Organizational Behavior and Human Decision Processes 101 (2006) 89–104 www.elsevier.com/locate/obhdp
Inaction inertia, regret, and valuation: A closer look Marcel Zeelenberg
a,*
q
, Bernard A. Nijstad b, Marijke van Putten a, Eric van Dijk
c
a
c
Department of Economic and Social Psychology, Tilburg University, P.O. Box 90153, 5000-LE Tilburg, The Netherlands b Department of Psychology, University of Amsterdam, Roetersstraat 15 1018-WB Amsterdam, The Netherlands Department of Social and Organizational Psychology, Leiden University, P.O. Box 9555, 2300-RB Leiden, The Netherlands Received 13 January 2005 Available online 4 January 2006
Abstract Inaction inertia is the phenomenon that one is not likely to act on an attractive opportunity after having bypassed an even more attractive opportunity. So far, all published work has assumed a causal role for the emotion regret in this effect. In a series of 5 experiments we found no support for this regret explanation. In these experiments factors that influenced regret did not influence inaction inertia, and factors that influenced inaction inertia did not influence regret. In addition, in two experiments we found evidence that missing the initial opportunity leads to a devaluation of the later offer. We propose that, in some cases, regret may be a by-product of this devaluation, rather than a cause of inaction inertia. 2005 Elsevier Inc. All rights reserved. Keywords: Inaction inertia; Regret; Valuation; Decision-making
The things we do have a clear impact on our lives. Not only as a result of the direct impact of these actions on the outcomes we obtain and the well-being that stems from it, but also more indirectly via its influence on subsequent behavior. This particular impact of actions on subsequent behavior is documented by a number of well-known phenomena, such as the sunk cost effect (Arkes & Blumer, 1985) and the foot-in-the-door effect (Freedman & Fraser, 1966). However, Tykocinski, Pittman, and Tuttle (1995) revealed that also actions we forgo can impact on our future behavior. These authors introduced the inaction inertia effect: the phenomenon that ‘‘forgoing an attractive action opportunity (initial inaction) decreases the likelihood that subsequent action will be taken in the same domain’’ (Tykocinski & Pittman, 1998, p. 607). q
This research is partly supported by the Netherlands Organization for Scientific Research (Grants # NWO 400-03-385 awarded to Marcel Zeelenberg and # NWO 452-04-311 awarded to Bernard Nijstad). * Corresponding author. Fax: +31 13 466 2067. E-mail address:
[email protected] (M. Zeelenberg). 0749-5978/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.obhdp.2005.11.004
The inaction inertia phenomenon can easily be illustrated by Tykocinski et al.’s (1995, p. 796) fitness center scenario. In this study, participants imagined that they considered becoming a member of a fitness club located 30 min away. Next, they read that they had failed to act on an earlier opportunity, namely becoming member of a fitness center that was located 25 min away (the small difference condition) or one that was located 5 min away (the large difference condition). Participants who failed to become member of the center that was only 5 min away were less likely to join the center that was currently under consideration. Thus, because the initial, more attractive action opportunity is missed (initial inaction), the second less attractive action opportunity is not taken (inertia). In other publications Tykocinski and Pittman (1998, 2001, 2004; Tykocinski, Israel, and Pittman, 2004) and other researchers (Arkes, Kung, & Hutzel, 2002; Butler & Highhouse, 2000; Kumar, 2004) showed similar effects for decisions about vacation trips, automobile purchases, renting apartments, ski passes, gambles, and etcetera. In all these cases, a large difference in attractiveness
90
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
between the initial and second opportunity led to a lower likelihood of taking the second opportunity than did a small difference between the initial and second opportunity. The most intriguing question regarding inaction inertia concerns the cause of this phenomenon. To our knowledge, all published work on inaction inertia suggests a causal role for the emotion regret. We will review the relevant literature and bring to light that evidence for the regret explanation is mixed. We further present a series of five experiments that systematically tests the regret explanation of inaction inertia and fails to provide support for this explanation. Interestingly, our two final experiments support an alternative explanation in terms of the devaluation of the second opportunity (see Arkes et al., 2002).
Regret and inaction inertia Regret is felt after realizing that our present situation would have been better if only we had acted differently (Zeelenberg, 1999). It is a comparison-based emotion (Van Dijk & Zeelenberg, 2005) and can be felt in response to bad decision outcomes and bad decision processes (Pieters & Zeelenberg, 2005). The key element of the regret explanation of inaction inertia is that missing the attractive initial opportunity to act leaves the decision maker with regret, because missing the initial opportunity is considered a loss. Tykocinski et al. (1995), in the general discussion of their original article on inaction inertia, already alluded to an explanation in terms of regret. They wrote: ‘‘Because the opportunity in question is now associated with a negative situation, individuals may be reluctant to process additional information that might reactivate this negative experience and provoke processes such as counterfactual regret’’ (p. 802). As a consequence of this regret, people reject the subsequent alternative promptly and without much consideration. In a later publication Tykocinski and Pittman (1998, p. 608); wrote: ‘‘participants in the large difference condition in these studies were probably entertaining upward counterfactual thoughts such as ‘If only I had joined the frequent flyer program before my first trip, by now I could have been close to getting a free ticket.’ Such upward counterfactuals are clearly unpleasant, provoking the experience of regret.’’ Thus, as it seems, regret over missing the initial opportunity is considered to be prime cause of inaction inertia. There is indeed some evidence that experiences of regret may influence subsequent behavioral decisions (Ratner & Herbst, 2005; Zeelenberg & Pieters, 1999; see for a review, Zeelenberg, Inman, & Pieters, 2001). But, to ascertain whether regret indeed has explanatory power for the inaction inertia phenomenon, it is relevant to be more specific about how regret might operate. The
regret explanation of inaction inertia implies at least two things. First, missing an initial very attractive opportunity should produce more regret than missing an opportunity that was less attractive. Second, there should be a negative correlation between regret and the likelihood of accepting the second opportunity (more precisely, there should be evidence for statistical mediation). Our reading of the literature revealed that the evidence appears to be mixed with respect to both aspects of this explanation. Before we turn to this empirical evidence, however, it is appropriate to note that there are two slightly different versions of the regret explanation. One stresses the avoidance of future regret, the other the escape of current regret. Tykocinski and colleagues and Butler and Highhouse (2000) seem to favor the avoidance explanation, whereas Arkes et al. (2002) and Kumar (2004) favor the escape explanation. Importantly, the difference between the two explanations is not always clear, both theoretically and empirically. Tykocinski and Pittman (1998) write in their introduction: ‘‘We propose that declining the subsequent action opportunity, with its inferior outcomes, represents an attempt to prevent or to put an end to the unpleasant psychological experience of regret’’ (p. 608, emphasis added). When discussing their results they write, ‘‘we cannot say which of these two processes, escape or avoidance, motivates inaction inertia’’ (p. 615). Importantly, although there are two slightly different regret explanations, we would like to note that in both explanations the regret stems from missing the first opportunity: In the one case regret is presently felt, in the other it is anticipated to be felt in the future. Further, according to both explanations, manipulations that amplify regret over missing the first offer, should also amplify the inaction inertia effect. Interestingly, no study to date has actually compared both types of regret within a single experiment. We will make the distinction between escaping current regrets and avoiding future regrets explicit in our Experiments 4 and 5. Let us now briefly review, in chronological order, the studies in which the regret explanation was tested and indicate for each study how regret was assessed.
Mixed support for the regret explanation Tykocinski and Pittman (1998, Experiment 4) were the first to explicitly test the regret explanation and to include a measure of regret. Participants in this experiment first read a scenario about booking a discounted trip to Tuscany after having missed a much more attractive or slightly more attractive offer. Next they indicated the likelihood of acting on the current discounted trip. After this they were asked to list the reasons that had led them to their decision. These were coded in different
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
categories, including regret. Although participants in the large difference condition on average reported more regret related thoughts that those in the small difference conditions, the thoughts did not correspond well to the likelihood of acting on the second action opportunity in the sense that participants who expressed regret were not significantly less likely to act. Butler and Highhouse (2000) had their participants imagine that that they owned a small business, and that they had received an offer from a big corporation to buy them out. The attractiveness of an earlier offer was manipulated. They did find a significant negative correlation between regret and action likelihood (‘‘r = .71’’: p. 228), but a limitation of this study is that no effect was found of prior inaction on anticipated regret. Regret was assessed with the combination of the following two items: ‘‘Imagine you accepted the offer. How happy would you be?’’, ‘‘Imagine you accepted the offer. How much would you regret your decision?’’ (Adam Butler, personal communication, November 2, 2004). Note that in this case the regret question does not refer to the first offer, although, theoretically it should. Tykocinski and Pittman (2001, Experiment 3) studied decisions to sign up for a course with an inexperienced teacher after having failed to sign up for the course with an excellent (large difference) or slightly above average teacher (small difference). They found that the large difference condition produced more regret and less action likelihood than the small difference condition, but they also found that the correlation between regret and action likelihood was not significant (‘‘r = .14, p < .30’’: p. 155). In that study participants were asked to imagine that they signed up for the course with the inexperienced teacher and that they found that the ‘‘lecturer’s lack of experience interferes with the course’s academic quality and impedes its progress’’ (p. 154). Next, regret was assessed by asking the participants to respond to the following statement: ‘‘In this situation I will probably regret having decided to sign up for this course’’ with end points, not at all descriptive of me [0], and very descriptive of me [10] (p. 154). In their first scenario experiment, Arkes et al. (2002) studied participants’ likelihood to act on a discounted pair of shoes ($90 instead of the usual $100) after they had missed an earlier sale in which the shoes cost even less ($40 vs. $80). They found that regret was more intense, and action likelihood was lower in the large difference condition than in the small difference condition. But, action likelihood and regret were measured in two different samples, so no correlation between the two could be computed. Regret was assessed with: ‘‘How much regret do you feel over missing the sale your friend told you about?’’ [0 = no regret at all, 10 = a great deal of regret] (p. 376). In Experiment 3, Arkes et al., measured action likelihood and regret in the same sample, and could establish that missing a very attractive offer
91
produced more regret then missing a less attractive offer, and that regret was negatively related to action likelihood. Mediation analyses showed that the effect of their difference in attractiveness manipulation on action likelihood was mediated by regret. Regret was assessed by asking participants to indicate the ‘‘amount of regret you would feel over having missed the earlier $40/$80 bargain’’ (p. 380). Finally, Kumar (2004) finds in his Study 1 that inaction inertia is mediated by regret over the first offer. Unfortunately, the other two studies only included the large difference conditions. This makes it hard to answer the question of whether regret mediates inaction inertia (i.e., the difference in action likelihood between the large and small difference conditions). In these studies, participants ‘‘reported the level of regret they would experience for having missed the previous sale, on an 11point (not at all/a lot) scale’’ (p. 178). Taken together, there is some support for the explanation of inaction inertia in terms of regret, but the evidence is mixed. In our research, to which we turn next, we continued the quest for the causes of inaction inertia. We aimed for a critical test of the escape from regret explanation. Hence, we did the following: we manipulated factors that influence regret, we assessed both regret and action likelihood and tested whether the factors that influenced regret also influenced action likelihood, and finally we investigated whether the effects on action likelihood were mediated by regret. A critical test of the regret explanation along these lines has, to our knowledge, not been performed before. We first focused uniquely on the escape from regret explanation in our Experiments 1–3. In Experiments 4 and 5 we also studied the avoidance of regret explanation and an explanation in terms of devaluation of the second offer, which was put forward by Arkes et al. (2002).
Experiment 1 The main impetus for our first experiment was a finding obtained by Tykocinski et al. (1995, Experiment 4). They report a study in which they crossed the standard manipulation of difference in attractiveness between the first and later opportunity with a manipulation of responsibility for missing the first opportunity. Interestingly, it was found that the manipulation of responsibility had an effect on the level of perceived personal responsibility (i.e., the manipulation check), but no effect on the likelihood of acting on the second opportunity. Thus, responsibility did not affect inaction inertia. This finding was replicated in Tykocinski et al.’s Experiment 5. This finding is of great relevance for the regret explanation of inaction inertia, because previous research has established, time and again, that there is a strong connection between responsibility and regret (see, Ordo´n˜ez
92
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
& Connolly, 2000; Zeelenberg, Van Dijk, & Manstead, 1998, 2000). Greater responsibility is associated with more regret. One may argue that responsibility is a defining characteristic of regret, such that if you see the responsibility for a bad outcome laying elsewhere you are likely to experience negative affect, but not the emotion regret. If regret plays a causal role in inaction inertia, personal responsibility should thus have led to an increase in regret and a decrease in action likelihood. Consequently, Tykocinski et al.’s findings seem to contradict the regret explanation. It is important to note, however, that Tykocinski et al. (1995) did not assess regret in this particular experiment. Thus, to critically examine the regret explanation, it would be worthwhile to replicate Tykocinski et al.’s Experiment 4 with the inclusion of a regret measure. Not only would such a replication reassure us about the occurrence of the null-effect for manipulated responsibility, it would also allow for a more direct test of the role of regret. This extended replication was the starting point of our inquiry. Method Participants and design Eighty students (39 males, 41 females, Mage = 21 years) at Tilburg University volunteered to participate in this study. They were randomly assigned to one of the four conditions of the 2 (Difference in Attractiveness: Small vs. Large) · 2 (Responsibility: High vs. Low) complete factorial design, with 20 participants per condition. Procedure and measures Participants were approached individually at several places on the University Campus (e.g., cafeteria’s, library). They were provided with a one-page questionnaire containing the scenario and the dependent measures. The scenario was closely modeled after Tykocinski et al.’s (1995, p. 789) Experiment 4. The scenario in the high responsibility and large [small] difference in attractiveness condition read as follows: You consider becoming a member of a fitness club located 30 min away. You could have joined another center, closer to home. Although you were told by a friend that this closer center was taking only a few more members and that you better hurry up or it might be too late, you missed this opportunity. The center that by now has closed its membership roll was located 5 [25] minutes away from your home. Although you passed this fitness center several times during the week after you talked to your friend, and thought about going to pick up a membership application form, you did not. When you wanted to register at the fitness center you were too late. You missed the opportunity.
In the low responsibility condition the last part of the scenario read as follows: ‘‘The day after you talked to your friend you had to leave town on a one week vacation with your family. When you wanted to register at the fitness center you were too late. You missed the opportunity.’’ The dependent variables in all our experiments were ratings on 11-point scales. Action likelihood was obtained directly after participants read the scenario. They were asked to indicate the likelihood that they would join the second fitness center (0 = not likely, 10 = very likely). Next participants rated the extent to which they felt personally responsible for missing the initial opportunity (0 = not responsible, 10 = very responsible). The last rating involved the extent to which they experienced regret over missing the opportunity to become member of the fitness center that was closer to home (0 = no regret, 10 = very much regret). Results and discussion The results are shown in Table 1 and were all analyzed using 2 · 2 ANOVAs. The responsibility ratings can be seen as a check of the responsibility manipulation. The analysis revealed only a significant main effect for responsibility, F (1, 76) = 16.70, p < .001, g2 = .18. As expected, participants in the high responsibility conditions reported higher levels of responsibility (M = 8.10, SD = 2.05) than those in the low responsibility conditions (M = 6.08, SD = 2.37). These ratings were not affected by the difference in attractiveness, neither by the interaction, Fs < 1. The results for likelihood of action were different. The ANOVA revealed a main effect of difference in attractiveness, F (1, 76) = 26.50, p < .001, g2 = .26. Participants in the large difference conditions were less likely
Table 1 Ratings on the dependent variables as a function of responsibility and difference size in Experiment 1 Dependent variables
Responsibility Low
High
M
SD
M
SD
Responsibility Small difference Large difference
5.60 6.55
(2.33) (2.37)
8.10 8.10
(2.07) (2.08)
Likelihood of action Small difference Large difference
7.55 5.00
(1.67) (2.34)
6.55 4.80
(1.57) (1.80)
Regret Small difference Large difference
4.50 6.05
(2.44) (2.74)
7.10 7.65
(1.87) (1.60)
Note. Ratings were made on 11-point scales, with endpoints labeled not responsible (0) and very responsible (10), not likely (0) and very likely (10), and labeled no regret (0) and very much regret (10).
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
to act (M = 4.90, SD = 2.06) than those in the small difference conditions (M = 7.05, SD = 1.68), revealing the inaction inertia effect. But there were no effects of Responsibility, F (1, 76) = 2.01, p = .155, g2 = .03, or the interaction, F (1, 76) = 0.92. This pattern of results closely resembles those obtained by Tykocinski et al. (1995, Experiment 4). Finally, the effects on regret were analyzed. Consistent with earlier findings, the analysis revealed a clear main effect for responsibility, F (1, 76) = 18.10, p < .001, g2 = .19. As expected, participants in the high responsibility conditions reported higher levels of regret (M = 7.38, SD = 1.74) than those in the low responsibility conditions (M = 5.27, SD = 2.68). Also a main effect of difference in attractiveness was found, F (1, 76) = 4.53, p < .05, g2 = .06. Participants in the large difference conditions indicated more regret (M = 6.85, SD = 2.36) than those in the small difference conditions (M = 5.80, SD = 2.51). There was no interaction, F (1, 76) = 1.03 p = .32, g2 = .01. Interestingly, the correlation between regret and action likelihood was not significant (r = .006, p = .96). These results clearly show that the manipulation of responsibility had a significant effect on regret, but not on the likelihood of action. At the same time, the difference in attractiveness of the initial and current opportunity did have an effect on likelihood of action. When a more attractive action opportunity was missed, the likelihood of action decreased, replicating the inaction inertia effect. The attractiveness of the initial opportunity also had an effect on the amount of regret. If experienced regret would be the cause of inaction inertia, then a factor that influences regret should also influence the occurrence of inaction inertia. In this study, the amount of responsibility did influence regret. However, it did not affect the likelihood of action.
93
assessed in that experiment, making it unclear whether the results are in favor of the regret explanation. Our Experiment 2 can be seen as a conceptual replication of their study with a measure of regret, and hence a more direct test of the experienced regret explanation of inaction inertia. Method Participants and design Sixty three students (11 males, 52 females, Mage = 21 years) at Leiden University volunteered to participate in this study. They were randomly assigned to one of the three conditions (Time Interval: short, long, control), with 21 participants per condition. Procedure and measures Participants, approached individually at several places on the University Campus, were asked if they wanted to participate in a study on buying behavior. After they agreed they were provided with a one-page questionnaire containing the scenario and the dependent measures. The scenario for the short [long] time interval read as follows: You would like to have a couch in your room. In the window of a furniture shop you saw some nice examples yesterday [two months ago] that had a 50% discount. Although you were interested you did not get to the shop right away. When you arrive at to the shop today, the owner tells you that you are a day [two months] late and the 50% discount does not apply anymore. He tells you that there is a 20% discount on the couch this coming week. Participants in the control condition did not read about the first opportunity and were only confronted with the 20% discount. The dependent variables, action likelihood and regret, were identical to those of Experiment 1.
Experiment 2 Results and discussion The results of Experiment 1 do not support the regret explanation. It may, however, be premature to conclude solely on the basis of this experiment that experienced regret is not the cause of inaction inertia. In the present experiment we therefore further explore the relation between regret and inaction inertia. In particular, we manipulate a factor that has been shown to influence the occurrence of inaction inertia, namely the time difference between missing the first offer and being confronted with the second offer (cf., Tykocinski & Pittman, 2001, Experiment 1). One reason why the time interval between the first and second offer affects action likelihood may be that the experience of regret becomes less intense with increased time between the missed and current opportunity. Unfortunately for our present purposes, regret over missing the first opportunity was not
The results are shown in Table 2. A one-way ANOVA on these action likelihood ratings in the three
Table 2 Mean ratings on the dependent variables as a function of experimental condition, in Experiment 2 Dependent variables
Time difference Short
Likelihood of action Regret
Long
Control
M
SD
M
SD
M
SD
1.95 8.05
(1.94) (2.11)
4.33 7.67
(2.39) (2.27)
7.19 —
(2.29) —
Note. Ratings were made on 11-point scales, with endpoints labeled not likely (0) and very likely (10), and no regret (0) and very much regret (10).
94
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
conditions yielded a significant effect, F (2, 60) = 29.39, p < .001, g2 = .50. Post hoc Tukey tests showed that the likelihood of action was highest in the control condition, which was significantly different from the long time interval condition and from the short time interval condition. This indicates that the inaction inertia effect occurred both after the initial offer was one day ago or two months ago. However, the likelihood of taking the second offer also differed significantly between the long interval condition and the short interval condition, indicating that the inaction inertia effect occurred, but less strong after a two month period than after one day. A t test on the regret ratings in the two experimental conditions1 did not yield a significant effect, t (40) = .56, ns. However, there was a significant positive correlation between the amount of experienced regret after missing the 50% discount and the likelihood of taking the 20% discount (r = .36, p < .02). Interestingly, this correlation thus goes against the regret explanation of inaction inertia. The more regret people experienced after missing the attractive offer, the more likely they were to act on the next offer. Experiments 1 and 2 suggest that regret experienced over the missed opportunity may not be the cause of inaction inertia. A manipulation that affected regret (responsibility; Experiment 1) did not affect action likelihood. A manipulation that affected action likelihood (time difference; Experiment 2) did not affect regret. As such, we have established a clear double dissociation between regret and action likelihood. Further evidence undermining the regret explanation is reported in Experiment 3, where we use a manipulation that may increase regret: uniqueness of the first offer.
Experiment 3 When we miss the chance to book a very cheap holiday arrangement, we may regret not booking that holiday sooner. When this is a regular offer that comes by more often, we feel regret and wait for the next offer. But what if the initial opportunity was a unique offer? What if that first opportunity was a once in a lifetime opportunity? We expect that after missing a unique opportunity the tendency to kick ourselves will be stronger and the feeling of regret will be more intense. After all, attractive opportunities come and go, but once in a lifetime opportunities only pass by once in a lifetime. The reasoning is that once in a lifetime offers are scarce. Research on commodity theory has shown that any commodity (i.e., anything that is useful to the owner and can be conveyed to another person) ‘‘will be valued to the extent that it is unavailable’’
1
Regret was not assessed in the control condition, because without a missed opportunity there was nothing to regret.
(Brock, 1968, p. 246). Put differently, the scarcer an items is, the more valuable it will be. Extending this to regret, we predict that the more valuable the missed offer is, the more regret one would feel. In the present experiment we manipulated whether the initial offer was unique, or not. A scenario was created in which participants imagined planning a holiday. They forgo a very attractive offer that for one half of the participants was a unique offer, due to an anniversary of the travel agency. The other half of the participants had no extra information on the offer, similar to former studies on inaction inertia. We predict that, after missing a more unique opportunity, people report more regret. As a consequence, assuming regret is the cause of inaction inertia, this amplified regret should increase the inaction inertia effect. If, however, experienced regret is not the cause of inaction inertia, this effect may not be observed. Method Participants and design Eighty students (23 males, 57 females, Mage = 22 years) at Tilburg University volunteered to participate in this study. They were randomly assigned to one the four conditions of the 2 (Difference in Attractiveness: Small vs. Large) · 2 (Offer Type: Unique vs. Control) complete factorial design. There were 20 participants in each condition. Procedure and measures Participants were approached individually at several places on the University Campus. They were provided with a one-page questionnaire containing the scenario and the dependent measures. In the condition with a large [small] difference in price and a unique offer the scenario read as follows: You adore Rome! Shortly you will have a whole week without lectures and you would very much like to visit Rome. A friend tells you that a local travel agency, because of its 25th anniversary, offers a completely organized three-day trip to Rome. You can book the trip this week for €100 [€165] instead of the usual €199. This includes traveling with the High Speed Train and two nights with breakfast at a four star hotel in the center of Rome. He also tells you that you have to hurry before it is too late and the trip is sold out. During the week after you have spoken to your friend, you pass by the travel agency several times and think about booking the trip. However, you do not do so. When you finally want to book the trip it is sold out. It was a unique offer because of the 25th anniversary. You are too late. The scenario in the control conditions did not mention the 25th anniversary and unique character of the offer. After reading the scenario, participants indicated
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
regret. Next they read: ‘‘Your friend calls you again and tells you that, although you missed the previous opportunity, you can still book the trip this week for €170.’’ After having read this, participants indicated their likelihood of action. Results The results are shown in Table 3. These were analyzed using 2 · 2 ANOVAs. The effects on regret revealed an offer type main effect, F (1, 76) = 5.74, p < .02, g2 = .07. As predicted, participants in the unique offer conditions reported more regret (M = 7.25, SD = 1.61) than participants in the control conditions (M = 6.10, SD = 2.53). The other effects were not significant, Fs < .02. The results for action likelihood revealed a main effect of difference in attractiveness, F (1, 76) = 46.59, p < .001, g2 = .38. Participants in the large difference conditions were less likely to act (M = 4.67, SD = 2.95) than those small difference conditions (M = 8.15, SD = 1.76), indicating the inaction inertia effect. There was also a main effect of offer type, F (1, 76) = 7.84, p < .01, g2 = .09, such that participants in the unique offer conditions were actually more likely to act on the second opportunity (M = 7.13, SD = 1.76) than participants in the control conditions (M = 4.67, SD = 2.95). The significant interaction effect, F (1, 76) = 4.88, p < .05, g2 = .06, shows that inaction inertia is much weaker in the unique offer conditions, F (1, 76) = 10.65, p < .005, g2 = .12, than in the control conditions, F (1, 76) = 40.82, p < .0001, g2 = .35. The correlation between regret and action likelihood was not significant (r = .075, p = .51). Discussion The type of initial offer (a unique opportunity or not) had the predicted results for regret. When the initial offer was unique, people indicated more regret than when it was not. At the same time, however, the likeli-
Table 3 Ratings on the dependent variables as a function of difference in price and offer type, in Experiment 3 Dependent variables
Offer type Control
Unique
M
SD
M
SD
Regret Small difference Large difference
6.00 6.20
(2.75) (2.35)
7.15 7.35
(1.53) (1.73)
Likelihood of action Small difference Large difference
8.00 3.40
(2.20) (2.72)
8.30 5.95
(1.22) (2.65)
Note. Ratings were made on 11-point scales, with endpoints labeled not likely (0) and very likely (10), and no regret (0) and very much regret (10).
95
hood of action also increased when the offer was unique. This clearly contradicts the escape from regret explanation of inaction inertia. We manipulated a factor that increased regret, and at the same time decreased the occurrence of inaction inertia. All in all, the experiments reported so far refute the explanation of inaction inertia in terms of regret. The results, however, do raise the question of what does cause the effect. Two explanations remain, avoidance of future regret and devaluation. Tykocinski and Pittman (1998, 2001) and Butler and Highhouse (2000) advanced the avoidance of regret (or anticipated regret) explanation. They argued that taking the second option would remind people of having missed the first and more attractive option (i.e., they would anticipate feeling like a ‘‘sucker’’ not taking the earlier option). Thus, these anticipated negative feelings reduce the anticipated decision outcomes and make the second offer less attractive, which is why it is not taken. In our Experiments 4 and 5, we therefore also measured anticipated regret (i.e., ‘‘how much regret do you expect to feel if you would take the second offer’’). Arkes et al. (2002) introduced another alternative explanation of inaction inertia. They have argued that a previous offer may function as an anchor against which the next offer is judged. When, for example, one is considering buying a couch (as in our Experiment 2) and one has previously learned that it was on sale for half the price, one may come to think that the couch is not worth the full price (see also, Zeelenberg & Van Putten, 2005). When the second opportunity is offered, one might reason that the price is above its actual value and consequently may decide not to buy it. This devaluation explanation was tested in Arkes et al.’s Experiment 3, and received support: valuation of the second offer mediated between their difference in attractiveness manipulation and action likelihood. Therefore, we also measured valuation of the second offer in our Experiments 4 and 5.
Experiment 4 In Experiment 4 we studied the effects of shallow versus deep thinking. Before answering the items of our questionnaire, participants were either instructed to give a spontaneous quick answer or to think deeply about the likelihood to act on the second offer. According to the regret explanation of inaction inertia, thinking deeply about the second offer would reactivate the regret felt about missing the first offer, and one might plausibly expect that regret would consequently be exacerbated (‘‘individuals may be reluctant to process additional information that might reactivate this negative experience and provoke processes such as counterfactual regret,’’ Tykocinski & Pittman, 1998, p. 802). However,
96
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
this has never been tested. Further, one might expect that increased regret would lead to more inertia in the deep thinking condition. In this experiment, we also tested the devaluation explanation (Arkes et al., 2002). We predicted, however, that thinking more carefully about the second offer would attenuate the tendency to devalue the second offer. Indeed, one may come to realize that, although the offer is not as good as the previous offer, it still is better than no offer at all. Also, when thinking deeply, one may come realize that the regret one feels about missing the first opportunity does not qualify as a valid reason not to act on the current opportunity. Thus, the second offer may not be considered a loss (as compared to the previous offer), but as a gain (as compared to the normal price; cf. Tykocinski et al., 1995, Experiment 6). If this reasoning were correct, one would expect an interaction between difference in attractiveness and instruction, such that the inaction inertia effect would be strongest in the large difference/shallow thinking condition. A similar interaction should be obtained for our measure of valuation of the second offer. Method Participants and design One hundred and twenty students at Tilburg University (51 males, 69 females, Mage = 21 years) volunteered to participate in this study. They were randomly assigned to one the four conditions of the 2 (Difference in Attractiveness: Small vs. Large) · 2 (Thinking: Shallow vs. Deep) complete factorial design, with 30 participants per condition. Procedure and measures Participants were approached individually at several places on the University Campus. They were provided with a two-page questionnaire containing the scenario and the dependent measures. In the conditions with a large [small] difference in attractiveness the scenario read as follows: You adore Paris! Next week you have a whole week without lectures and you have not yet planned anything. A friend calls you and informs you about a completely organized three-day trip to Paris that is available up until tomorrow for 175 [280] instead of the usual 350 Dutch Guilders.2 This trip includes traveling with the
2
In Experiment 4, the monetary amounts were expressed in Dutch Guilders, whereas in Experiment 3 (and 5) these were expressed in Euros. This was because Experiment 4 was conducted before introduction of the Euro (January 2002) whereas Experiments 3 and 5 were conducted after introduction of the Euro.
Thalys (High speed train) and two nights including breakfast at a two star hotel in the center of Paris. He asks you to join him next week. It sounded like a good idea, but you forgot to book the trip within these two days. Now your friend calls you again and tells you that, although you missed the prior opportunity, you can still book the trip this week for 315 Dutch Guilders. How likely is it that you will book this trip for 315 Dutch Guilders? Participants in the Shallow Thinking conditions were instructed to provide their spontaneous reaction. In the Deep Thinking conditions they were instructed think deeply before answering the question about their likelihood to act. In addition they were asked to provide a written explanation for their ratings. The first two dependent variables, action likelihood and regret, were identical to the earlier studies. In addition we assessed anticipated regret with the following question: ‘‘How much regret will you feel when you book the trip and realize that you could have booked it for less money’’ (0 = no regret, 10 = very much regret). The last dependent variable was a rating of the value participants attached to the Paris trip (cf. Arkes et al., 2002). This rating was also on an 11-point scale, with endpoints 175 Guilders and 350 Guilders. Results The results are shown in Table 4 and were all analyzed using 2 · 2 ANOVAs. The ANOVA for likelihood of action revealed a main effect of difference in attractiveness, F(1, 116) = 11.44, p < .001, g2 = .09. Participants in the large difference conditions were less likely
Table 4 Ratings on the dependent variables as a function of thinking and difference size in Experiment 4 Dependent variables
Thinking Shallow
Deep
M
SD
M
SD
Likelihood of action Small difference Large difference
6.67 4.47
(2.51) (2.57)
7.63 7.03
(2.34) (1.38)
Experienced regret Small difference Large difference
4.60 6.16
(2.51) (2.48)
4.80 5.50
(2.62) (2.36)
Anticipated regret Small difference Large difference
4.77 6.90
(2.54) (2.44)
5.13 5.63
(2.69) (2.63)
Valuation Small difference Large difference
6.77 4.16
(2.56) (3.34)
7.67 7.76
(2.40) (2.60)
Note. Ratings were made on 11-point scales, with endpoints labeled not likely (0) and very likely (10), no regret (0) and very much regret (10), and 175 Guilders (0) and 350 Guilders (10).
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
to act (M = 5.75, SD = 2.42) than those in the small difference conditions (M = 7.15, SD = 2.48), revealing the inaction inertia effect. There was also a main effect of thinking, F (1, 116) = 18.22, p < .001, g2 = .14, such that participants in the deep thinking conditions were more likely to act on the second opportunity (M = 7.33, SD = 1.96) than participants in the shallow thinking conditions (M = 5.57, SD = 2.75). The significant interaction effect, F (1, 116) = 3.75, p = .056, g2 = .03, shows the predicted pattern. Inaction inertia was observed in the shallow thinking conditions, F (1, 116) = 14.13, p < .01, g2 = .11, but not in the deep thinking conditions, F (1, 116) = 1.05, p = .31, g2 = .01. Next, the currently experienced regret data were analyzed. The ANOVA revealed only a main effect for difference in attractiveness, F (1,16) = 6.19, p < .02, g2 = .05. Participants in the large difference conditions indicated more regret (M = 5.83, SD = 2.42) than those in the small difference conditions (M = 4.70, SD = 2.56). The other effects were not significant, Fs < 0.3. Then, the effects on anticipated regret were analyzed. The effects on regret revealed only a main effect for difference in attractiveness, F (1,16) = 7.83, p < .01, g2 = .06. Participants in the large difference conditions anticipated more regret (M = 6.27, SD = 2.65) than those in the small difference conditions (M = 4.95, SD = 2.60). The other effects were not significant. Finally, the valuation data were analyzed. The ANOVA revealed a main effect of difference in attractiveness, F (1, 116) = 6.21, p < .02, g2 = .05. Participants in the large difference conditions placed less value on the second offer (M = 5.97, SD = 3.48) than those in the small difference conditions (M = 7.21, SD = 2.49). There was also a main effect of thinking, F (1, 116) = 20.12, p < .001, g2 = .15, such that participants in the deep thinking conditions put a higher value on the second opportunity (M = 7.72, SD = 2.48) than participants in the shallow thinking conditions (M = 5.47, SD = 3.23). The significant interaction effect, F (1, 116) = 7.24, p < .01, g2 = .06, shows that a difference in valuation between the large and small difference con-
97
ditions was observed in the shallow thinking conditions, F (1, 116) = 13.43, p < .01, g2 = .10, but not in the deep thinking conditions, F (1, 116) = .02. This pattern of results closely matches the pattern found for action likelihood. Mediation analyses The central question of this experiment is whether inaction inertia is caused by regret, anticipated regret or valuation. The data presented so far are most consistent with the explanation in terms of valuation, because the ANOVA for valuation showed the same effects as the ANOVA for action likelihood. However, we did not yet provide an adequate test of whether valuation (or anticipated or experienced regret) really mediated the effects on action likelihood. To test for mediation, a series of regression models were estimated (cf. Baron & Kenny, 1986). The results are presented in Table 5. The predictor variables in the models were the two manipulated factors and their interaction (Difference, Thinking, and Difference · Thinking). These were coded using effect coding. The hypothesized mediators were experienced regret, anticipated regret and valuation. Action likelihood served as the dependent variable. To examine the degree of mediation we first regressed the mediators on the predictor variables (columns 1–3). Next, we regressed the dependent variable on the predictor variables (column 4). Columns 1–4 simply replicate the ANOVAs reported earlier. Finally, we regressed the dependent variable on both the predictor variables and the mediators, separately for each potential mediator (columns 5–7). The degree to which the influence of the predictor variables (Difference, Thinking, and Difference · Thinking) on the dependent variable (Action Likelihood) is reduced when accounting for the influence of the mediators (Experienced Regret, Anticipated Regret, and Valuation) expresses the degree of mediation. The results in Table 5 (columns 5–7) clearly show that the effects of the predictor variables on the action likelihood are only affected by the inclusion of valuation. Sobel tests (1982) confirmed that the effects of both
Table 5 Mediation results of Experiment 4 Predictor variables
Difference Thinking Difference · Thinking Experienced regret Anticipated regret Valuation R2
Mediators
Dependent variable Mediation tests Action likelihood Experienced Anticipated Valuation Action likelihood Action likelihood Action likelihood (no mediators) regret regret (with experienced (with anticipated regret) (with valuation) regret) .22* .05 .09
.25** .09 .15
.20* .37*** .22**
.28*** .35*** .16 —
.21** .34*** .13 .31***
.23** .33*** .13
.16* .13 .02
.18* .06
.09**
.22***
.23***
.31***
Note. Standardized b-coefficients are reported. p < .06, *p < .05, **p < .01, ***p < .001.
.25***
.61*** .51***
98
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
manipulated factors and their interaction are significantly mediated by valuation, ZDifference = 2.38, p < .02, ZThinking = 3.94, p < .0001, ZDifference·Thinking = 2.56, p < .02. Discussion This experiment again provides data that are not in line with the explanation of inaction inertia in terms of regret experienced over the first offer. The results are also inconsistent with an explanation in terms of anticipated regret. We manipulated the amount of thinking and the difference in attractiveness between the first and second offer. For both anticipated regret and experienced regret we only found a main effect for difference in attractiveness. We found a markedly different pattern of results for the likelihood to act on the second offer. For this variable we found two main effects and an interaction, similar to the effects we obtained for valuation. This indicated that only devaluation (and not anticipated or experienced regret) could potentially cause the effects on action likelihood. Mediation analyses supported this interpretation.
Experiment 5 The results of Experiment 4 suggest that devaluation and not regret causes inaction inertia: the initial offer seems to work as an anchor against which the current offer is evaluated. However, it should be noted that in several experiments the notion of a simple anchoring effect has been studied and refuted. Both Tykocinski et al. (1995, Experiment 3) and Arkes et al. (2002, Experiment 1) found no inaction inertia when the original offer was in a far-away city. For example, when the missed offer of a very cheap ski-pass was at another location than the currently offered ski-pass (that was more expensive), no effect of difference in attractiveness on action likelihood was found. This seems to suggest that inaction inertia is not caused by a simple anchoring effect: the price at the other location apparently did not function as an anchor against which the current offer was valued. What it does suggest, in our view, is that a simple anchor is not enough, but that the anchor needs to be relevant to the current offer. There may be many reasons why a ski-pass at another location is cheaper, and this might not necessarily lead to a devaluation of the skipass currently under consideration. However, when there are no reasons why the earlier offer was so much better than the present one, devaluation of the offer might occur. Note that this is consistent with our findings in Experiment 3: the uniqueness of the offer was a clear reason why the offer was so good, and no inaction inertia was obtained. Some further evidence for this can
be found in Butler and Highhouse (2000), who found that action likelihood was higher (and inaction inertia was reduced) when the second offer came from a different source than the first offer. The reason, we suggest, is that there may be good reasons why an offer from someone else is worse, and devaluation of the subsequent offer does not occur. Unfortunately, valuation was not measured in either study, precluding a direct test of this reasoning. In our final test of whether experienced regret, anticipated regret and/or valuation are causes of inaction inertia we thus built on the research by Butler and Highhouse (2000) and manipulated whether the second offer comes from the same source or from a different source. In addition we considered another manipulation of the characteristics of the initial and second offer. Half of our participants were first offered an attractive trip to Paris, which they did not take, and was next offered a less attractive trip to Paris. The other half, however, were offered a trip to Berlin after missing the initial trip to Paris. If our reasoning is correct, inaction inertia should be significantly lower when the characteristics of the second offer are different from the first, and when the offer source is different. In that situation, the second offer may not be devaluated. Of course, valuation should also mediate the effects on action likelihood. Method Participants and design Participants were 159 students (65 male, 94 female, Mage = 22 years) of the University of Amsterdam, who participated for course credit. The study had a 2 (Difference in Attractiveness: Small vs. Large) · 2 (Offer Source: Same vs. Other) · 2 (Offer: Same vs. Other) complete factorial between-participants design. Procedure and materials Participants were invited to participate in a larger experimental session of which the current study was part. Upon arrival in the laboratory they were placed in separate cubicles. The presentation of the scenarios and the registration of the participants’ responses were all done via a computer. The basic text of the scenario was as follows in the large [small] difference in attractiveness condition: You enjoy traveling very much. Both Paris and Berlin are attractive destinations for you. Next week you have a whole week without lectures and you have not yet planned anything. A friend of you calls you and informs you about a completely organized three-day trip to Paris that is available from travel agency Student Travel up until tomorrow for €75 [€130] instead of the usual €150. This trip includes traveling with the Thalys (High speed train) and two nights including breakfast at a two
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
star hotel in the center of Paris. He asks you to join him next week. Although it sounded like a good idea, you postponed going to the travel agency, and you were too late; it was no longer possible to book the trip. Participants next rated experienced regret. After that they read another text. In the same source, same offer condition, the text was as follows: ‘‘Your friend calls again, and tells you that, although you have missed the offer, it is still possible to book the same trip at travel agency Student Travel for €140.’’ In the same source, different offer condition, they read: ‘‘Your friend calls again, and tells you that, although you have missed the offer, it is still possible to book another trip with Student Travel. This trip has Berlin as its destination, and is as attractive as the trip to Paris. The trip costs €140.’’ In the conditions with a different source, participants read an almost identical text, with the difference that the second offer did not come from Student Travel, but from another agency, called De Jong Travel. Next they rated action likelihood, anticipated regret, and valuation of the second offer (depending on conditions, the trip to Paris or to Berlin), as in the previous experiments. Results and discussion Action likelihood The results of this experiment are shown in Table 6. All items were analyzed with 2 (Difference in Attractiveness) · 2 (Offer Source) · 2 (Offer) ANOVAs. The ANOVA performed on action likelihood yielded a main effect of difference in attractiveness, F (1, 151) = 47.48, p < .001, g2 = .24. Participants in the large difference condition (M = 5.46, SD = 2.68) were much less likely to accept the second offer than those in the small differ-
99
ence condition (M = 8.03, SD = 2.01). The main effect of offer source was marginally significant, F (1, 151) = 2.84, p = .09, g2 = .02, and participants in the same source condition were somewhat less likely to accept the second offer (M = 6.44, SD = 3.00) than those in the different source condition (M = 7.06, SD = 2.31). The main effect of offer was not significant, F < 1. However, the interaction between source and offer was significant, F(1, 151) = 4.09, p < .05, g2 = .03. The difference between the two sources was not significant when both offered the same trip as the initial offer (i.e., to Paris; same source/same offer: M = 6.68, SD = 3.05; different source/same offer: M = 6.55, SD = 2.66). But there was a difference when the offer was different from the initial offer, and participants were more likely to accept the trip to Berlin when it was offered by a different source (M = 7.59, SD = 1.77) than when it was offered by the same source (M = 6.20, SD = 2.97). Thus, the inaction inertia effect disappeared when a different travel agency offered a different trip. The other interactions were not significant, Fs < 1.80, ps > .15. Regret Ratings of experienced regret after missing the first offer were not affected by any of the manipulations, all Fs < 1.80, ps > .15. Thus, there was no main effect of difference in attractiveness, F < 1, and regret cannot explain the effect of difference in attractiveness on action likelihood. The other manipulations could not have affected experienced regret, because they were introduced after regret had been measured. The ratings of anticipated regret showed a main effect of difference in attractiveness, F (1, 151) = 28.84,
Table 6 Ratings on the dependent variables as a function of condition in Experiment 5 Dependent variables
Same offer (Paris)
Different offer (Berlin)
Same source (Student Travel)
Different source (De Jong Travel)
Same source (Student Travel)
Different source (De Jong Travel)
M
SD
M
SD
M
SD
M
SD
Likelihood of action Small difference Large difference
8.05 5.30
(2.60) (3.16)
8.00 5.10
(2.60) (2.43)
7.80 4.60
(2.33) (2.70)
8.25 6.89
(1.41) (1.88)
Regret over first offer Small difference Large difference
6.90 7.75
(2.57) (2.90)
7.55 7.65
(2.46) (1.60)
8.05 7.70
(1.50) (2.23)
8.00 7.89
(1.45) (1.94)
Anticipated regret Small difference Large difference
2.10 4.90
(2.05) (3.55)
2.30 4.10
(2.25) (3.00)
2.60 4.55
(1.73) (2.86)
1.65 3.89
(1.60) (3.20)
Valuation Small difference Large difference
8.75 6.20
(2.31) (3.64)
8.95 5.20
(1.19) (3.01)
7.95 5.10
(2.28) (2.71)
7.95 7.21
(2.24) (2.44)
Note. Ratings were made on a 11-point scales, with endpoints labeled not likely (0) and very likely (10), no regret (0) and very much regret (10), and €75 (0) and €150 (10).
100
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
Table 7 Mediation results of Experiment 5 Predictor variables
Mediators Anticipated regret
Difference Offer · Source Anticipated regret Valuation R2
Valuation
Dependent variable Action likelihood (no mediators)
Mediation tests Action likelihood (with anticipated regret)
.40*** .05
.43*** .13
.48*** .14*
.38*** .13 .25***
.16***
.20***
.25***
.30***
Action likelihood (with valuation) .16** .05 .74*** .69***
Note. Standardized b-coefficients are reported. p < .10, *p < .05, **p < .01, ***p < .001.
p < .001, g2 = .16. Anticipated regret was higher with a large difference in attractiveness (M = 4.37, SD = 3.07) than with a small difference (M = 2.16, SD = 1.92). No further effects were found, Fs < 1.90, ps > .15. Valuation The offer valuation data (i.e., how much participants were prepared to pay for the second offer) revealed a main effect of difference in attractiveness, F (1, 151) = 36.90, p < .001, g2 = .20. The second offer was valued less in the large difference condition (M = 5.91, SD = 3.06) than in the small difference condition (M = 8.40, SD = 2.07). The main effects of source and offer were not significant, F < 1. There was no interaction between difference and source, F < 1, but there was a marginally significant source · offer interaction, F (1,151) = 3.20, p = .08, g2 = .02. This interaction was further qualified by a difference · source · offer interaction, F (1, 151) = 4.14, p < .05, g2 = .03. It appears that the offer · source interaction only occurred in the large difference in attractiveness condition, F (1, 75) = 5.33, p = .02, but not in the small difference condition, F = 0.05, ns. Mediation analyses There were two effects on action likelihood, one of difference in attractiveness and one of the interaction between source and offer. We tested whether these two effects were mediated by anticipated regret and offer valuation, using again the Baron and Kenny approach (1986). See Table 7 for the results. The predictor variables in the models were the difference in attractiveness and the interaction between offer and source (Difference, Offer · Source). The mediators were anticipated regret and valuation (the ANOVA results already indicated that experienced regret could not be a mediator). Action likelihood served as the dependent variable. To examine the degree of mediation we first regressed the mediators on the predictor variables (columns 1 and 2). Next, we regressed the dependent variable on the predictor variables (column 3). Columns 1–3 replicate the ANOVAs reported earlier. Finally, we regressed the dependent variable on both
the predictor variables and the mediators (columns 4 and 5). The degree to which the influence of the predictor variables (Difference, Offer · Source) on the dependent variable (action likelihood) is reduced when accounting for the influence of the mediators (anticipated regret and valuation) expresses the degree of mediation. The results in Table 7 (columns 4 and 5) clearly show that the effects of the predictor variables on the action likelihood are affected by the inclusion of valuation and not by anticipated regret. Sobel tests (1982) indicated that the effects of both manipulated factors and their interaction are mediated by valuation, ZDifference = 6.08, p < .001, ZOffer·Source = 1.73, p < .09. In a final test, action likelihood was regressed on all these variables. The explained variance was identical to the previous model (shown in column 5), in which anticipated regret was not used and the effect of anticipated regret was not significant anymore. In conclusion, offer valuation mediated the effects and the inclusion of anticipated regret did not increase the explained variance.
General discussion Foregoing an attractive opportunity can lead to a decreased likelihood of action in the same domain. Why should this be the case? Different researchers have argued for a causal role of regret. Because one regrets missing the first opportunity, the later (and less attractive) opportunity is rejected without giving it much thought, to escape from this negative emotion. However, in a series of five experiments we could not obtain evidence consistent with the idea that the experience of regret causes inaction inertia. In Experiment 1, a manipulation that strongly increased regret (i.e., responsibility) did not influence the occurrence of inaction inertia. In Experiment 2, a manipulation that affected inaction inertia (time difference between the first and second offer) did not affect regret. In Experiment 3, a manipulation of uniqueness of the initial offer even increased regret while at the same time decreased inaction inertia.
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
Also in Experiments 4 and 5 no support was found for a causal role of experienced regret. In Experiment 4, thinking shallow or deeply influenced inaction inertia and not regret, while in Experiment 5 difference in attractiveness affected inaction inertia but not experienced regret. In our opinion, these findings seriously question an explanation of inaction inertia solely in terms of the experience of regret. In Experiments 4 and 5 we also studied the explanation of inaction inertia in terms of anticipated regret. The reasoning was that people anticipate that they will regret taking the later offer, because it will remind them of missing the earlier opportunity (‘‘I could have had the same, but much cheaper’’). We found no evidence that anticipated regret mediated between our manipulation of difference in attractiveness and inaction inertia. At the same time our results provide support for an alternative explanation of inaction inertia. As Arkes and colleagues (2002) have argued, the initial offer can function as an anchor against which the current offer is judged. People may, as a consequence, devalue the current offer. For example, when offered a couch (our Experiment 2) or a vacation trip (Experiments 3–5) people may think that the current price is above the actual value of the product (‘‘Why else was it so cheap previously?’’). As a consequence, they decline the second opportunity. In Experiment 3 we found that when the initial offer consisted of a unique opportunity (the 25th anniversary of the travel agency), the inaction inertia effect was very much reduced, as compared to when the offer was not unique. This finding is consistent with the devaluation explanation, as it can be expected that a unique offer does not cause devaluation. However, because we did not measure valuation in that study, we did not have direct evidence to back this up in our Experiment 3. We measured valuation in Experiments 4 and 5, and obtained evidence showing that valuation was an important mediator of the inaction inertia effect. Indeed, in conditions in which the valuation of the second offer was lowest, action likelihood was lowest as well. In Experiment 4, we only found a significant inaction inertia effect in the condition with a large difference between the initial and later offer and in which participants did not think deeply. Valuation of the second offer was low there as well. Presumably, when thinking more deeply about the current offer, people come to see it as it is: an attractive opportunity. This may therefore attenuate the somewhat ‘irrational’ devaluation of the second offer, which may no longer be considered a loss as compared to the previous offer, but as a gain as compared to the normal situation (see for similar reasoning, Tykocinski et al., 1995, Experiment 6). In Experiment 5 we reasoned that another offer (the first offer was a trip to Paris, the second was to Berlin)
101
would not be devalued and therefore not result in inaction. Instead, however, we found that inaction inertia only disappeared when both the offer and the source of the offer (the travel agency) were different from the original offer. However, the same pattern of results was observed in the valuation data, and valuation mediated this interaction effect. This suggests that the more dissimilar the original and later offer are, the less the later offer is devalued, and the less likely it will be that inaction inertia will be observed. Reevaluating the regret explanations Where does this bring us with respect to the regret explanation? In defense of the regret explanation, one might offer several alternative explanations for some of our experiments. For example, one could argue that the manipulations of time interval in Experiment 2 and of offer type (unique or not) in Experiment 3 may at the same time have influenced expectations concerning the likelihood of future discounts. When a discount was missed some time ago one may think that discounts are not likely (but also note that when a discount was very recent, people may infer that it will take some time before the product will be discounted again). Also, when a missed discount was an exceptional offer, one may infer it is unlikely that a similar discount will occur in the near future. The effects of our manipulations on action likelihood may then simply be due to participants’ anticipations of these discounts. Also, one could argue that our manipulation of deep versus shallow thinking somewhat resembles manipulations of accountability (e.g., Tetlock, 1992) since participants are asked to write down reasons for their choice. Consistent with this reasoning, one may feel more accountable for one’s decisions (as measured via the action likelihood question) than for one’s emotion. This is indeed the pattern of results we obtained in Experiment 4 (and consistent with the reasoning we put forward in the introduction of Experiment 4). Importantly, however, although many of these alternative explanations may be consistent with some of our experiments we could not find a single alternative explanation that casts doubt on the whole series of five experiments. Moreover, we would like to argue that even if these alternative explanations would prove to be valid ones, the results remain unsupportive for a strong version of the regret explanation (i.e., that regret is the sole cause inaction inertia). Another as yet unresolved issue is the question as to how we can reconcile the present results with the fact that inaction inertia is sometimes found to be correlated with regret (e.g., in the research of Arkes et al. (2002) & Kumar (2004) that we discussed in the introduction). An interesting possibility, in our view, is that regret is not a cause of inaction inertia, but rather a consequence of the
102
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
devaluation of the second offer that occurs after having missed the first attractive offer. The reasoning is simple: when one thinks that the price of an offer is above its actual value, one will decide not to take it. When asked, ‘‘How would you feel when you take the offer anyway,’’ people are likely to anticipate regret. Of course: one would then accept a bad offer! We do not know how such a relation would have explanatory power for the studies of Arkes et al. (2002) and Kumar (2004). What we do know, however, is that the current data are consistent with this reasoning. We found a significant negative correlation between valuation and anticipated regret in both Experiment 4 (r = .45, p < .001) and Experiment 5 (r = .43, p < .001). Thus, devaluation of the second offer may have an effect on both action likelihood and on anticipated regret. This would create a spurious correlation between anticipated regret and action likelihood that would disappear when controlling for valuation. Indeed, in our data from Experiments 4 and 5 the partial correlations that controlled for valuation were not significant. Does this mean that there is no evidence whatsoever supporting the regret explanation? Of course we cannot make this claim, and it is not our intention to do so. There are at least two studies in which an explanation in terms of an anticipated regret mechanism seems very plausible (Experiments 1 and 2 of Tykocinski & Pittman, 1998). In these experiments it was manipulated whether or not forgoing the second opportunity protected the decision maker from a confrontation with missing the first (i.e., this feedback was avoidable or unavoidable). Tykocinski and Pittman found that when people know that, regardless of what they do, they will be constantly reminded of their initial failure to act, the inaction inertia effect was attenuated. This pattern of results fits quite nicely with the anticipated regret explanation in the sense that decision makers only seem to forgo a current opportunity when it shields them from regret causing feedback. Unfortunately, these studies did not include any measurements of regret. Taken together, we want to make explicit that, although we did not find evidence for a regret based explanation in the present experiments, we cannot rule out that under certain circumstances regret (or its anticipation) may influence the occurrence of inaction inertia. It may be the case that the inaction inertia effect, like many other complex psychological phenomena, is overdetermined in the sense that it may actually stem from a set of different mechanisms or processes that operate jointly. These mechanisms or processes may each be sufficient but not necessary causes for inaction inertia to occur. This would imply that our studies, like any other study, do not provide a definitive answer to question concerning ‘‘the’’ cause of inaction inertia. What the present research has shown, however, is that relative
contribution of regret to this effect may not be that large. Reevaluating the valuation explanation Although we feel that our results clearly imply that valuation provides an interesting explanation of inaction inertia, we acknowledge the fact that it does not explain all the results in the field. For example, it is unclear how this explanation deals with the findings of Tykocinski and Pittman’s (1998) Experiments 1 and 2, which we discussed above. Another problem with the devaluation explanation is that it seems limited to prices, at least that is how valuation was assessed in the current studies and in Arkes et al. (2002). This may be problematic since there are also inaction inertia studies in which the difference in attractiveness was not manipulated via different monetary values, but rather in spatial distances (e.g., a longer distance to the fitness club), temporal differences (see our Experiment 3) or free air miles (see Tykocinski et al., 1995). Clearly, strictly monetary devaluation does not apply here. In addition, it should be noted that some earlier results seem inconsistent with the valuation approach, in the sense that the valuation effects can be seen as contrast effects. Tykocinski et al. (1995) explicitly argued that inaction inertia is not caused by a simple contrast effect. In their Experiment 3 they showed that when the missed opportunity was at another location (in this case: the much cheaper ski-pass had been available at another ski resort far away), the inaction inertia effect was not obtained. Arkes et al. (2002, Experiment 1) report a similar finding. Thus, in these experiments either the valuation explanation does not hold, or the second offer was not devaluated. We suspect that the latter occurred. We believe that the devaluation process is indeed not a simple contrast effect, but rather a more or less reasoned process in which people try to make sense of the price difference between the first and second offer. We expect that a missed discount may result in devaluation of the current offer, when people consider the price of the missed opportunity informative about the current opportunity. But, when for some reason the difference in attractiveness is not informative or in some way justified (i.e., when there are plausible reasons why the earlier offer was so much better than the current offer) devaluation will not take place. Put differently, we suspect that a more complex contrast effect could underlie of inaction inertia. This contrast effect only occurs when the missed discount is used as an informative reference point. There may be several factors determining whether or not the attractiveness of the missed opportunity is informative. The attractiveness of the missed opportunity may be less informative when the current and the missed opportunity are not very similar (e.g., at another location, as in the
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
studies by Tykocinski et al., 1995, Experiment 3; and Arkes et al., 2002, Experiment 1; or when it is offered by a different source, as in Butler & Highhouse, 2000; and our Experiment 5) or when the missed discount is not informative about the ‘‘normal’’ price (e.g., because of differences in price level between locations, or because the initial price was the result of the 25th anniversary of the shop, as in our Experiment 3). Hence, there may be several conditions under which the missed discount will not be used as a reference point. Consistent with this more complex contrast effect are our findings of Experiment 4, in which reasoning deeply took away both devaluation and inaction inertia. Furthermore, previous research has shown that when reasons are given why the first offer was so cheap, the inaction inertia effect is weakened: Tykocinski and Pittman (1998, Experiment 4) obtained less inaction inertia when information suggested that the first offer was not so attractive after all (in this case: the low quality of the guide on a tour to Italy). This more reasoned process explanation that we put forward here, may also overcome the fact that the devaluation explanation only applies to prices. Our reasoning does also apply to non-monetary cases. It is not necessarily price that matters, but the judged attractiveness of the current offer. Thus, in the earlier mentioned fitness center scenario, people may ask themselves ‘‘how attractive is a fitness center located 30 min away?’’ and judge it to be less attractive if they previously had the opportunity to become a member of a center located only 5 min away. The same reasoning applies to, for example, free air miles, bets, or any other opportunity that can vary in attractiveness. Of course, we need to stress that our ideas about the devaluation process are still embryonic—these are post hoc interpretations—and hence it should be handled with care. Future research could investigate the validity of our reasoning and contrast it with other possible mechanisms for devaluation.3 It may, for example, also be that decision makers infer from their earlier failure to act that they actually do not value the current offer. This knowledge about their previous inaction may thus operate more or less as a heuristic or shortcut for determining the value of the subsequent offer. Although we believe that this is a plausible alternative explanation, we should also acknowledge that this explanation is at odds with the results of our Experiment 1 and those of Tykocinski et al. (1995, Experiments 4 and 5). In these studies it is found that inaction inertia occurs whether or not decision makers are responsible for missing the first action opportunity. If we were to use our first inaction as somehow informative for our valuation of the subsequent opportunity, we believe this first inaction
3 We thank an anonymous reviewer for stimulating us to address the different potential mechanisms underlying devaluation.
103
should have more impact when the behavior is freely chosen. Alternatively, one may argue that devaluation stems from being frustrated about missing the first opportunity. In such a sour grapes explanation, ‘‘preferences underlying a choice may be shaped by the constraints’’ (Elster, 1983, p. vii). Decision makers may comfort themselves with the idea that the missed opportunity (and therefore also the current opportunity, if it resembles the missed opportunity) was not that great after all. Future research may want to disentangle these different mechanisms and isolate their individual contribution to the phenomenon of devaluation. A final remark pertains to our use of scenario studies. Our primary aim was to further the understanding of the causes of inaction inertia. For this purpose we stayed close to the setup used, measures used, and findings obtained in previous studies on inaction inertia. This approach has the advantage of facilitating the comparison of our current findings with previous findings. Of course one might posit the more general question of whether the typical inaction inertia findings would also be replicated using different research paradigms, for example by presenting participants with a ‘‘real’’ decision context. Although for the present purposes—that so strongly relied on strict comparison with previous findings—we did not opt for such an approach of generalization, we do feel that it might be useful for future research to investigate actual decisions. In conclusion, the present research has thoroughly tested the explanation of inaction inertia in terms of regret. Both the escape from current regret and the avoidance of future regret version of the regret explanation appeared to be inconsistent with the data of our five experimental studies. We did find evidence consistent with the explanation in terms of a devaluation of the second offer, but noted in the general discussion that a more reasoned decision process might underlie the devaluation of the second offer. Future research may want to test this explanation directly. But, to return to the question that motivated our inquiry into the causes of inaction inertia, the current findings suggest that regret is not a crucial factor in the causation of inaction inertia.
References Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35, 124–140. Arkes, H. R., Kung, Y., & Hutzel, L. (2002). Regret, valuation, and inaction inertia. Organizational Behavior and Human Decision Processes, 87, 371–385. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182.
104
M. Zeelenberg et al. / Organizational Behavior and Human Decision Processes 101 (2006) 89–104
Brock, T. C. (1968). Implications of commodity theory for value change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychological foundations of attitudes (pp. 243–275). New York: Academic Press. Butler, A., & Highhouse, S. (2000). Deciding to sell: The effect of prior inaction and offer source. Journal of Economic Psychology, 21, 223–232. Elster, J. (1983). Sour grapes: Studies in the subversion of rationality. Cambridge, UK: Cambridge University Press. Freedman, J. L., & Fraser, S. C. (1966). Compliance without pressure: The foot-in-the-door technique. Journal of Personality and Social Psychology, 4, 195–202. Kumar, P. (2004). The effects of social comparison on inaction inertia. Organizational Behavior and Human Decision Processes, 95, 175–185. Ordo´n˜ez, L. D., & Connolly, T. (2000). Regret and responsibility: A reply to Zeelenberg et al., 1998. Organizational Behavior and Human Decision Processes, 81, 132–142. Pieters, R., & Zeelenberg, M. (2005). On bad decisions and deciding badly: When intention-behavior inconsistency is regrettable. Organizational Behavior and Human Decision Processes, 97, 18–30. Ratner, R. K., & Herbst, K. C. (2005). When good decisions have bad outcomes: The impact of affect on switching behavior. Organizational Behavior and Human Decision Processes, 96, 23–37. Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociological methodology (pp. 290–312). San Francisco: Jossey-Bass. Tetlock, P. E. (1992). The impact of accountability on judgment and choice: Toward a social contingency model. In M. Zanna (Ed.). Advances in experimental social psychology (Vol. 25, pp. 331–376). New York: Academic Press. Tykocinski, O., Israel, R., & Pittman, T. S. (2004). Inaction inertia in the stock market. Journal of Applied Social Psychology, 34, 1166–1175. Tykocinski, O. E., Pittman, T. S., & Tuttle, E. E. (1995). Inaction inertia: Foregoing future benefits as a result of an initial failure to act. Journal of Personality and Social Psychology, 68, 793–803. Tykocinski, O. E., & Pittman, T. S. (1998). The consequences of doing nothing: Inaction inertia as avoidance of anticipated
counterfactual regret. Journal of Personality and Social Psychology, 75, 607–616. Tykocinski, O. E., & Pittman, T. S. (2001). Product aversion following a missed opportunity: Price contrast or avoidance of anticipated regret? Basic and Applied Social Psychology, 23, 149–156. Tykocinski, O. E., & Pittman, T. S. (2004). The dark side of opportunity: Regret, disappointment, and the cost of prospects. In I. Brocas & J. D. Carillo (Eds.). The psychology of economic decisions. Volume 2: Reasons and choices (pp. 179–196). Oxford: Oxford University Press. Van Dijk, E., & Zeelenberg, M. (2005). On the psychology of ‘if only’: Regret and the comparison between factual and counterfactual outcomes. Organizational Behavior and Human Decision Processes, 97, 152–160. Zeelenberg, M. (1999). Anticipated regret, expected feedback and behavioral decision-making. Journal of Behavioral Decision Making, 12, 93–106. Zeelenberg, M., Inman, J. J., & Pieters, R. G. M. (2001). What we do when decisions go awry: Behavioral consequences of experienced regret. In E. U. Weber, J. Baron, & G. Loomes (Eds.), Conflict and tradeoffs in decision making (pp. 136–155). Cambridge, MA: Cambridge University Press. Zeelenberg, M., & Pieters, R. (1999). On service delivery that might have been: Behavioral responses to disappointment and regret. Journal of Service Research, 2, 86–97. Zeelenberg, M., Van Dijk, W. W., & Manstead, A. S. R. (1998). Reconsidering the relation between regret and responsibility. Organizational Behavior and Human Decision Processes, 74, 254–272. Zeelenberg, M., Van Dijk, W. W., & Manstead, A. S. R. (2000). Regret and responsibility Resolved? Evaluating Ordo´n˜ez and Connolly’s (2000) conclusions. Organizational Behavior and Human Decision Processes, 81, 143–154. Zeelenberg, M., & Van Putten, M. (2005). The dark side of discounts: An inaction inertia perspective on the post-promotion dip. Psychology and Marketing, 22, 611–622.