the current situation (all as 1 0.8).

Manipulation Check. Two measures assessed whether participants felt that they would have been able to buy a new CD of the exact concert at a later time and place using a nine-point scale anchored at 1 (not at all likely) and 9 (extremely likely). Thirteen participants in the limited condition provided a rating of either 8 or 9, indicating that they did not see the situation as limited, and their data were subsequently deleted. The following analyses are based on 92 participants (41 in the limited condition and 51 in the unlimited condition). The remaining participants in the limited condition still perceived at least some likelihood that they could buy the CD later (M p 3.8) but signicantly less so than those in the unlimited condition (M p 5.7; F(1, 90) p 13.67, p ! .001), rendering this a relatively conservative test of limited purchase opportunities.

Results
Value. We tested whether limiting the purchase opportunity increased the items perceived value. Participants beliefs about the cost of the CD were not affected by purchase limitation (MLim p $15.93, MUnlim p $13.84 ; F(1, 90) p 1.38, p 1 .2), although means are in the expected direction. Further, perceived value did not affect purchase intent (x 2 p 0.01, p 1 .9). Reactance. Participants in the limited condition did not experience greater feelings of reactance than those in the unlimited condition (MLim p 4.63, MUnlim p 4.04; F(1, 90) p 1.92, p 1 .15), although means are in the expected direction. Also, reactance did not inuence purchase intent (x 2 p 1.27, p 1 .25) and thus is unlikely to affect our ndings. Short-Term Regrets. Results are analyzed using ANOVA, with purchase situation and decision as betweensubjects factors. Nonpurchases (M p 2.7) were regretted more than purchases (M p 1.8; F(1, 88) p 10.10, p ! .01). More important, this main effect was qualied by an interaction of purchase situation by decision (F(1, 88) p 4.44,

348

JOURNAL OF CONSUMER RESEARCH

FIGURE 3 REGRET AS A FUNCTION OF PURCHASE SITUATION AND PURCHASE DECISION

p ! .05). Replicating our previous ndings for limited situations, not buying was regretted substantially more than buying (F(1, 39) p 12.3, p ! .01). In unlimited purchase situations, however, purchases and nonpurchases triggered similar levels of regret (F(1, 49) p 0.6, p 1 .4). See gure 3 for all means.

Temporal Changes in Regrets. Results are analyzed using a mixed ANOVA, with purchase situation and decision as between-subjects factors and time as a within-subject repeated measure. Since all participants learned at time 2 about the CDs poor sound quality, temporal changes in regret were driven by whether or not the CD was purchased, as indicated by a signicant time by decision interaction (F(1, 88) p 38.78, p ! .0001). Consistent with hypotheses 2 and 3, for low-utility items, regret decreased over time for nonbuyers (F(1, 28) p 14.37, p ! .001) and increased over time for buyers (F(1, 60) p 37.77, p ! .0001). The interaction of time and purchase situation was not signicant, indicating that changes in regret were not inuenced by limiting the purchase opportunity (F(1, 88) p 0.02, p 1 .8). See gure 3 for all means. Hot and Wistful Regrets. Assessing the different types of regret, we nd a stronger correlation of the general regret measure with hot regret (rST p 0.75 and rLT p 0.89, p ! .0001) than with wistful regret (rST p 0.36 and rLT p 0.48, p ! .001). An analysis of short-term regrets using general regret and hot regret as repeated factors revealed a main effect of measure (F(1, 88) p 20.94, p ! .0001) but no interactions with other experimental factors. Long-term regrets also revealed only a main effect of measure (F(1, 88) p 10.11, p ! .01). These results, which show only an intercept shift between measures, suggest that the general regret measure responds to purchase situation, decision, and time in the same manner as hot regret. Prior research has suggested that long-term life regrets, which are usually inaction regrets, tend to be more wistful in nature (Gilovich et al. 1998; Kahneman 1995). We nd that wistful regrets exist in the long term but that these regrets were greater for action than inaction (MBuy p 3.5, MNoBuy p 1.6; F(1, 88) p 23.34, p ! .0001), which is consistent with both the general and hot regret measures. In other words, long-term action regrets were both hot and wistful in nature. Thought Listing. By design, buyers experienced a negative outcome that triggered regret, while nonbuyers experienced minimal regret. Therefore, our analyses of the thought processes driving long-term regret focus on buyers. Buyers generated thoughts that emphasize negative aspects of their outcome (M p 1.19), but even more of their thoughts focused on coping activities that either attempt to minimize these negative aspects (M p 0.73) or emphasize positive aspects of their outcome (M p 0.56). Further, regression analyses show long-term regret increasing with thoughts that emphasize negative aspects of the outcome (b p 0.95, t(58) p 3.80, p ! .001) but decreasing with thoughts that minimize
those negative aspects (b p 0.61, t(58) p 2.04, p ! .05) and with thoughts that emphasize positive aspects of the outcome (b p 0.66, t(58) p 1.82, p ! .07). These results suggest that participants were successful with their psychological repair work.

Discussion
Study 3 compares consumers responses in limited versus unlimited purchase situations, which allows us to investigate two alternative mechanisms, scarcity and reactance. Although limited purchase opportunities are a type of scarcity (Brannon and Brock 2001; Cialdini 1993), we did not nd a signicant increase in item valuation. One explanation may be that the monetary value of the focal item (CD) is relatively unambiguous and has little variance. Further, our limited purchase situation did not trigger strong feelings of reactance, perhaps because limiting a new purchase opportunity is not equivalent to removing a previously existing opportunity. We found that nonbuyers felt greater short-term regret in limited but not in unlimited situations, supporting the idea that consumers appraise the purchase situation differently when it is limited. However, differences between limited and unlimited situations disappeared over time as long-term regret was inuenced only by the purchase decision. While the new, negative utility information that became available prior to assessing long-term regret may have overshadowed the effect of purchase situation, this nding may accurately reect that, over time, limitations present at choice become less salient relative to the decision outcome. Further, given the low item utility, it is not surprising that buyers had thoughts about negative aspects of the situation that increased regret. More interesting is our nding that most thoughts either minimized these negative aspects or found a silver lining in the outcome and that these coping thoughts

REGRET IN LIMITED PURCHASE OPPORTUNITIES

349

were effective at decreasing long-term regret. With regard to the type of regret experienced, we found that our general measure of regret corresponds most closely to hot regret. However, wistful emotions were also present in the long term, suggesting that purchase regrets may contain both types of emotions to varying degrees.

regretted more in the long term, we nd both hot and wistful elements in all situations where regret is present, although it was hot regret that most closely mirrored our general regret measure in the way it responded to purchase situation, decision, and time.

GENERAL DISCUSSION
Our research examines regrets in limited purchase opportunities. These situations are characterized by a restricted time window for purchase after which the purchase decision cannot readily be changed. Findings from a eld study and two laboratory studies suggest that limited purchase opportunities create a unique pattern of regrets over time that differs from what was previously observed for actions and inactions in life decisions. Our ndings contribute to the literature on regret in four important ways. First, following limited purchase opportunities, we nd a reversal of the omission bias (Kahneman and Tversky 1982) in the short term. Rather than greater regret for action, we nd greater regret for inaction, in this case nonpurchase. This is especially interesting because purchasing involves nancial expenditure and responsibility, which is generally associated with more regret. Limited purchase situations may render purchasing rather than nonpurchasing the norm (Kahneman and Miller 1986; Simonson 1992), at which point mental endowment (Carmon et al. 2003) may further heighten the personal responsibility of not purchasing. Thus it seems that limited purchase situations represent a boundary condition to the omission bias. Second, we show a different temporal pattern of regrets for limited purchase opportunities. Researchers previously found that action regrets decrease and inaction regrets may increase over time (Gilovich and Medvec 1995; Gilovich et al. 1998). We nd the opposite, that nonpurchase regrets generally decrease and purchase regrets may increase if the utility of the item is low. Further, we study a wider range of situations by focusing on purchase decisions that may or may not trigger regret and that are less intense than the general life regrets (e.g., education, marriage) studied previously. We also nd some evidence that long-term regrets stem more from the nature of the outcome (buy/no buy and utility) than from the nature of the purchase situation (limited/unlimited), but how far this generalizes is an area for future research. Third, we show that reframing a decision can minimize regret (study 2) and that consumers experiencing regret may instinctively use such framing as a coping mechanism (study 3). Thought protocols showed that buyers psychologically minimized negative aspects and emphasized positive aspects of their low-utility purchase to effectively reduce feelings of regret. Fourth, we contribute to the understanding of the nature of the regret experience. Kahneman (1995) argued that, for life regrets, greater short-term action regrets tend to be hot in nature, while greater long-term inaction regrets are more pleasantly wistful. In limited purchase situations, where inactions were regretted more in the short term and actions

LIMITATIONS AND FUTURE RESEARCH

All three studies presented here investigate limited purchase situations for products that have a somewhat pronounced hedonic component. Will our ndings generalize to purely utilitarian products, such as appliances, cleaning supplies, and newspapers? Although limited purchase opportunities could occur for such products, in developed nations, restrictions on utilitarian products seem much less common. If they occur, the magnitude of the lost purchase opportunity may be on a greater scale (e.g., postdisaster). Otherwise, utilitarian purchases may generally be perceived as unlimited purchase situations. Our ndings suggest that consumers frame the purchase decision differently in limited as opposed to unlimited situations. It would be interesting to determine how framing nonpurchase as a loss might t into a decision-process model alongside increased involvement and item valuation when items are scarce (Brannon and Brock 2001; Cialdini 1993). This research only begins to examine the thought processes that inuence long-term regrets. Future research should investigate not only how consumers cope with regrets over time but also how marketers can better assist in the prevention and reduction of regret. As such, it would be worthwhile to learn whether consumers cope with nonpurchase regrets by increasing purchase intent in the future (i.e., I wont make that mistake again). Our research focused on limited opportunities in purchase decisions. However, many other situations are characterized by a similar now-or-never frame (Thorngate 2003). Some of the most important decisions in life, such as whether or not to accept a job or marry someone, have a limited window of opportunity and are often not easily reversible. While some of these decisions may be among lifes greatest regrets (Gilovich and Medvec 1995), we believe that the limited aspect of these decisions represents an important and interesting area for future research.

APPENDIX A REACTANCE MEASURES

(Scale anchored at 1 [not at all] and 9 [very much], thus higher values indicate greater felt reactance.) To what extent did this situation force you to choose a particular alternative? To what extent did the situation restrict your ability to make a free and independent decision? To what extent did the situation trigger a sense of resistance in you?

350

JOURNAL OF CONSUMER RESEARCH

APPENDIX B CODING SCHEME FOR STUDY 3 CURRENT SITUATION1 Emphasize Positive Aspects
Positive emotions (e.g., glad, relieved, justied) Reminder/reminiscence/good times Unique/scarce/hard to nd again Frequency of usehigh Valuehigh (resale) At least I learned a lesson

Minimize Negative Aspects

It wasnt that expensive

Emphasize Negative Aspects

How situation could/should have been worse (general) More expensive elsewhere

Minimize Positive Aspects

How good could it really be? Cant be that good.

Other
Thoughts that do not t into any other category.

Minimize Negative Aspects

Not that big of a deal Didnt cost that much Not upset (i.e., minimize or negate a negative emotion)

REFERENCES
Aiken, Leona and Stephen West (1991), Multiple Regression: Testing and Interpreting Interactions, Thousand Oaks, CA: Sage. Brannon, Laura A. and Timothy C. Brock (2001), Limiting Time for Responding Enhances Behavior Corresponding to the Merits of Compliance Appeals: Refutations of Heuristic-Cue Theory in Service and Consumer Settings, Journal of Consumer Psychology, 10 (May), 13546. Brehm, Jack W. (1966), A Theory of Psychological Reactance, New York: Academic Press. Carmon, Ziv, Klaus Wertenbroch, and Marcel Zeelenberg (2003), Option Attachment: When Deliberating Makes Choosing Feel Like Losing, Journal of Consumer Research, 30 (June), 1529. Cialdini, Robert B. (1993), Inuence: Science and Practice, New York: Harper Collins. Cooke, Alan D. J., Tom Meyvis, and Alan Schwartz (2001), Avoiding Future Regret in Purchase-Timing Decisions, Journal of Consumer Research, 27 (March), 44759. Gilovich, Thomas and Victoria Husted Medvec (1994), The Temporal Pattern to the Experience of Regret, Journal of Personality and Social Psychology, 67 (September), 35765. (1995), The Experience of Regret: What, When, and Why, Psychological Review, 102 (April), 37995. Gilovich, Thomas, Victoria Husted Medvec, and Daniel Kahneman (1998), Varieties of Regret: A Debate and Partial Resolution, Psychological Review, 105 (July), 6025. Hetts, John J., Davis S. Boninger, David A. Armor, Faith Gleicher, and Ariel Nathanson (2000), The Inuence of Anticipated Counterfactual Regret on Behavior, Psychology and Marketing, 17 (April), 34568. Hong, Sung-Mook and Salvatora Faedda (1996), Renement of the Hong Psychological Reactance Scale, Educational and Psychological Measurement, 56 (February), 17382. Inman, J. Jeffrey, James S. Dyer, and Jianmin Jia (1997), A Generalized Utility Model of Disappointment and Regret Effects on Post-Choice Valuation, Marketing Science, 16 (2), 97111. Inman, J. Jeffrey and Leigh McAlister (1994), Do Coupon Expiration Dates Affect Consumer Behavior? Journal of Marketing Research, 31 (August), 42328. Kahneman, Daniel (1995), Varieties of Counterfactual Thinking, in What Might Have Been: The Social Psychology of Coun-

Emphasize Negative Aspects

Negative emotions (e.g., upset, unhappy, frustrated, angry) Poor quality Easy to nd elsewhere Frequency of uselow Valuelow (rip off, waste of money)

Minimize Positive Aspects

Doesnt matter how good it is if you dont use it.

Other
Thoughts that do not t into any other category.

ALTERNATIVE SITUATION2 Emphasize Positive Aspects

How situation could/should have been better (general) Obtain better quality elsewhere (at store, download from web) Obtain cheaper elsewhere Status quo was ne (i.e., undo decisionwould have been ne with alternative)
The current situation can be about oneself or the item. In the coding instructions, the alternative situation was dened as what one could have done but did not do, a forgone outcome. If the respondent decided to buy, then the alternative situations are either not buying at the concert or buying later somewhere else. If the respondent decided not to buy, then the alternative situation is buying at the concert.
2 1

REGRET IN LIMITED PURCHASE OPPORTUNITIES terfactual Thinking, ed. Neal J. Roese and James M. Olson, Hillsdale, NJ: Erlbaum, 37596. Kahneman, Daniel and Dale T. Miller (1986), Norm Theory: Comparing Reality to Its Alternatives, Psychological Review, 93 (April), 13653. Kahneman, Daniel and Amos Tversky (1982), The Psychology of Preferences, Scientic American, 246 (January), 16073. Kalwani, Manohar U. and Chi Kin Yim (1992), Consumer Price and Promotion Expectations: An Experimental Study, Journal of Marketing Research, 29 (February), 90100. Landman, Janet (1987), Regret and Elation Following Action and Inaction: Affective Responses to Positive versus Negative Outcomes, Personality and Social Psychology Bulletin, 13 (December), 52436. Oliver, Richard L. (1997), Satisfaction: A Behavioral Perspective on the Consumer, Boston: Irwin McGraw-Hill. Payne, John W., James R. Bettman, and Eric J. Johnson (1993),

351 The Adaptive Decision Maker, New York: Cambridge University Press. Simonson, Itamar (1992), The Inuence of Anticipating Regret and Responsibility on Purchase Decisions, Journal of Consumer Research, 19 (June), 10518. Sugden, Robert (1985), Regret, Recrimination, and Rationality, Theory and Decision, 19 (July), 7799. Taylor, Shelley E., Lien B. Pham, and Inna D. Rivkin (1998), Harnessing the Imagination: Mental Simulation, Self-Regulation, and Coping, American Psychologist, 53 (April), 42939. Thorngate, Warren (2003), Teaching Judgment and Decision Making: Take It or Leave It, Society for Judgment and Decision Making Newsletter, 22 (September), 911. Tsiros, Michael and Vikas Mittal (2000), Regret: A Model of Its Antecedents and Consequences in Consumer Decision Making, Journal of Consumer Research, 26 (March), 40117.