Content uploaded by Celia Moore
Author content
All content in this area was uploaded by Celia Moore on Jan 02, 2018
Content may be subject to copyright.
PERSONNEL PSYCHOLOGY
2012, 65, 1–48
WHY EMPLOYEES DO BAD THINGS: MORAL
DISENGAGEMENT AND UNETHICAL
ORGANIZATIONAL BEHAVIOR
CELIA MOORE
London Business School
JAMES R. DETERT
S.C. Johnson Graduate School of Management
Cornell University
LINDA KLEBE TREVI ˜
NO
Smeal College of Business
Pennsylvania State University
VICKI L. BAKER
Albion College
DAVID M . MAYER
Stephen M. Ross School of Business
University of Michigan
We examine the influence of individuals’ propensity to morally disen-
gage on a broad range of unethical organizational behaviors. First, we
develop a parsimonious, adult-oriented, valid, and reliable measure of
an individual’s propensity to morally disengage, and demonstrate the
relationship between it and a number of theoretically relevant constructs
in its nomological network. Then, in 4 additional studies spanning lab-
oratory and field settings, we demonstrate the power of the propensity
to moral disengage to predict multiple types of unethical organizational
behavior. In these studies we demonstrate that the propensity to morally
disengage predicts several outcomes (self-reported unethical behavior, a
decision to commit fraud, a self-serving decision in the workplace, and
supervisor- and coworker-reported unethical work behaviors) beyond
other established individual difference antecedents of unethical organi-
zational behavior, as well as the most closely related extant measure
of the construct. We conclude that scholars and practitioners seeking
to understand a broad range of undesirable workplace behaviors can
benefit from taking an individual’s propensity to morally disengage into
account. Implications for theory, research, and practice are discussed.
A host of ethical debacles across a wide range of contexts has in-
spired growing interest in studying and understanding why individuals
engage in the kind of behavior that leads to enormous costs—trillions
Correspondence and requests for reprints should be addressed to Celia Moore, London
Business School, Regent’s Park, London NW1 4SA, U.K.; cmoore@london.edu.
C2012 Wiley Periodicals, Inc.
1
2 PERSONNEL PSYCHOLOGY
of dollars annually—for organizations and society. As just one exam-
ple, the Association of Certified Fraud Examiners recently estimated that
businesses globally suffer annual losses of $2.9 trillion as a result of
fraudulent activity (2010). This is a huge sum, indicating that unethical
behavior is far more widespread than suggested by the intense focus on the
few high-profile scandals that are covered by major news outlets. Thus,
being able to better understand and predict who is likely to engage in
such behavior—that is, behaviors within organizations that directly cause
direct harm to another individual or that violate widely accepted moral
norms in society—is crucial for organizational leaders and for societal
well-being.
Organizational scholars have begun to identify a number of important
contextual drivers of unethical organizational behavior, such as ethical
leadership (Brown & Trevi˜
no, 2006), ethical climate (Mayer, Kuenzi,
& Greenbaum, 2009), and codes of conduct (Weaver & Trevi˜
no, 1999).
However, research thus far has failed to explain a substantial propor-
tion of the variance in unethical organizational behavior using contex-
tual variables alone (Kish-Gephart, Harrison, & Trevi˜
no, 2010). A range
of individual-level factors has also been used to aid in the explanation
of why people engage in unethical organizational behavior. The list of
these antecedents is long, including Machiavellianism (Christie & Geis,
1970; Shultz, 1993; Siegel, 1973), moral identity (Aquino & Reed, 2002;
McFerran, Aquino, & Duffy, 2010), cognitive moral development (Am-
brose, Arnaud, & Schminke, 2008; Greenberg, 2002; Kohlberg, 1969),
moral philosophies (Bird, 1996; Forsyth, 1980), empathy (Davis, 1983;
Gino & Pierce, 2009), and moral affect (Eisenberg, 2000; Paternoster &
Simpson, 1996; Tangney, 1990). However, effect sizes for their role in pre-
dicting unethical workplace behavior are generally small, leaving much
variance unexplained (Kish-Gephart et al., 2010).
In this paper, we propose that an important additional driver of un-
ethical behavior is an individual’s propensity to morally disengage—
that is, an individual difference in the way that people cognitively pro-
cess decisions and behavior with ethical import that allows those in-
clined to morally disengage to behave unethically without feeling distress
(Bandura, 1990a, 1990b, 1999, 2002). Broadly speaking, we know that
how individuals processs, frame, or understand information relevant to
ethically meaningful decisions plays an important role in their ethical and
unethical choices (Kern & Chugh, 2009; Tenbrunsel & Messick, 1999),
and recent reviews of the behavioral ethics literature have suggested that
scholars should attend more carefully to the role of cognitive processes in
unethical behavior (Tenbrunsel & Messick, 2004; Tenbrunsel & Smith-
Crowe, 2008). In the set of studies reported here, we heed these calls by
CELIA MOORE ET AL. 3
establishing a new measure of the propensity to morally disengage as a
uniquely important predictor of a broad range of unethical behaviors.
This research makes two major contributions. First, though there has
been a surge of interest in the concept of moral disengagement in the
past decade or so, researchers have not yet offered the field a carefully
validated, parsimonious, and easily administered measure of the general
propensity to morally disengage. To date, studies of moral disengage-
ment have depended on idiosyncratic methods of assessment, with new
measures employed for each study, often without systematic development
(e.g., Detert, Trevi˜
no, & Sweitzer, 2008; McFerran et al., 2010) or with-
out tapping the construct in a comprehensive way (Aquino, Reed, Thau,
& Freeman, 2007). A number of scholars have designed measures with
a specific audience in mind, such as children (Bandura, Barbaranelli,
Caprara, & Pastorelli, 1996), athletes (Boardley & Kavussanu, 2007;
Corrion, Scoffier, Gernigon, Cury, & d’Arripe-Longueville, 2010), com-
puter hackers (Rogers, 2001), or racial minorities (Pelton, Gound, Fore-
hand, & Brody, 2004), or with a specific outcome in mind, such as support
for military force (McAlister, 2001), the death penalty (Osofsky, Bandura,
& Zimbardo, 2005), or violating one’s duties to civic society (Caprara,
Fida, Vecchione, Tramontano, & Barbaranelli, 2009). Our first goal isthus
to provide a measure that can be easily administered and used generally—
that is, with any adult sample in any type of context (though we focus
here on the workplace context)—to successfully predict a wide set of un-
ethical behaviors. In doing so, we address a long-standing concern among
organizational ethics scholars about the lack of valid and reliable scales
to use in research (Mayer et al., 2009; Tenbrunsel & Smith-Crowe, 2008;
Trev i ˜
no, Weaver, & Reynolds, 2006).
Second, research on moral disengagement has often been conducted
without an adequate understanding of its role within the existing land-
scape of individual predictors of unethical behavior. We thus propose how
the propensity to morally disengage should relate to other individual dif-
ference constructs of three specific types: (1) morally revelant personality
traits, (2) moral reasoning abilities and orientations, and (3) dispositional
moral emotions. We empirically position the propensity to morally dis-
engage within the landscape of these other constructs and demonstrate
that, compared to them, it is a more powerful predictor of four different
measures of unethical organizational behavior. Finally, given the strength
of the propensity to morally disengage as a predictor of unethical be-
havior, and the fact that most studies with other individual difference
predictors have not included a strong overarching theoretical framework,
we suggest that researchers consider adopting Bandura’s (1986) theory
of self-regulation as a conceptual framework that may lead to better
4 PERSONNEL PSYCHOLOGY
understanding of how these individual differences operate and collec-
tively explain unethical behavior.
In what follows, we present the rationale and results of five studies that
build on and complement one another. We used Study 1 to develop and
establish the baseline psychometric properties of a parsimonious, general
measure of the propensity to morally disengage. Then, in each of Studies
2–5, we used an array of unethical behavior measures as dependent vari-
ables (measured independently from the propensity to morally disengage
in each case) to test the general hypothesis that the propensity to morally
disengage will explain a significant proportion of the variance in uneth-
ical organizational behavior after controlling for other variables within
its nomological net. The alternate predictors included here are among the
most theoretically relevant and among the most common individual differ-
ences used in ethical decision-making studies. However, they are almost
never studied together nor are they adequately controlled for (as we do
here) when demonstrating the utility of an additional construct. As part
of this endeavor, we also show the superior ability of our new measure
to predict both supervisor and coworker reported employee unethical be-
havior over the most relevant extant measure of the propensity to morally
disengage in the workplace (McFerran et al., 2010). Together, these stud-
ies offer robust evidence of the power of an individual’s propensity to
morally disengage to predict a host of unethical behaviors of interest to
organizational scholars and leaders.
Theoretical Background
Albert Bandura introduced the theory of moral disengagement as an
extension of his more general social cognitive theory (Bandura, 1986:
375–389). According to social cognitive theory, when self-regulatory ca-
pabilities are working properly, transgressive behavior is deterred through
the self-condemnation individuals anticipate they would suffer were they
to engage in behavior that conflicts with their internalized moral standards.
Moral disengagement theory explains how this self-regulatory process can
fail when moral disengagement mechanisms disable the cognitive links
between transgressive behavior and the self-sanctioning that should pre-
vent it (Bandura, 1986: 375–389, Bandura, 1990a, 1990b, 1999, 2002).
The moral disengagement process is theorized to play an important role
in explaining how individuals are able to engage in human atrocities such
as political and military violence (Bandura, 1990a, 1990b) or corporate
wrongdoing and corruption (Bandura, Caprara, & Zsolnai, 2000; Brief,
Buttram, & Dukerich, 2001; Moore, 2008b) without apparent cognitive
distress.
CELIA MOORE ET AL. 5
Moral Disengagement Mechanisms
Bandura proposed that moral disengagement occurs through a set of
eight interrelated cognitive mechanisms that facilitate unethical behavior.
Moral justification, euphemistic labeling, and advantageous comparison
are three mechanisms of moral disengagement that serve to cognitively
restructure unethical acts so that they appear less harmful. Moral justi-
fication cognitively reframes unethical acts as being in the service of a
greater good. Illustrations include the justifying of military atrocities as
serving a worthy goal (Kramer, 1990; Rapoport & Alexander, 1982) or the
recasting of inappropriate behavior such as unfair treatment as appropriate
to protect friends or an organization. Euphemistic labeling is the use of
sanitized language to rename harmful actions to make them appear more
benign (Bolinger, 1982). For example, in corrupt organizations, those who
collude are often positively labeled “team players” (see Jackall, 1988:
52–53). Advantageous comparison exploits the contrast between a behav-
ior under consideration and an even more reprehensible behavior to make
the former seem innocuous (Bandura, 2002). For example, misrepresent-
ing small lies on expense reports can be viewed as more acceptable when
compared with more egregious expense report violations.
The displacement and diffusion of responsibility mechanisms obscure
the moral agency of the (potential) actor. Displacement of responsibility
refers to the attribution of responsibility for one’s actions to authority
figures who may have tacitly condoned or explicitly directed behavior
(see Kelman & Hamilton, 1989; Milgram, 1974; Sykes & Matza, 1957).
Diffusion of responsibility works in a similar way but refers to dispersing
responsibility for one’s action across members of a group (see the descrip-
tion of the lead-up to the Challenger disaster recounted by Vaughan, 1996).
Distortion of consequences, dehumanization, and the attribution of
blame mechanisms serve to reduce or eliminate the distress one perceives
to be causing a victim (Sykes & Matza, 1957). Distortion of consequences
describes the minimization of the seriousness of the effects of one’s ac-
tions, thus providing “little reason for the self-censure to be activated”
(Bandura, 1999b: 199). This is illustrated by descriptions of stealing from
a large, profitable organization as a “victimless crime” (Benson, 1985).
Dehumanization is the framing of the victims of one’s actions as unde-
serving of basic human consideration. This is fostered by defining others
as members of an outgroup who are unworthy of moral regard (Deutsch,
1990; Opotow, 1990). Finally, in attribution of blame, responsibility is
assigned to the victims themselves, who are described as deserving what-
ever befalls them (Bandura, 2002: 110). It has been shown to describe
the cognition underlying unethical behavior in many contexts, including
types of white-collar crime (see Douglas, 1995).
6 PERSONNEL PSYCHOLOGY
Although others have discussed or studied similar cognitive mech-
anisms separately (e.g., euphemistic language, diffusion of responsibil-
ity; see Ashforth & Anand, 2003; Diener, 1976; Kelman, 1973), Ban-
dura conceptualized these eight moral disengagement mechanisms as a
coherent set of cognitive tendencies that influence the way individuals
may approach decisions with ethical import. This approach—which has
characterized most of Bandura’s work and represents our perspective
here—argues that individuals will systematically differ in their propen-
sities to use cognitive moral disengagement mechanisms when fac-
ing decisions with ethical import. In the set of studies reported here,
we focus on the trait instantiation of moral disengagement, examin-
ing the propensity to morally disengage as a generalized cognitive ori-
entation to the world that differentiates individuals’ thinking in a way
that powerfully affects unethical behavior. This stance is consistent the
majority of the empirical work on moral disengagement to date, from
Bandura and colleagues’ (1996) study of the propensity to morally dis-
engage in children and adolescents to a number of subsequent studies
carried out by other researchers using a range of measures of the individ-
ual propensity to morally disengage (e.g., Boardley & Kavussanu, 2007;
Caprara et al., 2009; Detert et al., 2008; Duffy, Tepper, & O’Leary-Kelly,
2002; McAlister, 2001).
The Nomological Network of the Propensity to Morally Disengage
A nomological network is a conceptual model that situates a construct
of interest within the landscape of constructs that are theoretically related
to it (Schwab, 1980). Though identifying and testing relationships within
a nomological network is often narrowly understood as part of the pro-
cess of construct validation and measurement development (Chronbach
& Meehl, 1955), the work of specifying a construct’s nomological net-
work for measurement purposes also helps researchers fully understand a
construct’s theoretical implications. In this section, we describe the nomo-
logical network of the propensity to morally disengage and, in so doing,
explain why the propensity to morally disengage is likely to be a particu-
larly strong predictor of unethical organizational behavior relative to other
constructs that share common conceptual space.
We identify three important categories of constructs within the nomo-
logical network of the propensity to morally disengage: (1) morally rev-
elant individual personality traits, (2) moral reasoning abilities and ori-
entations, and (3) dispositional moral emotions. For each category and
construct, we explain the rationale for inclusion and the expected rela-
tionship between the nomological network factor, the propensity to moral
disengage (our focal construct), and the key criterion variable (unethical
CELIA MOORE ET AL. 7
behavior). Our intention was not to exhaustively tap every possible cat-
egory or construct of potential relevance to the propensity to morally
disengage but to choose theoretically salient and well-studied representa-
tive constructs within each of these three conceptual categories.
Morally Relevant Individual Traits
As noted earlier, our focus is on the dispositional propensity of indi-
viduals to morally disengage (Bandura, 1990a, 1990b). Thus, other stable
dispositions that have been found to influence ethical and unethical behav-
ior and that describe orientations toward or ways of seeing behaviors with
ethical ramifications should fall within its nomological network. Three
in particular have been consistently identified as important predictors
of ethical and unethical behavior (Hoffman, 2000; Kish-Gephart et al.,
2010): Machiavellianism (Christie & Geis, 1970), moral identity (Aquino
& Reed, 2002), and empathy (Davis, 1983).
Machiavellianism represents an individual’s propensity to be manipu-
lative and ruthless in the pursuit of self-interested goals (Christie & Geis,
1970). We reason that those high in Machiavellianism will be more in-
clined to morally disengage because such cognitive mechanisms present
one means by which Machiavellians can more readily pursue their own
interests without self-censure. Machiavellianism has been shown to be
positively related to many transgressive behavioral tendencies, includ-
ing antisocial behavior, lying, and willingness to exploit others (Christie
& Geis, 1970; Sakalaki, Richardson, & Th´
epaut, 2007), as well as to a
broad range of unethical decisions in a recent meta-analysis (Kish-Gephart
et al., 2010). However, we believe that the propensity to morally disengage
may be a particularly strong predictor of unethical organizational behav-
ior because it captures an individual’s general tendency to disengage from
the self-sanctions that would otherwise prevent a wide range of unethical
behaviors rather than the narrower and more specific behaviors associated
with Machiavellianism.
Moral identity describes the extent to which one’s self-concept in-
corporates the importance of being a moral person (Aquino & Reed,
2002). Moral identity has a strong relationship with prosocial behavior
and has also been linked with reduced unethical behavior (Aquino & Reed,
2002; Shao, Aquino, & Freeman, 2008). We expect moral identity to be
negatively correlated with the propensity to morally disengage because
individuals with a highly salient moral identity should be more concerned
about harm to others and more likely to take responsibility for their behav-
ior, thereby making it more unlikely that they would disengage the moral
self-regulatory function (c.f., Detert et al., 2008). Further, we believe that
the propensity to morally disengage will have stronger effects on behavior
8 PERSONNEL PSYCHOLOGY
because moral identity’s effect requires an activated moral self-concept
(Aquino & Reed, 2002). However, many of the mechanisms of moral
disengagement disrupt the activation of the self-concept by, for exam-
ple, focusing on the victim (dehumanization, attribution of blame) or by
reducing personal agency (diffusion and displacement of responsibility).
Trait empathy, which includes sympathetic feelings, responsiveness to
others, and an ability to cognitively understand others’ perspectives, has
received particular attention as an individual difference that contributes
to ethical behavior and reduces unethical behavior (Eisenberg, 1986;
Eisenberg & Miller, 1987; Hoffman, 2000; Tangney, 1991). We expect
trait empathy to be negatively related to the propensity to morally dis-
engage because those predisposed to morally disengage should be less
likely to take others’ viewpoints or feel compassionate towards them (De-
tert et al., 2008). Thus, those lower in trait empathy (and thus less likely
to feel compassionately towards others) will likely demonstrate higher
propensities to morally disengage because the latter often also involves
ignoring or distorting others’ feelings, needs, or perspective. However,
compared to trait empathy, morally disengaged reasoning is relevant to
a much broader set of situations in work organizations. Therefore, we
expect the propensity to morally disengage to have even greater general
utility in the prediction of unethical organizational behavior.
Moral Reasoning Abilities and Orientations
The ethical decision-making literature has traditionally been domi-
nated by rational/deliberative models of moral reasoning, which suggest
that individuals move from awareness to deliberative judgment to mo-
tivation/intention and then to action (Rest, 1986). Constructs describing
aspects of reasoning about ethical issues should relate to the propensity
to morally disengage because the disengagement process is also inher-
ently cognitive and influences, probably less consciously, the framing and
making of ethically charged decisions. The three main constructs that
represent different deliberative modes of ethical reasoning are cognitive
moral development,idealism, and relativism.
Cognitive moral development (CMD) is considered theoretically im-
portant to the judgment phase of ethical decision making—that is, the
point when an individual decides what is right or wrong in a particular
situation. CMD is conceptualized in terms of a series of stages through
which individuals progress as they become more cognitively advanced
and autonomous in their moral reasoning (Kohlberg, 1969, 1984; Piaget,
1965). Higher levels of CMD have been found to be negatively associated
with unethical choice (see Kish-Gephart et al., 2010), but the construct’s
potential relationship to the overall moral disengagement process or
CELIA MOORE ET AL. 9
specific mechanisms of disengagement has never been studied. We ar-
gue that CMD is conceptually distinct from the propensity to morally
disengage because the former is a measure of the level of sophistication
with which one consciously reasons through ethical quandaries, whereas
the propensity to morally disengage describes a dispositional tendency
to use cognitive mechanisms that disengage moral self-regulatory sanc-
tions. We propose that CMD will be negatively related to the propensity to
morally disengage because those who engage in more complex, principled
reasoning are more likely to quickly invoke moral principles (e.g., justice,
the greater good) to evaluate ethical dilemmas and to decide what is right.
They are also more likely to think through ethical dilemmas autonomously
rather than relying on others for guidance. Therefore, they should also be
less inclined to displace or diffuse responsibility onto others.
Despite its prominence in theory and research for decades, a recent
meta-analysis reported that CMD is only moderately predictive of uneth-
ical choices (Kish-Gephart et al., 2010). This may be because the ability
to reason deliberatively and in a sophisticated way about ethical deci-
sions does not always lead one to behave more ethically (Rest, 1986).
One possible explanation for this finding is that many unethical acts are
better explained by impulsive or intuitive models (Haidt, 2001) rather
than the deliberative model assumed by CMD. Although understanding
of the overall process of moral disengagement is still unfolding, moral dis-
engagement appears to often involve little conscious deliberation. Thus,
we would expect not only a negative correlation between the propensity
to morally disengage and CMD but also that moral disengagement is
potentially “in play” in a wider variety of situations (where conscious de-
liberation is not involved) and therefore will show a stronger relationship
with unethical behavior than CMD.
It is also important to examine the relationship between the propen-
sity to morally disengage and the most common ethical philosophies
that guide human decision making in ethical dilemma situations. Ethical
philosophies describe “stated beliefs or personal preferences for particu-
lar normative frameworks” (Kish-Gephart et al., 2010: 3). Idealism and
relativism describe two moral philosophies that reflect stable individual
orientations toward ethical decision making. Forsyth (1980; Schlenker
& Forsyth, 1977) described idealism as an individual’s belief that “the
‘right’ action [can] always be obtained” and relativism as the degree to
which an individual “rejects universal moral rules...when drawing con-
clusions about moral questions” (Forsyth, 1980: 175–176). We reason
that the propensity to morally disengage will be positively correlated with
relativism because holding a relativist position is facilitated by morally
disengaged cognitions. Conversely, the propensity to morally disengage
should be negatively correlated with idealism because idealists are driven
10 PERSONNEL PSYCHOLOGY
to pursue absolute ethical standards and thus not be motivated to find ways
(cognitively) to skirt them. In addition, similar to cognitive moral develop-
ment, idealism and relativism are part of a rational/deliberative approach
to ethical decision making and would only be involved in situations in-
volving conscious deliberation. As a result, we expect the propensity to
morally disengage to apply across a wider range of situations including
those that do not involve conscious ethical deliberation.
Dispositional Moral Emotions
Because moral disengagement facilitates engaging in unethical be-
havior without feeling distress (Bandura, 1990a), moral affect/emotion
constructs should also fall within the nomological network of the propen-
sity to morally disengage. Moral emotions, including anticipatory guilt
and shame, have been recognized as important motivating factors under-
lying ethical conduct (Haidt, 2003; Tangney, Stuewig, & Mashek, 2007),
but they have never been studied in relation to moral disengagement. The
propensity to morally disengage is conceptually distinct from moral affect
because the former describes a set of relatively “cool” cognitive processes
stemming from parts of the brain not primarily involved in the “hot”
processing included in moral affect. However, as described below, we
expect that the propensity to morally disengage will be related to certain
dispositional moral emotions.
Guilt has been well established as a correlate of ethical behavior
(Tangney et al., 2007: 354), in part because it elicits a sense of personal
responsibility for one’s actions (Tangney, 1991). We expect the propensity
to morally disengage to be negatively related to dispositional guilt—the
tendency to experience a set of negative emotions upon judging one’s own
actions as harmful or immoral (Lewis, 1971; Tangney, 1991)—because
guilt is activated when one’s self-sanctions against unethical behavior are
working correctly, and morally disengaged reasoning weakens that self-
sanction trigger. Shame, however, works differently. It is a set of negative
emotions about one’s self rather than one’s actions (Lewis, 1971; Tangney,
1991). Because the propensity to morally disengage represents a tendency
to use certain types of cognitions to distance oneself from one’s actions
(and thus not see behavior in specific situations as being “reflective of”
or “revealing of” the self), we expect the propensity to morally disengage
to be unrelated to dispositional shame. As noted above, an individual’s
self-sanctions against unethical behavior must be working correctly in
order for guilt to be activated. By contrast, the propensity to morally dis-
engage can operate across a wide array of situations and assumes that self-
sanctions may be disengaged. Therefore, we expect that the propensity
CELIA MOORE ET AL. 11
to morally disengage will be more consistently related to unethical orga-
nizational behavior than will guilt.
Hypotheses
Thus far, we have provided theoretical rationale for expected relation-
ships between the propensity to morally disengage and other constructs
within its nomological network as well as its distinctiveness from these
other constructs. However, our ultimate interest is in better understanding
what explains organizationally embedded unethical behavior—behavior
that is widespread enough to lead to the huge annual financial losses
mentioned at the outset of this paper. Thus, we turn next to the rationale
for our expectation that the propensity to morally disengage is a dis-
tinct and powerful predictor of the types of unethical behavior relevant to
organizations.
The workplace provides ample opportunities for moral disengage-
ment: Organizations tend to be hierarchical, providing opportunities for
the displacement of responsibility; work is often undertaken within teams,
providing opportunities for the diffusion of responsibility; organizational
membership automatically defines the boundaries of an in-group, pro-
viding opportunities for moral justification (to protect the organization)
and the cognitive minimization of the consequences of one’s actions for
those who are outside the organization (and thus in an out-group). The
propensity to morally disengage might also be particularly damaging
in organizational life because work contexts have been documented as
triggering amoral frames of judgment (Gross, 1978). As Jackall (1988)
pointed out in Moral Mazes, organizations are particularly effective at
assisting individuals in bracketing off moral schemas that guide be-
havior elsewhere. Thus, the propensity to morally disengage is likely
to be particularly relevant in the prediction of unethical behavior in
organizations.
Though the link has been suggested previously (Bandura et al., 2000;
Moore, 2008b), little research on the negative outcomes of moral dis-
engagement extends directly to unethical behavior in organizations (see
Duffy, Aquino, Tepper, Reed, & O’Leary-Kelly, 2005, for an exception).
Among the few empirical studies to date that have documented specific
negative outcomes of the propensity to morally disengage, a couple link
this propensity to organizationally relevant generic behaviors such as
cheating, lying, and stealing (Detert et al., 2008). However, no study has
yet shown that a carefully validated measure of the propensity to morally
disengage can predict organizationally relevant unethical behavior or un-
ethical behavior among employed adults, after controlling for a full array
of other constructs likely to be related strongly to either the propensity to
12 PERSONNEL PSYCHOLOGY
morally disengage itself or to unethical behavior. Thus, we propose and
test here not merely that the propensity to morally disengage is distinct
from the other types of morally salient individual differences discussed
above but that this propensity will predict a variety of types of unethical
behavior after controlling for these alternative explanations.
Hypothesis 1: The propensity to morally disengage will be positively
related to unethical behavior after controlling for other
morally salient individual traits.
Hypothesis 2: The propensity to morally disengage will be positively
related to unethical behavior after controlling for de-
liberative moral reasoning capacity and moral philoso-
phies.
Hypothesis 3: The propensity to morally disengage will be positively
related to unethical behavior after controlling for dis-
positional moral emotions.
Beyond establishing that the propensity to morally disengage is a
nonredundant predictor of unethical behavior, when introducing a new
measure it is also important to demonstrate its value relative to existing
measures of the same construct. As noted earlier, most of the few extant
measures of the propensity to morally disengage are not clearly geared
toward adults or designed to tap all the mechanisms of the propensity to
morally disengage. The measure closest to achieving all of these aims
is the one developed by Duffy and colleagues (Duffy et al., 2005; Duffy
et al., 2002; McFerran et al., 2010). This measure was developed to explain
undesirable and unethical behaviors at work, such as social undermining
of colleagues and organizational deviance (Duffy et al., 2005; Duffy et al.,
2002). Thus, we also examine how our new measure of the propensity to
morally disengage compares to this existing measure, including whether
the new measure still predicts unethical behavior in a workplace context
after accounting for the variance explained by the alternative measure.
Hypothesis 4: The new measure of the propensity to morally disen-
gage offered here will be positively related to unethi-
cal employee behavior after controlling for an existing
measure of the propensity to morally disengage.
Measure Development and Assessment
Our development and validation of a new measure of the propensity to
morally disengage was undertaken to address several specific limitations
of existing measures. First, the original measure developed by Bandura
and colleagues (1996) was developed specifically for use with children and
CELIA MOORE ET AL. 13
is therefore not appropriate for adults—items, for example, include refer-
ence to schoolyard pranks and classroom teasing. Second, we also wanted
to develop a measure that incorporates all of the mechanisms of moral
disengagement rather than one or a few mechanisms (e.g., moral justifi-
cation; see Aquino et al., 2007). Third, we wanted a measure that would
be appropriate for a broad sample of adults. Other measures have been
developed for a particular type of sample or context, which significantly
limits their general use (e.g., Boardley & Kavussanu, 2007; Caprara et al.,
2009; Corrion et al., 2010; McAlister, 2001; Pelton et al., 2004; Rogers,
2001). Fourth, we aimed to provide the first systematic documentation
of the convergent, discriminant, and predictive validity (across a wide
range of samples) of a measure of the propensity to morally disengage
(e.g., Detert et al., 2008; McFerran et al., 2010). The final version of the
measure we introduce here is also significantly more parsimonious (only
eight items) than existing measures (which range from 15 to 34 items;
e.g., Bandura et al., 1996; Detert et al., 2008; Duffy et al., 2005), making
it highly advantageous for use in field research.
Consistent with Bandura’s theoretical claim that moral disengagement
is best understood to be “multifaceted” (Bandura et al., 1996: 367), not
multifactorial, and in line with both his (e.g., Bandura et al., 1996) as
well as subsequent published and unpublished work on moral disengage-
ment (Detert et al., 2008; Duffy et al., 2005; Moore, 2008a), our aim was
to create a unidimensional measure of the general propensity to morally
disengage. That is, while acknowledging that the eight individual mecha-
nisms of moral disengagement represent different facets of the construct,
our overarching goal was to tap these facets as part of a valid scale that
assesses the general propensity to morally disengage as a higher order
concept.
We began by pretesting a large pool of items (74) in a sample of 454
full-time employed adults (26% male, Mage =36.3 years, SD =9.8 years)
to ensure ample representation of the construct’s broad content domain
and to assess the properties of various items when rated by a wide range
of adults. We chose items to represent each of the mechanisms of moral
disengagement, to be understandable to adults across multiple contexts
and cultures, and to represent cognitions about general behaviors (such
as lying, cheating, or stealing)—thus measuring a general propensity to
morally disengage—rather than behaviors in specific contexts that might
not be relevant to all adults (e.g., lying in a real estate negotiation). In
developing these items we were guided by Bandura’s theoretical descrip-
tions of the eight moral disengagement mechanisms (Bandura, 1990a,
1990b, 1999, 2002) as well as by previous measures (Bandura et al., 1996;
Detert et al., 2008; Duffy et al., 2005; Moore, 2008a).We wrote new items
when we judged that existing items for a particular moral disengagement
14 PERSONNEL PSYCHOLOGY
mechanism might not meet our criteria for developing a general scale
appropriate for most adults. We used the pretest to select items with the
highest factor loadings on a single factor and with the widest and least
skewed distributions. The pretest left us with 47 items to further assess in
Study 1.
Study 1
Study 1 was conducted to further examine the psychometric properties
of the set of items developed in the pretest and to determine how parsimo-
niously the propensity to morally disengage could be measured while still
ensuring that the multifaceted nature of the construct was represented by
the scale. Prior to making decisions about scale length, it was also neces-
sary to thoroughly examine the dimensionality of the construct—that is,
to ensure that moving forward with a unidimensional measure was empiri-
cally justified. Having done this, we used statistical indices and researcher
judgment to select items that created three increasingly shorter measures
of the propensity to morally disengage (a 24-item, 16-item, and 8-item
version). We then compared these three scales’ psychometric properties
and preliminary validity evidence with the goal of determining the most
parsimonious way the construct could be measured without unacceptably
compromising on reliability or validity standards.
Methods and Measures
Participants and procedure. One hundred and ninety four adults, who
were recruited to participate through a university-based behavioral lab in
the United Kingdom, completed a Web-based survey including measures
of some of the variables within the construct’s nomological network. The
sample was diverse in age (Mage =25.6, SD =6.7) and background
(see Table 1 for demographic details). One hundred and sixteen of these
participants also participated subsequently in a second study in the lab.
Participants were paid £5 for completing the Web-based survey and £10
for participating in the lab study.
The propensity to morally disengage. The 47 items developed in the
pretest were used to create and compare potential versions of a propensity
to morally disengage measure: a 24-item version (3 items per moral dis-
engagement mechanism), a 16-item version (2 items per mechanism), and
an 8-item version (1 item per mechanism). Appendix A lists all the items
and specifies which are included in each version of the scale. Because
our aim was to create a measure that represents a higher-order concept
tapping each of the eight specific mechanisms of moral disengagement,
we ran a factor analysis model forcing all items onto a single factor.
CELIA MOORE ET AL. 15
TABL E 1
Overview of Samples Used in Studies 1–5, With Descriptive Statistics
Gender Age
Female Male MSD Race/background
Study 1: U.K.-based
university
behavioral lab
participants
(n=194)
114 80 25.6 6.7 North America =15,
Europe =98, Asia
=50, Latin/South
America =2,
Middle East =2,
Africa =21,
other/declined to
answer =6
Study 2:
Northeastern U.S.
undergraduates
(n=272)
109 163 19.1 0.6 White =228,
African American
=5, Asian =14,
Hispanic =5,
foreign country
native =12,
other/declined to
answer =8
Study 3:
International
MBA students
(n=304)
78 226 28.5 2.2 North America =
61, Europe =107,
Asia =67,
Latin/South
America =39,
Middle East =14,
Africa =8,
Australia/New
Zealand =8
Study 4:
Northeastern U.S.
undergraduates
(n=250)
105 145 18.9 1.0 White =187,
African American
=5, Asian =25,
Hispanic =9,
foreign country
native =18,
other/declined to
answer =6
Study 5: Employees
in the southern
U.S. (n=141)
69 67125.7 9.8 White =87,
Hispanic or
Latino/a =26,
African American
=15, Asian =6,
biracial =2,
other/declined to
answer =5
1Five respondents declined to state their gender.
16 PERSONNEL PSYCHOLOGY
The three highest-loading items representing each moral disengagement
mechanism were selected as a 24-item measure of moral disengagement
(α=.90). This 24-item measure was then trimmed to 16 items by dropping
one item for each moral disengagement mechanism (α=.88). Items were
dropped based on a combination of statistical (e.g., factor loadings) and
theoretical (e.g., ensuring each tactic was most clearly represented by the
items) considerations. Finally, the scale was trimmed to eight items (α=
.80), with each moral disengagement mechanism represented by a single
item. The choice of which items to drop was in this case based primarily
on theoretical grounds, attempting to keep a highly representative item for
each mechanism while ensuring that the measure still had broad content
coverage and acceptable estimated reliability.
Convergent and discriminant validity measures. Machiavellianism
was measured using the standard 20-item Mach IV (Christie, 1970),
which asks respondents to rate their level of agreement with a series
of statements (on a 5-point continuum from strongly disagree to strongly
agree). A sample item is, “It is hard to get ahead without cutting corners
here and there.” Estimated reliability of the measure in this sample (α
=.65) was low (though low reliabilities in the Mach IV are typical, see
Ray, 1983; Wrightsman, 1991).The measure of moral identity (Aquino
& Reed, 2002) presents a set of characteristics of “moral” people (e.g.,
caring, compassionate, fair) as stimuli and then asks respondents (using
a 7-point continuum from strongly disagree to strongly agree) to assess
how important it is to be viewed as an individual who shares those charac-
teristics. We used the five items (α=.87) that assess “internalization” of
a moral identity because this component measures the strength of individ-
uals’ self-concept as moral people. A sample item (rated after seeing the
list of ‘moral characteristics’) is, “I strongly desire to have these charac-
teristics.” We assessed empathy using two components of Davis’ (1983)
Interpersonal Reactivity Index (IRI): a seven-item measure of perspective
taking (PT), or the ability to adopt others’ viewpoints, and a seven-item
measure of empathetic concern, or the tendency to feel compassion to-
wards others. Respondents reported the degree to which they feel the items
are true about themselves, from 1 =not at all true to 7 =very true.A
sample item in the perspective-taking subscale (α=.75) is, “I try to look
at everybody’s side of a disagreement when I make a decision,” and in the
EC subscale (α=.78) is, “I am often quite touched by things that I see
happen.”
Because social desirability can affect reporting in studies of ethics-
related beliefs and behaviors, we wanted to ensure that our new measure
of the propensity to morally disengage was not highly correlated with
a measure of social desirability bias. Thus all participants completed
a 10-item version of the traditional 33-item Marlowe-Crowne measure
CELIA MOORE ET AL. 17
TABL E 2
Summary of Fit Indices for all CFA Models, Study 1
χ2df χ2/df RMSEA SRMR NNFI CFI
24 item measures
24 items on 1 factor 588 252 2.54 .090 .080 .90 .91
24 items on 8 factors 507 224 2.26 .089 .078 .90 .91
24 items on 8 factors on 1 second-order 560 244 2.30 .092 .081 .91 .92
16 item measures
16 items on 1 factor 254 104 2.44 .099 .075 .91 .93
16 items on 8 factors 193 76 2.54 .098 .070 .91 .94
16 items on 8 factors on 1 second-order 244 96 2.54 .100 .074 .91 .93
8 item measure
8 items on 1 factor 27 20 1.35 .045 .043 .98 .99
of social desirability (Crowne & Marlowe, 1960; Strahan & Gerbasi,
1972). For this measure, participants responded to a series of yes/no
questions about negative behaviors or cognitions that are thought to be
universal but socially embarrassing or unattractive (e.g., “I never resent
being asked to return a favor”) and therefore potentially elicit socially
desirable responses. The number of times that a participant answers “no”
is summed as a measure of his/her social desirability response bias.
Results
Prior to comparing results for three single-factor measures of differing
length, we first thoroughly explored the dimensionality of these three
measures of the propensity to morally disengage in a multistep process.
We began by estimating a series of confirmatory factor analysis models
using LISREL 8.8 (J¨
oreskog & S¨
orbom, 2009) for the two longer versions
of the scale: one that forced all the items to load onto a single latent variable
(a 1-factor model), one that forced the items to load onto eight different
factors representing each of the moral disengagement mechanisms (an
8-factor model), and one model where eight first-order factors (comprised
by forcing the items tapping each moral disengagement mechanism onto
separate factors) were forced to load onto a second-order latent variable
(the second-order factor model). The resultant fit indices for each of these
models are presented in Table 2. The first thing the table makes clear
is that all of the models fit the data adequately, with RMSEA values
less than .10, SRMR values less than .08, and CFI and NNFI indices all
greater than .90. In particular, the fit indices for a single-factor model run
using the 8-item version of the scale indicate a good fit of the data to the
model (χ2=26, df =20, ns, NNFI =.98, CFI =.99, SRMR =.04,
18 PERSONNEL PSYCHOLOGY
RMSEA =.05) according to current standards—values of less than .05
for the RMSEA (Browne & Cudeck, 1993; Hu & Bentler, 1999) and CFI
and NNFI values greater than .95 (Hu & Bentler, 1999).
The second thing evident from the factor analyses results summarized
in Table 2 is that statistical criteria alone do not provide strong evidence
in this case for preferring one factor structure over another. For the 16-
and 24-item versions of the scale, the CFA models that fit the data to eight
separate factors have very slightly better fit indices (only models with
the same observed variables are directly comparable). This is an expected
result given the two or three items assessing each specific mechanism are
intended to be more interrelated with each other than with those items
assessing the other mechanisms. Conversely, the fit indices for the models
that represent the best match to Bandura’s theoretical expectation that
moral disengagement should be conceived of as “multifaceted” (Bandura
et al., 1996: 367)—that is, as a higher-order construct that encompasses
eight different mechanisms in a unified way—are slightly lower than the
other models run with the same number of items. This slight decrement
in fit (compared to models involving only a first-order factor) is expected
mathematically (Marsh, 1987) and thus does not necessarily have practical
import. The simplest latent variable models for the 16- and 24-item ver-
sions of the scale—those that load all items onto a single first-order latent
factor—have fit indices in between those for the 8-factor and second-
order factor models. However, differences in fit on the various indices are
of an average magnitude of just .01, a size considered to indicate mea-
surement invariance between comparable models (Cheung & Rensvold,
2002). With this much similarity in fit among the comparable models,
along with Bandura’s theorization, the rule of parsimony suggests select-
ing a single-factor structure for this new measure of the propensity to
morally disengage (Kline, 1988).
To further assess the appropriateness of proceeding with a single-factor
solution (i.e., a unidimensional approach), we then examined whether
measuring the propensity to morally disengage in various multidimen-
sional and unidimensional ways resulted in any substantive empirical
differences. Specifically, we created several measures of the propensity
to morally disengage that were either (a) latent variable measures derived
from the factor loadings from the various first-order or second-order CFA
models or (b) simple average measures derived by taking the average of
the items forming the scale. The latent variable approach allows for differ-
ential weighing of each item and of each moral disengagement mechanism
(in the second-order model), whereas the simple average approach (where
each item is assigned equal weight) assumes a unifactorial structure of
the data with all items representing the single underlying factor equally
well. The three versions of the 24-item measure—one an average of the
CELIA MOORE ET AL. 19
scale items and the other two latent variable measures derived from the
factor loadings for a single factor model, and the hierarchical second-
order factor model, respectively—are all correlated with each other at r=
.98 or higher.1The three parallel versions (simple average and first- and
second-order latent variables) created using 16 items and the two paral-
lel versions (simple average and first-order latent variable) created using
eight items likewise have bivariate correlations all approaching 1.0. These
results also suggest that there is no practical difference, at any of the three
scale lengths, between the simplest item-average unidimensional mea-
sures and those that take into account differences in the weights assigned
to particular moral disengagement items or mechanisms in the formation
of latent variables.
Having concluded on statistical and practical grounds that measuring
the propensity to morally disengage in more complex ways produces no
meaningful advantage, we proceeded in a third step to compare different
length (24-, 16- and 8-item) versions of the propensity to morally dis-
engage scale constructed as simple averages of the items. As shown in
Table 3, the intercorrelations among the three different-length versions
of the scale are all above .90. Moreover, the correlations of the three
different-length versions of the propensity to morally disengage scales
with the other constructs assessed in Study 1 are strongly consistent with
theory and predictions. For example, the 8-item propensity to morally
disengage measure correlates positively with Machiavellianism (r=.44,
p<.01) and negatively with moral identity (r=−.55, p<.01) and two
facets of empathy: perspective taking (r=−.40, p<.01) and empathetic
concern (r=−.46, p<.01). This measure is also uncorrelated with social
desirability (r=.05, ns). All of these relationships are in the expected
direction, and none is so strong as to suggest that the propensity to morally
disengage is redundant with any of the other constructs. Importantly, the
correlations between the 8-, 16- and 24-item versions of the propensity to
morally disengage measure and all of the nomological network variables
assessed in Study 1 are substantively equivalent—that is, all are in the
hypothesized directions and of similar magnitude. Thus, we determined
that we could meet our objective of creating a more parsimonious measure
of the propensity to morally disengage than has previously been offered
1We did not include in these comparisons (for the 24- and 16-item scale versions) eight
first-order latent variables representing each moral disengagement mechanism separately
for two reasons: (1) the second-order latent variable is essentially doing this in the creation
of the eight first-order latent variables subsequently loaded onto the second-order latent
variable, and this approach better represents the theory of the propensity to morally dis-
engage as a single higher order construct with subdimensions; (2) reliabilities for the two-
and three-item scales representing each moral disengagement mechanism that result from
this approach are systematically far below the conventional cutoff of .70.
20 PERSONNEL PSYCHOLOGY
TABL E 3
Means, Standard Deviations, Correlations Among Study 1 Variables
Varia b l e MSD 123 4567
1. Propensity to
morally disengage
(8-items)a
2.57 .99 (.80)
2. Propensity to
morally disengage
(16-items)a
2.88 .92 .93∗∗ (.88)
3. Propensity to
morally disengage
(24-items)a
2.99 .88 .90∗∗ .98∗∗ (.90)
4. Machiavellianisma2.83 .42 .44∗∗ .52∗∗ .54∗∗ (.65)
5. Moral identitya6.13 1.01 −.55∗∗ −.54∗∗ −.54∗∗ −.34∗∗ (.87)
6. Perspective
takinga
4.80 .91 −.40∗∗ −.36∗∗ −.38∗∗ −.29∗∗ .38∗∗ (.75)
7. ECa5.22 .91 −.46∗∗ −.46∗∗ −.47∗∗ −.31∗∗ .56∗∗ .38∗∗ (.78)
8. Social
desirabilitya
3.65 1.54 .05 .00 −.01 .04 −.15∗−.16∗.00
aN=194, bN=116.∗∗p<.01, ∗p<.05. Where appropriate, alpha reliabilities appear
along the diagonal.
in the literature by proceeding with the 8-item version of the scale (which
has estimated reliability of .80 in Study 1).
To statistically confirm that our new 8-item measure of the propensity
to morally disengage is not redundant with related constructs, we followed
recommendations by Schwab (1980) and DeVellis (1991) in considering
results from several confirmatory factor analysis models estimated using
the eight items from the propensity to morally disengage measure as well
as all items assessing the other nomological network constructs included in
Study 1—namely, Machiavellianism, moral identity, perspective taking,
and empathetic concern. We compared fit indices for the hypothesized
model—that is, a five-factor model where we forced all the items to load
onto factors representing their theorized construct—with those for several
4-factor models where we forced the propensity to morally disengage
items to load on the same factor as the items from one of the related
constructs (i.e., a 4-factor model forcing the moral disengagement items
to load with the items for Machiavellianism, a 4-factor model forcing the
moral disengagement items to load with the moral identity items, etc.).
If any of the 4-factor models fit the data significantly better than the 5-
factor model, it would indicate that our new measure may be redundant
with another measure. Instead, chi-square difference tests indicate that
the data fit the hypothesized five-factor model (with the items measuring
CELIA MOORE ET AL. 21
the propensity to morally disengage restricted to their own latent factor;
χ2=1551, df =979) significantly better (p<.001 for all comparisons)
than any of the models where the moral disengagement items were forced
to load with Machiavellianism (χ2=1800, df =983), moral identity
(χ2=1723, df =983), perspective taking (χ2=1699, df =983), or
empathetic concern (χ2=1650, df =983). These results provide further
evidence that the new propensity to morally disengage scale is empirically
distinct from related constructs.
We therefore concluded from Study 1 that the propensity to morally
disengage can be appropriately assessed with the new 8-item unidimen-
sional scale. Appendix A presents these eight items (in bold), as well as
the other sixteen items used to form the 16- and 24-item versions also
considered in Study 1. In Studies 2–5, we continued the assessment of
the 8-item version’s psychometric properties, with particular focus on its
distinctness from and incremental validity beyond additional constructs
in its nomological network. Specifically, we used four separate samples to
test whether the new 8-item version of the propensity to morally disengage
scale predicts multiple types of unethical decisions after controlling for
sets of measures of constructs from each of the three nomological network
categories previously discussed, as well as a longer extant measure of the
propensity to morally disengage.
Incremental Predictive Validity Studies
Study 2
Study 2 tested Hypothesis 1—that the propensity to morally disen-
gage will have incremental validity in the prediction of unethical decision
making, after controlling for four morally relevant individual trait dif-
ferences (Machiavellianism, moral identity, and two facets of empathy)
and socially desirable response tendencies. The dependent variable was a
self-report measure of several types of unethical behavior (cheating, lying,
and stealing).
Methods and Measures
Participants and procedure. For extra credit, 272 students at a univer-
sity in the Northeastern U.S. completed one survey (with the propensity
to morally disengage and other independent variable measures) midway
through an undergraduate business course (not an ethics course); 242
of these students completed a second survey (including the unethical
behavior measure) 1 month later. Demographic data are presented in
Table 1.
22 PERSONNEL PSYCHOLOGY
The propensity to morally misengage. At Time 1, participants com-
pleted the 8-item propensity to morally disengage scale (α=.76 in
this sample). Fit indices from a CFA indicate a good fit of the data
to a 1-factor model (e.g., CFI =1.0, NNFI =1.0, RMSEA =0,
SRMR =.03).
Control variables. As in Study 1, Machiavellianism was measured
using the 20-item Mach IV (Christie, 1970, α=.69 in this sample), moral
identity was measured using Aquino and Reed’s (2002) 5-item internal-
ization measure (α=.84 in this sample), and perspective taking and
empathetic concern were measured using the two seven-item subscales
from Davis’ (1983) IRI (αs=.75 and .78, respectively, in this sample).
All participants also completed the same 10-item measure of social desir-
ability as in Study 1 (Strahan & Gerbasi, 1972).
Dependent variable. We first sought to test whether the new propen-
sity to morally disengage measure would show incremental validity in the
prediction of a broad range of unethical behavior, and thus used Detert and
colleagues’ (2008) measure of cheating, lying, and stealing. This measure,
obtained 1 month after the initial survey, asks respondents to indicate how
often they have engaged in 13 unethical behaviors (e.g., “Taking low-cost
items from a retail store”) using a 5-point scale ranging from never to
many times (α=.82).
Results
Means, standard deviations, alpha reliabilities (where appropriate),
and bivariate correlations for the Study 2 variables are reported in Table 4.
The correlations between the propensity to morally disengage scale and
other measures are consistent with expectations and with the results from
Study 1. In this sample as well, the propensity to morally disengage cor-
relates positively with Machiavellianism (r=.46, p<.01) and negatively
with moral identity (r=−.42, p<.01) and two facets of empathy: per-
spective taking (r=−.33, p<.01) and empathetic concern (r=−.48, p
<.01). Again, none of these relationships are so strong as to indicate that
the propensity to morally disengage is redundant with any of the other
constructs. We further confirmed this by estimating and comparing the re-
sults of five CFA models: one where the propensity to morally disengage
items and those tapping the other four nomological network variables are
all represented by their own latent factors (χ2=1950, df =1024), and
four that force the propensity to morally disengage items to load with
Machiavellianism (χ2=2010, df =1029), moral identity (χ2=2233,
df =1029), perspective taking (χ2=2290, df =1029), or empathetic
concern (χ2=2152, df =1029), respectively. Chi-square difference tests
CELIA MOORE ET AL. 23
TABL E 4
Means, Standard Deviations, Correlations Among Study 2 Variables
Varia b l e MSD 1234567
1. Propensity to
morally
disengagea
2.79 .87 (.76)
2. Machiavellianisma2.77 .41 .46∗∗ (.69)
3. Moral identitya6.29 .92 −.42∗∗ −.44∗∗ (.84)
4. Perspective
takinga
4.49 .98 −.33∗∗ −.21∗∗ .25∗∗ (.81)
5. Empathetic
concerna
5.02 .94 −.48∗∗ −.43∗∗ .55∗∗ .35∗∗ (.79)
6. Social
desirabilitya
3.61 1.33 −.16∗∗ .07 −.04 −.07 −.07
7. Self-reported
cheating, lying,
and stealingb
1.83 .50 .31∗∗ .20∗∗ −.21∗∗ −.18∗∗ −.16∗−.08 (.82)
aN=272 at Time 1, bN=245 at Time 2.∗∗p<.01, ∗p<.05. Where appropriate, alpha
reliabilities appear along the diagonal.
confirm that, in every case, the data fit the model with a separate factor for
the propensity to morally disengage significantly better (p<.001 for all
comparisons), providing further evidence that the propensity to morally
disengage is empirically distinct from related constructs.
The positive and significant bivariate relationship (r=.31, p<.01)
with self-reported cheating, lying, and stealing behavior provides initial
evidence of criterion validity for the new propensity to morally disen-
gage scale. In fact, the propensity to morally disengage has the strongest
bivariate relationship (among the Study 2 variables) with this measure
of unethical behavior. To test Hypothesis 1, we examined the additional
variance accounted for by the propensity to morally disengage in the
prediction of self-reported cheating, lying, and stealing, beyond that ex-
plained by four other morally relevant individual differences and social
desirability. In the first step of a linear regression (see the first and sec-
ond columns of Table 5), we entered the measures of Machiavellianism,
moral identity, perspective taking, empathetic concern, and social desir-
ability. In the second step, we added the propensity to morally disengage
measure. In neither step are any of the control variables individually pre-
dictive of self-reported cheating, lying, and stealing. However, in support
of Hypothesis 1, the propensity to morally disengage significantly predicts
self-reported unethical behaviors (β=.22, p<.01). Adding this variable
to the equation increases R2by 3% (F[1,236] =7.96, p<.01).
24 PERSONNEL PSYCHOLOGY
TABL E 5
Incremental Validity of the Propensity to Morally Disengage Over Other Ethical Decision Making Predictors (Studies 2–4)
Study 4
Study 2 Study 3 Decision to protect one’s
Self-reported cheating, Decision to intentionally self-interest over treating
lying, and stealing misrepresent facts to client subordinates fairly at work
Model 1 Model 2 Model 1 Model 2 Model 1 Model 2
1. Machiavellianism .11 .07
Moral identity −.14 −.10
Perspective taking −.13 −.09
Empathetic concern .00 .06
Social desirability −.08 −.04
2. Propensity to morally disengage .22∗∗
R2.08 .11
R2.03∗∗
1. CMD .99 (.97–1.01) 1.00 (.98–1.02)
Idealism .84 (.67–1.06) .92 (.72–1.17)
Relativism 1.00 (.78–1.28) .86 (.68–1.14)
Social desirability .95 (.74–1.22) .97 (.76–1.24)
2. Propensity to morally disengage 2.15 (1.35–3.43)∗∗
Model chi square (df) 3.16 (4) 14.06 (5)
chi square (df) 10.90 (1)∗∗
1. Dispositional shame 1.75 (.68–4.53) 1.41 (.534–3.74)
Dispositional guilt .17 (.06−.47)∗∗ .29 (.09−.86)∗
Social desirability .73 (.53–1.00) .71 (.52−.98)∗
2. Propensity to morally disengage 1.93 (1.13–3.31)∗
Model chi square (df) 14.21 (3) 20.14 (4)
chi square 5.84 (1)∗
Note. Odds ratios (Exp[B]) and 95% confidence intervals are presented for logistic regressions. Standardized coefficients are presented for the linear
regression. N=245 for Study 2, N=248 for Study 3, and N=225 for Study 4. ∗∗ p<.01, * p<.05.
CELIA MOORE ET AL. 25
Study 3
Study 3 tested Hypothesis 2—that the propensity to morally disengage
will show incremental validity in the prediction of unethical behavior after
controlling for three deliberative moral reasoning or ethical philosophy
factors (cognitive moral development, idealism and relativism) and so-
cially desirable response tendencies.
Methods and Measures
Participants and procedure. Three hundred and four students (74%
male, Mage =28.5 years, SD =2.2 years) in an international MBA program
participated in the study. Participants had an average of 5.6 years of work
experience prior to entering the program (SD =1.9) (see Table 1 for
additional demographics). Respondents completed a survey with the 8-
item propensity to morally disengage scale and other measures as part
of their course participation in a class on ethics and corporate social
responsibility. To minimize contamination of results due to the class’
subject matter, they completed the survey before the class began. The
measure of unethical behavior (the dependent variable) was collected as
the class began (Time 2), as part of the first assignment.
The propensity to morally disengage. We used the same 8-item
propensity to morally disengage scale developed in Study 1 (α=.70
in this sample). The fit indices from a confirmatory factor analysis again
revealed a good fit of the data to a 1-factor model (CFI =.99, NNFI =
.99, RMSEA =.02, SRMR =.04).
Control variables. Cognitive moral development was measured us-
ing the Defining Issues Test (DIT; Rest, 1986, 1990). The original test
presents respondents with six moral dilemmas, asking them to rank the
four most important considerations they would use in making a decision
about what to do in each dilemma. A score, which is computed from those
considerations ranked highest, is considered a measure of the proportion
of one’s reasoning that is at the highest (principled) level of moral rea-
soning. We used a short form version of the DIT that uses three of the six
scenarios; this version has been shown to correlate above .90 with scores
on the longer version (Rest, 1990).
Idealism and relativism were measured using Forsyth’s Ethics Position
Questionnaire (1980), which measures an individual’s level of agreement
with general statements on a 9-point continuum, ranging from completely
disagree to completely agree. Ten items measured idealism (α=.85,
sample item: “It is never necessary to sacrifice the welfare of others”) and
ten items measured relativism (α=.80, sample item: “What is ethical
26 PERSONNEL PSYCHOLOGY
TABL E 6
Means, Standard Deviations, Correlations Among Study 3 Variables
Varia b l e MSD 123 45
1. Propensity to morally disengagea2.12 .77 (.70)
2. Cognitive moral developmenta40.44 16.88 −.23∗∗
3. Idealisma5.84 1.40 −.27∗∗ .01 (.85)
4. Relativisma4.82 1.31 .29∗∗ −.10 −.18∗∗ (.80)
5. Social desirabilitya3.49 1.34 −.05 −.03 .03 −.03
6. Unethical decision (dummy
variable coded 1 if decided
to send fraudulent fax)b
.18 .23∗∗ −.07 −.10 .01 .05
aN=304 at Time 1, bN=258 at Time 2.∗∗p<.01, ∗p<.05. Where appropriate, alpha
reliabilities appear along the diagonal.
varies from one situation and society to another”). All participants also
completed the same social desirability measure used in Studies 1 and 2.
Dependent variable. The unethical outcome used in this study was
developed by coding MBA students’ responses to an ethically charged case
assignment (“Conflict on a Trading Floor,” Badaracco & Useem, 2006)
that asks participants to put themselves in the position of a junior trader at
a large bank whose supervisor is requesting them to send a fax containing
misleading and, in fact, fraudulent information to a client. Students were
asked (prior to the beginning of the class) to write one page explaining
what they would do if they were in the same situation and why. Responses
ranged from sending the fax, to voicing concerns to others, to refusing
to send the fax. These decisions about the case were coded by a trained
independent coder. An unethical decision was represented by a dummy
variable coded “1” if an individual explicitly stated that s/he would send
the fax (18% of the sample stated this).
Results
Means, standard deviations, alpha reliabilities (where appropriate),
and correlations among the Study 3 variables are reported in Table 6.
As expected, the propensity to morally disengage is modestly negatively
correlated with cognitive moral development (r=−.23, p<.01) and
idealism (r=−.27, p=<.01) and positively correlated with relativism
(r=.29, p<.01). Again, a CFA model where the moral propensity to
morally disengage items are forced onto their own latent factor (χ2=
827, df =347) fit significantly better (both chi-square difference tests
significant at p<.001)than models forcing the propensity to morally
disengage items to load on the same factor with the idealism items
CELIA MOORE ET AL. 27
(χ2=2010, df =1029) or the relativism items (χ2=2233, df =1029).
The computation of cognitive moral development makes it impossible to
include it in CFA models. But, as the least strongly correlated with the
propensity to morally disengage of the three nomological network vari-
ables included in this study (see Table 6), it is reasonable to conclude from
the other results that these two constructs are also distinct.
As predicted, the propensity to morally disengage is positively corre-
lated with the subsequently measured unethical decision to send the fax
(r=.23, p<.01). In fact, the propensity to morally disengage is the
only variable in this study that is significantly related to the dependent
variable. To test Hypothesis 2, we examined the variance accounted for,
beyond that explained by three moral reasoning and ethical philosophy
factors and social desirability, by the propensity to morally disengage in
the prediction of a work-related unethical decision. We used logistic re-
gression to examine the difference between the chi-square statistics for the
more and less restricted models (i.e., models that do and do not include
the propensity to morally disengage). In the first step of the regression
(see Column 3 of Table 5), the decision to send the fax was regressed onto
the measures of cognitive moral development, idealism, relativism, and
social desirability. In the second step (Column 4, Table 5), the propensity
to morally disengage measure was added to the equation. The difference in
the chi-square statistic between the first and second models is significant
(χ2=10.90 [1 d.f.], p<.01). Adding the variable for the propensity to
morally disengage (Exp[B] =2.15, p<.01) in the second step contributes
significantly to the prediction of whether an individual is likely to make
the unethical decision. An odds ratio of 2.15 for moral the propensity to
morally disengage means that, holding cognitive moral development, ide-
alism, relativism, and social desirability constant, for every one unit (on
the 1–7 scale) increase in an individual’s measured level of the propensity
to morally disengage, he or she is 115% more likely to send the fraudu-
lent fax. This result provides support for Hypothesis 2. Specifically, these
results provide evidence that the propensity to morally disengage predicts
an unethical decision beyond three important independent variables that
represent the traditional rational/deliberative approach to ethical decision
making.
Study 4
Study 4 tested Hypothesis 3—that the propensity to morally disengage
will have incremental validity in the prediction of unethical behavior, mea-
sured here as a self-serving decision in a work context, after controlling for
two measures of moral affect (dispositional shame and guilt) and socially
desirable response tendencies.
28 PERSONNEL PSYCHOLOGY
Methods and Measures
Participants and procedure. For extra credit, 250 undergraduate stu-
dents at a university in the Northeastern U.S. completed one survey (with
the propensity to morally disengage and social desirability measures)
midway through a business course (Time 1), and 225 of these students
completed a second survey (containing the other measures, including the
dependent variable) 1 month later (Time 2). Demographic data are pre-
sented in Table 1.
The propensity to morally disengage. All participants completed the
8-item propensity to morally disengage measure at Time 1 (α=.77 in this
sample). Fit indices from a confirmatory factor analysis again revealed
a good fit of the data to a one-factor model (CFI =.99, NNFI =.98,
RMSEA =.04, SRMR =.04).
Control variables. Dispositional shame and guilt were measured us-
ing the Test of Self-Conscious Affect (TOSCA; Tangney, Wagner, &
Gramzow, 1989). The scale is comprised of 16 brief scenarios (for exam-
ple, “You are driving down the road, and you hit a small animal”); after
each scenario there are two statements, one tapping guilt (in this exam-
ple, “You would feel bad if you hadn’t been more alert driving down the
road”) and the other tapping shame (in this example, “You would think:
‘I’m terrible.’”). The mean of each set of eight statements (assessed on
a 5-point scale ranging from not very likely to very likely) is considered
a measure of respondents’ dispositional guilt (α=.77) and shame (α=
.70), respectively. To control for socially desirable responding, all partici-
pants also completed the same 10-item measure of social desirability used
in Studies 1–3.
Dependent variable. The unethical outcome in this study was op-
erationalized using responses to an exercise that allowed respondents to
make a work-related self-serving decision. Participants read a one-page
description of a managerial dilemma in which they are asked to play the
role of a manager who discovers that a subordinate—though they do not
know which one—has made an error that will cost the company $90,000.
The manager’s own supervisor then demands that he or she either per-
sonally take the blame (resulting in a loss of 50% of his/her own bonus)
or blame a member of his or her team (costing the team member 50% of
his or her bonus). The choice was coded as a self-serving decision if the
respondent chose the latter option (11% made this choice) because the
manager in the exercise cannot fairly blame any specific team member;
doing so saves the manager’s (i.e., the respondent’s) bonus at the expense
of his or her subordinate.
CELIA MOORE ET AL. 29
TABL E 7
Means, Standard Deviations, Correlations Among Study 4 Variables
Varia b l e MSD 1234
1. Propensity to morally
disengagea
2.74 .86 (.77)
2. Dispositional shamea3.01 .51 .01 (.70)
3. Dispositional guilta3.98 .46 −.47∗∗ .31∗∗ (.77)
4. Social desirabilitya3.69 1.34 .08 −.02 −.14∗
5. Self-interested
decision (dummy
variable coded 1)b
.11 .23∗∗ .00 −.20∗∗ −.09
aN=250 at Time 1, bN=225 at Time 2.∗∗p<.01, ∗p<.05. Where appropriate, alpha
reliabilities appear along the diagonal.
Results
Means, standard deviations, alpha reliabilities (where appropriate),
and correlations among the Study 4 variables are reported in Table 7.
Bivariate relationships are consistent with expectations. The propensity
to morally disengage correlates negatively with dispositional guilt (r=
−.47, p<.01) and is uncorrelated with dispositional shame (r=.01,
ns). As in Studies 2 and 3, CFA models forcing the propensity to morally
disengage items to load onto their own latent factor (χ2=1641, df =737)
result in significantly better fit indices (both chi-square difference tests
are significant at p<.001)than models forcing the propensity to morally
disengage items to load onto factors measuring shame (χ2=1797, df =
739) or guilt (χ2=1761, df =739).
The positive relationship between the propensity to morally disengage
and the self-serving decision at work (r=.23, p<.01) provides further
evidence of criterion validity for the new propensity to morally disengage
scale. Again, the propensity to morally disengage has the strongest bi-
variate relationship (among the independent variables) with the outcome
measure. To test Hypothesis 3, we examined the additional variance ac-
counted for, beyond that explained by the measures of moral affect and
social desirability, by the propensity to morally disengage in the predic-
tion of the self-interested work decision. We used logistic regression to
examine the difference between the chi-square statistics for a more and
less restricted model (i.e., models that do and do not include the propensity
to morally disengage). In the first step of the regression (see the Column
5 of Table 5), the self-serving decision was regressed onto dispositional
shame, guilt, and social desirability. In the second step (Column 6 of
30 PERSONNEL PSYCHOLOGY
Table 5), the measure of the propensity to morally disengage was added
to the equation. The difference in the chi-square statistic between the first
and second models (χ2=5.84 [1 d.f.], p<.05) is significant. Adding
the propensity to morally disengage measure (Exp[B] =1.93, p<.05)
in the second step adds significantly to the prediction of the self-serving
decision. An odds ratio of 1.93 for the propensity to morally disengage
means that, holding shame, guilt, and social desirability constant, for each
one unit increase in an individual’s measured level of the propensity to
morally disengage, he or she is 93% more likely to make a self-interested
decision at the expense of a subordinate. This result provides support for
Hypothesis 3.
Study 5
Study 5 tested Hypothesis 4—that the propensity to morally disengage
will have incremental validity in the prediction of unethical behavior,
measured here as the unethical work behaviors of current employees as
rated by both their peers and supervisors, after controlling for the most
relevant extant measure of the propensity to morally disengage.
Methods and Measures
Participants and procedure. A total of 399 individuals—141 em-
ployees matched with 129 supervisors and 129 coworkers—participated
in the study. The employees worked at a variety of different organiza-
tions in the southeast U.S., including technology, government, education,
insurance, legal, financial, manufacturing, food service, and retail or-
ganizations. They averaged 25.7 years in age, 3.1 years of experience
in the organization, and 48.9% worked full time (see Table 1 for addi-
tional demographics). The employees’ supervisors were 63.5% male and
67.7% Caucasian (18.1% Hispanic or Latino/a, 6.3% African American,
and 2.4% Asian American), with an average age of 37.5 years and an
average organizational tenure of 8.8 years. The employees’ coworkers
were 42.2% male and 59.1% Caucasian (21.3% Hispanic or Latino, 7.9%
African American, and 7.1% Asian American), with an average age of
27.8 years and an average organizational tenure of 5.1 years.
We recruited participants using a snowball sampling procedure (e.g.,
Grant & Mayer, 2009; Skarlicki & Folger, 1997). Researchers sent an
electronic message to 277 (51% response rate) students in upper-level
management courses at a university in the southeastern U.S. and provided
them with the opportunity to partake in a study for extra credit. Students
who worked at least 20 hours per week in a job were allowed to partic-
ipate; otherwise, they were asked to invite a friend or family member to
CELIA MOORE ET AL. 31
participate. The “focal employees” were instructed to visit a Web site to
complete a survey and to send an electronic survey link to their supervisor
and a coworker familiar with their work. Respondents were assured that
their responses would remain confidential. Focal employees provided de-
mographic and other information as well as responses to items from two
scales measuring the propensity to morally disengage. Coworkers and su-
pervisors provided ratings for the unethical work behaviors of the focal
employee.
The propensity to morally disengage. Employees completed the 8-
item propensity to morally disengage scale described in Study 1 (α=.90
in this sample). Fit indices from a confirmatory factor analysis revealed
an acceptable fit of the Study 5 data to a one-factor model (CFI =.96,
NNFI =.94, RMSEA =.12, SRMR =.05).
Alternative measure of moral disengagement. We used Duffy and
colleagues’ measure (Duffy et al., 2005; Duffy et al., 2002; McFerran
et al., 2010) as an alternative measure of the propensity to morally dis-
engage. It is the only extant measure that was designed to measure the
propensity to morally disengage in adults (rather than children, as in Ban-
dura et al., 1996; or undergraduate students, as in Detert et al., 2008),
without being focused on a specific type of audience, such as athletes
(Boardley & Kavussanu, 2007; Corrion et al., 2010) or computer hackers
(Rogers, 2001). It is also the only measure of the propensity to morally
disengage that taps this disposition in a workplace context, thus making
it the strongest alternative to our measure. The measure has 15 items
(α=.90).
Social desirability. Participants also completed the same 10-item
measure of social desirability as in Studies 1–4 (Strahan & Gerbasi, 1972),
but in this study responses were measured on a 7-point scale ranging from
strongly disagree to strongly agree (α=.81).
Dependent variable. We collected data on employees’ unethical work
behavior from two sources, their immediate supervisor and a coworker.
Because there is no validated survey measure of employee unethical
behavior, we selected items from two extant measures of undesirable
work behavior—Robinson and O’Leary-Kelly’s antisocial work behav-
ior scale (1998) and Bennett and Robinson’s organizational deviance
scale (2000)—that most clearly represent unethical organizational be-
havior. Rather than relying solely on our own judgments, we obtained
ratings from a sample of 29 individuals with graduate-level manage-
ment training—including 16 faculty who teach and research in the area
of business ethics. We asked them to rate how unethical each of the
26 work behaviors on these two extant scales (7 from the antisocial
behavior scale and 19 from the organizational deviance scale) is. These
raters were told that although all of the 26 behaviors are clearly
32 PERSONNEL PSYCHOLOGY
undesirable, not all are necessarily unethical according to the definition of
unethical behavior as actions that cause direct harm to another individual
or that violate widely accepted moral norms in society. Respondents then
rated—using a 10-point scale ranging from 1 =not unethical in the least
to 10 =extremely unethical—how unethical they considered each of the
26 antisocial and deviance items. We then created a 7-item measure of
“unethical work behavior” using the items that received an average rating
of 8.0 or higher.2The mean “unethicality” of these seven items was 8.37
versus, by contrast, 5.33 for the 19 items not used.
The seven items selected for this measure of unethical work behavior
are: “Falsifying a receipt to get reimbursed for more money than you spent
on business expenses” (M=9.31, SD =1.00), “Discussing confidential
company information with an unauthorized person” (M=8.59, SD =
1.84), “Damaging property belonging to my employer” (M=8.25, SD =
1.76), “Taking property from work without permission” (M=8.24,
SD =1.17), “Saying or doing something to purposely hurt someone
at work” (M=8.17, SD =1.81), “Using an illegal drug or consuming
alcohol on the job ” (M=8.03, SD =2.01), and “Making ethnic, religious,
or racial remarks at work” (M=8.03, SD =1.80). These items indeed
seem different from an ethics perspective compared with excluded items
such as, “Spending too much time fantasizing or daydreaming instead of
working” (M=4.04, SD =1.86) and “Griping with coworkers” (M=
3.07, SD =1.79).
We then calculated, using the main sample, separate supervisor- (α=
.93) and coworker- (α=.93) rated employee unethical behavior scales
by taking an average of that rating source’s responses to the seven items
about the focal employee’s unethical work behavior. Supervisors and co-
workers reporting the frequency with which they had observed the focal
employee engage in each of the unethical behaviors, ranging from 1 =
never to 7 =very often.
Results
As expected, the two measures of the propensity to morally disengage
are highly correlated (r=.77, p<.01). However, despite this, the data fit
a two-factor CFA model that loads the two measures’ items onto separate
latent factors (χ2=518, df =225) better (chi-square difference test
significant at p<.001) than a one-factor model that forces all (23 of)
the propensity to morally disengage items to load together (χ2=747,
2As a robustness check, we ran all the same analyses reported in this section using a
second 11-item measure of unethical work behavior that instead uses a cutoff of 7.0 or
higher for item inclusion. The results are substantively unchanged.
CELIA MOORE ET AL. 33
TABL E 8
Incremental Validity of the Propensity to Morally Disengage Over Other
Predictors of Unethical Work Behavior (Study 5)
DV =
Supervisor-reported DV =Coworker-reported
employee unethical
behavior
employee unethical
behavior
Model 1 Model 2 Model 3 Model 4
Social desirability −.08 −.08 .06 .06
Alternative moral
disengagement
measure
.43∗∗ .22 .44∗∗ .07
Propensity to
morally
disengage
.27∗.49∗∗
R2.22 .25 .16 .26
R2.03∗.10∗∗
Standardized coefficients are presented. N=128 (supervisors), N=129 (coworkers).
∗∗p<.01, ∗p<.05.
df =230). The two measures correlate similarly with supervisor-reported
unethical behavior (r=.47, p<.01 for the new, shorter measure, and
r=.46, p<.01 for the Duffy et al. measure). However, the new, shorter
measure of the propensity to morally disengage is more strongly correlated
with coworker-reported unethical behavior (r=.52, p<.01) than the
alternative (Duffy et al.) measure (r=.41, p<.01). The alternative
measure is more strongly correlated with social desirability (r=−.46, p
<.01) than our new measure (r=−.33, p<.01) as well.
To test Hypothesis 4, we examined the variance in supervisor- and co-
worker-reported employee unethical behavior accounted for by our new
propensity to morally disengage measure beyond that explained by the al-
ternative measure of moral disengagement and socially desirable response
tendencies. As shown in Model 1 of Table 8, the alternative measure of
the propensity to morally disengage explains a significant amount of vari-
ance in employee unethical behavior when our new measure is absent.
However, the coefficient for the alternative measure becomes nonsignif-
icant (see Model 2, Table 8) when the new measure of the propensity
to morally disengage is added to the equation. In support of Hypothesis
4, the variable for the new propensity to morally disengage scale (β=
.27, p<.05) explains an additional 3% of the variance in supervisor-rated
employee unethical work behavior even after accounting for an alternative
measure of the propensity to morally disengage.
34 PERSONNEL PSYCHOLOGY
This same pattern of results emerges when using coworker ratings
of unethical work behavior as the dependent variable, though here the
incremental value of the new measure is much greater. As shown in
Table 8, the alternative measure of the propensity to morally disengage
is initially a significant predictor of coworker-rated employee unethical
behavior (Model 3) but ceases to be when our new measure of the propen-
sity to morally disengage is added to the equation (see Model 4). Again
in support of Hypothesis 4, the variable for the new propensity to morally
disengage scale (β=.49, p<.01) explains an additional 10% of the
variance in coworker-rated employee unethical work behavior even after
accounting for an alternative measure of the propensity to morally disen-
gage. This finding is particularly impressive because coworkers may be
in a significantly better position than supervisors to know about the var-
ious types of employee unethical behavior assessed here, as presumably
employees are more motivated to hide these behaviors from supervisors.
In both Models 2 and 4 (in Table 8), the standardized coefficients for
the new measure are larger than the coefficients for the alternative measure,
another means of comparing the relative importance of the two measures
in the prediction of unethical workplace behavior (Combs, 2010). Finally,
when models (not presented here) are run including our measure of the
propensity to morally disengage and social desirability in the first step,
and the alternative measure in the second step, in neither case does adding
the alternative measure explain a significant additional proportion of the
variance in unethical behavior. Collectively, these results suggest that the
new measure of the propensity to morally disengage is superior to the
closest current alternative.3
Discussion
The set of studies presented here demonstrates the potency of indi-
viduals’ propensity to morally disengage in the prediction of unethical
organizational behavior and, in the process, makes two major contribu-
tions to the literature. First, our finding—across five different samples
and using different dependent variables—that the propensity to morally
disengage consistently emerges as a significant predictor of a wide range
of organizationally relevant unethical behaviors, explaining additional
variance above and beyond many of the major alternative individ-
ual difference antecedents, provides compelling evidence that morally
3We also ran all models described in Study 5 using as dependent variables the full-
length scales for antisocial workplace behavior (Robinson & O’Leary-Kelly, 1998) and
organizational and interpersonal deviance (Bennett & Robinson, 2000). Results are similar
results to those reported here (and are available from the first author).
CELIA MOORE ET AL. 35
disengaged reasoning represents a critical factor in a wide range of uneth-
ical workplace behaviors.
Specifically, we demonstrated the predictive validity of the propensity
to morally disengage beyond the explanatory power of (a) morally salient
individual differences (Machiavellianism, moral identity, and empathy),
(b) rational-deliberative constructs such as capacity for moral reasoning
(cognitive moral development) and philosophical orientations (predis-
positions towards idealism and relativism), and (c) dispositional moral
emotions (shame and guilt) across an array of unethical behaviors, includ-
ing self-reported lying, cheating, and stealing; business-related unethical
and self-serving decision making; and supervisor- and coworker-reported
unethical employee behavior. In fact, compared to the numerous alterna-
tive predictors we selected and controlled for based on theoretical and
meta-analytic evidence, our measure of the propensity to morally disen-
gage is the strongest and most consistent predictor of multiple unethical
organizational outcomes in the studies reported here.4
Notably, most of the relationships between the propensity to morally
disengage and other constructs in its nomological network were assessed
for the first time, thus additionally contributing new knowledge about
the morally relevant individual difference correlates of the propensity
to morally disengage. As expected, the propensity to morally disengage
correlates positively with Machiavellianism and relativism; negatively
with moral identity, empathy, cognitive moral development, idealism, and
dispositional guilt; and is not significantly correlated with dispositional
shame. Given these findings, we suggest that future theoretical models,
empirical studies, and intervention efforts seeking to understand and re-
duce unethical outcomes include the propensity to morally disengage as
a valuable explanatory construct.
Our second major contribution is the development and validation of a
parsimonious, adult-oriented, easily administered measure of the propen-
sity to morally disengage. Despite increasing interest in this construct in
recent years (Aquino et al., 2007; Detert et al., 2008; Duffy et al., 2005;
Moore, 2008b), research has been hampered by the lack of a consistently
used, short, reliable, and clearly valid measure of the propensity to morally
4Note that in designing our studies to assess the incremental validity of our propensity
to morally disengage measure we chose the baseline controls (which are entered as a block
in the regressions) on the basis of theory and meta-analytic evidence. As shown in the
correlation tables for studies 2, 3, and 4 (Tables 4, 6, and 7), the majority these variables are
significantly correlated at the bivariate level with the dependent variable in these studies.
The bivariate relationships between cognitive moral development, idealism and relativism,
and the unethical organizational outcome in Study 3 are not significant, which makes the
incremental validity of the propensity to morally disengage demonstrated in this study more
limited.
36 PERSONNEL PSYCHOLOGY
disengage. Given consistent calls to improve measurement in the field of
behavioral ethics (Mayer et al., 2009; Tenbrunsel & Smith-Crowe, 2008;
Trev i ˜
no et al., 2006), we believe this contribution warrants special note.
We followed rigorous best practices for new measure development (Clark
& Watson, 1995; Hinkin, 1995; John & Benet-Martinez, 2000) to construct
and validate—using multiple samples diverse in respondent background
and demographics—a parsimonious and psychometrically sound measure
of the propensity to morally disengage in adults that should prove useful in
a broad spectrum of future research with adults. The measure is consistent
with Bandura’s theory of moral disengagement as a general process and
taps, in a unifactorial scale, the eight theorized mechanisms that sever the
link between internalized moral standards and unethical behavior.
In our analyses, we were also able to show that the new measure
outperforms the best available alternative, and much longer, measure of
the construct (McFerran et al., 2010). We also consistently controlled
for respondents’ tendency to respond in socially desirable ways, thereby
ruling out potential concerns that our findings might be contaminated
by social desirability bias. Further, to reduce concerns about respondent
fatigue, common method bias, and uncertain direction of causality as much
as possible, we collected data for Studies 2–4 using two surveys each, with
time separation of up to 1 month (Ostroff, Kinicki, & Clark, 2002). And,
in Study 5, we obtained ratings of unethical employee behavior from both
supervisors and coworkers.
Implications for Theory
In the series of studies presented here, the propensity to morally dis-
engage is a consistently stronger correlate of unethical decisions and
behavior than a wide array of other theoretically relevant predictors
and consistently explains variance in unethical decisions beyond these
other predictors in multivariate analyses. Theoretically, this suggests that
Bandura’s self-regulatory approach, and in particular, his articulation of
the cognitive means by which people’s self-regulation process can be
thwarted, provides a strong and useful foundation for understanding and
predicting unethical decisions and behavior in organizations. Thus, behav-
ioral ethics researchers should consider how other individual difference
constructs might fit within a social cognitive self-regulation framework.
For example, future research is needed to better understand how moral
traits and emotions, such as empathy and guilt, relate to self-regulatory
processes and morally disengaged reasoning. Such affective dispositions
may be part and parcel of anticipatory self-regulation that prevents morally
disengaged reasoning from occurring and/or may take effect and
thus counter the personal acceptance of such reasoning as it reaches
CELIA MOORE ET AL. 37
consciousness. Similarly, deliberative ethical reasoning strategies (as
driven by cognitive moral development or idealism/relativism) will likely
only be used if moral disengagement has not occurred before such deliber-
ation actually begins. Though the actual process of moral disengagement
remains poorly understood, it may well occur prior to a number of these
other processes, thus preempting them and helping to explain why the
propensity to morally disengage is consistently a stronger predictor of
unethical decisions and behavior in our studies.
Other constructs within the nomological network of the propensity to
morally disengage might also be further considered from a self-regulation
perspective. Machiavellianism, for example, may reflect either insufficient
internalization of widely shared moral standards or the internalization of
counternormative standards that often dominate decision making when
two or more standards conflict. In the first “Wall Street” movie, Gordon
Gekko (with, for example, his “greed is good” speech) appears to be a
strong Machiavellian who has internalized and is driven by self-interest
rather than concern for others or fairness standards that are expected to
have force in normally socialized individuals. In such a case, the propensity
to morally disengage may be used to readily override more prosocial self-
regulatory standards.
The social cognitive self-regulatory perspective may also provide the-
oretical insights if applied to some of the newer thinking and research in
behavioral ethics. For example, much attention has been paid recently to
less deliberative and more impulsive or intuitive models of ethical decision
making (Haidt, 2001; Reynolds, Leavitt, & DeCelles, 2010). Our focal
construct—the propensity to morally disengage—seems consistent with
and potentially valuable to research on these models because this way of
thinking likely happens automatically and, to some extent, at a subcon-
scious level. Individuals with higher propensities to morally disengage
may in fact be those with the most developed “self-justifying systems” for
rationalizing intuitively made decisions. All of the ideas proposed here
must be further theoretically developed and tested. Our point is to suggest
potential ways that behavioral ethics scholars might benefit from adopting
this overarching theoretical framework and, where appropriate, our new
scale for empirical testing of emergent hypotheses.
Implications for Research
Following Bandura, we conceptualized the propensity to morally dis-
engage as an individual difference that represents a generalized cogni-
tive orientation to the world. However, the propensity to morally disen-
gage is not necessarily invariant across contexts or time (Paciello, Fida,
Tramontano, Lupinetti, & Caprara, 2008). Consistent with his broader
38 PERSONNEL PSYCHOLOGY
social cognitive theory, and with most work on moral development
(Bandura, 1986; Kohlberg, 1984), Bandura described moral disengage-
ment as explicitly interactive, the result of the continued reciprocal influ-
ences of the individual, behavior, and the environment (Bandura, 2002).
Thus, acknowledging that disengaging self-sanctions through morally dis-
engaged reasoning may also be triggered by specific contextual factors
(suggesting moral disengagement may also have a state instantiation)
is consistent with Bandura’s theory (1999, 2002). For example, specific
moral disengagement mechanisms are most likely triggered by particular
circumstances (e.g., displacement of responsibility in response to dom-
inant authority figures). Researchers are beginning to investigate situa-
tionally motivated moral disengagement as triggered by unethical actions
(e.g., Shu, Gino, & Bazerman, 2011) or other aspects of a situation such
as how much personal gain or harm to others is involved (e.g., Trevi˜
no,
Detert, Sweitzer, & Gephart, 2008). A next step will be to combine these
approaches by simultaneously studying both dispositional and situational
influences, and their possible interaction, on morally disengaged rea-
soning (see Cronbach & Snow, 1977 for ideas on constructing research
designs that capture both individual and situational variation). Research
on moral disengagement that examines both the person and the situation
simultaneously would capture more directly the interactionist nature that
dominant theories agree characterize our moral selves (Bandura, 1990b;
Trev i ˜
no, 1986). Such approaches would also be consistent with thinking
about many other individual differences (Hoyle & Leary, 2009) and with
theories describing how traits can be activated in particular contexts (Tett
& Burnett, 2003; Tett & Guterman, 2000).
Overall, our theoretical approach and statistical results are consistent
with Bandura’s theorizing (Bandura, 1986, 1990a, 1990b, 1999, 2002)
that the specific mechanisms that underlie the propensity to morally dis-
engage represent a single higher-order construct. Nevertheless, future re-
search may reveal that valid and reliable first-order measures can also
be developed to tap the propensity to use specific moral disengagement
mechanisms (e.g., attribution of blame) that would be useful for par-
ticular research contexts or questions. For example, it may be valuable
to assess only diffusion of responsibility when studying teamwork or
collective action, or just displacement of responsibility when studying
unethical behavior stemming from the nature of authority relationships in
work hierarchies. Or, researchers may wish to compare various specific
moral disengagement mechanisms to see which predict the most unwanted
behavior in particular contexts. Because we focused on moral disengage-
ment as an overarching propensity, we did not set out to develop first-order
mechanism scales, and we did not produce subscales that presently meet
acceptable standards (e.g., reliabilities of .70 or higher). Nevertheless, we
CELIA MOORE ET AL. 39
list all 24 items tested in Study 1 in Appendix A as a potential starting
point for others who wish to pursue this research avenue.
Finally, this study and others have provided a great deal of information
about the relationships between the propensity to morally disengage and
many types of unethical behavior relevant to organizations as reported by
the self or by employees’ supervisors and coworkers. The linking here,
in Study 5, of the propensity to morally disengage with supervisor and
coworker reports of unethical work behavior is particularly impressive as it
is likely a conservative test of what employees actually do (because many
unethical behaviors are intentionally hidden, especially from authority
figures). Nonetheless, future studies that link the propensity to morally
disengage with objectively measured subsequent unethical behavior in
actual work contexts would be particularly valuable.
Implications for Practice
As Combs notes, the relevance of an empirical effect found in research
is often best understood in economic terms (2010). In our studies, the
additional variance in an outcome measure accounted for by our propensity
to morally disengage scale ranges from 3% to 10%. At first blush, these
effect sizes may seem to be of questionable practical value. Yet, given the
enormous costs of unethical behavior to organizations and societies, we
suggest that these results are actually very important. For example, 3% of
the estimated $350 billion that the U.S. government loses every year to
tax evasion (U.S. Internal Revenue Service, 2010) is $10.5 billion dollars,
and 3% of the $2.9 trillion estimated to be lost annually to global fraud
(Association of Certified Fraud Examiners, 2010) is $87 billion dollars.
Clearly these losses, and many others of similar size (e.g., those occurring
as a result of employee or consumer theft), are anything but trivial.
The magnitude of the effect sizes found here for the propensity to
morally disengage on unethical decision making relative to other domi-
nant predictors in the behavioral ethics literature (noting that the effects
for the propensity to morally disengage found here are beyond many other
theoretically important influences) is also worth considering (Aguinis
et al., 2010). In the Kish-Gephart et al. (2010) meta-analysis, the largest
average effect size for any of the traditionally studied individual differ-
ence predictors of unethical behavior was for Machiavellianism at ρ=
.27 (across 11 studies and 2,290 individuals). If the effects from the four
predictive validity studies in this paper are combined, the average cor-
rected correlation between our new propensity to morally disengage and
unethical outcomes (using the same calculation methods as used in the
40 PERSONNEL PSYCHOLOGY
Kish-Gephart analyses, cf. Hunter & Schmidt, 2004) is ρ=.36.5Though
this effect size only aggregates the effects of four studies and 857 indi-
viduals, it may be cautiously interpreted as suggesting that the propensity
to morally disengage is the strongest individual difference predictor of
unethical behavior identified to date.
The measure we developed in this study is also practically useful.
It is short, easy to complete, and appropriate for general adult samples.
The measure’s consistently low correlation with the most commonly used
social desirability measure allays concerns that individuals administered
this measure for job-related purposes will simply provide “expected” re-
sponses. Thus, this measure may be useful in lieu of the much longer
(100-plus items) and laborious administrations of more expensive com-
mercial integrity tests (cf., Berry, Sackett, & Wiemann, 2007). Admin-
istering the scale to all applicants but having those making hiring deci-
sions blind to the results may allow organizations to determine over time
how valuable this instrument is in predicting subsequent misbehavior and
whether its use as a selection tool is warranted.
Organizations should also investigate whether morally disengaged
thinking can be attenuated through intervention or training. The one
study that has examined moral disengagement longitudinally found ev-
idence that individual levels of moral disengagement are malleable to
external influences over time (Paciello et al., 2008), which suggests that
moral disengagement—even among those more predisposed toward such
reasoning—may well be receptive to training interventions. Such interven-
tions could have practical implications for organizations that are interested
in reducing the harm caused by morally disengaged thinking. For exam-
ple, if moral disengagement operates in ways similar to other cognitive
biases (an empirical question), it could be quite beneficial to train em-
ployees to be on the lookout for certain modes of thinking (e.g., “I have
to do it because my boss said so”) so that they can catch themselves or
others before unethical behavior occurs (Bazerman & Tenbrunsel, 2011).
One interesting question in this regard would be whether training focused
on those with initially high propensities to morally disengage (who might
learn how to lessen their own distorted thinking) or those with lower initial
propensities (who might learn more easily to recognize it and intervene
with others) is more effective in reducing the measurable harm done from
moral disengagement.
5We thank Jennifer Kish-Gephart for her assistance with this analysis. Note that the 95%
confidence intervals around the uncorrected meta-analytic correlations between unethical
outcomes and (a) the propensity to morally disengage (uncorrected mean r=.30, CI .19
to .41) and (b) Machiavellianism (uncorrected mean r=.22, CI .15 to .28) overlap, as
would be expected given the theoretical and empirical relationship between the propensity
to morally disengage and Machiavellianism.
CELIA MOORE ET AL. 41
Organizations might also consider instituting a variety of measures
designed to reduce the likelihood that morally disengaged reasoning goes
unchecked. For example, leaders can use formal and informal means to
increase individual accountability, making the displacement or diffusion
of responsibility and the attribution of blame onto others feel less valid
as justifications for behavior even among those with higher propensities
to morally disengage. Leaders can likewise clearly encourage the use of
ethical language and discourage the acceptance of euphemisms that cloud
judgments. In addition, leaders can make harm to other stakeholders more
real to employees so that dehumanization or blame for “bringing harm
onto themselves” is less likely. Of course, only with careful research will
we know whether these interventions are effective in minimizing moral
disengagement in those most predisposed to do so or whether ultimately
organizations are better off relying on careful selection and early screening
mechanisms.
Conclusion
This paper’s five complementary studies robustly demonstrate how
individuals’ propensity to morally disengage predicts multiple types of
work-relevant unethical behavior. We offer a systematically developed,
parsimonious, reliable, and valid new measure for assessing the propen-
sity to morally disengage so that scholars and practitioners can use it to
better understand and reduce unethical behavior. Collectively, our find-
ings establish the pervasive predictive power of the propensity to morally
disengage, illuminating the importance of this construct.
REFERENCES
Aguinis H, Werner S, Lanza Abbott J, Angert C, Park JH, Kohlhausen D. (2010). Customer-
centric science: Reporting significant research results with rigor, relevance, and
practical impact in mind. Organizational Research Methods,13, 515–539.
Ambrose M, Arnaud A, Schminke M. (2008). Individual moral development and ethical cli-
mate: The influence of person–organization fit on job attitudes. Journal of Business
Ethics,77, 323–333.
Aquino K, Reed A. (2002). The self-importance of moral identity. Journal of Personality
and Social Psychology,83, 1423–1440.
Aquino K, Reed A, Thau S, Freeman D. (2007). A grotesque and dark beauty: How moral
identity and mechanisms of moral disengagement influence cognitive and emotional
reactions to war. Journal of Experimental Social Psychology,43, 385–392.
Ashforth BE, Anand V. (2003). The normalization of corruption in organizations. Research
in Organizational Behavior,25, 1–52.
Association of Certified Fraud Examiners. (2010). Global fraud study: Report to the nations
on occupational fraud and abuse. Retrieved from www.acfe.com/rrtn.
Badaracco JL, Useem J. (2006). Conflict on a trading floor. Harvard Business School Case
(9–394-060). Cambridge, MA: Harvard Business School Press.
42 PERSONNEL PSYCHOLOGY
Bandura A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall.
Bandura A. (1990a). Mechanisms of moral disengagement. In Reich W (Ed.), Origins of
terrorism: Psychologies, ideologies, states of mind (pp. 161–191). New York, NY:
Cambridge University Press.
Bandura A. (1990b). Selective activation and disengagement of moral control. Journal of
Social Issues,46, 27–46.
Bandura A. (1999). Moral disengagement in the perpetuation of inhumanities. Personality
and Social Psychology Review,3, 193–209.
Bandura A. (2002). Selective moral disengagement in the exercise of moral agency. Journal
of Moral Education,31, 101–119.
Bandura A, Barbaranelli C, Caprara GV, Pastorelli C. (1996). Mechanisms of moral dis-
engagement in the exercise of moral agency. Journal of Personality and Social
Psychology,71, 364–374.
Bandura A, Caprara GV, Zsolnai L. (2000). Corporate transgressions through moral disen-
gagement. Journal of Human Values,6, 57–64.
Bazerman MH, Tenbrunsel AE. (2011). Blind spots: Why we fail to do what’s right and
what to do about it. Princeton, NJ: Princeton University Press.
Bennett RJ, Robinson SL. (2000). Development of a measure of workplace deviance.
Journal of Applied Psychology,85, 349–360.
Benson ML. (1985). Denying the guilty mind: Accounting for involvement in a white-collar
crime. Criminology,23, 583–607.
Berry CM, Sackett PR, Wiemann SA. (2007). A review of recent developments in integrity
test research. PERSONNEL PSYCHOLOGY,60, 270–301.
Bird FB. (1996). The muted conscience: Moral silence and the practice of ethics in business.
Westport, CT: Quorum.
Boardley ID, Kavussanu M. (2007). Development and validation of the moral disengage-
ment in sport scale. Journal of Sport and Exercise Psychology,29, 608–628.
Bolinger D. (1982). Language: The loaded weapon. London, UK: Longman.
Brief AP, Buttram RT, Dukerich JM. (2001). Collective corruption in the corporate world:
Toward a process model. In Turner ME (Ed.), Groups at work: Theory and research
(pp. 471–499). Mahwah, NJ: Erlbaum.
Brown ME, Trevi˜
no LK. (2006). Ethical leadership: A review and future directions. The
Leadership Quarterly,17(6), 595–616.
Browne MW, Cudeck R. (1993). Alternative ways of assessing model fit. In Bollen KA,
Long JS (Eds.), Testing structural equation models (pp. 136–162). Newbury Park,
CA: Sage.
Caprara GV, Fida R, Vecchione M, Tramontano C, Barbaranelli C. (2009). Assessing
civic moral disengagement: Dimensionality and construct validity. Personality and
Individual Differences,47, 504–509.
Cheung GW, Rensvold RB. (2002). Evaluating goodness-of-fit indexes for testing mea-
surement invariance. Structural Equation Modeling,9, 233–255.
Christie R. (1970). Scale construction. In Christie R, Geis FL (Eds.), Studies in Machiavel-
lianism (pp. 10–34). New York, NY: Academic Press.
Christie R, Geis FL. (1970). Studies in Machiavellianism. New York, NY: Academic Press.
Chronbach LJ, Meehl PE. (1955). Construct validity in psychological tests. Psychological
Bulletin,52, 281–302.
Clark LA, Watson D. (1995). Constructing validity: Basic issues in objective scale devel-
opment. Psychological Assessment,7, 309–319.
Combs JG. (2010). Big samples and small effects: Let’s not trade relevance and rigor for
power. Academy of Management Journal,53, 9–13.
CELIA MOORE ET AL. 43
Corrion K, Scoffier S, Gernigon C, Cury F, d’Arripe-Longueville F. (2010). D´
eveloppement
et validation d’une ´
echelle courte mesurant le d´
esengagement moral en sport
(ECDMS). L’Enc´
ephale,36, 495–503.
Cronbach L, Snow R. (1977). Aptitudes and instructional methods: A handbook for research
on interactions. New York, NY: Irgvington.
Crowne DP, Marlowe D. (1960). A new scale for social desirability independent of psy-
chopathology. Journal of Consulting Psychology,24, 349–354.
Davis MH. (1983). Measuring individual differences in empathy: Evidence for a mul-
tidimensional approach. Journal of Personality and Social Psychology,44, 113–
126.
Detert JR, Trevi˜
no LK, Sweitzer VL. (2008). Moral disengagement in ethical decision
making. Journal of Applied Psychology,93, 374–391.
Deutsch M. (1990). Psychological roots of moral exclusion. Journal of Social Issues,46,
21–25.
DeVellis RF. (1991). Scale development: Theory and applications.NewburyPark,CA:
Sage.
Diener E. (1976). Effects of prior destructive behavior, anonymity, and group presence on
deindividuation and aggression. Journal of Personality and Social Psychology,33,
497–507.
Douglas T. (1995). Scapegoats: Transferring blame. New York, NY: Routledge.
Duffy MK, Aquino K, Tepper BJ, Reed A, O’Leary-Kelly A. (2005, November). Moral
disengagement and social identification: When does being similar result in harm do-
ing? Paper presented at the Academy of Management Annual Conference, Honolulu,
Hawaii.
Duffy MK, Tepper BJ, O’Leary-Kelly A. (2002). Moral disengagement and antisocial be-
havior at work. Paper presented at the Annual Meetings of the Southern Management
Association, Atlanta, GA.
Eisenberg N. (1986). Altruistic emotion, cognition, and behavior. Hillsdale, NJ: Erlbaum.
Eisenberg N. (2000). Emotion, regulation, and moral development. Annual Review of
Psychology,51, 665–697.
Eisenberg N, Miller PA. (1987). The relationship of empathy to prosocial and related
behaviors. Psychological Bulletin,101, 91–119.
Forsyth DR. (1980). A taxonomy of ethical ideologies. Journal of Personality and Social
Psychology,39, 175–184.
Gino F, Pierce L. (2009). Dishonesty in the name of equity. Psychological Science,20,
1153–1160.
Grant AM, Mayer DM. (2009). Good soldiers and good actors: Prosocial and impression
management motives as interactive predictors of affiliative citizenship behaviors.
Journal of Applied Psychology,94, 900–912.
Greenberg J. (2002). Who stole the money and when? Individual and situational determi-
nants of employee theft. Organizational Behavior and Human Decision Processes,
89, 985–1003.
Gross E. (1978). Organizational crime: A theoretical perspective. Studies in Symbolic
Interaction,1, 55–85.
Haidt J. (2001). The emotional dog and its rational tail: A social intuitionist approach to
moral judgment. Psychological Review,108, 814–834.
Haidt J. (2003). The moral emotions. In Davidson RJ, Sherer KR, Goldsmith HH (Eds.),
Handbook of affective sciences (pp. 852–870). New York, NY: Oxford University
Press.
Hinkin TR. (1995). A review of scale development practices in the study of organizations.
Journal of Management,21, 967–988.
44 PERSONNEL PSYCHOLOGY
Hoffman ML. (2000). Empathy and moral development: Implications for caring and justice.
Cambridge: Cambridge University Press.
Hoyle RH, Leary MR. (2009). Methods for the study of individual differences in social
behavior. In Leary MR, Hoyle RH (Eds.), Handbook of individual differences in
social behavior (pp. 12–23). New York, NY: Guildford.
Hu L, Bentler PM. (1999). Cutoff criteria for fit indices in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling,6,
1–55.
Hunter JE, Schmidt FL. (2004). Methods of meta-analysis: Correcting errorbias in research
findings. Thousand Oaks, CA: Sage.
Jackall R. (1988). Moral mazes: The world of corporate managers. New York, NY: Oxford
University Press.
John OP, Benet-Martinez V. (2000). Measurement: Reliability, construct validation, and
scale construction. In Reis HT, Judd CM (Eds.), Handbook of research methods
in social and personality psychology (pp. 339–369). New York, NY: Cambridge
University Press.
J¨
oreskog KG, S¨
orbom D. (2009). LISREL 8.8: User’s reference guide. Chicago, IL: Sci-
entific Software International.
Kelman HC. (1973). Violence without moral restraint: Reflections on the dehumanization
of victims and victimizers. Journal of Social Issues,29(4), 25–61.
Kelman HC, Hamilton VL. (1989). Crimes of obedience. New Haven: Yale University
Press.
Kern MC, Chugh D. (2009). Bounded ethicality: The perils of loss framing. Psychological
Science,20, 378–384.
Kish-Gephart JJ, Harrison DA, Trevi˜
no LK. (2010). Bad apples, bad cases, and bad barrels:
Meta-analytic evidence about sources of unethical decisions at work. Journal of
Applied Psychology,95, 1–31.
Kline RB. (1988). Principles and practice of structural equation modeling. New York, NY:
Guildford.
Kohlberg L. (1969). Stage and sequence: The cognitive-developmental approach to so-
cialization. In DA. Goslin (Ed.), Handbook of socialization theory and research
(pp. 348–480). Chicago, IL: Rand McNally.
Kohlberg L. (1984). Moral stages and moralization: The cognitive developmental approach.
In Kohlberg L (Ed.), The psychology of moral development: The nature and validity
of moral stages (Vol. 2, pp. 170–205). San Francisco, CA: Harper & Row.
Kramer M. (1990). The moral logic of Hizballah. In Reich W (Ed.), Origins of terrorism:
Psychologies, ideologies, theologies, states of mind (pp. 131–157). Cambridge, UK:
Cambridge University Press.
Lewis HB. (1971). Shame and guilt in neurosis. New York, NY: International Universities
Press.
Marsh HW. (1987). The hierarchical structure of self-concept and the application of hi-
erarchical confirmatory factor analysis. Journal of Educational Measurement,24,
17–39.
Mayer DM, Kuenzi M, Greenbaum RL. (2009). Making ethical climate a mainstream
management topic: A review, critique, and prescription for the empirical research
on ethical climate. In Cremer DD (Ed.), Psychological perspectives on ethi-
cal behavior and decision making (pp. 181–213). Greenwich, CT: Information
Age.
McAlister AL. (2001). Moral disengagement: Measurement and modification. Journal of
Peace Research,38, 87–99.
CELIA MOORE ET AL. 45
McFerran B, Aquino K, Duffy M. (2010). How personality and moral identity relate to
individuals’ ethical ideology. Business Ethics Quarterly,20(1), 35–56.
Milgram S. (1974). Obedience to authority: An experimental view. New York, NY: Harper
&Row.
Moore C. (2008a). Moral disengagement in processes of organizational corruption.Doc-
toral dissertation, University of Toronto, Toronto, Canada.
Moore C. (2008b). Moral disengagement in processes of organizational corruption. Journal
of Business Ethics,80, 129–139.
Opotow S. (1990). Moral exclusion and injustice: An introduction. Journal of Social Issues,
46(1), 1–20.
Osofsky MJ, Bandura A, Zimbardo PG. (2005). The role of moral disengagement in the
execution process. Law and Human Behavior,29, 371–393.
Ostroff C, Kinicki AJ, Clark MA. (2002). Substantive and operational issues of response
bias across levels of analysis: An example of climate–satisfaction relationships.
Journal of Applied Psychology,87, 355–368.
Paciello M, Fida R, Tramontano C, Lupinetti C, Caprara GV. (2008). Stability and change of
moral disengagement and its impact on aggression and violence in late adolescence.
Child Development,79, 1288–1309.
Paternoster R, Simpson S. (1996). Sanction threats and appeals to morality: Testing a
rational choice model of corporate crime. Law and Society Review,30, 549–583.
Pelton J, Gound M, Forehand R, Brody G. (2004). The moral disengagement scale: Exten-
sion with an American minority sample. Journal of Psychopathology and Behavioral
Assessment,26, 31–39.
Piaget J. (1965). The moral judgment of the child. New York, NY: The Free Press.
Rapoport DC, Alexander Y. (Eds.). (1982). The morality of terrorism: Religious and secular
justification. Elmsford, NY: Pergamon Press.
Ray JJ. (1983). Defective validity of the Machiavellianism scale. Journal of Social Psy-
chology,119, 291–292.
Rest JR. (1986). Moral development: Advances in research and theory. Minneapolis, MN:
University of Minnesota Press.
Rest JR. (1990). DIT manual. Minneapolis, MN: University of Minnesota Press.
Reynolds SJ, Leavitt K, DeCelles KA. (2010). Automatic ethics: The effects of implicit
assumptions and contextual cues on moral behavior. Journal of Applied Psychology,
95, 752–760.
Robinson SL, O’Leary-Kelly AM. (1998). Monkey see, monkey do: The influence of work
groups on the antisocial behavior of employees. Academy of Management Journal,
41, 658–672.
Rogers M. (2001). A social learning theory and moral disengagement analysis of crimi-
nal computer behavior: An exploratory study. Doctoral dissertation, University of
Manitoba, Winnipeg, Canada.
Sakalaki M, Richardson C, Th´
epaut Y. (2007). Machiavellianism and economic oppor-
tunism. Journal of Applied Social Psychology,37, 1181–1190.
Schlenker BR, Forsyth DR. (1977). On the ethics of psychological research. Journal of
Experimental Social Psychology,13, 369–396.
Schwab DP. (1980). Construct validity in organizational behavior. Research in Organiza-
tional Behavior,2, 3–43.
Shao R, Aquino K, Freeman D. (2008). Beyond moral reasoning: A review of moral
identity research and its implications for business ethics. Journal of Business Ethics,
18, 513–540.
Shu LL, Gino F, Bazerman MH. (2011). Dishonest deed, clear conscience: When cheating
leads to moral disengagement and motivated forgetting. Personality and Social
Psychology Bulletin,37, 330–349.
46 PERSONNEL PSYCHOLOGY
Shultz CJ. (1993). Situational and dispositional predictors of performance: A test of the
hypothesized Machiavellianism structure interaction among sales persons. Journal
of Applied Social Psychology,23, 478–498.
Siegel JP. (1973). Machiavellianism, MBA’s and managers: Leadership correlates and
socialization effects. Academy of Management Journal,16, 404–411.
Skarlicki DP, Folger R. (1997). Retaliation in the workplace: The roles of distribu-
tive, procedural, and interactional justice. Journal of Applied Psychology,82,
434–443.
Strahan R, Gerbasi KC. (1972). Short, homogenous versions of the Marlowe-Crowne social
desirability scale. Journal of Clinical Psychology,28, 191–193.
Sykes GM, Matza D. (1957). Techniques of neutralization: A theory of delinquency. Amer-
ican Sociological Review,22, 664–670.
Tangney JP. (1990). Assessing individual differences in proneness to shame and guilt:
Development of the self-conscious affect and attribution inventory. Journal of Per-
sonality and Social Psychology,59, 102–111.
Tangney JP. (1991). Moral affect: The good, the bad and the ugly. Journal of Personality
and Social Psychology,61, 598–607.
Tangney JP, Stuewig J, Mashek DJ. (2007). Moral emotions and moral behavior. Annual
Review of Psychology,58, 345–372.
Tangney JP, Wagner P, Gramzow R. (1989). The test of self-conscious affect (TOSCA).
Fairfax, VA: George Mason University.
Tenbrunsel AE, Messick DM. (1999). Sanctioning systems, decision frames, and coopera-
tion. Administrative Science Quarterly,44, 684–707.
Tenbrunsel AE, Messick DM. (2004). Ethical fading. The role of self-deception in unethical
behavior. Social Justice Research,17, 223–236.
Tenbrunsel AE, Smith-Crowe K. (2008). Ethical decision making: Where we’ve been and
where we’re going. Academy of Management Annals,2, 545–607.
Tett RP, Burnett, DD. (2003). A personality trait-based interactionist model of job perfor-
mance. Journal of Applied Psychology,88, 500–517.
Tett RP, Guterman HA. (2000). Situation trait relevance, trait expression, and cross-
situational consistency: Testing a principle of trait activation. Journal of Research
in Personality,34, 397–423.
Trevi ˜
no LK. (1986). Ethical decision-making in organizations: A person-situation interac-
tionist model. Academy of Management Review,11, 601–617.
Trevi ˜
no LK, Detert JR, Sweitzer VL, Gephart J. (2008, August). Motivated moral disen-
gagement: Dispositional and situational influences on ethical reasoning and behav-
ior. Paper presented at the Academy of Management Annual Conference, Anaheim,
California.
Trevi ˜
no LK, Weaver GR, Reynolds SJ. (2006). Behavioral ethics in organizations: A review.
Journal of Management,32(6), 951–990.
U.S. Internal Revenue Service. (2010). Information reporting program advisory com-
mittee: Tax gap subgroup report. Retrieved from www.irs.gov/taxpros/article/
o,,id=228971,00.html.
Vaughan D. (1996). The Challenger launch decision: Risky technology, culture, and de-
viance at NASA. Chicago, IL: University of Chicago Press.
Wea ve r G R, Tr ev i ˜
no LK. (1999). Compliance and values oriented ethics programs: Influ-
ences on employee attitudes and behaviors. Business Ethics Quarterly,9, 315–335.
Wrightsman LS. (1991). Interpersonal trust and attitudes about human nature. In Robin-
son JP, Shaver PR, Wrightsman LS (Eds.), Measures of personality and social
psychological attitudes (Vol. 1, pp. 373–412). San Diego, CA: Academic Press.
CELIA MOORE ET AL. 47
APPENDIX A
Propensity to Morally Disengage Scale
Moral Justification
It is okay to spread rumors to defend those you care about.∗
It is alright to lie to keep your friends out of trouble.∗
Playing dirty is sometimes necessary in order to achieve noble ends.
Euphemistic Labelling
Taking something without the owner’s permission is okay as long as
you’re just borrowing it.∗
It’s okay to gloss over certain facts to make your point.∗
When you’re negotiating for something you want, not telling the whole
story is just part of the game.
Advantageous Comparison
Considering the ways people grossly misrepresent themselves, it’s
hardly a sin to inflate your own credentials a bit.∗
Compared to other illegal things people do, taking something small from
a store without paying for it isn’t worth worrying about.∗
Damaging property is no big deal when you consider that others are
assaulting people.
Displacement of Responsibility
People shouldn’t be held accountable for doing questionable things
when they were just doing what an authority figure told them to do.∗
People cannot be blamed for misbehaving if their friends pressured them
to do it.∗
You can’t blame people for breaking the rules if that’s what they were
taught to do by their leaders.
Diffusion of Responsibility
People can’t be blamed for doing things that are technically wrong
when all their friends are doing it too.∗
It’s okay to tell a lie if the group agrees that it’s the best way to handle the
situation.∗
In contexts where everyone cheats, there’s no reason not to.
48 PERSONNEL PSYCHOLOGY
Distortion of Consequences
Taking personal credit for ideas that were not your own is no big
deal.∗
Walking away from a store with some extra change doesn’t cause any
harm.∗
It is OK to tell small lies when negotiating because no one gets hurt.
Dehumanization
Some people have to be treated roughly because they lack feelings
that can be hurt.∗
It’s okay to treat badly somebody who behaves like scum.∗
Violent criminals don’t deserve to be treated like normal human
beings.
Attribution of Blame
People who get mistreated have usually done something to bring it on
themselves.∗
If a business makes a billing mistake in your favor, it’s okay not to tell
them about it because it was their fault.∗
If people have their privacy violated, it’s probably because they have not
taken adequate precautions to protect it.
Items measured on a 7-point Likert scale ranging from strongly disagree
to strongly agree.
Items in bold comprise the final 8-item measure. Items marked with *
comprise the 16-item measure.
Copyright of Personnel Psychology is the property of Wiley-Blackwell and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.
