Prospect Theory

Prospect theory (Kahneman & Tversky, 1979) predictions were examined in light of ethnocentrism and intergroup conflict. An experiment conducted at the outset of the 2003 invasion of Iraq by the U.S., U. K. and their allies explored American and British participants’ preferences for certain versus uncertain gains and losses concerning Iraqi, American, and British lives. In four conditions, participants showed the usual loss aversion when deciding between options that only affected Iraqi lives. Six other conditions examined choices between the lives of Americans, Britons, or Iraqis. Strong ethnocentric biases rather than risk-aversion occurred. Participants preferred policies that prioritized their own nationals’ and allies’ lives over Iraqi lives. War-related and other attitudes corresponded to participants’ decisions. The need to expand prospect theory to address intergroup relations is discussed. Author Note: Please address correspondence to Dr. Felicia Pratto, Department of Psychology, University of Connecticut, Storrs, CT 06269-1020, USA, [email protected], (860)-486-4772. Weighing the Prospects of War Policy decisions often require anticipating potential gains and losses for different groups of people. Prospect theory (e.g., Kahneman & Tversky, 1979) describes how people weigh potential gains and losses, which are called “prospects,” starting from one’s reference point or the status quo. According to prospect theory, gains are more appealing than losses, but a potential loss is even more aversive than a potential gain of the same magnitude is appealing. For example, losing € 5 feels more painful than finding € 5 feels enjoyable. One consequence of this asymmetry between preferences for potential losses and gains is that when people assume their outcomes will be losses, they are more willing to take risks, whereas when they assume their outcomes will be gains, they prefer certainty. In a classic demonstration of this risk-preference reversal, Tversky and Kahneman (1981) asked American participants to choose between two health policies designed to treat an outbreak of an unusual Asian disease that was expected to strike the U.S. and kill 600 people. The expected consequences of one policy were described as certain and fixed, whereas the expected consequences of the other policy were described in probabilistic terms as uncertain. Betweenparticipants, the choices were either framed as losses, by describing how many people would die as a result of the policy, or framed as gains, by describing how many people’s lives would be saved. For example, in the gain frame condition, the certain choice was described as saving exactly 200 lives, and the uncertain choice was described as having a 2/3 chance of saving 600 lives and a 1/3 chance of saving no lives. All four choices (two within the loss and gain conditions) had the same expected value (of 400 deaths and 200 lives saved). However, in the loss condition, 78% chose the uncertain loss over the certain loss, whereas in the gain condition, 72% chose the certain gain over the uncertain gain. Such preferences for uncertain losses and certain gains have been shown when people consider real and fictive monetary consequences for Prospect Theory and Ethnocentrism 2 themselves (e.g., Fennema & Wakker, 1997), access to medical care for others (Levin & Chapman, 1990, 1993), other people’s lives (Fagley & Miller, 1997), and public policies and political candidates (e.g., Quattrone & Tversky, 1988). The robustness of risk-preference reversals has led political scientists to recognize the utility of applying prospect theory to international relations (e.g., Berejikian, 2002; Farnham, 1992; Jervis, 1992; McDermott, 1998). For example, McDermott and Kugler (2001) used prospect theory to analyze U.S. President G.H.W. Bush’s 1991 Gulf War decisions. They reasoned that Bush’s desire to restore a loss (i.e., the Iraqi take-over of Kuwait) motivated him to pursue military action against Iraq. But, they reasoned, the prospect of further risks (e.g., loss of more American lives, domestic opposition) was not sufficiently motivating to pursue an additional gain (i.e., control of Iraq), and this led Bush to curtail military action short of toppling the Iraqi government. Hence, prospect theory seems to hold promise for understanding intergroup relations, potentially explaining when war will be pursued or curtailed (for more detailed commentaries see Boettcher, 1995; Farnham, 1994; McDermott, 2004a, 2004b). However, such post hoc case studies do not rule out other explanations for policy-maker’s decisions (Boettcher, 1995; McDermott & Kugler, 2001). Another serious limitation of extant research on prospect theory is that neither experiments nor case studies sufficiently consider whose prospects matter to those making decisions and why. The terms “loss” and “gain” might be understood to be general (i.e., occurring to anyone). But nearly all experiments have examined prospects only for the participant or for a single group on whose behalf the participant acts (e.g., citizens of the participant’s nation). Notable exceptions to this trend were provided by Levin and Chapman (1990, 1993). These researchers found that Americans were less likely to show preference-reversals when considering prospects concerning Iranians rather than Americans and when considering prospects involving people who contracted AIDS due to their own behavior rather than those who contracted AIDS due to medical practice. Lewin and Chapman’s (1990, 1993) results show that prospect theory does not fully describe preferences for prospects that affect denigrated outgroups, and suggest that the psychology of group prejudice may provide refinements to, or limiting conditions for, prospect theory. Prospect theory has not incorporated considerations of psychological others because it assumes that reference points are self-referential. But people in other social categories such as “foreigners” may be treated sometimes as out-group members and sometimes as in-group members, as when they are re-categorized psychologically (e.g., Gaertner, Dovidio, & Bachman, 1996; Gaertner, Mann, Murrell, & Dovidio, 1989). For example, Esses, Dovidio, Jackson, and Armstrong (2001) improved Canadian citizen’s attitudes towards immigration by reminding them that immigrants shared either common ethnic roots or a common present nationality with the participants. To provide a stronger test of the hypothesis that psychological inclusion of others influences how lives are valued, the present experiment not only compared the prospects of denigrated outgroups with those of co-nationals, but also compared the prospects of allies with both denigrated outgroups and with co-nationals. If allies are valued more than denigrated outgroups but comparably with co-nationals, this would provide direct support for the notion that psychological inclusion of others influences the evaluation of a given group’s prospects. One prediction from the group prejudice literature, then, is that individuals’ prospects will be more highly valued to the extent that those individuals are psychologically included in the Prospect Theory and Ethnocentrism 3 participants’ own group. This is often termed ethnocentrism. The group prejudice literature also implies that the way that decision choices are presented may invoke group prejudice. In particular, considerable research has identified two conditions that lead to negative treatment of outgroups: 1) intergroup competition over valued resources (e.g., Sherif & Sherif, 1953; Campbell, 1972) and 2) intergroup threat, that is, the presumption that the outgroup poses a risk of loss for the ingroup (e.g., Greenberg et al., 1990; Jackson & Esses, 2000; Sears, 1988). Although intergroup competition and perceived intergroup threat often co-occur, they are psychologically distinct, because competition does not always imply the threat of a loss, and losses can occur for reasons other than competition. Consider two nations who compete over land that neither controls. Such nations are in a situation of competition with potential gain for both, but neither is threatened. Alternatively, both nations may be threatened with losses from global warming or other environmental hazards that may reduce nations’ territories for reasons that are unrelated to competition. We address the empirical question of whether just one of these factors, competition or threat, or both together, are necessary to produce ethnocentric choices. These two conditions, competition and threat, may pertain to the framing effects examined by prospect theory as follows: First, if the prospect of a loss constitutes a psychological threat, and threat induces ingroup favoritism, then decisions in loss-frame conditions should be more groupprejudiced than decisions in gain-frame conditions. For example, when deciding between certain and uncertain gains for an enemy, participants may follow prospect theory and chose certain gains, but when deciding between certain and uncertain losses for an enemy, participants may prefer certain losses, in contradiction to prospect theory’s prediction that people prefer uncertainty in loss domains. Intergroup competition may also lead people to make more groupprejudiced decisions than they would in the absence of competition, such as when groups’ prospects are independent of one another. Group-prejudiced decisions should occur to the degree that outgroups’ prospects are in conflict with the prospects of ingroups. If competition (rather than threat) induces group prejudice, then participants may also make different decisions than prospect theory predicts. In particular they may be reluctant to choose an uncertain loss for the ingroup over a certain loss for the outgroup. Likewise they may be reluctant to choose a certain gain for the outgroup over an uncertain gain for the ingroup. We tested these predictions using a replication and extension of Tversky and Kahneman’s (1981) “Asian disease” experiment. The design of the experiment enabled us to test whether participants would apply prospect theory in a group-neutral matter, showing preference for certain gains and uncertain losses regardless of whose lives were at stake. Against this default hypothesis, we tested several alternative hypotheses derived from intergroup relations research. When making decisions about the outgroup’s prospects alone, if participants are indifferent between decision choices that provide equivalent expected values, they will show no risk-preference reversal. This would be evidence of ethnocentric indifference to the prospects of the outgroup. If participants prefer certain losses and uncertain gains for an enemy outgroup, this would be evidence that they value the enemy’s prospects negatively (i.e., ethnocentric aggression). Finally, if participants make similar decisions for allies as for co-nationals, this would be evidence of allocentrism in which ally’s prospects are valued similarly to those of co-nationals. American and British participants made decisions concerning American, British, and Iraqi lives at the outset of the March, 2003 U.S./U.K.-led invasion of Iraq. Concurrent national opinion Prospect Theory and Ethnocentrism 4 polls showed that 70% or more of the American public supported the war with Iraq (Gallup, 2003). Following majority opposition for over a year, British opinion swung suddenly in a prowar direction (63% support) six days into the war when Baghdad fell (Time, 2002; Travis, 2003a, 2003b). The divergence of opinion about the war within and between nations enabled us to test whether opinion about the war influenced how participants weighed prospects of gains and losses for the involved nations. Although social and political values have been robustly shown to influence other kinds of policy attitudes and decisions (e.g., Pratto & Cathey, 2002), we have found no studies that examine how independently-assessed values or beliefs influence how people make decisions in prospect-theory scenarios. Given that the national discourses in the U.S. and U.K. were focused on the war against Iraq, and that the conditions of the experiment spoke to the publics’ concerns in the U.K. and U.S., this experiment provides a particularly strong test of the predicted inclusion and framing effects.