Probability of Regret: Anticipation of Uncertainty Resolution in Choice

result’’ in the sense that one would be informed of the In choice between uncertain options, uncertainty outcome obtained for the chosen alternative: The value may be eventually resolved for all options, or just for of the stock one invested in or the data related to effithe chosen option. The findings described here demonciency subsequent to implementation of the efficiency strate that preference between uncertain options is measure. However, the two decision problems differ systematically affected by expectations concerning with respect to the extent of uncertainty resolution. In uncertainty resolution: Prospects which are less likely stock investment, one invariably knows both the value to yield worse outcome than the alternative are seof the chosen share and those of the alternative shares lected more often when all options are to be resolved one had considered. In implementing procedural (complete resolution). Two experiments examined changes designed to achieve better efficiency, one selchoice between statistically independent lotteries for which the chances of being worse off with the low-risk dom knows for certain what would have happened had alternative were nearly as high, or higher than the one chosen differently. The present study is designed chances of being worse-off with the high-risk alternato examine the hypothesis that anticipation of uncertive. For these gamble pairs it was found that subjects tainty resolution for all alternatives in the choice set, as who expected complete resolution chose the high-risk opposed to resolution of the chosen option only, affects high-gain option more often than subjects who expreferences among some options, due to the probability pected only their selected gamble to be resolved. A of regret. third experiment suggests that the preference changes Recent research provides some evidence for experidue to resolution are related to the probability of the enced regret, associated with comparing the outcomes low-risk gamble yielding the worse obtained outcome. The results are interpreted as compatible with the hyof the chosen and unchosen alternatives. Boles and pothesis that in choosing between uncertain options, Messick (1995) show that the judged goodness of an people are more inclined to compare outcomes per outcome is closely related to the alternative outcome. states of the world if they expect to learn what would Thus, a winner in one gamble was judged to be less have happened with each option. In those cases, the satisfied than a loser in another gamble, if the winner outcome of a foregone option is more likely to serve would have won more with the alternative, and the as a salient comparison point for evaluating the outloser would have lost more. Similarly, Baron and Ritov come of the selected option. An alternative explana(1994) found that comparison to the alternative outtion is also discussed. q 1996 Academic Press, Inc. come affected rating of the decision maker’s emotions, as well as judgment of the choice’s advisability. In parMany decision problems one faces, in real life as well ticular, the bias toward omission was greater when the as in laboratory experiments, involve a choice between act led to the worse of the alternative outcomes, even two or more uncertain alternatives. Whether one when this outcome represented a gain relative to the chooses to invest in a specific share in the stock market, previous status quo. or one makes a managerial decision in favor of applying If people are aware of the possibility of experiencing one efficiency measure over another, one normally exregret, it seems plausible to assume that they would pects to find out what the result of the choice will be. take this into consideration when making decisions Indeed, in both these examples, one would ‘‘know the whose outcomes are uncertain. Indeed, according to Regret Theory, anticipation of the expost experience afThe author thanks Jonathan Baron, Jane Beattie, Barbara Mellers, fects choice between two uncertain alternatives (Bell, and an anonymous reviewer for helpful comments. Address corre1982; Loomes & Sugden, 1982). The decision process spondence and reprint requests to Ilana Ritov, Department of Industhrough which anticipated regret is assumed to affect trial Engineering and Management, Ben-Gurion University, P.O.B. choice entails that for each decision problem a juxtapo653, 84105 Be’er Sheva, Israel. Fax: (972-7) 280-776. E-mail: ritov@b gumail.bgu.ac.il. sition of consequences in an action/state matrix can be