intrapersonal comparison in ultimatum games

Organizational Behavior and Human Decision Processes 90 (2003) 165–177 ORGANIZATIONAL BEHAVIOR AND HUMAN DECISION PROCESSES www.elsevier.com/locate/obhdp The salience of a recipientÕs alternatives: Inter- and intrapersonal comparison in ultimatum games Michel J.J. Handgraaf,a,* Eric van Dijk,b Henk A.M. Wilke,b and Ri€l C. Vermuntb e a Department of Economic and Social Psychology, Tilburg University, P.O. Box 90153, 5000 LE, Tilburg, The Netherlands b Department of Social and Organizational Psychology, Leiden University, The Netherlands Abstract The social utility model suggests that in social decision-making, both inter- and intrapersonal comparisons are important in assessing the utility of a decision outcome. In the ultimatum game both these comparisons play a role. This is especially true for recipients reacting to an unfair offer. We propose that the relative weights inter- and intrapersonal comparisons receive in ultimatum games depend on the way the decision is structured. In three studies we show that presenting recipients with a straightforward choice instead of the usual accept/reject question makes recipients more inclined to accept unfair offers. Moreover, the existence of an alternative outcome, i.e., the fact that refusal of the offer also leads to a substantial outcome, similarly raises the level of acceptances in a standard ultimatum game. Results are discussed in relation to the joint/separate evaluation disparity and the distinction between occurrences and non-occurrences. Ó 2003 Elsevier Science (USA). All rights reserved. Keywords: Ultimatum game; Social utility; Fairness; Self-interest 1. Introduction The social utility model (Blount, 1995; De Dreu, Lualhati, & McCusker, 1994; Loewenstein, Thompson, & Bazerman, 1989; Messick & Sentis, 1985; Van Dijk & Vermunt, 2000) suggests that two utility sources are important in distributive decision-making. The first is an absolute payoff component, which reflects the weight people assign to their own outcomes, the second a social comparative component, which reflects the weight people assign to their own outcome in relation to the outcomes of others. More specifically, the absolute payoff component may be interpreted as an inclination to act in a self-interested manner, whereas the comparative component may be interpreted as a taste for fairness, reflecting that people prefer equitable distributions over inequitable ones (cf. Bethwaite & Tompkinson, 1996; Kahneman, Knetsch, & Thaler, 1986; Loewenstein et al., 1989; Messick, 1995; Messick & Sentis, 1985). Insights from studies related to the social utility model suggest that it is both meaningful and useful to * Corresponding author. Fax: +31-0-13-466-2370. E-mail address: handgraaf@uvt.nl (M.J.J. Handgraaf). distinguish between several components of outcome utility instead of restricting theorizing to self-interest alone. The social utility model implies that the utility of an outcome depends on both own payoff and the comparison between oneÕs own and other peopleÕs outcomes, each with their own weight. Surprisingly, research to identify the situational factors that influence the differential weighting of self-interest and comparative outcomes is scarce (see Van Dijk & Vermunt, 2000). One of the aims of our present research is to identify factors that affect the relative importance of both components of the model, especially for decision-making in ultimatum games. 1.1. The ultimatum game A very appropriate paradigm used to investigate the discord between social motives like fairness and self-interest is the ultimatum game. In the standard ultimatum game, as originally developed by G€th, Schmittberger, u and Schwarze (1982), two players have to divide a certain amount of money between them. One player is allocator and proposes a division of the money; the other is recipient and can either accept or refuse the proposed 0749-5978/03/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0749-5978(02)00512-5 166 M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 division. If the recipient accepts, the money is divided as proposed. If the recipient refuses, both players receive nothing. With its simple structure and clear game theoretic predictions, the ultimatum game is an attractive tool to assess the relative importance of self-interest and fairness considerations in social decision-making. Theories that are based on the assumption that bargainers maximize their own outcome, like game theory, predict that allocators would offer the recipient the smallest possible amount and that recipients would accept this offer. After all, if recipients are only interested in maximizing their own outcomes, they should realize that accepting even the smallest possible offer will yield higher outcomes than the alternative of refusing, because refusing would mean no outcomes at all. As for the allocators, the results of a number of studies contradict game theoretic predictions. They show that allocatorsÕ offers are usually close to a 50-50 division. Only a small percentage (less than 1%) offers an amount that comes close to game theoretic predictions (G€th, 1995; see for reviews also Camerer & Thaler, u 1995; G€th & Tietz, 1990; Thaler, 1988). On first sight u this disconfirmation of the game theoretic prediction seems to imply that the comparative component of the social utility model carries great weight in ultimatum bargaining. That is, it seems that allocators do show a taste for fairness when making their offer. More recently, however, it has been argued that allocators may not truly want to be fair, but only try to appear fair to the recipient, because they fear rejection of their offer (Weg & Zwick, 1994). As Kagel, Kim, and Moser (1996) have shown, seemingly fair offers may actually reflect a motivation to maximize own outcome. In their study, participants were allocators in an ultimatum game in which the chips that were to be divided had a greater monetary value for the allocator than for the recipient. The results showed that when the recipient was aware of this differential value, the allocators compensated for this differential value. When the recipient was not aware of the differential value, however, and allocators were aware of this lack of information, allocators took advantage of the lack of information of the recipient and made offers that seemed fair, but actually were not. That is, they proposed a 50-50 division of the chips, resulting in an unfair monetary distribution, thus increasing their own outcomes without fear of rejection. This study supports the hypothesis that allocators are more inclined towards strategic and sometimes even exploitative behavior than toward fairness (see also Camerer & Thaler, 1995; Pillutla & Murnighan, 1995; Van Dijk & Vermunt, 2000). These results indicate that allocators in ultimatum games are trying to appear fair to the recipient instead of truly being fair (Pillutla & Murnighan, 1995). This seems to call for a return to models that assume that allocators are (only) motivated to maximize their own outcomes. Since much of the research done using ulti- matum games is biased in favor of the analysis of allocator behavior (Huck, 1999), results from ultimatum game studies may lead to a view of decision makers as being mainly concerned by own outcome maximization. However, these findings only concern half of the decision makers in ultimatum games, namely the allocators. They do not tell us much about the behavior and motivations of recipients. These studies tell us that allocators assume that recipients consider fairness an important issue and behave accordingly. This assumption could, however, be a misconception. Another reason we consider this research bias unfortunate is that tactical considerations, which may make interpretation of allocatorsÕ behavior less straightforward, are irrelevant to the behavior of recipients. Recipients in ultimatum games need not be influenced by tactical considerations, since they are the ones that make the final move. This implicates that for recipients the conflict between the comparative component and the absolute payoff component of the social utility model is not clouded by strategic considerations. It thus seems that studies on the behavior of recipients may fill gaps in our knowledge about the role of fairness in decisionmaking. It is for this reason that we wanted to focus our attention on recipients in ultimatum games. What do we know about the behavior of recipients? Is the fear of refusal shown by allocators justified? Several studies show that recipients often refuse substantial offers if the offers are unfair (Camerer & Thaler, 1995). Even when the total amount to divide was raised, as Hofmann, McCabe, and Smith (1996) did, from the usual $10 to $100, a substantial part of unfair offers were refused. Forty percent of the recipients in this experiment refused a 70-30 offer. Some of these participants actually gave up $30 to punish an unfair allocator. It is important to note that these refusals of unfair offers cannot be the result of fear of retaliation or other tactical considerations, since recipients have the last word (i.e., the game ends after their decision). In terms of the social utility model, it appears that refusals of unfair offers are probably motivated by concerns that are primarily related to fairness and interpersonal comparison and less to self-interest. 1.2. Separate vs. joint evaluation The foregoing seems to support the proposal that fairness considerations do play a substantial role in ultimatum game bargaining for recipients. When recipients choose to forego substantial amounts in order to punish an unfair allocator (Camerer, 1992), they apparently do this because they assign great weight to fairness. But why? Tactics cannot account for refusing recipientsÕ behavior (as they make the last and final move in the game), and therefore something other than strategy M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 167 should be the reason for their (according to game theory) irrational behavior. In order to make sense of the behavior of refusing recipients, it is first necessary to examine the kind of decision they face. At first sight it would seem that they make a choice between two possible distributions: They can either accept the unfair distribution offered by the allocator or they can refuse, which results in a 0-0 distribution. Although, in a formal sense, this is an adequate description of the decision faced by a recipient, we argue that it is an inexact description of the decision recipients actually make. Specifically, we maintain that one of these two behavioral options, particularly the offer, is more salient. This, we argue, results in the recipients reacting to the fairness characteristics of this offer, rather than making a choice between two possible distributions. This observation is important, because it points to a distinction between two modes of evaluation of decision alternatives. In their recent review Hsee, Loewenstein, Blount, and Bazerman (1999); see also Hsee (1996) argue that a large number of studies have documented systematic differences between situations in which options are presented and evaluated in isolation (separate evaluation mode, or SE) and situations in which multiple options are presented and evaluated simultaneously (joint evaluation mode, or JE). It should be noted that Hsee et al. argue that the distinction between these two types of evaluation (joint and separate) should be seen as a continuum. Most decisions fall between the extremes of this continuum. In several studies, the making of an isolated evaluation of an interpersonal distribution (SE) has been compared with choosing between two interpersonal outcomes (JE; see Bazerman, Loewenstein, & Blount White, 1992; Bazerman, Schroth, Shah, Diekmann, & Tenbrunsel, 1994). As an example, Bazerman et al. (1994) offered their participants (second year students) hypothetical jobs. Job 1 was at company A and involved a salary of $75.000. It was known that company A usually paid starters $75.000. The second job was at company B and involved a salary of $85.000. It was known that company B usually paid starters $95.000. Half of the participants rated these jobs independently (they were asked to indicate whether they would or would not accept this job), whereas the other participants got the job offers in pairs. Of each pair they had to indicate which one they would choose. Results showed that participants preferred job 1 (lower salary, but paid equal to other starters) when they had to rate the jobs separately, whereas the participants in the choice conditions preferred job 2 (the job with the higher salary). This indicates that fairness (and thus the social comparison component of the social utility model) is relatively more important when the decision at hand is a judgment of one single interpersonal distribution. Selfinterest is relatively more important when a comparison between two interpersonal distributions is made. According to this logic, if recipients in ultimatum games were really comparing the two distributions, selfinterest should prevail. But several studies (Camerer & Thaler, 1995) show that it does not. Why is this the case? The reason recipients focus on fairness may lie in the questions recipients ask themselves when making a decision. It is important to note that the two components of the social utility model require different comparisons. The first, the absolute component (i.e., self-interest), requires a comparison between the own outcome in one situation (e.g., the own outcome should one accept), and the own outcome in another situation (e.g., the own outcome should one refuse). This is an intra personal comparison. Utility is directly related to the difference between the outcomes in these situations. The comparative component requires a comparison between oneÕs own outcomes and those of relevant others. This is an inter personal comparison. These two comparisons are fundamentally different. This observation is the basis of our current framework. We argue that the specific way of presenting the recipientÕs decision in traditional ultimatum games focuses the recipient on the interpersonal component of the social utility model, thus setting the recipient in separate evaluation mode. After all, in a standard ultimatum game, the allocator makes an offer, and the recipient is subsequently asked to either accept or reject this offer. It is therefore mainly the proposed interpersonal division the recipient reacts to. The exit option, the 0-0 division (of which the recipient is aware, since the game has been thoroughly explained), is only implicitly there. When the recipient is asked to either accept or reject this proposed interpersonal division, the offer and its fairness characteristics are much more salient. This will probably increase the relative importance of the interpersonal (or comparative) component of the social utility model. Note that it is important to keep in mind that we do not suggest that decisions are made either solely out of intrapersonal considerations or solely out of considerations of an interpersonal nature. In line with the social utility model (Loewenstein et al., 1989) we assume that both oneÕs own outcome and someone elseÕs outcome are important motives in distributive decision making, and that it is more a question of the relative importance of these motives in the decision making process. Our reasoning about the differential relative importance of the interpersonal comparison under JE or SE is supported by a study by Blount and Bazerman (1996), which shows that the elicitation method of preferences for outcomes in ultimatum games may lead to inconsistencies in the weighting of fairness. They varied the elicitation method for recipientsÕ decisions in ultimatum games. Before learning the offer, half the participants were asked to indicate their minimal acceptable offer, and the other half were asked to react to several hypothetical offers by choosing between accepting or 168 M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 receiving zero. The results of the Blount and Bazerman study support our reasoning that the specific presentation of an ultimatum game may influence the relative importance of the interpersonal comparison component of the social utility model. Their study showed that recipients were more willing to accept unequal payoffs when they were asked to consider several possible offers than when they were asked to generate a minimum acceptable payoff. Of course this only constitutes indirect evidence for our reasoning. It should be noted that there were marked differences between their and our studies. The recipients in the Blount and Bazerman study are asked to either generate a minimum acceptable outcome (in the ‘‘minimal acceptable offer’’ condition), or to react to (several) hypothetical offers (in the ‘‘choice’’ condition). Both these elicitation methods differ from the method used in a standard ultimatum game, in which a recipient reacts to one non-hypothetical offer. It is our argument that it is the specific way of presenting the recipientÕs decision in traditional ultimatum games which sets the recipient in separate evaluation mode. The Blount and Bazerman study does not include such a standard ultimatum game, and therefore we use a design that does include such a standard ultimatum game in our first study. The question we address in this paper is: What happens if the exit option does play a more explicit role in the decision making process? Will recipients focus more on their own outcome and less on fairness and will they therefore be less reluctant to accept unfair offers? Will the intrapersonal comparison component of the social utility model indeed become more important at the cost of the interpersonal component? We consider these important questions because their answers may provide us with further insights into the factors that influence the weight each of the components of the social utility model receives in decision making. As Bazerman et al. (1994) have shown, interpersonal comparison is important when a single (interpersonal) distribution is judged, whereas individual outcome becomes more important when two interpersonal distributions are compared (see also Hsee, 1996; Hsee et al., 1999). It should therefore make a difference whether the recipients in an ultimatum game are presented with a clear choice between two possible distributions (i.e., the outcome distribution associated with acceptance and the outcome distribution associated with rejection) or are asked to accept or reject the offer the allocator makes. We predict that asking recipients to focus on the fact that there are two possible outcomes with different gains for both allocator and recipient will lead them to not only take into account the interpersonal comparison of the outcomes would they accept, but also the comparison between their own outcome in case of acceptance and their own outcome in case of refusal. This should lead to an increase in focus on intrapersonal comparison (or individual outcome) and a decrease in focus on interpersonal outcome comparison (or fairness considerations) and thus to a higher proportion of acceptances of unfair offers (cf. Bazerman et al., 1994). We will examine this in Study 1. If the salience of the exit option is indeed such an important factor in the weighting of the components of the social utility model, it may be interesting to look for other ways in which this salience may be influenced. We propose that there is an additional feature which may lead recipients in ultimatum games to focus on the interpersonal comparison: the fact that refusal by the recipient implies that no money is divided. Legrenzi, Girotto, and Johnson-Laird (1993) call this kind of outcome a non-occurrence. We propose that an exit option is more salient when this exit option is an occurrence. That is, if refusal by the recipient would mean that both players would still receive a certain amount (say 50c.), that should make this exit option more salient than an exit option which means that no money is distributed. Legrenzi et al. argue that individuals are likely to restrict their thoughts to what is explicitly represented in their mental models and often fail to consider less salient alternatives. We propose that the 0-0 division that results from refusal by a recipient in an ultimatum is such a less salient alternative. This may have an influence on decision-making in ultimatum games. This conjecture will be examined in Study 2. 2. Study 1 As a preliminary test of our ideas, we first examine the effects of presenting the recipients with a clear choice between two options instead of giving them the opportunity to refuse or accept an allocation. We argued that in a standard ultimatum game the recipient is asked to evaluate the offer in terms of acceptance or refusal and that this focuses the recipient on the fairness characteristics of the offer. This focus on the interpersonal component causes this component of the social utility model to dominate the decision making process. By presenting the decision to the recipient in such a way that the exit option is more explicitly available, i.e., by having recipients choose between two possible distributions (the allocatorÕs offer and the 0-0 outcome), we expect the intrapersonal component of the utility model to surface. Forcing the recipient to choose between two options will lead to a joint evaluation of both possible interpersonal distributions. The fact that a refusal will lead to a lower outcome will become more important in this joint evaluation. This should make recipients more reluctant to refuse the allocatorÕs offer. From the above we distilled the following expectations: For standard ultimatum games the focus will be on the interpersonal comparison, and therefore there will be many refusals of unfair offers. When the decision M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 169 the recipient has to make is explicitly presented as a choice between two possible distributions, intrapersonal comparison will become more important and interpersonal comparison will become less important. This will lead to an increase in acceptances of unfair offers as compared with the standard ultimatum game. 2.1. Method Participants and design. Thirty-four students from several programs of Leiden University volunteered to participate. Participants were randomly assigned to one of two conditions. In the first participants were recipients in a standard ultimatum game; in the second they were recipients in a slightly modified version, in which the decision they had to make was explicitly presented as a choice between two options. The most important dependent variable was the participantÕs reaction to the offer. Procedure. Upon arrival, participants (between 6 and 8 at a time) were seated behind computers in separate cubicles. Computers were used to present all information and to register the dependent measures. Participants were informed that they were going to divide an amount of money together with one of the other participants. The amount (10 Dutch guilders, or about $5 at the time of the experiment) was presented as consisting of 100 chips, each worth 0.10 Dutch guilders. We then explained the rules of the ultimatum game. In doing so we did not use the words ‘‘ultimatum’’ or ‘‘game’’. Participants were also told that they would remain anonymous to each other and that the number assigned to their cubicle would determine whether they would be an allocator or a recipient. In reality all participants were recipients (they were referred to as ÔPlayer BÕ) and always learned that the proposition the allocator (ÔPlayer AÕ) made was the following: 70 chips for player A, 30 for player B. Participants in the ‘‘standard’’ ultimatum conditions were informed that they could either accept, which would mean that the chips would be divided according to this proposition, or refuse, which would mean that both allocator and recipient would receive nothing. The question participants had to answer was: ‘‘The division Player A proposes is the following: Player A receives 70 chips, Player B (you) receives 30 chips. Do you accept this offer?’’ The possible answers were ‘‘Yes, I accept this offer’’ and ‘‘No, I refuse this offer’’. The answer to this question was the most important dependent variable. Participants in the ‘‘choice’’ ultimatum conditions were informed that they could choose between two options: option 1, which would mean that the chips would be divided according to this proposition, or option 2, which would mean that both allocator and recipient would receive nothing. The question participants had to answer was: ‘‘The division Player A proposes is the following: Player A receives 70 chips, Player B (you) receives 30 chips. Which of the following options do you choose?’’ The possible answers were: ‘‘I choose option 1: I receive 30 chips and Player A receives 70 chips’’ and ‘‘I choose option 2: I receive 0 chips and Player A receives 0 chips’’. Participants in all conditions were told that the allocator had received the same information as they and that they were also completely informed about the rules of the task. We also checked whether or not participants understood the game. We asked them to indicate which outcomes Players A and B would receive upon acceptance and upon rejection of the offer. After their reaction, participants were thoroughly debriefed. They were then thanked and received 10 guilders. All participants agreed to this procedure. 2.2. Results Manipulation checks. All participants answered the questions regarding the possible outcomes of Players A and B after acceptance and rejection of the offer correctly. Acceptance rates. Participants (all recipients) reacted to the unfair offer by either accepting or rejecting it (in the standard ultimatum condition), or (in the choice ultimatum condition) by choosing the proposed division or the 0-0 exit option. We expected that in the choice ultimatum conditions a greater proportion of the recipients would respond favorably to the unfair offer. RecipientsÕ reactions to the offer were analyzed by means of a logistic regression analysis. The results of this analysis show a significant effect for type of game (B ¼ À1:20; WaldÕs v2 ð1; N ¼ 34Þ ¼ 2:81, p < :05, one tailed). As expected, explicitly presenting the recipientÕs decision as a choice between two distributions resulted in a higher percentage of positive responses to the offer (63%, or 10 out of 16) than in the standard ultimatum game (33%, or 6 out of 18). 2.3. Discussion The results show that recipients in ultimatum games are more inclined to accept an unfair offer when they are explicitly made aware of the fact that refusal will lead them to go home empty-handed. This was done by presenting the decision as a choice between two distributions. From this we conclude that in standard ultimatum games, the decision of the recipient is not primarily based on the comparison between own outcome following acceptance and own outcome after refusal. The interpersonal characteristics of the offer itself are very important. The recipientÕs decision appears to be strongly based on the (un)fairness of the interpersonal outcome distribution as proposed by the allocator. It is this interpersonal comparison that makes many 170 M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 recipients refuse unfair offers in standard ultimatum games. When the ultimatum game is modified by presenting the decision as a clear choice between two distributions, the recipients appear to put less weight on the interpersonal comparison and more on the intrapersonal comparison. As a result, more recipients react positively to the (unfair) offer. 3. Study 2 We propose there is another feature that may lead recipients in ultimatum games to focus on the interpersonal comparison and give less attention to the intrapersonal comparison: the fact that the exit option implies a 0-0 division. To clarify this, it is necessary to explain the concept of a non-occurrence. In a standard ultimatum game refusal of an offer implies that no money is divided (both players receive nothing). One could therefore say that refusal leads to a non-occurrence. That is, the division as proposed by the allocator will not take place. This reformulation is important, because research shows that the non-occurrence of an event is often ignored when judging a situation. It has been shown that the non-occurrence of an event is less salient than the occurrence of an event and therefore receives less attention (cf. Becker, Ronen, & Sorter, 1974; Hoskin, 1983; Legrenzi et al., 1993; Northcraft & Neale, 1986). Our current theorizing is also related to work in the field of attitude formation (Allison & Messick, 1988; Fazio, Sherman, & Herr, 1982). People show the tendency to be influenced more by occurrences than by non-occurrences. This has been termed the feature-positive effect. For example, Fazio et al. asked participants to rate the funniness of cartoons. Part of the participants had to push a button to indicate that they felt a cartoon was funny, and to do nothing if they felt the cartoon was not funny. Another part was instructed to push a button if they felt a cartoon was not funny, and to do nothing if they felt the cartoon was funny. Results indicated that judgments afterward were more extreme when revealed by pushing a button than when revealed by doing nothing (see for similar findings Allison & Messick, 1988). Even more relevant is a study by Legrenzi et al. (1993) who argued that individuals are likely to restrict their thoughts to what is explicitly presented in their mental models, and often fail to consider relevant, but less salient alternatives. For example, Legrenzi et al. (1993) argued that if people have to decide on whether or not to go to a movie, people tend to focus on information about the movie, and not on the alternatives they could opt for should they decide not to go to the movie. Thus, even though a decision to act or not can be defined as a choice between at least two options (act or donÕt act; go to the movie, or donÕt go), people tend to focus on the option that changes the status quo. Based on this reasoning, we argue that when both refusal and acceptance in an ultimatum game lead to outcomes that change the status quo, this will affect the decision of the recipient. We expect that the existence of a substantial exit option will increase the recipientÕs focus on the comparison of the two possible distributions and thus the weight the intrapersonal component receives in the recipientsÕ decision process. Or, in terms of Hsee et al. (1999), a substantial exit option will make the decision of the recipient shift from the separate evaluation side of the decision continuum towards the joint evaluation end of this continuum. In order to investigate this reasoning, we had participants play the role of recipient in ultimatum games. In this ultimatum game 100 chips were to be divided. We manipulated both type of game and exit option. The manipulation of type of game was similar to the manipulation in Study 1; the participantÕs decision was either presented as an accept/reject decision or as a choice between two options (a 70-30 division, i.e., the offer, and an exit option). Exit option was manipulated by either telling subjects that both players would receive nothing upon refusal by the recipient or that both players would receive 15 chips if the recipient refused. The above led us to expect an interaction between type of game and exit option. More specifically, we expected the effect of type of game to be less pronounced when the exit option was 15-15 than when the exit option was 0-0. The rationale behind this expectation was that the effect of type of game we observed in Study 1 rests on the assumption that in the standard game the focus is on interpersonal comparison. Compared to this situation, the choice game should make the exit option more salient, thereby increasing the weight of the intrapersonal component. As we reasoned, this is particularly true if the exit option is zero (i.e., a nonoccurrence). Making this 0-0 exit option more salient should move the decision from the SE end of the JE–SE continuum toward the JE side of it, which should increase the weight the intrapersonal component of the social utility model receives. In the case of a 15-15 exit option, the decision already leans more towards the JE end of the JE–SE continuum. Therefore, intrapersonal comparison should already be more salient, even in the standard game. If so, the manipulation of game type should have less effect on the recipientsÕ reactions. It may be argued that changing the exit option from 0-0 to a distribution which entails a substantial outcome (15-15) has more effects than just change the exit option from a non-occurrence to an occurrence. It may change the attractiveness of refusal. That is, it is likely that a recipient who is focusing on an intrapersonal comparison (i.e., a comparison between own outcome following acceptance and own outcome following refusal) will be more inclined to refuse when the exit option is higher M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 171 than zero. After all, the amount recipients forego when they refuse is lower. Recipients for whom the intrapersonal comparison is important should therefore be influenced by the height of the exit option. As a consequence, the expected interaction between game type and exit option may also be described differently. We already argued that in the choice conditions the emphasis is on intrapersonal comparison. That being the case, it follows that an offer of 70-30 is more acceptable when the exit option is less attractive (i.e., zero). But what effect would the height of the exit option (15 vs. 0 chips) have in the standard game? We already reasoned that in the standard game with an exit option of zero, recipients are inclined to emphasize interpersonal comparison. With a shift to an exit option of 15 chips, the intrapersonal comparison should become more important, which could lead to an increase in acceptances. It should be noted, however, that this effect might be somewhat suppressed by the effect we described above (i.e., receiving 15 chips is more attractive than receiving nothing). 3.1. Method Participants and design. Ninety students from Leiden University volunteered to participate in this experiment. Participants were all recipients in ultimatum games and were randomly assigned to one of the cells of a 2 (Type of Game: standard vs. choice) Â 2 (Exit Option: 0 vs. 1.5 guilders) factorial design. The main dependent variable was their reaction to the offer. Procedure. The procedure was largely identical to the procedure described in Study 1. Six to eight participants were seated in separate cubicles behind a computer screen. Participants were informed that they were going to divide an amount of money together with one of the other participants. The amount (10 Dutch guilders, or about $5 at the time of the experiment) was presented as consisting of 100 chips, each worth 0.10 Dutch guilders. Similarly to Study 1, all participants were recipients (they were referred to as ÔPlayer BÕ) and always learned that the proposition the allocator (ÔPlayer AÕ) made was the following: 70 chips for player A, 30 for player B. Participants in the standard ultimatum conditions were informed that they could either accept, which would mean that the chips would be divided according to this proposition, or refuse, which would mean that both allocator and recipient would receive the exit option (i.e., nothing in the conditions with exit option 0, or 15 chips in the conditions with exit option 15). The question participants had to answer was: ‘‘The division Player A proposes is the following: Player A receives 70 chips, Player B (you) receives 30 chips. Do you accept this offer?’’ The possible answers were ‘‘Yes, I accept this offer’’ and ‘‘No, I refuse this offer’’. The answer to this question was the most important dependent variable. Participants in the choice ultimatum conditions received similar instructions. The main difference was the question participants had to answer: ‘‘The division Player A proposes is the following: Player A receives 70 chips, Player B (you) receives 30 chips. Which of the following options do you choose?’’ The possible answers were: ‘‘I choose option 1: I receive 30 chips and Player A receives 70 chips’’ and ‘‘I choose option 2: I receive 0 chips (or 15 chips, in the conditions with exit option 15) and Player A receives 0 chips (or 15 chips, in the conditions with exit option 15)’’. We again checked whether or not participants understood the game. We asked them to indicate which outcomes Players A and B would receive upon acceptance and upon rejection of the offer. 3.2. Results Manipulation checks. All participants answered the questions about the possible outcomes of the game for Players A and B correctly. Acceptance rates. Our main dependent variable was the participantsÕ reaction to the unfair offer of the allocator (i.e., the preprogrammed 70-30 offer). These reactions were analyzed by means of a logistic regression analysis. As expected, the results of the logistic regression analysis show a significant interaction effect for type of game and exit option (B ¼ À1:84; WaldÕs v2 ð1; N ¼ 90Þ ¼ 4:35, p < :05). A further analysis of the simple main effects showed that the results from Study 1 were replicated. The presentation of the decision in a choice format again resulted in more acceptances than presentation in the standard format when the exit option was 0-0. Subjects accepted more when the presentation had a choice format (65%) than when the presentation was like that of a standard ultimatum game (35%; B ¼ À1:26; WaldÕs v2 ð1; N ¼ 46Þ ¼ 4:12, p < :05). As expected, when recipients had an exit option of 15-15, the effect of type of game became smaller. In fact, there was no difference in percentage of acceptances between the subjects who played games that were presented as a choice (32%) and participants that played regular ultimatum games (46%; B ¼ :58; WaldÕs v2 ð1; N ¼ 44Þ ¼ :86, ns.). As we already noted, the interaction effect may also be interpreted in a different manner: As expected, presentation as a choice resulted in a higher percentage of acceptance in the condition with an exit option of 0-0 than in the condition with an exit option of 15-15, whereas this pattern was reversed when participants played a standard game. In the conditions with a presentation as a choice, an analysis of the simple main effect showed an effect of exit option: more acceptances when the exit option was 0-0 than when the exit option was 15-15 (65% vs. 32%; B ¼ À1:39; WaldÕs v2 ð1; N ¼ 45Þ ¼ 4:82, p < :05). In case of presentation as a standard ultimatum game this was reversed, the percentage 172 M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 of acceptances was higher when the exit option was 15-15 than when it was 0-0 (46% vs. 35%). Although the interaction effect was significant, analysis of the simple main effect showed that this latter difference did not reach significance. This is in line with the notion that the increase in acceptances may have been slightly suppressed by the fact that raising the exit option meant that refusal became more attractive (B ¼ .45; WaldÕs v2 ð1; N ¼ 45Þ ¼ :53, ns.). 3.3. Discussion The results of Study 2 confirm our hypotheses and show that it does make a difference whether the exit option entails an occurrence rather than a non-occurrence. Recipients in ultimatum games seem to give more weight to the intrapersonal comparison when the exit option means money is still divided than when the exit option means neither player receives anything. In terms of Legrenzi et al. (1993): When the exit option is a nonoccurrence, this exit option is not (or less) a part of recipientÕs mental models, which leads to a lack of focus on the intrapersonal comparison. This implicates that the existence of a substantial exit option has a similar effect as presentation in a choice format: Both move the decision from the SE end towards the JE end of the JE–SE continuum, which leads to an increase in focus on the intrapersonal comparison. The reason we find no difference between the standard ultimatum game with a 15-15 exit option and the choice ultimatum game with a 15-15 option may be that in the standard ultimatum game with a 15-15 exit option, the decision is already at the JE end of the continuum. In this case, changing this game into a choice ultimatum game does not move the decision much further towards the JE end of the continuum, which explains the similar acceptance rates in both conditions. When we concentrate on the choice ultimatum games, where the focus is already on the intrapersonal comparison, we found that that there are more refusals when there is a 15-15 exit option than when there is a 0-0 exit option. This can be explained by the fact that a refusal provides the recipient with a higher outcome when the exit option is 15-15 than when it is 0-0. When the presentation is like that of a standard ultimatum game, the fact that there is a substantial exit option focuses the recipient on the intrapersonal comparison, which leads to a higher proportion of acceptances, since acceptance always leads to a higher individual outcome. It should be noted, however, that the higher exit option in the condition with a 15-15 exit also had another effect: the higher exit option made refusal more attractive. As anticipated, this opposite force dimmed the effect of the increased emphasis on the intrapersonal comparison (which should lead to more acceptances). 4. Study 3 In the foregoing we proposed that both the presentation format of the game (either as a standard or as a choice ultimatum game) and the exit option (either 0 or a small but substantial amount) influences the relative weight self-interest and fairness receive in the decision making process. From the behavior towards the offer as displayed by the participants, we inferred that this was indeed the case. Both the existence of a substantial exit option and presentation of the decision as a choice raised the amount of positive responses toward the unfair offer, which indicates that fairness became relatively less important. In order to assess whether our reasoning is sound, and that it is indeed a decrease in relative weight of fairness which causes recipients to be more inclined to accept unfair offers in choice ultimatum games, we conducted a third study. In this study, we also measured how important recipients thought fairness was when they made their decision. If our reasoning is correct, the importance of fairness should be greater in a standard ultimatum game (hypothesis 1a). Moreover, the importance of fairness should mediate between presentation (as either a standard ultimatum game or a choice ultimatum game) and the acceptance rates, with a higher score on the importance of fairness measure leading to a decrease in acceptances of unfair offers (hypothesis 1b). In this third study we also wanted to rule out an alternative explanation for the results. There is a possibility, suggested by one of the reviewers, that the difference between the acceptance rates of the standard ultimatum game and the choice ultimatum game may be due to the specific wording of the recipientÕs alternatives. Indeed, specific wording may play a critical role in any bargaining situation. As Larrick and Blount (1997) put it: ‘‘Semantic differences may influence interpersonal behavior by evoking different understandings of responsibility and blame for an outcome’’ (p. 811). In the standard ultimatum game, one of the options is to refuse the offer, which may be an appropriate term to use, both conversationally and emotionally, when faced with an unfair offer. In contrast, in the choice ultimatum game, the corresponding option does not contain the word ‘‘refuse’’, or any other word with a similar emotional connotation. This difference may provide an alternative explanation for our data: when faced with an unfair offer, recipients may consider refusing an emotionally and conversationally appropriate response. Since the word ‘‘refuse’’ is only present in the standard ultimatum condition, this may cause the increase in refusals. In order to rule out this alternative explanation, we had participants play recipient in either a standard or a choice ultimatum game in which we added the words accept and refuse to the options recipients had in the choice ultimatum games. We expected that adding these M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 173 emotionally laden words would not lead to different findings than in the first two studies. We hypothesized that explicitly presenting the recipientÕs decision as a choice between two distributions would result in a higher percentage of positive responses to the offer than in the standard ultimatum game (hypothesis 2). 4.1. Method Participants and design. Thirty-five psychology students from Tilburg University volunteered to participate as part of a course requirement. Procedure. Procedures were similar to the ones reported in Study 1. Again participants always learned that the proposition the allocator (‘‘Player A’’) made was the following: 70 chips for player A, 30 for player B. Participants in the standard ultimatum conditions were informed that they could either accept, which would mean that the chips would be divided according to this proposition, or refuse, which would mean that both allocator and recipient would receive nothing. The question participants had to answer was: ‘‘The division Player A proposes is the following: Player A receives 70 chips, Player B (you) receives 30 chips. Do you accept this offer?’’ The possible answers were ‘‘Yes, I accept this offer’’ and ‘‘No, I refuse this offer’’. Participants in the choice ultimatum conditions were informed that they could choose between two options: option 1, which would mean that the chips would be divided according to this proposition, or option 2, which would mean that both allocator and recipient would receive nothing. The question participants had to answer was: ‘‘The division Player A proposes is the following: Player A receives 70 chips, Player B (you) receives 30 chips. Which of the following options do you choose?’’ The possible answers were: ‘‘I choose option 1, I accept: I receive 30 chips and Player A receives 70 chips’’ and ‘‘I choose option 2, I refuse: I receive 0 chips and Player A receives 0 chips’’. Thus, compared with Study 1, we added the words ‘‘accept’’ and ‘‘refuse’’ to the options the recipients had. To assess the relative importance of fairness, we subsequently asked participants how important fairness was when they made their decision. We asked them: ‘‘Thinking back to the moment you decided, how important was it at that moment whether the offer of Player A was fair?’’ This question was answered on a 7-point scale that went from ‘‘not at all important’’ to ‘‘very important’’. As in Study 1 and 2, participants were told that the allocator had received the same information as they and that they were also completely informed about the rules of the task. We also checked whether or not participants understood the game. We asked them to indicate which outcomes Players A and B would receive upon acceptance and upon rejection of the offer. After their reac- tion, participants were thoroughly debriefed. They were then thanked and received one point for their course requirement. All participants agreed to this procedure. 4.2. Results Manipulation checks. All participants answered the questions about the possible outcomes correctly. Acceptance rates. Participants (all recipients) reacted to the unfair offer by either accepting or refusing it (in the standard ultimatum condition), or (in the choice ultimatum condition) by choosing the proposed division or the 0-0 exit option. We expected similar results as in Study 1: in the choice ultimatum conditions a greater proportion of the recipients would respond favorably to the unfair offer (hypothesis 2). RecipientsÕ reactions to the offer were analyzed by means of a logistic regression analysis. The results of this analysis show a significant effect for type of game (Nagelkerke R2 ¼ :27; B ¼ À2:48, SE ¼ 1:14; WaldÕs v2 ð1; N ¼ 35Þ ¼ 4:71, p < :05). As in Study 1, explicitly presenting the recipientÕs decision as a choice between two distributions resulted in a higher percentage of positive responses to the offer (41%, or 7 out of 17) than in the standard ultimatum game (6%, or 1 out of 18). As expected, when the words ‘‘accept’’ and ‘‘refuse’’ were also present in the choice ultimatum game the pattern of results was the same as in Study 1. Mediation analysis. In order to assess whether it is indeed the relative importance of fairness that mediates the influence of presentation as either choice or standard ultimatum game, we also measured how important recipients thought fairness was at the moment of their decision. A regression analysis showed that the results for weight of fairness followed the same pattern as the acceptance rates (Adjusted R2 ¼ :29; B ¼ À1:49, SE ¼ :38; tð1; 33Þ ¼ À3:88, p < :01). Recipients rated fairness as more important in the standard ultimatum game (M ¼ 6:67) than in the choice ultimatum game (M ¼ 5:18). These results confirm hypothesis 1a. We expected (hypothesis 1b) that the weight of fairness would mediate the effect of type of game on acceptance rates, with a higher score on the importance of fairness measure leading to a decrease in acceptances of unfair offers. To test for mediation, a series of regression models were estimated according to the method described by Baron and Kenny (1986). The predictor variable in the models was type of game (either a standard ultimatum game or a choice ultimatum game). The hypothesized mediator was weight of fairness. The acceptance rates served as the dependent variable. To test for mediation we first had to show that the mediator (weight of fairness) is correlated with the predictor variable (type of game). We then had to show that the dependent variable (acceptance rate) was correlated with the predictor variable. Finally, we needed to show that the mediator affects the dependent variable, when con- 174 M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 trolled for the predictor variable. The degree to which the influence of type of game on acceptance rates is reduced when accounting for the weight of fairness expresses the degree of mediation by weight of fairness. In case of mediation the influence of type of game drops significantly and in case of full mediation the effect of type of game becomes non-significant. We already assessed that there is an effect for type of game on acceptance rate (see above). We also already assessed that there was an effect of type of game on weight of fairness (see above). We furthermore needed to establish whether adding the mediator to the simple model (which consists of type of game and acceptance rate) improves the fit of the regression model. This was done by performing a likelihood-ratio test. This test showed that adding weight of fairness to the regression model significantly improves the models fit (Nagelkerke R2 ¼ :402; v2 ð1; N ¼ 35Þ ¼ 3:91, p < :05). This shows that weight of fairness affects the dependent variable (acceptance rate), when controlling for the predictor variable. A logistic regression analysis furthermore showed that adding the weight of fairness measure to the regression model resulted in a drop in the effect of type of game on acceptance rates. In fact, it became nonsignificant (B ¼ À1:38, SE ¼ 1:33; WaldÕs v2 ð1; N ¼ 35Þ ¼ 1:08, ns.). This shows that, as expected, weight of fairness mediates the interaction effect of type of game on acceptance rates. 4.3. Discussion Study 3 tested the hypothesis that the difference in the acceptance rates between standard ultimatum games and choice ultimatum games in the first two studies was due to some sort of conversational or even emotional norm. After all, only in the standard ultimatum game did the participants encounter the emotionally and conversationally appropriate word ‘‘refuse’’. This possibly accounted for the larger proportion of refusals of recipients in standard ultimatum games. However, adding the words ‘‘accept’’ and ‘‘refuse’’ to the choice ultimatum game did not change the pattern of results we found in the first two studies. This rules out this alternative explanation. These findings further support our reasoning about the relative weight of the two components of the social utility model. The results of the first two studies indicated that whether the intra- or interpersonal comparison component of the social utility model is more important is influenced by the salience of the exit option. When the exit option is salient, either because it also entails a distribution of money, or because it is made more salient by using a different presentation-format, the intrapersonal comparison becomes more important to the recipient. This then results in an increase in the proportion of positive reactions to unfair offers. In order to further validate this reasoning, in this third study we also assessed whether the weight of fairness mediated the effect of type of game on acceptance rates. If it is indeed the case that the salience of alternatives influences the relative weight of both components of the social utility model in the proposed manner, this should lead to a decrease in weight of fairness in a choice ultimatum game, resulting in a corresponding increase in acceptance rates for unfair offers. Our results showed that weight of fairness did indeed mediate the effect of type of game on acceptance rates. 5. General discussion The results of all three studies show that participants are more inclined to accept an unfair offer when they give relatively more weight to the fact that refusal will lead them to go home with less. They are more inclined to refuse when they give more weight to the unfairness of an offer. In a standard ultimatum game the comparison of recipientsÕ own outcome in case of acceptance with the outcome of the other player in case of acceptance seems to play an important role. The comparison of their own outcome in case of acceptance with their own outcome in case of refusal receives relatively little weight. The results of the current studies indicate that whether intra- or interpersonal comparison is more important is influenced by the salience of the exit option. When the exit option is salient, either because it also entails a distribution of money, or because it is made more salient by using a different presentation-format, the intrapersonal comparison becomes more important to the recipient. This reasoning is supported by the fact that weight of fairness mediates between type of game and acceptance rates. As we mentioned in the Introduction, Blount and Bazerman (1996) report results supportive of this reasoning. Their findings seem to accord with the current findings observed in Studies 1 and 3 in the sense that they also found that the recipients in the ‘‘choice’’ condition were more inclined to accept lower offers. It should be noted, however, that our current setup also differed from the one used by Blount and Bazerman in some important ways. Firstly, the recipients in the Blount and Bazerman study are asked to either generate a minimum acceptable outcome (in the ‘‘minimal acceptable offer’’ condition), or to react to (several) hypothetical offers (in the ‘‘choice’’ condition). Both these elicitation methods differ from the method used in a standard ultimatum game, in which a recipient reacts to one non-hypothetical offer. The difference in elicitation method in itself may have influenced the outcomes of the games. In our study, all recipients reacted to one (non-hypothetical) offer of the allocator, as is usually the M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 175 case in ultimatum game research. Furthermore, in the Blount and Bazerman study, participants indicated their decisions prior to actually being confronted with the offer, whereas in the typical ultimatum game study, recipients react after being confronted with a specific offer. Studies on a variety of experimental paradigms have shown that timing effects may have a profound effect on social decision making (see for example Camerer, Weber, & Knez, 1996; Rapoport, Budescu, & Suleiman, 1993; Roth, 1988). In the current studies, we investigated the typical ultimatum bargaining situation in which recipients know what offer the allocator made when deciding on their reaction. Participants were always aware of the offer, and thus were always in a position that they could react to this specific (nonhypothetical) offer. Thus, the original structure of the game was left intact, both in the standard ultimatum conditions and in the choice ultimatum conditions. Another major difference of course is that our current theorizing is not restricted to possible effects of the elicitation method, but also extends to the possible effect of having a substantial (i.e., non-zero) exit option. The results suggest that this manipulation may also make the exit option more salient. This induces a shift in focus from the inter- to the intrapersonal comparison component of the social utility model, resulting in higher acceptance rates. The mechanism that is responsible for this shift in focus on either the intra- or the interpersonal comparison in our studies may be the same mechanism that is the basis of the evaluability hypothesis (Hsee et al., 1999). The evaluability hypothesis posits that: ‘‘it is more difficult to evaluate the desirability of values on some attributes than on others and that, compared with easyto-evaluate attributes, difficult-to-evaluate attributes have a greater impact in JE than in SE’’ (Hsee et al., 1999, p. 576). Hsee et al. argue that the distinction between these two types of evaluation (joint and separate) should be seen as a continuum. Most decisions fall between the extremes of this continuum. The modifications we made to the ultimatum game (either presenting the decision as a choice or adding a substantial exit option) move the decision a recipient in an ultimatum game has to make towards the joint evaluation side of the continuum. In a standard ultimatum game, a recipient faces a proposed division and has to decide whether to accept or reject this proposal. This decision, we argue, falls at the separate evaluation extreme of the JE–SE continuum. We argue that this specific way of presenting the decision focuses the recipient on the interpersonal comparison. This is in line with the evaluability hypothesis, which would predict that the attribute that is easiest to assess is more important in separate evaluation. In a situation where the recipient is asked to accept or reject an interpersonal distribution (a standard ultimatum game), it is the interpersonal component that is easily assessed and which therefore receives most attention. After all, what the recipients get to see when they have to decide is the proposed interpersonal division. The exit option, the 0-0 division (of which the recipient is aware, since the game has been thoroughly explained) is only implicitly there. When the exit option is made more explicit, the decision moves toward the joint evaluation side of the continuum, which means the difficult-to-evaluate attributes (in this case the intrapersonal comparison) gain importance. This leads to a higher proportion of acceptances of unfair offers. In case of an ultimatum game in which the recipient has a substantial exit option, the decision faced by this recipient leans more towards the joint evaluation side of the JE–SE continuum. This is even the case if presentation resembles that of a standard accept/reject ultimatum game, since the mere existence of this alternative makes it more salient. Changing the game into an explicit choice ultimatum game (with a substantial exit option) does not move the decision of the recipient much further along the JE–SE continuum. This explains why, at a behavioral level, there is not much difference in acceptance rates between the standard and choice ultimatum game when a substantial exit option is present. It seems that adding a substantial exit option for recipients in a standard ultimatum game makes the effect of presentation as either a separate or a joint evaluation disappear. It should be noted, however, that we do not think this means that the JE–SE distinction goes away when there is a substantial exit option. It is simply the case that a substantial exit option moves the decision of the recipient towards the JE end of the JE–SE continuum. A way of assessing whether this easy or difficult to evaluate dimension is indeed an important factor in a recipientÕs decision making process would be to directly manipulate the ease with which either the intra- or interpersonal comparison can be made. If our theorizing is correct, making the interpersonal comparison more difficult should lead to a greater proportion of acceptances of unfair offers. Making the interpersonal comparison more difficult should lead to an increase in focus on the intrapersonal features of an offer, and thus to more acceptance of unfair offers. This could, for instance, be done by making sure the recipient receives a different type of outcome than the allocator (let us say, apples for the allocator and oranges for the recipient). In this case it will be more difficult for the recipients to assess the difference between their own outcome upon acceptance and the allocators outcome upon acceptance (we all know how difficult it is to compare apples with oranges). This should lead to an increased focus on intrapersonal comparison, and thus to more acceptance of unfair offers. We are currently investigating this hypothesis. Another interesting finding in the current research is the drop in acceptance rates in choice ultimatum games when a substantial exit option is added to the game. This 176 M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 difference in acceptance rates does not occur in standard ultimatum games. We explain this drop in acceptance rates by economic considerations. When an exit option is available, refusing is less costly, which makes it more likely that recipients refuse. This is what our results showed. We think the reason this difference stemming from economical considerations is absent in the standard version is because the participants are less likely to make this economic calculation in a standard ultimatum game with a 0-0 exit option. Since the 0-0 exit option is not very salient in the standard ultimatum game, refusals in this case are motivated by fairness considerations, whereas refusals in the standard game with a 15-15 exit are more influenced by economic considerations. At a behavioral level, the two lead to similar acceptance rates. We do think that additional considerations such as not loosing face or being considered unintelligent may play a role, but we also think that these considerations by themselves do not constitute an adequate explanation for our results. After all, one would still need to explain why and when it is that these considerations play a role in the choice ultimatum game and not in the standard game. We argue that loosing face or being considered unintelligent are considerations that are closely linked to our key concepts of fairness and self-interest. Loosing face may be related to not standing up to unfair behavior (but also to not behaving in oneÕs own interest), whereas being considered unintelligent probably has to do with giving the impression of not being able to act in a rational economic way (i.e., in a self-interested manner). Since the concepts of fairness and self-interest are central in our argumentation, we do not think that incorporation of these considerations changes the argument. Nor does it enable more precise theorizing. In this respect, we suggest that the current argument that concentrates on the components of the social utility model is more parsimonious. In the current article we show that very simple variations in the structure of the decision-making situation may have great impact on the ensuing behavior of the recipient. Interestingly, Larrick and Blount (1997) recently drew attention to the influence the presentation of decisions may have in mixed motive situations. They compared decision making in social dilemma games and ultimatum games and focused on the differences between the required procedures in these games, claiming an amount in a social dilemma game and either proposing (for players 1) or accepting or rejecting (for players 2) in an ultimatum game. They found that players in social dilemma games were more generous, even though the games are structurally equivalent. The present article, with its focus on the presentation of decisions, is related to this approach. There is an important difference, however, in that we concentrate on the perception of recipients in ultimatum games and compare results for modified games with those of the standard game. In their discussion, Larrick and Blount (1997) state that a ‘‘factor that might underlie the difference is that accepting or rejecting is evaluative and draws more attention to fairness’’ (p. 823). In our studies we have found similar results that seem to agree with this explanation. We did not compare different games, but modified the ultimatum game in such a way that the likelihood that the recipient would focus on intrapersonal comparison was varied. It turned out that when recipients focused more on their own outcomes and therefore less on the interpersonal comparison that is required for the assessment of fairness, they became less inclined to reject unfair offers. In short, these studies show that the importance of both components of the social utility model may be influenced by both the presentation of the decision and by the availability of exit options. A real world illustration of this may be that when it is desirable that a recipient accepts an unfair distribution, and there exist a substantially lower alternative, it may sometimes be a good strategy to make sure that this recipient is aware of this exit option. Recipients are then more likely to realize that they are better off by accepting the unfair offer and therefore more likely to accept. If this lower alternative is not present, it may even be feasible to create one. With this tentative example, attention returns to the allocator. One of the interesting questions the current studies do not address is how these manipulations of game type and of the height of the exit option would influence the allocatorÕs decision. It would be very interesting to determine whether allocators, for instance, would change their offers in response to these manipulations and whether or not they prefer a situation in which the recipient has a small but substantial exit option to a situation where the recipient doesnÕt have this substantial exit option. Such an approach, focusing on both allocators and recipients, may be the next step toward a better understanding of the relation between the two components of the social utility model and their connection to the concepts of occurrence, non-occurrence and evaluability. References Allison, S. T., & Messick, D. M. (1988). The feature-positive effect, attitude strength, and degree of perceived consensus. Personality and Social Psychology Bulletin, 14, 231–241. 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, 1173–1183. Bazerman, M. H., Loewenstein, G. F., & Blount White, S. (1992). Reversals of preference in allocation decisions: Judging an alternative versus choosing among alternatives. Administrative Science Quarterly, 37, 220–240. Bazerman, M. A., Schroth, H. A., Shah, P. P., Diekmann, K. A., & Tenbrunsel, A. E. (1994). The inconsistent role of comparison M.J.J. Handgraaf et al. / Organizational Behavior and Human Decision Processes 90 (2003) 165–177 others and procedural justice in reactions to hypothetical job descriptions: Implications for job acceptance decisions. Organizational Behavior and Human Decision Processes, 60, 326–352. Becker, S. W., Ronen, J., & Sorter, G. H. (1974). Opportunity costs— an experimental approach. Journal of Accounting Research, 12, 317–329. Bethwaite, J., & Tompkinson, P. (1996). The ultimatum game and non-selfish utility functions. Journal of Economic Psychology, 17, 259–271. Blount, S. (1995). When social outcomes arenÕt fair: The effect of causal attributions on preferences. Organizational Behavior and Human Decision Processes, 63, 131–144. Blount, S., & Bazerman, M. H. (1996). The inconsistent evaluation of absolute versus comparative payoffs in labor supply and bargaining. Journal of Economic Behavior and Organization, 30, 227–240. Camerer, C. F. (1992). The rationality of prices and volume in experimental markets. Organizational Behavior and Human Decision Processes, 51, 237–272. Camerer, C. F., & Thaler, R. H. (1995). Ultimatums, dictators and manners. Journal of Economic Perspectives, 9, 209–219. Camerer, C. F., Weber, R. A., & Knez, M. (1996). Timing and virtual observability in ultimatum bargaining and Ôweak linkÕ coordination games. Unpublished manuscript. De Dreu, C. K. W., Lualhati, J. C., & McCusker, C. M. (1994). Effects of gain-loss frames on satisfaction with self-other outcomedifferences. European Journal of Social Psychology, 24, 497– 510. Fazio, R. H., Sherman, S. J., & Herr, P. M. (1982). The featurepositive effect in the self-perception process: Does not doing matter as much as doing? Journal of Personality and Social Psychology, 42, 404–411. G€th, W. (1995). On ultimatum bargaining experiments: A personal u review. Journal of Economic Behavior and Organization, 27, 329– 344. G€th, W., Schmittberger, R., & Schwarze, B. (1982). An experimental u analysis of ultimatum games. Journal of Economic Behavior and Organization, 3, 367–388. G€th, W., & Tietz, R. (1990). Ultimatum bargaining behavior: A u survey and comparison of experimental results. Journal of Economic Psychology, 11, 417–449. Hofmann, E., McCabe, K., & Smith, V. L. (1996). Social distance and other-regarding behavior in dictator games. American Economic Review, 86, 653–660. Hoskin, R. E. (1983). Opportunity costs and behavior. Journal of Accounting, 21, 78–95. Hsee, C. (1996). The evaluability hypothesis: An explanation for preference reversals between joint and separate evaluations of 177 alternatives. Organizational Behavior and Human Decision Processes, 67, 247–257. Hsee, C. K., Loewenstein, G. F., Blount, S., & Bazerman, M. H. (1999). Preference reversals between joint and separate evaluations of options: A review and theoretical analysis. Psychological Bulletin, 125, 576–590. Huck, S. (1999). Responder behavior in ultimatum offer games with incomplete information. Journal of Economic Psychology, 20, 183–206. Kagel, J. H., Kim, C., & Moser, D. (1996). Fairness in ultimatum games with asymmetric information and asymmetric payoffs. Games and Economic Behavior, 13, 100–110. Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1986). Fairness and the assumptions of economics. Journal of Business, 59, 285–300. Larrick, R. P., & Blount, S. (1997). The claiming effect: Why players are more generous in social dilemmas than in ultimatum games. Journal of Personality and Social Psychology, 72, 810–825. Legrenzi, P., Girotto, V., & Johnson-Laird, P. N. (1993). Focusing in reasoning and decision making. Cognition, 49, 37–66. Loewenstein, G. F., Thompson, L., & Bazerman, M. H. (1989). Social utility and decision making in interpersonal contexts. Journal of Personality and Social Psychology, 57, 426–441. Messick, D. M. (1995). Equality, fairness, and social conflict. Social Justice Research, 8, 153–173. Messick, D. M., & Sentis, K. P. (1985). Estimating social and nonsocial utility functions from ordinal data. European Journal of Social Psychology, 15, 389–399. Northcraft, G. B., & Neale, M. A. (1986). Opportunity costs and the framing of resource allocation decisions. Organizational Behavior and Human Decision Processes, 37, 348–356. Pillutla, M. M., & Murnighan, J. K. (1995). Being fair or appearing fair: Strategic behavior in ultimatum bargaining. Academy of Management Journal, 38, 1408–1426. Rapoport, A., Budescu, D. V., & Suleiman, R. (1993). Sequential requests from randomly distributed shared resources. Journal of Mathematical Psychology, 37, 241–265. Roth, A. E. (1988). Laboratory experiments in economics: A methodological overview. The Economic Journal, 98, 974–1031. Thaler, R. H. (1988). The ultimatum game. Journal of Economic Perspectives, 2, 195–206. Van Dijk, E., & Vermunt, R. (2000). Strategy and fairness in social decision making: Sometimes it pays to be powerless. Journal of Experimental Social Psychology, 36, 1–25. Weg, E., & Zwick, R. (1994). Toward the settlement of the fairness issues in ultimatum games: A bargaining approach. Journal of Economic Behavior and Organization, 24, 19–34. Received 22 May 2001