Journal of Experimental Social Psychology 49 (2013) 1027–1033

Contents lists available at SciVerse ScienceDirect

Journal of Experimental Social Psychology

journal homepage: www.elsevier.com/locate/jesp

The whistleblower's dilemma and the fairness–loyalty tradeoff

Adam Waytz a,⁎, James Dungan b, Liane Young b
a
Northwestern University, USA
b
Boston College, USA

H I G H L I G H T S

• The tradeoff between fairness and loyalty corresponds to whistleblowing decisions.

• Experimental and dispositional variation in this tradeoff maps onto whistleblowing.
• Five studies demonstrate this previously undocumented relationship.
• These results shed light on a novel psychological determinant of whistleblowing.

a r t i c l e i n f o a b s t r a c t

Article history: Whistleblowing – reporting another person's unethical behavior to a third party – often constitutes a conflict be-
Received 4 December 2012 tween competing moral concerns. Whistleblowing promotes justice and fairness but can also appear disloyal. Five
Revised 20 June 2013 studies demonstrate that a fairness–loyalty tradeoff predicts people's willingness to blow the whistle. Study 1
Available online 11 July 2013
demonstrates that individual differences in valuing fairness over loyalty predict willingness to report unethical be-
havior. Studies 2a and 2b demonstrate that experimentally manipulating endorsement of fairness versus loyalty
Keywords:
Ethics
increases willingness to report unethical behavior. Study 3 demonstrates that people recall their decisions to report
Morality unethical behavior as driven by valuation of fairness, whereas people recall decisions not to report unethical behav-
Loyalty ior as driven by valuation of loyalty. Study 4 demonstrates that experimentally manipulating the endorsement of
Fairness fairness versus loyalty increases whistleblowing in an online marketplace. These findings reveal the psychological
Whistleblowing determinants of whistleblowing and shed light on factors that encourage or discourage this practice.
© 2013 Elsevier Inc. All rights reserved.

Introduction processes underlying people's decision to blow the whistle or not. Specif-
ically, we propose that differences in people's valuation of moral norms,
The decision to report another person's unethical behavior to a third fairness versus loyalty, contribute to whistleblowing decisions.
party – to engage in whistleblowing – presents a dilemma. Although Fairness and loyalty alike represent basic moral values, as reflected
some whistleblowers receive heroic acclaim (Johnson, 2003), other in developmental and evolutionary approaches to moral cognition. In-
whistleblowers face revenge from their community (Dyck, Adair, & fants endorse distributive and retributive justice — before age two, chil-
Zingales, 2010). For example, a recent editorial reprimanding National dren expect resources to be divided fairly among individuals according
Security Agency whistleblower, Edward Snowden, stated that Snowden to each individual's effortful contribution in a group task (Kanngiesser &
“faced a moral dilemma” and ultimately “betrayed his employers,” con- Warneken, 2012; Sloane, Baillargeon, & Premack, 2012). Furthermore,
tributing to “the fraying of social fabric” (Brooks, 2013). 8-month-olds prefer to reward helpful, prosocial behavior and punish
What then drives whistleblowing decisions? Previous research has selfish, antisocial behavior (Hamlin, Wynn, Bloom, & Mahajan, 2011).
investigated structural and organizational factors that influence At the same time, young children's adherence to fairness and justice
whistleblowing, including the professional status of whistleblowers, or- norms are powerfully modified by group membership — children
ganizational support for whistleblowing (Dozier & Miceli, 1985; Near & share disproportionate resources with family and friends over strangers
Miceli, 1985; Vadera, Vadera, & Caza, 2009), and the type of behavior (Olson & Spelke, 2008) and often choose to act loyally versus fairly, es-
that people deem unethical and therefore reportable (Gino & Bazerman, pecially when expectations for friendship are made salient (Smetana,
2009). Existing research has not, however, investigated the psychological Killen, & Turiel, 1991). Research on third-party judgments shows that
determinants of whistleblowing. Here, we investigate the cognitive infants also prefer those who harm dissimilar others and help similar
others (Hamlin, Mahajan, Liberman, & Wynn, 2013), and toddlers pre-
⁎ Corresponding author at: Northwestern University, 2001 Sheridan Rd, Evanston, IL
fer those who behave loyally (i.e., who reciprocate) to those who be-
60208, USA. have fairly in certain competitive contexts (Shaw, DeScioli, & Olson,
E-mail address: [email protected] (A. Waytz). 2012). Finally, whereas toddlers consider tattling in some cases to be

0022-1031/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jesp.2013.07.002
1028 A. Waytz et al. / Journal of Experimental Social Psychology 49 (2013) 1027–1033

a just, prosocial act (Ingram & Bering, 2010), adolescents, who place a Method
premium on group membership, respond far more negatively to peer
tattlers (Friman et al., 2004). Notably, precursors to both fairness (e.g., Eighty-three individuals (Mage = 35.72, SDage = 13.93; 65% fe-
Brosnan, Schiff, & de Waal, 2005) and loyalty have been observed in male) participated via Amazon.com's Mechanical Turk in exchange
our primate ancestors as well (Mahajan et al., 2011), revealing the fun- for a small payment; all subsequent studies used the same methodo-
damental nature of both moral norms. logical approach. Participants completed three measures to assess
Although fairness and loyalty represent basic moral values (Haidt, their valuation of fairness versus loyalty. In this study and subsequent
2007; Walker & Hennig, 2004), they do, at times, conflict. At their core, studies, we included only participants who completed all measures.
norms for fairness and justice demand that all persons and groups be The first measure consisted of six-point Likert scale items from the
treated equally. By contrast, loyalty norms dictate preferential treatment, Moral Foundations Questionnaire (MFQ; Graham et al., 2011),
a responsibility to favor one's own group over other groups. Studies have assessing valuation of fairness and loyalty. Two fairness items and
shown that fairness norms typically dominate behavior but may be over- two loyalty items assessed the relevance of various considerations
ridden in contexts that pit fairness against loyalty. For example, factors for judgments of right and wrong (e.g., “Whether or not someone
such as psychological closeness (Batson, Klein, Highberger, & Shaw, acted unfairly,” “Whether or not someone was denied his or her
1995), national culture (Miller & Bersoff, 1992), residential mobility rights” and “Whether or not someone did something to betray his
(Lun, Oishi, & Tenney, 2012), perceived duty (Baron, Ritov, & Greene, or her group,” “Whether or not someone showed a lack of loyalty”);
2013), and relationship type (Rai & Fiske, 2011) modulate people's prefer- one fairness item and one loyalty item asked about agreement with
ence for loyalty versus fairness (see also Shaw et al., 2012). Because of the moral statements (“Justice is the most important requirement for a
fundamental tension between these norms, the present research assesses society” and “People should be loyal to their family members, even
the loyalty–fairness tradeoff rather than assessing each in isolation. when they have done something wrong”). Following Graham et al.
We propose that fairness and loyalty norms clash during (2011), we averaged the three loyalty and three fairness items sepa-
whistleblowing decisions. Our definition of whistleblowing corresponds rately and subtracted the loyalty score from the fairness score to pro-
to organizational definitions of this behavior as well as with definitions duce a composite values score (these items were embedded among
of “tattling” from the social cognitive development perspective (Ingram three other MFQ items irrelevant to the current hypothesis).
& Bering, 2010) and “snitching” from a legal perspective (Natapoff, The second measure asked, “Objectively speaking, who do you
2004). We take whistleblowing to involve two key components: (1) think is the more morally good person?” with a forced-choice re-
reporting unethical behavior (2) to a third party (e.g., an authority figure). sponse option: “Someone who is fair and just, impartial and
On the one hand, whistleblowers may act in the service of fairness unprejudiced” (fairness; coded 1) or “Someone who is loyal and faith-
and justice when exposing corporate wrongdoing (Miceli & Near, ful, devoted and dependable” (loyalty; coded 0). This constituted par-
1992; Near & Miceli, 1985), neighborhood crime (Natapoff, 2004), or ticipants' judgment score.
scientific fraud (Vogel, 2011; Yong, 2012). On the other hand, The third measure asked, “Who would you rather be friends
whistleblowing may constitute an act of disloyalty, depending also on with?” with a forced-choice response option: “Someone who is fair
the relationship between the offender and the whistleblower. Indeed, and just to others, who is impartial and unprejudiced regardless of
the vast majority of corporate whistleblowers face negative outcomes how it affects their family and friends” (fairness; coded 1) or “Some-
as a result of their actions: revenge, reassignment, firing, and personal one who is loyal and faithful to their family and friends, who is devot-
distress (Dyck et al., 2010) and such “moral rebels” are often ostracized ed and dependable regardless of how it affects outsiders” (loyalty;
(e.g., Minson & Monin, 2012; Monin, Sawyer, & Marquez, 2008; Parks & coded 0). This constituted participants' friendship score.
Stone, 2010). Would-be whistleblowers are thus faced with the dilem- We standardized and averaged the three scores (α = .64) to com-
ma of choosing between competing demands. Whereas fairness norms pute a composite fairness-versus-loyalty score. Higher values indicate
typically require that people report and punish wrongdoing, loyalty a preference for fairness over loyalty, whereas lower values indicate a
norms – even in the abstract – indicate that reporting another person preference for loyalty over fairness.
to a third party may constitute an act of betrayal, associated with poten- To measure attitudes toward whistleblowing, we asked partici-
tial repercussions as detailed above. pants about seven violations ranging in severity2:
We propose that whistleblowing behavior constitutes a tradeoff be-
1. Stealing $1 from a restaurant's tip jar.
tween fairness and loyalty. A direct prediction of this proposal is that
2. Embezzling $1000 from their work place.
the endorsement of fairness versus loyalty tracks subsequent decisions
3. Robbing a woman of her cell phone and wallet.
to blow the whistle. Evidence from five studies supports this prediction.
4. Cheating on their final exam in college.
First, individual differences in the endorsement of fairness versus loyal-
5. Spraying rude graffiti on the side of a local store.
ty correspond to decisions to blow the whistle (Study 1). Second, exper-
6. Using and selling marijuana to other adults.
imentally manipulating concern for fairness versus loyalty predicts
7. Fatally stabbing a convenience store owner.
willingness to blow the whistle (Studies 2a and 2b). Third, people de-
scribe real-life decisions to blow the whistle as motivated by concerns For each scenario, participants indicated (1 = Very unlikely; 7 =
for fairness more than loyalty, whereas they describe decisions to not Very likely) how likely they would be to blow the whistle if the perpe-
blow the whistle as motivated by concerns for loyalty more than fair- trator were:
ness (Study 3). Finally, experimental inductions of fairness versus loyal-
1. A total stranger you've never met.
ty predict real-life whistleblowing in an online marketplace (Study 4).
2. An acquaintance you see occasionally.
3. A close friend you've known for years.
Study 1: individual differences 4. A family member you're very close to.

Study 1 assessed individual differences in valuation of fairness Specifically, for each offense, participants read a version of the fol-
versus loyalty and the relation to whistleblowing.1 lowing, “Imagine that you witness someone stealing $1 from a
restaurant's tip jar. How likely would you be to report the perpetrator
1 of this incident if this perpetrator were… a stranger/acquaintance/
Following Simmons, Nelson, and Simonsohn (2012), we report how we deter-
mined our sample size, all data exclusions (if any), all manipulations, and all measures.
2
Sample sizes were determined separately for each study based on prior similar studies Unrelated pilot items for a separate study were randomly presented before or after
in the literature. the whistleblowing scenarios and did not affect results.
 A. Waytz et al. / Journal of Experimental Social Psychology 49 (2013) 1027–1033 1029

Table 1
Correlations between fairness-minus-loyalty score and whistleblowing score for each offense (Study 1).

Stealing Embezzlement Robbery Cheating Graffiti Marijuana Stabbing

.21 (p = .055) .26 (p = .019) .20 (p = .073) .26 (p = .019) .29 (p = .008) .07 (p = .52) .13 (p = .26)

close friend/family member?” each accompanied by the 7-point specified the relevant authority to whom the offense would be reported
scale. We computed an overall whistleblowing score, by averaging (see supplementary materials).
all 28 responses (four levels of closeness for each of seven scenarios)
(α = .93). We also computed separate whistleblowing scores for Results and discussion
each target (i.e., stranger, acquaintance, friend, and family member)
by averaging responses over all seven scenarios (all αs N .74). For ex- Independent-samples t-tests revealed that overall willingness to
ploratory purposes, we also computed separate whistleblowing blow the whistle (α = .94) was higher in the fairness condition
scores for each offense by averaging over all four levels of closeness (M = 4.98, SD = 1.10) than in the loyalty condition (M = 4.47,
(all αs N .77). SD = 1.06), t(140) = 2.78, p = .006, d = .47. This finding replicates
the pattern of Study 1 using an experimental manipulation of
Results and discussion fairness-versus-loyalty.
We then conducted a 2 (condition: fairness/justice vs. loyalty) × 4
Our primary hypothesis was that fairness-versus-loyalty valuation (target: stranger, acquaintance, friend, family member) mixed-effects
would be associated with positive whistleblowing decisions.3 As pre- ANOVA on participants' willingness to blow the whistle averaged
dicted, participants' fairness-versus-loyalty score was correlated with across the seven scenarios (all αs N .69 for composite scores for
whistleblowing overall, r(81) = .28, p = .011. each target). We observed a main effect of target (multivariate:
For exploratory purposes, we also explored relations between F(3,138) = 52.61, p b .001, η2p = .53; univariate: F(3, 420) =
fairness-versus-loyalty and whistleblowing for specific targets (i.e., 131.89, p b .001, η2p = .49): the closer the relationship between par-
levels of relationship closeness) and specific offenses. Although willing- ticipant and violator, the less willing participants were to blow the
ness to blow the whistle decreased as relationship closeness increased whistle (stranger: M = 5.42, SD = 1.01; acquaintance: M = 5.16,
(stranger: M = 5.32, SD = 1.10, acquaintance: M = 5.02, SD = 1.11, SD = 1.09; friend: M = 4.31, SD = 1.39; family member: M = 4.00,
friend: M = 4.48, SD = 1.25, and family member: M = 4.27, SD = SD = 1.54). We also observed a significant interaction between con-
1.35; paired-samples t-tests demonstrated that all differences were sig- dition and target (multivariate: F(3,138) = 3.74, p = .013, η2p =
nificant, ps b .005), fairness-versus-loyalty scores correlated with .075; univariate: F(3, 420) = 8.58, p b .001, η2p = .058): the effect
whistleblowing at each level of closeness (stranger: r(81) = .22, p = of the loyalty versus fairness prime was more pronounced for close
.051, acquaintance: r(81) = .28, p = .01, friend: r(81) = .26, p = versus distant targets.
.023, and family member: r(81) = .25, p = .022). See Table 1 for corre- Independent-samples t-tests on dependent measures showed that
lations for specific offenses. Overall, individual differences in the en- condition predicted whistleblowing for family member (M = 4.44,
dorsement of fairness over loyalty predicted willingness to blow the SD = 1.50 vs. M = 3.56, SD = 1.46; p = .001), friend (M = 4.65,
whistle on a range of perpetrators committing crimes of varying SD = 1.34 vs. M = 3.96, SD = 1.35; p = .003), marginally for ac-
severity. quaintance (M = 5.33, SD = 1.01 vs. M = 4.99, SD = 1.14; p =
.062), and non-significantly for stranger (M = 5.48, SD = 1.05 vs.
Study 2: inducing fairness and loyalty M = 5.37, SD = 0.98; p N .50). See Table 2 for whistleblowing scores
by condition for each offense for this study and Study 2b.
Study 1 tested whether individual differences in moral valuation Participants reported more willingness to blow the whistle when
could predict whistleblowing decisions. Study 2 tested whether exper- primed to consider fairness versus loyalty. This result further sup-
imentally manipulating subjective valuation of fairness/justice versus ports our hypothesis that endorsement of fairness versus loyalty
loyalty could produce the same pattern of differences. Study 2a provides norms tracks with subsequent decisions to blow the whistle.
an initial test, while Study 2b provides a replication of the effect.
Study 2b
Study 2a
Method
Method
One hundred fifty-one MTurk workers (Mage = 33.74, SDage =
12.77; 45% female) participated as in Study 1. Study 2b was identical
One hundred forty-two MTurk workers (Mage = 33.19, SDage =
to Study 2a, with two exceptions. (1) To alleviate task demands, we
12.01; 51% female) participated as in Study 1. Participants were first
told participants at the start of the assignment that they would be
randomly assigned to write three short essays on the importance of
completing two studies: a first study consisting of a writing task on
fairness/justice or the importance of loyalty, designed to induce partic-
values and a second study involving answering questions about social
ipants to prioritize the target norm. For example, in the fairness/justice
decision-making. The writing section, which involved the experimen-
condition, participants were asked, “Please write a few sentences about
tal manipulation, was clearly labeled, “STUDY 1: VALUES,” and the
why it is more important to be just than to be loyal” whereas in the loy-
section that included the whistleblowing measures was clearly la-
alty condition, participants were asked, “Please write a few sentences
beled, “STUDY 2: SOCIAL DECISION MAKING.” Second, we included
about why it is more important to be loyal than to be just” (see supple-
additional questions at the study's end asking whether participants
mentary materials for full instructions). Participants then completed
had previously seen any of the study materials as well as a series of
the same measures of whistleblowing from Study 1; however, we
suspicion checks to assess potential effects of task demands (Inbar,
3
Pizarro, Gilovich, & Ariely, 2013). First, we asked whether participants
The forced choice nature of the judgment and friendship items precludes us from
regressing whistleblowing separately on the dimensions of fairness and loyalty. How-
thought there was anything “strange, confusing or suspicious about
ever, separate value scores for fairness (p = .98) and loyalty (p = .14) did not predict these studies” (Yes/No). If participants answered, “yes,” they saw an
whistleblowing overall. item asking whether they thought that the first and second study
1030 A. Waytz et al. / Journal of Experimental Social Psychology 49 (2013) 1027–1033

Table 2
Whistleblowing scores for each offense, by condition (Study 2).

Study/condition Stealing Embezzlement Robbery Cheating Graffiti Marijuana Stabbing

M (SD) M (SD) M (SD) M (SD) M (SD) M (SD) M (SD)

2a/Fairness 4.34a 5.61a 6.01a 4.27a 4.57a 3.56a 6.48a

(1.63) (1.36) (1.15) (1.94) (1.78) (2.08) (0.90)
2a/Loyalty 3.82b 5.02b 5.88a 3.37b 3.95b 2.91a 6.33a
(1.51) (1.27) (1.13) (1.73) (1.79) (1.96) (1.05)
2b/Fairness 3.78a 5.37a 5.96a 3.76a 4.35a 3.17a 6.33a
(1.80) (1.46) (1.02) (1.77) (1.72) (2.12) (1.04)
2b/Loyalty 3.65a 4.96a 5.35b 3.38a 3.85a 2.91a 5.97b
(1.69) (1.30) (1.25) (1.66) (1.64) (1.95) (1.07)

Means that do not share a subscript across condition within each study differ significantly. In Study 2a, marijuana differs by condition at p = .060. In Study 2b, embezzlement differs
at p = .073 and graffiti varies at p = .069.

were related at all. If participants answered, “yes,” they were asked to that someone you knew – a friend, family member, co-worker, or ac-
write about how they were related. Only three people answered, quaintance – did something that was morally wrong, and you had the
“yes” to any question; excluding these individuals did not alter option to report that person, and chose to report the person.” In the
results. WB −, participants were asked to recall a similar incident in which
they had the option to report that person and chose not to do
Results and discussion anything.
Participants were asked to write a short paragraph about what
Independent-samples t-tests again revealed that overall willingness was going through their mind when they did or did not decide to re-
to blow the whistle (α = .93) was higher in the fairness condition port that person, their reasons for their decision and whether their
(M = 4.67, SD = 1.07) than in the loyalty condition (M = 4.30, decision was driven by any particular moral or ethical values. On a
SD = 1.04), t(149) = 2.78, p = .029, d = .36. separate screen, participants indicated, “How difficult was your deci-
We also conducted the same 2 (condition: fairness/justice vs. loy- sion to (not) report the person?” (1 = Very easy; 7 = Very difficult),
alty) × 4 (target: stranger, acquaintance, friend, family member) and the extent to which their decision was driven by (1) the value
mixed-effects ANOVA on participants' willingness to blow the whistle of “fairness/justice” and (2) the value of “loyalty” (1 = Not at all;
averaged across the seven scenarios (all αs N .72 for composite scores 10 = Very much).
for each target). A main effect emerged for target (multivariate:
F(3,147) = 55.76, p b .0001, η2p = .53; univariate, Greenhouse– Results and discussion
Geisser correction: F(1.34, 200.31) = 141.77, p b .0001, η2p = .49):
the closer the relationship between participant and violator, the less In exploring participants' free responses, we created categories for
willing participants were to blow the whistle (stranger: M = 5.23, any violation that appeared more than once; violations that appeared
SD = 1.05; acquaintance: M = 4.89, SD = 1.07; friend: M = 4.09, only once were categorized as “other”. Participants were also given
SD = 1.30; family member: M = 3.74, SD = 1.46). Sphericity was vi- the opportunity not to identify the violation (to prevent distress
olated (p b .0001), and the condition × target interaction was non- over revealing sensitive information and to prevent data loss from
significant using the multivariate approach (F = 1.51, p = .21) but people dropping out). Frequencies for this categorization scheme for
was marginally significant using the univariate approach (Green- the WB + vs. WB − conditions were: stealing/theft (25 vs. 20),
house–Geisser correction, F(1.34, 200.31) = 3.40, p = .054, η2p = cheating (6 vs. 8), infidelity/sex (5 vs. 9), substance abuse (1 vs. 6),
.067), revealing the same pattern of effects as in Study 2a. Specifically, property damage (4 vs. 2), lying (3 vs. 3), physical abuse (3 vs. 3),
independent-samples t-tests on dependent measures showed that sexual abuse/harassment (2 vs. 2), missing work (0 vs. 3), other
condition predicted whistleblowing for family member (M = 4.04, (3 vs. 7), and those who chose not to identify the violation (9 vs.
SD = 1.43 vs. M = 3.45, SD = 1.45; p = .012), friend (M = 4.34, 11). Binomial tests confirmed that these counts did not differ by con-
SD = 1.27 vs. M = 3.84, SD = 1.28; p = .017), marginally for ac- dition (all ps N .12). In addition, the difficulty of the decision did not
quaintance (M = 5.03, SD = 1.07 vs. M = 4.74, SD = 1.06; p = differ by condition, t(133) = 0.530, p N .50.
.099), and non-significantly for stranger (M = 5.29, SD = 1.07 vs. We analyzed participants' written responses using Linguistic Inqui-
M = 5.17, SD = 1.04; p N .47). Again, these findings reveal that ry and Word Count (LIWC; Pennebaker, Chung, Ireland, Gonzales, &
priming fairness versus loyalty norms corresponds to subsequent Booth, 2007), a text analysis software program. To investigate the
whistleblowing decisions. values driving whistleblowing behavior, we used the Moral Founda-
tions Dictionary (MFD, developed by Graham, Haidt, & Nosek, 2009),
Study 3: recalling past whistleblowing opportunities which includes lexical categories for fairness and loyalty. For each par-
ticipant, we calculated the proportion of words in their response that
Studies 1 and 2 measured whistleblowing behavior in hypotheti- fell into each MFD category. This analysis, described in supplementary
cal scenarios. Study 3 extended this work by asking participants materials, revealed that, as predicted, WB- participants used a larger
about whether fairness and loyalty have contributed to real-life proportion of loyalty versus fairness words. However, given that a
whistleblowing opportunities. low percentage of participants used words captured by the MFD and
that this analysis appeared to be driven by outliers, we pursued a
Method more sensitive analysis strategy next, analyzing the open-ended data
by employing human coders.
One hundred thirty-five MTurk workers (Mage = 30.15, SDage = Coders were blind to our hypotheses; however, because condition
9.82; 34% female) participated. Participants were randomly assigned (WB+, WB−) was easily determined from participants' responses,
to either the whistleblowing (WB +) condition or to the no we assigned coders to only one condition. Four independent coders
whistleblowing (WB −) condition. In the WB + condition, partici- (paid participants; two per condition) characterized the reason each
pants were asked to recall “an incident when you became aware participant gave for reporting/not reporting. Specifically, each response
 A. Waytz et al. / Journal of Experimental Social Psychology 49 (2013) 1027–1033 1031

was coded using four binary-measures, i.e., whether or not (1) fairness,
(2) loyalty, (3) a disregard for fairness, or (4) a disregard for loyalty
played a part in the participant's decision. We emphasized that the
codes should be based on the participant's final decision whether to re-
port or not, and that any response could be coded as any combination of
the four measures. Some participants were coded as having mentioned
none of the principles, and some were coded as having mentioned more
than one principle. This coding scheme allowed us to account for the
richness of participants' responses. For example, some participants
appeared to be contemplating the merits of both fairness and loyalty,
whereas others showed both regard and disregard for loyalty (or fair-
ness) simultaneously. Coders were given an additional section to com-
ment if they felt any other reason played a role in participants'
decisions; this approach revealed no notable differences. After coding
the stories independently, coders within each pair discussed discrepan-
cies between their codes and came to an agreement.
Fig. 1. Participants instructed to recall past decisions to blow the whistle (WB+ condi-
We analyzed final codes by conducting binomial tests on each vari- tion) reported that their decision was driven more by fairness than by loyalty, whereas
able – fairness, loyalty, disregard for fairness, disregard for loyalty – for participants instructed to recall past decisions to not blow the whistle (WB− condition)
each condition (see Table 3 for frequencies; see supplementary mate- reported that their decision was driven more by loyalty than by fairness (Study 3). Error
rials for an alternative analysis). For the WB+ condition, significant dif- bars indicate standard error.

ferences emerged in the predicted direction for fairness (72.1% were

identified as using this principle, p = .001), loyalty (23%, p b .0001), amongst your fellow workers” (see supplementary materials for com-
and disregard for fairness (78.7%, p b .0001) (disregard for loyalty, plete essay prompts).
p = .31). For the WB− condition, significant differences emerged in Following the essay task, participants were that told they would
the predicted direction for fairness (5.4%, p b .0001), disregard for fair- be instructed to complete a typing task for piloting purposes. In this
ness (62.2%, p = .047), and disregard for loyalty (94.6%, p b .0001) task, they were asked to write out, in English, all numbers, 1–30. Par-
(loyalty, p = .91). Chi-square tests that compare each variable by con- ticipants read that they would then see the work of the previous par-
dition were significant as well, χ2s N 9.46, ps b .01, Φs N .26. ticipant. All participants were presented with an example of
We also analyzed the Likert scale questions of “to what extent the substandard work, which participants later had the chance to report
decision was driven by fairness or loyalty.” A 2(condition: WB+ vs. (Fig. 2).
WB −) × 2(value: fairness vs. loyalty) mixed-effects ANOVA revealed Participants proceeded with the task and then afterward wrote
the predicted condition × value interaction, F(1,133) = 112.191, about their impressions of the task and the previous participant's
p b .001, η2p = .46. WB + participants expressed higher levels of fair- work. In these open-ended questions, six participants expressed sus-
ness (M = 7.98, SD = 2.11) and lower levels of loyalty (M = 4.72, picion about the existence of this previous participant (e.g., “It was
SD = 3.03), compared to WB− participants; WB − participants not a real person”) or the purpose of the study (e.g., “I'm guessing
expressed higher levels of loyalty (M = 7.55, SD = 2.42) and lower you were trying to prime me for fairness and seeing how I reacted
levels of fairness (M = 4.51, SD = 2.24). All simple effects were to a person who acted unfairly”); excluding these participants did
significant (independent-samples t-tests, ps b .001) (Fig. 1). No not alter the results. Finally, participants responded to two key
main effects emerged. Overall, these findings suggest that fairness items on a 7-point scale (1 = definitely not; 7 = definitely yes). The
and loyalty constitute core values considered for whistleblowing first item stated, “Please tell us, would you recommend that we
decisions. block this previous participant from future MTURK studies?” The sec-
ond item stated, “The present study is a pilot study for future studies.
Study 4: modulating whistleblowing behavior In the future we plan to weed out rule violators. Please report: Did the
previous participant violate any rules or codes of MTURK?” These
Study 4 combined methods from Studies 2 and 3 to test whether items were averaged together (r(155) = .53, p b .0001) as a compos-
priming fairness versus loyalty could modulate actual whistleblowing ite measure of whistleblowing.
behavior in an online environment.
Results and discussion
Method
An independent-samples t-test revealed that participants in the
One hundred fifty-seven MTurk workers (Mage = 35.09, SDage = fairness condition (M = 4.36, SD = 1.83) engaged in more whistle-
13.61; 48% female) were randomly assigned to the fairness or loyalty blowing behavior than participants in the loyalty condition (M =
condition, in which they wrote an essay about the importance of pro- 3.72, SD = 1.71), t(155) = 2.29, p = .023, d = .37. Only ten partici-
moting a sense of fairness or loyalty among Amazon.com Mechanical pants (6% of the sample) did not properly complete the typing task,
Turk's workplace. For example, in the fairness condition, participants five per condition (thus, completion rates seem unlikely to have af-
were asked to “write about why it is important to be just and fair fected whistleblowing behavior). We suggest instead that the present
amongst your fellow workers”, whereas in the loyalty condition, par- effects resulted from experimental priming of fairness versus loyalty
ticipants were asked to “write about why it is important to be loyal prior to participants' evaluation of a substandard worker. This result
provides further support that the fairness–loyalty tradeoff contrib-
utes to real-world whistleblowing decisions.
Table 3
Frequency of principles that coders identified by condition, N = 61 in WB+ condition,
N = 74 in WB− condition (Study 3). General discussion
Condition Fairness Loyalty Disregard for fairness Disregard for loyalty
The present research provides a novel characterization of
WB+ 44 14 13 26 whistleblowing as a tradeoff between acting fairly and acting loyally.
WB− 4 36 46 4
Five studies demonstrate that moral behavior appears to be influenced
1032 A. Waytz et al. / Journal of Experimental Social Psychology 49 (2013) 1027–1033

Fig. 2. The substandard work of the ostensible prior participant that all actual participants saw in Study 4.

not simply by the moral norms that people hold but by how people outgroup) (Cohen, Montoya, & Insko, 2006), suggesting that guilt
trade off different moral norms. The current work thus follows a tradi- may also decrease whistleblowing in adherence to loyalty norms.
tion of research that describes decision-making as involving tradeoffs The current results illuminate these seemingly contradictory findings.
between competing values (Keeney & Raiffa, 1976; von Neumann & When moral norms conflict, as in the case of fairness and loyalty, the
Morgenstern, 1947). In this case, we examine the influence of values relationship between moral emotions and behavioral outcomes will
that people may naturally trade off (Jost, 2009; see also, Graham, vary depending on which moral norm takes precedent.
2011). The present goal was to examine the fairness–loyalty tradeoff The current work also informs our understanding of the cross-
for whistleblowing decisions, yet it will be important for future work cultural perceptions of unethical behavior. For example, although
to determine the separate influence of these values, the influence of ad- most Western cultures are individualistic, endorsing fairness over
ditional values (e.g., concerns regarding harm, authority, purity, and lib- loyalty, many Asian cultures are collectivistic, exalting the importance
erty), as well as these values' influence on decisions related to of loyalty (Graham et al., 2011; see also, Miller & Bersoff, 1992). Asian
whistleblowing, such as confronting the offender directly (Czopp, cultures might therefore preferentially tolerate unfair but “loyal” be-
Monteith, & Mark, 2006). havior. Indeed, existing research suggests that the United States, an
Our research raises a question concerning the precise meaning of individualistic country, views whistleblowing as more ethical than
loyalty in the context of whistleblowing. Theoretical discussion of or- collectivistic countries such as Japan (Brody, Coulter, & Mihalek,
ganizational whistleblowing sees “whistleblowing as entirely com- 1998), China (Chiu, 2003), and Taiwan (Brody, Coulter, & Lin, 1999;
patible with employee loyalty” (Larmer, 1992, p. 125) and indicates see also Christie, Kwon, Stoeberl, & Baumhart, 2003).
that “whistle-blowers have been argued to be more loyal to the orga- The present research may also be relevant for ongoing normative
nization than inactive observers” (Vadera et al., 2009, p. 558). Indeed, debates about the very nature of morality. On one account, a key as-
loyalty may reflect loyalty not only to an offending individual but also pect of morality is impartiality (DeScioli & Kurzban, 2009, 2012;
to one's organization at large; thus the whistleblower may believe Kurzban, DeScioli, & Fein, 2012). This account suggests that being
that reporting wrongdoing benefits the organization and its reputa- moral consists of being fair, rather than being loyal. Despite this,
tion. Loyalty may therefore be targeted at groups of varying scope — adults and children are prone to behaving loyally rather than fairly
from one's immediate friends and family to society at large. Fairness or impartially (Bloom, 2004; Haidt, 2012; Hamlin et al., 2013; Shaw
too may reflect diverse values, from equality to meritocracy (Rai & et al., 2012). Indeed, our moral senses may have evolved from pro-
Fiske, 2011). Notably, the present studies relied on fairly broad defini- cesses that function to favor the ingroup, such as kin selection
tions of fairness and loyalty, in line with existing literature in moral (Bloom, 2011; Churchland, 2011).
psychology (e.g., Moral Foundations Theory; Graham et al., 2009; Finally, the current findings offer recommendations for how to pro-
Graham et al., 2011; Haidt & Josephs, 2004). Importantly, our findings mote fairness and to encourage whistleblowing. One suggestion is to en-
suggest that valuation of fairness, broadly construed, over loyalty, gage brute-force deliberate reasoning to override prepotent partiality-
broadly construed, increased whistleblowing decisions. based responses of ingroup favoritism and bias (Feinberg, Willer,
Given the results of Studies 1 and 2, we also qualify that this Antonenko, & John, 2012; Kahneman, 2011). Another method may be to
tradeoff may be most pertinent when the perpetrator of wrongdoing reframe whistleblowing as demonstrating a “larger loyalty” (Rorty, 1997)
and the whistleblower are socially close (e.g., relatives). Although — an allegiance to the superordinate group of society as a whole and the
unpredicted, this interaction effect can be understood in the context greater good. Loyalty can reflect allegiance toward a distinctive ingroup or
of prior work on relationship types and moral norms (Haidt, 2007; toward a more universal social circle, and promoting loyalty toward a
Rai & Fiske, 2011), which suggests that the principle of loyalty may larger social circle may in fact reflect norms of fairness and promote
apply more to close relationships than distant relationships. Because whistleblowing behavior. Reconciling the conflict between fairness and
loyalty represents a key component of the fairness–loyalty tradeoff, loyalty in these terms may improve perceptions of whistleblowing and ul-
this tradeoff can be expected to play a significant role in cases of timately encourage ethical behavior across cultures.
close relationships. Future work should additionally examine wheth-
er this effect is modulated by the nature of the loyalty — loyalty to the Acknowledgments
individual (e.g., family member, friend) or the relevant group (e.g.,
family, social circle). We thank Davis Vo, Howard Gardner, Nicolas Baumard, and Ellen
Winner for advice and assistance on this work. This project was sup-
Implications of the present research ported in part by a grant from the John Templeton Foundation.

The tradeoff illustrated here has important implications for re- Appendix A. Supplementary materials
search on the influence of moral emotions, such as guilt. On one
hand, guilt proneness is associated with unwillingness to behave Supplementary materials to this article can be found online at
unethically (Cohen, Panter, & Turan, 2012), suggesting guilt increases http://dx.doi.org/10.1016/j.jesp.2013.07.002.
willingness to blow the whistle on fairness violations (see also,
Ellemers, Spears, & Doosje, 2002). On the other hand, when ingroup References
concerns are made salient, guilt-prone individuals show increased Baron, J., Ritov, I., & Greene, J. D. (2013). The duty to support nationalistic policies. Journal
concern for the welfare of their ingroup (at the expense of the of Behavioral Decision Making, 26, 128–138.
 A. Waytz et al. / Journal of Experimental Social Psychology 49 (2013) 1027–1033 1033

Batson, C. D., Klein, T. R., Highberger, L., & Shaw, L. L. (1995). Immorality from Inbar, Y., Pizarro, D. A., Gilovich, T., & Ariely, D. (2013). Moral masochism: On the con-
empathy-induced altruism: When compassion and justice conflict. Journal of Per- nection between guilt and self-punishment. Emotion, 13, 14–18.
sonality and Social Psychology, 68, 1042–1054. Ingram, G. P. D., & Bering, J. M. (2010). Children's tattling: The reporting of everyday
Bloom, P. (2004). Descartes' baby: How the science of child development explains what norm violations in preschool settings. Child Development, 81, 945–957.
makes us human. New York, NY: Basic Books. Johnson, R. A. (2003). Whistleblowing: When it works and why. Boulder, CO: Lynne
Bloom, P. (2011). Family, community, trolley problems, and the crisis in moral psychol- Riener Publishers.
ogy. The Yale Review, 99, 26–43. Jost, J. T. (2009). Group morality and ideology: Left and right, right and wrong. Paper
Brody, R. G., Coulter, J. M., & Lin, S. (1999). The effect of national culture on presented to the Society for Personality and Social Psychology Annual Conference,
whistle-blowing perceptions. Teaching Business Ethics, 3, 385–400. Tampa, FL.
Brody, R. G., Coulter, J. M., & Mihalek, P. H. (1998). Whistle-blowing: A cross-cultural Kahneman, D. (2011). Thinking, fast and slow. New York: Farrar, Straus, & Giroux.
comparison of ethical perceptions of U.S. and Japanese accounting students. Amer- Kanngiesser, P., & Warneken, F. (2012). Young children consider merit when sharing
ican Business Review, 16, 14–21. resources with others. PLoS One, 7, 1–5.
Brooks, D. (2013). Katrina's silver lining. New York Times, 23. Keeney, R. L., & Raiffa, H. (1976). Decisions with multiple objectives: Preferences and
Brosnan, S. F., Schiff, H. C., & de Waal, F. B. M. (2005). Tolerance for inequity may in- value tradeoffs. New York: Wiley.
crease with social closeness in chimpanzees. Proceedings of the Royal Society: Bio- Kurzban, R., DeScioli, P., & Fein, D. (2012). Hamilton vs. Kant: Pitting adaptations for al-
logical Sciences, 272, 253–258. truism against adaptations for moral judgment. Evolution and Human Behavior, 33,
Chiu, R. K. (2003). Ethical judgment and whistleblowing intention: Examining the 323–333.
moderating role of locus of control. Journal of Business Ethics, 43, 65–74. Larmer, R. A. (1992). Whistleblowing and employee loyalty. Journal of Business Ethics,
Christie, P. M. J., Kwon, I. G., Stoeberl, P. A., & Baumhart, R. (2003). A cross-cultural 11, 125–128.
comparison of ethical attitudes of business managers: India, Korea, and the United Lun, J., Oishi, S., & Tenney, E. R. (2012). Residential mobility moderates preferences for egal-
States. Journal of Business Ethics, 46, 263–287. itarian versus loyal helpers. Journal of Experimental Social Psychology, 48, 291–297.
Churchland, P. (2011). Braintrust: What neuroscience tells us about morality. : Princeton Mahajan, N., Martinez, M. A., Gutierrez, N. L., Diesendruck, G., Banaji, M. R., & Santos, L.
University Press. (2011). The evolution of intergroup bias: Perceptions and attitudes in rhesus ma-
Cohen, T. R., Montoya, R. M., & Insko, C. A. (2006). Group morality and intergroup rela- caques. Journal of Personality and Social Psychology, 100, 387–405.
tions: Cross-cultural and experimental evidence. Personality and Social Psychology Miceli, M. P., & Near, J. P. (1992). Blowing the whistle. New York, NY: Lexington Books.
Bulletin, 32, 1559–1572. Miller, J. G., & Bersoff, D. M. (1992). Culture and moral judgment: How are conflicts be-
Cohen, T. R., Panter, A. T., & Turan, N. (2012). Guilt proneness and moral character. Cur- tween justice and interpersonal responsibilities resolved? Journal of Personality
rent Directions in Psychological Science, 21, 355–359. and Social Psychology, 62, 541–554.
Czopp, A. M., Monteith, M. J., & Mark, A. Y. (2006). Standing up for a change: Reducing Minson, J. A., & Monin, B. (2012). Do-gooder derogation: Putting down morally-
bias through interpersonal confrontation. Journal of Personality and Social Psychol- motivated others to defuse implicit moral reproach. Social and Psychological and
ogy, 90(5), 784–803. Personality Science, 3, 200–207.
DeScioli, P., & Kurzban, R. (2009). Mysteries of morality. Cognition, 112, 281–299. Monin, B., Sawyer, P. J., & Marquez, M. J. (2008). The rejection of moral rebels: Resenting
DeScioli, P., & Kurzban, R. (2012). A solution to the mysteries of morality. Psychological those who do the right thing. Journal of Personality and Social Psychology, 95, 76–93.
Bulletin, 22, 1151–1164. Natapoff, A. (2004). Snitching: The institutional and communal consequences. Univer-
Dozier, J. B., & Miceli, M. P. (1985). Potential predictors of whistleblowing: A prosocial sity of Cincinnati Law Review, 73, 645–703.
behavior perspective. Academy of Management Review, 10, 823–836. Near, J. P., & Miceli, M. P. (1985). Organizational dissidence: The case of
Dyck, A., Adair, M., & Zingales, L. (2010). Who blows the whistle on corporate fraud? whistle-blowing. Journal of Business Ethics, 4, 1–16.
Journal of Finance, 65, 2213–2254. Olson, K. R., & Spelke, E. S. (2008). Foundations of cooperation in young children. Cog-
Ellemers, N., Spears, R., & Doosje, B. (2002). Self and social identity. Annual Review of nition, 108, 222–231.
Psychology, 53, 161–186. Parks, C. D., & Stone, A. B. (2010). The desire to expel unselfish members from the
Feinberg, M., Willer, R., Antonenko, O., & John, O. P. (2012). Liberating reason from the group. Journal of Personality and Social Psychology, 99, 303–310.
passions: Overriding intuitionist moral judgments through emotion reappraisal. Pennebaker, J. W., Chung, C. K., Ireland, M., Gonzales, A., & Booth, R. J. (2007). The devel-
Psychological Science, 23, 788–795. opment and psychometric properties of LIWC2007. Austin, TX: LIWC.Net.
Friman, P. C., Woods, D. W., Freeman, K. A., Gilman, R., Short, M., McGrath, A. M., et al. (2004). Rai, T. S., & Fiske, A. P. (2011). Moral psychology is relationship regulation: Moral
Relationships between tattling, likeability, and social classification: A preliminary inves- motives for unity, hierarchy, equality, and proportionality. Psychological Review,
tigation of adolescents in residential care. Behavior Modification, 28, 331–348. 118, 57–75.
Gino, F., & Bazerman, M. H. (2009). When misconduct goes unnoticed: The acceptabil- Rorty, R. (1997). In R. Bonketoe, & M. Stepaniants (Eds.), ‘Justice as a larger loyalty’ in
ity of gradual erosion in others' unethical behavior. Journal of Experimental Social justice and democracy: Cross-cultural perspectives (pp. 9–22). Hawaii: University
Psychology, 45, 708–719. of Hawaii Press.
Graham, J. (2011). Left gut, right gut: Ideology and automatic moral reactions. (Doctoral Shaw, A., DeScioli, P., & Olson, K. R. (2012). Fairness versus favoritism in children. Evo-
dissertation). : University of Virginia. lution and Human Behavior, 33, 736–745.
Graham, J., Haidt, J., & Nosek, B. (2009). Liberals and conservatives use different sets of Simmons, J., Nelson, L., & Simonsohn, U. (2012). A 21-word solution. Dialogue, 26, 4–12.
moral foundations. Journal of Personality and Social Psychology, 96, 1029–1046. Sloane, S., Baillargeon, R., & Premack, D. (2012). Do infants have a sense of fairness?
Graham, J., Nosek, B. A., Haidt, J., Iyer, R., Koleva, S., & Ditto, P. H. (2011). Mapping the Psychological Science, 23, 196–204.
moral domain. Journal of Personality and Social Psychology, 101, 366–385. Smetana, J. G., Killen, M., & Turiel, E. (1991). Children's reasoning about interpersonal
Haidt, J. (2007). The new synthesis in moral psychology. Science, 316, 998–1002. and moral conflicts. Child Development, 62, 629–644.
Haidt, J. (2012). The righteous mind: Why good people are divided by politics and religion. Vadera, A. K., Vadera, R. V., & Caza, B. B. (2009). Making sense of whistle-blowing's an-
New York, NY: Pantheon Books. tecedents: Learning from research on identity and ethics programs. Business Ethics
Haidt, J., & Joseph, C. (2004). Intuitive ethics: How innately prepared intuitions gener- Quarterly, 19, 553–586.
ate culturally variable virtues. Daedalus, 133, 55–66. Vogel, G. (2011). Psychologist accused of fraud on an astonishing scale. Science, 334, 579.
Hamlin, J. K., Mahajan, N., Liberman, Z., & Wynn, K. (2013). Not like me = bad: Infants von Neumann, J., & Morgenstern, O. (1947). Theory of games and economic behavior
prefer those who harm dissimilar others. Psychological Science, 24, 589–594. (2nd ed.)Princeton, NJ: Princeton University Press.
Hamlin, J. K., Wynn, K., Bloom, P., & Mahajan, N. (2011). How infants and toddlers react Walker, L. J., & Hennig, K. H. (2004). Differing conceptions of moral exemplarity: Just,
to antisocial others. Proceedings of the National Academy of Sciences of the United brave, and caring. Journal of Personality and Social Psychology, 86, 629–647.
States of America, 108, 19931–19936. Yong, E. (2012). The data detective. Nature, 487, 18–19.