ATTITUDES AND SOCIAL COGNITION

The Cheater’s High: The Unexpected Affective Benefits of

Unethical Behavior

Nicole E. Ruedy Celia Moore

University of Washington London Business School

Francesca Gino Maurice E. Schweitzer

Harvard University University of Pennsylvania

Many theories of moral behavior assume that unethical behavior triggers negative affect. In this article,
we challenge this assumption and demonstrate that unethical behavior can trigger positive affect, which
we term a “cheater’s high.” Across 6 studies, we find that even though individuals predict they will feel
guilty and have increased levels of negative affect after engaging in unethical behavior (Studies 1a and
1b), individuals who cheat on different problem-solving tasks consistently experience more positive
affect than those who do not (Studies 2–5). We find that this heightened positive affect does not depend
on self-selection (Studies 3 and 4), and it is not due to the accrual of undeserved financial rewards (Study
4). Cheating is associated with feelings of self-satisfaction, and the boost in positive affect from cheating
persists even when prospects for self-deception about unethical behavior are reduced (Study 5). Our
results have important implications for models of ethical decision making, moral behavior, and self-
regulatory theory.

Keywords: cheating, dishonesty, ethics, unethical behavior, affect

About morals, I know only that what is moral is what you feel good decision making (Haidt, 2001; Kohlberg, 1969; Monin, Pizarro, &
after and what is immoral is what you feel bad after. Beer, 2007; Pizarro, 2000; Rest, 1986; Tangney, Stuewig, &
—Ernest Hemingway, Death in the Afternoon Mashek, 2007; Treviño, 1986), in part to explain the pervasiveness
I was heady with happiness. Since I hadn’t yet had my first taste of
of unethical behavior (e.g., Callahan, 2004; Mazar, Amir, & Ari-
alcohol, I couldn’t compare the feeling to a champagne high, but it ely, 2008; Mazar & Ariely, 2006; Murdock & Anderman, 2006).
was the most delightful sensation I’d ever experienced. Many of these models adopt Hemingway’s view, presuming that
—Frank Abagnale, Catch Me If You Can unethical behavior triggers negative feelings, such as guilt, shame,
and anxiety (DePalma, Madey, & Bornschein, 1995; Eisenberg,
These two quotations offer contradictory perspectives on the 2000; Massi, 2005), and that the prospect of experiencing these
emotional consequences of unethical behavior. Hemingway’s per- negative feelings curbs unethical behavior (Baumeister, Vohs,
spective is the more commonly espoused: immoral behavior DeWall, & Zhang, 2007; Wright, 1971).
causes individuals to feel bad. However, Frank Abagnale, the Notably, none of the extant models of ethical decision making
swindler made famous in the movie Catch Me If You Can, presents can account for Abagnale’s experience of feeling “heady with
an alternate view when describing his first successful scam: im- happiness.” In this article, we challenge the fundamental assump-
moral behavior can elicit positive affect (Abagnale & Redding, tion that unethical behavior necessarily triggers negative affect.
2000). A number of scholars have proposed models of ethical We propose that although many individuals believe that engaging
in unethical behavior will result in increased negative affect, acting
unethically can predictably trigger positive feelings. Across six
This article was published Online First September 2, 2013. experiments, we find support for our predictions.
Nicole E. Ruedy, Foster School of Business, University of Washington;
Celia Moore, London Business School, London, United Kingdom; Franc- The Role of Affect in Ethical Decision Making
esca Gino, Harvard Business School, Harvard University; Maurice E.
Emotions are a critical component of any decision (Schwarz,
Schweitzer, Wharton School, University of Pennsylvania.
The authors are grateful for helpful comments from Joseph Gaspar.
2000). This is particularly true of ethical decisions, which are often
Correspondence concerning this article should be addressed to Francesca personal and highly self-relevant. Departing from early theorizing
Gino, Baker Library, Bloomberg Center 447, Soldiers Field Road, Harvard that viewed ethical decision making as a primarily cognitive pro-
Business School, Harvard University, Cambridge, MA 02163. E-mail: cess (Kohlberg, 1969; Rest, 1986; Treviño, 1986), a growing
[email protected] literature has begun to explore the role of affect in these choices
Journal of Personality and Social Psychology, 2013, Vol. 105, No. 4, 531–548
© 2013 American Psychological Association 0022-3514/13/$12.00 DOI: 10.1037/a0034231

531
532 RUEDY, MOORE, GINO, AND SCHWEITZER

(Haidt, 2001; Monin et al., 2007; Pizarro, 2000; Tangney et al., victims, are more likely to result in PTSD. Research on emotional
2007). For instance, recent work in moral psychology shows that consequences of ethical transgressions in the workplace has also
ethical decisions are frequently informed by one’s feelings and focused on actions that cause direct harm to salient victims, such
intuitions (Greene & Haidt, 2002; Haidt, 2001). In fact, scholars as workplace bullying, harassment, discrimination, injustice, and
and philosophers alike have long assumed a link between affect victimization (Giacalone & Promislo, 2010).
and ethical decision making (see Plato, Republic, 2:359a–360d, In summary, empirical work across laboratory and field studies
10:612b; see also Doris, 2002, Chapter 8; Rawls, 1971). have supported the prevalent assumption that unethical behavior
Accounts of this link typically presume that immoral acts trigger triggers negative affect, but these studies have focused on acts that
negative affect (such as guilt, shame, and remorse) and that the harm a salient victim, often at the directive of someone else. What
anticipation of negative affect represents an expected cost that role does harm play in unethical behavior? Certainly, perceptions
curbs unethical behavior (Bandura, 1990; Baumeister et al., 2007; of harm play a critical role in moral judgment. For example, Gray,
McGraw, 1987; Schwarz, 2000; Wright, 1971). For example, Young, and Waytz’s (2012) model of the moral dyad suggests that
recall studies about emotional reactions to past transgressions have individuals’ prototype of wrongdoing involves interpersonal harm.
asked participants to recall times they felt a specific negative This means that the moral aspects of an act are more cognitively
emotion, such as guilt or shame (e.g., Baumeister, Stillwell, & accessible when it involves interpersonal harm. However, restrict-
Heatherton, 1995; Tracy & Robins, 2006). Similarly, prediction ing the definition of unethical behavior to acts that cause direct
studies have found that when individuals expect to experience guilt interpersonal harm would be problematic. Many acts of theft and
following a particular behavior, they are less likely to engage in dishonesty—including computer piracy, tax evasion, insurance
that behavior (Massi, 2005). These study designs are unlikely to fraud, workplace theft, and cheating in scholarly endeavors— do
detect positive affective reactions. The one study that solicited a not cause (obvious) direct harm to a salient victim. Yet, they do
broader set of emotional reactions asked participants to recall contravene a moral principle, such as fairness or honesty, and are
episodes of academic dishonesty (Whitley, 2001). In this study as therefore widely recognized as unethical, even without an element
well, participants reported that they had experienced more negative of interpersonal harm.
than positive affect. Thus, building on traditions in both moral psychology (Uhl-
However, the focus of these studies on recall and prediction mann, Pizarro, Tannenbaum, & Ditto, 2009) and philosophy (Hare,
limit the conclusions they can draw about the actual affective 1981; Kant, 1785/1993), we define unethical behavior as acts that
consequences of unethical behavior. Recalling affective experi- violate widely held moral principles (such as honesty or fairness).1
ences is difficult; individuals are notoriously inaccurate when This definition includes behaviors such as cheating, dishonesty,
recalling emotions (Thomas & Diener, 1990). In addition, individ- and stealing. Many behaviors that violate widely held principles
uals tend to both mispredict and misremember unethical behavior lack a specific, identifiable victim but harm the broader commu-
(Tenbrunsel, Diekmann, Wade-Benzoni, & Bazerman, 2010). As a nity and constitute unethical behavior (Alicke, 2012; Pizarro,
result, participants’ reports in these types of studies may reflect lay Tannenbaum, & Uhlmann, 2012). For instance, unethical actions
theories about affective reactions to unethical behavior or self- may undermine a norm (Gray, Waytz, & Young, 2012) or violate
presentational concerns rather than their actual reactions at the abstract concepts such as “community” (Rozin, Lowery, Imada, &
time they engaged in the behavior. Similarly, the few laboratory Haidt, 1999) or “purity” (Haidt, 2001).
studies that have directly tested the relationship between unethical Surprisingly, very little prior work has examined the affective
behavior and affect have examined a very specific type of uneth-
consequences of voluntary unethical behavior without obvious
ical behavior—acts that comply with an authority figure’s request
harm or a salient victim. This is an important omission, not only
to cause direct harm to a salient victim (e.g., administering an
because these types of unethical behavior are common (Callahan,
electric shock to a confederate at the request of the experimenter).
2004) and costly (Hollinger & Langton, 2007; U.S. Internal Rev-
These studies consistently found increases in negative affect fol-
enue Service, 2010) but also because the affective consequences of
lowing this type of unethical behavior (Brock & Buss, 1964; Buss
these acts may be very different. Thus, in contrast to the tradition-
& Brock, 1963; Okel & Mosher, 1968; Ring, Wallston, & Corey,
al—and Hemingway’s—perspectives, we propose that voluntary
1970). To our knowledge, only one study has examined the affec-
unethical acts without salient victims and obvious harm not only
tive consequences of elicited unethical behavior that did not in-
fail to elicit negative affect but may actually evoke positive affect,
volve direct harm to a salient victim (Wallington, 1973), and this
a phenomenon we term the “cheater’s high.” The idea that uneth-
study found no significant effects of unethical behavior on partic-
ical behavior can trigger positive affect is consistent with many
ipants’ negative affect. The author did not measure positive affect.
anecdotal accounts of dishonesty, theft, and fraud. These accounts
The small set of field studies that have investigated affective
include wealthy individuals who delight in shoplifting affordable
consequences of unethical behavior has also focused on behavior
goods (Seagrave, 2001), joy-riders who steal cars for the thrill
that involves harm to salient victims (i.e., where it is clear to the
(Katz, 1988), and fraudsters who revel in their misdeeds (Abagnale
perpetrator that they have directly hurt someone else by their
& Redding, 2000).
actions). For instance, Evans, Ehlers, Mezey, and Clark (2007)
We note that the “cheater’s high” is related to two streams of
found that out of 105 perpetrators of violent crimes, 46% reported
that they were haunted by distressing memories, and 6% showed prior research. First, it is related to Ekman’s (2001) concept of
signs of Posttraumatic Stress Disorder (PTSD), a condition char- “duping delight,” or the exhilaration caused by successfully de-
acterized in part by negative emotional responses. Byrne’s (2003)
work on trauma reactions in perpetrators of violent crimes suggests 1
Consistent with their etymological roots as synonyms, we also use we
that certain types of crimes, including those that involve known use the terms “immoral” and “unethical” interchangeably.
 CHEATER’S HIGH? 533

ceiving others, which to date, lacks an empirical demonstration altruistic actions in contexts such as donating time and money to
(DePaulo et al., 2003). Second, it is related to research on “for- charity (Epley & Dunning, 2000) and participating in a blood drive
bidden fruit,” which suggests that people find objects and experi- (Trope & Liberman, 2003).
ences that are forbidden or taboo to be more attractive and alluring In the same way that the “should” self mispredicts ethical
than those that are not forbidden (Alberts, Mulkens, Smeets, & behavior, we expect the “should” self to mispredict that unethical
Thewissen, 2010; Fishbach, 2009; Mann & Ward, 2001; Pech- behavior will evoke negative affect. We postulate that when people
mann & Shih, 1999). This work does not address the affective predict what their affective state will be after they act unethically,
consequences of unethical behavior but does suggest a contribut- their “should” self and their desire to be moral will be salient. The
ing factor: People may enjoy engaging in behaviors that are discrepancy between their moral self-image and the immoral acts
prohibited. they imagine engaging in will result in projected dissonance. From
We explore the link between unethical behavior and affect in six this perspective, focusing on ideals and values, the “should” self
experimental studies. We investigate the proposition that affective will overestimate the psychic cost that they will bear and will
reactions to unethical behavior are different from what prior work underestimate their reactions to any psychological or material
has assumed and what individuals predict. Specifically, we chal- benefits that result from the unethical behavior. Thus, we hypoth-
lenge a fundamental assumption of the ethical decision making esize that when individuals predict their affective reactions to
literature by considering the possibility that unethical behavior can unethical behavior, they will predict that they will experience
trigger positive affective consequences. Our findings break new heightened negative affect (Hypothesis 1).
ground by demonstrating that theoretical models of ethical deci-
sion making should account for a broader range of affective Positive Affective Consequences of
consequences than they currently do.
Actual Unethical Behavior

Negative Affective Consequences of Affective forecasts (e.g., how much we will enjoy a planned
beach vacation) are commonly biased (Gilbert, Gill, & Wilson,
Predicted Unethical Behavior
2002; Kurtz, Wilson, & Gilbert, 2007; Wilson & Gilbert, 2003),
The pervasiveness and frequency with which people engage in and we expect individuals to mispredict how they will feel fol-
unethical behavior (Callahan, 2004) contrasts with the desire to lowing unethical behavior. In the case of affective reactions to
maintain a positive self-concept both privately and publicly (Adler, unethical behavior that lack a salient victim, we expect people to
1930; Allport, 1955; Rogers, 1959). People are motivated to re- predict they will experience heightened negative affect after be-
solve this tension, even when doing so requires a degree of having unethically when in fact they will experience a boost in
self-deception or pretense (Schlenker, Forsyth, Leary, & Miller, positive affect.
1980; Tajfel, 1982). Moral values constitute a central aspect of an We expect the “should” self to guide predictions but the “want”
individual’s positive self-image (Chaiken, Giner-Sorolla, & Chen, self to guide ethical decisions (Tenbrunsel et al., 2010). People in
1996), and consistent with the motivation to maintain a positive a “want” state are tempted by something they acutely desire. They
self-image, most people perceive themselves as honest and deserv- may feel a strong pull to engage in behavior that confers an
ing and believe firmly in their own morality (Aquino & Reed, immediate reward, even if the behavior is not consistent with their
2002; Greenwald, 1980; Sanitioso, Kunda, & Fong, 1990). ideals and long-term goals. In a classic self-control situation, an
The misalignment between actions (e.g., one’s unethical behav- individual in the “want” state might sabotage their weight loss
ior) and internal desires (e.g., one’s desire to be ethical) creates a plans by eating a rich dessert. In the moment, the immediate
psychological threat for individuals as well as a sense of disso- reward is more compelling than adherence to their long-term goal.
nance (Festinger, 1957). Behavioral ethics scholars argue that In the context of ethics, someone who is tempted to engage in an
when people face ethical dilemmas, they experience a tension unethical behavior, guided by the “want” self, will be more likely
between a person’s “want” and “should” selves (Bazerman, Ten- to focus on the immediate benefits they might accrue rather than
brunsel, & Wade-Benzoni, 1998; Mead, Baumeister, Gino, the potential costs that they may incur later, such as diminished
Schweitzer, & Ariely, 2009; Tenbrunsel et al., 2010). The “want” self-image or risk to their reputation (Loewenstein, 1996). If an
self is defined by short-term goals, impulsivity, and “hot” deci- individual is not attentive to long-term consequences, unethical
sions, whereas the “should” self involves long-term, rational, and acts may fail to induce negative affect in the moment in which an
“cold” decisions. Tenbrunsel et al. (2010) propose that the individual makes a decision.
“should” self dominates the prediction phase (when people con- In the “want” state, individuals may focus on the short-term
sider future decisions) and that the “want” self dominates the benefits they derive from a behavior. As a result, in the “want”
action phase. As a result, people overestimate their future ethical- state unethical behavior may elicit positive affect. There are sev-
ity. eral reasons why individuals may derive psychological benefits
This prediction is consistent with findings that ethical forecasts from engaging in unethical behavior. First, unethical behavior
reflect one’s hopes and ideals (Kivetz & Tyler, 2007; Newby- often confers an advantage, such as affording access to additional,
Clark, Ross, Buehler, Koehler, & Griffin, 2000) and undervalue undeserved resources. Undeserved gains from unethical behavior
both situational constraints (Diekmann, Tenbrunsel, & Galinsky, may be encoded as windfalls, whether they represent financial
2003) and one’s motivations in the heat of the moment (Diekmann gains (e.g., additional money), social gains (e.g., beating an oppo-
et al., 2003; Diekmann, Walker, Galinsky, & Tenbrunsel, 2012). nent), or psychological gains (e.g., better grades). These windfalls
As a result, predictions about one’s ethical behavior tend to be are positive events that can trigger positive affect (Arkes et al.,
overly optimistic. For example, people overestimate their future 1994; Meloy, Russo, & Miller, 2006).
534 RUEDY, MOORE, GINO, AND SCHWEITZER

Second, unethical behavior may confer psychological benefits find further support for our thesis, and in Study 5, we examine the
derived from a sense of greater autonomy and influence. Engaging underlying mechanism for the cheater’s high. We find that after
in unethical behavior allows actors to circumvent rules by which cheating, people feel more satisfied with their actions, even in the
others are bound, giving cheaters an expanded range of options and presence of explicit acknowledgment that the experimenters are
greater control over their outcomes. For instance, deception en- aware that participants may have cheated.
ables individuals to take advantage of others by manipulating the
information that others use to make decisions (Lewicki, 1983). As Study 1: Affective Predictions
a result, unethical behavior may provide an enhanced sense of
control, which can increase positive affect (Carver & Scheier, We conducted two studies to gauge affective forecasts following
1990). unethical behavior. In the first study, we asked participants to
Third, unethical behavior often involves the challenge of break- predict how they would feel after cheating in an experiment. In the
ing rules and overcoming systems that are designed to constrain second, we asked participants to predict how they would feel after
behavior. Convincing one’s boss of a lie or finding a loophole in lying on a timesheet to their employer. In both cases, we varied
a tax code may represent an interesting and enjoyable challenge. In whether participants forecasted their own feelings or the feelings
a study of cheating at work, a supermarket cashier who consis- of someone else. We included this manipulation to test whether
tently embezzled from her register explained that the behavior individuals predict different affective reactions for others than they
made her job “more interesting; it gave her new targets and a sense predict for themselves. This manipulation also helps us determine
of challenge” (Mars, 1982, p. 31). Many people enjoy engaging in whether individuals’ affective predictions are influenced by self-
effortful cognitive endeavors (Cacioppo, Petty, Feinstein, & Jarvis, presentational concerns.
1996), and succeeding in challenging tasks can engender a sense of
pride. Study 1a: Method
To date, the fields of behavioral ethics and criminology have
paid scant attention to the possibility that unethical behavior trig- Participants and design. We recruited 290 participants on
gers positive affect. In one exception using first-hand accounts, Mechanical Turk (hereinafter MTurk; Mage ⫽ 29.29, SD ⫽ 8.91;
Katz (1988) describes the emotional seduction or “sneaky thrills” 65% male; 33% students) to complete an online survey for a
that individuals derive from crimes, such as joyriding in stolen payment of $0.50. We dropped eight participants with duplicate IP
cars, vandalism, and shoplifting affordable goods. Katz (1988) addresses, and we report results for the remaining 282.2
notes that in many first-hand descriptions of these events, the We randomly assigned participants to one of four conditions in
euphoria of successfully getting away with the crime overshadows a 2 (self vs. other) ⫻ 2 (cheat vs. honest) between-subjects design.
any material gain from the behavior. Other criminology research- In each condition, participants read a description of an experiment
ers also note that “getting away with it” can feel good (Matza & involving a task that allows participants the opportunity to cheat.
Sykes, 1961; Moore & Gullone, 1996; Scully & Marolla, 1985). Half the participants imagined themselves in the study (self con-
The fact that one’s behavior is forbidden may enhance this expe- ditions), and half imagined someone else in the study (other
rience (Wood, Gove, Wilson, & Cochran, 1997). This explanation conditions).
is consistent with Frank Abagnale’s opening quote. We expect that Procedure. We told participants that we were interested in
unethical behavior that does not immediately cause interpersonal their predictions of how they or someone else might react to an
harm to create psychological benefits and increase positive affect experiment we were planning to run in the near future. We de-
(Hypothesis 2). scribed the experiment to them, and told them that participants in
this study would have the opportunity to act unethically. The
instructions read,
Overview of the Research
In the study, we will give participants a simple math test to complete
We tested our predictions in six experiments. We first con- under time pressure. The test sheet consists of 20 matrices, each based
ducted two studies that asked individuals to predict how they on a set of 12 three-digit numbers. We will tell participants that they
would feel if they acted unethically. In these two prediction have 5 minutes to find two numbers per matrix that add up to 10. Each
studies (Studies 1a and 1b), participants imagined having the correctly solved pair pays them $1. In addition to the test sheet, each
opportunity to behave unethically and deciding either to act participant will receive an envelope with 20 one-dollar bills. At the
honestly or dishonestly. In both studies, participants predicted end of the study, participants will have to take money according to
they (or someone else) would experience negative affect after their performance and turn back the envelope to the experimenter with
behaving unethically. the unearned money. About 10 –15 participants will participate in
Next, we conducted four experiments that afforded participants each session.
the opportunity to cheat. Contrary to what people predicted, across Once the 5 minutes are up, we will give participants a collection slip
these studies, engaging in unethical behavior (compared to ethical on which they are asked to report the number of correctly identified
behavior) increased positive affect without an accompanying in- pairs they’ve found. We will then instruct participants to first recycle
crease in negative affect. In Study 2, we identify a “cheaters high”: their test sheet (using the recycling box in the room) and then fill out
individuals who over-reported their performance experienced a
boost in positive affect and no boost in negative affect, compared 2
Following a reviewer’s suggestion, we dropped online participants
to non-cheaters. Study 3 replicates the cheater’s high using novel with duplicate IP addresses because we could not ensure that they did not
experimental methods to address self-selection issues. In Study 4, complete the study twice. However, in every case, the results are un-
we disentangle the effects of cheating and monetary rewards to changed when we leave in these participants.
 CHEATER’S HIGH? 535

their collection slips. Once they complete the collection slip, partici- than those in the other condition, F(1, 278) ⫽ 14.84, p ⬍ .001,
pants will pay themselves accordingly and take the unearned money in ␻2 ⫽ 0.047; however, the interaction with cheat versus honest was
the envelope to the experimenter as they leave the room. That means not significant, F(1, 278) ⫽ 0.00, p ⫽ .99.
that in this study participants can over-report their performance by The analogous univariate ANOVA for negative affect revealed
reporting a score that is higher than their actual performance on the
that participants across self and other conditions predicted that
task, and thus make more money on the task.
cheating would evoke greater negative affect (M ⫽ 2.84, SD ⫽
Manipulation. In the self-cheat condition, we asked partici- 0.93) than would honest behavior (M ⫽ 1.50, SD ⫽ 0.63), F(1,
pants “Please imagine being a participant in this experiment. 278) ⫽ 198.73, p ⬍ .001, ␻2 ⫽ 0.412. We found no difference for
Imagine that you decided to over-report your performance, and participants in the self or other conditions, F(1, 278) ⫽ 0.05, p ⫽
you report having solved more matrices correctly than you actually .821, and the interaction between self versus other and cheat
did solve correctly. How likely would you be to experience each of versus honest was not significant, F(1, 278) ⫽ 0.99, p ⫽ .32.
the following emotions right after reporting the score?” In the
self-honest condition, we asked participants to “Imagine that you Study 1b: Method
decided to report your performance accurately, and you report
having solved exactly the number of matrices correctly that you Participants. One hundred and thirty-seven students (45%
actually did solve correctly. How likely would you be to experi- male; Mage ⫽ 20.4, SD ⫽ 1.89) completed a survey as part of a
ence each of the following emotions right after reporting the laboratory session in exchange for $10.
score?” Procedures in the other-cheat and other-honest conditions Design and procedure. Participants read one of four versions
were identical except that the questions related to someone else, of a scenario from a 2 (self vs. other) ⫻ 2 (cheat vs. honest)
for example, “Please imagine a participant in this experiment. between-subjects design. In the self condition, participants imag-
Imagine that this person decided to over-report their performance ined they were Pat, a consultant at a large consulting company. In
. . .” the other condition, participants imagined they were Pat’s man-
Measures. Participants completed the 20-item Positive and ager. In the vignette, Pat completes a time sheet that will determine
Negative Affectivity Schedule (PANAS; Watson, Clark, & Telle- whether or not Pat will earn a $500 bonus. Pat is 5 hours short of
gen, 1988), which measures both positive and negative affect (PA: the 500 billable hours required for the bonus. However, Pat could
␣ ⫽ .88, NA: ␣ ⫽ .94) using a 5-point scale (ranging from 1 ⫽ bill 5 hours spent on a training course to meet the goal without
Very slightly or not at all to 5 ⫽ Extremely). Participants also anyone finding out, although this is against company policy. In the
answered demographic questions about their age and gender. cheat condition, participants read “Imagine that [you/Pat] billed 5
Pilot study. To verify that participants viewed the over- training hours. As a result, [you/Pat] reported 500 hours for the
reporting we described in this scenario as unethical, we conducted quarter and reached the target for the bonus.” In the honest
a pilot test. We recruited a separate MTurk sample of 107 partic- condition, participants read, “Imagine that [you/Pat] did not bill
ipants (Mage ⫽ 30.40, SD ⫽ 9.13; 51% male; 22% students) for a any training hours. As a result, [you/Pat] reported 495 hours for
payment of $0.25. (There were no duplicate IP addresses in this the quarter and did not reach the target for the bonus.”
sample.) In this study, we asked participants to rate the ethicality After reading the vignette, participants reported the extent to
of over-reporting one’s performance in this same scenario. That is, which they or Pat would feel positive and negative affect after
participants read the same description as used in Study 1a, and they reporting hours, using the PANAS (Watson et al., 1988, PA: ␣ ⫽
then rated the extent to which they thought over-reporting perfor- .78, NA: ␣ ⫽ .89).
mance in this situation was morally wrong, unethical, and dishon-
est (␣ ⫽ .88) on a 7-point scale (ranging from 1 ⫽ Not at all to 7 ⫽ Results and Discussion
Very much).
Participants rated over-reporting performance as very unethical. We conducted a univariate ANOVA using positive affect as the
On the 7-point scale, the average combined rating was 5.87 (SD ⫽ dependent measure and self versus other and cheat versus honest
1.32); this average rating is significantly higher than 5, t(106) ⫽ as the two between-subjects factors. We found that participants
6.77, p ⬍ .001, but not significantly lower than 6, t(106) ⫽ ⫺1.05, predicted that cheating would lead to greater positive affect (M ⫽
p ⫽ .30. These results demonstrate that participants perceive 2.61, SD ⫽ 0.70) than honesty (M ⫽ 2.31, SD ⫽ 0.71),
over-reporting in this domain as unethical. F(1, 133) ⫽ 5.97, p ⫽ .016, ␻2 ⫽ 0.035. Participants in the self
condition reported lower levels of positive affect after cheating
than those in the other condition, F(1, 133) ⫽ 7.99, p ⫽ .005, ␻2 ⫽
Results and Discussion
0.049; however, the interaction with cheating condition was not
Participants in Study 1a predicted that they and others would significant, F(1, 133) ⫽ 0.00, p ⫽ .99.
feel significantly lower levels of positive affect and higher levels We conducted a similar univariate ANOVA for negative affect.
of negative affect after cheating than after acting honestly. We This ANOVA revealed that participants across self and other
conducted a univariate analysis of variance (ANOVA) using pos- conditions predicted that negative affect would be greater after
itive affect as the dependent measure and our two manipulations as cheating (M ⫽ 2.93, SD ⫽ 1.01) than after honest behavior (M ⫽
between-subjects factors. Supporting our first hypothesis, partici- 2.12, SD ⫽ 0.58), F(1, 133) ⫽ 31.94, p ⬍ .001, ␻2 ⫽ 0.184.
pants predicted that cheating would evoke less positive affect Participants in the self condition reported marginally lower nega-
(M ⫽ 2.58, SD ⫽ 0.83) than honesty (M ⫽ 3.05, SD ⫽ 0.75), F(1, tive affect than those in the other condition, F(1, 133) ⫽ 3.19, p ⫽
278) ⫽ 24.33, p ⬍ .001, ␻2 ⫽ 0.076. Participants in the self .076, but the interaction between self versus other and cheating
condition reported lower levels of positive affect after cheating condition was not significant, F(1, 133) ⫽ 0.16, p ⫽ .69.
536 RUEDY, MOORE, GINO, AND SCHWEITZER

Consistent with the results of Study 1a, results from this study Method
demonstrate that participants expect to experience more negative
Participants. One-hundred seventy-nine individuals (40%
affect after behaving unethically than after behaving honestly. Our
male; Mage ⫽ 25, SD ⫽ 8.80) from a large Northeastern university
findings regarding positive affect were mixed. In Study 1a, par-
participated in the study for a $10 payment plus a bonus based on
ticipants predicted lower positive affect for cheaters than for
their performance.
non-cheaters, but in Study 1b, participants predicted higher posi-
Procedure. Participants worked in individual cubicles. They
tive affect for cheaters. These results did not change depending on first completed an abbreviated 10-item version of the Positive and
whether participants predicted their own affective state or the Negative Affectivity Schedule (PANAS; Watson et al., 1988).
affective state of another person. The results of Study 1a are Specifically, we used the items excited, enthusiastic, interested,
consistent with existing models of unethical behavior; results from strong, and determined to measure positive affect (␣ ⫽ .86) and
Study 1b imply that participants expect cheating to evoke emo- upset, hostile, ashamed, jittery, and scared to measure negative
tional ambivalence. Neither study predicts a “cheater’s high.” affect (␣ ⫽ .75).
Participants did not predict that cheating, compared to acting Next, participants engaged in a timed anagram task. The task,
honestly, increases positive affect without an accompanying in- adapted from Ruedy and Schweitzer (2011) involved unscram-
crease in negative affect. bling as many of 15 words as participants could in 4 min. Partic-
ipants earned a $1 bonus for every correctly unscrambled word.
The page that listed the scrambled words was stapled to a packet
Criteria for Establishing the Cheater’s High of sheets and a manila folder. In the packet were two sheets of
To document the cheater’s high, we conduct three sets of tests: carbonless copy paper that, unbeknown to participants, created an
(a) We compare changes in positive affect across time for both imprint of their writing. At the end of 4 min, the experimenter
asked participants to tear the top sheet out of packet. The experi-
cheaters and non-cheaters. Specifically, we expect cheaters to
menter then collected the packets and folders. Participants were
experience higher levels of positive affect at Time 2 (post-
not able to see that the packets the experimenter collected con-
cheating) than at Time 1 (baseline), and we expect non-cheaters
tained a complete imprint of the words they had unscrambled.
not to experience a boost in positive affect. (b) We compare
Next, the experimenter handed out answer keys, and participants
positive affect after the cheating opportunity between cheaters and checked their work in complete privacy. This stage of the exper-
non-cheaters. We expect cheaters to report higher positive affect iment afforded participants the opportunity to cheat; as they
than non-cheaters at Time 2, but not to differ from non-cheaters at checked their answers, they could add answers to their answer
Time 1. (c) We compare the difference between positive and sheet. The experimenter determined whether or not participants
negative affect for cheaters and non-cheaters across time. We had cheated by comparing the imprint of the answers participants
expect this difference to vary across time; in the analyses, this had written down in the first 4 min of the study with the answer
means finding a significant three-way interaction of affect valence sheets they ultimately submitted for payment after checking their
(positive and negative), cheating (yes and no), and time (1 and 2).3 work.
In particular, we expect that at Time 1, the two-way interaction After, participants answered some brief questions about how
between affect valence and cheating will not be significant. At difficult and enjoyable they found the task. Then they completed
Time 2, we expect a significant simple two-way interaction be- the same 10-item affect measure as before (␣PA ⫽ .90, ␣NA ⫽ .66).
tween affect valence and cheating, with a significant difference in Participants were also asked to provide any comments they had
affective valence for cheaters (with positive affect higher than about the session so far. At the end of the session, participants
negative affect), and a non-significant or weaker difference be- were again asked to provide comments. No participants expressed
tween positive and negative affect for non-cheaters. We report suspicion that the purpose of the study was to study cheating
results for each of these three criteria in the four studies that behavior.
follow. Given the fact that our paradigms involve performing
cognitive tasks and that performance may influence both cheating Results and Discussion
(McCabe & Treviño, 1997) and affect (Carver & Scheier, 1990),
we include performance as a covariate in all the studies. We excluded six participants for failing to complete the second
affect measure. On average, participants correctly unscrambled
3.84 (SD ⫽ 2.08) of the 15 anagrams. Seventy-one (41.0%)
Study 2 participants cheated by writing in additional answers after the time
to work on the task had ended. Participants who cheated added
In Study 2, we consider affective reactions to actual cheating.
2.48 answers on average (SD ⫽ 1.98).
The experimental design affords participants an opportunity to Criterion 1. We expect cheaters, but not non-cheaters, to
cheat by overstating performance for an additional payment. This experience higher levels of positive affect at Time 2 (post-
behavior involves dishonesty and taking unearned money, behav- cheating) than they do at Time 1. That is, we expect a model with
ior that falls well within our definition of unethical behavior and
was also rated as unethical in the pilot study reported with Study
3
1a. We use repeated measures analyses to disentangle the influence We thank Linda Skitka, our associate editor, for guidance in develop-
ing these criteria.
of positive affect on cheating from the effect of cheating on 4
We also conducted our analyses without controlling for performance,
positive affect.4 and our results for positive and negative affect are nearly identical.
 CHEATER’S HIGH? 537

positive affect as the dependent variable to reveal a two-way task as a covariate. The results revealed a significant three-way
interaction between Time (1 vs. 2) and Cheating (yes, no). In interaction between cheating, affect valence, and time, F(1, 170) ⫽
addition, we expect simple effects tests to show a significant 5.53, p ⫽ .020, ␻2 ⫽ .025. To explore the nature of this three-way
increase in positive affect across time periods for cheaters but not interaction, we tested the simple interaction of Cheating ⫻ Affec-
for non-cheaters. We tested this by conducting a mixed ANOVA tive Valence separately at Time 1 and Time 2. At Time 1, the
with positive affect at Time 1 and 2 as the within-subjects factor, simple interaction was not significant, F(1, 170) ⫽ 0.38, p ⫽ .54.
condition as the between-subjects factor, and actual performance At Time 2, it approached marginal significance, F(1, 170) ⫽ 2.29,
on the matrix task as a covariate. Supporting the cheater’s high, the p ⫽ .13. Examining the simple effects at Time 2, the difference
two-way interaction was significant, F(1, 170) ⫽ 5.18, p ⫽ .024, between positive and negative affect was significant for both
␻2 ⫽ .024, suggesting that there were differences in the changes in cheaters (M ⫽ 1.87, SE ⫽ 0.12); F(1, 170) ⫽ 261.93, p ⬍ .001,
positive affect across conditions (see Figure 1). Simple effects ␻2 ⫽ .601, and non-cheaters (M ⫽ 1.65, SE ⫽ 0.10); F(1, 170) ⫽
testing confirmed that those who cheated experienced an increase 290.32, p ⬍ .001, ␻2 ⫽ .626, but the effect size was larger for
in positive affect, F(1, 170) ⫽ 46.77, p ⬍ .001, ␻2 ⫽ .175. In this cheaters than non-cheaters (albeit not significantly so).5
study, those who did not cheat also experienced an increase in In this study, all participants experienced increases in both
positive affect across rounds, F(1, 170) ⫽ 21.57, p ⬍ .001, ␻2 ⫽ positive and negative affect during the course of the experiment.
.106. This was perhaps due to the fact that the performance However, for those who decided to cheat, their boost in positive
bonuses were substantial, and even those who did not cheat earned affect was larger than the increase experienced by non-cheaters.
additional money. However, the effect size for time in the cheating Traditional significance testing did not support all of the criteria of
condition (␻2 ⫽ .175) was greater than the effect size for time in the cheater’s high hypothesis in this study. However, all the results
the no cheating condition (␻2 ⫽ .106), a result consistent with our were nonetheless in the predicted direction, providing suggestive
prediction. evidence for a cheater’s high. The Study also has two additional
Criterion 2. We expect cheaters to report more positive affect limitations that we address in our following studies. First, cheating
than non-cheaters at Time 2 but not to differ from non-cheaters at in this study was a self-selected behavior. This is broadly true of
Time 1. Simple effects testing showed that cheaters were direc- unethical behavior, but we seek to document the cheater’s high in
tionally higher in positive affect than non-cheaters at Time 2, a paradigm with random assignment. Second, cheating in this
though this difference was not significant, F(1, 170) ⫽ 2.76, p ⫽ study was associated with higher levels of payment. Though this is
.099, ␻2 ⫽ .001. Cheaters and non-cheaters did not differ in also a common feature of cheating behavior, we seek to replicate
positive affect at Time 1, F(1, 170) ⫽ 0.04, p ⫽ .84. the cheater’s high in a context that disentangles cheating from
Criterion 3. To test our third criterion, we conducted a mixed financial rewards.
ANOVA with cheating (yes, no) as the between-subjects factor,
affective valence (positive and negative) and Time (1 and 2) as Study 3: Ruling Out Self-Selection
within-subjects factors, and actual performance on the anagram
An important challenge for scholars investigating unethical be-
havior is that unethical behavior is often self-selected: People
decide whether to behave dishonestly or honestly. As a result, the
cheater’s high we observe in Study 2 could reflect self-selection.
We address this concern in Study 3 with a novel experimental
approach for studying ethical decision making. In this study, we
pair each participant with a confederate and have them work
together and share a joint reward. The confederate partners either
do or do not over-report their joint performance. Thus, in this
study, we investigate the “cheater’s high” for participants ran-
domly assigned to either a cheating or no-cheating condition. We
also use a very different sample population in a different country.

Method
Participants. We recruited 47 participants (33% male;
Mage ⫽ 27, SD ⫽ 9.0; 62.5% students) at a university in London,
England. Participants earned £10 ($15.50) for participating in the

5
One possible concern is that the negative affect items in the PANAS
might not represent the types of negative affect associated cheating. To
address this concern, we conducted a separate pilot study on Amazon
MTurk with 103 participants (61% male, Mage ⫽ 30, SD ⫽ 10.7 years) who
were paid $.50. We asked each participant to identify affect items he or she
thought someone would experience after cheating. Re-running our analyses
Figure 1. Study 2 means (⫾1 SE) of positive affect before and after the for Studies 2, 3, 4, and 5 with the subset of negative affect items that a
cheating opportunity for cheating and honest participants. SE ⫽ standard majority of participants identified as relevant (guilty, ashamed, jittery, and
error. nervous), we found identical results to those we report.
538 RUEDY, MOORE, GINO, AND SCHWEITZER

study and had the chance to earn up to an additional £10 in bonus

based on their performance.
Procedure. We randomly assigned participants to one of two
conditions (cheating or no-cheating). In both conditions, partici-
pants completed a problem-solving task that included 20 matrices,
each based on a set of 12 three-digit numbers (adapted from Mazar
et al., 2008). We gave participants 5 min to find two numbers in
each matrix that summed to 10; this duration was not sufficient for
anyone to solve all 20 matrices (Msolved ⫽ 5.0, SD ⫽ 3.2,
Maxsolved ⫽ 13.0).
Participants always completed the matrix task in a room with a
confederate. When the experimenter brought the confederate and
the participant to the room, she left them with a timer set for 5 min
and informed them that when the timer went off, they should
switch sheets and grade each other’s worksheet. We also informed
participants that they would each earn a bonus based upon the total
number of matrices they and their (confederate) partner scored on
the task, £0.25 ($0.39) for each correct answer, up to a total of £10.
When the experimenter returned to the room, she first asked the
participant to report the confederate’s score. The confederate al-
ways answered five matrices correctly, and every participant ac-
curately reported five. Then, the experimenter asked the confed-
erate to report the participant’s score. Figure 2. Study 3 means (⫾1 SE) of positive affect pre- and post-study,
In the no-cheating condition, the confederate accurately reported by cheating condition.
the participant’s score. In the cheating condition, the confederate
always added five to the total number of matrices that the partic-
ipant had legitimately solved. This over-statement corresponded to effects testing, we find that cheaters experienced marginally higher
an additional payment of £1.25 ($1.94) that participants had not positive affect than non-cheaters at Time 2, F(1, 43) ⫽ 3.60, p ⫽
legitimately earned. None of the participants in the cheating con- .064, ␻2 ⫽ .053, but not at Time 1, F(1, 43) ⫽ 0.24, p ⫽ .63.
dition corrected the confederate’s over-reporting of their score, and Criterion 3. We conducted a mixed ANOVA with cheating
none reported suspicion that the purpose of the study was to test as the between-subjects factor, affective valence, and Time as
their reactions to the confederate’s behavior. within-subjects factors, and actual performance on the matrix
We collected affect measures (using the abbreviated PANAS; task as a covariate. The results revealed a marginally significant
Mackinnon et al., 1999) upon participants’ arrival (Time 1, ␣PA ⫽ three-way interaction between cheating, affect valence, and
.81; ␣NA ⫽ .85) and immediately after the cheating manipulation time, F(1, 43) ⫽ 2.95, p ⫽ .093, ␻2 ⫽ .040. At Time 1, the
(Time 2, ␣PA ⫽ .85; ␣NA ⫽ .77). simple interaction of cheating and affective valence was not
significant, F(1, 43) ⫽ 0.01, p ⫽ .92, as we expected. At Time
2, however, it approached marginal significance, F(1, 43) ⫽
Results and Discussion
2.54, p ⫽ .12. Examining the simple effects at Time 2, the
One participant took their matrix task with them so we do not difference between positive and negative affect was significant
have a measure of their true performance. We report analyses for for both cheaters (M ⫽ 1.70, SE ⫽ 0.21); F(1, 43) ⫽ 68.99, p ⬍
the remaining 46 participants. We expected participants in the .001, ␻2 ⫽ .596, and non-cheaters (M ⫽ 1.22, SE ⫽ 0.21); F(1,
cheating condition, compared to those in the no-cheating condi- 43) ⫽ 32.55, p ⬍ .001, ␻2 ⫽ .407, but again was (non-
tion, to experience the cheater’s high and report a boost in positive significantly) larger for cheaters than non-cheaters. As in Study
affect. 2, traditional significance testing did not support Criterion 3 of
Criterion 1. We conducted a mixed ANOVA with positive the cheater’s high hypothesis, but the results were in the pre-
affect at Time 1 and 2 as the within-subjects factor, condition as dicted direction.
the between-subjects factor, and actual performance on the matrix In Study 3, we introduce a novel approach for studying unethical
task as a covariate. This analysis revealed a marginally significant behavior with a method that randomly assigns participants to a
two-way interaction, F(1, 43) ⫽ 3.03, p ⫽ .089, ␻2 ⫽ .042, cheating or no-cheating condition. By assigning participants to a
suggesting that there were differences in the changes in positive cheating condition, results from Study 3 document a cheater’s high
affect across conditions. Simple effects testing confirmed that that cannot be explained by self-selection. Although our approach
those in the all-cheating condition experienced a significant in- in this study addresses the problem of self-selection, we acknowl-
crease in positive affect, F(1, 43) ⫽ 4.26, p ⫽ .045, ␻2 ⫽ .066, and edge that this method introduces two issues that may influence our
that those in the no-cheating condition did not, F(1, 43) ⫽ 0.22, results. First, participants in the all-cheating condition may have
p ⫽ .64 (see Figure 2). felt morally superior to the confederate who misreported their
Criterion 2. We next examine whether or not cheaters expe- score. This social comparison may have increased positive affect
rience similar positive affect to non-cheaters at Time 1 but greater and amplified our effect. However, this design also may have
positive affect than non-cheaters at Time 2. In additional simple created social pressure for participants to conform with the con-
 CHEATER’S HIGH? 539

federate. Succumbing to social pressure is an aversive experience Problem-solving task. After answering the short emotion
(Brehm, 1966), which would have dampened our effects in the questionnaire, participants received instructions for the problem-
all-cheating condition. As a result, characteristics of the experi- solving task. The task required them to answer 20 math and logic
mental design may have both amplified as well as dampened the problems. Participants had 25 s to read each problem and choose
cheater’s high. We therefore sought to create an additional exper- one of five possible answers. A timer on the computer screen
imental paradigm that would allow us to address self-selection helped them keep track of time on each problem. Participants were
issues in a different way, as well as address concerns that unearned also told that, “for each problem, when time is up, you will be
financial rewards are driving our result. automatically transferred to the next problem. If you select one of
the answers before time is up in any given round, you will
automatically be moved to the next problem. Note that wrong
Study 4: Effects of Incentives
answers will not harm your score.” After these instructions, par-
In Study 4, we disentangle the effects of financial rewards from ticipants were given an example so that they would become
the effects of cheating. In this study, participants have no financial familiar with the task. Next, participants read the following instruc-
incentives for cheating. We test whether or not participants expe- tions, “NOTE that this problem solving task is an INTELLIGENCE
rience a boost in positive affect after cheating even in the absence TEST that is designed to assess your cognitive abilities. High
of financial incentives. Study 4 also addresses concerns about scores on this test correlate with overall intelligence and predict
self-selection, using a different approach than the one we used in personal and professional success. The computer will record the
Study 3. In Study 4, we randomly assign participants to either an number of problems you solve correctly and will show you your
experimental condition, where cheating is possible, or a control performance at the end of the task.”
condition, where cheating is not possible. This design allows us to In the control condition, participants then moved to the first
compare participants who chose to cheat with participants who did problem. In the experimental condition, they received the follow-
not have the option to cheat. In this way, we can address the ing additional piece of information before moving to the first
concern that choosing not to cheat might have consequences for problem: “During the problem solving task, some browsers will
positive affect and that this effect might account for the difference display the correct answer on the screen if you click on the button
in positive affect between cheaters and non-cheaters. We also test labeled ‘CORRECT ANSWER.’ Just disregard that button if you
whether a psychological sense of relief after not being caught for see it on the screen and solve the problems on your own.”
cheating may be driving the cheater’s high. After completing the problem-solving task, participants re-
sponded to the second set of affect measures, reported their level
of relief and then answered a few demographic questions. They
Method
were then shown their overall score on the problem-solving task.
Participants and design. We recruited 161 students (69% Finally, they answered two questions assessing their level of
male; Mage ⫽ 20.65, SD ⫽ 2.41) from a large Southeastern suspicion. In particular, they indicated what they thought the study
university and paid them $8 for their participation, with no oppor- was about and reported any additional comments they had.
tunity to win additional money depending on performance. Partic-
ipants were randomly assigned to one of two conditions: a control
Results and Discussion
condition and an experimental condition where they were given
the opportunity to cheat. Suspicion. Analysis of participants’ answers to the question
Procedure. We told participants that the study included a assessing potential suspicion revealed that no participants thought
variety of tasks that they would complete sequentially. The in- the study was testing cheating or the link between cheating and
structions informed participants that we were interested in exam- emotions.
ining different problem solving techniques and their link to per- Proportion of cheating participants. In the experimental
sonality. In addition, the instructions informed participants that we condition, 68% of participants (52 out of 77) cheated in at least one
were also interested in studying emotions across short periods of round. In Studies 2 and 3, we compared participants who cheated
time and that they would complete short surveys asking them to participants who did not cheat. In this design, we compare
about their feelings more than once throughout the study. We participants who cheated in the experimental condition to partici-
collected affect measures using the full PANAS (Watson et al., pants in the control condition, which afforded no opportunity to
1988) upon arrival (Time 1, ␣PA ⫽ .89; ␣NA ⫽ .79) and immedi- cheat. We note, however, that the nature and significance of our
ately after the cheating manipulation (Time 2, ␣PA ⫽ .88; ␣NA ⫽ results do not change if we conduct analyses comparing cheaters in
.87). our experimental condition to control participants and non-
In addition to the items measuring positive and negative affect, cheaters in our experimental condition (as part of the same group).
we also included an item measuring relief (“relieved”) both at Criterion 1. We conducted a mixed ANOVA with positive
Time 1 and at Time 2. At Time 2, we also asked participants to affect at Time 1 and Time 2 as the within-subjects factor, and
indicate their agreement with three items measuring relief (“I feel cheating versus control condition as the between-subjects factor. In
as if a weight has been lifted off my shoulders,” “I can breathe this study, we did not control for task performance, because unlike
easily now,” and “I feel like I have just experienced a release of Studies 2 and 3, participants in this study did not know how they
tension,” ␣ ⫽ .91). We included these relief measures to rule out performed on the problem-solving task until the end of the study.
an alternative explanation for the effects of cheating on positive (However, our results do not change if we control for perfor-
affect we observed in Studies 2 and 3: that cheaters experience a mance.) This analysis revealed a significant two-way interaction,
sense of relief after not being caught for their unethical behavior. F(1, 134) ⫽ 7.73, p ⫽ .006, ␻2 ⫽ .047, suggesting that there were
540 RUEDY, MOORE, GINO, AND SCHWEITZER

differences in the changes in positive affect across conditions. of relief (M ⫽ 2.78, SD ⫽ 0.95) as did participants in the control
Simple effects testing confirmed that those who cheated signifi- condition who did not have the opportunity to cheat (M ⫽ 2.65,
cantly increased in positive affect, F(1, 134) ⫽ 8.92, p ⫽ .003, SD ⫽ 1.02), t(134) ⫽ 0.75, p ⫽ .46.
␻2 ⫽ .055, whereas those in the no-cheating condition did not, These findings provide further support for a link between cheat-
F(1, 134) ⫽ 0.49, p ⫽ .49 (see Figure 3). ing and positive affect. First, these results demonstrate that cheat-
Criterion 2. We next examine whether or not cheaters report ing triggers an emotional “high” independent of its associated
more positive affect than non-cheaters at Time 1 and Time 2. financial rewards. Second, the use of the control condition as the
Consistent with the cheater’s high, simple effects tests reveal that comparison group demonstrates that the differences in affect are
cheaters significantly differed from non-cheaters in positive affect driven by a positive affect boost in cheaters, rather than by affec-
at Time 2, F(1, 134) ⫽ 9.51, p ⫽ .002, ␻2 ⫽ .059, but not at Time tive reactions for those who chose not to cheat when they have the
1, F(1, 134) ⫽ 0.17, p ⫽ .68. opportunity. Third, these results indicate that the cheater’s high is
Criterion 3. We conducted a mixed ANOVA with cheating as not driven by a sense of relief after not getting caught cheating.
the between-subjects factor, and affective valence and time as
within-subjects factors. The results revealed a significant three-
Study 5: Getting Away With It
way interaction between cheating, affect valence, and time, F(1,
134) ⫽ 6.48, p ⫽ .012, ␻2 ⫽ .039. As we expect, the simple In Study 5, we investigate the underlying mechanism of the
interaction of cheating and affective valence was not significant at cheater’s high. In this study, we consider the possibility that the
Time 1, F(1, 134) ⫽ 0.78, p ⫽ .38. However, consistent with our cheater’s high is driven by the thrill of getting away with unethical
predictions, it was significant at Time 2, F(1, 134) ⫽ 11.69, p ⫽ behavior. To test for this possibility, we explicitly acknowledge
.001, ␻2 ⫽ .073. Examining the simple effects at Time 2, the the potential for cheating in our study to some of our participants.
difference between positive and negative affect was significant for In prior studies of unethical behavior, experimenters have not
both cheaters (M ⫽ 0.98, SE ⫽ 0.11); F(1, 134) ⫽ 75.36, p ⬍ .001, explicitly confronted cheaters with the acknowledgment that over-
␻2 ⫽ .353, and non-cheaters (M ⫽ 0.49, SE ⫽ 0.09); F(1, 134) ⫽ reporting performance in the experiment is possible and that it
30.30, p ⬍ .001, ␻2 ⫽ .177, but was larger for cheaters than constitutes cheating. Schweitzer and Hsee (2002) and Shalvi,
non-cheaters. This pattern of results is consistent with Criterion 3 Dana, Handgraaf, and De Dreu (2011) identify available justifica-
of the cheater’s high hypothesis. tions as an essential antecedent to lying, and a number of other
Relief. We then conducted a mixed ANOVA with the single researchers suggest that individuals strive to preserve their moral
item measure of relief at Time 1 and Time 2 as the within-subjects self-concept while engaging in unethical behavior through self-
factor, and cheating condition as the between-subjects factor. This deception or moral rationalization (Gino, Ayal, & Ariely, 2009;
analysis revealed no significant effects (all ps ⬎ .22). Next, we Mazar et al., 2008; Mead et al., 2009).
examined the three-item measure of relief we collected after par- If our effects are driven by participants’ self-deception or moral
ticipants had the opportunity to cheat. Participants who cheated in rationalizations, then confronting participants with the experiment-
the experimental condition reported experiencing the same amount er’s recognition that over-reporting performance would constitute
cheating should prevent the possibility of preserving a positive
moral self-concept, increasing guilt and negative affect and damp-
ening positive affect. If, however, the cheater’s high is triggered by
the thrill of getting away with unethical behavior, increases in
positive affect should be amplified when participants are made
aware that they did “get away with” cheating, engendering a
feeling of self-satisfaction. In order to investigate this second
explanation, we measured self-satisfaction as a potential mediator
of the cheater’s high.
Thrill is very different from relief, which we tested in Study 4.
Relief is triggered by avoiding an aversive experience (Sweeny &
Vohs, 2012) and reflects a shift from a state of high negative
affect, defined as “subjective distress and unpleasurable engage-
ment,” to one of lower negative affect, defined as “a state of
calmness and serenity” (Watson et al., 1988, p. 1063). In contrast,
the thrill of the cheater’s high is likely to be associated with
approach-oriented emotions, such as the positive affect associated
with “getting away with” cheating (Carver, 2001; Watson, Wiese,
Vaidya, & Tellegen, 1999). For these reasons, we conceptualize
the “thrill” of getting away with something as distinct from relief.

Method
Participants. Two hundred and five (46% male; Mage ⫽ 34,
Figure 3. Study 4 means (⫾1 SE) of positive affect pre- and post-study, SD ⫽ 11.6) participants were recruited from MTurk to complete an
by condition. online study. Participants earned $0.50 for participating in the
 CHEATER’S HIGH? 541

study and had the chance to earn up to an additional $4.50 in felt clever, capable, accomplished, satisfied, and superior (mea-
bonuses based on their reported performance. sured on the same 5-point scale as the PANAS, from “not at all”
Design and procedure. The study was presented as a test of to “extremely”). We averaged the responses to these five items to
performance under time pressure. We randomly assigned partici- create a measure of feeling clever or self-satisfied (␣ ⫽ .93).
pants to one of two conditions (“reality check” vs. control). Par-
ticipants in both conditions completed the full PANAS (Watson et
Results and Discussion
al., 1988, PA: ␣ ⫽ .92, NA: ␣ ⫽ .89) and then completed an
anagram task (adapted from Wiltermuth, 2011), designed to detect We excluded two participants for failing an attention filter that
cheating in an online environment. The anagram task presents preceded our manipulation. This question listed a series of activ-
participants with a list of nine jumbled words and instructions that ities that people might enjoy on a daily basis, but the instructions
read, for the questions required that they only check the activity “cook-
ing” from a list of 14 activities. We excluded an additional four
Each jumble has only one correct answer, and the jumbles must be who participated using duplicate IP addresses, though the nature
solved in order. In other words, if you successfully unscramble the
and significance of our results do not change if we include them.
first three word jumbles but not the fourth, you will earn a bonus only
Fifty-two percent of participants reported they successfully un-
for the first three— even if you also successfully unscramble the fifth,
sixth, and seventh word jumbles. scrambled the third anagram (or more), reporting on average
having solved 4.61 of the anagrams (SD ⫽ 2.80).
Participants were asked to solve the jumbles in order, not Self-rated reliability. Participants in the “reality check” con-
proceeding to the next one unless the prior one had been solved. dition were asked to rate the reliability of their results. Within this
They were also instructed to “use only your own mind, and no condition, cheaters rated their responses as less reliable (M ⫽ 4.67,
other tools, to solve the jumbles” so that it would be clear that SD ⫽ 0.58) than did honest participants (M ⫽ 4.95, SD ⫽ 0.31),
using online anagram unscramblers was not permitted. t(97) ⫽ 2.91, p ⫽ .004.
The third jumble in the list was “UNAAGT,” for which the only If our effect is driven by the ability of individuals to deceive
solution is “TAGUAN,” a species of flying squirrel. Previous themselves about the nature of their actions, the “cheater’s high”
pretesting has confirmed that the likelihood of correctly identify- should be moderated when participants are confronted with an
ing this solution is miniscule (Wiltermuth, 2011), and even the explicit acknowledgment that the experimenters are aware that
largest online web-based anagram solving programs do not return they could be cheating. However, if participants revel in the fact
“TAGUAN” as an answer. that they are cheating and “getting away with it,” as we predict,
Participants had 10 min to solve as many of the jumbles as then our manipulation will not curtail the cheater’s high. In fact, it
possible in the order in which they were presented. Next, they should highlight the fact that they have transgressed, boosting the
turned to a page on which they reported how many jumbles they effect and producing a greater increase in positive affect after
correctly unscrambled, again being reminded that they should only cheating.
report the number they had been able to consecutively solve. We Criterion 1. Since we have no way to accurately measure
interpret any answer of 3 or more to be indicative of cheating on performance in this study, we could not include it as a covariate in
the task. these analyses. Consistent with the results from Studies 2, 3, and
Experimental manipulation. In this experiment, we manip- 4, a mixed ANOVA with positive affect at Time 1 and 2 as the
ulated whether participants were confronted with the experiment- within-subjects factor and condition and cheating as the between-
er’s explicit acknowledgment that participants could misrepresent subjects factors revealed a significant two-way interaction, F(1,
their performance in the anagram task. After reporting their an- 195) ⫽ 39.00, p ⬍ .001, ␻2 ⫽ .160, indicating that there were
swers, all participants read the following: differences in the changes in positive affect between cheaters and
non-cheaters. Simple effects testing confirmed that cheaters expe-
Thank you for your participation in this series of tasks. You have one rienced a significant increase in positive affect from Time 1 to
more task to complete before the end. Your data is important to us as Time 2, F(1, 195) ⫽ 4.51, p ⫽ .035, ␻2 ⫽ .017, while non-cheaters
we are trying to understand performance under time pressure. experienced a significant decrease in positive affect, F(1, 195) ⫽
In the “reality check” condition, participants also read: 43.63, p ⬍ .001, ␻2 ⫽ .176 (see Figure 4).
Criterion 2. In addition, simple effects testing revealed that
We realize we cannot check your answers (i.e., whether you actually cheaters experienced significantly more positive affect than non-
solved correctly the jumbles you told us you solved) and that you may cheaters at Time 2, F(1, 195) ⫽ 29.32, p ⬍ .001, ␻2 ⫽ .125, but
have cheated on this task by inflating your performance. We hope you not at Time 1, F(1, 195) ⫽ 1.99, p ⫽ .16.
reported your answers honestly. Criterion 3. We conducted a mixed ANOVA with cheating
To emphasize the fact that their answers would not be reliable if (yes, no) and condition (reality check vs. not) as the between-
they had misrepresented their true performance, we also asked subjects factors and affective valence and time as within-subjects
participants in this condition, “To what extent can we rely on your factors. The results revealed a significant three-way interaction
answers?” Responses were provided on a 5-point scale ranging between cheating, affect valence, and time, F(1, 195) ⫽ 47.93, p ⬍
from 1 ⫽ “Not at all” to 5 ⫽ “Completely.” .001, ␻2 ⫽ .191. At Time 1, the simple interaction of affect
Post-manipulation measures. After this manipulation, par- valence and cheating was not significant, F(1, 195) ⫽ 0.55, p ⫽
ticipants completed the PANAS again (PA: ␣ ⫽ .93, NA: ␣ ⫽ .46, but again, it was at Time 2, F(1, 195) ⫽ 31.44, p ⬍ .001, ␻2 ⫽
.79). To assess the extent to which cheating may provide a feeling .133. Examining the simple effects at Time 2, the difference
of self-satisfaction, we also asked the extent to which participants between positive and negative affect was significant for both
542 RUEDY, MOORE, GINO, AND SCHWEITZER

resents the degree to which participants’ positive affect changes

from baseline. The mediation results indicate a positive indirect
effect of cheating on positive affect at Time 2 through self-
satisfaction, point estimate ⫽ .38; 95% CI ⫽ .25 to .53.
In this study, we reminded participants that over-reporting per-
formance constitutes cheating. This manipulation failed to curb the
cheater’s high, evidence that the cheaters high does not depend on
self-deception. Additionally, though the manipulation did not elicit
significantly greater positive affect compared to cheaters in the
control condition, it appears to have increased the self-satisfaction
cheaters enjoyed, further evidence that awareness one is cheating
may enhance, rather than dampen, the cheater’s high. We also
found that self-satisfaction mediates the relationship between
cheating and positive affect. These findings suggest that the cheat-
er’s high reflects positive feelings that derive from having “gotten
away” with something, rather than self-deception or the ability to
rationalize unethical behavior.

General Discussion
Existing models of ethical decision making have assumed that
unethical behavior triggers negative affect and that the negative
affective consequences of engaging in unethical behavior promote
ethical decision making. Our findings challenge these assumptions
Figure 4. Study 5 means (⫾1 SE) of cheaters’ and non-cheaters’ positive
and demonstrate that some unethical behaviors not only fail to
affect before and after the cheating opportunity.
trigger negative affect but can in fact trigger positive affect.
In two studies, we find that people fail to forecast how they will
cheaters (M ⫽ 2.03, SE ⫽ 0.10); F(1, 195) ⫽ 419.57, p ⬍ .001, feel after engaging in unethical behaviors. Individuals in one study
␻2 ⫽ .678, and non-cheaters (M ⫽ 1.23, SE ⫽ 0.10); F(1, 195) ⫽ predicted that they would feel higher levels of negative affect and
138.94, p ⬍ .001, ␻2 ⫽ .409, but again was larger for cheaters than lower levels of positive affect after engaging in unethical behavior,
non-cheaters. This pattern of results is also consistent with Crite- and individuals in another study predicted higher levels of both
rion 3 of the cheater’s high hypothesis. positive and negative affect, representing emotional ambivalence.
Self-satisfaction. We next examined self-satisfaction. We
conducted an ANOVA with self-satisfaction at Time 2 as the
dependent variable and the two between-subjects factors (reality
check vs. control; cheater vs. not). We identify a main effect of
cheating on self-satisfaction at Time 2, F(1, 195) ⫽ 105.71, p ⬍
.001, ␻2 ⫽ .345, such that those who cheated reported higher
levels of self-satisfaction (M ⫽ 3.81, SD ⫽ 1.25) than those who
did not (M ⫽ 2.31, SD ⫽ 1.07), controlling for condition. We
found no interaction between cheating and condition, F(1, 195) ⫽
1.34, p ⫽ .25. However, the main effect for condition approached
statistical significance, F(1, 195) ⫽ 3.38, p ⫽ .068, ␻2 ⫽ .012, and
a post hoc contrast effect reveals that cheaters in the reality-check
condition were more self-satisfied than cheaters in the control
condition, F(1, 195) ⫽ 4.71, p ⫽ .031, ␻2 ⫽ .018. This contrast
suggests that the cheater’s high may be triggered by the thrill of
getting away with something rather than by self-deception or
moral rationalization. In other words, for cheaters, greater aware-
ness of their own cheating appears to increase self-satisfaction and
fuel rather than curb the cheater’s high (see Figure 5).
We tested whether self-satisfaction mediated the relationship
between cheating and positive affect using Preacher and Hayes’s
(2008) bootstrapping technique. We ran a model with positive
affect at Time 2 as the dependent variable, cheating (vs. honest) as
the independent variable, and self-satisfaction as the mediator,
including positive affect at Time 1 and condition (reality-check vs.
control) as covariates. By including positive affect at Time 1 as a Figure 5. Study 5 means (⫾1 SE) of cheaters’ and non-cheaters’ self-
covariate, the dependent variable (positive affect at Time 2) rep- satisfaction after the cheating opportunity, by reality check condition.
 CHEATER’S HIGH? 543

In our studies with actual behavior, we demonstrate that these ence associated with meeting immediate urges (Loewenstein,
beliefs reflect an error in prediction. Interestingly, this mispredic- 1996).
tion persists even when individuals forecast how others will feel Our findings also speak to the literature on affective forecasting
after engaging in unethical behavior, suggesting that this predic- (Gilbert et al., 2002; Wilson & Gilbert, 2003). Participants in our
tion is not motivated by self-presentational concerns. studies mispredicted the affective consequences of unethical be-
Across four additional studies that afforded participants the havior. This finding adds to our understanding about how predict-
opportunity to cheat in a setting with no obvious victim, we find ing the affective consequences of decisions may be more compli-
that participants who cheated experienced more positive affect cated than we intuitively believe (Barsade & Gibson, 2007;
(and no more negative affect) than those who did not cheat. We Loewenstein & Lerner, 2003; Schwarz, 2000). Future work should
term this effect the cheater’s high. Across these studies, we inves- explore affective forecasts related to both unethical and ethical
tigate several alternative explanations for the cheater’s high. Spe- decision making.
cifically, we consider and rule out self-selection (only individuals Finally, our results deepen our understanding of the conse-
who choose to cheat experience the cheater’s high), financial quences of unethical behavior. Research on ethical behavior has
incentives (the cheater’s high is driven by an undeserved financial largely focused on identifying predictors rather than outcomes
windfall), and rationalization (the cheater’s high is experienced by (Tenbrunsel & Smith-Crowe, 2008). Our findings underscore the
individuals who do not acknowledge their behavior as cheating). importance of examining consequences of unethical behavior in
Instead, our findings suggest that the cheater’s high reflects the moral psychology.
thrill of having gotten away with cheating.
We note that we set a high evidentiary bar in testing for the
Limitations
cheater’s high, examining whether data from four different studies
satisfy each of the three criteria we identified. Our results for the One limitation of our studies is our laboratory context. Our
first two criteria were supported across all four studies (excepting experimental designs afforded control over the opportunity to
one test in Study 2 for which p ⫽ .10): Cheaters reported an cheat and clear measurement, but we cannot be sure how partici-
increase in positive affect from Time 1 to Time 2 and reported pants conceptualized the opportunity to cheat in this context. One
higher positive affect than non-cheaters at Time 2. Results from concern is that participants thought that cheating in our experi-
Studies 4 and 5 supported Criterion 3, which required that the mental setting was acceptable, and this may have diminished their
difference between positive and negative affect be greater for negative affective responses. However, our data do not support this
cheaters than for non-cheaters at Time 2, but not at Time 1. Results explanation. In the pilot study we ran to assess whether partici-
from Studies 2 and 3 were directional, but not significant (p ⫽ .12 pants thought cheating in exactly this type of laboratory setting
and p ⫽ .13, respectively). It is possible that the relatively small was unethical (reported with Study 1a), our results revealed that
sample size in Study 3 and specific characteristics of our procedure participants rated this behavior as very unethical. It is also possible
in Study 2 may have contributed to the lack of support for this third that an experiment may allow participants to rationalize their
criterion. Overall, findings from our studies support our thesis that cheating in a way that they would not do outside of an experiment.
victimless cheating can cause people to feel good. Given the pervasiveness of unethical behavior in everyday life, we
believe that rationalizing unethical behavior is common both in-
side and outside the controlled setting of the laboratory. Taken
Theoretical Contributions
together, we believe that even though our results come from a
Existing models of ethical decision making presume that uneth- laboratory setting, our findings may have broad external validity.
ical behavior triggers negative feelings. Our findings not only A second limitation of our studies is that participants may have
challenge this assumption but also highlight the importance of reported low negative affect because they believed that reporting
incorporating emotion more broadly in ethical decision making high levels of negative affect would reveal that they had cheated,
research (e.g., Gaudine & Thorne, 2001; Schweitzer & Gibson, either to themselves or the experimenter. Two factors suggest this
2008; Warren & Smith-Crowe, 2008). In addition to considering may not be the case. First, the opportunity to cheat was not explicit
the material benefits and risks related to unethical behavior, the in any of the studies except for Study 5, and participants’ post-
psychic costs and benefits should be considered. The cheater’s experimental remarks across the studies did not reveal any suspi-
high we document may also help to explain puzzling findings, such cion that the experiment was studying cheating behavior. Second,
as the pervasiveness of low stakes unethical behavior (Mazar et al., in Study 5, cheaters in the explicit condition were told that cheat-
2008). The emotional boost conferred by cheating may be one ing was possible. Cheaters in this condition reported no less
reason people are motivated to cheat even when the financial negative affect than those in the control condition.
payoff is small. Third, our studies revealed that cheating led to some compli-
Our results also contribute to research on the temporal nature of cated effects in terms of negative affect. In Studies 2 and 4,
morality (Tenbrunsel et al., 2010). Though the “should” self is negative affect increased for both cheaters and non-cheaters, and in
likely to inform predictions about feelings after engaging in un- Study 5, negative affect increased for non-cheaters. Our designs
ethical behavior, in the moment when people make ethical deci- were unable to distinguish whether participants in these studies
sions, the “want” self may dominate the “should” self (Bazerman may have experienced an increase in negative affect due to some
et al., 1998; Tenbrunsel et al., 2010). People may succumb to aspect of the experiment (e.g., stress from engaging in a timed
immediate temptation (Mead et al., 2009), allowing their self- task) or due to reactions to deciding to cheat or not to cheat.
interested impulses to prevail over their larger ideals (Loewen- However, the increase in negative affect for control participants in
stein, 1996). In addition, individuals may enjoy a hedonic experi- Study 4 who had no opportunity to cheat suggests that aspects of
544 RUEDY, MOORE, GINO, AND SCHWEITZER

some of our tasks unrelated to cheating may have increased neg- ity of the act. Similarly, individual differences may moderate the
ative affect. In addition, the criterion for which we found the most cheater’s high. For instance, an individual’s level of moral identity
equivocal empirical support was Criterion 3, the only one that (Aquino & Reed, 2002) or prosocial orientation (Van Lange, 1999)
implicated the role of positive and negative affect jointly as part of might decrease the effect, whereas other attributes such as moral
the cheater’s high. These results suggest that future research disengagement, Machiavellianism, and psychopathy (McIlwain et
should explore more fully how positive and negative affect are al., 2012; Moore, Detert, Treviño, Baker, & Mayer, 2012; Wil-
jointly affected by the decision to cheat, or indeed the decision not liams, Nathanson, & Paulhus, 2010) may exacerbate it.
to cheat. Future work should also examine whether the positive affect
people experience after cheating may motivate them to continue
cheating in the future. A growing body of research in behavioral
Additional Directions for Future Research
ethics has identified a number of psychological processes that
The cheater’s high is likely to be moderated by contextual allow individuals to easily “stumble into bad behavior” (Bazerman
factors. Although cheating was pervasive in our studies, the effect & Tenbrunsel, 2011). For instance, individuals are more likely to
sizes of the cheater’s high varied across our studies. One possible behave unethically when they have recently had an opportunity to
explanation is that the context in which the studies took place assert what good people they are (Monin & Miller, 2001), when
enhanced or mitigated the high. The size of the cheater’s high is the offense is trivial enough that the transgressor can maintain a
larger in Studies 2 and 5 than it is in Studies 3 and 4. In Study 3, positive self-concept (Mazar et al., 2008), when people become
the effect may be confounded with social pressure, and in Study 4, acclimated to unethical behavior slowly (Gino & Bazerman,
the effect is not accompanied by a related boost in undeserved 2009), when they are motivated by goals (Schweitzer, Ordóñez, &
rewards, both of which may have dampened the cheater’s high. In Douma, 2004), when tension about committing the offense is
contrast, Study 5 took place online, an inherently anonymous misattributed to another source (Dienstbier, 1984; Dienstbier &
environment. Anonymity can amplify unethical behavior (Diener, Hunter, 1971), when the context is ambiguous (Schweitzer &
1976; Diener, Fraser, & Beaman, 1976; Zimbardo, 1970) and may Hsee, 2002), or when people can use others to help establish a
have also increased the positive affect associated with cheating in justification for their behavior (Gino, Gu, & Zhong, 2009). Our
this study. Certainly, conditions that amplify or dampen the cheat- results suggest that the positive affect boost that accompanies these
er’s high represent one important avenue for future work to ex- (often unintended) behaviors may reinforce them, making future
plore. unethical behavior more likely. Future studies could examine
If people derive psychological benefits from outsmarting others, whether initial unethical behavior promotes subsequent unethical
overcoming challenges, and solving puzzles, some contexts may behavior as a result of a cheater’s high.
promote unethical behavior more than others. One such context is In our studies, we explored positive and negative affect. In
online security. In addition to potential material gains, hackers are future work, it would be interesting to explore specific emotions.
often motivated by solving puzzles and earning bragging rights For example, cheaters may experience guilt and anxiety or a
(Richmond, 2011, August 21). This is an especially vexing issue mixture of emotions in anticipation of, during, and after cheating.
given that if outsmarting the system is likely part of the appeal, In addition, individuals who recognize opportunities to cheat but
complex control systems designed to curb this unethical behavior refrain from cheating may also experience specific emotions. For
may also promote it. Additionally, as online fraud becomes more instance, deciding not to cheat may result in regret or feeling like
sophisticated and requires the coordinated efforts of larger groups a “sucker.” Though we were not able to disentangle these effects
of people (Moore, Clayton, & Anderson, 2009), the experience of with the current designs, the emotional impact of being aware of
a cheater’s high may involve social emotions as well. Future and yet not taking an opportunity to cheat is an avenue that merits
research should investigate how the cheater’s high scales to further exploration.
groups. Quite possibly, when groups of people coordinate an effort Finally, future work should also delve deeper into the temporal
to cheat the system, it could exacerbate the cheater’s high by aspects of the cheater’s high. Though our main focus here was on
diffusing responsibility for negative outcomes and building a sense the positive affective consequences of unethical behavior, our
of camaraderie from cheating together. studies provide some evidence that there may be emotional am-
Fortunately, our findings do not suggest that all unethical be- bivalence associated with cheating (or, indeed, perhaps the deci-
havior will lead to positive affect. The cheater’s high may be sion to avoid cheating as well). One possibility is that participants
restricted to certain types of unethical behavior. The type of feel different emotions in the moment of cheating to those that
unethical behavior we examine involves actions that are clearly emerge when we measured them, a few minutes later. It would be
understood as unethical (as our pre-test findings show), but they do worthwhile to explore more immediate measures of emotion, dur-
not involve direct harm to a salient victim. We only examined ing the act itself, to see if the emotional reactions change.
cheating by over-reporting one’s performance. While this is an It would also be interesting to investigate the long-term affective
important type of unethical behavior—particularly given the rash consequences of unethical behavior. In the short term, the “want”
of recent academic cheating scandals (Pérez-Peña, 2013; Yee, self, driven by impulsive behavior is less motivated to behave
2012)— unethical behavior that more clearly involves harming a ethically (Tenbrunsel et al., 2010) and more likely to experience
salient other may be less likely to trigger a cheater’s high. the cheater’s high. As individuals transition to the “should” self,
Other contextual factors may also moderate the cheater’s high. over time and perhaps through self-reflection, they may become
For example, the cheater’s high may be mitigated by moral inten- more likely to attend to ethical considerations and feel differently
sity (Jones, 1991), such as the magnitude of the consequences of about their actions. Perhaps prompting individuals to ruminate
unethical behavior and the social consensus regarding the ethical- about their actions may increase the negative affect that individ-
 CHEATER’S HIGH? 545

uals experience from engaging in unethical behavior. This possi- Brock, T. C., & Buss, A. H. (1964). Effects of justification for aggression
bility is consistent with prior work that has linked unethical be- and communication with the victim on post-aggression dissonance.
havior and negative affect in recall studies (e.g., Baumeister et al., Journal of Abnormal and Social Psychology, 68, 403– 412. doi:10.1037/
1995; Tracy & Robins, 2006). h0042571
Buss, A. H., & Brock, T. C. (1963). Repression and guilt in relation to
aggression. Journal of Abnormal and Social Psychology, 66, 345–350.
Conclusion doi:10.1037/h0043707
Across six experiments, we find that although people expect to Byrne, M. (2003). Trauma reactions in the offender. International Journal
of Forensic Psychology, 1, 59 –70.
experience negative affect after over-reporting their performance
Cacioppo, J. T., Petty, R. E., Feinstein, J. A., & Jarvis, B. G. (1996).
on cognitive tasks, they actually experience a boost in positive
Dispositional differences in cognitive motivation: The life and times of
affect. These findings challenge existing models of ethical decision
individuals varying in need for cognition. Psychological Bulletin, 119,
making and offer cause for concern. Many ethical decisions are 197–253. doi:10.1037/0033-2909.119.2.197
made privately and are difficult to monitor. Individuals who rec- Callahan, D. (2004). The cheating culture: Why more Americans are doing
ognize, perhaps from experience, that they can derive both mate- wrong to get ahead. New York, NY: Harcourt.
rial and psychological rewards from engaging in unethical behav- Carver, C. S. (2001). Affect and the functional bases of behavior: On the
ior may be powerfully motivated to behave unethically. It is dimensional structure of affective experience. Personality and Social
imperative that we develop our understanding of how emotions Psychology Review, 5, 345–356. doi:10.1207/S15327957PSPR0504_4
influence our moral behavior, how our moral behavior influences Carver, C. S., & Scheier, M. F. (1990). Origins and functions of positive
our emotions, and how people expect these relationships to work. and negative affect: A control-process view. Psychological Review, 97,
19 –35. doi:10.1037/0033-295X.97.1.19
Chaiken, S., Giner-Sorolla, R., & Chen, S. (1996). Beyond accuracy:
References Defense and impression motives in heuristic and systematic information
Abagnale, F. W., & Redding, S. (2000). Catch me if you can: The true story processing. In P. M. Gollwitzer & J. A. Bargh (Ed.), The psychology of
of a real fake. New York, NY: Broadway Books. action: Linking cognition and motivation to behavior (pp. 553–578).
Adler, A. (1930). Individual psychology. Oxford, United Kingdom: Clark New York, NY: Guilford Press.
University Press. DePalma, M. T., Madey, S. F., & Bornschein, S. (1995). Individual
Alberts, H. J. E. M., Mulkens, S., Smeets, M., & Thewissen, R. (2010). differences and cheating behavior: Guilt and cheating in competitive
Coping with food cravings. Investigating the potential of a mindfulness- situations. Personality and Individual Differences, 18, 761–769. doi:
based intervention. Appetite, 55, 160 –163. doi:10.1016/j.appet.2010.05 10.1016/0191-8869(94)00208-A
.044 DePaulo, B. M., Lindsay, J. J., Malone, B. E., Muhlenbruck, L., Charlton,
Alicke, M. D. (2012). Self-injuries, harmless wrongdoing, and morality. K., & Cooper, H. (2003). Cues to deception. Psychological Bulletin,
Psychological Inquiry, 23, 125–128. doi:10.1080/1047840X.2012 129, 74 –118. doi:10.1037/0033-2909.129.1.74
.666720 Diekmann, K. A., Tenbrunsel, A. E., & Galinsky, A. D. (2003). From
Allport, G. W. (1955). Becoming: Basic considerations for a psychology of self-prediction to self-defeat: Behavioral forecasting, self-fulfilling
personality. New Haven, CT: Yale University Press. prophecies, and the effect of competitive expectations. Journal of Per-
Aquino, K., & Reed, A. (2002). The self-importance of moral identity. sonality and Social Psychology, 85, 672– 683. doi:10.1037/0022-3514
Journal of Personality and Social Psychology, 83, 1423–1440. doi: .85.4.672
10.1037/0022-3514.83.6.1423 Diekmann, K. A., Walker, S. D. S., Galinsky, A. D., & Tenbrunsel, A. E.
Arkes, H. R., Joyner, C. A., Pezzo, M. V., Nash, J. G., Siegel-Jacobs, K., (2012). Double victimization in the workplace: Why observers condemn
& Stone, E. (1994). The psychology of windfall gains. Organizational passive victims of sexual harassment. Organization Science, 98, 579 –
Behavior and Human Decision Processes, 59, 331–347. doi:10.1006/ 592. doi:10.1287/orsc.1120.0753
obhd.1994.1063
Diener, E. (1976). Effects of prior destructive behavior, anonymity, and
Bandura, A. (1990). Mechanisms of moral disengagement. In W. Reich
group presence on deindividuation and aggression. Journal of Person-
(Ed.), Origins of terrorism: Psychologies, ideologies, states of mind (pp.
ality and Social Psychology, 33, 497–507. doi:10.1037/0022-3514.33.5
161–191). New York, NY: Cambridge University Press.
.497
Barsade, S. G., & Gibson, D. E. (2007). Why does affect matter in
Diener, E., Fraser, S. C., & Beaman, A. L. (1976). Effects of deindividu-
organizations? Academy of Management Perspectives, 21, 36 –59. doi:
ation variables on stealing among Halloween trick-or-treaters. Journal of
10.5465/AMP.2007.24286163
Personality and Social Psychology, 33, 178 –183. doi:10.1037/0022-
Baumeister, R. F., Stillwell, A. M., & Heatherton, T. F. (1995). Personal
narratives about guilt: Role in action control and interpersonal relation- 3514.33.2.178
ships. Basic and Applied Social Psychology, 17, 173–198. Dienstbier, R. A. (1984). The role of emotion in moral socialization. In C.
Baumeister, R. F., Vohs, K. D., DeWall, N., & Zhang, L. (2007). How Izard, J. Kagan, & R. B. Zajonc (Eds.), Emotions, cognition, and
emotion shapes behavior: Feedback, anticipation, and reflection, rather behavior (pp. 484 –518). New York, NY: Cambridge University Press.
than direct causation. Personality and Social Psychology Review, 11, Dienstbier, R. A., & Hunter, P. O. (1971). Cheating as a function of the
167–203. doi:10.1177/1088868307301033 labeling of natural arousal. Journal of Personality and Social Psychol-
Bazerman, M. H., & Tenbrunsel, A. E. (2011, April 20). Stumbling into ogy, 17, 208 –213. doi:10.1037/h0030384
bad behavior. The New York Times. Retrieved from http://www.nytimes Doris, J. M. (2002). Lack of character: Personality and moral behavior.
.com/2011/04/21/opinion/21bazerman.html?_r⫽0 Cambridge, United Kingdom: Cambridge University Press.
Bazerman, M. H., Tenbrunsel, A. E., & Wade-Benzoni, K. (1998). Nego- Eisenberg, N. (2000). Emotion, regulation, and moral development. Annual
tiating with yourself and losing: Making decisions with competing Review of Psychology, 51, 665– 697. doi:10.1146/annurev.psych.51.1
internal preferences. The Academy of Management Review, 23, 225–241. .665
Brehm, J. W. (1966). A theory of psychological reactance. New York, NY: Ekman, P. (2001). Telling lies: Clues to deceit in the marketplace, mar-
Academic Press. riage, and politics (3rd ed.). New York, NY: Norton.
546 RUEDY, MOORE, GINO, AND SCHWEITZER

Epley, N., & Dunning, D. (2000). Feeling “holier than thou”: Are self- approach to socialization. In D. A. Goslin (Ed.), Handbook of social-
serving assessments produced by errors in self- or social predictions? ization theory and research (pp. 348 – 480). Chicago, IL: Rand McNally.
Journal of Personality and Social Psychology, 79, 861– 875. doi: Kurtz, J. L., Wilson, T. D., & Gilbert, D. T. (2007). Quantity versus
10.1037/0022-3514.79.6.861 uncertainty: When winning one prize is better than winning two. Journal
Evans, C., Ehlers, A., Mezey, G., & Clark, D. M. (2007). Intrusive of Experimental Social Psychology, 43, 979 –985. doi:10.1016/j.jesp
memories in perpetrators of violent crime: Emotions and cognitions. .2006.10.020
Journal of Consulting and Clinical Psychology, 75, 134 –144. doi: Lewicki, R. J. (1983). Lying and deception: A behavioral model. In M. H.
10.1037/0022-006X.75.1.134 Bazerman & R. J. Lewicki (Eds.), Negotiating in organizations (pp.
Festinger, L. (1957). A theory of cognitive dissonance. Evanston, IL: Row, 68 –90). Beverly Hills, CA: Sage.
Peterson. Loewenstein, G. (1996). Out of control: Visceral influences on behavior.
Fishbach, A. (2009). The function of value in self-regulation. Journal of Organizational Behavior and Human Decision Processes, 65, 272–292.
Consumer Psychology, 19, 129 –133. doi:10.1016/j.jcps.2009.02.005 doi:10.1006/obhd.1996.0028
Gaudine, A., & Thorne, L. (2001). Emotion and ethical decision-making in Loewenstein, G., & Lerner, J. (2003). The role of affect in decision
organizations. Journal of Business Ethics, 31, 175–187. doi:10.1023/A: making. In R. J. Davidson, K. R. Scherer, & H. H. Goldsmith (Eds.),
1010711413444 Handbook of affective science (pp. 619 – 642). Oxford, England: Oxford
Giacalone, R. A., & Promislo, M. D. (2010). Unethical and unwell: University Press.
Decrements in well-being and unethical activity at work. Journal of Mackinnon, A., Jorm, A. F., Christensen, H., Korten, A. E., Jacomb, P. A.,
Business Ethics, 91, 275–297. doi:10.1007/s10551-009-0083-3 & Rodgers, B. (1999). A short form of the Positive and Negative Affect
Gilbert, D. T., Gill, M. J., & Wilson, T. D. (2002). The future is now: Schedule: Evaluation of factorial validity and invariance across demo-
Temporal correction in affective forecasting. Organizational Behavior graphic variables in a community sample. Personality and Individual
and Human Decision Processes, 88, 430 – 444. doi:10.1006/obhd.2001 Differences, 27, 405– 416. doi: 10.1016/S0191-8869(98)00251-7
.2982 Mann, T., & Ward, A. (2001). Forbidden fruit: Does thinking about a
Gino, F., Ayal, S., & Ariely, D. (2009). Contagion and differentiation in prohibited food lead to its consumption? International Journal of Eating
unethical behavior: The effect of one bad apple on the barrel. Psycho- Disorders, 29, 319 –327. doi:10.1002/eat.1025
logical Science, 20, 393–398. doi:10.1111/j.1467-9280.2009.02306.x Mars, G. (1982). Cheats at work: An anthropology of workplace crime.
Gino, F., & Bazerman, M. H. (2009). When misconduct goes unnoticed: London, England: George Allen & Unwin.
The acceptability of gradual erosion in others’ unethical behavior. Jour- Massi, L. L. (2005). Anticipated guilt as behavioral motivation. Human
nal of Experimental Social Psychology, 45, 708 –719. Communication Research, 31, 453– 481. doi:10.1111/j.1468-2958.2005
Gino, F., Gu, J., & Zhong, C.-B. (2009). Contagion or restitution? When .tb00879.x
bad apples can motivate ethical behavior. Journal of Experimental Matza, D., & Sykes, G. M. (1961). Juvenile delinquency and subterranean
Social Psychology, 45, 1299 –1302. doi:10.1016/j.jesp.2009.07.014 values. American Sociological Review, 26, 712–719. doi:10.2307/
Gray, K., Waytz, A., & Young, L. (2012). The moral dyad: A fundamental 2090200
template unifying moral judgment. Psychological Inquiry, 23, 206 –215. Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people:
doi:10.1080/1047840X.2012.686247 A theory of self-concept maintenance. Journal of Marketing Research,
Gray, K., Young, L., & Waytz, A. (2012). Mind perception is the essence 45, 633– 644. doi:10.1509/jmkr.45.6.633
of morality. Psychological Inquiry, 23, 101–124. doi:10.1080/1047840X Mazar, N., & Ariely, D. (2006). Dishonesty in everyday life and its policy
.2012.651387 implications. Journal of Public Policy & Marketing, 25, 117–126. doi:
Greene, J., & Haidt, J. (2002). How (and where) does moral judgment 10.1509/jppm.25.1.117
work? Trends in Cognitive Sciences, 6, 517–523. doi:10.1016/S1364- McCabe, D. L., & Treviño, L. K. (1997). Individual and contextual
6613(02)02011-9 influences on academic dishonesty: A multicampus investigation. Re-
Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of search in Higher Education, 38, 379 –396. doi:10.1023/A:
personal history. American Psychologist, 35, 603– 618. doi:10.1037/ 1024954224675
0003-066X.35.7.603 McGraw, K. M. (1987). Guilt following transgression: An attribution of
Haidt, J. (2001). The emotional dog and its rational tail: A social intuition- responsibility approach. Journal of Personality and Social Psychology,
ist approach to moral judgment. Psychological Review, 108, 814 – 834. 53, 247–256. doi:10.1037/0022-3514.53.2.247
doi:10.1037/0033-295X.108.4.814 McIlwain, D., Evans, J., Caldis, E., Cicchini, F., Aronstan, A., Wright, A.,
Hare, R. M. (1981). Moral thinking. Oxford, England: Oxford University & Taylor, A. (2012). Strange moralities: Vicarious emotion and moral
Press. doi:10.1093/0198246609.001.0001 emotions in Machiavellian and psychopathic personality styles. In R.
Hemingway, E. (1932). Death in the afternoon. New York, NY: Scribner. Langdon & C. Mackenzie (Eds.), Emotions, imagination, and moral
Hollinger, R. C., & Langton, L. (2007). National retail security survey: reasoning (pp. 119 –148). New York, NY: Psychology Press.
Final report. Gainesville, FL: University of Florida. Mead, N. L., Baumeister, R. F., Gino, F., Schweitzer, M. E., & Ariely, D.
Jones, T. M. (1991). Ethical decision making by individuals in organiza- (2009). Too tired to tell the truth: Self-control resource depletion and
tions: An issue-contingent model. The Academy of Management Review, dishonesty. Journal of Experimental Social Psychology, 45, 594 –597.
16, 366 –395. doi:10.1016/j.jesp.2009.02.004
Kant, I. (1993). Grounding for the metaphysics of morals (J. W. Ellington, Meloy, M. G., Russo, J. E., & Miller, E. G. (2006). Monetary incentives
Trans., 3rd ed.). Indianapolis, IN: Hackett. (Original work published and mood. Journal of Marketing Research, 43, 267–275. doi:10.1509/
1785) jmkr.43.2.267
Katz, J. (1988). Seductions of crime. New York, NY: Basic Books. Monin, B., & Miller, D. T. (2001). Moral credentials and the expression of
Kivetz, Y., & Tyler, T. R. (2007). Tomorrow I’ll be me: The effect of time prejudice. Journal of Personality and Social Psychology, 81, 33– 43.
perspective on the activation of idealistic versus pragmatic selves. Or- doi:10.1037/0022-3514.81.1.33
ganizational Behavior and Human Decision Processes, 102, 193–211. Monin, B., Pizarro, D. A., & Beer, J. S. (2007). Deciding versus reacting:
doi:10.1016/j.obhdp.2006.07.002 Conceptions of moral judgment and the reason-affect debate. Review of
Kohlberg, L. (1969). Stage and sequence: The cognitive-developmental General Psychology, 11, 99 –111. doi:10.1037/1089-2680.11.2.99
 CHEATER’S HIGH? 547

Moore, C., Detert, J. R., Treviño, L. K., Baker, V. L., & Mayer, D. M. Sanitioso, R., Kunda, Z., & Fong, G. T. (1990). Motivated recruitment of
(2012). Why employees do bad things: Moral disengagement and un- autobiographical memories. Journal of Personality and Social Psychol-
ethical organizational behavior. Personnel Psychology, 65, 1– 48. doi: ogy, 59, 229 –241. doi:10.1037/0022-3514.59.2.229
10.1111/j.1744-6570.2011.01237.x Schlenker, B. R., Forsyth, D. R., Leary, M. R., & Miller, R. S. (1980).
Moore, S., & Gullone, E. (1996). Predicting adolescent risk behavior using Self-presentational analysis of the effects of incentives on attitude
a personalized cost-benefit analysis. Journal of Youth and Adolescence, change following counterattitudinal behavior. Journal of Personality
25, 343–359. doi:10.1007/BF01537389 and Social Psychology, 39, 553–577. doi:10.1037/0022-3514.39.4.553
Moore, T., Clayton, R., & Anderson, R. (2009). The economics of online Schwarz, N. (2000). Emotion, cognition, and decision making. Cognition
crime. Journal of Economic Perspectives, 23, 3–20. doi:10.1257/jep.23 & Emotion, 14, 433– 440. doi:10.1080/026999300402745
.3.3 Schweitzer, M. E., & Gibson, D. E. (2008). Fairness, feelings, and ethical
Murdock, T. B., & Anderman, E. M. (2006). Motivational perspectives on decision-making: Consequences of violating community standards of
student cheating: Toward an integrated model of academic dishonesty. fairness. Journal of Business Ethics, 77, 287–301. doi:10.1007/s10551-
Educational Psychologist, 41, 129 –145. doi:10.1207/ 007-9350-3
s15326985ep4103_1 Schweitzer, M. E., & Hsee, C. K. (2002). Stretching the truth: Elastic
Newby-Clark, I. R., Ross, M., Buehler, R., Koehler, D. J., & Griffin, D. justification and motivated communication of uncertain information.
(2000). People focus on optimistic scenarios and disregard pessimistic Journal of Risk and Uncertainty, 25, 185–201. doi:10.1023/A:
scenarios while predicting task completion times. Journal of Experimen- 1020647814263
tal Psychology: Applied, 6, 171–182. doi:10.1037/1076-898X.6.3.171 Schweitzer, M. E., Ordóñez, L., & Douma, B. (2004). Goal setting as a
Okel, E., & Mosher, D. L. (1968). Changes in affective states as a function motivator of unethical behavior. Academy of Management Journal, 47,
of guilt over aggressive behavior. Journal of Consulting and Clinical 422– 432. doi:10.2307/20159591
Psychology, 32, 265–270. doi:10.1037/h0025917 Scully, D., & Marolla, J. (1985). “Riding the bull at Gilley’s”: Convicted
Pechmann, C., & Shih, C. F. (1999). Smoking scenes in movies and rapists describe the rewards of rape. Social Problems, 32, 251–263.
antismoking advertisements before movies: Effects on youth. Journal of doi:10.2307/800685
Marketing, 63, 1–13. doi:10.2307/1251772 Seagrave, K. (2001). Shoplifting: A social history. Jefferson, NC: McFar-
Pérez-Peña, R. (2013, February 13). Students disciplined in Harvard scan- land.
dal. The New York Times. Retrieved from http://www.nytimes.com/ Shalvi, S., Dana, J., Handgraaf, M. J. J., & De Dreu, C. K. W. (2011).
2013/02/02/education/harvard-forced-dozens-to-leave-in-cheating- Justified ethicality: Observing desired counterfactuals modifies ethical
scandal.html?smid⫽pl-share perceptions and behavior. Organizational Behavior and Human Deci-
Pizarro, D. (2000). Nothing more than feelings? The role of emotions in sion Processes, 115, 181–190. doi:10.1016/j.obhdp.2011.02.001
moral judgment. Journal for the Theory of Social Behaviour, 30, 355– Sweeny, K., & Vohs, K. D. (2012). On near misses and completed tasks:
375. doi:10.1111/1468-5914.00135 The nature of relief. Psychological Science, 23, 464 – 468. doi:10.1177/
Pizarro, D. A., Tannenbaum, D., & Uhlmann, E. (2012). Mindless, harm- 0956797611434590
less, and blameworthy. Psychological Inquiry, 23, 185–188. doi: Tajfel, H. (1982). Social psychology of intergroup relations. Annual Re-
10.1080/1047840X.2012.670100 view of Psychology, 33, 1–39. doi:10.1146/annurev.ps.33.020182
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling .000245
strategies for assessing and comparing indirect effects in multiple me- Tangney, J. P., Stuewig, J., & Mashek, D. J. (2007). Moral emotions and
diator models. Behavior Research Methods, 40, 879 – 891. doi:10.3758/ moral behavior. Annual Review of Psychology, 58, 345–372. doi:
BRM.40.3.879 10.1146/annurev.psych.56.091103.070145
Rawls, J. (1971). A theory of justice. Cambridge, MA: Harvard University Tenbrunsel, A. E., Diekmann, K. A., Wade-Benzoni, K. A., & Bazerman,
Press. M. H. (2010). The ethical mirage: A temporal explanation as to why we
Rest, J. R. (1986). Moral development: Advances in research and theory. are not as ethical as we think we are. Research in Organizational
Minneapolis, MN: University of Minnesota Press. Behavior, 30, 153–173. doi: 10.1016/j.riob.2010.08.004
Richmond, R. (2011, August 21). Web site ranks hacks and bestows Tenbrunsel, A. E., & Smith-Crowe, K. (2008). Ethical decision making:
bragging rights. The New York Times. Retrieved from www.nytimes.com/ Where we’ve been and where we’re going. The Academy of Manage-
2011/08/22/technology/web-site-ranks-hacks-and-bestows-bragging- ment Annals, 2, 545– 607. doi:10.1080/19416520802211677
rights.html?_r⫽0 Thomas, D. L., & Diener, E. (1990). Memory accuracy in the recall of
Ring, K., Wallston, K., & Corey, M. (1970). Mode of debriefing as a factor emotions. Journal of Personality and Social Psychology, 59, 291–297.
affecting subjective reaction to a Milgram type obedience experiment: Tracy, J. L., & Robins, R. W. (2006). Appraisal antecedents of shame, guilt
An ethical inquiry. Representative Research in Social Psychology, 1, and pride: Support for a theoretical model. Personality and Social
67– 88. Psychology Bulletin, 32, 1339 –1351. doi:10.1177/0146167206290212
Rogers, C. (1959). A theory of therapy, personality and interpersonal Treviño, L. K. (1986). Ethical decision-making in organizations: A person-
relationships as developed in the client-centered framework. In S. Koch situation interactionist model. The Academy of Management Review, 11,
(Ed.), Psychology: A study of a science: Vol. 3. Formulations of the 601– 617.
person and the social context (pp. 184–256). New York, NY: McGraw Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological
Hill. Review, 110, 403– 421. doi:10.1037/0033-295X.110.3.403
Rozin, P., Lowery, L., Imada, S., & Haidt, J. (1999). The CAD triad Uhlmann, E. L., Pizarro, D. A., Tannenbaum, D., & Ditto, P. H. (2009).
hypothesis: A mapping between three moral emotions (contempt, anger, The motivated use of moral principles. Judgment and Decision Making,
disgust) and three moral codes (community, autonomy, divinity). Jour- 4, 476 – 491.
nal of Personality and Social Psychology, 76, 574 –586. doi:10.1037/ U.S. Internal Revenue Service. (2010). Information Reporting Program
0022-3514.76.4.574 Advisory Committee: Tax Gap Subgroup report. Washington, DC: Au-
Ruedy, N. E., & Schweitzer, M. E. (2011). In the moment: The effect of thor.
mindfulness on ethical decision making. Journal of Business Ethics, 95, Van Lange, P. A. M. (1999). The pursuit of joint outcomes and equality in
73– 87. outcomes: An integrative model of social value orientation. Journal of
548 RUEDY, MOORE, GINO, AND SCHWEITZER

Personality and Social Psychology, 77, 337–349. doi:10.1037/0022- Wiltermuth, S. S. (2011). Cheating more when the spoils are split. Orga-
3514.77.2.337 nizational Behavior and Human Decision Processes, 115, 157–168.
Wallington, S. A. (1973). Consequences of transgression: Self-punishment doi:10.1016/j.obhdp.2010.10.001
and depression. Journal of Personality and Social Psychology, 28, 1–7. Wood, P. B., Gove, W. R., Wilson, J. A., & Cochran, J. K. (1997).
Warren, D. E., & Smith-Crowe, K. (2008). Deciding what’s right: The role Nonsocial reinforcement and habitual criminal conduct: An extension of
of external sanctions and embarrassment in shaping moral judgments in learning theory. Criminology, 35, 335–366. doi:10.1111/j.1745-9125
the workplace. Research in Organizational Behavior, 28, 81–105. doi: .1997.tb00879.x
10.1016/j.riob.2008.04.004 Wright, D. (1971). The psychology of moral behavior. Baltimore, MD:
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali- Penguin.
dation of brief measures of positive and negative affect: The PANAS
Yee, V. (2012, September 25). Stuyvesant students describe the how and
scales. Journal of Personality and Social Psychology, 54, 1063–1070.
the why of cheating. The New York Times. Retrieved from http://www
doi:10.1037/0022-3514.54.6.1063
.nytimes.com/2012/09/26/education/stuyvesant-high-school-students-
Watson, D., Wiese, D., Vaidya, J., & Tellegen, A. (1999). The two general
describe-rationale-for-cheating.html?_rmoc.semityn.www⫽&adxnnl⫽
activation systems of affect: Structural findings, evolutionary consider-
ations, and psychobiological evidence. Journal of Personality and Social 1&ref⫽education&adxnnlx⫽1361189274-⫹4bRpdwmGcQ0eeQq
Psychology, 76, 820 – 838. doi:10.1037/0022-3514.76.5.820 suhAvw
Whitley, B. E. (2001). Gender differences in affective responses to having Zimbardo, P. G. (1970). The human choice: Individuation, reason, and
cheated: The mediating role of attitudes. Ethics & Behavior, 11, 249 – order vs. deindividuation, impulse, and chaos. In W. J. Arnold & D.
259. doi:10.1207/S15327019EB1103_4 Levine (Eds.), 1969 Nebraska Symposium on Motivation (pp. 237–307).
Williams, K. M., Nathanson, C., & Paulhus, D. L. (2010). Identifying and Lincoln, NE: University of Nebraska Press.
profiling scholastic cheaters: Their personality, cognitive ability, and
motivation. Journal of Experimental Psychology: Applied; Journal of
Experimental Psychology: Applied, 16, 293–307. doi:10.1037/a0020773
Wilson, T. D., & Gilbert, D. T. (2003). Affective forecasting. In M. P. Received June 29, 2012
Zanna (Ed.), Advances in experimental social psychology (Vol. 35, pp. Revision received July 29, 2013
345– 411). San Diego, CA: Academic Press. Accepted July 30, 2013 䡲