Turillo Et Al. - 2002 - Is Virtue Its Own Reward Self-Sacrificial Decisions For The Sake of Fairness

Organizational Behavior and Human
Decision Processes 89 (2002) 839865
ORGANIZATIONAL
BEHAVIOR
AND HUMAN
DECISION PROCESSES
www.academicpress.com
Is virtue its own reward? Self-sacricial

decisions for the sake of fairness
Carmelo Joseph Turillo, Robert Folger,* James J. Lavelle,
Elizabeth E. Umphress, and Julie O. Gee
A. B. Freeman School of Business, Tulane University, New Orleans, LA 70118-5669, USA
Abstract
We investigate the ways in which concern for fairness inuences decision-making. We use a
paradigm previously shown to illustrate circumstances under which a decision maker sacrices
some of his or her own potential for nancial gain to punish or reward someone who has
demonstrated a prior intent to be either unfair or fair to another person. By ruling out alternative hypotheses related to the original nding, we obtain evidence that virtue is its own
reward: Decision makers make self-sacricing allocations, despite the absence of short- or
long-term benets for doing so. Extending the generality of this eect, we also identify circumstances under which the desire for virtuous fairness produces decisions that are not selfsacricial and do reward someone whose motives seemingly include a willingness to exploit
others. These special circumstances apparently indicate the decision makers belief that two
wrongs dont make a right. Thus, these studies show that the fairness motive and moral
concerns can inuence decisions that have economic impact. We extend the range of eects in
other studies to include condemnation of interactional injustice and we discuss implications of
the overall set of studies in terms of three new foci for attention: A focus on the perpetrator, a
focus on the victim, and a focus on the oensiveness of the act itself. 2002 Elsevier Science
(USA). All rights reserved.
1. Introduction
Moral judgment and the condemnation of others, including ctional others and others who
have not harmed the self, is a universal and essential feature of human social life (Rozin,
Lowery, Imada, & Haidt, 1999, p. 574).
People undertake to punish others when they have no concrete or immediate interest in
doing sowhen they have nothing directly to gain by punishing, and there may be some risk
or cost in doing so (Fiske, 1991a, p. 192).
Are humans inevitably selsh? Can all human behaviors be reduced to self-interest? As the opening quotations suggest, we think not. They indicate that people
sometimes seek to punish the moral transgressions of otherseven those with whom
they have no relationshipnot only without any instrumental self-benet but also
(at times) despite burdens imposed. Violations of fairness norms t that description,
*
Corresponding author.
E-mail address: robert.folger@tulane.edu (R. Folger).
0749-5978/02/$ - see front matter 2002 Elsevier Science (USA). All rights reserved.
PII: S 0 7 4 9 - 5 9 7 8 ( 0 2 ) 0 0 0 3 2 - 8
840
C.J. Turillo et al. / Organizational Behavior and Human Decision Processes 89 (2002) 839865
but, as we will show, existing organizational justice literature derives fairness-oriented reactions by assuming universal self-interest. We propose that such reactions
reect the presence of endogenous deontic emotions (from the Greek root deon for
obligation or duty)emotions elicited by perceiving the transgression itself, just as
aesthetic emotions (e.g., the enjoyment elicited by viewing a dramatic sunset) are
endogenous and intrinsic to the experience itself. The intrinsic, endogenous nature of
such reactions is expressed in a common colloquial expression regarding moralistic
reactions: Virtue is its own reward.1
To preclude self-interest, our research has respondents function as observers of a
moral transgression between two parties with whom they have no relationship (i.e., as
disinterested third parties in the sense of the opening quote from Fiske, 1991a). Our
respondents never know the identity of either of the other two parties, the victim or
the perpetrator. Respondents who express antipathy toward the perpetrator of a
transgression or who punish that perpetrator at the cost of their own self-interests
therefore act consistently with the opening quotes from theorists who hypothesized
the existence of such behavior. In our nal discussion, we speculate about possible
links between our results and emergent theories on deonance (Folger, 2001; see also
Cropanzano, Byrne, Bobocel, & Rupp, 2001), which aim to explain moral emotions
and reactions to others actions as morally creditable or blameworthy (see also the
attribution-of-blame based Fairness Theory of accountability, Folger & Cropanzano,
1998, 2001). Our next section contextualizes the deontic approach by distinguishing it
from two others.
2. A triad of justice approaches

Thinking about justice in moral terms beyond self-interest contrasts with two
traditional approaches. We refer to the three as material, relational-identity, and
deontic. Even though the moral or deontic approach (e.g., Folger, 1994, 1998, 2001)
is much more recent than the other two, some commentaries already distinguish
among them (e.g., Colquitt & Greenberg, 2001; Cropanzano, Rupp, Mohler, &
Schminke, 2001). We start with the former commentary.
As an explanation of procedural justice (Thibaut & Walker, 1975) and as characterized by Lind and Tyler (1988), the rst approach contends that people attend
to fairness insofar as it allows them to gauge the degree to which their long-term
economic interests are enhanced or protected (Colquitt & Greenberg, 2001, p. 221).
The second, or group value or relational model (Lind & Tyler, 1988; Tyler & Lind,
1992), contends that people care about fairness insofar as it allows them to gauge
the degree to which they are valued by the collectives to which they belong
1
The phrase virtue is its own reward merely emphasizes the endogenous, intrinsic (internally
motivated) nature of a for-its-own-sake response, consistent with how Fiske describes moral motives to
respond in accord with norms of sociality: Each of the relational models of sociality, which give rise to
norms, comprises an autonomous motivational goal [which] People seek. . .for its own sake (Fiske,
1991b, p. 183). Similarly, Batson makes the following distinctions: An instrumental goal is sought as a
means to reach some other goal; an ultimate goal is sought as an end in itself; an unintended consequence
is a result of acting to reach a goal but is not itself sought as a goal (1994, p. 604). We treat the deontic
goal of punishing moral transgressions as an ultimate goal, which is all that virtue is its own reward is
meant to imply. In Batsons terms, retribution is an end in itself rather than the instrumental means to
achieving some other ultimate goal (e.g., self-interest)although, as Batson notes, benets serving selfinterest might result as unintended consequences from pursuing a deontic ultimate goal. Batson refers to
principlism as a driving force of human nature independent of egoism, relational concerns, or
empathya motivation with the ultimate goal of upholding some moral principle, such as justice
(1994, p. 608).
841
(Colquitt & Greenberg, 2001, p. 221).2 The third, or deontic, model (Folger, 1998,
2001) contends people value justice simply because it is moral (Colquitt &
Greenberg, 2001, p. 221).
Cropanzano, Rupp et al. (2001) refer to these as alternate roads to organizational justicethree reasons to care about fairness. Two self-interest roadscalled
instrumental and interpersonalcontend that people care for only what justice does
for them. The third, called a moral principle road, is the only one contending that
human motives can rise above self-interest.3 Following Cropanzano, Rupp et al.
(2001), we associate the rst two justice roads or motivations with types of selfinterest benet sought (instrumental or interpersonal). We deviate from their account only in calling the rst material rather than instrumental (because both can be
instrumental) and the second relational-identity rather than interpersonal (because
both can be interpersonal).
In contrast with either of the rst two approaches, a moral-principles or deontic
approach emphasizes commitment to ethical standards (Cropanzano, Byrne et al.,
2001; Folger, 2001). Such a commitment can be called deontic because it creates an
obligation to disregard the self-interest of material and relational-identity benets
when pursuing whether such benets would conict with honoring that commitment.
Put another way, commitment to an ethical standard can entail self-sacrice, which
is the opposite of self-interest. Finally, note that the very notion of fairness itself
involvesby denitionan ethical standard conceived independently of self-interest
so that it can act as a constraint against an otherwise unfettered self-interest capable
of threatening fairness. To show that the concept of fairness has any genuine ethical
meaning as a constraint on self-interest therefore entails showing negative reactions
against unfairness when such reactions cannot garner any self-interested benets of
either the material or self-relational variety.
2
Technically speaking, the instrumental and relational models refer to explanations of the eects of
procedural justice (Lind & Tyler, 1988). It is also a mistake, however, to collapse all previous justice
orientations into these two categories. Even when it comes to explaining procedural justice eects, the
popularity of referring only to these two categories mistakenly overlooks other possible explanations,
including referent cognitions theory (RCT), which actually addressed procedural justice eects in a
dierent manner, prior to development of the relational approach (Folger, 1986, 1987). Moreover, the
RCT explanation accounted for the eects of both procedural and distributive justice (describing the basis
for their interaction as illustrated in a series of initial studies, e.g., Cropanzano & Folger, 1989; Folger &
Martin (1986); for a later review that demonstrates the overwhelming support of cumulative ndings, see
Brockner & Wiesenfeld, 1996). Notably, RCT indicates two properties of procedures neglected by the
instrumental and relational approaches: (a) information contributing toward identication of the source of
mistreatment and, hence, assignment of blame for harm; (b) a type of harm itself, namely the violation
of a moral code (cf. Folger & Cropanzano, 1998, 2001). The latter attribute of unfair procedures need not
involve damage to self-esteem, hence it is distinguishable from the relational model and actually is a
precursor to the outlook advanced in the present undertaking.
3
Concepts and terminology seemingly related to, or at least consistent with, a deontic approach have
appeared in many prior discussions (e.g., Fiske, 1991a,b; Fiske & Tetlock, 1997; Kim & Mauborgne, 1998;
Skitka, 2000; Trevino & Bies, 1997) and even that parenthetical listing undoubtedly omits many important
contributors. To briey elaborate, Fiskes work emphasizes the importance of jural or normative models
of social relations. In particular, in discussing transgressions, Fiske argues that third parties take account
of such violations whether or not they have a relationship with the victim and whether or not the victim is
even aware that such a violation has occurred. In addition, the experience of should has received special
attention in other works as well (e.g., Folger & Cropanzano, 1998, 2001; Folger & Skarlicki, 1998).
Deontic standards can also be conceptualized as counterfactual referentsthat is, an ideal or ought to be
this way state which, when imagined, stands in contrast to some actual state of aairs, such as an
apparently intended act of unfair mistreatment (for related work on counterfactuals as comparison
standards, see Folger & Kass, 2000).
Finally, approaching the study of organizational justice from a deontic perspective is also consistent
with recent work by Trevino and Bies (1997) who call for organizational scientists to move beyond instrumental explanations of actions and behavior. Specically, they state, [F]ailure to include normative
considerations results in underspecied and biased models of organizational behavior (p. 440).
842
Thus, step one of the agenda for our research program was to remove possible
material and relational-identity benets that might otherwise account for self-sacricial reactions against injusticea tactic of stripping away any possible self-interest from our respondents actions (e.g., punishing unfairness). This reductio logic,
standard for proofs in math and logic, operates like that employed by Sherlock
Holmes: When the possible and the probable have been ruled out, then the impossible, however unlikely, must be true (for an illustration of this logic applied to
demonstrate human altruism mediated by empathy, see Batson, 1995).4
We used this logic to study third parties who, observing a perpetrator wrong a
victim, had no vested interest. The absence of an observers relationship with the
other parties helped establish conditions under which we could assess punishment by
third parties who had no concrete or immediate interest in doing so (Fiske, 1991a,
p. 192). In this situation, the parties to the oense had no prior relationship with the
observer, who did not know the other two parties identities. Not knowing whether
they would ever interact with these other people, who also did not know that their
fates had been in the hands of our third-party observers, the observers knew that the
way they treated those other parties could have no predictable, instrumental eect on
the way either the victim or the perpetrator might subsequently treat them. In other
words, observers could not use reward or punishment instrumentally in trying to
shape the future behavior of either the victim or the perpetrator toward them, so no
self-interested benets could result from punishment. A third-party paradigm designed by Kahneman, Knetsch, and Thaler (KKT; 1986) can be administered along
these lines, making it ideal for the study of deontic phenomena not reducible to
material or relational self-interest explanations. We develop variations to ensure
eects devoid of either material or relational benets, thereby increasing our condence that the eects are deontic by ruling out the other categories of explanations.
In these ways, we explore a simple alternative to self-interest as grounds for our
predictions: People hold some degree of allegiance to shared moral standards that
they expect others to hold as well; when transgressions dishonor such standards,
people feel an antipathy toward the transgressor and a concomitant punitive urge.
3. A research program on motives transcending self-interest: KKT and beyond

Talk is cheapit costs nothing to claim to endorse fairness versus having to pay
to endorse that norm. Sacricing money to punish unfairness is not motivated by
material self-interest. This behavior was the object of the KKT research: They
studied whether people would willingly sacrice $1 for the sake of punishing an act
of intended unfairness (and rewarding a prior act of intended fairness). In a rst part
of some KKT versions, students chose how to divide $20 with an anonymous other.
Their only options were either to designate (a) $18 for themselves and $2 for the
classmate, or (b) to divide the $20 equally, each person receiving $10. (Participants
knew that because of limited funds, only a random 10% of student-pairs would
actually be paid as designated by the chosen option.) In another version, this rst
part was eliminated; participants were led to believe, however, that the rst part had
taken place. As with our design, these students participated only in the crucial second
part, which enabled them to punish intended-but-unimplemented injustice. During
the second part, participants believed that they had been matched with two students
not selected for payment in the rst part of the experiment: One student (called U,
for uneven, in KKT) had opted for $18; the other (E for even), $10. Participants
chose between taking $6 and giving $6 to U, or taking $5 and giving $5 to E (cf. our
We thank the editors for reminding us about Holmes and Batson.
843
relabeling in Table 1). Thus, the options were either (a) to pay a dollar to split money
with a stranger who endorsed fairness (equality) or (b) to split the money without
cost with a stranger who had made the self-interested choice in the rst part of the
experiment. The modal response (across variations, consistently averaging around
75%) was self-sacricial punishment of U.
4. Study 1
The KKT design (Table 1) begins to eliminate material or relational motives
thereby yielding an endogenous end-in-itself response devoid of self-interest. We rst
focus on a confound in the KKT design to distinguish between (a) the possibility of
an altruistic, generous, or positive-other-regarding response (rewarding fairness) and
(b) the unique motivation to apply negative sanctions in response to moral transgressions (punishing unfairness). Even though other scholars have referred to the
KKT paradigm as the punishment game (Eckel & Grossman, 1996, p. 143), the
KKT design is actually confounded with respect to inferences about the desire for
punishing unfairness, because the only opportunity to do so (sacricially splitting the
smaller amount) coincided with a simultaneous opportunity to reward fairness.
4.1. Prolog to reward vs. punish: KKT improvements and a pooled replication
We also addressed the KKT methodology. KKT scripts were unavailable, so we
devised variants of the basic paradigm (Table 1 payos), hoping that dierences
would be irrelevant and would allow a pooled, single-cell replication condition (a
robust estimate of replicability) to compare with Study 1s factorial conditions addressing the reward punishment confound. Thus, we began with a conceptual-replication baseline and then added variations in anonymity and access to information.
The baseline and variations used Table 1 payos (in contrast to Study 1s factorial
design) plus KKT improvements. First, we changed Even (E) to $10/$10 and
Uneven (U) to $18/2 in reference to Part 1 students, thereby removing salient
fairness cues.
Second, participants did not put their names on questionnaires. We stapled a
double-stubbed ticket to each questionnaire for payment by stub number. One stub
remained on the questionnaire and the participant kept the other one. Because
participants names appeared only on consent forms, gathered separately, the participants knew that they could collect money without identifying themselves (i.e.,
from pay envelopes numbered to match ticket stubs). We even stressed to students
that as a special anonymity precaution, they could let a friend present the stub.
Because no one could link identities with responses, we eliminated the self-interest
from impression-management and broadcasting ones good deeds to others (for the
sake of basking in praise or gaining a reputation for virtue). For virtue to be its own
reward, no one must know of it.
We also varied descriptions about information availability, as follows: (a) Only
$18$2 students from the rst part would allegedly learn the results, regardless of
Table 1
Payout matrix with two partners and chance to punish, reward simultaneously
Option
1
2
Persons receiving outcomes

You
Student A (who made 18/2 choice)
Student B (who made 10/10 choice)
6
5
6
0
0
5
844
the payos the respondent chose (i.e., $10$10 students would not, unless they received the $5 amount rather than 0); (b) only $10$10 would allegedly learn the
results, regardless of the choice ($18$2 students would not, unless they received the
$6 amount rather than 0); (c) both the $18$2 and the $10$10 would receive such
information; (d) neither partner would know anything about the participants choice
unless scheduled to receive payment (i.e., the choice of the $0 punishment option
would not be revealed); or (e) the Experimenter would require a signature at the time
of payment, eliminating the participants anonymity. An ANOVA showed no differences among ae, for an average (73%) nearly the same as KKT; F 4; 93 1:47,
p :22. We thus pooled them as an overall replication cell (n 98, randomly assigned in conjunction with factorial cells).
This replication condition ruled out sources of self-interest that failed to inuence
behavior: (a) Public image. Remaining anonymous, our respondents could not get
any credit from anyone else by doing good for goodness sake. (b) Public message.
Because no one other than the researchers would see the raw data, respondents were
not able to use their punishment as a public message that crime does not pay,
thereby negating eects of general deterrence (sending a message to the general
public) that might otherwise have been possible. (c) Perpetrator deterrence. To have
any hope that the punishment might deter the perpetrator from committing a similar
transgression in the future, the respondent would have to believe that the perpetrator
was aware of being punished (being allocated $0 rather than $6). But, as indicated
above, this variation had no eect, indicating that self-interested deterrence was not
inuential.
The 73% self-sacricial punishment result also speaks against self-interest based
on the KKT paradigm itself. Respondents thought the Part 1 $18$2 student had
received nothing and had not actually deprived a partner. Sacricially punishing an
$18$2 therefore did not redress either ones own or anothers inequity. Rather, the
punishment deprived the $18$2 person because of his/her presumed intent to take
$18 of $20 if randomly chosen for payment (although because these are presumed
intentions, we manipulate intent in Study 4). These third-party observers, who knew
of $18$2s unfair intentions but not his or her identity, were also not in an ongoing
relationship; anonymity therefore precluded any long-run benet from punishment.
For such reasons, the 73% self-sacrice choicevirtually identical with the KKT
datasuggests actually doing good for goodnesss sake, beyond fairness-endorsement as a cost-free act (consistent with fairness as an end in itself rather than for
instrumental purposes). Other commentators have characterized such a response as a
tendency for people to act on principle. . . to conform to or enforce social norms
(Eckel & Grossman, 1996, pp. 143 & 144). We thus used the replication-cell condition
as an adjunct to our factorial reward-versus-punish design.
4.2. Reward vs. punish, Study 1 design: A two-person paradigm as a rst
unconfounding attempt
Study 1 (see design in Table 2) is our rst attempt to examine punish-unfairness
vs. reward-fairness motives. To try to untangle their confounding in KKT, we
changed the KKT three-party allocation into a two-party allocation. For half of
these two-party conditions, we paired participants with an $18/$2 (Uneven) person
allegedly remaining from the rst part of the study. For the other half of the twoparty allocations, we paired participants with a $10/$10 (Even) person. The design
otherwise used the same dollar amounts ($6, $5, and $0) as in the original KKT, with
the exception of displaying only tworather than threesuch amounts (e.g., 6-6
rather than 6-6-0 and 5-0 rather than 5-0-5see Table 2, Matrix #3, the Uneven/182 Punish entries). Thus, two-person pairing of the respondent with the $18/$2
student from the rst part indicated a choice between (a) $6 for that person and $6
845
Table 2
2 (Partner: Even vs. Uneven) 2 (Payout: Punish vs. Reward) payout matrices, Study 1
Option

You
Student who had

made 18/2 choice
Student who had

made 10/10 choice
1
2
Matrix #1: Even (10-10), Punish

6
5
6
0
1
2
Matrix #2: Even (10-10), Reward

6
5
0
5
1
2
Matrix #3: Uneven (18-2), Punish

6
5
6
0
1
2
Matrix #4: Uneven (18-2), Reward

6
5
0
5
Note. Where entries are blank (e.g., for the 10/10 Student in the Uneven-Unrelated Condition), participants saw neither that blank column nor the heading that appears above it in this table.
for the respondent or (b) $0 (nothing) for that person and $5 for the respondentthereby testing for the strength of a desire to punish unfairness when the
(previously confounded) simultaneous chance to reward fairness did not exist.
Our dependent variable is the percentage of sacricial behavior that violates
material self-interestthat is, $5 choices by participants who thereby declined to
obtain $6. We separately paired the participants with either an Uneven (18/2) or an
Even (10/10) partner. In a factorially crossed fashion, we also linked the $5-for-self,
sacricial option with either of the following consequences for the partner: (a) a $5
reward rather than nothing$0for the partner, or (b) a $0 punishment rather than
$6 for the partner. This 2 (Even vs. Uneven) 2 (Reward vs. Punish) design thus
yielded the payout matrices of Table 2. But why sacrice to reward unfairness? If a
participant makes the sacrice of taking $5 rather than $6, it obviously makes more
sense if that sacrice rewarded a fair (Even, 10/10) person (Matrix #2) rather than if
it rewards an unfair (Uneven, 18/2) person (Matrix #4). This logic yields the following two-part hypothesis.
Hypothesis 1a. In Reward conditions, respondents will more often sacrice $1 to
reward an Even partner (Matrix #2) than an Uneven partner (Matrix #4).
Similar logic with respect to punishment yields the following parallel hypothesis.
Hypothesis 1b. In Punish conditions, respondents will more often sacrice $1 to
punish an Uneven partner sacricially (Matrix #3) than an Even partner sacricially
(Matrix #1).
Note that our two-party conditions also allow for a test of the relative strength of
tendencies to reward fairness and to punish unfairness. We did not believe that a
prediction could be derived from existing theory with respect to these two conditions, however, so we do not propose a hypothesis regarding their relative strengths.
Instead, we adopt a neutral stance, expecting the original results obtained by
Kahneman et al. to be mimicked both by sacricial punishment of the Uneven
partner (Uneven/Punishment condition) and by sacricial reward of the Even
partner (Even/Reward condition). That stance yields the following respective predictions.
846
Hypothesis 2a. Compared with the 73% pooled-replication cell involving the same
three parties as in the Kahneman et al. (1986) study (viz., self, Even, and Uneven), the Uneven-Punishment condition should not dier in sacricial unfairness-punishment.
Hypothesis 2b. Compared with the 73% pooled-replication cell involving the same
three parties as in the Kahneman et al. (1986) study (viz., self, Even, and Uneven),
the Even-Reward condition should not dier in the strength of fairness-rewarding
tendencies.
4.3. Method
4.3.1. Research participants
Additional MBA students from a mid-sized university in the southern United
States (n 167) took part in the factorial conditions for partial fulllment of course
requirements and the chance to win money. Participants did not provide demographics (consistent with anonymity), but aggregate statistics indicate that the
population of students from which this sample was drawn is 76% male, 10% minority, and, on average, 26 years old. Two of the participants failed to understand
the directions correctly, leaving a nal sample size of 165.
4.3.2. Experimental conditions and dependent variable
We randomly assigned participants to cells in a 2 (Partner: Even vs. Uneven) 2
(Payout: Punish vs. Reward) design. The dependent variable was the percentage
(means between 0 and 1) choosing $5 instead of $6 for themselves.
4.3.3. Procedure
Participants completed an anonymous questionnaire that also included instructions and manipulations. Questionnaires had a double-stubbed ticket stapled
to them and instructed participants to keep one stub for their chance at payment
later in the semester. Because we did not have a large enough budget to pay
everybody, we decided to pay 10% of the tickets, chosen at random, according to
the participants choice among allocation-decision options (described below).
Paying 10% of the participants has been shown to yield results that do not dier
signicantly from paying all participants based on their choices (Kahneman et al.,
1986).
Respondents read that an earlier study ostensibly had already taken place in
which anonymous students had been paired with each other as follows. One of each
pair, X, was provisionally budgeted $20. X was informed that he or she had to split
the $20 with the other member of the pair, Y, in one of the two ways: Either so that
both X and Y received $10, or so that X received $18 and Y received $2. Y had no
choice but to accept Xs oer. Respondents learned that in the earlier study ten
percent of the (ctional) pairs had been chosen randomly to be paid according to
their decisions. The instructions then gave specics about the study in which our
participants were actually involved, which diered according to condition. Participants learned that they had been anonymously paired with a player X who had not
received money in the earlier study, but whose earlier preference had been either an
$18$2 or a $10$10 choice.
Participants then read about the two allocation-decision options. One of the
options always gave the participant $5 and the other $6; hence, our dependent
variable (% choosing $5) was always a self-sacricial behavior opposed to a material
self-interest behavior of taking the higher monetary outcome ($6). In the punish
conditions, the payout matrix was such that the participant sacriced $1 ($5 instead
of $6) to allocate $0 instead of $6 to a stranger. In the reward conditions, the par-
847
ticipants sacricial choice of $5 for himself or herself allowed $5 instead of $0 going

to the stranger. These two conditions represent a deconstruction of the original
Kahneman et al. payout matrix, which we used in our replication cell (Table 1).
Table 2 shows the 2 (Partner: Even vs. Uneven) 2 (Payout: Punish vs. Reward)
portion of our Study 1 design.
Our instructions copied the baseline-replication in stating that unless the participant chose a non-0 allocation for his or her partner and their pair was one of the
10% chosen randomly to be paid based on the participants choice, nobody would
learn of the participants choice (e.g., the rst part student would not even know
that this second study existed). Recall that in all the variations of our replication
conditions, we used the original three-person design of KKT in which participants
had both an Even partner and an Uneven partner. Although the partners knowing
(vs. not knowing) about the punishment had no eect within those replication
variations, we wanted to be sure that this would extend to our two-person version of
the original three-person paradigm. For that reason, we duplicated our entire 2 2
design (in this case called Partner Knowledge: Low) with parallel conditions in which
the partner would know if punished (Partner Knowledge: High). Analysis showed
that the results from the extra conditions (High Partner Knowledge), which we
added only to verify the lack of information-based eects as revealed by prior results
within our replication-condition variations, did not dier from those in the main
conditions of this two-partner design in which such knowledge was absent (Low
Partner Knowledge), F < 1. Given this reassurance that a deterrence-based explanation was equally invalid regardless of using a three- or two-partner design (also
indicated by lack of two- or three-way interactions), we pooled both sets of data in
our remaining analyses and the results section reports ndings from this collapsing
of a 2 2 2 design into a 2 2 design (see Table 3, however, for the full report of
means from all eight cells).
After initial instructions, participants were given a manipulation check to ensure
that they had understood the information. Participants were then presented with the
payout matrix in table form (in correspondence with Table 1 or Table 2, depending
on condition) and asked to circle their choice. Participants also had to indicate the
dollar amounts that they and their partner(s) would receive according to the tableagain to ensure that they had understood their options correctly. The experiment concluded once participants had made their decisions. Participants were
thanked for their time, but not yet debriefed for fear of information spreading to
other prospective participants. After all the data were collected, we conducted a
lottery and paid 10% of the participants. All winners were paid $6, regardless of their
choice. We distributed a list of all the winning lottery numbers along with a written
debrieng to all of the participants.
Table 3
Means and standard deviations of choice to sacrice $1 in Study 1
Partner knowledge
Payout matrix
Reward
Punish
SD
SD
High partner
Even
Uneven
.95
.75
.22
.44
21
20
.00
.20
.00
.41
22
20
Low partner
Even
Uneven
.90
.71
.31
.46
20
21
.10
.10
.30
.31
21
20
Note. Cell means represent the respective percentages of participants who chose $5 instead of $6.
848
4.4. Results
From a Partner X Payout ANOVA on the percentage of participants $5 choices,
the Payout factor produced a signicant main eect, F 1; 157 196:09, p < :01.
Participants more often chose $5 to reward (M :83, possible range of 01) than to
punish (M :10)see Table 4 and note that .83 is the average of 93% and 73%,
whereas .10 is the average of 15% and 5% (all $6 percentages, of course, would be
100 ) those $5 percentages). We also obtained a signicant Partner X Payout interaction, F 1; 157 8:08, p < :01. Fig. 1 shows that participants chose $5 to a
dierent degree in each payout condition, depending on the partner condition.
Table 4
2 (Partner: Even vs. Uneven) 2 (Payout: Punish vs. Reward) payout matrices and percentages of each
option chosen, Study 1
Option

You
18/2 Student
10/10 Student
1
2
Matrix #1: Even (10-10), Punish

6
5
6
0
95%
5%*
1
2
Matrix #2: Even (10-10), Reward

6
5
0
5
7%
93%*
1
2
Matrix #3: Uneven (18-2), Punish

6
5
6
0
85%
15%*
1
2
Matrix #4: Uneven (18-2), Reward

6
5
0
5
27%
73%*
Note. Asterisks indicate the percentage who chose the self-sacricial ($5) option, which constituted the
score from that condition for the sake of analysis, whereas the corresponding non-asterisked entry indicates the percentage of remaining participants ($6 choosers).
Fig. 1. Mean choice to sacrice $1 as a function of past evenness of partner and payout matrix (Study 1).
849
We then analyzed the simple main eects from that interaction (cf. Pedhazur &
Schmelkin, 1991). Reward-condition participants signicantly more often sacriced $1
to reward an Even partner (M :93) than to reward an Uneven partner (M :73),
conrming Hypothesis 1a, F 1; 161 7:05, p < :01. The test of Hypothesis 1b showed
that Punish-condition participants more often sacriced $1 in punishing an Uneven
partner condition (M :15) than in punishing an Even partner (M :05)results in
the predicted direction short of conventional signicance, F 1; 161 2:0, p :16.
Punishing an Uneven partner fell to such a huge degree (15%, vs. KKT replication 73%!) that it was hard for the Even-punishment cell to be signicantly lower
(also reecting power dierences of .23 vs. .09 for those two tests, respectively).
In fact, this huge drop in punishing unfairness completely violates Hypothesis 2a.
We had assumed that in comparison with the replication design of Table 1, which
involved the same three parties as the original KKT study (viz., self, Even, and
Uneven), the Uneven-Punishment condition would not dier in the strength of
sacricial-punishing tendencies. Table 4, Matrix #3 shows that the Uneven-Punish
condition instead produced results (15% choosing the option of $5 for themselves
and $0 for the 18/2, Uneven partner) that diered signicantly from those of the
replication cell (73% choosing that option, which also gave $5 to themselves and $0
to the 18/2, Uneven partnerbut which simultaneously provided $5 to the 10/10,
Even partner).
The results also disconrmed Hypothesis 2b. We assumed that in comparison
with the Table 1 replication design (same three parties as KKT), the Even-Reward
condition would show the same fairness-rewarding tendencies. Our replication design produced a 73% majority of participants who chose to sacrice $1 to reward
fairness (which, of course, was confounded with the motive to punish unfairness).
That diered signicantly from the unconfounded tendency to reward fairness sacricially in our Even-Reward condition, M :93, F 1; 137 6:68, p < :02.
4.5. Discussion
The results suggest that individuals sacrice $1 more readily to punish partners
for past intended greed than for past intended generosity, although this nding did
not reach signicance. The results support the hypothesis that when sacricing $1,
individuals are more likely to reward partners for past intentions of generosity than
for past intentions of greed (which, on fairness grounds, would not merit reward).
Moreover, anonymity ruled out impression-management ends.
In addition, the results across all variations in our replication conditions mirror
the original ndings obtained by Kahneman et al. (1986): Decision-makers in the
three-party design of Table 1 are more likely to split a relatively smaller sum with a
stranger who has shown generous intent in the past than a larger amount with a
stranger who has shown greedy intent in the past. Those data seem to rule out such
motivations as a desire to please the experimenter or a desire to put a miscreant on
notice. The punitive decision-maker was not making the world a better place,
increasing the chances that the strangers future interactions with people would be
fairer. In other words, there were no clear long-term benets of such a decision. One
of the only remaining explanations seems to be that virtue, for its own sake, was an
end in itself for the decision-makers. People prefer fair behavior in others and make
nancial sacrices accordingly.
Our results show an unexpected Payout main eect between Reward conditions,
in which taking $5 rather than $6 gives $5 to the partner (whether an Even or
Uneven person), and Punish conditions, in which taking $5 imposes a $0 deprivation
on the partner (whether Even or Uneven). The dierence between the proportion of
$5 self-sacrice decisions in Reward conditions (.83) and Punish conditions (.10)
reveals an inclination so overwhelming that it tends to swamp reactions toward
850
Uneven versus Even partners. We did not count on so strong an impulse because we
failed to consider that our two-party Reward conditions confounded the $5 sacrice
with an egalitarian 50/50 allocation ($5 to self and $5 to partner), whereas our twoparty Punishment conditions confounded the otherwise more self-aggrandizing
choice of $6 with the same type of egalitarianism (except that it involved a $6/$6 split
in that case).
This strong egalitarian urge shows that participants were reluctant to take all the
money (either $5/0 or $6/0) if sharing money equally with someone else was an option.
Participants in our Partner Knowledge, two-person conditions might have viewed
their $5/0 or $6/0 (100%/0%) option as egregious self-aggrandizement or winner-takeall greed. Avoiding these options would therefore conict with punishing someone
elses greed, for example, in the same sense that two wrongs dont make a right.
Thus, our design in the Partner Knowledge or two-person conditions inadvertently
pitted one moral motive (sharing equally rather than greedily consuming all available
resources) against another (punishing someones prior attempt to be unfair).
In addition to some success at untangling reward and punishment, we also saw
that more than one moral motive might operate in KKT-like contexts (cf. choices
between the lesser of two evils). Our Table 2 design changed the three-party allocation of the original paradigm into a two-party version that disentangled reward
and punishment, but that two-party version also introduced a winner-take-all option
lacking in the original KKT. We reasoned that if it were possible to disentangle
reward from punishment while still using a three-party allocation that eliminated any
winner-take-all options, the revised paradigm would simultaneously (a) provide
evidence for a punish-unfairness motive unconfounded with a reward-fairness motive and (b) reveal the extent to which the Payout main eect in Study 1 resulted from
a winner-take-all option.
Study 2 thus contained three conditions, each with an 18-2 potential recipient: (a)
Uneven-AloneThe 18-2 person was the sole recipient other than the respondent,
duplicating Matrix #3 of Table 2 (Study 1, Uneven-Punish). (b) Uneven/EvenBoth
other recipients were the same as in our baseline replication of KKT (Table 1). (c)
Uneven/UnrelatedThis condition was like Table 1, except that the third recipient
now came from an unrelated study on goals.
Based on the Study 1 KKT-replication results, we hypothesized that respondents
would be likely to punish a partner who had shown greedy intent in the past when, by
doing so, they could still split money evenly with another person. We did not expect,
however, that it would matter whether that person was someone who had shown
generous intentions in the past or someone who had been in an unrelated study without
allocations. That is, we expected a boundary condition on punishment: Retributive
justice would take place only if it did not upset the equilibrium of distributive justice (cf.
Batson, Bowers, Leonard, & Smith, 2000). Thus we predicted the following: Whereas
participants in a two-person design tend to avoid punishing an Uneven partner who is
the only other person in the allocation, participants in a three-person design will punish
an Uneven partner more often if they can still split money evenly with someone else,
whether an Even partner or someone from an unrelated study.
5. Study 2
5.1. Method
5.1.1. Participants and design
Forty-seven MBA students from a mid-sized university in the southern United
States participated for partial fulllment of course requirements and a chance to win
money. To ensure as much anonymity as possible, we asked no individuating in-
851
formation; nevertheless, the population of MBA students at the university from

which this sample was drawn is 76% male, 10% minority, and the average age is 26
years. Seven of the participants failed to understand the directions correctly, leaving
a sample size of 40. These were randomly assigned to the Uneven-Alone, Uneven/Even
[KKT], or Uneven/Unrelated conditions.
5.1.2. Procedure
Participants completed a questionnaire containing instructions and manipulations
and, except as noted, the rest of this study followed the Study 1 protocol (e.g., with
regard to 10% payment by ticket stubs and instructions about an ostensible prior
experiment). The major exception involved participants in an Uneven-Unrelated
participant condition, whom we told about another prior (ctional) study that varied
goals and measured performance. A similar 10% payment ostensibly also took place
in that prior goal-setting study and no participants from the prior allocation study
had participated. Participants in the Uneven-Alone and the Uneven-Even conditions
were not told of the goal-setting study. The instructions then diered further as
participants were told that they had been anonymously matched (a) alone with a
player X who had made an $18$2 decision (Uneven-Alone); (b) with an X who had
made an $18$2 decision and a player X who had made a $10$10 decision (Uneven/
Even); or (c) with an X who had made an $18$2 decision and a participant from the
unrelated goal-setting study (Uneven/Unrelated).
For the payout-option description in all conditions, one option meant that the
participant would receive $5 and another meant that the participant would receive
$6. In the Uneven-Alone condition, the payout matrix was such that the participant
would sacrice $1 (choosing $5 instead of $6) to allocate $0 instead of $6 to his or her
partner (see the parallel Uneven-Punish condition in Table 2). In the Uneven-Even
condition, the participant would sacrice a dollar to allocate $0 (instead of $6) to the
uneven participant and $5 (instead of $0) to the even partner (this condition was a
replication of the original study; see Table 1). In the Uneven-Unrelated condition, it
was the same as in the UnevenEven condition, except that the third participant was
unrelated to the original allocation experiment, instead of being an even partner (see
Table 5).
5.2. Results
Seven participants failed to ll in appropriate dollar amounts and were excluded
from further analyses, which began with a one-way ANOVA revealing a signicant
Table 5
Payout matrices by condition, Study 2
Option

You
18/2 Student
1
2
Uneven-unrelated condition
6
6
5
0
1
2
Uneven-even condition
6
6
5
0
1
2
Uneven-alone condition
6
6
5
0
10/10 Student
Goal (unrelated) Student

0
5
0
5
Note. Where entries are blank (e.g., for the 18/2 Student in the Even, Punish Condition), participants
saw neither that blank column nor the heading that appears above it in this table.
852
Fig. 2. Percentage of participants who chose to sacrice $1 to punish a partner who had made an uneven
allocation decision in the past, as a function of who the partner(s) is(are) (Study 2).
eect of partner, F 2; 37 15:49, p < :01. Following the omnibus analysis, we

conducted planned pairwise contrasts of means (see Fig. 2). The choice to sacrice $1
was signicantly lower in the Uneven-Alone condition (M 0, from a possible range
of 01) than in either the Uneven-Even condition (M :62) or the Uneven-Unrelated condition (M :75), p < :01 for both. The $1-sacrice levels in the UnevenEven condition and the Uneven-Unrelated condition did not dier signicantly,
p > :15 (where the power to detect dierences was .06).
5.3. Discussion
The results conrm our hunch that people incline to punish someones unfair
intent if they can still divide money evenly with another personboth when the
observer/allocator knew about the latters prior (generous) intent and when that
other person came from an unrelated study (intent moot). Just being able to split the
sum of money evenly with someone who is not known to have had unfair intentions
in the past is enough to inuence decision-makers to make nancial sacrices to
punish someone who is known to have had unfair intentions in the past. These results support our reasoning that individuals will only engage in retributive justice to
punish unfairness in the past when, by doing so, they are not themselves being unfair
in the present.
6. Study 3
The Study 1 and 2 results show that respondents take into account others past
intentions (greedy vs. generous) in addition to their own economic utility. These
ndings clarify and extend past ndings (Kahneman et al., 1986) in that they not
only address virtue as an end in itself, but also under what circumstances other
peoples past behaviors become important for observers. Study 3 extends this line of
research beyond the domain of past behavior involving violations of distributive
justiceespecially because much of the contemporary study of fairness in organizations has instead focused on procedural justice (e.g., Lind & Tyler, 1988) and on
interactional justice (e.g., Bies, 1987). Various reviews (e.g., Folger & Cropanzano,
1998) suggest that interactional injustice in particularthe absence of interpersonal
853
sensitivity or the presence of callous indierence to anothers feelings, sometimes to

the point of abusive mistreatmenthas a special sting in the workplace. Given the
advantage of using renements to the Kahneman et al. (1986) paradigm as a means
of studying reactions to injustice, we thought that we would be remiss not to include
a parallel investigation on the punishment of interactional fairness violations.
6.1. Method
6.1.1. Procedure
Our new instructions asked undergraduates in an accounting class (n 14) to
choose between options X and Y, with the associated consequences indicated below:
Your option X:
Your option Y:
Results for you
Results for
person A
Results for
person B
$6
$5
$6
$0
$6*
$5
Note rst that we changed the asterisked entry from the Kahneman et al. (1986)
value of $0. Note also that at rst blush, every conceivable line of argumentrational self-interest, maximal benet to both A and B, equal benet to all,
and greatest good for the greatest number (cf. Pareto optimality)points clearly to
Option X as the obvious preference. On the other hand, presumably it might
matter who A is and who B is. Suppose B is an anonymous stranger chosen at
random but the question, What kind of person is A? becomes crucial. Option
Xs advantages seem diminished, for example, if Person A is someone so agrantly
unfair that it makes your blood boil (e.g., a terrorist; someone who delights in
torturing innocent children). Once we identify A as someone willing to engage in
reprehensibly unfair eorts that elicit your moral outrage and righteous indignation, then the money at stake for youmuch less the shape of the Option-X
distributionhas declining impact on your choice tendencies. Quite plausibly, for
example, a large proportion of U.S. citizens would pick Option Y over Option X if
they thought Person A was, say, Saddam Hussein, even if all the numbers were
multiplied by factors of 1000 or more!
As in Study 2, the consequences in the Results for Person B column applied
to a neutral personsomeone whose conduct is unreported, with no information
regarding whether this person has either good or bad proclivities. Specically, our
cover story led our participants to believe that their choice between options X and
Y would have monetary consequences for two supervisors at an unnamed company in Falls Church, Virginia, and nothing was reported about Person B other
than that this person was one of those supervisors chosen at random. Unlike in the
Kahneman et al. study, our participants could not reward B with $5 instead of
nothing: Option X has the higher reward for B of $6 (see the asterisk above),
whereas Option Y gives B only $5. Thus, any self-sacricial participants would
actually be depriving B of a dollar at the same that they would be depriving
themselves of a dollar. Also, unlike the Kahneman et al. participants who gave $5
to an E (fair) person by depriving themselves of $1, any of our students who
give themselves $5 and B $5 do not select B to receive at least some money because
of anything meritorious or E-like (fair) about Person B. After all, this particular
Person B was simply picked at random out of a B pool and no one in pool B had
any special characteristics (fair or unfair).
The description of Person A, however, revealed a supervisor whose unfair conduct
toward employees constituted interactional injustice. In fact, it was identied as such
for us by Robert Bies, the originator of that concept, based on quoted remarks from
an employee whom Bies interviewed about an actual supervisor at the Virginia
854
company. When we asked for interview text indicating extreme outrage about interactional injustice by that employees supervisor, Bies provided the following
quote: He takes great pleasure in belittling me and ridiculing me in public. We told
participants that in making selections for Person A, we chose only supervisors about
whom all employees shared the same opinion. Instructions contained the quoted
remark about ridiculewhich each participant thought pertained only to his or her
designated Person Aand called it a representative, consensual description of
Person A as a supervisor.
6.2. Results and discussion
Quite notably, this design yielded a 50% self-sacrice rate (M :50). Half the
participants chose Option Y ($5 to themselves and $5 to the Person B manager)
thereby allocating $0 to the ridiculing, Person A manager. Although lower than the
73% self-sacricial punishment of our Study 1 replication conditions, this punishment of interactional injustice by half our participants is noteworthy not only because it provides an extension of the Kahneman et al. paradigm to that additional
fairness domain but also because in this instance, the self-sacricial punishment
occurs despite even more reasons to keep $6 to oneself than in the original paradigm.
That is, we deliberately gave our participants more excuses for taking $6 when we
changed the Person B entry for Option X from $0 (the Kahneman et al. entry) to $6
(the entry in our new matrix abovewithout the asterisk, of course, for our participants). As mentioned earlier, we thought that change would make Option X more
attractive than in our earlier studies because it now could nd support from rationales such as maximal benet to all, equal benet to all, and greatest good for the
greatest number. Remarkably, half our participants resisted the temptation of using
such reasons as rationalizations allowing them to keep $6 free of guilt.
Interactional justice is sometimes considered a form of procedural justice, but our
participants punishment of interactional injustice also stands out by virtue of the
mismatch between that punishment reaction and the two most common explanations
for why people care about procedural injusticeconcerns about material self-interest or psychic self-identity. According to the rst type of explanation, procedural
injustice evokes punishment if the instrumental (outcome-determining) character of
the procedure threatens the material well-being of the person experiencing that injustice. But why would half our respondents sacrice $1 to punish a ridiculing
managers procedural misconduct on such grounds? If they were concerned about
their material short-term interest, why not simply keep $6 instead of $5? Protecting
material long-term interest is an even more implausible reason to self-sacricially
punish such misconduct, given the fact that our participants had not suered any
direct consequences themselves and had no way of anticipating any detrimental
consequences from that supervisor (whom they did not know) in the future.
Writings on the group-value/relational model (Lind & Tyler, 1988; Tyler & Lind,
1992) have called self-identity concerns an alternative to material self-interest for
explaining why people care about procedures. Perhaps, for example, people might
want to punish an authority whose procedurally unjust conduct signies exclusion to
subordinates rather than their inclusion as full-edged members of a group. Concern
about procedural injustice would encourage punishment reactions to it, according to
that explanation, if the group seemed desirable enough so that being a member in
good standing would reect favorably on self-identity. The isolation of our (Louisiana) students from an unnamed company in Virginia, however, also makes implausible various types of explanations based on concern about self-identity. They
were not the subordinates ridiculed by the Person A manager. The ridiculing behavior they punished as interactionally or procedurally unfair did not threaten to
exclude them from the membership group of managers, or subordinates, at this
855
unknown company in another state. It would be impossible for them to estimate the
odds of their ever working in that company, much less for them to know whether
full-edged membership in it would reect favorably on their self-identities.
Thus, we have again shown that people can have a morally principled reason for
sacricially punishing injustice, extending beyond distributive injustice to include the
unfair violation of procedural and/or interactional norms. Nonetheless, at least two
types of issues relevant to that additional realm remain unsettled. First, identitybased concerns seem implausible as the most directly applicable psychological
mechanism, but variations on related themes might still apply (e.g., empathy as
identication with victims). Second, ruling out material and relational-identity selfinterest as explanations does not in itself directly examine variables relevant to deontic reactions that can shed further light on a fairness-as-virtue motive. Instead, we
decided that further reasoning and an additional investigation would be required to
address those two types of issues.
Moreover, our fourth study was designed to expand beyond the Kahneman et al.
(1986) paradigm in one more way, but this time abandoning its basic structure rather
than stretching it to t circumstances where its applicability would be forced and
articial. The rst three studies used a unique, dichotomous variable to measure
subjects responses to unfairness, whereas we would expect the virtue of punishing
unfairness to generalize beyond any peculiarities associated with the pay allocation
measure itself. In Study 4 as dependent variables therefore we employ continuous
measures of subjects perceived oensiveness of the perpetrators behavior, willingness to help the perpetrator (reverse coded), and the extent to which the perpetrators
behavior makes the subject angryall of which we treat in an exploratory fashion as
possible ways to tap the more general virtue of condemning unfairness rather than
the specic act of punishing it.
7. Study 4
Study 4 aims to evaluate more thoroughly the role of identication as regards
third-party responses to injustice. On the one hand, Study 3 showed that no direct
form of group identication is needed to obtain self-sacricial punishment. On the
other hand, Study 3 did not manipulate group identication; rather, it was held
constant at a low (virtually absent) level. Because we did not manipulate possible
dierences in identication with the membership category of the victim, however,
our Study 3 did not allow for a separate, independent main eect possible when
variations in identication with groups are measured or manipulatedan eect that
might be suggested by the self-identity motive of the group value and relational
models, or by dierences in empathy with victims of dierent categories. For example, Tyler and Degoey (1995) found that citizens who identied strongly with their
community showed stronger procedural justice eects in reference to water conservation decisions than did those who identied less strongly. Similar eects were
found by Brockner, Tyler, and Cooper-Schneider (1992) in a layo context.
To paraphrase Folger and Cropanzano (2001), however, we do not believe a
concern for justice stops at the door of a close in-group. Our deontic perspective
on justice, informed in part by fairness theory (for a detailed review see Folger &
Cropanzano, 2001), suggests that responses to injustice involve accountability to
moral principles rather than threats to self-identity. Civilizations hold people accountable to moral norms, such as fairness, as standards of conduct. We do not deny
the importance of group identication as an independent eect, such as the eect of
greater care and concern for a relative as the victim rather than a mere acquaintance
(e.g., degree of interest in the victims condition because of degrees of variation in the
ability to identify with or empathize with, or care about, the victim). An eective
856
guide to conduct should be internalized, however, rather than being completely

subject to the whims of benets available only from external sourcessuch as the
benet to self-identity from acceptance by a group. Because fairness is a moral
standard for evaluating conduct, people can be held accountable for the extent to
which their intentions show an unwillingness to act in accord with such standards.
Thus, in Study 4, we cross identication conditions with variations in perpetrator
intent as a second possible antecedent to injustice reactionsone that, in our
opinion, is more central.
Recent work on Fairness Theory (Folger & Cropanzano, 1998, 2001) points to
accountability attributions as crucial to understanding injustice responses. Perpetrator characteristics such as motives inuence accountability and culpability assessments; thus, intent can generate antipathy toward perpetrators based on blame
attributions (e.g., Aquino, Tripp, & Bies, 2001; Bies, Tripp, & Kramer, 1997;
Bradeld & Aquino, 1999; Folger & Cropanzano, 2001; Folger & Skarlicki, 1998).
Attributions of intent involve assessing the extent to which the perpetrator willfully
acted in a manner considered oensive. Research shows that when perpetrators intend to harm, victims respond in a particularly negative manner toward them (see
review by Miller, 2000). For example, victims who attribute intentional harm to the
harmdoer report more feelings of anger (Weiner, Amirkhan, Folkes, & Verette,
1987), blame the perpetrator more (Quigley & Tedeschi, 1996), and are more likely to
retaliate (Dyck & Rule, 1978). This attributional logic was also incorporated by
referent cognitions theory (e.g., Folger, 1987), which is the intellectual ancestor to
Fairness Theory: Attribution-based conceptualizations suggest that people are
more resentful when someone deprives them intentionally (the cause of the other
persons action is internal) than when the cause of the other persons action is attributed externally to situational constraints (Folger, Roseneld, & Robinson,
1983, p. 269citing, e.g., Kulik & Brown, 1979).
As noted, the studies referenced above examined victims responses to intended or
unintended acts of harm. Consistent with Fairness Theory and a deontic perspective
on justice, however, we believe that third party responses, motivated by a desire to
hold others accountable for injustice, will also be inuenced by perpetrator intent.
We thus predict greater condemnation under conditions that indicate intended rather than unintended unfair treatment of others.
7.1. Method
7.1.1. Participants
Seventy-three undergraduates from a mid-sized university in the U.S. participated. Students were recruited from a psychology course and received course credit
for participation. Twenty-ve of the respondents were male and 48 were female. The
mean age was 20.
7.1.2. Design
Participants were randomly assigned to one of the four cells of a 2 (Group
Identication: High or Low) 2 (Intent: High or Low) factorial design. Three dependent variables were measured: Perceived oensiveness of the perpetrators behavior, willingness to help the perpetrator, and participants feelings of anger. Both
independent variables (group identication and intent) were manipulated by asking
participants to read one of the four ctitious situations (vignettes).
7.1.3. Procedure
Participants read about either a group of international gymnasts (low group
identication condition) or a group of students from their university (high group
857
identication condition). Low group-identication participants were asked to

imagine the following situation:
Members of an internationally competitive gymnastics team were asked to evaluate and
comment on their coach. All of these gymnasts made very similar and consistent comments.
The following representative comment captures the tone of those comments: He takes great
pleasure in belittling me and ridiculing me in public.
We asked the high group-identication participants to imagine a dierent situation, as follows:

Students at [their] University were asked to evaluate and comment on one of their professors. All of these students made very similar and consistent comments. The following representative comment captures the tone of those comments: He takes great pleasure in
belittling me and ridiculing me in public.
In all conditions, the intent manipulation followed next, then the questionnaire
items.
Low-intent participants read the following: Others who knew the professor/
coach well were then asked to comment on his treatment of the students/gymnasts.
They all said essentially the same thing: Knowing him, I can tell you that Im sure he
doesnt act that way on purpose, deliberately. Thats just not characteristic of him.
High intent participants read the following: Others who knew the professor/coach
well were then asked to comment on his treatment of the students/gymnasts. They all
said essentially the same thing: Knowing him, I can tell you that Im sure he acts
that way on purpose, deliberately. Its never accidental with him.
7.1.4. Manipulation checks
We adapted Tyler, Degoey, and Smith (1996) group identication measure as a
manipulation check: [My] University (Being a member of a gymnastics team) is
important to the way I think of myself as a person. The following two items served
as manipulation checks on perpetrator intent: The (coach) professor intended to be
oensive; The (coach) professor meant to be insulting. Participants rated the
extent to which they agreed with manipulation check items on a seven-point scale
(1 strongly disagree; 7 strongly agree).
7.1.5. Dependent variables
To measure the oensiveness of the perpetrators behavior, we had respondents
rate the extent to which the following adjectives described the (coachs) professors
actions: Acceptable, shocking, oensive, appalling, excusable, insulting, respectful,
rude, and polite. Participants rated each adjective using a ve-point scale (1 not at
all; 5 extremely). We also used seven-point response scales (disagree/agree format)
with an item about willingness to help and an item about anger. Participants willingness to help the perpetrator (i.e., the coach or professor) was measured with the
following item: If the (coach) professor spilled the contents of his briefcase on the
oor in front of me I would help him pick it up. Participants anger was measured
with the following item: The (coachs) professors actions make me angry.
7.2. Results
Manipulation checks. Both manipulation checks succeeded. The group identication manipulation check item showed that participants in the low group identication condition (gymnastics condition) reported signicantly lower levels of group
identication than those in the high group identication condition (university condition): Low group identication M 2:76, high group identication condition M
4:03; F 1; 71 7:87, p < :01. The two intent manipulation check items showed high
internal consistency (Cronbachs alpha .80) and were averaged together. Partici-
858
pants rated the (coach) professor as having signicantly more intent in the high
intent condition (M 5:41) than the low intent condition (M 3:71),
F 1; 71 36:9, p 6 :001.
7.2.1. Hypothesis testing
Scale-reliability analyses on the adjectives measuring the oensiveness of the
perpetrators behavior showed that the polite item tended not to correlate with the
others, so we deleted it. Reliability of the remaining eight items showed acceptable
internal consistency (Cronbachs alpha .86); we combined them as the perpetrator
oensiveness measure.
We predicted that participants would rate the perpetrators actions as more offensive, report less willingness to help the perpetrator, and convey more anger about
the perpetrators actions if the perpetrator intended harm (high intent) than if s/he
did not intend harm. We conducted a 2 (intent: High or low) 2 (group identication: High or low) Multivariate Analysis of Variance (MANOVA) on the three
dependent variables to test this prediction. We found the predicted main eect of
intent at both the multivariate and univariate levels: Multivariate F 3; 67 6:89,
p < :001; univariate for perceived oensiveness, F 1; 69 15:01, p < :001; univariate for willingness to help, F 1; 69 8:55, p < :01; univariate for anger,
F 1; 69 10:59, p < :01.
A MANOVA showed no main eect of group identication nor was the interaction between intent and group identication signicant (for group identication,
Multivariate F 3; 67 :64, p > :1; for the interaction between intent and group
identication, Multivariate F 3; 67 1:02, p > :1; the power of these tests to detect
dierences was .18 and .26, respectively). See Table 6 for means and standard deviations of all dependent variables.
7.3. Discussion
We designed Study 4 to see if group identication and perpetrator intent inuence
third-party reactions to injustice. We found that perpetrator intent signicantly inTable 6
Cell means and standard deviations of dependent measures, Study 4
Condition
Measure
n
High group identication/high

perpetrator intent
M
SD
18
High group identication/low

perpetrator intent
M
SD
18
Low group identication/high

perpetrator intent
M
SD
19
Low group identication/low

perpetrator intent
M
SD
18
Perceived
oensiveness
Willingness to
help
Anger
4.19
.64
4.67
1.97
5.28
1.49
3.34
.60
5.83
.92
4.17
1.42
3.80
.82
4.53
1.50
5.05
1.43
3.42
.60
5.39
1.33
4.00
1.33
Note. The higher the mean, the more in the label direction. Perceived oensiveness used a 15 (not at
all/extremely) scale, whereas the other two measures used 7-point, disagree/agree scales.
859
uenced third party responses to injustice whereas group identication did not. The
main eect of intent is consistent with prior theorizing by Folger and Cropanzano
(1998, 2001), suggesting that third parties who infer perpetrators malevolent intent
will be motivated to hold them accountable and morally blameworthy, and that such
responses do not depend on self-identity concerns. Moreover, the lack of a main
eect of identication implies that people can value fairness for its own sake, even in
the absence of self-identity concerns. We studied third party responses to injustice
where the impact of identication is likely to be less direct than had we studied the
responses of the victims themselves, however, and studies on the victims themselves
might be more likely than observer studies to nd main eects consistent with a
group-value/relational explanation.
We also explored the interaction between intent and identication. Although it
was not signicant, the perceived oensiveness means indicate the most negative
reactions to injustice when both perpetrator intent and group identication were
high. In this regard, our ndings suggest that group identication might magnify
certain already existing (independently induced) tendencies to condemn injustice on
grounds other than group identity, and that those responses are based at least in part
on the intent of the perpetratorwhich suggests that holding other people accountable for their intentions is one such independent source of condemnation
tendencies. Note also that not obtaining signicant eects for that interaction might
be due to lack of power.
The results of Study 4 further suggest a need to consider alternative explanations
for why fairness matters, beyond existing theories based on economic or relational
self-interest. Further exploration of accountability and moral virtue should provide a
more complete conceptualization of how and why people respond to injustice. This is
the topic of our concluding sections.
8. Conclusions
We conclude by rst reviewing how these studies rule out alternative hypotheses
and thus support a deontic (non-self-interested) perspective. We also address the selfinterest of impression management, including managing ones own impressions of
oneself toward oneself.
8.1. Reviewing the evidence
An author of the Kahneman et al. (1986) study described it as showing that participants sacrice money to punish an Allocator who behaved unfairly to someone
else (Thaler, 1992, p. 24, emphasis in original). That is, their study focused on thirdparty observers rather than the victims of unfairness themselves. Before moving on to
a further commentary about our own investigations, however, we must add two
qualications to Thalers description. First, the reference to an Allocator who behaved unfairly mischaracterizes an important feature of the original study: That
allocator (U for uneven, a person indicating a preference for an $18$2 division)
behaved unfairly in a more indirect manner than has most often been studied in the
literature on organizational justice. The unfair behavior was only the indication of
an intention (a preference for unfairness) rather than the iniction of actual harm,
which makes the punishment of such unfairnessespecially when made self-sacriciallyeven more dramatic than in the usual studies of organizational justice. We
investigated the role of intent as a specic focus in Study 4.
Second, we must qualify the characterization of the reaction as the willingness to
sacrice money to punish intended unfairness. In the original Kahneman et al.
study, those who sacricially punished intended unfairness also simultaneously re-
860
warded intended fairnessthe confound addressed in our rst two studies. The results of our attempt to disentangle punishment and reward in Study 1 conrmed
Hypothesis 1a (predicting that participants would more often reward an Even
partner sacricially than an Uneven partner sacricially) signicantly, whereas the
means for Hypothesis 1b (predicting that participants would be more likely to punish
an Uneven partner sacricially than an Even partner sacricially) were in the right
direction but did not reach conventional levels of signicance. The latter tendency
reected a marked overall unwillingness to punish anyone in the two-person (Partner
Knowledge) conditions we created as a special decomposition of the original
Kahneman et al. paradigm. This interpretation is also consistent with the unexpected
main eect for Payout in Study 1, such that an average of 83% of our respondents
chose to reward sacricially and only 10% chose to punish sacricially. The dramatically reduced punishment in the two-person design led to the Study 2 logicthat
we would have to eliminate a two wrongs dont make a right conict between two
moral motives.
Study 2 used three conditions: (a) a replication condition from Study 1 of the twoparty opportunity to punish an Uneven partner; (b) a replication condition from the
KKT design of a three-party opportunity to punish Uneven and reward Even simultaneously; and (c) a new three-party condition, involving Uneven and someone
allegedly from a prior experiment unrelated to allocations (revealing nothing about
that persons intentions). The results conrmed our prediction. Again, no punishment of Uneven occurred in the two-party condition, which diered signicantly
from punishment that was equally high, regardless of whether the three-party condition contained an unrelated or Even partner. These results are the rst in the KKT
tradition to reveal self-sacricial punishment of intended injustice unconfounded
with a chance to reward justice.
We turned next to generalizing this core nding in Studies 3 and 4. Study 3
extended that nding to the realm of punishing interactional injustice and also
provided a further opportunity to remove that punishment from other eects
achievable simultaneously. Prior to Study 3, all three-party variations on the
Kahneman et al. paradigm linked the $6-for-oneself option not only with $6 for
Uneven but also with $0 for someone else (e.g., in Study 2, either Even or a person
from an unrelated experiment). The unrelated partner in Study 2 was not rewarded
for fairness by the choice of the self-sacricial option, but choosing it did prevent
the unrelated partner from being punished by the $0 allocation that would have
resulted from choosing the $6-for-onself option instead. One of the unique contributions of Study 3 therefore was the change that substituted $6 (rather than $0)
for the unrelated partner in the $6-for-oneself option. Another contribution was the
evidence for punishing an interactional injustice, thereby generalizing the original
paradigm beyond circumstances in which the dependent variableallocations of
outcomesremained in the same justice domain as that of the (prior) intended
unfairness.
Study 4 extended further generalizability in terms of both the circumstances of the
prior unfairness and the dependent variable. Rather than focusing exclusively on a
dichotomous measure of punishment allocations (which of two options chosen), we
altered dependent measurement to tap a broader construct regarding the condemnation of misconduct (its oensiveness rated on continuous measurement scales).
Study 4 also included manipulations of both group identication and intent.
Whereas we had assumed a high level of intent held constant in the rst three studies,
its manipulation and signicant eects in Study 4 provided conrming evidence to
support that assumption and to demonstrate the important role of perpetrator
motives more generally. Study 4 did not nd eects related to the manipulation of
group identication, but further exploration regarding boundary conditions for its
impact might be worthwhile.
861
Taken together, the results of these four studies cast doubt on the reasoning that
self-interest, either material or social, is the sole, or even prime, motivator for
people to attend to issues of fairness. First, in studies 13, a signicant number of
participants were willing to sacrice their material self-interest in reaction to unfairness, even though there were no short- or long-term material benets for doing
so. We chose the Kahneman et al. paradigm as our starting point precisely because
it provides such a severe challenge to the view of humans as maximizers of material
outcomes. Its eects, especially in the expanded range revealed by our variations on
that paradigm, also challenge a host of other conventional theories. Punishing
unfairness in these studies does not reect an equity-restoration response (cf. Adams, 1965), for example, because the punished person never actually created an
inequity.
Second, social self-interest, or group identication, also does not seem to explain these results. Note the role of the person with unfair intent in the original
Kahneman et al. paradigm as well as in our initial extensions of it (Study 1 and
Study 2). That other persons exact identity was unclear; in our versions, for
example, s/he was allegedly a student in some other class who participated in a
prior experiment. The intended unfairness was in that person trying to keep $18
out of $20 for him- or herself. Why would that threaten respondents self-identity?
They surely would not see that persons unfair intentions as an eort to exclude
them from membership in a group important to their self-identities. True, selfidentity based on acceptance by a group was developed as an explanation for
violations of procedural justice rather than the distributive violations reected by
the punished persons unfair intentions. The evidence inconsistent with material
self-interest as a motive, however, also runs counter to the only other explanation
yet proposed in cases where material self-interest has failed to account for obtained ndings!
We also varied the identity of the unfair person both between and within studies.
We can see no indication that this identity mattered to our respondents. In the rst
two studies, for example, the unfair person was another studentand perhaps it
might be argued that students were willing to punish unfair intentions because the
punished person had attempted to deprive a fellow student. Study 3 therefore created
a situation psychologically removed from our undergraduates close identication
(i.e., full-time employees at an unnamed organization in a distant city). Despite some
degree of identity-based remoteness, that study still produced a substantial indication of the desire to hold the perpetrator accountable for acts of interactional injustice.
Similarly, Study 4 found condemnation of intentional unfairness (signicantly
stronger than when intent was absent), despite a context with no self-identity implications from group membershipnamely when interactional injustice aected
international gymnasts. We think it strains credulity to explain those negative reactions as based on a desire to feel accepted into the community of gymnasts, or to
assume that our respondents would base self-identity on membership in that particular group. Note that we also failed to nd an eect of group identication in
Study 4, despite signicant manipulation-check evidence for the impact of that independent variable. Null eects can be dicult to interpret, however, and we certainly would not be surprised to nd the intensity of reactions varying as a function
of empathy-based identication in other studies (e.g., caring more about what
happened to a friend than to an enemy).
Our studies open the door to a search for motives that go beyond both material
self-interest and self-identity based on group membership. Although we do not claim
that these investigations yield a proof for the existence of other motives or denitive
clarication of their nature, we hope others will nd this search worthwhile in light
of the apparent inapplicability of motives addressed in the prior justice literature. In
862
particular, we suggest discounting self-interest, either material or social, as the only

explanation for the importance of justice.
Is it possible, however, that we have just found another guise for self-interest?
Following Greenberg (2001), Colquitt and Greenberg (2001) subsume all three
viewsnot just the material and relational-identity but also the deonticunder selfinterest. Greenberg (2001) argued that some roots of selshness could still be the
underlying motive, even when people seem to be acting deontically on principle
(called ostensibly moral motives for behaving fairly, p. 214). For example, virtuous-appearing deeds might simply reect actions taken by people who want to
protect their self-images or to avoid feelings of guilt (Colquitt & Greenberg, 2001,
p. 221).
We disagree. Holley (1999) is one of the many philosophers showing cogently that
self-interest (called egoism in philosophy) as a universal explanation for all behavior
becomes a denitional tautology incapable of disproof: If all else fails, there is a
strategy used by advocates of universal egoism to make their case airtight against
any counterexamples. It involves claiming that a person must be motivated by selfinterested desires because what moves her to act is always her own desires (p. 42). In
eect, this claim says what I mean by a self-interested desire is simply any desire of
the person who performs an action (p. 42). The vacuous circularity in this tautological denition is clear: In that case, the egoist has given us a way to claim that
egoism is true, but the cost of defending the thesis in this way is to render it trivial
(p. 42).5
8.2. Beyond self-interest: Quo vadis?

If self-interest does not account for our resultsif virtue really is its own rewardwhat is the next step? We do not pretend to answer that question but instead
point in three directions other than concerns that focus exclusively on the self: Toward other people, toward the oender, and toward the moral precept that the offender violates. The rst of those future directions, pointed toward other people,
suggests focusing on victims characteristics rather than exclusively on consequences
aecting the self directly. Simply put, this arms common sense: People sometimes
experience outrage about an oense, even when it aects others with whom they have
We quote without comment some of Holleys other remarks about self-interest:

Ideas can be enticing. Sometimes we nd ourselves attracted by an idea, not so much on the
basis of evidence, but because it provides a way of thinking that seems so congenial. When
we think in accordance with the idea, it is as if we gain an insight that claries and simplies
matters. We can scarcely even imagine how things could be otherwise. Even if someone gives
us a proof that the idea is mistaken, we nd ourselves slipping back into the comfortable
patterns of thought it provides, believing that the idea must be true nevertheless.
One of those enticing ideas that exerts a powerful inuence on many people is the view that
self-interested motivations underlie all human actions. This idea oers a way of interpreting
human behavior that radically simplied our understanding of why people do what they do.
According to this view, alternative motivations for behavior, such as concern for the good of
others, do not exist. When we appear to display such concern, what we really exhibit is a
disguised form of self-interest, for we only seek the good of others as a means to achieving
our own good. From the mind of the college sophomore, who believes he has found an
important truth about human nature, to the works of scholars in the social sciences, the idea
of universal egoism is accepted as an obvious truth. Whether this truth is supported by the
evidence is often a matter of indierence to those who are convinced of it. They nd it
dicult even to imagine the possibility of being wrong (Holley, 1999, pp. 2930).
Holley provides useful references to other attacks on egoism by Feinberg (1996) and Rachels (1986).
Other references can be found in discussions by Cropanzano, Byrne et al. (2001) and Folger (2001).
863
very little direct aliation (e.g., a heterosexuals outrage about unfairness toward
homosexuals or the oensiveness of genocide that does not threaten ones own ethnic
group).
The second suggested directionattention focused on the oender rather than
focused exclusively on the selfis more novel than the rst and more unique than
it might seem if viewed supercially. We argued against self-identity interpretations
based on group inclusion or acceptance, but we do acknowledge that failing to
condemn injustice could threaten a persons self-identity, which some might see as
a nuance rather than a meaningful dierence. Consider, however, the dierence
between observing someones intended mistreatment of another person and asking
the following two kinds of questions: (a) What kind of person does that make
me? (e.g., in contemplation of doing nothing) versus (b) What kind of person
does that make you? (e.g., referring to the perpetrator of oensive misconduct).
We argue even more generally that responses to injustice may be driven by answers
to perpetrator-focused questions such as, Who does she think she is? and What
does that act of injustice say about him? in addition to the current emphasis on
self-as-victim questions such as, Am I a valued and respected member of this
group? and How well will my long-term economic interests be served by this
group?
Past theorizing and research on reactions to injustice typically have put the self in
the victim position, whereas we call for treating those perspectives dierently as a
way to opening up new investigation arenas. We think that even a shift of attention
to others as victims will fall short of providing a comprehensive perspective on
justice unless the role of perpetrators also gets special attention. The Study 4 manipulation of perceived intent illustrates a perpetrator focus.
We think a third future direction may have the most innovative implicationsin
part because it is the least explored. In abstract terms, we had referred earlier to a
new direction pointing toward the moral precept that the oender violates.
Concrete examples include sacrilegious grati or the mutilation of dead bodies; or
imagine a corporate entity that buys the Grand Canyon and converts it to an
amusement park. All that we are saying is that certain actions seem horric in and of
themselves, and the capacity for moral outrage about such actions might transcend
the specics about the oender, the victim, and the group-based grounds for our own
self-identity.
Lest we seem to have traveled too far from the domain of organizational justice,
we end on an applied note about implications for organizations. Consider a litigious society in which insurance companies and large conglomerates become the
targets for law suits including class actions. If found guilty in U.S. courts, such
rms are subject to payment for two types of damages: Compensatory and punitive. Compensatory damages provide payments to victims; punitive damages
constitute an independent judgment about the oensiveness of the act itself (and
also serve purposes such as expressing moral outrage over injury to the innocent).
By pursuing research directions such as those illustrated by the current investigations and our speculative theoretical commentary, we think organizational scholars
in the future can have much more to say about the determinants of such judgments
than can be said based on the contemporary literature. If nothing else, we think
our call for attention to virtue-as-reward eects regarding reactions to unfairness
might help shed new light on citizen reactions to corporate conduct and perceived
misconduct.
We have shown that people sometimes sacrice nancial well-being to condemn
(even if only privately) wrongful intent. The same people who give up a dollar in the
laboratory will one day, if not already, own stocksand might refrain from investing in a company on moral grounds, even if it oered a more attractive investment opportunity than that investors other options.
864
References
Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.), Advances in experimental social
psychology (Vol. 2) (pp. 267299). New York: Academic Press.
Aquino, K., Tripp, T. M., & Bies, R. J. (2001). How employees respond to personal oense: The eects of
blame attribution, victim status, and oender status on revenge and reconciliation in the workplace.
Journal of Applied Psychology, 86, 5259.
Batson, C. D. (1994). Why act for the public good? Four answers. Personality and Social Psychology
Bulletin, 20, 603610.
Batson, C. D. (1995). Prosocial motivation: Why do we help others. In A. Tesser (Ed.), Advanced social
psychology (pp. 333381). Boston, MA: McGraw-Hill.
Batson, C. D., Bowers, M. J., Leonard, E. A., & Smith, E. C. (2000). Does personal morality exacerbate or
restrain retaliation after being harmed? Personality and Social Psychology Bulletin, 26, 3545.
Bies, R. J. (1987). The predicament of injustice: The management of moral outrage. In L. L. Cummings, &
B. M. Staw (Eds.), Research in organizational behavior (Vol. 9) (pp. 289319). Greenwich, CT: JAI
Press.
Bies, R. J., Tripp, T., & Kramer, R. M. (1997). At the breaking point: Cognitive and social dynamics of
revenge in organizations. In J. Greenberg, & R. Giacalone (Eds.), Anti-social behavior in organizations
(pp. 1836). Thousand Oaks, CA: Sage.
Bradeld, M., & Aquino, K. (1999). The eects of blame attributions and oender likeableness on revenge
and forgiveness in the workplace. Journal of Management, 25, 607631.
Brockner, J., Tyler, T. R., & Cooper-Schneider, R. (1992). The inuence of prior commitment to an
institution on reactions to perceived unfairness: The higher they are, the harder they fall.
Administrative Science Quarterly, 37, 241261.
Brockner, J., & Wiesenfeld, B. M. (1996). An integrative framework for explaining reactions to decisions:
The interactive eects of outcomes and procedures. Psychological Bulletin, 120, 189208.
Colquitt, J. A., & Greenberg, J. (2001). Doing justice to organizational justice: Forming and applying
fairness judgments. In S. Gilliland, D. Steiner, & D. Skarlicki (Eds.), Theoretical and cultural
perspectives on organizational justice. Greenwich, CT: Information Age Press.
Cropanzano, R., Byrne, Z. S., Bobocel, D. R., & Rupp, D. E. (2001). Moral virtues, fairness heuristics,
social entities, and other denizens of organizational justice. Journal of Vocational Behavior, 58(2), 164
209.
Cropanzano, R., & Folger, R. (1989). Referent cognitions and task decision autonomy: Beyond equity
theory. In R. M. Steers, & L. W. Porter (Eds.), Motivation and work behavior (5th ed.., pp. 131143).
New York: McGraw-Hill.
Cropanzano, R., Rupp, D. E., Mohler, C. J., & Schminke, M. (2001). Three roads to organizational
justice. In G. Ferris (Ed.), Research in personnel and human resources management (Vol. 20) (pp. 1
113). Greenwich, CT: JAI Press.
Dyck, R. J., & Rule, B. G. (1978). Eect on retaliation of causal attributions concerning attack. Journal of
Personality and Social Psychology, 36, 521529.
Eckel, C. C., & Grossman, P. J. (1996). The relative price of fairness: Gender dierences in a punishment
game. Journal of Economic Behavior and Organization, 30, 143158.
Feinberg, J. (1996). Psychological egoism. In J. Feinberg (Ed.), Reason and responsibility: Readings in some
basic problems of philosophy (9th ed.). Belmont, CA: Wadsworth Publishing Company.
Fiske, A. P. (1991a). Structures of social life: The four elementary forms of human relations. New York:
Free Press.
Fiske, A. P. (1991b). The cultural relativity of selsh individualism: Anthropological evidence that humans
are inherently sociable. In M. Clark (Ed.), Prosocial behavior (pp. 176214). Newbury Park, CA: Sage.
Fiske, A. P., & Tetlock, P. E. (1997). Taboo tradeos: Reactions to transactions that transgress spheres of
exchange. Journal of Political Psychology, 18, 255297.
Folger, R. (1986). Rethinking equity theory: A referent cognitions model. In H. W. Bierho, H. C. Cohen,
& J. Greenberg (Eds.), Justice in social relations (pp. 145162). New York: Plenum.
Folger, R. (1987). Reformulating the preconditions of resentment: A referent cognitions model. In J. C.
Masters, & W. P. Smith (Eds.), Social comparison, justice, and relative deprivation: Theoretical,
empirical, and policy perspectives (pp. 183215). Hillsdale, NJ: Lawrence Erlbaum Associates.
Folger, R. (1998). Fairness as a moral virtue. In M. Schminke (Ed.), Managerial ethics: Morally managing
people and processes (pp. 1334). Mahwah, NJ: Lawrence Erlbaum Associates.
Folger, R. (2001). Fairness as deonance. In S. W. Gilliland, D. D. Steiner, & D. P. Skarlicki (Eds.),
Research in social issues in management (pp. 331). Information Age Publishers.
Folger, R., & Cropanzano, R. (1998). Organizational justice and human resource management. Thousand
Oaks, CA: Sage.
Folger, R., & Cropanzano, R. (2001). Fairness theory: Justice as accountability. In J. Greenberg, & R.
Cropanzano (Eds.), Advances in organizational justice (pp. 155). Stanford, CA: Stanford University
Press.
865
Folger, R., & Kass, E. (2000). Social comparison and fairness: A counterfactual simulations perspective.
In J. Suls, & L. Wheeler (Eds.), Handbook of social comparison: Theory and research (pp. 423441).
New York: Plenum.
Folger, R., & Martin, C. (1986). Relative deprivation and referent cognitions: Distributive and procedural
justice eects. Journal of Experimental Social Psychology, 22, 532546.
Folger, R., Roseneld, D., & Robinson, T. (1983). Relative deprivation and referent cognitions. Journal of
Personality and Social Psychology, 45, 172184.
Folger, R., & Skarlicki, D. P. (1998). A popcorn metaphor for employee aggression. In R. Grin, A.
OLeary-Kelly, & J. Collins (Eds.), Dysfunctional behavior in organizations: Violent and deviant
behavior. Monographs in organizational behavior and industrial relations (pp. 4381). Stamford, CT:
JAI.
Greenberg, J. (2001). The seven loose can(n)ons of organizational justice. In J. Greenberg, & R.
Cropanzano (Eds.), Advances in organizational behavior (pp. 245271). Stanford, CA: Stanford
University Press.
Holley, D. M. (1999). Self-interest and beyond. St. Paul, MN: Paragon House.
Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1986). Fairness and the assumptions of economics.
Journal of Business, 59, s285s300.
Kim, W. C., & Mauborgne, R. (1998). Procedural justice, strategic decision making, and the knowledge
economy. Strategic Management Journal, 19, 323338.
Kulik, J. A., & Brown, R. (1979). Frustration, attribution of blame, and aggression. Journal of
Experimental Social Psychology, 15, 183194.
Lind, E. A., & Tyler, T. R. (1988). The social psychology of procedural justice. New York: Plenum Press.
Miller, D. T. (2000). Disrespect and the experience of injustice. Annual Review of Psychology, 52, 527553.
Pedhazur, E. J., & Schmelkin, L. P. (1991). Measurement, design, and analysis: An integrated approach.
Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers.
Quigley, B. M., & Tedeschi, J. T. (1996). Mediating eects of blame attributions on feelings of anger.
Personality and Social Psychology Bulletin, 22, 12801288.
Rachels, J. (1986). The elements of moral philosophy. New York: Random House.
Rozin, P., Lowery, L., Imada, S., & Haidt, J. (1999). The CAD triad hypothesis: A mapping between three
moral emotions (contempt, anger, disgust) and three moral codes (community, autonomy, divinity).
Journal of Personality and Social Psychology, 76(4), 574586.
Skitka, L. J. (2000, September). The moral mandate hypothesis. Paper presented in K. van den Bos (Chair)
Recent Advances in Social Psychological Theories of Justice, a symposium conducted at the
International Society for Justice Research, Rishon LeZion, Israel.
Thaler, R. H. (1992). The winners curse: Paradoxes and anomalies of economic life. New York: Free Press.
Thibaut, J. T., & Walker, L. (1975). Procedural justice: A psychological perspective. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Trevino, L. K., & Bies, R. J. (1997). Through the looking glass: A normative manifesto for organizational
behavior. In C. L. Cooper, & S. E. Jackson (Eds.), Creating tomorrows organizations (pp. 439452).
New York: Wiley.
Tyler, T. R., & Lind, E. A. (1992). A relational model of authority in groups. In M. P. Zanna (Ed.),
Advances in experimental social psychology (vol. 25) (pp. 115191). San Diego, CA: Academic Press.
Tyler, T., & Degoey, P. (1995). Collective restraint in social dilemmas: Procedural justice and social
identication eects on support for authorities. Journal of Personality and Social Psychology, 69, 482
497.
Tyler, T., Degoey, P., & Smith, H. (1996). Understanding why the justice of group procedures matters: A
test of the psychological dynamics of the group-value model. Journal of Personality and Social
Psychology, 70, 913930.
Weiner, B., Amirkhan, J., Folkes, V. S., & Verette, J. A. (1987). An attributional analysis of excuse giving:
Studies of a nave theory of emotion. Journal of Personality and Social Psychology, 52, 316324.

Turillo Et Al. - 2002 - Is Virtue Its Own Reward Self-Sacrificial Decisions For The Sake of Fairness

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Turillo Et Al. - 2002 - Is Virtue Its Own Reward Self-Sacrificial Decisions For The Sake of Fairness

Enviado por

Direitos autorais:

Formatos disponíveis

Organizational Behavior and Human

Decision Processes 89 (2002) 839865

Is virtue its own reward? Self-sacricial

2. A triad of justice approaches

3. A research program on motives transcending self-interest: KKT and beyond

We thank the editors for reminding us about Holmes and Batson.

Persons receiving outcomes

Student A (who made 18/2 choice)

Student B (who made 10/10 choice)

Persons receiving outcomes

Student who had

Student who had

Matrix #1: Even (10-10), Punish

Matrix #2: Even (10-10), Reward

Matrix #3: Uneven (18-2), Punish

Matrix #4: Uneven (18-2), Reward

ticipants sacricial choice of $5 for himself or herself allowed $5 instead of $0 going

Persons receiving outcomes

Matrix #1: Even (10-10), Punish

Matrix #2: Even (10-10), Reward

Matrix #3: Uneven (18-2), Punish

Matrix #4: Uneven (18-2), Reward

formation; nevertheless, the population of MBA students at the university from

Persons receiving outcomes

Goal (unrelated) Student

eect of partner, F 2; 37 15:49, p < :01. Following the omnibus analysis, we

sensitivity or the presence of callous indierence to anothers feelings, sometimes to

Results for you

guide to conduct should be internalized, however, rather than being completely

identication condition). Low group-identication participants were asked to

We asked the high group-identication participants to imagine a dierent situation, as follows:

High group identication/high

High group identication/low

Low group identication/high

Low group identication/low

particular, we suggest discounting self-interest, either material or social, as the only

8.2. Beyond self-interest: Quo vadis?

We quote without comment some of Holleys other remarks about self-interest:

Você também pode gostar