ArticlePDF Available

The (Non)Relation Between Empathy and Aggression: Surprising Results From a Meta-Analysis

Authors:

Abstract and Figures

Assumptions regarding the importance of empathy are pervasive. Given the impact these assumptions have on research, assessment, and treatment, it is imperative to know whether they are valid. Of particular interest is a basic question: Are deficits in empathy associated with aggressive behavior? Previous attempts to review the relation between empathy and aggression yielded inconsistent results and generally included a small number of studies. To clarify these divergent findings, we comprehensively reviewed the relation of empathy to aggression in adults, including community, student, and criminal samples. A mixed effects meta-analysis of published and unpublished studies involving 106 effect sizes revealed that the relation between empathy and aggression was surprisingly weak (r = -.11). This finding was fairly consistent across specific types of aggression, including verbal aggression (r = -.20), physical aggression (r = -.12), and sexual aggression (r = -.09). Several potentially important moderators were examined, although they had little impact on the total effect size. The results of this study are particularly surprising given that empathy is a core component of many treatments for aggressive offenders and that most psychological disorders of aggression include diagnostic criteria specific to deficient empathic responding. We discuss broad conclusions, consider implications for theory, and address current limitations in the field, such as reliance on a small number of self-report measures of empathy. We highlight the need for diversity in measurement and suggest a new operationalization of empathy that may allow it to synchronize with contemporary thinking regarding its role in aggressive behavior. (PsycINFO Database Record (c) 2013 APA, all rights reserved).
Content may be subject to copyright.
Psychological Bulletin
The (Non)Relation Between Empathy and Aggression:
Surprising Results From a Meta-Analysis
David D. Vachon, Donald R. Lynam, and Jarrod A. Johnson
Online First Publication, December 23, 2013. doi: 10.1037/a0035236
CITATION
Vachon, D. D., Lynam, D. R., & Johnson, J. A. (2013, December 23). The (Non)Relation
Between Empathy and Aggression: Surprising Results From a Meta-Analysis. Psychological
Bulletin. Advance online publication. doi: 10.1037/a0035236
The (Non)Relation Between Empathy and Aggression: Surprising Results
From a Meta-Analysis
David D. Vachon
University of Minnesota
Donald R. Lynam and Jarrod A. Johnson
Purdue University
Assumptions regarding the importance of empathy are pervasive. Given the impact these assumptions
have on research, assessment, and treatment, it is imperative to know whether they are valid. Of particular
interest is a basic question: Are deficits in empathy associated with aggressive behavior? Previous
attempts to review the relation between empathy and aggression yielded inconsistent results and
generally included a small number of studies. To clarify these divergent findings, we comprehensively
reviewed the relation of empathy to aggression in adults, including community, student, and criminal
samples. A mixed effects meta-analysis of published and unpublished studies involving 106 effect sizes
revealed that the relation between empathy and aggression was surprisingly weak (r–.11). This finding
was fairly consistent across specific types of aggression, including verbal aggression (r–.20), physical
aggression (r–.12), and sexual aggression (r–.09). Several potentially important moderators were
examined, although they had little impact on the total effect size. The results of this study are particularly
surprising given that empathy is a core component of many treatments for aggressive offenders and that
most psychological disorders of aggression include diagnostic criteria specific to deficient empathic
responding. We discuss broad conclusions, consider implications for theory, and address current
limitations in the field, such as reliance on a small number of self-report measures of empathy. We
highlight the need for diversity in measurement and suggest a new operationalization of empathy that
may allow it to synchronize with contemporary thinking regarding its role in aggressive behavior.
Keywords: empathy, verbal aggression, physical aggression, sexual aggression, meta-analysis
In the broadest sense, empathy is an adaptive behavior within
the repertoire of all mammals that includes social responses to
emotional expressions of pain, fear, and hunger, such as isolation
calls and hunger cries (Carter, Harris, & Porges, 2009). In con-
temporary cognitive neuroscience, empathy is typically character-
ized as a uniquely human trait dependent on higher brain structures,
although its underlying physiological substrates are necessarily
shared with more general aspects of emotionality and sociability
that depend on lower brain structures and the autonomic nervous
system (Carr, Iacoboni, Dubeau, Mazziotta, & Lenzi, 2003; Carter
et al., 2009; Decety & Jackson, 2004; Porges, 2007). In humans,
the ability to use affective information to predict others’ behavior
or regulate one’s own behavior makes empathy essential to adap-
tive social and moral development.
One of the most basic suppositions regarding the role of empa-
thy in humans is that it both inhibits antisocial behavior (Jolliffe &
Farrington, 2004; P. A. Miller & Eisenberg, 1988) and facilitates
prosocial behavior (Eisenberg & Miller, 1987). Highly empathic
people are thought to use information about others’ affective states
to avoid engaging in potentially harmful behavior and to alleviate
the suffering of others. Conversely, those lacking empathy cannot
use such information to guide their behavior. Some of the most
frequently described mechanisms underlying deficient empathy
include difficulty recognizing others’ emotional expressions,
adopting their perspective, sharing their emotional experience, or
feeling concerned about their distress. Clinical descriptions of
individuals lacking empathy also include features such as shallow
affect and the tendency to view the suffering of others with
indifference, contempt, or enjoyment. Researchers and theorists
have long struggled to disentangle these ideas and distinguish
empathy from similar constructs; as a result, the definition and
conceptualization of empathy has varied considerably over the last
50 years. This history is reviewed in detail elsewhere (Batson,
2009; Wispé, 1990), although some have suggested that the effort
to distinguish precise forms of empathy has been overemphasized
to the point of distraction (Preston & de Waal, 2002). Despite these
difficulties, consistent evidence has emerged for two broad forms
of empathy at least partially separable at the neural and cognitive
levels: cognitive empathy, the ability to detect or understand
emotions, and affective empathy, the tendency to feel the emotions
of others.
Cognitive and Affective Empathy: Theory of Mind
and Simulation
The distinction between cognitive and affective empathy might
be best understood by reference to two competing theoretical
views of how humans understand other humans’ mental states
David D. Vachon, Department of Psychology, University of Minnesota;
Donald R. Lynam and Jarrod A. Johnson, Department of Psychological
Sciences, Purdue University.
Correspondence concerning this article should be addressed to David D.
Vachon, University of Minnesota, Department of Psychology, N503 Elliott
Hall, Minneapolis, MN 55455. E-mail: dvachon@umn.edu
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Psychological Bulletin © 2013 American Psychological Association
2013, Vol. 140, No. 1, 000 0033-2909/13/$12.00 DOI: 10.1037/a0035236
1
more generally. The theory of mind perspective describes an
evolved psychological capacity for understanding behavior using
something akin to scientific theory (Gopnik & Meltzoff, 1998).
Using a fairly cold form of information processing, humans apply
a system of rules derived from their own experiences to represent
the mental states of others. The accurate representation of others’
thoughts and feelings confers obvious benefits, because it allows
us to accurately monitor others’ intentions, predict their behavior,
and enjoy the various advantages associated with social living. For
example, the evolutionary brain development of primates has
enabled them to act gregariously and enjoy the benefits of group
living, such as better protection from predation and food sharing
(Jolly, 1966). However, because group living occurs at the expense
of competition for resources and sexual partners, theory of mind
and other complex forms of social intelligence are needed to help
infer the mental states of putative allies or competitors (Fehr &
Fischbacher, 2004; Trivers, 1971). Functional brain imaging re-
search supports this view, as brain areas activated during theory of
mind tasks are nearly identical to those activated during the pris-
oner’s dilemma game and other tasks involving reciprocal ex-
change (McCabe, Houser, Ryan, Smith, & Trouard, 2001; Rilling,
Sanfey, Aronson, Nystrom, & Cohen, 2004).
In contrast to the theory of mind perspective, the simulation
perspective suggests that we instinctively mimic others’ mental
activity and use our own reactions to understand what they are
thinking and feeling (Gallese & Goldman, 1998). The automatic
and unconscious activation of neural representations matching the
perceived mental state of others has been referred to as the
perception-action mechanism (PAM; Preston & De Waal, 2002).
The PAM provides a bottom-up, phylogenetically ancient mecha-
nism for representing the mental states of others that is far less
intellectually demanding and cold than the top-down processes
described by the theory of mind account. The PAM allows for
social functioning without undue reliance on controlled processes,
explaining behaviors that are crucial for the reproductive success
of animals living in groups, such as mother–infant responsiveness,
social communication, altruistic behavior, and emotional conta-
gion (De Waal, 2008). The simulation perspective has garnered
initial support through the discovery of mirror neurons, which
respond when a particular action is both performed and observed
(Gallese & Goldman, 1998), although some have challenged the
role of mirror neurons in disorders such as autism (Fan, Decety,
Yang, Liu, & Cheng, 2010). Regardless of the mechanism, the
ability to experience the mental state of others, and the survival
advantage that this confers, represent the proximal and ultimate
explanations of simulation theory (Preston & De Waal, 2002).
The differences between theory of mind and simulation perspec-
tives (Shamay-Tsoory, 2009) are consistent with the distinction
typically drawn between cognitive and affective empathy. The
broadest definitions of cognitive empathy are actually redundant
with the theory of mind perspective and not specific to emotions,
while narrower definitions limit the scope of cognitive empathy to
detecting emotions (“empathic accuracy”; Ickes, 1993), projecting
oneself into another’s situation (“Einfühlung”; Lipps, 1903), or
imagining how another person is feeling (“perspective taking”;
Eslinger, 1998). In any case, cognitive empathy represents a top-
down analysis reliant on higher brain structures. For example,
general theory of mind tasks are associated with activation in the
medial prefrontal cortex (Fletcher et al., 1995), superior temporal
sulcus, and temporal poles (Gallagher & Frith, 2003), while theory
of mind tasks specific to emotions are associated with the ventro-
medial prefrontal cortex (Shamay-Tsoory, Tomer, Berger, Gold-
sher, & Aharon-Peretz, 2005).
Definitions of affective empathy also vary. The conceptualiza-
tion most similar to simulation theory is that of “emotional con-
tagion”—the exact matching of emotions (Hatfield, Cacioppo, &
Rapson, 1994). Although other definitions focus on the synchrony
of physiological responses with no necessity for emotion matching
(“shared physiology”; Levenson & Ruef, 1992), most conceptual-
izations refer to resonant emotional experiences that are congruent
in valence rather than identical (“affective empathy”; Eisenberg &
Strayer, 1987). Closely related to this conception of affective
empathy are sympathy and personal distress, which the psycho-
logical literature distinguishes on the basis of their social conse-
quences. Both include a negative emotional reaction caused by
witnessing another’s distress; however, sympathy is characterized
by an altruistic motivation to reduce that person’s suffering (Eisen-
berg, 2000), while personal distress connotes a selfish motivation
to reduce one’s own suffering— either by helping the other person
or escaping the situation, whichever is less costly (Batson, 1991).
In research, clinical practice, and everyday use, affective empathy
often refers to the former reaction and is used interchangeably with
sympathy (Eisenberg & Strayer, 1987; Preston & De Waal, 2002;
Wispé, 1986). Like cognitive empathy, affective empathy is asso-
ciated with a specific neural network, which includes the insula,
amygdala, and anterior cingulate cortex (Jackson, Rainville, &
Decety, 2006; Singer et al., 2004; Wicker et al., 2003).
Empathy and Aggression
Although the history of empathy is marked by conceptual and
operational disagreements, assumptions regarding its role in ag-
gressive behavior are remarkably congruent. These assumptions
extend beyond the realm of popular culture and its ubiquitous
cold-blooded criminals. Perpetrators of antisocial behavior, vio-
lence, and sexual aggression are regularly described as having
insufficient empathy (Hogan, 1973; Kohlberg, 1963; Marshall,
Hudson, Jones, & Fernandez, 1995; P. A. Miller & Eisenberg,
1988). Empathy also plays an important role in several external-
izing syndromes included in the Diagnostic and Statistical Manual
of Mental Disorders (5th ed.; DSM-5; American Psychiatric As-
sociation, 2013): Conduct disorder, antisocial personality disorder,
and narcissistic personality disorder are all characterized by low
empathy. Although not included in the DSM, psychopathy is a
personality disorder associated with extreme violence and antiso-
cial behavior; it is also the disorder most associated with empathy
deficits (Cleckley, 1941; Hare & Neumann, 2008; Lykken, 1995).
Empathy-externalizing research continues to flourish and clinical
applications are widespread. The purpose of the current meta-
analysis is to examine the validity of assumptions regarding the
role of empathy in aggressive behavior; before doing so, however,
we consider the theoretical bases of these assumptions and their
impact on clinical practice.
Theoretical Bases
Despite the conceptual uncertainty that surrounds empathy, its
hypothesized role in aggression is quite clear: Coming to know the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2VACHON, LYNAM, AND JOHNSON
internal state of others and vicariously experiencing their distress
will encourage supportive behavior and deter harmful behavior
(Eisenberg & Miller, 1987; P. A. Miller & Eisenberg, 1988).
Conversely, difficulties knowing or experiencing others’ emotions
will lead to deficient social and moral development. From this
perspective, empathy acts as an internal control. Like other internal
controls—such as the tendency to worry about consequences, the
ability to delay gratification, and an appreciation of the need for
rules—the capacity for understanding and being moved by the
feelings and well-being of others contributes to healthy socializa-
tion and moral development. Hare (1993) describes these internal
controls as “inner policemen” (p. 75) that regulate behavior even
in the absence of external controls, such as the law. Under control
theories of crime, the likelihood of aggression increases as controls
are reduced or eliminated. In the case of low empathy, impairment
in the ability to recognize, understand, feel, or care about others’
distress represents a missing internal control; in the absence of
such a control, selfish impulses are free to be acted upon. For this
reason, deficient empathy is a core feature of psychopathy—a
disorder of chronic antisocial behavior that represents the pan-
absence of internal controls.
From a learning perspective, empathic people will find their
own aggression vicariously punishing. Because the victim’s dis-
tress is aversive, stimulus-reinforcement learning occurs and ag-
gressive behavior is inhibited. Blair and colleagues (R. J. R. Blair,
2004; R. J. R. Blair & Blair, 2009; J. Blair, Mitchell, & Blair,
2005) have suggested that the amygdala is central to this process.
The amygdala is important in stimulus-reinforcement learning,
particularly fear-based conditioning (Flor, Birbaumer, Hermann,
Ziegler, & Patrick, 2002; LeDoux, 2000), and there is significant
amygdala dysfunction in psychopathy (J. Blair et al., 2005). People
with high psychopathy scores show reduced amygdala activity
during aversive conditioning (Birbaumer et al., 2005) and other
forms of emotional learning (Kiehl et al., 2001), as well as dys-
functional stimulus–reinforcement based on instrumental learning
tasks that require the functional integrity of the amygdala (K. S.
Blair, Morton, Leonard, & Blair, 2006). Amygdala dysfunction is
expected to prevent the aversive learning caused by victims’
expressions of distress. Other neuroimaging studies of moral de-
cision making have implicated the orbital frontal cortex in empa-
thy (Greene, Sommerville, Nystrom, Darley, & Cohen, 2001; Luo
et al., 2006). The orbital frontal cortex complements the amygdala
by representing the reinforcement expectations provided by the
amygdala and enabling the individual to avoid or reduce aggres-
sive behavior (R. J. R. Blair, 2004). Research also links empathy
to directly experienced pain. For example, meta-analytic research
suggests that empathy for pain is associated with the bilateral
anterior insular cortex and medial/anterior cingulate cortex and
that activation in these areas overlaps with activation during di-
rectly experienced pain (Lamm, Decety, & Singer, 2011).
Finally, expectations regarding the association between empathy
and aggression are also reasonable from a trait perspective. In
five-factor model trait space, for example, empathy is best char-
acterized by strong positive correlations with agreeableness (Moo-
radian, Davis, & Matzler, 2011; Nettle, 2007). In contrast, chronic
disorders of aggression, such as antisocial personality disorder and
psychopathy, are best characterized by strong negative correlations
with agreeableness (Lynam & Derefinko, 2006; J. Miller, Lynam,
Widiger, & Leukefeld, 2001; Samuel & Widiger, 2008), and the
stability, variety, and onset of conduct problems are most strongly
related to low agreeableness (Jones, Miller, & Lynam, 2011).
Furthermore, meta-analytic research using multiple measures of
personality suggests that traits related to empathy, nurturance, and
tendermindedness yield the largest sex differences of all traits,
with women scoring significantly higher than men (meta d0.97,
Feingold, 1994). This difference is constant across ages, years of
data collection, educational levels, and nations, and it parallels the
ubiquitous sex differences in violent and sexually violent crime,
with men committing the majority of simple assaults (71%), ag-
gravated assaults (84%), robberies (86%), and rapes (95%; U.S.
Department of Justice, 2007). Taken together, data from multiple
levels of analysis suggest that empathy should play a role in
aggression.
Clinical Applications
Expectations regarding the importance of empathy influence
treatment and assessment. Modules designed to increase empathy
in offenders are standard components of treatment in many cor-
rectional settings (Ross & Ross, 1995; Serin & Kuriychuk, 1994),
although they are most prevalent in the treatment of sexual offend-
ers (Marshall, 1999). Over $500 million each year is spent on
therapy for sex offenders that features “victim awareness and
empathy” training as its most frequent component (93% of men’s
programs and 95% of women’s programs; McGrath, Cumming,
Burchard, Zeoli, & Ellerby, 2010). Empathy training is also a core
feature of therapy programs outside correctional settings, including
violence prevention curricula for elementary school students
(Grossman et al., 1997), anger management therapy for youth
(A. P. Goldstein, Glick, & Gibbs, 1998; Pecukonis, 1990), and
treatment for perpetrators of domestic violence (Fruzzetti & Lev-
ensky, 2000). Despite the central role of empathy training in the
treatment of aggressive behavior, there is little evidence that
changes in empathy are sizable, stable, or predictive of lower
recidivism, sexual violence, or physical aggression (Day, Casey, &
Gerace, 2010; Hanson & Morton-Bourgon, 2005).
Empathy also plays an important role in the clinical assessment
of child and adult externalizing disorders, including conduct dis-
order, antisocial personality disorder, narcissistic personality dis-
order, and psychopathy. Conduct disorder in DSM-5 now comes
with an explicit specifier for “callous—lack of empathy,” and
antisocial personality disorder is characterized as a pattern of
disregard for and violation of the rights of others. The DSM-5 text
description for antisocial personality disorder also explicitly de-
scribes low empathy as an associated feature: “Individuals with
antisocial personality disorder frequently lack empathy and tend to
be callous, cynical, and contemptuous of the feelings, rights, and
sufferings of others” (American Psychiatric Association, 2013, p.
660). Narcissistic personality disorder is defined as a pervasive
pattern of grandiosity, need for admiration, and lack of empathy.
The seventh diagnostic criterion for narcissistic personality disor-
der is, “Lacks empathy: is unwilling to recognize or identify with
the feelings or needs of others” (American Psychiatric Association,
2013, p. 670). The narcissistic individual is described as having
difficulty recognizing the desires, subjective experiences, and feel-
ings of others; when recognized, these feelings are likely to be
viewed disparagingly as signs of weakness or vulnerability. Fi-
nally, psychopathy is characterized by an extreme lack of empathy.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
The eighth item of the Psychopathy Checklist—Revised (PCL–R;
Hare, 2003), the gold standard measure of psychopathy, is “cal-
lous/lack of empathy.” This item describes an individual whose
attitudes and behavior indicate a profound lack of empathy and a
callous disregard for others’ feelings and welfare. This individual
views others as objects to be manipulated, and any appreciation of
the pain, anguish, or discomfort of others is merely abstract and
intellectual.
The recent change to DSM-5 further elevates the prominence of
empathy. As mentioned above, conduct disorder in DSM-5 now
comes with an explicit specifier for low empathy. Also, although
proposals to reconfigure the personality disorder section in DSM-5
were ultimately rejected due to a lack of study, they also highlight
the importance of empathy in research and practice. Moved to
Section 3 of DSM-5 for further study, the trait-specific methodol-
ogy for the diagnosis of all personality disorders requires impair-
ment in two of four areas of personality functioning (Criterion A):
identity, self-direction, empathy, and intimacy. Following this, the
individual must exhibit at least one of five pathological personality
trait domains or 27 specific trait facets within these domains
(Criterion B; American Psychiatric Association, 2013). Thus, un-
der the proposed trait approach, empathy is a superordinate domain
to all personality traits. Somewhat surprisingly, the new trait
system defines empathy in purely cognitive terms: “Comprehen-
sion and appreciation of others’ experiences and motivations;
tolerance of differing perspectives; understanding of the effects of
own behavior on others” (American Psychiatric Association, 2013,
p. 762). Examples of impaired empathy are given for each of six
proposed personality disorders, although they vary considerably
from this central definition and from one another. Consider the
following descriptions: (a) antisocial personality disorder—“lack
of concern for feelings, needs, or the suffering of others; lack of
remorse after hurting or mistreating another” (p. 764); (b) narcis-
sistic personality disorder—“impaired ability to recognize or iden-
tify with the feelings and needs of others; excessively attuned to
reactions of others, but only if perceived as relevant to self; over-
or underestimate of own effect on others” (p. 767); and (c)
avoidant personality disorder—“preoccupation with, and sensitiv-
ity to, criticism or rejection, associated with distorted inference of
others’ perspectives as negative” (p. 765). These descriptions
focus on affective reactions, cognitive abilities, or content outside
the typical realm of empathy (preoccupation with criticism). Be-
cause empathy may eventually become a superordinate criterion
for the diagnosis of all personality disorders, there is a growing
need to define its conceptual boundaries and clarify its relation to
basic social behaviors, such as aggression.
Estimating the Strength of the Empathy–Aggression
Association
Because of the impact that assumptions about empathy have on
research, treatment, and assessment, it is important to know
whether empathy acts as it should. Of primary interest to the
current analysis is a basic and clinically relevant question: Is
empathy associated with aggressive behavior? This is a difficult
question to answer for several reasons. First, definitions of empa-
thy vary widely, and the association between empathy and aggres-
sion likely varies according to how empathy is operationalized.
Some conceptions are narrow (facial mimicry, shared physiology,
etc.) and do not fully represent the broader empathy construct. The
specificity of these narrow constructs also precludes a strong
association with broad behavioral tendencies because of mis-
matched conceptual resolution. Other definitions are broad, con-
struing empathy as an agglomeration of related constructs. How-
ever, we believe that the most useful and theoretically coherent
conceptions of empathy describe it in top-down cognitive terms
(Eslinger, 1998; Ickes, 1993), bottom-up affective terms (Batson,
1991; Eisenberg & Strayer, 1987), or both (M. H. Davis, 1980;
Preston & de Waal, 2002). These definitions place empathy at a
level of conceptual resolution that is suited to the study of social
behavior, connects it with established theories of how we under-
stand others’ behavior (theory of mind and simulation), and asso-
ciates it with separable neural networks. Because disorders of
aggression such as psychopathy have been associated with deficits
in affective empathy but not cognitive empathy (R. J. R. Blair,
2005), measures of empathy that are too narrow, too broad, or
based solely on cognitive definitions are expected to bear less of a
relation to antisocial behavior.
Measurement differences also make it difficult to determine
whether empathy inhibits aggression. For example, the various
modes of measuring empathy are often confounded with devel-
opmental period, such as picture/story methods for children and
questionnaire methods for adults (P. A. Miller & Eisenberg,
1988). Even within a developmental period, methods of mea-
suring empathy differ in important ways, such as whether the
stimuli involve specific hypothetical events, whether partici-
pants are tasked with perceiving emotional expressions, and
whether participants are required to report their feelings to an
experimenter—a procedure more susceptible to demand char-
acteristics than other empathy indexes. Another issue is that the
empathy literature is saturated with measures developed and
used in only one or two studies. Only three questionnaires of
dispositional empathy have been frequently related to aggres-
sion; these include the Interpersonal Reactivity Index (IRI;
M. H. Davis, 1983), the Hogan Empathy Scale (HES; Hogan,
1969), and the Questionnaire Measure of Emotional Empathy
(QMEE; Mehrabian & Epstein, 1972)—also referred to as the
Emotional Empathic Tendency Scale (EETS) and as the Meh-
rabian and Epstein Empathy Scale (MEES). As a result, most of
what is known about the relation between empathy and aggres-
sion in adults is derived from studies using the IRI, HES,
QMEE, and a collection of infrequently used measures. These
problems notwithstanding, there have been attempts to summa-
rize the relation between empathy and aggression.
Previous Meta-Analyses
The first systematic review of this issue was conducted 25
years ago. In their meta-analysis, P. A. Miller and Eisenberg
(1988) integrated research examining the association between
empathy and a broad range of externalizing outcomes, primarily
aggression but also delinquency, crime, and psychopathology
(conduct disorder, psychopathy, sociopathy, narcissism, and
Machiavellianism). Miller and Eisenberg analyzed 30 studies of
children, adolescents, and adults, and found that empathy was
significantly negatively related to aggression in studies that
used questionnaire methods for assessing empathy (average r
–.18). However, aggression was nonsignificantly related to all
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4VACHON, LYNAM, AND JOHNSON
other modes of assessing empathy, including picture/story mea-
sures (average r–.06), facial/gestural reactions (average r
–.06), and experimental inductions of empathy (average r
–.08). Miller and Eisenberg concluded that empathy is nega-
tively related to aggression, although their observed effect size
was small.
In a more recent meta-analysis, Jolliffe and Farrington (2004)
examined the relation of empathy to offending, excluding all other
aggressive behaviors and externalizing disorders. They also dif-
ferentiated affective empathy (the vicarious sharing of an emotion)
from cognitive empathy (the understanding of others’ emotions).
Jolliffe and Farrington defined offending behaviors as official or
reported criminal acts that, if detected, would be serious enough to
warrant legal action that could result in a conviction. Although
Jolliffe and Farrington’s analysis was limited to offending behav-
iors, their definition of offending includes both violent and non-
violent offenses. Using data from 25 studies, Jollife and Farrington
suggested that offending bore a moderate negative relation to
overall empathy (average d⫽⫺0.28), with a strong negative
relationship for cognitive empathy (average d 0.48) and a
weak negative relationship for affective empathy (average d
0.14). These authors also provided evidence that age moderated
the relation between overall empathy and offending, with empathy
being more weakly related in adult samples (d 0.17) than in
adolescent samples (d– 0.39).
Finally, in a review of 17 studies, Lovett and Sheffield (2007)
examined the relation between affective empathy and aggres-
sion in children and adolescents. Lovett and Sheffield con-
cluded that extant research offered conflicting findings. Specif-
ically, of the relations in five child samples, one was positive,
two were negative, and two were nonsignificant; of 11 effect
sizes in adolescent samples, seven were negative and five were
nonsignificant. This “vote counting” approach of tallying sig-
nificant and nonsignificant findings, however, fails to represent
the mean effect size across studies; as a result, the strength of
the relation between affective empathy and aggression in youth
is unknown.
These previous attempts to review the relation between em-
pathy and aggression yield somewhat inconsistent results. One
likely reason for the absence of convergence is that they differ
substantially in terms of measurement and sampling. In P. A.
Miller and Eisenberg (1988), for example, empathy was mea-
sured as a unitary construct and its relation to all forms of
aggression and related behaviors was studied in children, ado-
lescents, and adults. Their only significant finding was for the
nine studies that used self-report measures of empathy, seven of
which relied on the QMEE. In Jolliffe and Farrington (2004),
self-reported empathy was organized as a multifaceted con-
struct but was related to criminal offending in adolescents and
adults that included nonviolent crimes. They found stronger
negative correlations in adolescents compared to adults, and in
measures of cognitive empathy compared to measures of affec-
tive empathy. In Lovett and Sheffield (2007), empathy was
defined only in affective terms and related to aggression in
samples of children and adolescents. The authors concluded
that the relation between affective empathy and aggression was
inconsistent in their samples.
The Current Study
In an effort to clarify these mixed findings, we comprehensively
reviewed the relation of affective and cognitive empathy to phys-
ical, verbal, and sexual aggression in adults. There are several
reasons why the adult literature is the best starting point for
understanding the association between empathy and aggression.
First, empathy is a dispositional characteristic, and because per-
sonality traits are more stable in adults compared to children
(Roberts & DelVecchio, 2000), adult samples will likely yield
more reliable estimates of empathy. Second, adults have an estab-
lished history of aggression and a broader and more dangerous
repertoire of aggressive behaviors; therefore, aggression can be
measured more reliably in adults, and various forms of aggression
can be differentiated. Third, adults also have better reading ability,
which is important because virtually all of the significant effects
yielded by previous reviews were found when self-report measures
of empathy were used. There are also some methodological con-
cerns related to younger samples. For example, while methods are
more diverse in younger samples, they are also less well validated.
Also, measurement differences between adults (typically self-
reports) and children (typically lab tasks and other-reports) are
confounded with age.
There are several features that make the current meta-
analysis more comprehensive than previous reviews. For exam-
ple, all measures of empathy and aggression were included; as
such, various methods of measurement and multiple types of
empathy (cognitive and affective), aggression (verbal, physical
and sexual), and crime (physical and sexual) were considered.
Our analysis is also much more extensive than previous efforts:
It includes 86 published and unpublished studies that yield 106
independent effect sizes. Previous analyses included no adult
studies (Lovett & Sheffield, 2007), nine adult studies (P. A.
Miller & Eisenburg, 1988), and 14 adult studies (Jolliffe &
Farrington, 2004).
The size of the current meta-analysis also allowed for an
examination of several potentially important moderators, in-
cluding age, sex, race, criminal status, education, aggression
measure, aggression type, empathy measure, and empathy con-
tent. Given the frequent distinction between cognitive and af-
fective empathy and the potential mismatch between how a
measure is named and the content of its items, in the current
study we coded empathy content in terms of cognitive and
affective content for each empathy measure using average item-
level scores from trained raters; these ratings were used as
moderators of the relation between empathy and aggression.
Few moderators have been examined in previous meta-
analyses, and there is a question of whether empathy predicts
aggression in a generalizable way. Although mean levels of
empathy differ across some populations (e.g., men and women),
our analysis will determine whether the empathy–aggression
relation generalizes across age, sex, race, and criminal status.
We also examine education as a moderator because most mea-
sures of empathy are self-reports that require reading. Although
self-report measures of empathy have yielded the largest effect
sizes in previous meta-analyses, they may not work as well in
populations with low education.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
Hypotheses
Based on theoretical expectations and clinical applications of
empathy, we have several hypotheses about its relation to aggres-
sion.
Hypothesis 1: Empathy will be moderately related to aggres-
sion. This hypothesis is based on two subhypotheses for
affective and cognitive empathy.
Hypothesis 1a: Affective empathy will be strongly related to
aggression. In the DSM and elsewhere (e.g., psychopathy
assessments), violent offenders are described as callous, cold-
blooded, and remorseless. These features refer to the affective
component of empathy, as do theories based on stimulus-
reinforcement learning and amygdala function.
Hypothesis 1b: Cognitive empathy will be weakly related to
aggression. Although the ability to recognize and under-
stand emotions is a necessary component of empathic re-
sponding, it is also insufficient. Knowing or imagining that
someone will feel pain is the purpose of most violent
behavior; in this case, cognitive empathy will be positively
related to aggression. Expectations regarding a strong neg-
ative relation between cognitive empathy and aggression
may be unrealistic, because included within this expecta-
tion is the assumption that knowledge of others’ emotions
will guide behavior in a prosocial direction. It seems more
likely that those who are unable to recognize or understand
the emotional experiences of others will be socially un-
skilled and have difficulty developing or maintaining rela-
tionships (e.g., autistic spectrum disorders), rather than act
aggressively, per se.
Hypothesis 2: Empathy will be more strongly related to verbal
aggression than physical or sexual aggression. This expecta-
tion is based solely on a statistical factor—verbal aggression is
more normally distributed than physical and sexual aggres-
sion, which are rarer occurrences with positively skewed
distributions. Because mismatch in the distribution of two
variables attenuates the strength of their association, normally
distributed measures of empathy will therefore bear stronger
relations to verbal aggression than physical or sexual
aggression.
Hypothesis 3: When aggression is approximated using group
differences rather than directly measured, effect sizes will be
lower due to a reduction in variability through
dichotomization.
Hypothesis 4: Demographic variables will not moderate the
relation between empathy and aggression. Despite mean level
differences in empathy across populations, its relation to all
types of aggression will generalize across age, race, and sex.
Hypothesis 5: Education will moderate the relation between
empathy and aggression. Because most empathy measures use
a self-report method, smaller effect sizes are expected in
samples with lower education and reading ability.
Method
Sample of Studies
The initial literature search was conducted with the database
PsycINFO, which has a broad coverage of psychology and social
science journals as well as unpublished dissertations. Search terms
included the various keywords related to empathy and aggression.
Empathy search terms were crossed with aggression search terms.
The empathy terms consisted of the following: “empathy or em-
pathic or empathetic or callous or callousness or sympathy or
sympathetic or emotional contagion or perspective taking or theory
of mind or compassion or emotion recognition.” Aggression search
terms were “externaliz
or aggress
or antisocial or fight
or
assault or murder or homicide or violen
or rape or molest
or
physical abuse or spous
abuse or domestic abuse or child abuse
or crime or criminal
or incarcerat
or theft or stealing or robbery
or burglary or fraud or forgery or shoplifting or vandalism or
arson.” This search yielded 2,602 studies, although this number
was reduced to 2,396 when exclusionary criteria were imposed.
Specifically, studies were excluded from the analysis if they were
not written in English, if they were published prior to 1960, or if
they exclusively examined nonhuman populations or humans un-
der the age of 18 years. Because not all studies in PsycINFO were
marked with age group information, the search limit “participants
18only” recovered too few studies; as such, all abstracts were
manually screened and only those studies with adult participants
were retained.
The remaining abstracts were examined and included if they
met the following criteria: (a) the study was empirical and
included at least 20 participants; (b) the study included a
questionnaire or lab task designed to measure empathy; (c)
questionnaires of empathy had at least adequate reliability
(Cronbach’s ␣⬎.60); and (d) the study included a question-
naire or lab task designed to measure aggression, included
historical information regarding aggressive behavior, or in-
cluded groups inherently distinguished by differences in ag-
gression. Where abstracts did not provide sufficient information
to establish whether they met the inclusion criteria, they were
included in the next stage of the selection, in which the entire
document was examined using the above criteria.
One hundred sixty-one articles were downloaded or re-
quested through interlibrary loan. If an article met the inclusion
criteria but lacked sufficient data for an effect size to be
computed, authors were contacted by e-mail. Six such requests
were made with five usable responses. Additionally, the refer-
ence sections of all studies were reviewed to identify any
publications that may have been missed. For the current review,
dissertations were included but unpublished data were not so-
licited. Ultimately, 86 articles (56 published, 30 unpublished)
were included in the meta-analysis. Studies that reported rele-
vant relations using several different samples were treated as
independent samples, and each was included in the meta-
analysis for a total of 106 independent samples and 17,354
individuals.
Coding the Studies
For each study, a range of variables was coded. These include
the number of participants; demographic variables (age, pro-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6VACHON, LYNAM, AND JOHNSON
portion male, proportion White, proportion without high school
education, and socioeconomic status); IQ; and criminal vari-
ables (proportion with a criminal offense, proportion with a
violent offense, and proportion with a sex offense). Of 102
effect sizes, information regarding socioeconomic status and IQ
were available in only six and five cases, respectively, and these
variables were dropped from the moderator analysis. Also
coded were variables related to the measurement of empathy
and aggression.
Empathy coding. Coded were the name and type of each
empathy measure. The vast majority of studies used self-report
measures of empathy (93%, k80). Only six studies used
laboratory methods (e.g., emotion recognition, affect matching)
or a mixture of self-report and laboratory tasks. No study
examined an interview-based or informant report of empathy.
The content of the empathy measures was also coded. A goal of
the current investigation was to examine whether empathy
content— cognitive and affective—moderates the relation to
aggression. Some measures of empathy have item content that
is not relevant to the type of empathy they supposedly measure.
For example, the HES is a measure of cognitive empathy but
has few items that seem relevant to cognitive empathy. An
important question is whether the content validity of empathy
measures moderates their prediction of aggression: Do mea-
sures with more affective empathy content better predict ag-
gression than measures with less affective empathy content? To
answer this question, 19 independent raters coded every item of
the IRI, the HES, and the QMEE, at least one of which was used
in 82% of the studies included in the meta-analysis. After a
training session conducted by the first author, these 19 raters
coded each item on a 0 –2 scale for cognitive empathy and a 0 –2
scale for affective empathy, with ratings ranging from not at all
related to the specific form of empathy (0) to highly related to
the specific form of empathy (2). During training, cognitive
empathy and affective empathy were described in broad terms:
Cognitive empathy is the ability to detect or understand emo-
tions (using perspective-taking or not), while affective empathy
is the tendency to feel an emotional response that matches
another’s emotions in valence. The items from all three mea-
sures were unlabeled, randomized, and administered to raters
during a single session. An intraclass correlation coefficient
(ICC) was calculated as a measure of interrater agreement for
these content ratings; in the current study, agreement among the
19 raters was high (ICC .93).
Aggression coding. In all studies, the type of aggression
was coded (verbal, physical, sexual, or general). In 40% of
studies (k33), aggression was directly measured using one of
several methods, including self-report (k24), archival (e.g.,
criminal record; k4), laboratory (e.g., Taylor aggression
paradigm, k2), and a combination of methods (k3). In
60% of studies (k53), aggression was not directly measured,
but rather inferred based on group status. In general, these
studies either compared (a) physically/sexually violent criminal
samples to nonviolent criminal samples, or (b) general/sex
criminal samples to noncriminal samples (college students or
community members). In these cases, studies were coded under
the assumption that violent criminals are more aggressive than
nonviolent criminals and that criminals are more aggressive
than college students or community members. The first assump-
tion is based on a criterion-group approach, where all of the
violent samples have been convicted of an aggressive offense,
compared to none of the nonviolent samples (although it is
likely that some of these offenders have committed violent
crimes and not been caught). The second assumption is based
on evidence that aggression and antisocial disorders are more
common in criminal samples than student or community sam-
ples. For example, scores on aggression measures are signifi-
cantly higher for violent criminals than nonviolent criminals,
and significantly higher for nonviolent criminals than college
students or community members (Diamond & Magaletta, 2006;
Smith, Waterman, & Ward, 2006). Antisocial personality dis-
order and psychopathy are also over 10 times more prevalent in
criminal samples than in the community. The prevalence of
antisocial personality disorder is approximately 40%– 60% in
criminals (Fazel & Danesh, 2002; Hart & Hare, 1989), com-
pared to 4% in the community (Grant et al., 2004; Robins, Tipp,
& Przybeck, 1991). The prevalence of psychopathy is approx-
imately 15%–25% in criminals (Hare, 2003; Harpur & Hare,
1994), compared to 1% in the community (Coid, Yang, Ullrich,
Roberts, & Hare, 2009; Hare, 1998).
In addition to these two typical group comparisons (violent vs.
nonviolent criminals and criminals vs. noncriminals), a few studies
compared a violent criminal sample to a general criminal sample.
The description of the general criminal sample was ambiguous in
these studies; it was unclear whether they were composed entirely
of nonviolent criminals or whether some portion of them were also
violent. In these cases, the violent groups (100% violent) were still
assumed to be more aggressive than the general groups, which
were either composed of a nonviolent criminal sample (0% vio-
lent) or a general criminal sample (3% violent on average; U.S.
Department of Justice, 2010). Therefore, all studies were coded
under this set of assumptions regarding aggression: violent crim-
inal samples nonviolent or general criminal samples student
or community samples.
The magnitude of differences between these groups was
calculated using the standardized mean difference effect size,
which was converted to a Pearson correlation coefficient. All of
the studies were independently coded by the first and third
authors; regular discussions between the authors were used to
bring up coding discrepancies and to review details that one
coder had missed. Consensus meetings resolved all coding
discrepancies, leaving perfect agreement between the two cod-
ing logs. If we had taken a less iterative approach—waiting
until the coding of all studies was complete before having a
consensus meeting—we would have been able to calculate a
reliability statistic; however, given the large number of studies
included in our meta-analysis, we chose to regularly meet and
resolve discrepancies. See Table 1 for the studies included in
the current meta-analysis.
Statistical Analyses
Individual effect sizes. Because the underlying distributions
of empathy and aggression are assumed to be continuous, all effect
sizes were represented using the Pearson correlation coefficient
(r). Where rvalues were not reported, rwas calculated either by
converting existing parametric statistics such as F(1 effect size),
t(1 effect size), and dvalues (2 effect sizes) or directly from
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
Table 1
Studies Included in the Meta-Analysis and Sample Characteristics
Study NSample type Empathy Aggression measure ES
Published
Abbey et al. (2006) 163 Comm IRI SES, general delinquency 0.231
Abbey et al. (2007) 123 Comm IRI SES 0.271
Bartholow et al. (2005) 200 Stud IRI AQ-verbal, AQ-physical 0.210
Book et al. (2004) 116 Comm, Crim IRI Group difference 0.135
Book et al. (2004) 109 Stud, Crim IRI Group difference 0.006
Bovasso et al. (2002) 254 Psych IRI Violent, non-violent charges 0.095
Clements et al. (2007) 71 Comm Unique measure CTS-physical 0.152
Cohen et al. (2002) 44 Comm, Sex DAPI-Q Group difference 0.253
Covell et al. (2007) 107 Crim IRI CTS 0.078
Deardorff et al. (1975)-a 46 Stud, Crim HES Group difference 0.419
Deardroff et al. (1975)-b 45 Comm, Crim HES Group difference 0.219
DeGue & DiLillo (2004) 304 Stud, Sex IRI Group difference 0.156
DeWolf et al. (1988) 86 Comm, Crim HES, QMEE Group difference 0.290
Dolan & Fullam (2004) 109 Comm, Crim IRI Group difference 0.003
Elliott et al. (2009) 1,031 Norm, Sex IRI Group difference 0.094
Elsegood & Duff (2010) 92 Comm, Sex RME, MCET Group difference 0.173
Eysenck & McGurk (1980) 1,016 Comm, Crim QMEE Group difference 0.230
Fisher et al. (1999) 221 Comm, Sex IRI, VEDS Group difference 0.115
Francis & Wolfe (2008) 48 Comm, Viol IRI Group difference 0.344
Giancola (2003) 204 Comm IRI Taylor aggression paradigm 0.110
Goldstein & Higgins-D’alessandro (2001)-a 245 Comm, NVNS IRI Group difference 0.087
Goldstein & Higgins-D’alessandro (2001)-b 201 Comm, Viol IRI Group difference 0.004
Goldstein & Higgins-D’alessandro (2001)-c 186 NVNS, Viol IRI Group difference 0.083
Hanson & Scott (1995)-a 117 Stud, NVNS EWT Group difference 0.152
Hanson & Scott (1995)-b 274 Stud, Sex EWT Group difference 0.315
Hanson & Scott (1995)-c 125 Comm, NVNS EWT Group difference 0.095
Hanson & Scott (1995)-d 282 Comm, Sex EWT Group difference 0.115
Hanson & Scott (1995)-e 239 NVNS, Sex EWT Group difference 0.190
Hayashino et al. (1995)-a 53 Comm, NVNS IRI Group difference 0.086
Hayashino et al. (1995)-b 102 Comm, Sex IRI Group difference 0.121
Hayashino et al. (1995)-c 103 NVNS, Sex IRI Group difference 0.186
Heilbrun (1982) 168 Comm, Viol HES Group difference 0.055
Hoppe & Singer (1977)-a 71 NVNS, Sex QMEE Group difference 0.173
Hoppe & Singer (1977)-b 60 NVNS, Viol QMEE Group difference 0.146
Hoppe & Singer (1977)-c 99 Sex, Viol QMEE Group difference 0.021
Hosser & Bosold (2006) 105 Sex, Viol IRI Group difference 0.117
Kavussanu & Boardley (2009) 106 Comm IRI PABS 0.354
Kurtines & Hogan (1972) 249 Stud, Crim HES Group difference 0.466
Langevin et al. (1988) 98 Stud, Sex QMEE IBS-agg, History violence 0.005
Lauterbach & Hosser (2007) 839 NVNS, Viol IRI Group difference 0.148
Lengua & Stormshak (2000) 250 Stud IRI Number of antisocial acts 0.074
Lisak & Ivan (1995) 198 Stud, Sex QMEE Group difference 0.194
Loeffler et al. (2010) 115 Norm, Viol IRI History violence 0.484
Loudin et al. (2003) 300 Stud IRI Werner & Crick aggression 0.166
Marshal et al. (1993)-a 592 Stud, Sex IRI Group difference 0.003
Marshal et al. (1993)-b 230 Comm, Sex IRI Group difference 0.149
Marshal et al. (1993)-c 40 Comm, Sex IRI Group difference 0.012
Marshall & Maric (1996) 58 Comm, Sex HES, QMEE Group difference 0.411
Martínn et al. (2005) 196 Stud IRI SES 0.151
McGinley & Carlo (2007) 252 Stud IRI Delinquency measure 0.472
Mullins-Nelson et al. (2006) 174 Stud IRI # antisocial acts 0.020
Nussbaum et al. (2002)-a 163 NVNS, Viol TCI-empathy Group difference 0.147
Nussbaum et al. (2002)-b 65 NVNS, Sex TCI-empathy Group difference 0.348
Nussbaum et al. (2002)-c 140 NVNS, Sex TCI-empathy Group difference 0.438
Parton & Day (2002) 20 Norm, Sex IRI Group difference 0.069
Pithers (1994) 20 Norm, Sex IRI Group difference 0.070
Pithers (1999) 30 Norm, Sex IRI Group difference 0.042
Proctor & Beail (2007) 50 Comm, Viol IRI, TEP Group difference 0.053
Rice et al. (1994) 28 Comm, Sex HES, QMEE Group difference 0.182
Richardson et al. (1994) 189 Stud IRI AQ-verbal, AQ-physical 0.140
Richardson et al. (1998) 130 Stud IRI Verbal aggression lab task 0.250
Sandoval et al. (2000) 96 Crim QMEE AQ-verbal, AQ-physical 0.158
(table continues)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8VACHON, LYNAM, AND JOHNSON
published or provided means and standard deviations. Two effect
sizes were estimated as zero where nonsignificant empathy–
aggression relations were reported and no relevant statistics could
be located.
Mean effect sizes. The requirement of statistical indepen-
dence of observations dictates that the same sample not be
included multiple times when computing an aggregate effect
size. Although many studies included in the meta-analysis used
multiple measures of empathy and/or aggression in a single
sample, aggregating effect sizes by measure does not violate
this requirement of independence. When single samples yielded
multiple effect sizes, therefore, an average correlation was used
to represent that sample. Because correlation coefficients are
nonadditive, they were first transformed into z=units using the
Fisher z=transformation (Lipsey & Wilson, 2001) before being
averaged:
Table 1 (continued)
Study NSample type Empathy Aggression measure ES
Schaffer et al. (2009) 244 Stud IRI Delinquency measure 0.177
Schweinle et al. (2010) 86 Comm Unique measure CTS sexual 0.200
Sergeant et al. (2006) 182 Comm EQ AQ-verbal, AQ-physical 0.234
Seto et al. (1993) 36 Comm, Sex HES, QMEE Group difference 0.175
Simons et al. (2002) 188 Sex EWT Number of victims 0.406
Smallbone et al. (2003) 88 Norm, Sex IRI Group difference 0.091
Stuewig et al. (2010)-a 65 Stud IRI ARI-verbal, ARI-physical 0.311
Stuewig et al. (2010)-b 355 Crim IRI PAI-verbal, PAI-physical 0.148
Teten et al. (2008) 38 Psyc IRI IPAS, AQ-verbal, AQ-physical 0.307
Watt et al. (2000) 65 Comm, Crim HES Group difference 0.029
Unpublished
Barry (2003) 120 Stud IRI BDHI 0.150
Bench (1997) 127 Norm, Sex QMEE Group differences 0.213
Covell (2002) 202 Viol, Sex, NVNS IRI Group differences 0.029
Cremer (1996) 20 Sex, Comm IRI Group differences 0.177
Daly (2004) 448 Stud QMEE AQ-verbal, AQ-physical 0.061
Davis (2010) 71 Viol IRI # convictions, IPAS PM scale 0.142
DeGue (2007) 360 Viol, Sex, NVNS IRI History violence, crim history 0.316
Dietzel (2008) 102 Stud BEES, JACFEE SES, CTS-sexual 0.089
D’Orazio (2002) 60 Comm, Sex IRI Group difference 0.074
Goldstein (1996) 184 Viol, NVNS IRI History violence, crim history 0.016
Gynn-Orenstein (1981) 34 Viol, NVNS QMEE History of violence 0.205
Harmon (2001) 64 Viol, Comm EQI Group differences 0.205
Haugen (1998) 20 Norm, Sex IRI Group differences 0.032
Johnson (1996) 30 Viol, Comm BEES History violence 0.192
Kupferberg (2002) 348 Comm IRI AQ-verbal, AQ-physical 0.005
Layman (1995) 83 Viol, Sex, Stud MEES Criminal/violent history 0.031
Mattek (2003) 196 Stud QMEE, RES SES 0.000
McGinley (2007) 60 Viol, NVNS IRI, Unique measure Group differences 0.063
Pickett (2006) 150 Sex, NVNS IRI Group differences 0.255
Sartin (2004) 189 Viol IRI CTS, # arrests, # convictions 0.160
Sever & Harriette (2006) 39 Sex, NVNS IRI, DANVA Group differences 0.050
Shaw (1995)-a 14 Stud IRI Shocks administered 0.000
Shaw (1995)-b 18 Stud Unique measure Shocks administered 0.340
Shoss (2006) 88 Norm, Sex IRI Group differences 0.091
Simon (2002) 59 Viol, Comm IRI Group differences 0.115
Sturgeon (2003) 130 Viol, Sex, NVNS IRI Group differences 0.008
Travis (2008) 157 Viol, Sex IRI Group differences 0.187
Vachon & Lynam (2013) 579 Stud IRI, BES, ACME AQ, RPQ 0.235
Waldorf (1997) 26 Viol, Comm Unique measures History of violence 0.407
Warkentin (2008) 514 Sex, Stud IRI Group differences 0.121
Yardley (1997) 48 Viol, NVNS Unique measures Group differences 0.119
Note. Unpublished unpublished dissertation. Sample type: Comm community sample; Stud student sample; Psych psychiatric sample; Norm
Norm value for empathy scale in the community; Viol violent criminal sample; Sex sexual criminal sample; NVNS nonviolent, nonsexual criminal
sample; Crim general criminal sample. Empathy measure: IRI Interpersonal Reactivity Index; HES Hogan Empathy Scale; QMEE Questionnaire
Measure of Emotional Empathy; EWT Empathy for Women Test; EQ Empathy Quotient; VEDS Victim Empathy Distortion Scale; TCI-Empathy
Temperament and Character Inventory, empathy subscale; TEP Test of Emotional Perception; DAPI-Q Dimensional Assessment of Personality
Impairment-Questionnaire, Empathy subscale; RME Reading the Mind in the Eyes Test; MCET Mind in a Child’s Eyes Task; DANVA Diagnostic
Accuracy of Nonverbal Accuracy, faces subscale; MEES Mehrabian and Epstein Empathy Scale; JACFEE Japanese and Caucasian Facial Expression
of Emotion; BEES Balanced Emotional Empathy Scale; BES Basic Empathy Scale; ACME Affective and Cognitive Measure of Empathy.
Aggression measure: Group difference no explicit measure of aggression was administered and group comparison represents the measure of aggression;
SES Sexual Experiences Survey; AQ Aggression Questionnaire; CTS Conflict Tactics Scale; IPAS Impulsive Premeditated Aggression Scale;
PM scale Premeditated scale; IBS-agg Interpersonal Behavior Scale, aggression subscale; ARI Anger Response Inventory; PABS Prosocial and
Antisocial Behavior in Sport; PAI Personality Assessment Inventory; BDHI Buss Durke Hostility Inventory; RPQ Reactive Proactive
Questionnaire; ES effect size (Pearson r).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
z.5[ln(1 r)ln(1 r)].
In order to derive a mean effect size across samples, each effect
size (ES
i
) was weighted by the inverse of its variance (w
i
). The
general formula for the weighted mean effect size is as follows
(Lipsey & Wilson, 2001):
ES
(wiESi)
wi
.
When multiple types of aggression were measured within a sam-
ple, effect sizes were averaged for the overall aggression analysis
but preserved for the specific aggression analyses. At both levels
of analysis, however, only one effect size from each sample was
included: At the overall aggression level of analysis, effect sizes
associated with all forms of aggression were averaged within a
sample to produce a single effect size; at the specific aggression
level of analysis, effect sizes associated with a specific form of
aggression (e.g., sexual aggression) were averaged within a sample
to produce a single effect size.
Outlier analysis of effect sizes was performed separately for
empathy for each aggression outcome (total, verbal, physical, and
sexual). All rvalues in an analysis were transformed to z=units and
values of z=outside the range of –2.5 and 2.5 standard deviations
were classified as outliers and subsequently removed from analy-
sis. Four outliers were identified using this approach, reducing the
total number of effect sizes from 106 to 102.
1
Analysis of heterogeneous distributions of effect size. The
distribution of effect sizes in a meta-analysis is analyzed using a
fixed-, random-, or mixed-effects model. Under the fixed-effect
model, all studies in the meta-analysis are assumed to share a
common (true) effect size. The observed effect size varies between
studies only because of the random sampling error inherent in each
study. In a fixed-effect model, the summary effect is an estimate of
this common effect size and all observed dispersion reflects sam-
pling error; study weights are assigned with the goal of minimizing
this subject-level sampling error.
Under the random-effects model, studies in the meta-analysis
are assumed to have multiple true effects that are normally dis-
tributed around a grand mean. Studies are included in the meta-
analysis because they have enough in common that it makes sense
to synthesize the information. In random-effects models, error
consists of variation in both true effects and subject-level sampling
error. Variation of true effects is assumed to be random and has the
same value across all studies in the meta-analysis, while sampling
error changes across studies and is estimated from the variance of
the observed effect size from each study. In a random-effect
model, the summary effect is an estimate of the mean of true
effects; study weights are assigned with the goal of minimizing
both true effect variance and sampling error.
Under the mixed-effects model, studies in the meta-analysis
are also assumed to have multiple true effects, but variation in
these true effects has both random and systematic components.
Thus, the mixed-effect model partitions variance into system-
atic true score variance, random true score variance, and subject-
level sampling error. Potential sources of systematic true score
variance, such as sample or measurement differences, can be
estimated as moderators of the summary effect.
In the current study, a homogeneity test was conducted to
determine whether variation in the effect sizes was larger than
what would be expected on the basis of sampling error alone. This
was accomplished using the Qstatistic, which tests for equality of
effect sizes within each analysis following a chi-square distribu-
tion with k– 1 degrees of freedom (Hedges & Olkin, 1985). The
formula for Qrepresents a weighted sum of squares estimate:
Qwi(ESiES
)2.
A nonsignificant Qsuggests that the various effect sizes all
estimate a common effect size and that the dispersion of the effect
sizes around the mean can be explained by subject-level sampling
error alone. When effect sizes are considered homogenous, a
fixed-effect model may be considered. However, even if Qis
nonsignificant, fixed effects may not be appropriate because they
are best used when all the studies have similar methods and the
purpose of the analysis is to compute the common effect size for
the identified population rather than generalize to other popula-
tions (Borenstein, Hedges, Higgins, & Rothstein, 2009). Con-
versely, a significant Qstatistic indicates that differences among
the effect sizes have some source other than subject-level sampling
error.
While Qcan be useful, it does not estimate the magnitude of
heterogeneity, and it has poor power to detect true heterogeneity
among studies in a small meta-analysis and excessive power to
detect negligible variability in a large meta-analysis (Higgins,
Thompson, Deeks, & Altman, 2003). Another statistic, I
2
, esti-
mates the magnitude of heterogeneity rather than its presence or
absence. The I
2
statistic can also be compared across meta-
analyses of different sizes, of different types of study, and using
different types of outcome data (Higgins et al., 2003):
I2Q(k1)
Q
100%,for Q(k–1)
I20, for Q(k–1)
.
When effect sizes are significantly heterogeneous, a random- or
mixed-effect model is appropriate. The mixed-effect model is used
when systematic factors can be identified and coded with some
consistency across studies; otherwise, a random-effect model is
used. Given the results of our heterogeneity test and the availabil-
ity of data on potentially important moderators to the empathy–
aggression effect, the current study used a mixed-effect model.
This model was used to partition potentially important systematic
variance from random variance and subject-level sampling error.
Systematic variance was estimated using several moderators, in-
cluding sample differences in demographics and the measurement
of empathy and aggression. Random variance was estimated using
the residual variability after accounting for systematic variance
rather than total variability, as in the random-effect model. The
1
Because we only had four outliers, we followed a common procedure
for handling a small number of outliers described by Lipsey and Wilson
(2001). Specifically, we eliminated the outlier studies from the effect size
distribution and compared the overall mean effect size from this trimmed
distribution to that of an untrimmed distribution. If the difference between
these mean effect sizes was sizable, we had planned to bring in the outliers
and Windsorize their effects; if the difference was negligible, we had
planned to simply exclude the outliers. Because the overall weighted mean
effect size for the empathy–aggression association was r–.113 in the
trimmed distribution and r–.106 in the untrimmed distribution, the
difference was negligible, and we simply eliminated the outliers.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 VACHON, LYNAM, AND JOHNSON
value of the random component in the mixed model was estimated
using a matrix algebra macro for modified weighted multiple
regression (Wilson, 2011).
Statistical software. The rvalues and Qstatistics were cal-
culated with Microsoft Excel, and the moderator analyses were
conducted using IBM SPSS (Version 19). In cases where rwas
calculated directly from published results or from in-text means
and standard deviations, Wilson’s (2011) online effect size calcu-
lator was used. Mixed-effect moderator analyses were conducted
using macros also available at this site.
Results
Test of Heterogeneity
The heterogeneity of the effect sizes in this meta-analysis was
significant, Q414.67 (101), p.001. The Qtest indicated that
there is significantly more heterogeneity of effect sizes than would
be expected on the basis of sampling error alone. The magnitude
of this heterogeneity, assessed using the I
2
statistic, was large (I
2
76%). Thus, Qand I
2
both suggest rejection of the homogeneity
assumption and its associated test—the fixed-effects model—and
consideration of the random-effects and mixed-effects models.
Given the availability of data on potentially important moderators
to the empathy–aggression effect, the current study used a mixed-
effect model to partition potentially important systematic variance
from random variance and subject-level sampling error.
Mean Effect Sizes
Table 2 reports the weighted mean effect sizes between various
measures of empathy and types of aggression, along with the
number of effect sizes, total number of participants, standard
errors, and 99% confidence intervals. Across all 86 studies, there
was a statistically significant
2
but small negative correlation be-
tween empathy and aggression (r–.11). The most frequently
used measure of empathy, the IRI, generally exhibited weak rela-
tions at the total and subscale levels (range of rs–.05 to –.11).
Only the HES yielded a substantially stronger mean effect size
(r–.42). With regard to type of aggression, mean effect sizes
were largest for verbal aggression (r–.20), followed by physical
aggression (r–.12) and sexual aggression (r–.09).
Publication Bias
Figure 1 displays a funnel plot of the effect sizes which helps to
detect potential bias due to underrepresentation of studies with
small samples. Such samples are underrepresented in the literature
because of their low statistical power and resultant nonsignificant
results. Publication bias thus tends to censor small effect sizes,
reducing the number of effect sizes in the region of the funnel
display where samples are small and the effect is close to zero. The
funnel plot in Figure 1 shows no evidence of publication bias.
Similarly, in any meta-analysis there is the potential for an
upward bias of the mean effect size due to sampling bias— on
average published studies have a larger mean effect size that
unpublished ones. In an effort to avoid this problem, both pub-
lished (n56) and unpublished (n30) studies were included.
Because the mean effect size is based on both published and
unpublished studies, the risk of sampling bias is substantially
mitigated. Additionally, a comparison of mean effect sizes in
published studies (r–.11) and unpublished studies (r–.10)
indicates that these effect sizes are equivalent, suggesting that
sampling bias is not a concern in the current analysis. A statistical
test of this difference is nonsignificant.
Moderator Analyses
Moderators included variables related to demographics (age, %
male, % White, and % no high school), crime (% criminal, %
violent criminal, and % sexual criminal), empathy (measure and
content rating), and aggression (type and measurement). Associa-
tions among the moderators were generally small, with some
predictable exceptions. Samples with a higher proportion of of-
fenders were less educated (r
% no high school
.74) and had a higher
proportion of males (r
% male
.39), while samples with a higher
proportion of sex offenders were older (r
age
.62) and more
educated (r
% no high school
–.50). Empathy variables were mod-
estly predicted by demographic variables (average r.01, range
rs–.21 to .18), and empathy content ratings were negatively
associated (r–.40) with one another; empathy scales with higher
affective ratings received lower cognitive ratings, and vice versa.
Tables 3 and 4 summarize the results of the regression and
analyses of variance (ANOVA) for continuous and categorical
moderators, respectively. Since multiple moderators were tested,
the significance level was set at a more conservative value, p
.01; confidence intervals were set at 99%. Homogeneity of effect
size analysis revealed significant variation in the weighted mean
effect sizes beyond what could be explained by sampling error.
Because significant heterogeneity in effect size was present, mod-
erator analyses were undertaken using mixed-effects modeling
estimated via the method of moments. Moderators were examined
individually and only at the level of total aggression, where infor-
mation was most consistently available.
Table 3 provides the results of these analyses for continuous
moderators analyzed using weighted multiple regression; Table 4
provides the results of these analyses for categorical moderators
analyzed using analysis of variance.
Continuous moderators. For each moderator in the regres-
sion analyses, Table 3 reports k, the number of effect sizes; n, the
number of participants; Mand SD, the mean and standard devia-
tion for continuous moderators, used to determine the high and low
values of rlater in the table; B, the unstandardized regression
coefficient that represents the amount of change in rfor a one unit
increase in the moderator; SE, the standard error associated with
the Bcoefficient; , the standardized regression coefficient; CI, the
confidence interval surrounding ;r low and r high, the values of
rat low (–1 SD) and high (1 SD) values of the moderator,
respectively; Q_bet, the variance accounted for by the moderator
variable; and Q_with, the residual variability in effect size.
Age example. For continuous moderators, a separate analysis
was conducted for each moderator. Using the example of age, the
empathy–aggression effect size was regressed onto age across
studies using a mixed-model effects regression. The studies that
included information on age yielded a total of 79 effect sizes (K)
2
Statistical significance is indicated by the fact that the 99% confidence
does not include zero.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
and had 13,174 participants (N). The mean age in these studies was
29.1 years (M), with a standard deviation of 7.9 years (SD). When
age was used as a moderator of the empathy–aggression correla-
tion, a 1-year increase in age predicted a .005 increase in the
correlation (B), with a standard error of .003 (SE). In standardized
units, a standard deviation increase in age predicted a .215 stan-
dard deviation increase in the correlation (); there is a 99%
chance that estimates a true value that falls between –.009 (CI
lower) and .422 (CI upper). The mean association between
empathy and aggression in younger adults is r–.13 (rlow) and
–.05 (rhigh) in older adults. Although not significant as a mod-
erator, age accounted for 4.6% of the variance in the empathy–
aggression association (Qbetween/[Qbetween Qwithin]; al-
ternatively,
2
).
Categorical moderators. For each moderator in the ANOVA
analysis, Table 4 reports k;n; mean weighted ES,the mean
weighted effect size; SE, the standard error of the mean weighted
effect size; p;Q_bet;Q_with; and r
2
.
Empathy measure example. For categorical moderators, a
separate analysis was conducted for each moderator, represented
by a set of dummy codes. Using the example of “empathy mea-
sure,” the empathy–aggression effect size was simultaneously re-
gressed onto three dummy codes representing the type of each
empathy measure using a mixed-model effects regression. The
studies that included information on type of empathy measure
yielded a total of 102 effect sizes and had 16,773 participants (N).
The mean empathy–aggression correlation was r–.101 for the
IRI, with a standard error of 0.023 and a confidence interval
ranging from r–.160 to – 0.042; r–.332 for the HES, with a
standard error of 0.078 and a confidence interval ranging from r
–.532 to –.132; r–.106 for the QMEE, with a standard error of
0.068 and a confidence interval ranging from r–.281 to 0.069;
and r–.061 for the other measures of empathy, with a standard
error of 0.029 and a confidence interval ranging from r–.136 to
0.014. Type of empathy measure accounted for 9.1% of the vari-
ance in the empathy–aggression association (Qbetween/[Qbe-
tween Qwithin]).
Demographics. Of the demographic variables examined, only
education moderated the empathy–aggression relation. In samples
characterized by low education, the already small negative asso-
Table 2
Meta-Analytic Relations Between Empathy and Aggression
Measure KN r SE
99% CI
Lower Upper
Total aggression
All empathy measures 102 16,882 0.113
0.008 0.133 0.093
IRI—4 subscale total 18 3,414 0.080
0.017 0.124 0.035
IRI—2 subscale total (PT EC) 52 10,439 0.072
0.010 0.098 0.047
IRI—Perspective Taking (PT) 48 9,572 0.111
0.010 0.138 0.085
IRI—Empathic Concern (EC) 46 9,051 0.048
0.011 0.075 0.021
Hogan Empathy Scale 9 781 0.416
0.036 0.510 0.322
QMEE 13 19,017 0.163
0.023 0.222 0.103
Empathy for Women Test 6 1,225 0.198
0.029 0.272 0.124
TCI, empathy subscale 3 368 0.161
0.053 0.026 0.229
Other measures 19 1,970 0.064
0.023 0.123 0.005
Specific aggression—Verbal
All empathy measures 13 2,880 0.204
0.019 0.253 0.156
IRI—2 subscale total (PT EC) 8 1,939 0.169
0.023 0.213 0.110
IRI—Perspective Taking (PT) 9 2,112 0.256
0.022 0.312 0.199
IRI—Empathic Concern (EC) 7 1,901 0.208
0.023 0.268 0.149
Specific aggression—Physical
All empathy measures 33 5,933 0.116
0.013 0.150 0.083
IRI—2 subscale total (PT EC) 20 4,455 0.150
0.015 0.188 0.111
IRI—Perspective Taking (PT) 18 4,143 0.137
0.015 0.176 0.098
IRI—Empathic Concern (EC) 20 4,416 0.128
0.015 0.166 0.090
Specific aggression—Sexual
All empathy measures 39 6,424 0.092
0.013 0.124 0.059
IRI—2 subscale total (PT EC) 23 4,407 0.076
0.015 0.115 0.036
IRI—Perspective Taking (PT) 21 4,121 0.132
0.016 0.173 0.092
IRI—Empathic Concern (EC) 21 4,121 0.010 0.016 0.030 0.051
Hogan Empathy Scale 3 122 0.487
0.094 0.729 0.244
QMEE 5 643 0.161
0.040 0.264 0.057
Note.Knumber of samples; Nsample size; SE standard error; CI confidence interval. Other measures include the Mind in the Eyes task (k
2), Balanced Emotional Empathy Scale (k1), Diagnostic Accuracy of Nonverbal Accuracy, Faces scale (k1), Rape Empathy Scale, Social Emotional
Questionnaire (k1), Empathy Quotient (k1), Victim Empathy Distortion Scale (k1), Test of Emotional Perception (k1), Dimensional Assessment
of Personality Impairment-Questionnaire, Empathy scale (k1), an empathy adjective list (k1), an empathic inference task (k1), Affective and
Cognitive Measure of Empathy (k1); four used multiple measures including measures not listed above. IRI Interpersonal Reactivity Index; PT
Perspective Taking subscale; EC Empathic Concern subscale; TCI Temperament and Character Inventory; QMEE Questionnaire Measure of
Emotional Empathy.
p.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 VACHON, LYNAM, AND JOHNSON
ciation between empathy and aggression was reduced nearly to
zero. The effect of education accounted for 17% of the variance in
the empathy–aggression association. Other demographic modera-
tors (age, sex, and race) and criminal variables (% criminal, % sex
criminal, % sex criminal) were not significant and accounted for
between 0% and 5% of the variance in the empathy–aggression
relation (see Table 3).
Measurement of empathy. Two moderator analyses were
conducted to determine whether the measurement of empathy
affected its association with aggression. First, item-level content
ratings of the empathy scales and subscales were used as moder-
ators. As described in the method section, the Cognitive Empathy
and Affective Empathy scores represent average ratings of the item
content for each measure and its subscales. High scores for Cog-
nitive Empathy and Affective Empathy suggest that items in the
measure are relevant to cognitive and affective empathy, respec-
tively, while low scores suggest that items are not particularly
relevant to empathy. Regression analyses for ratings of cognitive
content and for affective content were separately conducted; nei-
ther significantly accounted for variance in the mean effect size,
with r
2
ranging from 0% to 4% (see Table 3).
Second, the mean effect sizes across different empathy mea-
sures were compared. ANOVAs were used to compare the
effects from the IRI only, HES only, QMEE only, and all other
measures of empathy. Effect sizes were coded as other when
any measure other than the IRI, HES, or QMEE was used, or
when multiple measures of empathy were combined to produce
the effect size (including the IRI, HES, and QMEE). Of 102
effect sizes, 56 were coded as IRI only, 5 as HES only, 6 as
QMEE only, and 35 as other. The results of this analysis
indicated that differences in mean effect size across measures of
empathy were statistically significant, accounting for 9% of
the variability in the empathy–aggression relation. Specifically,
the empathy–aggression relation was significantly stronger for
the HES (see Table 4).
Measurement of aggression. Two moderator analyses were
also conducted to determine whether the measurement of ag-
gression affected its association with empathy. First, the mean
effect sizes across different types of aggression were compared:
physically aggressive only, sexually aggressive only, or gener-
ally aggressive.
3
Effect sizes were coded as generally aggres-
sive when the measure used general items (e.g., “I am an
aggressive person”) or combined items or scales representing
different forms of aggression (e.g., the Aggression Question-
naire total score, which represents an amalgamation of physical
and verbal aggression scores), as well as when several effect
sizes within a study that measured different types of aggression
were combined into a single effect size. Of 102 effect sizes, 24
were coded as physically aggressive only, 37 as sexually ag-
gressive only, and 41 as generally aggressive. The results of this
analysis indicated that differences in mean effect size across
types of aggression were not statistically significant, accounting
for less than 1% of the variability in the empathy–aggression
relation (see Table 4).
Second, a moderator analysis was conducted to compare the 43
effect sizes obtained from studies that measured aggression di-
rectly (self-reports, archival data, laboratory tasks) to the 59 effect
sizes obtained by comparing groups distinguished by their levels
of aggression (e.g., violent criminals vs. nonviolent criminals). The
results of this analysis revealed a marginally significant effect (p
.02), suggesting that the negative relationship between empathy
and aggression was stronger when direct measures of aggression
were used (r–.13) compared to proxies of aggression based on
group comparisons (r–.07). Within direct measures of aggres-
sion, there was no significant difference for self-report versus
archival data, and there were too few effect sizes derived from
laboratory measures of aggression to include it in the moderator
analysis.
Published versus unpublished studies. The association be-
tween empathy and aggression in published studies (r–.11) was
no different than in unpublished studies (r–.10). These effect
sized are nearly equivalent, and a moderator test suggests that the
effect sizes from published and unpublished studies were not
significantly different in magnitude.
Discussion
The results of this meta-analysis supported all but one of our
primary hypotheses. As expected, empathy predicted verbal ag-
gression more strongly than physical or sexual aggression (Hy-
pothesis 2); empathy was more strongly related to aggression when
aggression was directly measured than when it was indirectly
3
Because there was only a single effect size for the Verbal-only cate-
gory, this category was dropped.
Figure 1. Funnel plot of effect sizes: Pearson correlations between em-
pathy and aggression by sample size.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
13
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
assessed using group membership (Hypothesis 3); the association
between empathy and aggression generalized across age, race, and
sex (Hypothesis 4); and education moderated this association, with
smaller effect sizes in samples with lower education (Hypothesis
5). However, our primary hypothesis, that empathy and aggression
would be at least moderately related (Hypothesis 1), was not
supported by the meta-analysis. Collectively, across all measures
of empathy and aggression, only 1% of the variance in aggression
was explained by empathy. Prediction across specific forms of
aggression, including verbal (4%), physical (1%), and sexual ag-
gression (1%), was consistently low. Built into our primary hy-
pothesis were two subhypotheses: first, that cognitive empathy
would weakly predict aggression, and second, that affective em-
pathy would strongly predict aggression. The meta-analysis sug-
Table 3
Multiple Regression Analyses for Continuous Moderator Variables Using a Mixed Effects Model
Predictor KN MSD B SE
CI ()rQ
Lower Upper Low High Between Within
Demographics
Age 79 13,174 29.10 7.90 0.005 0.003 0.215 0.009 0.422 0.13 0.05 4.19 (1) 86.14 (77)
% male 96 16,025 0.84 0.27 0.054 0.063 0.085 0.109 0.280 0.11 0.09 0.74 (1) 100.57 (94)
% White 51 7,605 0.63 0.23 0.135 0.118 0.158 0.113 0.430 0.13 0.07 1.30 (1) 50.81 (49)
% no high school 46 7,733 0.24 0.31 0.213 0.073 0.413 0.136 0.690 0.15 0.03 8.53 (1)
41.52 (44)
% criminal 68 10,358 0.73 0.34 0.121 0.070 0.211 0.027 0.449 0.11 0.04 3.01 (1) 64.68 (66)
% violent criminal 55 9,123 0.31 0.33 0.002 0.032 0.004 0.247 0.255 0.11 0.11 0.01 (1) 61.05 (53)
% sex criminal 46 7,523 0.49 0.37 0.015 0.072 0.029 0.309 0.251 0.11 0.11 0.04 (1) 48.95 (44)
Empathy content
Cognitive rating 75 12,776 0.35 0.16 0.001 0.083 0.001 0.225 0.224 0.11 0.11 0.00 (1) 76.17 (73)
Affective rating 75 12,776 0.88 0.21 0.125 0.071 0.195 0.024 0.414 0.13 0.09 3.06 (1) 77.27 (73)
Note.Kthe number of effect sizes; Nthe number of participants; Mand SD the mean and standard deviation for continuous moderators, used
to determine the high and low values of r;Bthe unstandardized regression coefficient that represents the amount of change in rfor a one-unit increase
in the moderator; SE the standard error associated with the Bcoefficient; ␤⫽the standardized regression coefficient; CI the confidence interval
surrounding ;rlow and rhigh the values of rat low (–1 SD) and high (1SD) values of the moderator, respectively; Qbetween the variance
accounted for by the moderator variable; Qwithin the residual variability in effect size. Values in parentheses are degrees of freedom.
p.01.
Table 4
Analysis of Variance for Categorical Moderator Variables Using a Mixed Effects Model
Predictor KN r SE
CI Q
Lower Upper Between Within
Empathy measure
IRI only 56 16,773 0.101 0.023 0.160 0.042 10.47 (3)
105.01 (98)
HES only 5 0.332 0.078 0.532 0.132
QMEE only 6 0.106 0.068 0.281 0.069
Other 35 0.061 0.029 0.136 0.014
Aggression type
Physical only 24 16,673 0.104 0.039 0.204 0.004 0.25 (2) 104.03 (99)
Sexual only 37 0.082 0.029 0.157 0.007
General 41 0.098 0.026 0.165 0.031
Aggression measurement
Direct measurement (0) 43 16,773 0.133 0.026 0.200 0.066 3.85 (1) 104.49 (100)
Group difference (1) 59 0.065 0.023 0.124 0.006
Aggression method
Self-report (0) 29 9,067 0.128 0.029 0.203 0.053 0.68 (1) 50.22 (37)
Archival (1) 10 0.139 0.051 0.270 0.008
Publication type
Unpublished (0) 30 16,773 0.104 0.024 0.166 0.042 0.37 (1) 107.97 (100)
Published (1) 72 0.114 0.016 0.155 0.073
Note.Knumber of samples; Nsample size; reffect size; SE standard error; CI the confidence interval surrounding r;Qbetween the
variance accounted for by the moderator variable; Qwithin the residual variability in effect size; IRI Interpersonal Reactivity Index; HES Hogan
Empathy Scale; QMEE Questionnaire Measure of Emotional Empathy; Other any empathy measure other than the IRI, HES, and QMEE, or a
combination of measures. “General” refers to any nonspecific measure of aggression or combination of measures. “Direct measurement” means that
aggression was measured using self-report, laboratory task (e.g., Taylor shock paradigm), or historical data (e.g., number of crimes); “Group difference”
means that empathy was compared across two groups that were characterized by different levels of aggression (e.g., violent vs. nonviolent criminals).
Values in parentheses are degrees of freedom.
p.01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
14 VACHON, LYNAM, AND JOHNSON
gests that this distinction was unimportant because all measures of
empathy weakly predicted aggression. Furthermore, item-level
ratings of cognitive and affective content did not moderate this
finding. For example, the Empathic Concern subscale of the IRI,
which received the highest affective empathy item ratings and is
the most common measure of affective empathy across all studies,
predicted only 0.25% of the variance in aggression.
There are at least two alternative explanations for our failure to
find a moderate relation between empathy and aggression: (a) the
true association between empathy and aggression is weak, and our
results simply reflect this fact; or (b) the true association between
empathy and aggression is strong, but its manifestation is dimin-
ished by measurement problems. Before considering these alter-
natives, we turn our attention to the one clear exception to our
findings—the HES, which yielded moderate to large relations with
aggression.
Why the HES Works
Although the HES did a better job of predicting aggression than
all other measures of empathy, it is a poor measure of empathy.
Developed by Hogan (1969) over 40 years ago, the HES is a
64-item self-report measure constructed by combining 31 items
from the California Psychological Inventory (CPI; Gough, 1964),
25 items from the Minnesota Multiphasic Personality Inventory
(MMPI; Hathaway & McKinley, 1943), and 8 new items. These 64
items were selected because they were the best at distinguishing
groups rated as having high, medium, and low levels of empathy.
These groups were formed by taking a sample of 211 men,
including military officers (n100), research scientists (n45),
and graduate student engineers (n66), and splitting them into
high, medium, and low groups on the basis of observation by 8 –10
trained spectators who recorded their impressions of the partici-
pants over the course of a weekend “living-in” assessment. Ob-
servers were trained to think of the “empathic man” as one who (a)
is socially perceptive, (b) seems to be aware of the impression he
makes, (c) is skilled in social techniques of imaginative play, (d)
has insight into his own motives and behavior, and (e) evaluates
the motivation of others in interpreting situations. According to
Hogan (1969), “the content of these items is clearly relevant to
empathy: all reflect insight, perceptiveness, and social acuity” (p.
309). In contrast, observers were trained to think of the “unem-
pathic man” as one who (a) does not vary roles and relates to
everyone in the same way, (b) judges himself and others in
conventional terms like “popularity” and “the correct thing to do,”
(c) is uncomfortable with uncertainty and complexities, (d) is
extrapunitive and tends to project blame, and (e) handles anxiety
and conflicts by refusing to recognize their presence. According to
Hogan, “These items embody a complex dimension of conven-
tionality, anti-intraception, and reliance on the defensive tech-
niques of projection and rejection” (p. 309).
There are several issues worth noting about this procedure. First,
Hogan’s (1969) empathic man seems to represent a blend of high
cognitive empathy, social ability, and insight. However, Hogan’s
unempathic man is not a person lacking in these traits but rather a
mix of rigidity and blame externalization. Therefore, the trained
observers in Hogan’s study were asked to split a group of men into
high, medium, and low empathy groups using a conceptualization
of empathy that differs markedly from typical definitions and
includes a healthy dose of interpersonal antagonism at the low end
of measurement. This difference in itself brings empathy closer to
aggression. Second, items from the CPI and MMPI were selected
solely on the basis of predicting membership to these empathy
groups. No consideration was given to the content validity of the
selected items, as is evident from many HES items (e.g., “I liked
Alice in Wonderland by Lewis Carroll,” “I am afraid of deep
water,” “I would like the job of a foreign correspondent for a
newspaper,” “I prefer a shower to a bathtub”). Third, several items
from the HES tap externalizing traits, such as rule-breaking (e.g.,
“Sometimes I rather enjoy going against the rules and doing things
I’m not supposed to”), dominance (e.g., “I am usually a leader in
my group”), manipulation (e.g., “I have a natural talent for influ-
encing people”), narcissism (e.g., “I am an important person”), and
aggression (e.g., “sometimes I enjoy hurting persons I love”). This
extra-empathy item content almost certainly increases the ability
of the HES to predict aggression.
Taken together, these issues suggest that the HES lacks content
validity. This deficiency was also noted in the current meta-
analysis by the 19 raters who judged the item content of each
empathy measure: Their average item ratings for the HES (scored
from 0 not at all related to empathy to2highly related to
empathy) were 0.18 for cognitive empathy and 0.25 for affective
empathy. In comparison, cognitive and affective ratings were 0.27
and 1.52 for the IRI empathic concern subscale, 0.75 and 0.27 for
the IRI perspective taking subscale, and 0.39 and 1.11 for the
QMEE, a measure of affective empathy; because content ratings
worked for other empathy measures, low content ratings for the
HES were not a result of rating problems. Previous research also
suggests that the reliability of the HES is unsatisfactory. For
example, Froman and Peloquin (2001) found that the HES had an
alpha reliability of .57. Cross and Sharpley (1982) also found low
reliability (␣⫽.60), and that of 64 HES items, 30 were uncorre-
lated with the total score and 13 items had a negative correlation
with the total score.
Why Empathy and Aggression Are Weakly
Correlated: Two Alternatives
Aside from the exception of the HES, empathy and aggression
are weakly correlated. Based on theory, research, and practice, a
weak association with cognitive empathy is not particularly sur-
prising; however, a weak association with affective empathy is
startling. Similarly, a weak association with general empathy,
which includes measures with both cognitive and affective items,
seems to defy clinical wisdom. However, there are two very
different ways of interpreting this surprising finding.
Alternative 1: The true association between empathy and
aggression is weak. The first way of interpreting our findings is
that they reflect reality—the true association between empathy and
aggression is weak. Within this alternative, the results are an
accurate representation of the true association and previous expec-
tations regarding the role of empathy in aggression are too high.
We have already argued that expectations regarding a strong
negative relation between cognitive empathy and aggression are
probably unrealistic. Measures of cognitive empathy assess the
ability to detect and understand emotions, which is necessary but
insufficient for empathic responding. We cannot assume that
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
15
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
knowledge of others’ emotions will necessarily guide behavior in
a prosocial direction.
Similarly, expectations regarding the role of affective empathy
in aggression might also be too high. Measures of affective em-
pathy are designed to assess an individual’s ability to vicariously
experience emotions. Here, the assumption is that those who
experience less of this resonant emotional response will be more
likely to act without regard for the feelings of others. This line of
reasoning quickly extends beyond the content of the measure to
describe a callous, unemotional person who cares little about the
welfare of others. However, even if people who lack affective
empathy experience less of an emotional reaction, it does not mean
that they do not care about the welfare of others. There are
emotions and considerations outside of empathy, and there are
many reasons to care about others. While it is reasonable to expect
that those who lack affective empathy are less emotionally respon-
sive to others’ feelings and perhaps less motivated to inhibit their
aggression as a result, it may not be reasonable to expect that low
affective empathy will strongly predict aggression. Similarly, it
may be unreasonable to expect that high affective empathy will
strongly inhibit aggression. After all, parents are able to override
empathic responses to their children’s cries in order to discipline
them.
In the case of both cognitive and affective empathy, it is as-
sumed that the ability to understand and/or vicariously experience
others’ emotions inhibits aggressive behavior by causing concern
for the welfare of others. However, the mediating motivational
construct— concern for the welfare of others—is rarely repre-
sented in the content of popular empathy measures. Instead, these
measures assume that the comprehension and vicarious experience
of others’ emotions naturally leads to increased concern regarding
their welfare and a higher likelihood of prosocial behavior (or
lower likelihood of antisocial behavior). This string of assumptions
may be wrong— understanding emotions may empower one to
manipulate others, and feeling their discomfort may cause one to
flee the situation rather than help (i.e., personal distress). Because
empathic responding, versus empathic feeling, is informed by a
myriad of intervening forces that distance the initial flush of
empathy from its behavioral endpoint, items that ask about the
simple detection or experience of others’ emotions may be too
remote from behavior to predict it.
Alternative 2: The true association between empathy and
aggression is strong, but the observed association is diminished
by measurement problems. The second way of interpreting our
findings is that the true association between affective empathy and
aggression is stronger than observed in the present meta-analysis,
but its observable association is diminished by measurement prob-
lems. There are at least five measurement issues related to empathy
and aggression that might attenuate their observed association: (a)
Measures of empathy and/or aggression have low reliability; (b)
group differences in aggression may be small; (c) measures of
empathy and aggression have mismatched distributions; (d) mea-
sures of empathy rely on self-reports; and (e) measures of empathy
represent a construct that is too narrow. Each of these potential
measurement issues will be considered.
Measures of empathy and/or aggression have low reliability.
Because of imperfect reliability, the correlation between observed
scores on two measures will be smaller than the correlation be-
tween their respective unobservable true scores (J. Cohen, Cohen,
West, & Aiken, 2003). Although low reliability is a sufficient
cause of the failure to find anticipated relations, it is an unlikely
explanation for the present results. Studies with empathy or ag-
gression measures that reported an alpha .60 were excluded
from the meta-analysis; across studies that reported internal con-
sistency estimates, the average alpha was .76 for empathy mea-
sures and .84 for aggression measures. Even after correcting for
attenuation by artificially boosting the reliability of our empathy
and aggression measures to a perfect alpha of 1.00 (by dividing the
correlation by the square root of the product of the reliabilities), the
overall effect size only increases from r–.11 to r–.14.
Similarly, increases in effect size would be modest for verbal
aggression (from r–.20 to r–.25), physical aggression (from
r–.12 to r–.15), and sexual aggression (from r–.09 to r
–.11).
Group differences in aggression may be small. In the current
meta-analysis, 60% of the studies estimated aggression through
group comparison (violent vs. nonviolent or criminal vs. noncrim-
inal). A typical limitation to such criterion-group designs is that the
magnitude and reliability of group differences is ambiguous. Stud-
ies using dichotomous group comparisons rarely include continu-
ous measures of the proxy construct because the criterion group is
already characterized as an extreme manifestation of that construct
(e.g., violence for an aggression measure, self-harm for a depres-
sion measure, or overdose for an addiction measure). Although
there is little doubt that such extreme groups differ from their
comparison group, there is a question of the magnitude and con-
sistency of these differences across studies, particularly when the
composition of the comparison groups is vague. In the current
meta-analysis, for example, it is unclear how much more aggres-
sive violent criminals were than nonviolent criminals.
In any case, the potential cause for concern is that group
differences in aggression are so slight that empathy differences are
also small, thus lowering the overall effect size between empathy
and aggression. However, this seems not to be the case because a
moderator analysis comparing direct measures of aggression to the
group comparison method yielded no significant difference in the
empathy–aggression association. This result suggests that the two
methods of assessing aggression produce comparable relations to
empathy and that the overall small effect size is not attenuated by
the group comparison method.
Measures of empathy and aggression have mismatched
distributions. The more dissimilar the distribution of two vari-
ables, the greater the reduction in the maximum possible correla-
tion (J. Cohen et al., 2003). Because physical and sexual violence
are rare occurrences even in criminals, measures of these con-
structs are positively skewed (Huesmann, Eron, Lefkowitz, &
Walder, 1984). This contrasts with self-report measures of empa-
thy, which have approximately normal distributions. In the present
study, the relation between empathy and more normally distributed
verbal aggression (r–.20) was higher than the relations for more
skewed physical aggression (r–.12) and sexual aggression (r
–.09). However, while mismatch between the distributions of
empathy and aggression may attenuate their association, this ex-
planation alone is unsatisfactory. If mismatch between a normally
distributed predictor and a positively skewed outcome reduced
truly strong associations to weak observed effect sizes, most per-
sonality traits would fail to predict aggression. This is not the
case—a range of normally distributed personality traits produce
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
16 VACHON, LYNAM, AND JOHNSON
moderate to large effect sizes with aggression (e.g., antagonism,
hostility, straightforwardness, altruism, compliance, deliberation,
and urgency; J. D. Miller, Zeichner, & Wilson, 2012).
Measures of empathy rely on self-reports. In the current
meta-analysis, the vast majority of studies used a self-report mea-
sure of empathy (93%, k82). Only six studies used laboratory
methods (e.g., emotion recognition, affect matching) or a mixture
of self-report and laboratory tasks, and none used interviews or
informant reports of empathy. Because most studies used a self-
report measure of empathy, it is important to consider whether
there is a problem with the self-report method itself. There is little
evidence that this method is responsible for undermining a true
association between empathy and aggression. Other self-report
measures, including self-reported personality measures, do a good
job of predicting aggression. Previous meta-analytic research also
suggests that self-reports of empathy predict aggression more
strongly than other methods. For example, across children, ado-
lescents, and adults, P. A. Miller and Eisenberg (1988) found
significant relations between aggression and self-reported empathy
(average r–.18), but nonsignificant relations for all other modes
of assessing empathy, including picture/story measures (average
r–.06), facial/gestural reactions (average r–.06), and exper-
imental inductions of empathy (average r–.08).
This is not to suggest that self-report methods should be the
exclusive means of assessing empathy or any other construct.
Behavioral and psychophysiological measures, for example, may
offer more objectivity than self-report measures and can be used in
individuals for whom self-report measures are unreliable or un-
suitable (prisoners, children, etc.). Future research using a multi-
method approach combining self-report, behavioral, and psycho-
physiological measurements has the potential to significantly
broaden and advance our understanding of empathy.
Another concern is that self-report empathy measures may
prompt socially desirable responding, causing an upward shift in
the distribution of responses. This shift would compress the range
of responses around the top of the scale and the restricted variance
in scale scores would result in the attenuation of correlations. This
is not a major concern, however, because self-report measures of
empathy generally have a normal distribution and are uncorrelated
with measures of social desirability (Lawrence, Shaw, Baker,
Baron-Cohen, & David, 2004; Mehrabian & Epstein, 1972; Meh-
rabian & O’Reilly, 1980).
Another potential concern with self-reports is reading level,
because an inability to read the measure’s items will lead to greater
measurement error, lower reliability, and attenuated correlations.
Our results suggest that education significantly moderates the
association between empathy and aggression, with a correlation
approaching zero in samples with low education. Because there is
little research on the reading level of empathy measures, we
conducted a readability analysis of the three most common self-
report measures of empathy—the IRI, QMEE, and HES, used in
82% of the studies in this meta-analysis. The Flesch-Kincaid index
(Flesch, 1948) and SMOG index (McLaughlin, 1969) yielded
grade-level scores of 8.0 and 10.6 for the IRI, 7.1 and 9.8 for the
QMEE, and 6.8 and 9.2 for the HES. Across measures and indexes,
the average grade level of the most common self-report measures
of empathy is 8.6, which may be too high for some of the samples
relevant to the study of empathy and aggression (e.g., prison
samples) but not others.
Current measures of empathy assess a construct that is too
narrow. In the area of personality, one can conceive of narrow
traits (e.g., talkativeness), somewhat broader concepts (e.g., asser-
tiveness), and general dispositions (e.g., extraversion). Scales can
be developed to assess constructs at each level of abstraction.
Consequently, a key issue to be resolved in the initial develop-
mental stage is the scope or bandwidth of the construct. The
predictive capacity of a construct is maximized when its band-
width matches that of the criterion. Broad, global traits ought to do
a good job of predicting broad outcomes, while narrow, specific
traits ought to do a great job of predicting narrow, conceptually
similar outcomes. For example, comparing conscientiousness and
one of its subordinate traits, dutifulness, one would expect consci-
entiousness to better predict a general outcome like job perfor-
mance, and dutifulness to better predict a specific and relevant
outcome, like job attendance.
Accordingly, the bandwidth of empathy as it is typically con-
ceived and measured may be too narrow to predict aggression, a
broad and complex construct. This would explain why aggression
is robustly predicted by a broad trait like agreeableness, which
incorporates various narrow traits (trust, straightforwardness, al-
truism, compliance, modesty, and tendermindedness). Broadening
empathy to include similar constructs would likely increase its
ability to predict aggression, although doing so comes at the risk of
blurring its boundaries with similar constructs and straying from
theory. On the other hand, if empathy is construed as a very narrow
construct, it may be unreasonable to expect it to strongly predict
something as broad as aggression.
Choosing Between Alternative Explanations
When choosing between the alternatives explanations described
above—that the true association between empathy and aggression
is weak, or that the true association is strong but diminished by
measurement problems—we advocate a middle position. Specifi-
cally, we believe that the true association between aggression and
a narrow conceptualization of affective empathy is stronger than
we found and that measurement problems (related to reliability,
distributional match, reading level, etc.) attenuate the relation. The
question is whether to (a) leave empathy alone, try to minimize
measurement problems, and change our expectations regarding its
relation to aggression; or (b) reconceptualize empathy and increase
its breadth. Neither option is intrinsically superior—maintaining
fidelity to traditional conceptualizations of empathy leaves avail-
able a wealth of research on its development and correlates, while
reconceptualizing empathy may allow it to synchronize with con-
temporary thinking regarding its relation to antisocial and aggres-
sive behavior.
A New Conceptualization of Empathy
Our own recent work reconceptualizing empathy suggests that
the latter approach may be a useful one (Vachon & Lynam, 2013).
We believe that current conceptions of empathy are censored and
fail to capture the full range of the construct. Traditional theories
of empathy focus on how much a person’s feelings resonate with
those of others. Scales range from a high level of emotional
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
17
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
resonance to a low level of resonance. However, clinical descrip-
tions of low empathy in psychopathy, antisocial personality disor-
der, and narcissistic personality disorder extend beyond this range
to include more maladaptive manifestations of very low empathy,
such as callous disregard for the feelings of others, lack of remorse
for the misery caused by one’s actions, and scorn toward others’
emotional experience. Those who enjoy seeing others in pain are
afflicted with more than a mere lack of empathy; this dissonant
emotion, pleasure at another’s pain, has not been incorporated into
current measures of empathy. Constructing affective empathy as a
broader construct could accommodate basic and clinical concep-
tualizations of empathy. Specifically, at the high end of the scale
is a resonant response (empathetic, sympathetic, compassionate,
etc.), at the midpoint is a lack of response (callous, unemotional,
uncaring, etc.), and at the low end is a dissonant response (sadistic,
scornful, schadenfreude, etc.). This reconceptualization merges
similar constructs at the high end that have probably been distin-
guished to the point of distraction. It is also theoretically appealing
because it does not drift far from traditional views of empathy:
affective empathy is the resonance (or dissonance) of emotional
response.
Whether resonant and dissonant responses belong on the same
continuum is an empirical question, as is the question of whether
dissonance is really empathy at all. Preliminary evidence supports
the view that they belong on the same continuum. For example, the
Affective and Cognitive Measure of Empathy (ACME; Vachon &
Lynam, 2013) is a new measure that includes both Affective
Resonance and Affective Dissonance as separate scales. However,
these scales are very highly correlated (r–.85), showing that
they tap the same construct and fit together to form a single scale
of affective empathy. Importantly, the dissonance scale does a
much better job of predicting aggression than the resonance scale
or other measures of affective empathy, suggesting that broad-
ening the affective empathy construct into a maladaptive range
allows it to align with expectations regarding its role in aggres-
sion. Although more research is needed, these findings provide
evidence that assumptions regarding the role of empathy in
aggression may not be misguided after all.
Conclusion
The results of our meta-analysis seem to defy conventional
wisdom about the association between empathy and aggression.
The two constructs, as currently measured, share only the barest
sliver of variance. This finding is made all the more troubling by
the integral role empathy currently plays in the diagnosis of
externalizing disorders, assessment of future risk, and treatment of
offenders. We suggest that such applications of the empathy con-
struct have slipped the leash of science. How can we justify
spending several hundred million dollars each year on empathy
training programs for sex offenders when the accumulated science
of empathy yields such modest associations with aggression? We
know even less about the ability of empathy to predict recidivism,
and there are almost no data on the responsiveness of empathy to
training programs.
Although applications of empathy have outpaced its science, we
do not believe they are unfounded—strong theory, a long history
of clinical observation, and data from multiple levels of analysis
indicate a role for empathy deficits in aggressive behavior. Rather,
we suggest that the true association between empathy and aggres-
sion is diminished by measurement problems and further dimin-
ished by an overly narrow conceptualization of affective empathy.
We have provided preliminary evidence that broadening the affec-
tive empathy construct beyond resonant responses to include cal-
lous and dissonant responses unifies basic research on empathy
with clinical research on callousness. Broadening the construct
into the maladaptive range may also allow it to synchronize with
contemporary thinking regarding the role of empathy deficits in
aggressive behavior. We submit this new conceptualization of
empathy as a priority for research, particularly given the elevated
prominence of empathy throughout the DSM-5.
References
References marked with an asterisk indicate studies included in the
meta-analysis.
Abbey, A., Parkhill, M. R., Clinton-Sherrod, A. M., & Zawacki, T.
(2007). A comparison of men who committed different types of sexual
assault in a community sample. Journal of Interpersonal Violence, 22,
1567–1580. doi:10.1177/0886260507306489
Abbey, A., Parkill, M. R., BeShears, R., Clinton-Sherrod, A. M., &
Zawacki, T. (2006). Cross-sectional predictors of sexual assault perpe-
tration in a community sample of single African American and Cauca-
sian men. Aggressive Behavior, 32, 54 – 67. doi:10.1002/ab.20107
American Psychiatric Association. (2013). Diagnostic and statistical man-
ual of mental disorders (5th ed.). Arlington, VA: American Psychiatric
Publishing. doi:10.1176/appi.books.9780890425596.744053
Barry, C. T. (2003). Aggression and alternative responses following
performance feedback: The influences of self-esteem, narcissism, feed-
back valence, and evaluative standard (Doctoral dissertation). Available
from ProQuest Dissertations and Theses database. (No. AAI3101733)
Bartholow, B. D., Sestir, M. A., & Davis, E. B. (2005). Correlates and
consequences of exposure to video game violence: Hostile personality,
empathy, and aggressive behavior. Personality and Social Psychology
Bulletin, 31, 1573–1586. doi:10.1177/0146167205277205
Batson, C. D. (1991). The altruism question: Toward a social-
psychological answer. Hillsdale, NJ: Erlbaum.
Batson, C. D. (2009). These things called empathy: Eight related but
distinct phenomena. In J. Decety & W. Ickes (Eds.), The social neuro-
science of empathy (pp. 3–17). Cambridge, MA: MIT Press.
Bench, L. L. (1997). Rehabilitating offenders through the use of literature
and computer technology (Doctoral dissertation). Available from Pro-
Quest Dissertations and Theses database. (No. AAM9733985)
Birbaumer, N., Veit, R., Lotze, M., Erb, M., Hermann, C., Grodd, W., &
Flor, H. (2005). Deficient fear conditioning in psychopathy: A functional
magnetic resonance imaging study. Archives of General Psychiatry, 62,
799 – 805. doi:10.1001/archpsyc.62.7.799
Blair, J., Mitchell, D., & Blair, K. (2005). The psychopath: Emotion and
the brain. Oxford, England: Wiley-Blackwell.
Blair, K. S., Morton, J., Leonard, A., & Blair, R. J. R. (2006). Impaired
decision-making on the basis of both reward and punishment informa-
tion in individuals with psychopathy. Personality and Individual Differ-
ences, 41, 155–165. doi:10.1016/j.paid.2005.11.031
Blair, R. J. R. (2004). The roles of orbital frontal cortex in the modulation
of antisocial behavior. Brain and Cognition, 55, 198 –208. doi:10.1016/
S0278-2626%2803%2900276-8
Blair, R. J. R. (2005). Responding to the emotions of others: Dissociating
forms of empathy through the study of typical and psychiatric popula-
tions. Consciousness and Cognition: An International Journal, 14, 698 –
718. doi:10.1016/j.concog.2005.06.004
Blair, R. J. R., & Blair, K. S. (2009). Empathy, morality, and social
convention: Evidence from the study of psychopathy and other psychi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
18 VACHON, LYNAM, AND JOHNSON
atric disorders. In J. Decety & W. Ickes (Eds.), The social neuroscience
of empathy (pp. 139 –152). Cambridge, MA: MIT Press.
Book, A. S., & Quinsey, V. L. (2004). Psychopaths: Cheaters or warrior-
hawks? Personality and Individual Differences, 36, 33– 45. doi:10.1016/
S0191-8869%2803%2900049-7
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009).
Introduction to meta-analysis. Chichester, United Kingdom: Wiley. doi:
10.1002/9780470743386
Bovasso, G. B., Alterman, A. I., Cacciola, J. S., & Rutherford, M. J.
(2002). The prediction of violent and nonviolent criminal behavior in a
methadone maintenance population. Journal of Personality Disorders,
16, 360 –373. doi:10.1521/pedi.16.4.360.24124
Carr, L., Iacoboni, M., Dubeau, M. C., Mazziotta, J. C., & Lenzi, G. L.
(2003). Neural mechanisms of empathy in humans: A relay from neural
systems for imitation to limbic areas. Proceedings of the National
Academy of Sciences, USA, 100, 5497–5502. doi:10.1073/pnas
.0935845100
Carter, C. S., Harris, J., & Porges, S. W. (2009). Neural and evolutionary
perspectives on empathy. In J. Decety & W. Ickes (Eds.), The social
neuroscience of empathy (pp. 169 –182). Cambridge, MA: MIT Press.
Cleckley, H. (1941). The mask of sanity. St. Louis, MO: Mosby.
Clements, K., Holtzworth-Munroe, A., Schweinle, W., & Ickes, W.
(2007). Empathic accuracy of intimate partners in violent versus nonvi-
olent relationships. Personal Relationships, 14, 369 –388. doi:10.1111/
j.1475-6811.2007.00161.x
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple
regression/correlation analysis for the behavioral sciences (3rd ed.).
Mahway, NJ: Erlbaum.
Cohen, L. J., McGeoch, P. G., Watras-Gans, S., Acker, S., Poznansky, O.,
Cullen, K.,...Galynker, I. (2002). Personality impairment in male
pedophiles. Journal of Clinical Psychiatry, 63, 912–919. doi:10.4088/
JCP.v63n1009
Coid, J., Yang, M., Ullrich, S., Roberts, A., & Hare, R. D. (2009).
Prevalence and correlates of psychopathic traits in the household pop-
ulation of Great Britain. International Journal of Law & Psychiatry, 32,
65–73. doi:10.1016/j.ijlp.2009.01.002
Covell, C. N. (2002). Empathic deficits in sexual offenders: An integration
of affective, social, and cognitive constructs (Doctoral dissertation).
Available from ProQuest Dissertations and Theses database. (No.
AAI3074073)
Covell, C. N., Huss, M. T., & Langhinrichsen-Rohling, J. (2007). Em-
pathic deficits among male batterers: A multidimensional approach.
Journal of Family Violence, 22, 165–174. doi:10.1007/s10896-007-
9066-2
Cremer, J. M. (1996). Empathy and nurturance: A study of incest offend-
ers (Doctoral dissertation). Available from ProQuest Dissertations and
Theses database. (No. AAM9634645)
Cross, D. G., & Sharpley, C. F. (1982). Measurement of empathy with the
Hogan Empathy Scale. Psychological Reports, 50, 62. doi:10.2466/pr0
.1982.50.1.62
Daly, B. (2004). An investigation of empathy in adults as a function of
variables in three clusterings: Dispositional, situational, and biophilial
(Doctoral dissertation). Available from ProQuest Dissertations and The-
ses database. (No. AAINR00118)
Davis, C. E. (2010). The affective features of psychopathy in incarcerated
women (Doctoral dissertation). Available from ProQuest Dissertations
and Theses database. (No. AAI3401774)
Davis, M. H. (1980). A multidimensional approach to individual differ-
ences in empathy. Retrieved from http://www.uv.es/~friasnav/
Davis_1980.pdf
Davis, M. H. (1983). The effects of dispositional empathy on emotional
reactions and helping: A multidimensional approach. Journal of Per-
sonality, 51, 167–184. doi:10.1111/j.1467-6494.1983.tb00860.x
Day, A., Casey, S., & Gerace, A. (2010). Interventions to improve empathy
awareness in sexual and violent offenders: Conceptual, empirical, and
clinical issues. Aggression and Violent Behavior, 15, 201–208. doi:
10.1016/j.avb.2009.12.003
Deardorff, P. A., Finch, A. J.Jr., Kendall, P. C., Lira, F., & Indrisano, V.
(1975). Empathy and socialization in repeat offenders, first offenders,
and normals. Journal of Counseling Psychology, 22, 453– 455. doi:
10.1037/h0077127
Decety, J., & Jackson, P. L. (2004). The functional architecture of human
empathy. Behavioral and Cognitive Neuroscience Reviews, 3, 406 – 412.
doi:10.1177/1534582304267187
DeGue, S. (2007). Examining a predictive model of male sexual coercion
(Doctoral dissertation). Available from ProQuest Dissertations and The-
ses database. (No. AAI3250375)
DeGue, S., & DiLillo, D. (2004). Understanding perpetrators of nonphys-
ical sexual coercion: Characteristics of those who cross the line. Vio-
lence and Victims, 19, 673– 688. doi:10.1891/vivi.19.6.673.66345
de Waal, F. B. (2008). Putting the altruism back into altruism: The
evolution of empathy. Annual Review of Psychology, 59, 279 –300.
doi:10.1146/annurev.psych.59.103006.093625
DeWolfe, T. E., Jackson, L. A., & Winterberger, P. (1988). A comparison
of moral reasoning and moral character in male and female incarcerated
felons. Sex Roles, 18, 583–593. doi:10.1007/BF00287962
Diamond, P. M., & Magaletta, P. R. (2006). The short-form Buss-Perry
Aggression Questionnaire (BPAQ-SF): A validation study with federal
offenders. Assessment, 13, 227–240. doi:10.1177/1073191106287666
Dietzel, B. J. (2008). Emotion recognition and the propensity to engage in
sexually coercive behaviors: A study with college males (Doctoral dis-
sertation). Available from ProQuest Dissertations and Theses database.
(No. AAI3323520)
Dolan, M., & Fullam, R. (2004). Theory of mind and mentalizing ability
in antisocial personality disorders with and without psychopathy. Psy-
chological Medicine, 34, 1093–1102. doi:10.1017/S0033291704002028
D’Orazio, D. M. (2002). A comparative analysis of empathy in sexually
offending and non-offending juvenile and adult males (Doctoral disser-
tation). Available from ProQuest Dissertations and Theses database.
(No. AAI3104307)
Eisenberg, N. (2000). Emotion, regulation, and moral development. Annual
Review of Psychology, 51, 665– 697. doi:10.1146/annurev.psych.51.1
.665
Eisenberg, N., & Miller, P. A. (1987). The relation of empathy to prosocial
and related behaviors. Psychological Bulletin, 101, 91–119. doi:
10.1037/0033-2909.101.1.91
Eisenberg, N., & Strayer, J. (Eds.). (1987). Empathy and its development.
New York, NY: Cambridge University Press.
Elliott, I. A., Beech, A. R., Mandeville-Norden, R., & Hayes, E. (2009).
Psychological profiles of Internet sexual offenders: Comparisons with
contact sexual offenders. Sexual Abuse: A Journal of Research and
Treatment, 21, 76 –92. doi:10.1177/1079063208326929
Elsegood, K. J., & Duff, S. C. (2010). Theory of mind in men who have
sexually offended against children: A U.K. comparison study be-
tween child sex offenders and nonoffender controls. Sexual Abuse: A
Journal of Research and Treatment, 22, 112–131. doi:10.1177/
1079063209359926
Eslinger, P. J. (1998). Neurological and neuropsychological bases of em-
pathy. European Neurology, 39, 193–199. doi:10.1159/000007933
Eysenck, S. B., & McGurk, B. J. (1980). Impulsiveness and venturesome-
ness in a detention center population. Psychological Reports, 47, 1299 –
1306. doi:10.2466/pr0.1980.47.3f.1299
Fan, Y. T., Decety, J., Yang, C., Liu, J., & Cheng, Y. (2010). Unbroken
mirror neurons in autistic spectrum disorders. Journal of Child Psychol-
ogy and Psychiatry, 54, 981–988. doi:10.1111/j.1469-7610.2010
.02269.x
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
19
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
Fazel, S., & Danesh, J. (2002). Serious mental disorder in 23,000 prisoners:
A systematic review of 62 surveys. Lancet, 359, 545–550. doi:10.1016/
S0140-6736%2802%2907740-1
Fehr, E., & Fischbacher, U. (2004). Third-party punishment and social
norms. Evolution and Human Behavior, 25, 63– 87. doi:10.1016/S1090-
5138%2804%2900005-4
Feingold, A. (1994). Gender differences in personality: A meta-analysis.
Psychological Bulletin, 116, 429 – 456. doi:10.1037/0033-2909.116.3
.429
Fisher, D., Beech, A., & Browne, K. (1999). Comparison of sex offenders
to nonoffenders on selected psychological measures. International Jour-
nal of Offender Therapy and Comparative Criminology, 43, 473– 491.
doi:10.1177/0306624X99434006
Flesch, R. (1948). A new readability yardstick. Journal of Applied Psy-
chology, 32, 221–233. doi:10.1037/h0057532
Fletcher, P. C., Happe, F., Frith, U., Baker, S. C., Dolan, R. J., Frackowiak,
R. S., & Frith, C. D. (1995). Other minds in the brain: A functional
imaging study of “theory of mind” in story comprehension. Cognition,
57, 109 –128. doi:10.1016/0010-0277%2895%2900692-R
Flor, H., Birbaumer, N., Hermann, N., Ziegler, C., & Patrick, C. J. (2002).
Aversive conditioning in psychopaths: Peripheral and central correlates.
Psychophysiology, 39, 505–518. doi:10.1111/1469-8986.3940505
Francis, K. J., & Wolfe, D. A. (2008). Cognitive and emotional differ-
ences between abusive and non-abusive fathers. Child Abuse & Neglect,
32, 1127–1137. doi:10.1016/j.chiabu.2008.05.007
Froman, R. D., & Peloquin, S. M. (2001). Rethinking the use of the Hogan
Empathy Scale: A critical psychometric analysis. American Journal of
Occupational Therapy, 55, 566 –572. doi:10.5014/ajot.55.5.566
Fruzzetti, A. E., & Levensky, E. R. (2000). Dialectical behavior therapy for
domestic violence: Rationale and procedures. Cognitive and Behavioral
Practice, 7, 435– 447. doi:10.1016/S1077-7229%2800%2980055-3
Gallagher, H. L., & Frith, C. D. (2003). Functional imaging of “theory of
mind”. Trends in Cognitive Sciences, 7, 77– 83. doi:10.1016/S1364-
6613%2802%2900025-6
Gallese, V., & Goldman, A. (1998). Mirror neurons and the simulation
theory of mind reading. Trends in Cognitive Sciences, 2, 493–501.
doi:10.1016/S1364-6613%2898%2901262-5
Giancola, P. R. (2003). The moderating effects of dispositional empathy
on alcohol-related aggression in men and women. Journal of Abnormal
Psychology, 112, 275–281. doi:10.1037/0021-843X.112.2.275
Goldstein, A. P., Glick, B., & Gibbs, J. C. (1998). Aggression replacement
training: A comprehensive intervention for aggressive youth (Rev. ed.).
Champaign, IL: Research Press.
Goldstein, H. M. (1996). Empathy and attachment in violent and nonvi-
olent criminal offenders (Doctoral dissertation). Available from Pro-
Quest Dissertations and Theses database. (No. AAM9631037)
Goldstein, H., & Higgins-D’alessandro, A. (2001). Empathy and attach-
ment in relation to violent vs. non-violent offense history among jail
inmates. Journal of Offender Rehabilitation, 32, 31–53. doi:10.1300/
J076v32n04_03
Gopnik, A., & Meltzoff, A. N. (1998). Words, thoughts, and theories.
Cambridge, MA: MIT Press.
Gough, H. G. (1964). Manual for the California Psychological Inventory,
Revised. Palo Alto, CA: Consulting Psychologists Press.
Grant, B. F., Hasin, D. S., Stinson, F. S., Dawson, D. A., Chou, S. P., Ruan,
W. J., & Pickering, R. P. (2004). Prevalence, correlates, and disability of
personality disorders in the United States: Results from the National
Epidemiologic Survey on Alcohol and Related Conditions. Journal of
Clinical Psychiatry, 65, 948 –958. doi:10.4088/JCP.v65n0711
Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., & Cohen,
J. D. (2001). An fMRI investigation of emotional engagement in moral
judgment. Science, 293, 2105–2108. doi:10.1126/science.1062872
Grossman, D. C., Neckerman, H. J., Koepsell, T. D., Liu, P., Asher, K. N.,
Beland, K.,...Rivara, F. P. (1997). Effectiveness of a violence
prevention curriculum among children in elementary school. Journal of
the American Medical Association, 277, 1605–1611. doi:10.1001/jama
.1997.03540440039030
Gynn-Orenstein, J. (1981). The relationship between moral reasoning,
locus of control, emotional empathy, and parenting profile in physically
abusing mothers (Doctoral dissertation). Available from ProQuest Dis-
sertations and Theses database. (No. 1982-72567-001)
Hanson, R. K., & Morton-Bourgon, K. E. (2005). The characteristics of
persistent sexual offenders: A meta-analysis of recidivism studies. Jour-
nal of Consulting and Clinical Psychology, 73, 1154 –1163. doi:10.1037/
0022-006X.73.6.1154
Hanson, R. K., & Scott, H. (1995). Assessing perspective-taking among
sexual offenders, nonsexual criminals, and nonoffenders. Sexual Abuse:
A Journal of Research and Treatment, 7, 259 –277. doi:10.1007/
BF02256831
Hare, R. D. (1993). Without conscience: The disturbing world of the
psychopaths among us. New York, NY: Pocket Books.
Hare, R. D. (1998). Psychopathy, affect and behaviour. In D. Cooke, A.
Forth, & R. Hare (Eds.), Psychopathy: Theory, research and implica-
tions for society (pp. 105–137). Dordrecht, the Netherlands: Kluwer.
doi:10.1007/978-94-011-3965-6_6
Hare, R. D. (2003). The Hare Psychopathy Checklist—Revised (2nd ed.).
Toronto, Ontario, Canada: Multi-Health Systems.
Hare, R. D., & Neumann, C. S. (2008). Psychopathy as a clinical construct.
Annual Review of Clinical Psychology, 4, 217–246. doi:10.1146/annurev
.clinpsy.3.022806.091452
Harmon, P. A. (2001). Why do men batter women? Assessing empathy,
self-regard and narcissism levels, and attitudes toward women, men’s
roles and family of origin experiences among middle to upper class male
batterers (Doctoral dissertation). Available from ProQuest Dissertations
and Theses database. (No. AAI3036365)
Harpur, T. J., & Hare, R. D. (1994). Assessment of psychopathy as a
function of age. Journal of Abnormal Psychology, 103, 604 – 609. doi:
10.1037/0021-843X.103.4.604
Hart, S. D., & Hare, R. D. (1989). Discriminant validity of the Psychopathy
Checklist in a forensic psychiatric population. Psychological Assess-
ment, 1, 211–218. doi:10.1037/1040-3590.1.3.211
Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1994). Emotional conta-
gion. New York, NY: Cambridge University Press.
Hathaway, S. R., & McKinley, J. C. (1943). Manual for the Minnesota
Multiphasic Personality Inventory. New York, NY: Psychological Cor-
poration.
Haugen, R. E. (1998). The effects of perspective-taking training on em-
pathy development in adult male sex offenders (Doctoral dissertation).
Available from ProQuest Dissertations and Theses database. (No.
AAM9917742)
Hayashino, D. S., Wurtele, S. K., & Klebe, K. J. (1995). Child molesters:
An examination of cognitive factors. Journal of Interpersonal Violence,
10, 106 –116. doi:10.1177/088626095010001007
Hedges, L. V., & Olkin, I. (1985). Nonparametric estimators of effect size
in meta-analysis. Psychological Bulletin, 96, 573–580. doi:10.1037/
0033-2909.96.3.573
Heilbrun, A. B., Jr. (1982). Cognitive models of criminal violence based
upon intelligence and psychopathy levels. Journal of Consulting and
Clinical Psychology, 50, 546 –557. doi:10.1037/0022-006X.50.4.546
Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003).
Measuring inconsistency in meta-analysis. BMJ, 327, 557–560. doi:
10.1136/bmj.327.7414.557
Hogan, R. (1969). Development of an empathy scale. Journal of Consult-
ing and Clinical Psychology, 33, 307–316. doi:10.1037/h0027580
Hogan, R. (1973). Moral conduct and moral character: A psychological
perspective. Psychological Bulletin, 79, 217–232. doi:10.1037/
h0033956
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
20 VACHON, LYNAM, AND JOHNSON
Hoppe, C. M., & Singer, R. D. (1977). Interpersonal violence and its rela-
tionship to some personality measures. Aggressive Behavior, 3, 261–270.
doi:10.1002/1098-2337(1977)3:3261::AID-AB24800303073.0.CO;
2-U
Hosser, D., & Bosold, C. (2006). A comparison of sexual and violent
offenders in a German youth prison. The Howard Journal, 45, 159 –170.
doi:10.1111/j.1468-2311.2006.00412.x
Huesmann, L. R., Eron, L. D., Lefkowitz, M. M., & Walder, L. O. (1984).
Stability of aggression over time and generations. Developmental Psy-
chology, 20, 1120 –1134. doi:10.1037/0012-1649.20.6.1120
Ickes, W. (1993). Empathic accuracy. Journal of Personality, 61, 587– 610.
doi:10.1111/j.1467-6494.1993.tb00783.x
Jackson, P. L., Rainville, P., & Decety, J. (2006). To what extent do we
share the pain of others? Insight from the neural bases of pain empathy.
Pain, 125, 5–9. doi:10.1016/j.pain.2006.09.013
Johnson, L. K. (1996). The effects of watching a videotaped crying child
on the emotions and attributions of physically child abusive and matched
comparison mothers (Doctoral dissertation). Available from ProQuest
Dissertations and Theses database. (No. AAM9716544)
Jolliffe, D., & Farrington, D. P. (2004). Empathy and offending: A sys-
tematic review and meta-analysis. Aggression and Violent Behavior, 9,
441– 476. doi:10.1016/j.avb.2003.03.001
Jolly, A. (1966). Lemur social behavior and primate intelligence. Science,
153, 501–506. doi:10.1126/science.153.3735.501
Jones, S. E., Miller, J. D., & Lynam, D. R. (2011). Personality, antisocial
behavior, and aggression: A meta-analytic review. Journal of Criminal
Justice, 39, 329 –337. doi:10.1016/j.jcrimjus.2011.03.004
Kavussanu, M., & Boardley, I. D. (2009). The prosocial and antisocial
behavior in sport scale. Journal of Sport & Exercise Psychology, 31,
97–117. Retrieved from http://journals.humankinetics.com/jsep
Kiehl, K. A., Smith, A. M., Hare, R. D., Mendrek, A., Forster, B. B., Brink,
J., & Liddle, P. F. (2001). Limbic abnormalities in affective processing
by criminal psychopaths as revealed by functional magnetic resonance
imaging. Biological Psychiatry, 50, 677– 684. doi:10.1016/S0006-
3223%2801%2901222-7
Kohlberg, L. (1963). Moral development and identification. In H. W.
Stevenson, N. B. Henry, & H. G. Richey (Eds.), Child psychology: The
sixty-second yearbook of the National Society for the Study of Education
(Part 1, pp. 277–332). Chicago, IL: National Society for the Study of
Education.
Kupferberg, S. L. (2002). The relation between alexithymia and aggres-
sion in a nonclinical sample (Doctoral dissertation). Available from
ProQuest Dissertations and Theses database. (No. AAI3057030)
Kurtines, W., & Hogan, R. (1972). Sources of conformity in unsocialized
college students. Journal of Abnormal Psychology, 80, 49 –51. doi:
10.1037/h0033391
Lamm, C., Decety, J., & Singer, T. (2011). Meta-analytic evidence for
common and distinct networks associated with directly experienced pain
and empathy for pain. NeuroImage, 54, 2492–2502. doi:10.1016/j
.neuroimage.2010.10.014
Langevin, R., Wright, P., & Handy, L. (1988). Empathy, assertiveness,
aggressiveness, and defensiveness among sex offenders. Annals of Sex
Research, 1, 533–547. doi:10.1007/BF00854715
Lauterbach, O., & Hosser, D. (2007). Assessing empathy in prisoners: A
shortened version of the Interpersonal Reactivity Index. Swiss Journal of
Psychology, 66, 91–101. doi:10.1024/1421-0185.66.2.91
Lawrence, E. J., Shaw, P., Baker, D., Baron-Cohen, S., & David, A. S.
(2004). Measuring empathy: Reliability and validity of the Empathy
Quotient. Psychological Medicine, 34, 911–920. doi:10.1017/
S0033291703001624
Layman, C. A. (1995). Convicted rapists: How they differ in affect,
empathy, and attitudes from other criminals and college students (Doc-
toral dissertation). Available from ProQuest Dissertations and Theses
database. (No. AAM9601624)
LeDoux, J. E. (2000). Emotion circuits in the brain. Annual Review of
Neuroscience, 23, 155–184. doi:10.1146/annurev.neuro.23.1.155
Lengua, L. J., & Stormshak, E. A. (2000). Gender, gender roles, and
personality: Gender differences in the prediction of coping and psycho-
logical symptoms. Sex Roles, 43, 787– 820. doi:10.1023/A:
1011096604861
Levenson, R. W., & Ruef, A. M. (1992). Empathy: A physiological
substrate. Journal of Personality and Social Psychology, 63, 234. doi:
10.1037/0022-3514.63.2.234
Lipps, T. (1903). Einfühlung, inner Nachahmung, und Organ-
empfindungen [Empathy, inner imitation, and physical sensation]. Ar-
chiv für die gesamte Psychologie, 1, 185–204.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand
Oaks, CA: Sage.
Lisak, D., & Ivan, C. (1995). Deficits in intimacy and empathy in sexually
aggressive men. Journal of Interpersonal Violence, 10, 296 –308. doi:
10.1177/088626095010003004
Loeffler, C. H., Prelog, A. J., Unnithan, N., & Pogrebin, M. R. (2010).
Evaluating shame transformation in group treatment of domestic vio-
lence offenders. International Journal of Offender Therapy and Com-
parative Criminology, 54, 517–536. doi:10.1177/0306624X09337592
Loudin, J. L., Loukas, A., & Robinson, S. (2003). Relational aggression in
college students: Examining the roles of social anxiety and empathy.
Aggressive Behavior, 29, 430 – 439. doi:10.1002/ab.10039
Lovett, B. J., & Sheffield, R. A. (2007). Affective empathy deficits in
aggressive children and adolescents: A critical review. Clinical Psychol-
ogy Review, 27, 1–13. doi:10.1016/j.cpr.2006.03.003
Luo, Q., Nakic, M., Wheatley, T., Richell, R., Martin, A., & Blair, R. J. R.
(2006). The neural basis of implicit moral attitude? An IAT study using
event-related fMRI. NeuroImage, 30, 1449 –1457. doi:10.1016/j
.neuroimage.2005.11.005
Lykken, D. T. (1995). The antisocial personalities. Hillsdale, NJ: Erlbaum.
Lynam, D. R., & Derefinko, K. J. (2006). Psychopathy and personality. In
C. J. Patrick (Ed.), Handbook of psychopathy (133–155). New York,
NY: Guilford Press.
Marshall, W. L. (1999). Current status of North American assessment and
treatment programs for sexual offenders. Journal of Interpersonal Vio-
lence, 14, 221–239. doi:10.1177/088626099014003002
Marshall, W. L., Hudson, S. M., Jones, R., & Fernandez, Y. M. (1995).
Empathy in sex offenders. Clinical Psychology Review, 15, 99 –113.
doi:10.1016/0272-7358%2895%2900002-7
Marshall, W. L., Jones, R., Hudson, S. M., & McDonald, E. (1993).
Generalized empathy in child molesters. Journal of Child Sexual Abuse,
2, 61– 68. doi:10.1300/J070v02n04_04
Marshall, W. L., & Maric, A. (1996). Cognitive and emotional compo-
nents of generalized empathy deficits in child molesters. Journal of
Child Sexual Abuse, 5, 101–110. doi:10.1300/J070v05n02_06
Martínn, A. F., Vergeles, M. R., Acevedo, V. D., Sánchez, A. D., & Visa,
S. L. (2005). The involvement in sexual coercive behaviors of Spanish
college men: Prevalence and risk factors. Journal of Interpersonal
Violence, 20, 872– 891. doi:10.1177/0886260505276834
Mattek, P. W. (2003). The role of sexual fantasy in the confluence model
of sexual aggression (Doctoral dissertation). Available from ProQuest
Dissertations and Theses database. (No. AAI3092261)
McCabe, K., Houser, D., Ryan, L., Smith, V., & Trouard, T. (2001). A
functional imaging study of cooperation in two-person reciprocal ex-
change. Proceedings of the National Academy of Sciences, USA, 98,
11832–11835. doi:10.1073/pnas.211415698
McGinley, M., & Carlo, G. (2007). Two sides of the same coin? The
relations between prosocial and physically aggressive behaviors. Journal
of Youth and Adolescence, 36, 337–349. doi:10.1007/s10964-006-
9095-9
McGinley, T. G. (2007). Alexithymia, perception of nonverbal emotional
cues, and empathy: The prosocial skill deficits of substance abusing
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
21
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
offenders and their responsivity to treatment (Doctoral dissertation).
Available from ProQuest Dissertations and Theses database. (No.
AAI3297909)
McGrath, R., Cumming, G., Burchard, B., Zeoli, S., & Ellerby, L. (2010).
Current practices and emerging trends in sexual abuser management:
The Safer Society 2009 North American Survey. Brandon, VT: Safer
Society Press.
McLaughlin, G. H. (1969). SMOG grading: A new readability formula.
Journal of Reading, 12, 639 646. Retrieved from http://www.jstor.org/
stable/40011226
Mehrabian, A., & Epstein, N. (1972). A measure of emotional empathy.
Journal of Personality, 40, 525–543. doi:10.1111/j.1467-6494.1972
.tb00078.x
Mehrabian, A., & O’Reilly, E. (1980). Analysis of personality measures in
terms of basic dimensions of temperament. Journal of Personality and
Social Psychology, 38, 492–503. doi:10.1037/0022-3514.38.3.492
Miller, J., Lynam, D. R., Widiger, T. A., & Leukefeld, C. (2001). Person-
ality disorders as extreme variants of common personality dimensions:
Can the five-factor model adequately represent psychopathy? Journal of
Personality, 69, 253–276. doi:10.1111/1467-6494.00144
Miller, J. D., Zeichner, A., & Wilson, L. F. (2012). Personality correlates
of aggression: Evidence from measures of the five-factor model, UPPS
model of impulsivity, and BIS/BAS. Journal of Interpersonal Violence,
27, 2903–2919. doi:10.1177/0886260512438279
Miller, P. A., & Eisenberg, N. (1988). The relation of empathy to aggres-
sive and externalizing/antisocial behavior. Psychological Bulletin, 103,
324 –344. doi:10.1037/0033-2909.103.3.324
Mooradian, T. A., Davis, M., & Matzler, K. (2011). Dispositional empathy
and the hierarchical structure of personality. American Journal of Psy-
chology, 124, 99 –109. doi:10.5406/amerjpsyc.124.1.0099
Mullins-Nelson, J. L., Salekin, R. T., & Leistico, A. R. (2006). Psychop-
athy, empathy, and perspective-taking ability in a community sample:
Implications for the successful psychopathy concept. International Jour-
nal of Forensic Mental Health, 5, 133–149. doi:10.1080/14999013.2006
.10471238
Nettle, D. (2007). Empathizing and systemizing: What are they, and what
do they contribute to our understanding of psychological sex differ-
ences? British Journal of Psychology, 98, 237–255. doi:10.1348/
000712606X117612
Nussbaum, D., Collins, M., Cutler, J., Zimmerman, W., Farguson, B., &
Jacques, I. (2002). Crime type and specific personality indicia: Clon-
inger’s TCI impulsivity, empathy, and attachment subscales in non-
violent, violent, and sexual offenders. American Journal of Forensic
Psychology, 20, 23–56.
Parton, F., & Day, A. (2002). Empathy, intimacy, loneliness and locus of
control in child sex offenders: A comparison between familial and
non-familial child sexual offenders. Journal of Child Sexual Abuse, 11,
41–57. doi:10.1300/J070v11n02_03
Pecukonis, E. V. (1990). A cognitive/affective empathy training program
as a function of ego development in aggressive adolescent females.
Adolescence, 25, 59 –76.
Pickett, L. A. (2006). Trait versus state empathy: Evaluating the stability
of empathy and the relationship between negative affect and empathy in
an incarcerated sex offender population (Doctoral dissertation). Avail-
able from ProQuest Dissertations and Theses database. (No.
AAI3255336)
Pithers, W. D. (1994). Process evaluation of a group therapy component
designed to enhance sex offenders’ empathy for sexual abuse survivors.
Behaviour Research and Therapy, 32, 565–570. doi:10.1016/0005-
7967%2894%2990146-5
Pithers, W. D. (1999). Empathy: Definition, enhancement, and relevance
to the treatment of sexual abusers. Journal of Interpersonal Violence, 14,
257–284. doi:10.1177/088626099014003004
Porges, S. W. (2007). The polyvagal perspective. Biological Psychology,
74, 116 –143. doi:10.1016/j.biopsycho.2006.06.009
Preston, S. D., & de Waal, F. B. M. (2002). Empathy: Its ultimate and
proximate bases. Behavioral and Brain Sciences, 25, 1–72. doi:10.1017/
S0140525X02000018
Proctor, T., & Beail, N. (2007). Empathy and theory of mind in offenders
with intellectual disability. Journal of Intellectual & Developmental
Disability, 32, 82–93. doi:10.1080/13668250701373331
Rice, M. E., Chaplin, T. C., Harris, G. T., & Coutts, J. (1994). Empathy
for the victim and sexual arousal among rapists and nonrapists.
Journal of Interpersonal Violence, 9, 435– 449. doi:10.1177/
088626094009004001
Richardson, D. R., Green, L. R., & Lago, T. (1998). The relationship
between perspective-taking and nonaggressive responding in the face of
an attack. Journal of Personality, 66, 235–256. doi:10.1111/1467-6494
.00011
Richardson, D. R., Hammock, G. S., Smith, S. M., Gardner, W., & Signo,
M. (1994). Empathy as a cognitive inhibitor of interpersonal aggres-
sion. Aggressive Behavior, 20, 275–289. doi:10.1002/1098-
2337%281994%2920:4%3C275::AID-AB2480200402%3E3.0.CO;2-4
Rilling, J. K., Sanfey, A. G., Aronson, J. A., Nystrom, L. E., & Cohen, J. D.
(2004). The neural correlates of theory of mind within interpersonal
interactions. NeuroImage, 22, 1694 –1703. doi:10.1016/j.neuroimage
.2004.04.015
Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistency
of personality traits from childhood to old age: A quantitative review of
longitudinal studies. Psychological Bulletin, 126, 3–25. doi:10.1037/
0033-2909.126.1.3
Robins, L. N., Tipp, J., & Przybeck, T. (1991). Antisocial personality. In
L. N. Robins & D. A. Regier (Eds.), Psychiatric disorders in America
(258 –290). New York, NY: Free Press.
Ross, R., & Ross, R. (1995). Thinking straight: The reasoning and reha-
bilitation program for delinquency prevention and offender rehabilita-
tion. Ottawa, Ontario, Canada: Air Training.
Samuel, D. B., & Widiger, T. A. (2008). A meta-analytic review of the
relationships between the five-factor model and DSM-IV-TR personality
disorders: A facet level analysis. Clinical Psychology Review, 28, 1326 –
1342. doi:10.1016/j.cpr.2008.07.002
Sandoval, A. R., Hancock, D., Poythress, N., Edens, J. F., & Lilienfeld, S.
(2000). Construct validity of the Psychopathic Personality Inventory in
a correctional sample. Journal of Personality Assessment, 74, 262–281.
doi:10.1207/S15327752JPA7402_7
Sartin, R. M. (2004). Characteristics associated with domestic violence
perpetration: An examination of factors related to treatment response
and the utility of a batterer typology (Doctoral dissertation). Available
from ProQuest Dissertations and Theses database. (No. AAI3142099)
Schaffer, M., Clark, S., & Jeglic, E. L. (2009). The role of empathy and
parenting style in the development of antisocial behaviors. Crime &
Delinquency, 55, 586 –599. doi:10.1177/0011128708321359
Schweinle, W., Ickes, W., Rollings, K., & Jacquot, C. (2010). Maritally
aggressive men: Angry, egocentric, impulsive, and/or biased. Journal of
Language and Social Psychology, 29, 399 – 424. doi:10.1177/
0261927X10377988
Sergeant, M. J. T., Dickins, T. E., Davies, M. N. O., & Griffiths, M. D.
(2006). Aggression, empathy, and sexual orientation in males. Person-
ality and Individual Differences, 40, 475– 486. doi:10.1016/j.paid.2005
.07.002
Serin, R. C., & Kuriychuk, M. (1994). Social and cognitive deficits in
violent offenders: Implications for treatment. International Journal of
Law and Psychiatry, 17, 431– 441. doi:10.1016/0160-2527(94)90018-3
Seto, M. C., & Barbaree, H. E. (1993). Victim blame and sexual arousal
to rape cues in rapists and nonoffenders. Sexual Abuse: A Journal of
Research and Treatment, 6, 167–183. doi:10.1007/BF00849559
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
22 VACHON, LYNAM, AND JOHNSON
Sever, W., & Harriette, A. (2006). Affect regulation in adult male sex
offenders: Alexithymia as a risk factor (Doctoral dissertation). Available
from ProQuest Dissertations and Theses database. (No. AAI3230483)
Shamay-Tsoory, S. G. (2009). Empathic processing: Its cognitive and
affective dimensions and neuroanatomical basis. In J. Decety & W. Ickes
(Eds.), The social neuroscience of empathy (pp. 215–232). Cambridge,
MA: MIT Press.
Shamay-Tsoory, S. G., Tomer, R., Berger, B. D., Goldsher, D., & Aharon-
Peretz, J. (2005). Impaired “affective theory of mind” is associated with
right ventromedial prefrontal damage. Cognitive and Behavioral Neu-
rology, 18, 55– 67. doi:10.1097/01.wnn.0000152228.90129.99
Shaw, L. L. (1995). Empathy as an inhibitor of aggression (Doctoral
dissertation). Available from ProQuest Dissertations and Theses data-
base. (No. AAM9623368)
Shoss, N. E. (2006). Personality pathology moderates the relationship
between length of time in treatment and levels of empathy in incarcer-
ated male sex offenders (Doctoral dissertation). Available from ProQuest
Dissertations and Theses database. (No. AAI3248397)
Simon, J. L. (2002). Analysis of the relationship between shame, guilt and
empathy in intimate relationship violence (Doctoral dissertation). Avail-
able from ProQuest Dissertations and Theses database. (No.
AAI3057623)
Simons, D., Wurtele, S. K., & Heil, P. (2002). Childhood victimization
and lack of empathy as predictors of sexual offending against women
and children. Journal of Interpersonal Violence, 17, 1291–1307. doi:
10.1177/088626002237857
Singer, T., Seymour, B., O’Doherty, J., Kaube, H., Dolan, R. J., & Frith,
C. D. (2004). Empathy for pain involves the affective but not sensory
components of pain. Science, 303, 1157–1162. doi:10.1126/science
.1093535
Smallbone, S. W., Wheaton, J., & Hourigan, D. (2003). Trait empathy and
criminal versatility in sexual offenders. Sexual Abuse: A Journal of
Research and Treatment, 15, 49 – 60. doi:10.1023/A:1020615807663
Smith, P., Waterman, M., & Ward, N. (2006). Driving aggression in
forensic and non-forensic populations: Relationships to self-reported
levels of aggression, anger and impulsivity. British Journal of Psychol-
ogy, 97, 387– 403. doi:10.1348/000712605X79111
Stuewig, J., Tangney, J. P., Heigel, C., Harty, L., & McCloskey, L.
(2010). Shaming, blaming, and maiming: Functional links among the
moral emotions, externalization of blame, and aggression. Journal of
Research in Personality, 44, 91–102. doi:10.1016/j.jrp.2009.12.005
Sturgeon, C. L. (2003). Typology of violence and alexithymia, empathy,
perfectionism, and substance abuse in federal offenders (Doctoral dis-
sertation). Available from ProQuest Dissertations and Theses database.
(No. AAINQ82829)
Teten, A. L., Miller, L. A., Bailey, S. D., Dunn, N. J., & Kent, T. A.
(2008). Empathic deficits and alexithymia in trauma-related impulsive
aggression. Behavioral Sciences and the Law, 26, 823– 832. doi:
10.1002/bsl.843
Travis, R. (2008). The relationship between empathy and sexual arousal
in civilly-committed sex offenders (Doctoral dissertation). Available
from ProQuest Dissertations and Theses database. (No. AAI3317870)
Trivers, R. L. (1971). The evolution of reciprocal altruism. Quarterly
Review of Biology, 46, 35–57. doi:10.1086/406755
U.S. Department of Justice. (2007). Criminal victimization in the United
States, 2007: Statistical tables. Retrieved from http://bjs.ojp.usdoj.gov/
content/pub/pdf/cvus0702.pdf
U.S. Department of Justice. (2010). Federal justice statistics, 2008: Sta-
tistical tables. Retrieved from http://www.bjs.gov/content/pub/html/fjsst/
2008/tables/fjs08st502.pdf
Vachon, D. D., & Lynam, D. R. (2013). Development and validation of the
Affective and Cognitive Measure of Empathy (ACME). Manuscript sub-
mitted for publication.
Waldorf, V. A. (1997). The relationship between neuropsychological
deficits and spouse abuse: An examination of both abuser and victim
(Doctoral dissertation). Available from ProQuest Dissertations and The-
ses database. (No. AAM9813411)
Warkentin, J. B. (2008). Dating violence and sexual assault among
college men: Co-occurrence, predictors, and differentiating factors
(Doctoral dissertation). Available from ProQuest Dissertations and The-
ses database. (No. AAI3375098)
Watt, M. C., Frausin, S., Dixon, J., & Nimmo, S. (2000). Moral intelli-
gence in a sample of incarcerated females. Criminal Justice and Behav-
ior, 27, 330 –355. doi:10.1177/0093854800027003004
Wicker, B., Keysers, C., Plailly, J., Royet, J. P., Gallese, V., & Rizzolatti,
G. (2003). Both of us disgusted in my insula: The common neural basis
of seeing and feeling disgust. Neuron, 40, 655– 664. doi:10.1016/S0896-
6273(03)00679-2
Wilson, D. B. (2011). Meta-analysis macros for SAS, SPSS, and Stata
[Computer software]. Retrieved July 1, 2012, from http://mason.gmu
.edu/~dwilsonb/ma.html
Wispé, L. (1986). The distinction between sympathy and empathy: To call
forth a concept, a word is needed. Journal of Personality and Social
Psychology, 50, 314 –321. doi:10.1037/0022-3514.50.2.314
Wispé, L. (1990). History of the concept of empathy. In N. Eisenberg & J.
Strayer (Eds.), Empathy and its development (pp. 17–37). New York,
NY: Cambridge University Press.
Yardley, S. L. (1997). The perception of emotion and empathic ability: A
neuropsychological approach to the study of violent and nonviolent
offenders (Doctoral dissertation). Available from ProQuest Dissertations
and Theses database. (No. AAM9825808)
Received July 31, 2012
Revision received August 22, 2013
Accepted October 4, 2013
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
23
THE (NON)RELATION BETWEEN EMPATHY AND AGGRESSION
... SD = 6.14), living in the urban area of the city of Palermo (Italy), participated in the study. The sample size was derived based on a theoretical correlation coefficient of r = -.113 between empathy and aggression, as suggested by the recent meta-analyses of Vachon et al. (2014). By establishing the threshold probability for rejecting the null hypothesis (σ two-tailed, Type I error rate) at .05 and the probability of failing to reject the null hypothesis under the alternative hypothesis (Type II error rate) at .20, the exact sample size was computed to be 612. ...
... When considering the association between the two constructs, higher levels of empathy were linked to lower levels of aggression. This result appears in line with the traditional idea that sharing others' internal state and vicariously experiencing their distress would encourage supportive behaviours, and discourage aggressive behaviours (Vachon et al., 2014). However, the introduction of the attachment style profile as a predictive variable of both empathy and aggression made their association non-significant, supporting the assumption that the negative relation between empathy and aggression might be accounted for by the underlying role of the attachment style. ...
... Furthermore, the study was based only on self-report measures, and this may have produced an overestimation/un derestimation of the studied relationships. In this line, Vachon et al. (2014) suggested that the study of the relation between empathy and aggression would be more valid if direct detection tools, such as observation, were used. Additionally, even if we forced cluster analysis on ASQ scores to obtain theoretically guided insecure attachment styles, our findings seem to suggest that ASQ is more suitable for pondering only a general group of insecurely attached adults (e.g., Fuchshuber et al., 2018). ...
Article
Full-text available
This study aimed to examine the explaining and moderating role of attachment style profiles on the association between empathy and aggression. Participants were 548 Italian adults (M = 47.62 years, SD = 6.14) who completed a survey measuring attachment, empathy, and aggression. Using cluster analytic methods, initial results indicated two attachment style profiles to be considered (secure vs. insecure). However, we also extracted a more theoretically guided four-cluster solution including preoccupied, secure, fearful, and dismissing profiles. Moreover, structural equation modelling showed that higher levels of empathy linked to lower levels of aggression. Nonetheless, when introducing in the model the dichotomous or the multi-categorical attachment style profile variable as predictive of both empathy and aggression, their association became not significant, while secure attachment profile significantly presented higher levels of empathy and lower levels of aggression compared to the other profiles. Furthermore, attachment style profile moderated the link between empathy and aggression. Specifically, in the secure group empathy and aggression were negatively related, but no significant association was evidenced in the other groups. Findings are discussed in the light of the literature.
... In the context of aggressive behaviors, several schools of thought explain the relationship between empathy and aggression. For example, control theories of crime postulate that empathy can act as a form of internal control whereby the more empathic an individual is, the less likely they will be to engage in aggression, and vice versa (Vachon et al., 2014). When considered from a learning theory perspective, empathic individuals are more likely to feel that their own aggressive behaviors vicariously punish a perceived outgroup member (Vachon et al., 2014). ...
... For example, control theories of crime postulate that empathy can act as a form of internal control whereby the more empathic an individual is, the less likely they will be to engage in aggression, and vice versa (Vachon et al., 2014). When considered from a learning theory perspective, empathic individuals are more likely to feel that their own aggressive behaviors vicariously punish a perceived outgroup member (Vachon et al., 2014). Prior research has found empathy to be a key factor in preventing and diminishing the likelihood of aggression more generally (Miller & Eisenberg, 1988) and in specific contexts, such as reducing aggression towards racial outgroups (Cikara et al., 2011). ...
Article
Full-text available
Prejudice and bias-motivated aggression (BMA) are pervasive social problems. Scholars have tested numerous competing theoretical models to demonstrate the key predicates of prejudice and BMA, including intergroup contact, dual process (i.e., right-wing authoritarianism and social dominance orientation), perceived injustice, peer socialization, and empathy. Yet, studies to date have not empirically examined the comparative strength of these theoretical perspectives to explain the correlates of (a) prejudice and (b) BMA. This study seeks to address this gap. Utilizing a sample of young 1,001 Belgian participants, this study explores the association between key constructs from different theoretical perspectives to better understand prejudice and BMA towards immigrant populations. Findings show that when accounting for all models of prejudice and BMA, the strongest predictors of prejudice emerge from the dual-process model, the empathy model (outgroup empathy), and the quality (not frequency) of intergroup contact. Yet, prejudice and exposure to peer outgroup hostility are the strongest predictors of BMA. We discuss the implications of our findings and suggest that drawing on criminological theories of prejudice and BMA can be integrated to provide a more nuanced understanding of the nature of prejudice and BMA than what is currently known. We conclude by highlighting some directions for future research on prejudice and BMA.
... According to the empathy-altruism hypothesis, Batson (2011) argued that individuals' empathy is associated with their altruistic motivation and prosocial behavior. People who are in low levels of empathy are usually more aggressive and suffer interpersonal problems (Vachon et al., 2014;Mitsopoulou and Giovazolias, 2015). Meanwhile, moral elevation seems to be an effective way of mitigating the detrimental effects of low empathy since it is effective in instigating subsequent prosocial actions (Haidt, 2003). ...
Article
Full-text available
The relationship between exposure to prosocial media content and prosocial behavior has been extensively explored. However, previous studies mainly explore the effect of prosocial media content exposure by comparing an individual’s exposure to the different types of content (i.e., prosocial content, or neutral content), and generally focus on traditional media and video games, with less attention given to the increasingly popular new media platforms. In this study, we explored new dimensions by considering individuals’ exposure to different consequences of the same prosocial behavior (i.e., reward, punishment, or no consequences) in the context of short videos. Drawing upon social cognitive theory and the general learning model, this experimental study identified the effect of such exposure on subsequent prosocial behavior among adolescents. We found that compared to the no consequences group, exposure to the reward consequence did not significantly predict moral elevation and subsequent prosocial behavior. Meanwhile, exposure to the punishment consequence significantly predicted subsequent prosocial behavior via moral elevation. Furthermore, the results revealed that empathy moderated the relationship between moral elevation and prosocial behavior, and moral elevation only positively predicted prosocial behavior among those with low empathy. Theoretically, this study deepens our understanding of the impact of exposure to different consequences of prosocial behavior on adolescents’ subsequent prosocial behavior, and highlights the importance of moral elevation and empathy to understand the underlying mechanism. The study also provides some practical implications for parents and practitioners to nurture prosocial behavior among adolescents.
... Moreover, the TP and its FC have been suggested to play a core role in social referring and social emotion (e.g., empathy) (Olson et al., 2007). The low DC and the FC strength may be related to a low level of social referring and empathy, which may lead to increased aggression (e.g., DA) (Austin et al., 2017;Vachon et al., 2014). The thalamus may be related to the increased sensitivity to threats (Pessoa & Adolphs, 2010;Yamamura et al., 2016),which leads to an individual being easily provoked by an innocent third party and carrying out DA. ...
Article
Background Hostile attribution bias (HAB) has been considered as a risk factor of various types of psychosocial adjustment problem, and contributes to displaced aggression (DA). The neural basis of HAB and the underlying mechanisms of how HAB predicts DA remain unclear. Methods The current study used degree centrality (DC) and resting-sate functional connectivity (RSFC) to investigate the functional connection pattern related to HAB in 503 undergraduate students. Furthermore, the “Decoding” was used to investigate which psychological components the maps of the RSFC-behavior may be related to. Finally, to investigate whether and how the RSFC pattern, HAB predicts DA, we performed mediation analyses. Results We found that HAB was negatively associated with DC in bilateral temporal poles (TP) and positively correlated with DC in the putamen and thalamus; Moreover, HAB was negatively associated with the strength of functional connectivity between TP and brain regions in the default mode network (DMN), and positively related to the strength of functional connectivity between the thalamus and regions in the DMN. The “Decoding” showed the maps of the RSFC-behavior may involve the theory mind, autobiographic, language, comprehension and working memory. Mediation analysis further showed that HAB mediated the relationship between some neural correlates of the HAB and DA. Limitations The current results need to be further tested by experimental methods or longitudinal design in further studies. Conclusions These findings shed light on the neural underpinnings of HAB and provide a possible mediation model regarding the relationships among RSFC pattern, HAB, and displaced aggression.
... SCM a BIAS Map predikujú podmienky, za ktorých sa ľudia majú v úmysle zapojiť do facilitujúceho správania voči iným skupinám. Ľudia, ktorí prejavujú aktívnu facilitáciu voči stigmatizovanej skupine (pomáhajúce/prosociálne správanie) sú motivovaní vnímanou srdečnosťou, pričom sprostredkujúce sú pociťované emócie ako obdiv alebo ľútosť.Štúdie preukázali, že prosociálne správanie je pozitívne spojené s empatiou (napr.Chaparro, & Grusec, 2016), ktorá predstavuje kritickú zručnosť umožňujúcu deťom adaptívne fungovať v sociálnom kontexte a je negatívnym prediktorom agresie (napr.Björkqvist et al., 2000;Vachon et al., 2014), šikany (napr. van Noorden et al., 2014) v detstve, a je tiež negatívnym prediktorom asociálneho správania v dospievaní a dospelosti(Bernstein et al., 1996). ...
Conference Paper
Full-text available
V príspevku sa zameriavame na pilotné overenie meracích nástrojov v rámci výskumu zameraného na viacnásobnú sociálnu kategorizáciu a stigmatizáciu znevýhodnených skupín na Slovensku. Prostredníctvom kognitívnych interview sme kvalitatívne overovali upravenú verziu BIAS Map, znenie otázok na prosociálne správanie, kontakt a vinety opisujúce minoritného rovesníka. Kognitívnych interview sa zúčastnilo 25 participantov (14 - 18 rokov). V príspevku diskutujeme ťažkosti a návrhy na zmeny, ktoré účastníci mali. Výsledky naznačujú, že účastníci chápu predložené položky a otázky tak, ako ich zamýšľame. Účastníci sa zhodli na tom, čo otázky znamenajú a na čo sa ich pýtame. Po hlbšom dopytovaní sme dostali od respondentov návrhy na zmeny, ktoré sú tiež prezentované a diskutované v tejto štúdii. Kľúčové slová: Bias Map, viacnásobná kategorizácia a stigmatizácia, kognitívne interview, pilotný výskum
... Difficulties in empathy, defined as the ability to understand and share another person's emotional experience, seemed to play a key role in the maladaptive behavioral regulation of IPV perpetrators [10,11]. Indeed, impaired empathy has been linked to the occurrence of violent behaviors [12,13], including IPV perpetration [14,15]. Specifically, difficulties in understanding the thoughts and feelings of the partner could predispose someone to IPV by promoting affective distress when dealing with emotional conflicts [16][17][18][19]. ...
Article
Full-text available
Empathy deficits have been proposed to be an important factor for intimate partner violence (IPV). IPV perpetrators have shown a differential change in salivary oxytocin (sOXT), testosterone (sT), and cortisol (sC), following empathic and stress tasks, compared to non-violent men. However, the influence of empathic deficits in those hormones after an emotion-induction task in IPV perpetrators remains unclear. We analyzed the effects of an empathic induction task on endogenous sOXT, sT and sC levels, as well as their hormonal ratios, in IPV perpetrators (n = 12), and compared them to controls (n = 12). Additionally, we explored the predictive capacity of em-pathy-related functions (measured with the interpersonal reactivity index) in the hormonal responses to the task. IPV perpetrators presented lower sOXT changes and higher total sT levels than controls after the task, lower sOXT/T change and total sOXT/T levels, as well as higher total sT/C levels. Notably, for all participants, the lower the perspective taking score, the lower the total sOXT levels and sOXT changes and the higher the sT changes were. Low perspective taking also predicted smaller sOXT/T and sOXT/C changes in the empathic induction task, and higher total sT/C levels for all participants. Therefore, our results could contribute to furthering our ability to focus on new therapeutic targets, increasing the effectiveness of intervention programs and helping to reduce IPV recidivism in the medium term.
Article
Teen dating violence (TDV) refers to a wide range of partner-directed harmful behaviors among adolescents. Since it was first documented in the 1980s, there is a growing interest in TDV due to its potentially devastating mid- and long-term consequences. Aiming at the early detection of TDV onset with prevention purposes, research has focused on the detection of typologies of perpetrators and/or victims as well as on identifying risk and protective factors for its occurrence. Research with Spanish adolescents, however, is very limited. To fill this gap, we recruited a total of 2,319 adolescents from different regions in Spain, out of which, 1,079 reported having had a romantic partner during the last year. These filled out measures of TDV (perpetration and victimization), school aggression, hostile and benevolent sexism, empathy, assertiveness, psychological inflexibility (general measures), and psychological inflexibility with prejudice thoughts. A cluster analysis revealed that adolescents could be divided into two clusters as a function of their TDV profile: Cluster 1, including close to 76% of the sample (boys and girls), presented low TDV perpetration and victimization; Cluster 2, including 24% of the sample (boys and girls), presented higher TDV perpetration and victimization. Regression analyses revealed that, as compared to those in Cluster 1, adolescents in Cluster 2 were more likely to be older boys who scored high in benevolent sexism, overt and relational school aggression, and personal distress, and low in behavior regulation skills, perspective taking, and practical personal ability. We discuss the implications of these findings for the design of evidence-based TDV prevention campaigns.
Article
Full-text available
To investigate the effects of left-behind children’s empathy and coping styles on school adjustment, 605 left-behind children in the third grade from three rural elementary schools in Suzhou, Anhui Province were selected to complete the Chinese version of the Interpersonal Reaction Index Scale, the Coping Style Scale, and the School Adjustment Behavior Scale for Primary and Secondary School Students. The results showed that (1) emotional empathy positively predicted children’s social competence, and negatively predicted children’s antisocial behavior; cognitive empathy positively predicted children’s social competence; (2) the role of positive coping styles mediated the relationship between cognitive empathy and social competence, and that between cognitive empathy and antisocial behavior; negative coping styles mediated the relationship between cognitive empathy and antisocial behavior; both positive and negative coping styles mediated the relationship between emotional empathy and social competence, and that between emotional empathy and antisocial behavior. The findings of the study are valuable for understanding the relationship between empathy and school adjustment, which also helps to enhance the school adjustment of left-behind children.
Article
BACKGROUND: Clinical empathy leads to improved patient satisfaction and better clinical outcomes. Currently, there are multiple empathy scales with minimal or no efforts to produce an integrated definition of clinical empathy which can be assessed sufficiently by only a few scales. Moreover, there is an unclear overall reliability of these empathy scales, hence limiting comparative evaluation. AIM: To examine which empathy scales have been used in healthcare students and to estimate their overall internal consistency. METHODS: A systematic review was performed with inclusion criteria any empirical study with quantitative data examining empathy of healthcare students toward patients between 2012 and 2016. A random effects model was used to produce a pooled estimate of the Cronbach’s alphas. The Hakstian-Whalen transformation was used for analyses based on the Rodriguez-Maeda method. Heterogeneity was quantified using the I2 statistic and further investigated with subgroup analysis and meta-regression. Publication bias was assessed using funnel plots, Egger’s test, Begg’s test, and the trim and fill analysis. RESULTS: Thirteen scales have been used to assess clinical empathy in healthcare students from forty nine studies with total sample size 49384 students. The most frequently used scale is the Jefferson Scale of Physician Empathy followed by Davis’ Interpersonal Reactivity Index. The overall reliability was 0.805 (95%CI 0.786-0.823), which is acceptable, but there was heterogeneity and publication bias. Some heterogeneity was explained by the different countries of the studies under investigation and student types but most heterogeneity remained unexplained. CONCLUSION: The results indicate that scales have satisfactory internal consistency but there are a multitude of scales, definitions and empathy components. Future research should focus on standardizing scales and creating consensus statements regarding the definition of empathy and use of appropriate scales.
Article
This research aimed to (a) develop a measure of prosocial and antisocial behavior in sport, (b) examine its invariance across sex and sport, and (c) provide evidence for its discriminant and concurrent validity. We conducted two studies. In study 1, team sport athletes (N = 1,213) recruited from 103 teams completed questionnaires assessing demographics and prosocial and antisocial behaviors in sport. Factor analyses revealed two factors representing prosocial behavior and two factors representing antisocial behavior. The model had a very good fit to the data and showed configural, metric, and scalar invariance across sex and sport. The final scale consisted of 20 items. In Study 2, team-sport athletes (N = 106) completed the scale and measures of empathy and goal orientation. Analyses provided support for the discriminant and concurrent validity of the scale. In conclusion, the new scale can be used to measure prosocial and antisocial behaviors in team sport.
Article
Four meta-analyses were conducted to examine gender differences in personality in the literature (1958-1992) and in normative data for well-known personality inventories (1940-1992). Males were found to be more assertive and had slightly higher self-esteem than females. Females were higher than males in extraversion, anxiety, trust, and, especially, tender-mindedness (e.g., nurturance). There were no noteworthy sex differences in social anxiety, impulsiveness, activity, ideas (e.g., reflectiveness), locus of control, and orderliness. Gender differences in personality traits were generally constant across ages, years of data collection, educational levels, and nations.
Article
Four meta-analyses were conducted to examine gender differences in personality in the literature (1958-1992) and in normative data for well-known personality inventories (1940-1992). Males were found to be more assertive and had slightly higher self-esteem than females. Females were higher than males in extraversion, anxiety, trust, and, especially, tender-mindedness (e.g., nurturance). There were no noteworthy sex differences in social anxiety, impulsiveness, activity, ideas (e.g., reflectiveness), locus of control, and orderliness. Gender differences in personality traits were generally constant across ages, years of data collection, educational levels, and nations.
Chapter
This chapter focuses on autonomic and neuroendocrine processes that underlie social behaviors and emotional states, including those that are believed to reflect empathy in humans. Empathy has been considered a unique characteristic of human consciousness, but evidence suggests that emotional contagion and consolation exist in other mammalian species, including social primates such as bonobo chimpanzees. The chapter argues that empathy is a trait shared by humans with other mammals and linked to the neural circuits that emerged during the evolutionary transition from reptiles to mammals. It discusses empathy in relation to the evolution of social awareness in mammals, along with the neuroendocrine correlates of sociality, prosocial behaviors in highly social mammals, the role of neuropeptides in selective sociality, and possible mechanisms for sex differences in sociality or empathy.
Chapter
Psychopathy is a socially devastating personality disorder defined by a constellation of affective, interpersonal, and behavioral characteristics, including egocentricity, manipulativeness, deceitfulness, lack of empathy, guilt or remorse, and a propensity to violate social and legal expectations and norms (Cleckley, 1976; Hare, 1995, 1996). In this chapter I selectively review recent research on the role played by emotional processes in the disorder. Because some of the most illuminating insights into the emotional life of psychopaths are provided by close scrutiny of their psycholinguistic processes, I emphasize work that has implications for understanding the complex interplay of the psychopath’s language, affect, and predatory behavior.
Article
The present study investigated personality differences in violent, non-violent and sexual offenders incarcerated at a medium security federal penitentiary. The Temperament and Character Inventory was administered to 185 male inmates specifically to obtain, among other data, personality measures of impulsiveness, attachment, and empathy. Criminal records were reviewed and crime type was assigned according to offense history. Age at first offense was also examined. Violent offenders were found to be more impulsive and less empathic than nonviolent offenders. Sexual offenders were found to be less impulsive, more empathic, more attached, and to have a later age of onset than all other offenders. Identifying variables associated with different types of criminal behavior may have important implications for treatment.