Journal of Experimental Social Psychology 42 (2006) 273–289

www.elsevier.com/locate/jesp

On preferences and doing the right thing: Satisfaction with advantageous inequity when cognitive processing is limited �

Kees van den Bos a,¤, Susanne L. Peters a, D. Ramona Bobocel b, Jan Fekke Ybema c

a Department of Social and Organizational Psychology, Utrecht University, Utrecht, The Netherlandsb University of Waterloo, Canada

c Netherlands Organization for Applied ScientiWc Research, Hoofddorp, The Netherlands

Received 6 May 2004; revised 19 February 2005Available online 13 June 2005

Abstract

We examine outcome satisfaction in situations in which people receive better outcomes than comparable other persons. Buildingon classical and modern social psychological theories, we argue that when reacting to these arrangements of advantageous inequity,judging the advantage is quick and easy as preferences are primary. We further propose that adjusting this appraisal requires cogni-tive resources as it entails integrating fairness concerns with the initial preference appraisal. Extending the literature on cognitivebusyness, we therefore predict that people should be more satisWed with advantageous inequity when cognitive processing isstrongly—as opposed to weakly—limited. Findings across several diVerent experimental paradigms support our predictions. Takentogether, our Wndings shed light on the psychological processes underlying the intriguing interplay between egoism- and fairness-based considerations when evaluating outcomes, as well as on more general preference and adjustment processes.© 2005 Elsevier Inc. All rights reserved.

Keywords: Fairness; Justice; Equity; Preferences; Cognitive busyness

Numerous scientiWc disciplines, including philosophy,sociology, political sciences, economics, and psychology,had good reasons to focus on the issue of how peopleform evaluations of the outcomes they have received. Avital view in both classic and modern social psychologyemphasizes the importance of drawing a distinctionbetween two diVerent reference points in the outcomeevaluation process (e.g., Adams, 1965; Blau, 1964; Diek-mann, Samuels, Ross, & Bazerman, 1997; Epley& Caruso, 2004; Epley, Keysar, Van Boven, & Gilovich,2004; Frey, Benz, & Stutzer, 2003; Loewenstein,

� We thank Henk Aarts, Johan Karremans, Mel Lerner, WolfgangStroebe, Remco Wijn, and several anonymous reviewers for their com-ments on previous versions of this paper, and Sophie de Vente, Erik Ei-zema, and Parcival von Schmid for their assistance with Experiment 4.

* Corresponding author. Fax: +31 30 253 4718.E-mail address: k.vandenbos@fss.uu.nl (K. van den Bos).

0022-1031/$ - see front matter © 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jesp.2005.04.003

Thompson, & Bazerman, 1989; Messick & Sentis, 1979,1983; Moore & Loewenstein, 2004; Smeesters, Warlop,Van Avermaet, Corneille, & Yzerbyt, 2003; StouVer,Suchman, DeVinney, Star, & Williams, 1949; Thibaut &Kelley, 1959). One important reference point is usuallyreferred to as people’s preferences and is based on whatoutcome arrangements make people pleased or dis-pleased (Messick & Sentis, 1979, 1983). Another impor-tant reference point has to do with what outcomedistributions are fair or unfair, equitable or inequitable,just or unjust (Van den Bos, Wilke, Lind, & Vermunt,1998).

Arrangements that make people pleased can be diVer-ent from those that are fair, equitable, or just (e.g., Mes-sick & Sentis, 1983). In some cases, what makes onepleased and what one believes to be right coincide per-fectly. In other instances, one’s reactions toward out-come distributions are characterized by a conXict

274 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

between one’s preferences and what is right. It is thisconXict that we are focusing on in this paper. More spe-ciWcally, we do this by examining what we think is one ofthe most intriguing issues in our discipline: people’s sat-isfaction with arrangements in which their own out-comes are inequitable but better than the outcomes ofcomparable other persons.

Ever since Adams’ (e.g., 1963a, 1963b, 1965; Adams &Jacobsen, 1964; Adams & Rosenbaum, 1962) Wrst obser-vations of the eVects that these and other arrangementsof advantageous inequity had on the reactions ofemployees of General Electric, social psychologists havebeen interested in people’s reactions to advantageousinequity (see, e.g., Berkowitz & Walster, 1976). We wouldlike to propose here that because of the mixed-motivequality of advantageous inequity arrangements(Jacques, 1961; Peters, Van den Bos, & Karremans,2004), studying these arrangements may shed furtherlight on the relationship between people’s preferencesand their inclinations of doing the right thing. To thisend, we Wrst scrutinize how people usually evaluatearrangements of advantageous inequity. After this, wespecify what we think are interesting conditions thatmay help to better understand this evaluation processand the conXicting forces that are exerted on it.

How do people usually evaluate advantageous inequity?

Following earlier equity studies (e.g., Austin,McGinn, & Susmilch, 1980; Van den Bos, Lind, Ver-munt, & Wilke, 1997; Van den Bos et al., 1998), we studypeople’s evaluations of advantageous inequity by expos-ing participants to situations in which there is anotherperson who is comparable to the participants themselveswith respect to the amount of input they have provided.The outcome that participants themselves receive is heldconstant across conditions and, to arrange for equitableand inequitable arrangements, we vary the outcome thatthe other person in the situation receives: Participantswho are confronted with advantageous inequity areinformed that their own outcome is better than the out-come of the other person. Participants who are facedwith disadvantageous inequity are told that their out-come is worse than the other person’s outcome. Partici-pants who Wnd themselves in an equitable situation areinformed that their outcome is equal to the otherperson’s outcome.

DiVerent types of human reactions have been investi-gated in equity studies. The two dependent variables thathave been most frequently used measure how satisWedpeople are with equitable and inequitable arrangementsand how just people judge these arrangements to be (e.g., Adams, 1965; Austin et al., 1980; Van den Bos et al.,1997, 1998). Although the concepts of justice and satis-faction are interrelated, they are diVerent concepts, and

it is important not to confuse them (Austin et al., 1980;Blau, 1964; Messick & Sentis, 1983; Van den Bos et al.,1997, 1998). Most important for the current purposes isthe Wnding that the conXict between preferences anddoing the right thing can be most clearly seen when peo-ple’s satisfaction with advantageous inequity is studied.

That is, previous studies have shown that, withrespect to people’s judgments of outcome justice, a per-son who receives an outcome that is equal to the out-come of another person (who has provided acomparable amount of inputs) receives an equitableoutcome and hence receives a just outcome, whereas aperson who receives a better outcome than the com-parison other and a person who receives a worse out-come than the comparison other both receiveinequitable outcomes and hence both receive unjustoutcomes. As a result, previous equity studies havefound that a person who receives an outcome that isequal to the outcome of the comparison other will per-ceive his or her outcome to be more just than either aperson who receives a better outcome than the com-parison other or a person who receives a worse out-come than the comparison other, and that outcomejustice judgments in the better than other conditionswill not diVer from the outcome justice judgments inthe worse than other conditions (see, e.g., Austin et al.,1980; Van den Bos et al., 1997, 1998).

More interesting for the current purposes is people’ssatisfaction with arrangements of advantageous inequity.On the basis of equity theory (e.g., Adams, 1965; Austinet al., 1980; Austin & Walster, 1974; Buunk & Van Ype-ren, 1989; Van den Bos et al., 1997, 1998), it can be rea-soned that individuals who are faced with inequity willfeel distressed and will be less satisWed about thisarrangement than individuals who are faced with equity.As noted by Adams (1965): “There can be little doubtthat inequity results in dissatisfaction” (p. 283). Further-more, it can be argued that people who are confrontedwith disadvantageous inequity do not have to deal withconXicting social motives when forming judgments ofoutcome satisfaction whereas people who are faced withadvantageous inequity do. That is, when people are madeangry by disadvantageous inequity there are two sourcesof negative aVect: (a) the unfairness, and (b) the relativedeprivation of lacking what the other person has received(cf. Van den Bos et al., 1997, 1998). However, when expe-riencing advantageous inequity, there is one source ofpositive aVect and one source of negative aVect: (a) thepositive source is the egoism-based pleasure of receiving arelatively good outcome, whereas (b) the negative sourceis the fairness-based feeling of being unfairly advantaged(cf. Van den Bos et al., 1997, 1998). When confrontedwith equitable arrangements there are mainly sources ofpositive aVect (caused by the experiences of fairness andequity) and no or weak sources of negative aVect (Adams,1965; Buunk & Van Yperen, 1989).

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 275

At least three implications follow from the aboveanalysis of people’s satisfaction with advantageous ineq-uity. The Wrst implication underlies the prediction thattypically is made in equity studies (e.g., Buunk & VanYperen, 1989; Van den Bos et al., 1997, 1998): that twosources of negative experience are bound to sum to lesssatisfaction than are one source of a positive emotionalexperience and one source of a negative emotional expe-rience. Therefore, persons who are confronted withadvantageous inequity should feel uncomfortable butwill be more satisWed than persons who are confrontedwith disadvantageous inequity (Peters et al., 2004). Thus,in equity studies it is usually proposed and found that aperson who receives an equitable outcome will be moresatisWed than a person who receives an advantageousinequitable outcome, and that the latter individual willbe more satisWed than a person who receives a disadvan-tageous inequitable outcome (e.g., Buunk & Van Yperen,1989; Van den Bos et al., 1997, 1998).

The second implication is that the above-describedanalysis suggests that the conXict between preferencesand doing the right thing is likely to be most evidentwhen inspecting people’s satisfaction with advantageousinequity. That is, whereas people judge equitablearrangements to be just and satisfying and disadvanta-geous inequitable arrangements to be unjust and dissat-isfying, they typically judge advantageous inequity to beas unjust as disadvantageous inequity but tend to bemore satisWed with advantageous inequity than with dis-advantageous inequity. It follows then that when peopleare asked to judge the justice of arrangements of advan-tageous inequity, their justice judgments tend to be inXu-enced predominantly by the unfairness aspects of thesearrangements. In contrast, when people are evaluatinghow satisWed they are with these arrangements, bothpreferences and fairness considerations are inXuencingtheir satisfaction evaluations. This suggests that whenpeople are reacting to advantageous inequity, satisfac-tion evaluations are likely to be more easily inXuencedthan justice judgments by the two conXicting forces inwhich we are interested in the present research. Giventhen that satisfaction ratings may be particularly wellsuited to study the conXict between egoism-based prefer-ences and fairness considerations, satisfaction withadvantageous inequity will constitute our primarydependent variable in all of the current experiments.Additionally, we will explore the distinction between jus-tice judgments and satisfaction evaluations in the studiesthat follow after Experiment 2, and we will come back tothis issue in the General discussion.

The third implication is that we argue here that theabove-described analysis of how people evaluate howsatisWed they are with advantageous inequity is impor-tant because it can be related to theories that suggestthat people usually will know whether their outcomegives them pleasure before they have insight into the

fairness aspects of the outcome distribution (see,e.g., Epley & Caruso, 2004; Epley et al., 2004; Messick &Sentis, 1979, 1983; Moore & Loewenstein, 2004). Mes-sick and Sentis, for instance, argue convincingly that therelationship between preference and fairness is such thatpeople generally have more immediate access to orknowledge of their preferences in a situation than towhat is fair. As stated by Messick and Sentis (1983):“one usually knows one’s preferences before one knowswhat is fair” (p. 88). This view led the authors to proposethat preference is primary (cf. Zajonc, 1980) and thatpeople assess whether and how fairness is relevant in alater phase (possibly almost immediately). Related tothis, Moore and Loewenstein (2004) argue that self-interest is automatic, viscerally compelling, and typicallyunconscious whereas, in contrast, paying attention tofairness concerns is usually a more thoughtful process.Similarly, Epley and Caruso (2004; Epley et al., 2004)propose that people automatically interpret objects andevents egocentrically and only subsequently correct oradjust that interpretation when necessary. According toEpley et al. (2004), the automatic default occurs rapidlybut correction requires time and attentional resources.

The theories noted above are primarily general con-ceptual statements, have not focused on people’s satis-faction with advantageous inequity, and have notprovided empirical data on this issue (for details, see Epley& Caruso, 2004; Epley et al., 2004; Messick & Sentis, 1979,1983; Moore & Loewenstein, 2004), but combining theprocesses these models are stipulating with our examina-tion of the social psychology of advantageous inequitylead us to argue that when people are evaluating howsatisWed they are with an arrangement of advantageousinequity they will infer that the outcome distributiongives them egoism-based pleasure before they realizethat the arrangement is unfair (cf. Messick & Sentis,1979, 1983). What we propose to do here is to explorethe implications of this proposition by investigating peo-ple’s evaluations of outcome satisfaction when reactingto arrangements of advantageous inequity while theircognitive processing has been strongly limited (versusweakly limited). To this end, we will integrate our analy-sis of the social psychology of advantageous inequitywith the research on the eVects of cognitive busyness onperson evaluations and will note some interesting paral-lels between these two lines of thought.

Satisfaction with advantageous inequity and cognitive busyness

Gilbert and others have argued convincingly thatwhen people form evaluations of having seen anotherperson performing a certain behavior they typicallydraw initial character inferences from the behavior andthen correct these initial evaluations by taking into

276 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

account the inXuences of external forces that may haveinXuenced the person’s behavior (e.g., Gilbert &Osborne, 1989; Gilbert, Pelham, & Krull, 1988; Ham& Vonk, 2003). These authors have further put forwardthat correction requires more cognitive resources thandoes the forming of initial inferences. As a result, form-ing person evaluations while simultaneously performinga resource-consuming task should impair the formerprocess more than the latter process. Gilbert et al.’sexperiments indeed show that person evaluations are lessinXuenced by external information under conditions ofhigh cognitive busyness than under conditions of lowcognitive busyness (e.g., Gilbert & Osborne, 1989;Gilbert et al., 1988).1

In both our analysis of the process by which peopleevaluate how satisWed they are with advantageous ineq-uity (see also Epley & Caruso, 2004; Epley et al., 2004;Messick & Sentis, 1979, 1983; Moore & Loewenstein,2004) and Gilbert et al.’s (1988; Gilbert & Osborne,1989) work, some correction is postulated to take placein the evaluation processes under investigation. Further-more, Gilbert et al. showed persuasively that the implica-tions of this correction process can be investigated byexploring people’s evaluations under conditions of highversus low cognitive busyness. Integrating our analysisof the social psychology of advantageous inequity withGilbert et al.’s (1988; Gilbert & Osborne, 1989) workleads to the following hypothesis regarding people’s sat-isfaction with advantageous inequity: When people reactto arrangements of advantageous inequity their evalua-tions of outcome satisfaction should be more positiveunder conditions where their cognitive processing isstrongly (as opposed to weakly) limited.

We investigated this hypothesis in a number of diVer-ent experiments using diVerent paradigms and diVerentmanipulations. Common elements in our experimentswere that we varied that participants’ cognitive process-ing was either strongly or weakly limited while respond-ing to the stimulus materials (see, e.g., Gilbert &Osborne, 1989; Gilbert et al., 1988; see also Wegner& Erber, 1992). Furthermore, in all experiments, advan-tageous inequity conditions were included in which par-ticipants received an outcome that was better than theoutcome of a comparable other person and the maindependent variable was participants’ outcome satisfac-tion evaluations.

In Experiments 1–3, we also included conditions ofdisadvantageous inequity, in which participants’ ownoutcomes were worse than the comparable other person,and equitable conditions, in which own outcomes were

1 Because our research hypotheses and research designs were in-spired by the work by Gilbert and his colleagues (e.g., Gilbert &Osborne, 1989; Gilbert et al., 1988), we will adopt their convention anduse the term “cognitive busyness” in the present paper, rather than“cognitive load.”

equal to the other person’s outcome. We incorporatedthese conditions because they are often—although notalways (see, e.g., Rivera & Tedeschi, 1976)—included inequity studies and because we wanted to explorewhether eVects of the cognitive busyness manipulationswould be found in these conditions as well.

Experiment 1

In Experiment 1, participants responded to outcomedistributions that were constructed following earlierresearch by Van den Bos (1999, Experiment 1). The out-come that participants received was held constant acrossconditions, and we varied whether the outcome was equalto the outcome of a comparable other person (equal-to-other condition), better than the outcome of the other per-son (better-than-other condition), or worse than the otherperson’s outcome (worse-than-other condition). Buildingon the work by Gilbert and others (e.g., Gilbert et al.,1988), half of the participants rehearsed a string of eightsymbols while reading and responding to the stimulusmaterials (high-busyness condition) whereas the otherhalf of the participants rehearsed 1 symbol (low-busynesscondition). Following previous equity studies (e.g., Vanden Bos et al., 1997, 1998), the dependent variable wasparticipants’ judgments of outcome satisfaction.

Method

Participants and designOne hundred and thirty-eight students (49 men and

89 women)2 at the Free University Amsterdam partici-pated in the experiment and were paid for their partici-pation. Participants were randomly assigned to one ofthe conditions of the 2 (cognitive busyness: low vs.high)£3 (outcome: equal to other vs. better than othervs. worse than other) factorial design.

Experimental procedureParticipants completed the experiment before and

after participating in other, unrelated experiments. Theexperiments lasted a total of 1.5 h, and participants werepaid 15 Dutch guilders for their participation (1 Dutchguilder equaled approximately $0.40 U.S. at the time theexperiment was conducted). On arrival at the laboratory,participants were led to separate cubicles, each of whichcontained a computer with a monitor and a keyboard.The computers were used to present the stimulus infor-mation and to measure the dependent variables. In all

2 In all experiments presented here, gender was proportionally dis-tributed among conditions. Gender had no main or interaction eVectson the dependent variables of the experiments reported here and hencewas dropped from the analyses that are presented in this paper.

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 277

experiments of this paper, participants’ responses wereanonymous and participants knew this.

In the Wrst part of the experiment, we manipulatedcognitive busyness. Building on the work by Gilbert etal. (1988), participants in the high-busyness conditionwere asked to rehearse a string of eight symbols: @ * %# ? $ § §. Participants in the low-busyness conditionwere asked to rehearse one symbol: @. In both condi-tions, participants were given 25 s to rehearse the sym-bols and they were asked to hold the symbols in memoryuntil the computer would later ask them to reproducethem (cf. Gilbert et al., 1988).

After this, participants were asked to imagine the fol-lowing situation (cf. Van den Bos, 1999):

Last summer you had a job together with a fellow stu-dent. The two of you worked together in a pair. Therewere a large number of such pairs in the organizationwhere you worked. You and your fellow student haveworked equally hard and performed equally well.Because the organization has performed well last sum-mer, it is announced on the last day of summer that abonus of 10,000 Dutch guilders will be distributedamong all employees. A certain amount of money hasbeen allocated to you and your fellow student. It has tobe decided how this amount of money will be distributedbetween you and your fellow student.

This was followed by the manipulation of outcome.Participants read the following sentences (manipulatedinformation in italics):

A week after this employees are paid. You receive abonus of 500 Dutch guilders. Your fellow studentreceives a bonus of 250/500/750 Dutch guilders.

After this, participants were asked questions pertain-ing to the dependent variable: Participants were askedhow satisWed they were with the bonus they received(1D very dissatisWed, 7D very satisWed) and how happythey were with the bonus they received (1D veryunhappy, 7D very happy). Because participants’ answersto these two items were highly correlated (rD .86,p < .001), we averaged their answers to form a reliablescale of perceived outcome satisfaction (�D .93). Aftercompleting these measures, participants were asked towrite down the symbols they were supposed to rehearse.When the participants had done this, and had completedthe other experiments in which they would participate,they were thoroughly debriefed and paid for their partic-ipation.

Results

Recalled symbolsAs expected, the large majority of the participants

were able to accurately recall the symbols they were sup-posed to rehearse: Ninety-nine percent of the participants

in the low-busyness condition accurately recalled thesymbol they had to rehearse. Ninety-six percent of theparticipants in the high-busyness condition recalled atleast six symbols accurately (19% of those accuratelyrecalled six symbols, 48% accurately recalled seven sym-bols, and 33% accurately recalled eight symbols); thenumber of recalled symbols in this condition did notdiVer as a function of the outcome manipulation,F (2,67)D1.27, n.s. These results indicate that partici-pants took the memorization task seriously and made aneVort to recall the symbols while completing theexperiment.

Outcome satisfactionA 2£ 3 analysis of variance (ANOVA) on the out-

come satisfaction scale showed main eVects of outcome,F (2, 132)D236.56, p < .001, and cognitive busyness,F (1, 132)D4.36, p < .04; these eVects were qualiWed bythe predicted interaction, F (2, 132)D3.38, p < .04. Tointerpret these eVects we performed a least signiWcantdiVerence test for multiple comparisons between means(p < .05), with the six cells of our design serving as theindependent variable (Kirk, 1982). Table 1 shows theresult of this test as well as the means and standard devi-ations of the outcome satisfaction scale. As hypothe-sized, results showed that within the better-than-othercondition, participants were more satisWed with theiroutcome in the high-busyness condition than in the low-busyness condition. There were no eVects of cognitivebusyness within the equal-to-other and worse-than-other conditions.

Additionally, it can be noted here that, within bothhigh- and low-busyness conditions, perceived outcomesatisfaction was higher in the equal-to-other conditionsthan in the better-than-other conditions and was higherin the better-than-other conditions than in the worse-than-other conditions (see Table 1). This pattern ofmeans is in accordance with previous equity Wndings(e.g., Buunk & Van Yperen, 1989; Van den Bos et al.,1997, 1998). It should be noted, however, that as pre-dicted the mean satisfaction rating within the better-than-other condition was closer to the equal-to-other

Table 1Means and standard deviations of outcome satisfaction as a functionof cognitive busyness and outcome (Experiment 1)

Note. Means are on 7-point scales, with higher values indicating higherlevels of outcome satisfaction. Means with no subscripts in commondiVer signiWcantly (p < .05), as indicated by a least signiWcant diVerencetest for multiple comparisons between means (Kirk, 1982).

Cognitive busyness

Outcome

Equal to other Better than other Worse than other

M SD M SD M SD

Low 6.5a 0.6 3.0c 1.2 1.6d 0.6High 6.3a 0.5 4.0b 1.7 2.0d 1.1

278 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

condition under conditions of high cognitive busynessthan under conditions of low cognitive busyness.

Discussion

The reported Wndings are supportive of our line ofreasoning presented earlier: As predicted, Wndings showthat when people are reacting to arrangements of advan-tageous inequity their judgments of outcome satisfactionare more positive under conditions of high cognitivebusyness than under conditions of low cognitive busy-ness. Before strong conclusions are drawn on the basis ofthese Wndings, however, it is important to replicate themin a second experiment.

Experiment 2

In Experiment 2, we used an arrangement of advanta-geous inequity that was constructed following Van denBos et al. (1998, Experiment 1). As in Experiment 1 ofthe current paper, the outcome that participants receivedin Experiment 2 was held constant, and we manipulatedwhether this outcome was equal to the outcome of acomparable other person (equal-to-other condition),better than the outcome of the other person (better-than-other condition), or worse than the other person’soutcome (worse-than-other condition). To get an indica-tion of the robustness of the eVects reported here, weused a diVerent manipulation to vary limited cognitiveprocessing among our participants: Building on thework by Wegner and Erber (1992, Experiment 1), half ofthe participants were asked to respond to the questionsfollowing the scenarios as quickly as possible (high timepressure condition) whereas the other half of the partici-pants were not asked this (low time pressure condition).The dependent variable was participants’ judgments ofoutcome satisfaction.

Method

Participants and designOne hundred and Wfty-four students (60 men and 94

women) at the Free University Amsterdam participatedin the experiment and were paid for their participation.Participants were randomly assigned to one of the con-ditions of the 2 (time pressure: low vs. high)£ 3 (out-come: equal to other vs. better than other vs. worse thanother) factorial design.

Experimental procedureParticipants completed the experiment before and

after participating in other, unrelated experiments. Theexperiments lasted a total of 40 min, and participantswere paid 15 Dutch guilders for their participation. Onarrival at the laboratory, participants were led to sepa-

rate cubicles, each of which contained a computer with amonitor and a keyboard. The computers were used topresent the stimulus information and to measure thedependent variables.

At the start of the experiment, all participants wereinformed that in this study they would be asked to readabout a situation and answer questions about this situa-tion. Participants in the high time pressure conditionwere asked to respond to the questions as quickly aspossible whereas participants in the low time pressurecondition were not asked this.

After this, participants were asked to imagine thefollowing situation (cf. Van den Bos et al., 1998):

In the near future, you are going to live in a new rentedhouse. The rent of this house has yet to be determined.To decide on the rent, each individual tenant has toappear before a rent tribunal. The rent tribunal willdecide on the monthly rent that you will have to pay. Todetermine this rent, your neighbor, who will rent a com-parable house, also has to appear before the renttribunal.

This was followed by the outcome manipulation. Par-ticipants read the following sentences (manipulatedinformation in italics):

A week after this you are informed that the rent that youwill have to pay is 750 Dutch guilders. Your neighbor’srent also has been determined: The rent he will have topay is 500/750/1000 Dutch guilders.

After this, participants were asked questions pertain-ing to the dependent variable: Participants were askedhow satisWed they were with the rent they had to pay(1D very dissatisWed, 7D very satisWed) and how happythey were with the rent they had to pay (1D veryunhappy, 7D very happy). The answers to these two ques-tions were highly correlated (rD .76, p < .001) and wereaveraged to form a reliable scale of perceived outcomesatisfaction (�D .92). When participants had answeredthese questions, and had completed the other experi-ments in which they would participate, they were thor-oughly debriefed and paid for their participation.

Results

Response latenciesParticipants in the high time pressure condition of

Experiment 2 were asked to respond to the outcome sat-isfaction items as quickly as possible whereas partici-pants in the low time pressure condition were not askedthis. The time participants needed to answer the two out-come satisfaction items of Experiment 2 was measuredby the computers. As expected, a 2£ 3 multivariate anal-ysis of variance (MANOVA) on the response latenciesyielded only main eVects of time pressure: multivariateF (2,147)D 7.39, p < .01; for outcome satisfaction,

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 279

F (1, 148)D4.66, p < .04; for outcome happiness,F (1, 148)D14.02, p < .001. In accordance with instruc-tions, participants in the high time pressure conditiontook less time to answer both the outcome satisfactionquestion (MD 2.1 s, SDD1.8) and the outcome happi-ness question (MD3.1 s, SDD 2.0) than participants inthe low time pressure condition (MsD 2.9 and 4.9 s,SDsD2.6 and 3.7, respectively). This shows that, asintended, participants in the high time pressure condi-tions answered the outcome satisfaction items morequickly than did participants in the low time pressureconditions.

Outcome satisfactionA 2£3 ANOVA on the outcome satisfaction scale

yielded a main eVect of outcome, F(2,148)D55.52, p<.001;an eVect that was qualiWed by the predicted interaction,F(2,148)D5.10, p<.01. The main eVect of the time pres-sure manipulation was not signiWcant, F<1. To interpretthese eVects we performed a least signiWcant diVerence testfor multiple comparisons between means (p<.05), with thesix cells of our design serving as the independent variable(Kirk, 1982). Table 2 shows the result of this test and themeans and standard deviations of the outcome satisfactionscale. As hypothesized, Wndings showed that, within thebetter-than-other condition, participants were more satis-Wed with their outcome in the high time pressure conditionthan in the low time pressure condition. There were noeVects of the time pressure manipulation within the equal-to-other and worse-than-other conditions.

It can further be noted that within the low time pres-sure condition, perceived outcome satisfaction washigher in the equal-to-other condition than in the better-than-other condition and was higher in the better-than-other condition than in the worse-than-other condition(see Table 2). This pattern of means is in accordancewith previous equity Wndings (e.g., Buunk & Van Ype-ren, 1989; Van den Bos et al., 1997, 1998). In the hightime pressure condition, participants in the better-than-other condition were also more strongly satisWed withtheir outcome than those in the worse-than-othercondition, but they were as satisWed as those in theequal-to-other condition.

Table 2Means and standard deviations of outcome satisfaction as a functionof cognitive busyness and outcome (Experiment 2)

Note. Means are on 7-point scales, with higher values indicating higherlevels of outcome satisfaction. Means with no subscripts in commondiVer signiWcantly (p < .05), as indicated by a least signiWcant diVerencetest for multiple comparisons between means (Kirk, 1982).

Time pressure Outcome

Equal to other Better than other Worse than other

M SD M SD M SD

Low 4.7a 1.4 3.8b 1.4 2.0c 0.8High 4.1a,b 1.5 4.7a 2.0 1.5c 0.8

Discussion

The Wndings of Experiment 2 replicate the results ofExperiment 1 and provide supportive evidence for ourline of reasoning: As hypothesized, Wndings reveal thatin situations of advantageous inequity, people’s evalua-tions of outcome satisfaction are more positive underconditions of high time pressure than under conditionsof low time pressure. Furthermore, these results havebeen found using diVerent stimulus materials and adiVerent manipulation of limited cognitive processingthan in Experiment 1. This helps to establish the robust-ness of the Wndings reported here.

In addition, whereas in Experiment 1 participantsfaced a net gain that was, depending on conditions,equal, better, or worse than their partner’s, in Experi-ment 2 we varied that participants had to pay an equalamount of money as their partner’s or had to pay eithermore or less money than their partner’s. In other words,whereas in Experiment 1 participants faced a gain per-spective, in Experiment 2 they were confronted with aloss frame. Based on the literature on gain and lossframes (see, e.g., De Dreu, Lualhati, & McCusker, 1994)one could well argue that this is important because test-ing our hypothesis in loss domains represents a strongtest of the line of reasoning we put forward in this paper.Obtaining evidence for our hypothesis using both gain-frame (Experiment 1) and loss-frame (Experiment 2)materials further contributes to the generalizability ofour analysis of the social psychology of advantageousinequity.

It should be noted, however, that in both Experiments1 and 2 participants read and responded to hypotheticalsituations. One might wonder whether similar resultswould be obtained when participants were exposed to asituation in which they directly experienced a distribu-tion of outcomes. As a third test of our predictions,therefore, we manipulated the cognitive processing ofparticipants in an experiment in which outcome distribu-tions were directly experienced by them.

Experiment 3

Experiment 3 was constructed following the experi-mental paradigm developed by Van den Bos et al. (1997,Experiment 2). In this experimental situation, partici-pants completed tasks together with another participantand learned that their own task performance was com-parable to the other participant’s task performance. Asin Experiments 1 and 2, the outcome that participantsreceived was held constant and we varied whether thisoutcome was equal to the outcome of the other partici-pant, better than the outcome of the other participant, orworse than the other participant’s outcome. Thecognitive busyness manipulation was the same as in

280 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

Experiment 1: Half of the participants rehearsed a stringof eight symbols while responding to the distribution ofoutcomes and half of the participants rehearsed onesymbol. The dependent variable was again participants’judgments of outcome satisfaction, but, to improve thereliability of participants’ satisfaction with equitable andinequitable arrangements, we included two additionalquestions (asking participants how satisWed and happythey were with the distribution of outcomes), yielding areliable four-item scale of participants’ evaluations ofoutcome satisfaction. Additionally, a similar scale ofparticipants’ outcome justice judgments was included inthe experiment to check whether our outcome manipula-tion would yield the justice judgments eVects commonlyfound in equity studies (see, e.g., Austin et al., 1980; Vanden Bos et al., 1997, 1998).

Method

Participants and designOne hundred and sixty-eight students (68 men and

100 women) at the Free University Amsterdam partici-pated in the experiment and were paid for their partici-pation. Participants were randomly assigned to one ofthe conditions of the 2 (cognitive busyness: low vs.high)£3 (outcome: equal to other vs. better than othervs. worse than other) factorial design.

Experimental procedureParticipants completed the experiment before partici-

pating in other, unrelated experiments. The experimentslasted a total of 45 min, and participants were paid 10Dutch guilders for their participation. Participants wereinvited to participate in a study on how people performtasks. On arrival at the laboratory, participants were ledto separate cubicles, each of which contained a computerwith a monitor and a keyboard. The computers wereused to present the stimulus information and to measurethe manipulation checks and the dependent variables.

In the Wrst part of the instructions, participants wereinformed that they would be participating in the experi-ment with another person, referred to as “Other.” Theexperimental procedure was then outlined to the partici-pants: After the experimental tasks were explained, par-ticipants would work on the tasks for 10 min.Furthermore, participants were informed that they andOther would get a certain amount of money at the end ofthe experiment if they both had performed well on thetasks. The amount of money would be divided asbonuses among the participant and Other. What exactamount of money would be allocated to the participantswould become known at the end of the experiment. Afterparticipants had completed the 10-min work round, theexperimenter would divide bonuses between the partici-pants and Other. Three practice questions were posed toensure comprehension of the course of events. If partici-

pants gave a wrong answer to a question, the correctanswer was disclosed, and main characteristics of thecourse of events were repeated.

Then the tasks were explained to the participants: Par-ticipants were asked to answer questions that measuredgeneral knowledge for 10 min. They were told that boththey and Other had to answer the same questions. Afterthe work round had ended, participants were told howmany questions they had answered correctly, and it wascommunicated to the participants that Other hadanswered an equivalent amount of questions correctly(cf. Van den Bos et al., 1997). To assess whether partici-pants thought of Other as a person who was comparablein the amounts of inputs he or she provided (cf. Van denBos, 1999; Van den Bos et al., 1997), they were asked towhat extent Other had performed well when answeringthe general knowledge questions relative to the perfor-mance of the participant self (1Dmuch worse, 4D equally,7Dmuch better), to what extent Other did his/her bestwhen answering the knowledge questions relative to theparticipant self (1Dmuch less, 4D equally, 7Dmuchmore), and to what extent Other was good in performingthe knowledge questions relative to the participant self(1Dmuch worse, 4D equally, 7Dmuch better).

After this, cognitive busyness was manipulated in thesame way as in Experiment 1: Participants in the high-busyness condition were asked to remember a string ofeight symbols and those in the low-busyness conditionwere asked to remember one symbol.

Participants were then told that the experimenterwould divide bonuses between them and Other. It wascommunicated to the participants that they receivedthree bonuses. This was followed by the manipulation ofoutcome of Other. In the better-than-other condition,participants were informed that Other received onebonus. In the equal-to-other condition, participants wereinformed that Other received three bonuses. In theworse-than-other condition, participants were informedthat Other received Wve bonuses.

Participants were then asked questions pertaining tothe dependent variable and manipulation checks. Theprimary dependent variable was assessed by asking par-ticipants how satisWed they were with their three bonuses(1D very dissatisWed, 7D very satisWed), how happy theywere with their three bonuses (1D very unhappy, 7D veryhappy), how satisWed they were with the division ofbonuses (1D very dissatisWed, 7D very satisWed), and howhappy they were with the division of bonuses (1D veryunhappy, 7D very happy). These items were highly corre-lated (rs > .51, ps < .001) and were averaged to form areliable scale of perceived outcome satisfaction (�D .86).

As a check on the manipulation of the three outcomeconditions, participants were asked two questions foreach condition. SpeciWcally, to check the manipulationof the better-than-other condition, participants wereasked to what extent they agreed with the statement that

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 281

they received more bonuses than the other participant(1D strongly disagree, 7D strongly agree) and to whatextent they agreed with the statement that the other par-ticipant received less bonuses than they received(1D strongly disagree, 7D strongly agree). Answers tothese two items were strongly correlated (rD .95,p < .001) and were averaged to form a reliable check ofthe better-than-other condition (�D .97). To check themanipulation of the equal-to-other condition, partici-pants were asked to what extent they agreed with thestatement that they received an equal number of bonusesas the other participant (1D strongly disagree,7D strongly agree) and to what extent they agreed withthe statement that the other participant received anequal number of bonuses as they received (1D stronglydisagree, 7D strongly agree). Answers to these two itemswere strongly correlated (rD .94, p < .001) and averagedto form a reliable check of the equal-to-other condition(�D .97). Finally, to check the manipulation of theworse-than-other condition, participants were asked towhat extent they agreed with the statement that theyreceived less bonuses than the other participant(1D strongly disagree, 7D strongly agree) and to whatextent they agreed with the statement that the other par-ticipant received more bonuses than they received(1D strongly disagree, 7D strongly agree). Answers tothese two items were strongly correlated (rD .90,p < .001) and averaged to form a reliable check of theworse-than-other condition (�D .95).

To further check the manipulation of outcome, par-ticipants’ outcome justice judgments were solicited byasking participants how just they considered the threebonuses they received (1D very unjust, 7D very just), howfair they considered the three bonuses they received(1D very unfair, 7D very fair), how just they judged thedivision of the bonuses (1D very unjust, 7D very just),and how fair they judged the division of the bonuses(1D very unfair, 7D very fair). These four items werehighly correlated (rs > .67, ps < .001) and were averagedto form a reliable scale of outcome justice judgments(�D .91).

As check on the manipulation of cognitive busyness,participants were asked to what extent they agreed withstatements that during the study they were busy trying toremember the symbols (1D strongly disagree,6D strongly agree), that it was diYcult to remember thesymbols (1D strongly disagree, 6D strongly agree), thattheir thoughts were kept busy remembering the symbols(1D strongly disagree, 6D strongly agree), and that ittook much eVort to remember the symbols (1D stronglydisagree, 6D strongly agree). These items were highlycorrelated (rs > .47, ps < .001) and were averaged to forma reliable scale of cognitive busyness (�D .87). Aftercompleting these measures, participants were asked towrite down the symbols they were supposed to rehearse.When the participants had done this, and had completed

the other experiments in which they would participate,they were thoroughly debriefed and paid for theirparticipation.

Results

Relative outcomesA 2£3 MANOVA on the three manipulation checks

of outcome yielded only a main eVect of outcome at boththe multivariate level and the univariate levels, multivari-ate F(6,320)D585.77, p < .001; for the better-than-othercheck, F (2,162)D826.38, p <.001; for the equal-to-othercheck, F (2,162)D589.09, p< .001; for the worse-than-other check, F(2,162)D943.93, p <.001. To interpret theseeVects, we performed a least signiWcant diVerence test formultiple comparisons between means (p< .05) for eachmanipulation check, with the three conditions of the out-come manipulation serving as the independent variable(Kirk, 1982). This showed that participants in the better-than-other condition agreed more with the statementsthat they received an outcome that was better than theother participant’s outcome (MD 6.6, SDD1.2) thanparticipants in the equal-to-other (MD1.2, SDD0.5)and worse-than-other conditions (MD 1.2, SDD 0.6) andthat no other diVerences between conditions were signiW-cant. Participants in the equal-to-other condition agreedmore with the statements that their outcome was equalto the outcome of the other participant (MD6.8,SDD0.7) than participants in the better-than-other(MD1.3, SDD1.1) and worse-than-other conditions(MD1.3, SDD1.0); no other diVerences between condi-tions were signiWcant. Participants in the worse-than-other condition agreed more with the statements thattheir outcome was worse than the other participant’soutcome (MD6.7, SDD 1.0) than participants in theequal-to-other (MD1.2, SDD 0.7) and better-than-otherconditions (MD1.2, SDD0.5); no other diVerencesbetween conditions were signiWcant. These Wndings sug-gest that outcome was successfully operationalized.

Justice judgmentsParticipants’ outcome justice judgments yielded only

a main eVect of outcome, F (2, 162)D99.57, p < .001. Themain eVect of cognitive busyness, F (1, 162)D 0.04,p > .83, and the interaction eVect between cognitive busy-ness and outcome, F (2, 162)D0.28, p > .75, were bothnonsigniWcant. Table 3 shows the means and standarddeviations of the outcome justice scale and the results ofa least signiWcant diVerence test for multiple compari-sons between means (p < .05), with the six cells of ourdesign serving as the independent variable (Kirk, 1982).As expected, participants in the equal-to-other conditionjudged their outcome to be more just (MD5.7, SDD0.8)than those in the better-than-other (MD3.2, SDD1.1)and worse-than-other conditions (MD 3.1, SDD 1.3) andno other diVerences between conditions were signiWcant.

282 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

These Wndings are in accordance with previous equitystudies (e.g., Van den Bos et al., 1997, 1998) and yieldadditional evidence that the manipulation of outcomewas perceived as intended.

Recalled symbolsAs expected, the large majority of the participants

were able to accurately recall the symbols they were sup-posed to rehearse: All participants in the low-busynesscondition accurately recalled the symbol they had torehearse. Ninety-three percent of the participants in thehigh-busyness condition recalled at least six symbolsaccurately (28% of those accurately recalled six symbols,41% accurately recalled seven symbols, and 31% accu-rately recalled eight symbols); the number of recalledsymbols in this condition did not diVer as a function ofthe outcome manipulation, F < 1. These results indicatethat participants took the memorization task seriouslyand made an eVort to recall the symbols while complet-ing the experiment.

Cognitive busynessA 2£ 3 ANOVA on the manipulation check of cogni-

tive busyness yielded only a main eVect of busyness,F (1,162)D 201.20, p < .001. Inspection of the means indi-cated that participants in the high-busyness conditionwere cognitively busier with remembering the symbols(MD3.1, SDD1.0) than participants in the low-busynesscondition (MD1.4, SDD1.0). This shows that themanipulation of cognitive busyness was successful ininXuencing the relative strength of cognitive busyness inways that were intended with this manipulation.

Comparability measuresThe answers that participants gave on the questions

that assessed whether participants thought of the otherparticipant as a comparable person did not yield any sig-niWcant results at either the multivariate level or the uni-variate levels. Inspection of the means indicated thatparticipants thought that the other participant hadperformed equally well when answering the generalknowledge questions (MD 4.0, SDD0.4), had doneequally his/her best when answering the questions

Table 3Means and standard deviations of outcome justice judgments as afunction of cognitive busyness and outcome (Experiment 3)

Note. Means are on 7-point scales, with higher values indicating higherlevels of outcome justice. Means with no subscripts in common diVersigniWcantly (p < .05), as indicated by a least signiWcant diVerence testfor multiple comparisons between means (Kirk, 1982).

Cognitive busyness

Outcome

Equal to other Better than other Worse than other

M SD M SD M SD

Low 5.6a 0.9 3.3b 1.3 3.0b 1.1High 5.6a 0.9 3.2b 1.0 3.2b 1.3

(MD4.0, SDD0.2), and was equally good in answeringthe questions (MD 4.0, SDD0.3). It can be concludedthat participants thought of the other person as a com-parable person with respect to the tasks they completedin the experiment.

Outcome satisfactionA 2£ 3 ANOVA on the outcome satisfaction scale

yielded a main eVect of outcome, F (1, 162)D34.45,p < .001, and a signiWcant interaction eVect,F (2,162)D 3.40, p < .04. The main eVect of cognitivebusyness was not signiWcant, F < 1. To interpret theseeVects we performed a least signiWcant diVerence test formultiple comparisons between means (p < .05), with thesix cells of our design serving as the independent variable(Kirk, 1982). Table 4 shows the result of this test and themeans and standard deviations of the outcome satisfac-tion scale. As hypothesized, Wndings revealed that withinthe better-than-other condition, participants were moresatisWed with their outcome in the high-busyness condi-tion than in the low-busyness condition. There were noeVects of cognitive busyness within the equal-to-otherand worse-than-other conditions.

Additionally, it can be noted here that within the low-busyness conditions, outcome satisfaction was higher inthe equal-to-other conditions than in the better-than-other condition and was higher in the better-than-othercondition than in the worse-than-other condition (seeTable 4). This pattern of means is in accordance withprevious equity Wndings (e.g., Buunk & Van Yperen,1989; Van den Bos et al., 1997, 1998). In the high-busy-ness condition, participants in the better-than-other con-dition were also more strongly satisWed than those in theworse-than-other condition, but they were as satisWed asthose in the equal-to-other condition.

Discussion

The Wndings of Experiment 3 replicate and extend theresults of Experiments 1 and 2 and are supportive of ourline of reasoning: As hypothesized, Wndings reveal thatpeople are more satisWed with arrangements of advanta-geous inequity under conditions of high cognitive

Table 4Means and standard deviations of outcome satisfaction as a functionof cognitive busyness and outcome (Experiment 3)

Note. Means are on 7-point scales, with higher values indicating higherlevels of outcome satisfaction. Means with no subscripts in commondiVer signiWcantly (p < .05), as indicated by a least signiWcant diVerencetest for multiple comparisons between means (Kirk, 1982).

Cognitive busyness

Outcome

Equal to other Better than other Worse than other

M SD M SD M SD

Low 5.3a 1.0 4.6b 0.8 3.9c 1.0High 5.0a,b 0.8 5.1a 0.9 3.7c 1.0

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 283

busyness than under conditions of low cognitive busy-ness. Furthermore, results of Experiment 3 have beenobtained using an experimental paradigm in which par-ticipants directly experienced the distribution of out-comes. Thus, the predicted eVects of cognitive busynesson reactions to advantageous inequity can be found usingdiVerent scenario experiments (Experiments 1 and 2),using an experiment in which people directly experiencethe outcome distribution (Experiment 3), and usingdiVerent manipulations of limiting of cognitive process-ing (Experiments 1 and 3 vs. 2). It can now be concludedthat all three experiments reported here show that peo-ple are more satisWed with arrangements of advanta-geous inequity when their cognitive processing has beenstrongly (as opposed to weakly) limited.

A critic, however, might note two objections towardthe Wndings presented thus far. First, a close inspection ofthe Wndings of Experiment 3 shows that, in the advanta-geous inequity conditions, our dependent variable of out-come satisfaction was inXuenced by the cognitivebusyness manipulation whereas the manipulation checkof outcome justice judgments was not. As noted earlier inthe introduction of this paper and in the research litera-ture (see, e.g., Messick & Sentis, 1983; Van den Bos et al.,1997, 1998), this dissociation is not surprising: Theoreti-cally, one would expect that satisfaction evaluations aremore susceptible to the conXict between egoism-basedpreferences and what is right, the conXict of interest in thepresent research. Moreover, the dissociation is consistentwith other research demonstrating that satisfaction evalu-ations and justice judgments are quite diVerent variablesto which people can react quite diVerently (e.g., Austinet al., 1980; Blau, 1964). More speciWcally, research hasshown that people often judge advantageous inequity tobe unfair (indeed as unfair as disadvantageous inequity)although simultaneously they are moderately satisWedwith advantageous inequity (relative to disadvantageousinequity) (see, e.g., Van den Bos et al., 1997, 1998).

Although the results of Experiment 3 are in line withour expectations, we conducted an additional experimentto rule out the alternative hypothesis that the Wndingsobtained in that experiment were an artifact of the order inwhich the satisfaction and justice measures were measuredin Experiment 3. Thus, to assess whether the manipulationof cognitive busyness would have inXuenced evaluations ofoutcome justice had the justice items been administeredprior to the measurement of participants’ satisfaction, par-ticipants in the additional experiment reacted to the samestimulus materials as in Experiment 1, including the samearrangements of advantageous inequity and the same con-ditions of either high or low cognitive busyness used inExperiment 1. Furthermore, we measured the same out-come satisfaction ratings as in Experiment 1 (asking partic-ipants how satisWed and happy they were with the bonusthey received; �D .95), assessed similar items that solicitedparticipants’ outcome justice judgments (asking partici-

pants how just and fair they judged the bonus theyreceived; �D .81), and, most importantly, varied the orderin which ratings of outcome satisfaction and judgments ofoutcome justice were assessed.

Results of this additional experiment showed that par-ticipants’ evaluations of outcome satisfaction yielded only amain eVect of cognitive busyness, F(1,91)D4.23, p<.05.The main eVect of order of measurement, F(1,91)D0.48,p>.48, and the interaction eVect between cognitive busy-ness and order, F(1,91)D0.00, p>.97, were not signiWcant.As expected, participants were more satisWed with theiroutcome in the high-busyness condition (MD3.5, SDD1.4)than in the low-busyness condition (MD2.9, SDD1.7).

Furthermore, participants’ judgments of outcomejustice yielded no signiWcant eVects: The main eVect ofcognitive busyness, F (1, 91)D0.79, p > .37, the maineVect of order of measurement, F (1, 91)D0.07, p > .79,and the interaction eVect between cognitive busynessand order, F (1, 91)D0.07, p > .79, were all nonsigniWcant(MD2.9, SDD1.6). Thus, these results show that vary-ing the order in which participants rated outcome justiceand outcome satisfaction did not yield diVerent resultsthan those obtained in Experiment 3: Satisfaction evalu-ations did show signiWcant eVects of the busyness manip-ulation whereas justice judgments did not.

A second objection that might be raised against ourseries of experiments is that one can wonder whetherthe predicted eVects of cognitive busyness on satisfac-tion with advantageous inequity would also show up insituations that are more similar to those faced by peoplein real interactions. To this end, we conducted yetanother experiment in which we used an experimentalparadigm that was very diVerent from those used ear-lier. In brief, in Experiment 4, participants actuallyinteracted with the experimenter and the other partici-pant, and we induced advantageous inequity in a waythat more closely resembled what may happen outsideof the lab. The manipulation of cognitive busyness wasalso diVerent from those used in the earlier experiments,and it had greater mundane realism. Finally, in contrastto Experiment 3, where participants were told explicitlythat the other participant’s performance was equivalentto theirs, in Experiment 4, we examined the generaliz-ability of our eVect to a situation in which there wasmore ambiguity about the other participant’s inputs orperformance.

Experiment 4

Method

Participants and designForty students (11 men and 29 women) at Utrecht

University participated in the experiment and were paidfor their participation. Participants were randomly

284 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

assigned to one of the conditions of cognitive busynessmanipulation (low vs. high).

Experimental procedureParticipants of Experiment 4 were invited by means of

Xyers on diVerent spots at the campus of Utrecht Univer-sity to participate in a marketing study, ostensibly con-ducted by a marketing company called “TestMe-Research”to assess people’s evaluations of diVerent brands of choco-late. On the Xyers, participants were informed that thestudy would take about 20min of their time and that theywould be paid 3 Euros for their participation (about $3U.S. at the time the experiment was conducted).

When a participant arrived at the rooms where themarketing research would be conducted, the participantwas asked to wait in the waiting room. After 1 min asame-sex other student walked into the waiting room.In reality, this student was not a real participant butwas a confederate that we used to create an arrange-ment of advantageous inequity in a real-life way. Oneminute later, the experimenter came into the waitingroom and informed the real participant and the bogusparticipant that they would participate together in themarketing study. The experimenter then led the twoparticipants to a small room of 3£ 3 m where the mar-keting study would be conducted. In this room, the par-ticipants were seated 1 m apart from each other in sucha way that the real participant could easily see thebogus participant.

The experimenter then explained that the researchwas conducted in commission of a marketing companycalled “TestMe-Research.” Participants were told thatthis marketing company conducted many studies andthat the particular study in which they would participatetoday would assess people’s evaluations of diVerentbrands of chocolate. To this end, the experimenterexplained, the participants would be asked to taste 10diVerent pieces of chocolate and to evaluate these pieceson diVerent dimensions. Participants were informed thateach piece of chocolate would be handed over to themwhile the experimenter would mention the particularbrand of that piece. After carefully having tasted a piece,participants would be asked to evaluate the piece interms of sweetness, structure, whether it tasted good ornot, and what grade they would give to the piece’s Xavor.Participants would then drink some water and wouldtaste and evaluate the next piece of chocolate.

In the high-busyness condition, both the real partici-pant and the bogus participant were asked to recall all10 brand names that the experimenter would mention tothem while handing over the pieces of chocolate. In thelow-busyness condition, both real and bogus partici-pants were asked to recall only the Wrst brand name. Inboth busyness conditions, both real and bogus partici-pants were asked to recall the brand names until theywould be asked to reproduce them.

Participants were told that after having tasted the 10pieces they would be paid for their participation. As partof standard practice at “TestMe-Research,” participantsthen would be asked to evaluate the particular studythey participated in. Participants were informed that theresults of this evaluation would be used to improvefuture studies by “TestMe-Research.”

After it was made sure that participants understoodthe course of events, participants tasted and evaluatedthe 10 pieces of chocolate. After this, the experimenterpaid participants for their participation. The experi-menter’s behavior was then used to create an arrange-ment of advantageous inequity: The experimenter took 6Euro coins and laid down 4 Euros in front of the realparticipant and 2 Euros in front of the bogus partici-pant. The experimenter laid down the Euros in such away that the real participants clearly saw the amount ofEuros they and the bogus participants received.

The experimenter then handed over the evaluationforms to the participants. In these forms, participantswere asked several questions that pilot testing hadrevealed could well be part of a general marketing ques-tionnaire. Embedded in these questions were our depen-dent variables and manipulation checks: The maindependent variable was assessed by asking participantshow satisWed (1D very dissatisWed, 7D very satisWed) andhappy (1D very unhappy, 7D very happy) they were withtheir payment. Participants’ answers to these two itemswere highly correlated (rD .63, p < .001) and were aver-aged to form a reliable scale of perceived outcome satis-faction (�D .77).

Participants’ outcome justice judgments were solic-ited by asking participants how just (1D very unjust,7D very just) and fair (1D very unfair, 7D very fair) theyjudged their payment to be. These items were highly cor-related (rD .78, p < .001) and were averaged to form areliable scale of outcome justice judgments (�D .87). Ascheck on the manipulation of cognitive busyness, partic-ipants were asked to what extent it was diYcult toremember the brands of chocolate they were asked toremember (1D strongly disagree, 7D strongly agree) andthat it took much eVort to remember the brands(1D strongly disagree, 7D strongly agree). These itemswere highly correlated (rD .82, p < .001) and were aver-aged to form a reliable scale of cognitive busyness(�D .90). Finally, participants were asked to write downthe brand names they were supposed to rehearse. Whenthe participants had done this, they were thoroughlydebriefed and thanked for their participation.

Results

Recalled symbolsAs in the previous experiments, the majority of the par-

ticipants were able to accurately recall the brand namesthey were supposed to rehearse: All participants in the

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 285

low-busyness condition accurately recalled the brand namethey had to rehearse. Seventy-Wve percent of the partici-pants in the high-busyness condition recalled at least sixbrand names accurately (27% of those accurately recalledsix brands, 20% recalled seven brands, 27% recalled eightbrands, 20% recalled nine brands, and 7% recalled 10brands). These results indicate that participants took thememorization task seriously and made an eVort to recallthe symbols while completing the experiment.

Cognitive busynessAs expected, the cognitive busyness scale (excluding

one participant with missing values on one of the items ofthe scale) yielded a signiWcant eVect of the cognitive busy-ness manipulation, F(1,37)D8.22, p < .01. Inspection ofthe means indicated that participants in the high-busynesscondition were cognitively busier with remembering thebrand names (MD4.8, SDD1.7) than participants in thelow-busyness condition (MD3.2, SDD1.9). This showsthat the manipulation of cognitive busyness was successfulin inXuencing the relative strength of cognitive busyness inways that were intended with this manipulation.

Justice judgmentsParticipants’ outcome justice judgments yielded no

eVect of the cognitive busyness manipulation,F (1, 38)D 0.00, p > .96, thus revealing that perceived jus-tice of the payment participants received did not diVer asa function of being in the high-busyness condition(MD5.2, SDD 1.6) or the low-busyness condition(MD5.2, SDD 2.0). These Wndings are in accordancewith our justice judgment data presented earlier.

Outcome satisfactionAs predicted, the outcome satisfaction scale showed a

signiWcant eVect of the manipulation of cognitive busy-ness, F (1,38)D5.38, p < .03. Inspection of the meansrevealed that, as hypothesized, participants were moresatisWed with their payment in the high-busynesscondition (MD 5.5, SDD 1.0) than in the low-busynesscondition (MD4.8, SDD0.7).

General discussion

The question of how people deal with conXicts betweenegoism-based preferences and fairness-based concerns hasfascinated and puzzled philosophers and social theoristsfor centuries (see, e.g., Adams, 1965; Beauchamp, 2001;Blau, 1964; Cohen, 1986; Diekmann et al., 1997; Epley& Caruso, 2004; Epley et al., 2004; Frey et al., 2003; Jac-ques, 1961; Loewenstein et al., 1989; Messick & Sentis,1979, 1983; Moore & Loewenstein, 2004; Peters et al., 2004;Smeesters et al., 2003; StouVer et al., 1949; Thibaut & Kel-ley, 1959). The experimental Wndings that have been pre-sented here may shed new light on how to answer this

question as they reveal that people are more satisWed withadvantageous inequity when cognitive processing isstrongly as opposed to weakly limited. These Wndings arein accordance with our analysis of the social psychology ofadvantageous inequity. Moreover, these results wereobtained building on our integration of the research onperson inference and correction processes with frameworkson outcome evaluation and preference and adjustmentprocesses. The fact that we found converging evidence forour hypothesis in diVerent experimental paradigms usingdiVerent experimental manipulations is compelling andunderscores the reliability of the Wnding revealed here thatpeople are more satisWed with advantageous inequityunder conditions of high cognitive busyness.

In Experiment 4, we tried to speak to the importanceof cognitive busyness eVects on satisfaction with anadvantageous unfair situation in real life. Because it right-fully has been argued that social interactions often requireconsiderable attention of people to monitor all the thingsthat are going around (see, e.g., Bargh, 1994; Bargh &Thein, 1985; Gilbert & Osborne, 1989), the Wndings of theexperiments that were presented here may well indicatethat when people have a lot on their minds, they mayreact in a more self-centered way to unfair arrangementsthat are to their own advantage than when they have thecapacity to ponder about these things more deeply.

In previous equity studies (e.g., Buunk & Van Yperen,1989; Van den Bos et al., 1997, 1998) it has usually beenfound that people are more satisWed with equitablearrangements than with advantageous inequitablearrangements and that they are more satisWed withadvantageous inequity than with disadvantageous ineq-uity. In the present paper, we have replicated these eVectsin the conditions of Experiments 1–3 where cognitiveprocessing was only weakly limited. More interesting,however, in all three experiments, we found that in con-ditions where cognitive processing was strongly limitedpeople’s evaluations of outcome satisfaction when react-ing to advantageous inequity were closer to the evalua-tions of people reacting to equitable arrangements. Infact, Wndings of Experiments 2 and 3 even show that, inconditions were cognitive processing was strongly lim-ited, people’s satisfaction with advantageous inequitymay not be statistically diVerent from their satisfactionwith equitable arrangements. The (small) diVerences inthe results across our experiments may have been causedby diVerences in operationalizations, but it is noteworthythat in all of our studies we found that variations incognitive processing exert reliable eVects on people’ssatisfaction with arrangements of advantageous inequity.

Implications

Extending on the earlier-mentioned frameworks(Epley & Caruso, 2004; Epley et al., 2004; Messick &Sentis, 1979, 1983; Moore & Loewenstein, 2004) and the

286 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

work by Zajonc (1980) and others (e.g., De Houwer,Thomas, & Bayens, 2001; Stapel et al., 2002), an impor-tant implication of the Wndings presented here may bethat when reacting to arrangements of advantageousinequity, judging the advantage is quick and easy aspreferences are primary whereas adjusting this appraisalrequires cognitive resources as it entails integrating fair-ness concerns with the initial preference appraisal. Thisimplication would suggest a two-phase model of peo-ple’s reactions to advantageous inequity, in which peo-ple’s very Wrst reaction when confronted withadvantageous inequity is one of pleasure (“wow, I getmore than someone else, that’s great!”). This egoism-based, gut reaction may happen in an automatic manner(Bargh, 1994). Furthermore, we propose that it is onlyafter this Wrst automatic reaction of pleasure that peopleconsider the fairness of the situation (“hey, but that’s notfair!”) This latter fairness-based reaction is not as auto-matic and fast as the Wrst egoism-based reaction. Peopleneed just a little bit of time (e.g., some seconds) to dis-cover, understand, and respond to the unfairness of a sit-uation in which they are better oV than others, and thiscorrection process can only take place, therefore, whenpeople have enough cognitive resources available.

Besides helping in getting more insight into the pro-cess with which people may react to advantageous ineq-uity in lab experiments in particular, and being better oVthan others in society more generally, another importantimplication of our line of reasoning and the two-phaseprocess it suggests, is that our Wndings have revealed thatthe hypothesis we tested here is conditional on the typeof outcome distribution that is being evaluated and theexact type of evaluation that is being made. That is, ourWndings indicate that the eVects of cognitive busynesswere found reliably when satisfaction with advantageousinequity is assessed (and not when satisfaction withequity or disadvantageous inequity or when justice judg-ments of advantageous inequity are measured). This sug-gests that the relationship between preferences andfairness concerns is dependent on some speciWc condi-tions outlined in the current paper and in this way ourWndings may contribute to further specifying modelsrelated to people’s preferences and their adjustment pro-cesses (e.g., Epley & Caruso, 2004; Epley et al., 2004;Messick & Sentis, 1979, 1983; Moore & Loewenstein,2004; Stapel, Koomen, & Ruys, 2002; Zajonc, 1980).

As noted earlier, the Wndings of both Experiments 3and 4, as well as those discussed in Discussion of Experi-ment 3, showed that cognitive busyness inXuenced evalu-ations of outcome satisfaction but not ratings ofoutcome justice. In various experiments presented here,therefore, judgments of outcome justice were not inXu-enced by the manipulation of cognitive busynesswhereas evaluations of outcome satisfaction were. In ouropinion, these diVerential Wndings highlight the impor-tance of distinguishing between judgments of justice and

perceived satisfaction (cf. Austin et al., 1980; Blau, 1964;Messick & Sentis, 1983; Van den Bos et al., 1997, 1998).Moreover, as outlined in detail in the general introduc-tion, this may imply that satisfaction evaluations aremore susceptible to the conXict between egoism-basedpreferences and doing the right thing, the conXict ofinterest in the present research. Related to this, it seemsreasonable to argue that explicitly asking participants toindicate their justice judgments may well have a higherlikelihood of making the fairness aspects of advanta-geous inequitable arrangements salient than doesmeasuring satisfaction evaluations, hence making satis-faction evaluations a better dependent variable for thehypothesis under consideration. Although we are con-vinced that the patterns of our Wndings are generalizableto other social contexts and experimental manipulations,it may be of interest for researchers to explore theboundary conditions of the eVects reported here in thefuture. This said, however, for now, our research demon-strates that particular eVects can occur, more speciWcally,that people tend to be more satisWed with advantageousinequity when cognitive processing is limited. This maystimulate future research into other conditions in whichsuch eVects are or are not found.

In Experiments 1–3, we included conditions of equita-ble and disadvantageously inequitable arrangements todetermine whether eVects of the cognitive busynessmanipulations would be found in these conditions aswell. Although we found consistent eVects of cognitivebusyness in the advantageous conditions, in none of thethree experiments did we Wnd any eVects of cognitivebusyness in the other conditions (even when we testedthe eVects with one-tailed t tests). These may be impor-tant Wndings because they indicate that it is especiallywhen people are reacting toward advantageous inequitythat their reactions are likely to inXuenced by being cog-nitively busy or not.

The absence of reliable cognitive busyness eVectswithin the disadvantageous inequity conditions is espe-cially interesting in this respect. After all, one couldargue that when people experience disadvantageousinequity there are two sources of negative aVect: (a) thedispleasing feeling caused by the relative deprivation oflacking what the other person has received, and (b) theunfairness of the outcome distribution (cf. Van den Boset al., 1997, 1998). Furthermore, because one could pro-pose that people generally have more immediate accessto or knowledge of their preferences than to what is fair(e.g., Messick & Sentis, 1983), one might reason thatwhen people are evaluating disadvantageous inequitywhile their cognitive resources are simultaneously beingdepleted, their satisfaction evaluations would be inXu-enced only or primarily by the displeasing aspects of theoutcome distribution. In the absence of cognitive busy-ness, however, the individual’s satisfaction should beinXuenced not only by the displeasing aspects of the

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 287

arrangements but also be corrected for the unfairness ofthe distribution as well.

Thus, one could argue that because some correctionmay be assumed to take place in the disadvantageousinequity conditions, it should be possible to Wnd eVectsof cognitive busyness in these conditions. However, it isalso conceivable that correcting one’s initial impressionin an opposite direction (as we argue is the case whenreacting to advantageous inequity) takes up more cogni-tive resources than does correcting in the same directionas one’s initial impression. In short, it is much more cog-nitively demanding to decide one’s reactions to advanta-geous inequity than to decide one’s reactions todisadvantageous inequity. Accordingly, one mightexpect the eVects of cognitive busyness to be more evi-dent in the advantageous inequity condition as com-pared to the disadvantageous inequity condition, asfound in our research.

This latter account explains why consistent busynesseVects were found within the advantageous conditionsand unreliable eVects within the disadvantageous condi-tions. Related to this, people have a very strong negativereaction to disadvantageous inequality under virtuallyall circumstances (e.g., whether they are dealing with anabsolute stranger or a friend; Peters et al., 2004). Thesame is not true of feelings toward advantageousinequality, which depend in complex ways on the exactsituation (e.g., feelings toward advantageous inequalitytend to be more positive when dealing with a friend asopposed to a stranger; see Peters et al., 2004).

Furthermore, it is noteworthy that the fact that weconsistently found cognitive busyness eVects only in theadvantageous inequity conditions discounts a more gen-eral explanation of our data, speciWcally, that the resultsare merely due to general attentional demands (cf. Messick& Sentis, 1983). Such an explanation would state that ego-istic evaluations (beneWts to the self) may be thought ofas requiring few attentional resources and that theseevaluations may be conceived of as involving auto-matic processes that begin early in life and that areobservable across species. On the other hand, concernfor the welfare of others (fairness and ethical con-cerns) requires deliberative, reXective appraisal pro-cesses that make greater attentional demands(Messick & Sentis, 1983).

Had this more general attentional demands explana-tion been true, cognitive busyness eVects would havebeen likely to be found in disadvantageous inequity con-ditions as well. Instead, the current Wndings suggest thata more speciWc explanation that incorporates the direc-tion in which evaluations are corrected, better explainsthe psychology of satisfaction evaluations with outcomedistributions. This does not rule out the possibility thatwith stronger manipulations of cognitive busyness(e.g., asking participants to recall 16 as opposed to eightsymbols) it will be possible to Wnd reliable busyness

eVects on evaluations of disadvantageous inequity; nev-ertheless, on the basis of the Wndings reported here, wewould predict that, even when including conditions ofvery high cognitive busyness, larger eVects of busynessmanipulations will be found when people are reacting toadvantageous inequity.3

All in all, it seems reasonable to conclude that theWndings reported here are in line with what was pre-dicted on the basis of integrating the work by Messickand Sentis (1979, 1983), Moore and Loewenstein (2004),and Epley c.s. (Epley & Caruso, 2004; Epley et al., 2004)with the work by Gilbert and others (e.g., Gilbert& Osborne, 1989; Gilbert et al., 1988; Ham & Vonk,2003), and the research reveals consistently that peopleare more satisWed with advantageous inequity underconditions of strongly limited cognitive processing.Future research may want to explore the implications ofthe Wndings presented here.

Future research

It is noteworthy that our suggestion that preferencesare more immediate than fairness considerations corre-sponds with Zajonc’s (1980) position that feelings andpreferences may be more primary than are thoughts andinferences (see also De Houwer et al., 2001; Stapel et al.,2002). We hasten to say that the exact processes underly-ing Zajonc’s position have been disputed (for anoverview of this discussion, see, e.g., Winkielman& Cacioppo, 2001) and similarly that the implicationsthat follow from our research should be validated infuture research. In particular, more research is needed, ofcourse, to elucidate the subtle but potentially excitingpsychological processes involved when people are facedwith conXicts between preferences and their inclinationsof doing the right thing. It is our hope that our experi-mental Wndings provide a new perspective on the empiri-cal study of preferences and fairness concerns and thatthe current paper may therefore stimulate researchers toinvestigate the implications of the studies presented here.

Furthermore, it should be pointed out explicitly thatthe current paper conWrms patterns of important eVectsthat have been predicted by the classical and modernsocial psychological frameworks mentioned earlier, butthe patterns reported here have implications for whetheror not the processes implicated by these frameworkshold up in the long run. Indeed, in the future, researchersmay productively investigate the psychological processes

3 Indeed, in Experiment 1 there was some tendency for participantsto be more satisWed with disadvantageous inequity under conditions ofhigh as opposed to low cognitive busyness. However, this eVect was notstatistically signiWcant, t (41) D 1.27, p < .11 (one-tailed), was not repli-cated in the other experiments, and—as would be predicted by ouraccount—was weaker than the cognitive busyness eVect within the ad-vantageous inequity condition.

288 K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289

suggested by our research Wndings. We hope that ourWndings stimulate further interest in the captivatinginterplay between egoism- and fairness-based factors ofoutcome evaluations as well as more general preferenceand adjustment processes.

A candidate for another process explaining people’sreactions to advantageous inequity may be suggested bya study by Rivera and Tedeschi (1976). Participants inthe bogus pipeline condition of this study were led tobelieve that a bogus apparatus could detect their truefeelings by implicit muscle responses whereas partici-pants in a paper-and-pencil condition were not giventhis impression. Furthermore, in the bogus pipeline con-dition, dependent variables were measured by askingparticipants to indicate their ratings by turning a dialmoving a pointer along a meter. In the paper-and-pencilcondition, dependent variables were assessed using thenormal paper-and-pencil procedures. Findings indi-cated that participants reported more happiness withadvantageous inequity in the bogus pipeline conditionthan in the paper-and-pencil condition. Rivera andTedeschi’s account for their Wndings is that when paper-and-pencil procedures are used people’s reactions arepublic whereas when bogus pipeline procedures are usedreactions are private. The authors further argue that,because people want to create positive impressions ofthemselves to others, they will report less happiness withadvantageous inequity in public circumstances.

It should be emphasized, however, that the Riveraand Tedeschi (1976) results have received an enormousamount of criticism (see, e.g., Ellard, Meindl, & Lerner,2004), and that more recent research Wndings suggestthat fairness is important to people even in completelyprivate circumstances (see, e.g., Turillo, Folger, Lavelle,Umphress, & Gee, 2002). Furthermore, there are severalissues that limit the validity of Rivera and Tedeschi’s(1976) interpretation of their Wndings, such as the con-found between the bogus pipeline procedure (present vs.absent) and the way in which dependent variables wereassessed (dial equipment vs. paper and pencil). More-over, it is not clear whether participants indeed perceivedthe paper-and-pencil conditions to be more public.

In addition, Roese and Jamieson (1993) noted intheir thorough review of the bogus pipeline researchthat when researchers wish to import the bogus pipelineprocedure to their research domain in an eVort toreduce impression management and social desirabilityeVects, they should Wrst demonstrate that some form ofimpression management bias indeed inXuences reac-tions in the domain of interest. Roese and Jamieson(1993) suggest that an obvious way to accomplish thiswould be to associate reactions in the domain of interestwith responses on impression management or socialdesirability scales. To the best of our knowledge, therehave been no equity studies that have incorporated oneof these scales. If future research Wndings would indeed

support Rivera and Tedeschi’s (1976) account of thediVerence between public and private reactions toadvantageous inequity (but see Ellard et al., 2004;Turillo et al., 2002), we would suggest that this accountis congruent with the line of reasoning we have put for-ward here: In private circumstances, preferences mayinXuence people’s outcome evaluations more stronglythan in public situations. In public (as opposed to pri-vate) circumstances people may be more strongly moti-vated to incorporate fairness-based considerations toconvey a positive impression of themselves to others,especially when they have enough cognitive capacity fordoing so. We explicitly want to note here that both thesocial desirability eVects suggested by the Rivera andTedeschi (1976) study as well as the preferences versusfairness considerations line of thinking presented earlierin this paper need further research, as does the relation-ship between these two lines of thought and the pro-cesses they suggest.

In concluding, it is important to note that our line ofreasoning does not imply that preferences and pleasureare always selWsh, as people sometimes may prefer or bepleased to see that other persons than themselves (suchas their children or persons in need) receive more of avalued resource than they themselves receive. Althoughour Wndings suggest that people’s primitive core maysometimes (e.g., when their cognitive capacities havebeen severely limited) push them in an egocentric direc-tion, it may well be the case that frequently people try tofree cognitive resources to do the right thing. Thus, weare not suggesting that fairness concerns are a myth.Quite the contrary, we work from the assumption thatfairness is frequently a very real concern to people (Vanden Bos & Lind, 2002; see also Miller & Ratner, 1998).That said, we also think that researchers should thor-oughly study the conditions under which fairness is moreof a concern to people and when it is less important tothem, as well as the processes underlying people’s con-siderations of fairness. It is our hope that with this paperwe have contributed a bit to what we see as the process-oriented future of the social psychology of preferencesand doing the right thing.

References

Adams, J. S. (1963a). Toward an understanding of inequity. Journal ofAbnormal Social Psychology, 67, 422–436.

Adams, J. S. (1963b). Wage inequities, productivity, and work quality.Industrial Relations, 3, 9–16.

Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.),Advances in experimental social psychology (Vol. 2, pp. 267–299).New York: Academic Press.

Adams, J. S., & Jacobsen, P. R. (1964). EVects of wage inequities onwork quality. Journal of Abnormal Social Psychology, 69, 19–25.

Adams, J. S., & Rosenbaum, W. B. (1962). The relationship of workerproductivity to cognitive dissonance about wage inequities. Journalof Applied Psychology, 46, 161–164.

K. van den Bos et al. / Journal of Experimental Social Psychology 42 (2006) 273–289 289

Austin, W., McGinn, N. C., & Susmilch, C. (1980). Internal standardsrevisited: EVects of social comparisons and expectancies on judg-ments of fairness and satisfaction. Journal of Experimental SocialPsychology, 16, 426–441.

Austin, W., & Walster, E. (1974). Reactions to conWrmations and dis-conWrmations of expectancies of equity and inequity. Journal ofPersonality and Social Psychology, 30, 208–216.

Bargh, J. (1994). The four horsemen of automaticity: Awareness,eYciency, intention, and control in social cognition. In R. S. WyerJr. & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., pp. 1–40). Hillsdale, NJ: Erlbaum.

Bargh, J. A., & Thein, R. D. (1985). Individual construct accessibility,person memory, and the recall-judgment link: The case of informa-tion overload. Journal of Personality and Social Psychology, 49,1129–1146.

Beauchamp, T. L. (2001). Philosophical ethics: An introduction to moralphilosophy (3rd ed.). Boston: McGraw-Hill.

Berkowitz, L., & Walster, E. (Eds.). (1976). Advances in experimentalsocial psychology, Volume 9: Equity theory—Toward a general the-ory of social interaction. New York: Academic Press.

Blau, P. M. (1964). Exchange and power in social life. New York: Wiley.Buunk, B. P., & Van Yperen, N. W. (1989). Social comparison, equality,

and relationship satisfaction: Gender diVerences over a ten-yearperiod. Social Justice Research, 3, 157–180.

Cohen, R. L. (1986). Justice: Views from the social sciences. New York:Plenum Press.

De Dreu, C. K. W., Lualhati, J. C., & McCusker, C. (1994). EVects ofgain-loss frames on satisfaction with self-other outcome diVerences.European Journal of Social Psychology, 24, 497–510.

De Houwer, J., Thomas, S., & Bayens, F. (2001). Associative learningof likes and dislikes: A review of 25 years of research on humanevaluative conditioning. Psychological Bulletin, 127, 853–869.

Diekmann, K. A., Samuels, S. M., Ross, L., & Bazerman, M. H. (1997).Self-interest and fairness in problems of resource allocation: Allo-cators versus recipients. Journal of Personality and Social Psychol-ogy, 72, 1061–1074.

Ellard, J. H., Meindl, J. R., & Lerner, M. J. (2004). Attributing egoisticmotives to norm-elicited behavior: Perpetuating the hegemony of self-interest. Submitted.

Epley, N., & Caruso, E. M. (2004). Egocentric ethics. Social JusticeResearch, 17, 171–188.

Epley, N., Keysar, B., Van Boven, L., & Gilovich, T. (2004). Perspectivetaking as egocentric anchoring and adjustment. Journal of Person-ality and Social Psychology, 87, 327–339.

Frey, B. S., Benz, M., & Stutzer, A. (2003). Introducing procedural util-ity: Not only what, but also how matters. Unpublished manuscript.

Gilbert, D. T., & Osborne, R. E. (1989). Thinking backward: Some cur-able and incurable consequences of cognitive busyness. Journal ofPersonality and Social Psychology, 57, 940–949.

Gilbert, D. T., Pelham, B. W., & Krull, D. S. (1988). On cognitive busy-ness: When person perceivers meet persons perceived. Journal ofPersonality and Social Psychology, 54, 733–740.

Jacques, E. (1961). Equitable payment. London: Heinemann Educa-tional Books.

Ham, J., & Vonk, R. (2003). Smart and easy: Co-occurring activationof spontaneous trait inferences and spontaneous situational infer-ences. Journal of Experimental Social Psychology, 39, 434–447.

Kirk, R. E. (1982). Experimental design: Procedures for the behavioralsciences (2nd ed.). Belmont, CA: Brooks/Cole.

Loewenstein, G. F., Thompson, L., & Bazerman, M. H. (1989). Socialutility and decision making in interpersonal contexts. Journal ofPersonality and Social Psychology, 57, 426–441.

Messick, D. M., & Sentis, K. (1979). Fairness and preference. Journal ofExperimental Social Psychology, 15, 418–434.

Messick, D. M., & Sentis, K. (1983). Fairness, preference, and fairnessbiases. In D. M. Messick & K. S. Cook (Eds.), Equity theory: Psy-chological and sociological perspectives (pp. 61–94). New York:Praeger.

Miller, D. T., & Ratner, R. K. (1998). The disparity between the actualand assumed power of self-interest. Journal of Personality andSocial Psychology, 74, 53–62.

Moore, D. A., & Loewenstein, G. (2004). Self-interest, automaticity,and the psychology of conXict of interest. Social Justice Research,17, 189–202.

Peters, S. L., Van den Bos, K., & Karremans, J. C. (2004). On thepsychology of the advantaged: How people react to being overpaid.Submitted.

Rivera, A. N., & Tedeschi, J. T. (1976). Public versus private reactionsto positive inequity. Journal of Personality and Social Psychology,34, 895–900.

Roese, N. J., & Jamieson, D. W. (1993). Twenty years of bogus pipelineresearch: A critical review and meta-analysis. Psychological Bulle-tin, 114, 363–375.

Smeesters, D., Warlop, L., Van Avermaet, E., Corneille, O., & Yzerbyt,V. (2003). Do not prime hawks with doves: The interplay of con-struct activation and consistency of social value orientation oncooperative behavior. Journal of Personality and Social Psychology,84, 972–987.

Stapel, D. A., Koomen, W., & Ruys, K. (2002). The eVects of diVuseand distinct aVect. Journal of Personality and Social Psychology, 83,60–74.

StouVer, S. A., Suchman, E. A., DeVinney, L. C., Star, S. A., & Wil-liams, R. M. (1949). The American soldier: Adjustment during Armylife (Vol. 1). Princeton: Princeton University Press.

Thibaut, J., & Kelley, H. H. (1959). The social psychology of groups.New York: Wiley.

Turillo, C. J., Folger, R., Lavelle, J. J., Umphress, E., & Gee, J. (2002). Isvirtue its own reward? Self-sacriWcial decisions for the sake of fair-ness. Organizational Behavior and Human Decision Processes, 89,839–865.

Van den Bos, K. (1999). What are we talking about when we talk aboutno-voice procedures? On the psychology of the fair outcome eVect.Journal of Experimental Social Psychology, 35, 560–577.

Van den Bos, K., & Lind, E. A. (2002). Uncertainty management bymeans of fairness judgments. In M. P. Zanna (Ed.), Advances inexperimental social psychology (Vol. 34, pp. 1–60). San Diego, CA:Academic Press.

Van den Bos, K., Lind, E. A., Vermunt, R., & Wilke, H. A. M. (1997).How do I judge my outcome when I do not know the outcome ofothers? The psychology of the fair process eVect. Journal of Person-ality and Social Psychology, 72, 1034–1046.

Van den Bos, K., Wilke, H. A. M., Lind, E. A., & Vermunt, R. (1998).Evaluating outcomes by means of the fair process eVect: Evidencefor diVerent processes in fairness and satisfaction judgments. Jour-nal of Personality and Social Psychology, 74, 1493–1503.

Wegner, D. M., & Erber, R. (1992). The hyperaccessibility of sup-pressed thoughts. Journal of Personality and Social Psychology, 63,903–912.

Winkielman, P., & Cacioppo, T. (2001). Mind at ease puts a smile onthe face: Psychophysiological evidence that processing facilitationelicits positive aVect. Journal of Personality and Social Psychology,81, 989–1000.

Zajonc, R. B. (1980). Feeling and thinking: Preferences need no infer-ences. American Psychologist, 35, 151–175.