A Meta-Analysis Exploring the Relationship Between Psychopathy and Instrumental Versus Reactive Violence

Abstract

The present meta-analysis explored the relationship between psychopathy and instrumental and reactive violence with a focus on factor and facet scores. A total of 53 studies (reporting on 55 unique samples, N = 8,753) from both published and unpublished sources were included. Results from random-effects analyses indicated moderate and significant relationships between psychopathy and both instrumental and reactive violence. There was some evidence that the Interpersonal facet was more important for instrumental violence, while Factor 2 (social deviance) was more important for reactive violence. The Lifestyle facet appeared important in explaining both violent outcomes. Effect sizes were significantly smaller for clinical rating scales compared with informant and self-report scales. Significant between-study variability was partly explained by mean age of the sample and type of outcome measure. The current findings do not support the conclusion that psychopathy is more related to instrumental violence as opposed to reactive violence.

Keywords

psychopathy instrumental violence reactive violence meta-analysis

Psychopathy is a personality disorder characterized by interpersonal, affective, behavioral, and antisocial features. It is estimated that 10% to 15% of the total offender population (Hare, 2003) and less than 1% of the general population (Coid, Yang, Ullrich, Roberts, & Hare, 2009) would meet the criteria for psychopathy. Although psychopaths represent a small subset of the overall offender population, their impact on the criminal justice system is exponential; it is estimated that a relatively small number of psychopathic offenders are responsible for a disproportionate amount of crime (Hare, 2003). Research has also demonstrated a relationship between psychopathy and violence in both offender (e.g., Asscher et al., 2011) and non-offender samples (e.g., Walters, 2003a, 2003b).

Violence can be further classified according to the instrumental versus reactive dimension. Reactive violence is thought to be marked by an impulsive emotional response to provocation, while instrumental violence is thought to be predominantly controlled, proactive, and goal oriented (Cornell et al., 1996). A great deal of research has demonstrated that psychopaths tend to engage in more instrumental violence as opposed to reactive violence (Cornell et al., 1996; Flight & Forth, 2007; Glenn & Raine, 2009; Vitacco, Neumann, & Caldwell, 2010; Vitacco, Neumann, Caldwell, Leistico, & Van Rybroek, 2006; Walsh, Swogger, & Kosson, 2009; Williamson, Hare, & Wong, 1987; Woodworth & Porter, 2002). Despite the prevalence of this finding, no study to date has systematically examined whether the relationship between psychopathy and instrumental violence is consistent across studies and across psychopathy facets or whether this relationship is dependent on sample characteristics such as age, ethnicity, or gender.

Measuring Psychopathy

The measurement scales created to assess psychopathy can be divided into three broad categories: clinical rating scales, self-report scales, and informant scales. The Psychopathy Checklist–Revised (PCL-R; Hare, 2003) and its derivations (i.e., PCL: Youth Version [YV]; Forth, Kosson, & Hare, 2003, and PCL: Screening Version [SV]; Hart, Cox, & Hare, 1995) are classified as clinical rating scales. These scales require raters to integrate information from interviews, collateral sources, and files to make inferences about traits (Hare, 2003). Using factor analysis, two factors have been identified within the PCL-R (Hare, 1991, 2003; Harpur, Hakstian, & Hare, 1988). Factor 1 represents the interpersonal (e.g., manipulative) and affective (e.g., callousness) components of psychopathy, while Factor 2 describes the antisocial lifestyle (e.g., early antisocial behavior) and behavioral (e.g., need for stimulation) components of psychopathy. More recently, these two factors have been further divided into four facets: Interpersonal, Affective, Lifestyle, and Antisocial (Hare, 2003). Subsequent research has found support for these four facets as a means of more thoroughly capturing the nature of psychopathy in a variety of forensic, clinical, and community populations (e.g., Babiak, Neumann, & Hare, 2010; Kosson et al., 2013; Neumann, Kosson, Forth, & Hare, 2006; Neumann, Kosson, & Salekin, 2007).

Additional research has been conducted in an attempt to identify possible alternate factor structures for psychopathy. Cooke and Michie (2001) have studied the validity of a three-factor model: arrogant and deceitful interpersonal style, deficient affective experience, and impulsive and irresponsible behavioral style. Weaver, Meyer, Van Nort, and Tristan (2006) conducted an examination of the original two-factor model (Harpur et al., 1988) as well as the three-factor model proposed by Cooke and Michie (2001), and the four-facet model proposed by Hare (2003). Confirmatory factor analysis revealed that the best fit to the data (sample of 1,566 male sex offenders) was represented by the three-factor model; the four-facet model was also found to be a better fit than the original two-factor model (Weaver et al., 2006).

Despite their frequent use, there are limitations associated with clinical rating scales. For example, scoring clinical rating scales is time-consuming given that interviews and reviews of file information are both necessary. Also, with community samples, access to any collateral or file information is often difficult to obtain. Alternatively, self-report scales represent a more cost-effective method of assessing psychopathy and can be administered to both criminal and non-criminal populations (Lilienfeld, 1994). In addition, self-report scales can be completed without fear of observer bias or subjectivity in scoring. The main limitation of self-report scales, especially when assessing psychopathy, is the possibility of dishonest responding (Lilienfeld & Fowler, 2006). Despite this limitation, a recent meta-analysis found that most self-report psychopathy scales are actually negatively correlated with social desirability responding (Ray et al., 2013). Although psychopaths may be willing to admit to negative traits in the context of a research study, they may be much more likely to engage in impression management when there are meaningful consequences such as during a forensic clinical assessment.

Various self-report scales have been developed to assess psychopathy. Importantly, these scales tend to correlate significantly with total scores derived from clinical rating scales and also represent similar underlying factor and facet structures. For example, the factor structure of the Self-Report Psychopathy Scale (SRP; Paulhus, Neumann, & Hare, in press) is best represented by the four facets of the PCL-R (Neal & Sellbom, 2012). In addition, a study by Brinkley, Schmitt, Smith, and Newman (2001) found that the primary and secondary factors of the Levenson Self-Report Psychopathy Scale (LSRP; Levenson, Kiehl, & Fitzpatrick, 1995) moderately correlate with Factors 1 and 2 of the PCL-R. Recent research has also suggested that a three-factor solution may be appropriate for the LSRP (Sellbom, 2011).

Another common self-report scale for youth, the Youth Psychopathic Traits Inventory (YPI; Andershed, Kerr, Stattin, & Levander, 2002), is also representative of the three-factor model of psychopathy. More specifically, the three factors of the YPI (Affective, Interpersonal, and Lifestyle) correlate with the Interpersonal, Affective, and Lifestyle factors of psychopathy (Cooke & Michie, 2001). Finally, the two underlying factors found within the Psychopathic Personality Inventory (PPI; Lilienfeld & Andrews, 1996) have been shown to significantly correlate with both factors of the PCL-R (Benning, Patrick, Hicks, Blonigen, & Krueger, 2003; Berardino, Meloy, Sherman, & Jacobs, 2005; Poythress, Edens, & Lilienfeld, 1998). Evidence suggests that although self-report scales differ in administrative procedure, they are able to capture the underlying characteristics of psychopathy.

Informant scales have been created to assess psychopathy among children and adolescents. They provide researchers with the ability to accurately assess a child through regular interaction and observation (Lynam, 1997). For example, the Antisocial Process Screening Device (APSD; Frick & Hare, 2001) and the Childhood Psychopathy Scale (CPS; Lynam, 1997) can be completed by a parent, guardian, or teacher. Like scores from self-report scales, scores from informant scales have shown acceptable convergent validity with other measures of psychopathy (Falkenbach, Poythress, & Heide, 2003; Frick & Hare, 2001). In addition, the factor structure of informant scales also relates to the factor structure of clinical rating scales. For example, confirmatory factor analysis has demonstrated that the APSD has three underlying factors (i.e., callous-unemotional, narcissism, and impulsivity), which are closely related to the three-factor model of psychopathy (Frick, Bodin, & Barry, 2000; Frick & Hare, 2001).

Psychopathy and Violence

Psychopathy is one of the most researched risk factors for violence (Guy, Douglas, & Hendry, 2010; Harris, Rice, & Cormier, 1991; Salekin, Rogers, & Sewell, 1996; Weiler & Widom, 1996). Several meta-analyses examining this relationship have been conducted using adult samples (Guy, Edens, Anthony, & Douglas, 2005; Kennealy, Skeem, Walters, & Camp, 2010; Salekin et al., 1996), youth samples (Asscher et al., 2011; Edens, Campbell, & Weir, 2006; Olver, Stockdale, & Wormith, 2009), or a combination of both (Walters, 2003a, 2003b, 2006). Although the majority of the samples represent forensic populations (offender or psychiatric), two of the meta-analyses also included high school students (Walters, 2003a, 2003b). The most common outcome measures included general and violent recidivism, delinquency, and institutional misconduct. Overall, previous meta-analyses have identified a moderate relationship between psychopathy and violence.

It is important to note that considerable variability between effect sizes is often cited and several significant moderators have been identified. For example, the relationship between psychopathy and violence is larger for youth compared with adult samples (Walters, 2006) and smaller for samples with higher ethnic heterogeneity (Edens et al., 2006). Gender has not consistently been found as a significant moderator, with Olver and colleagues (2009) reporting no significance and Edens and colleagues (2006) reporting smaller effect sizes for samples of female juvenile offenders. When examining sample origin, studies from the United States and Canada tend to have significantly larger effect sizes (Kennealy et al., 2010; Olver et al., 2009); however, this is not always consistent (Edens et al., 2006). Finally, several meta-analyses have reported that the relationship between psychopathy and risk of violence is significantly larger for Factor 2 compared with Factor 1 (e.g., Kennealy et al., 2010; Walters, 2003a, 2003b).

Reactive and Instrumental Violence

In psychological research, a distinction is often made between instrumental or predatory violence and reactive or hostile violence. Instrumental violence is defined by aggression that is controlled, proactive in nature, and is oriented by an external goal. Therefore, the consequence of violence is secondary to a goal such as monetary gain or elevated social status (Berkowitz, 1993). Emotions are thought to play a minimal role in instrumental violence (Glenn & Raine, 2009; Meloy, 2006). Conversely, reactive violence refers to aggression emanating from an extreme emotional response (usually fear or anger) to a real or perceived threat or provocation. In addition, this violence is committed without a foreseeable external goal by the offender (Berkowitz, 1993; Cornell et al., 1996). The operationalization of instrumental and reactive violence is not without controversy. Although the above definitions suggest that violent acts can be classified into one category, in reality, violence is more likely the result of a complex interaction of emotions and motives (Bushman & Anderson, 2001). Despite the simplicity of the dichotomous model, neurological and biological evidence for its validity is evident (see Meloy, 2006, for a review), and the distinction is still widely used in research settings.

Research has consistently found that although psychopathic offenders may commit both types of violence, psychopathy is more related to instrumental violence in both youth (Bezdjian, Tuvblad, Raine, & Baker, 2011; Fite, Raine, Stouthamer-Loeber, Loeber, & Pardini, 2010; Fite, Stoppelbein, & Greening, 2009; Flight & Forth, 2007; Murrie, Cornell, Kaplan, McConville, & Levy-Elkon, 2004) and adult offenders (Chase, O’Leary, & Heyman, 2001; Cima & Raine, 2009; Cornell et al., 1996; Porter & Woodworth, 2007). The relationship between psychopathy and instrumental violence has also been replicated in samples of female offenders (Lehmann & Ittel, 2012; Marsee & Frick, 2007). A narrative review of this literature has suggested that psychopathy may in fact serve as a potential protective factor against reactive violence (Reidy, Shelley-Tremblay, & Lilienfeld, 2011).

Although meta-analyses have reported Factor 2 to be more predictive of general violence than Factor 1, the opposite finding is reported when examining the type of violence; Factor 1 scores appear to be more predictive of instrumental violence compared with Factor 2 scores (Bezdjian et al., 2011; Cornell et al., 1996; Declercq, Willemsen, Audenaert, & Verhaeghe, 2012; Hodges & Heilbrun, 2009). For example, a study conducted by Flight and Forth (2007) with young offenders found that the potential for instrumental violence increased with higher scores on the interpersonal/affective (IA) factor as opposed to the social deviance (SD) factor. When examining facet scores, inconsistent results have been reported with some studies showing that the Interpersonal facet is more strongly related to instrumental violence (e.g., Declercq et al., 2012; Walsh et al., 2009), while others indicating that the Antisocial facet is more strongly related to instrumental violence (e.g., Hodges & Heilbrun, 2009).

The purpose of the present study was to systematically examine the relationship between psychopathy and instrumental and reactive violence using meta-analysis. To further understand these relationships, we calculated meta-analytic effects for total, factor, and facet psychopathy scores and for the three main types of measurement scales. In addition, moderator analyses were conducted to determine whether the association between psychopathy and instrumental and reactive violence was influenced by certain participant (e.g., age, gender, and ethnicity) and study characteristics (e.g., sample type and outcome measure).

Method

Study Selection

Both published and unpublished sources available before January 1, 2013, were included in the present meta-analysis. The following databases were searched: PsycINFO, National Criminal Justice Reference System, Web of Science, Criminal Justice Abstracts, Dissertations and Theses: Fulltext, and Dissertations and Theses: UK and Ireland. In addition, forensic journals not included in PsycINFO were searched individually, and the reference lists for accepted articles were also searched for any missing studies. The search procedure involved combining psychopathy-related terms (e.g., psychopath*, PCL*, and Psychopathy Checklist-Revised) with synonyms of violence with specific reference to instrumental or reactive violence (e.g., instrument*, predator*, proactive*, premeditate*, planned, impulsive*, reactive*, violen*, and aggress*).

To be included, every study must have adhered to both of the following inclusion criteria. First, the study must have explored the relationship between psychopathy (as measured by any valid and reliable scale) and a measure of either instrumental or reactive violence (studies reporting on criminal thoughts as opposed to action were excluded). Studies had to specifically state the type of violence (either instrumental or reactive) under study. Studies examining impulsivity and psychopathy without specifying impulsive violence were excluded. Studies did not have to examine both types of violence outcomes to be included, but if a study involved a comparison between instrumental and reactive violence, both effect sizes were coded.

Second, each study must have provided enough information for at least one correlation coefficient between the variables of interest to be extracted or converted from the existing information. Additional information was requested from the authors when needed. Information on all studies included in the present meta-analysis can be found in Table 1.¹ If multiple studies reported on the same sample (or a subsample of the total sample), the study with the largest sample size was chosen as the primary study. Only non-overlapping effect sizes were then coded from the other studies.

Table 1:

List of Studies Included in the Meta-Analysis

Study No.	Authors	Year	Peer Reviewed	Country of Origin
1	Barr	2008	No	USA
2, 2.01	Bezdjian, Tuvblad, Raine, and Baker	2011	Yes	USA
3, 3.01	Bobadilla, Wampler, and Taylor	2012	Yes	USA
4	Chase, O’Leary, and Heyman	2001	Yes	USA
5	Cima and Raine	2009	Yes	The Netherlands
6	Cornell, Warren, Hawk, Stafford, Oram, and Pine	1996	Yes	USA
7	Declercq, Willemsen, Audenaert, and Verhaeghe	2012	Yes	Belgium
8	Fite, Raine, Stouthamer-Loeber, Loeber, and Pardini	2010	Yes	USA
9	Fite, Stoppelbein, and Greening	2009	Yes	USA
10	Flight and Forth	2007	Yes	Canada
11	Hodges and Heilbrun	2009	Yes	USA
12, 12.01	Kerig and Stellwagen	2010	Yes	USA
13	Laurell, Belfrage, and Hellström	2010	Yes	Sweden
14	Lehmann and Ittel	2012	Yes	Germany
15	Marsee and Frick	2007	Yes	USA
16	McDermott, Quanbeck, Busse, Yastro, and Scott	2008	Yes	USA
17	Murrie, Cornell, Kaplan, McConville, and Levy-Elkon	2004	Yes	USA
18	Nouvion, Cherek, Lane, Tcheremissine, and Lieving	2007	Yes	USA
19	Porter and Woodworth	2007	Yes	Canada
20	Ragatz, Anderson, Fremouw, and Schwartz	2011	Yes	USA
21	Raine et al.	2006	Yes	USA
22	Reidy, Zeichner, Miller, and Martinez	2007	Yes	USA
23	Seals, Sharp, Ha, and Michonski	2012	Yes	USA
24	Stafford and Cornell	2003	Yes	USA
25	Stanford, Houston, and Baldridge	2008	Yes	USA
26	Swogger, Walsh, Houston, Cashman-Brown, and Conner	2010	Yes	USA
27	Taubner, White, Zimmerman, Fonagy, and Nolte	2012	No	Germany
28	Vitacco, Neumann, Caldwell, Leistico, and Van Rybroek	2006	Yes	USA
28.1^a	Leistico and Salekin	2003	Yes	USA
29	Vitacco, Van Rybroek, Rogstad, Yahr, Tomony, and Saewert	2009	Yes	USA
29.1^a	Vitacco, Gonsalves, Tomony, Smith, and Lishner	2012	Yes	USA
30	Walsh, Swogger, and Kosson	2009	Yes	USA
31, 31.01	Yeh, Chen, Raine, Baker, and Jacobson	2011	Yes	USA
32	Verschuere, Candel, Van Reenen, and Korebrits	2012	Yes	The Netherlands
33	Ostrov and Houston	2008	Yes	USA
34	van Baardewijk, Vermeiren, Stegge, and Doreleijers	2011	Yes	The Netherlands
35	Bjørnebekk	2007	Yes	Norway
36	Barry, Thompson, Barry, Lochman, Adler, and Hill	2007	Yes	USA
37	Steadham	2012	No	USA
38	Phillips	2003	No	USA
39	Preston	1998	No	USA
40	Barker	2003	No	USA
41	Crawley and Martin	2006	Yes	New Zealand
42	Davis	2010	No	USA
43	Fanti, Frick, and Georgiou	2009	Yes	Cyprus
44	Jones and Paulhus	2010	Yes	Canada
45	Muñoz, Frick, Kimonis, and Aucoin	2008	Yes	USA
46	Reidy, Zeichner, and Martinez	2008	Yes	USA
47	Reidy, Zeichner, and Seibert	2011	Yes	USA
48	Serin	1991	Yes	Canada
49	Williamson, Hare, and Wong (demographics from Wong, 1984)	1987	Yes	Canada
50	Wirth	2010	No	USA
51	Woodworth and Porter	2002	Yes	Canada

Note. Additional information was obtained from both S. Bezdjian and L. Bobadilla. Studies with two sample numbers represent independent samples within each study (i.e., males and females analyzed separately).

Same sample as previous. Only non-overlapping effect sizes were coded.

Measures of Psychopathy

Measures of psychopathy were divided into three broad categories: clinical rating scales, informant measures, and self-report scales. Clinical rating scales included the PCL-R and its derivative scales (PCL: YV and PCL: SV). Informant measures included the APSD (Frick & Hare, 2001) and CPS (Lynam, 1997). Several self-report scales were identified including the self-report scores from both the APSD and CPS. To ensure independence of effect sizes, only one effect size for each type of scale was coded for each study. If a study included multiple effect sizes for the same type of scale, the average of the two effect sizes was taken. For example, Study 17 included both the APSD self-report and the Millon Adolescent Clinical Inventory (MACI; Millon, 1993).

Although the overall relationship between psychopathy and instrumental or reactive violence was of interest, we also coded effect sizes for any study that included information on the relationship between psychopathy factor and facet scores and both types of violence (i.e., Factor 1, Factor 2, Interpersonal, Affective, Lifestyle, and Antisocial; a study did not need to include factor or facet scores to be included). For informant and self-report scales, factors and facets closely related to the original factors and facets derived from the PCL-R were coded. Table 2 provides information on all the scales included in the study as well as how each scale component was averaged to assess effect sizes for factor and facet scores.

Table 2:

Individual Scales Contributing to Psychopathy Total, Factor, and Facet Scores Divided by Type of Scale

Scale Type	Factors and Facets	Study
Clinical
Total scores	PCL-R; PCL: SV; PCL: YV; P-SCAN	1, 6, 7, 10, 11, 13, 14, 16, 17, 18, 19, 24, 26, 28.1, 29, 30, 37, 41, 42, 48, 49, 51
Factor 1	PCL-R Factor 1; PCL: SV Factor 1; PCL: YV Factor 1	1, 6, 7, 10, 11, 13, 26, 28, 29.1, 51
Factor 2	PCL-R Factor 2; PCL: SV Factor 2; PCL: YV Factor 2	1, 6, 7, 10, 11, 13, 26, 28, 29.1, 51
Facet 1 (Interpersonal)	PCL-R Facet 1; PCL: SV Facet 1; PCL: YV Facet 1	7, 10, 11, 14, 16, 28, 29, 30, 37, 42
Facet 2 (Affective)	PCL-R Facet 2; PCL: SV Facet 2; PCL: YV Facet 2	7, 10, 11, 14, 16, 28, 29, 30, 37, 42
Facet 3 (Lifestyle)	PCL-R Facet 3; PCL: SV Facet 3; PCL: YV Facet 3	7, 10, 11, 14, 16, 28, 29, 30, 37, 42
Facet 4 (Antisocial)	PCL-R Facet 4; PCL: SV Facet 4; PCL: YV Facet 4	7, 10, 11, 14, 16, 28, 29, 30, 37, 42
Informant
Total scores	APSD; CPS	2, 2.01, 17, 21, 31, 31.01
Factor 1	—	—
Factor 2	—	—
Facet 1 (Interpersonal)	APSD Machiavellian; APSD Narcissism	9, 12, 12.01, 36
Facet 2 (Affective)	APSD Callous-unemotional traits	9, 12, 12.01, 36, 38
Facet 3 (Lifestyle)	APSD Impulsive	9, 12, 12.01, 36
Facet 4 (Antisocial)	—	—
Self-report
Total scores	APSD Self-Report; CPS-R; LSRP; MACI; MCMI-II; mCPS; NEO; PPI; SRP; YPI	1, 2, 2.01, 3, 3.01, 4, 5, 8, 17, 22, 23, 25, 27, 32, 34, 39, 40, 44, 46, 47, 50
Factor 1	LSRP Primary Psychopathy; mCPS Factor 1; PPI Factor 1	3, 3.01, 5, 20, 22, 25, 32, 33, 35, 39, 46
Factor 2	LSRP Secondary Psychopathy; mCPS Factor 2; PPI Factor 2	3, 3.01, 5, 20, 22, 25, 32, 33, 35, 39, 46
Facet 1 (Interpersonal)	YPI Interpersonal	23, 34
Facet 2 (Affective)	ICU; YPI Affective	15, 23, 34, 43, 45
Facet 3 (Lifestyle)	YPI Lifestyle	23, 34
Facet 4 (Antisocial)	—	—

Note. PCL-R = Psychopathy Checklist–Revised; PCL: SV = Psychopathy Checklist: Screening Version; PCL: YV = Psychopathy Checklist: Youth Version; APSD = Antisocial Process Screening Device; CPS = Childhood Psychopathy Scale; CPS-R = Chidhood Psychopathy Scale-Revised; LSRP = Levenson Self-Report Psychopathy Scale; MACI = Millon Adolescent Clinical Inventory; MCMI-II = Millon Clinical Multiaxial Inventory-II; mCPS = Modified Child Psychopathy Scale; PPI = Psychopathic Personality Inventory; SRP = Self-Report Psychopathy Scale; YPI = Youth Psychopathic Traits Inventory; ICU = Inventory of Callous and Unemotional Traits.

Coding Procedures

Each eligible study was coded for study demographics (e.g., country of origin, year of publication), sample statistics (e.g., sample type, mean age), and effect sizes for the variables of interest. The first and second authors coded all studies independently. Interrater statistics were calculated for 6 studies (11%). The kappa statistic was used when assessing the reliability of categorical variables, and a two-way random-effects model intraclass correlation coefficient (ICC; absolute agreement) was used when assessing the reliability of ordinal or continuous variables. When kappa could not be calculated, a percent agreement between raters was calculated. Acceptable values were found for study and sample variables and for effect size calculations. The kappa statistics ranged from .60 to 1.00 (80% of kappa values equaled 1.00; percent agreement ranged from 71.4 to 85.7) and ICC values ranged from .82 to 1.00. Given that both the first and second authors coded all studies, a consensus coding was obtained whenever a disagreement occurred.

Effect Size Calculation

Correlation coefficients (r) were used as effect size estimates for the relationship between psychopathy and instrumental or reactive violence. As described by Borenstein, Hedges, Higgins, and Rothstein (2009), the correlation coefficient for each effect was converted to the Fisher’s z scale (z_r) and aggregated using the variance of Fisher’s z (all z_r values weighted by the inverse of the variance). The aggregated effect sizes were then converted to the original correlation coefficient for ease of interpretation. In interpreting correlation coefficients, values below .30 can be considered “small,” values between .30 and .50 can be considered “moderate,” and any values above .50 can be considered “large” (Cohen, 1988). If the 95% confidence interval (CI) does not include 0, the effect size is significant at the .05 level.

Both fixed-effect and random-effects models were calculated. The fixed-effect model is based on the assumption that there is one true effect expressed by all of the studies used in the meta-analysis; any variance between samples is attributed to sampling error. Conclusions are therefore restricted to the particular sample of studies included in the meta-analysis (Borenstein et al., 2009). The random-effects model assumes that the observed effect sizes represent a random sample of all possible estimates of the true population effect. Under this model, observed effects vary as a function of study methodology (e.g., differences in samples), and this between-study variability is subsequently incorporated into the error term. Results of fixed-effect and random-effects analyses converge as between-study variability decreases (Borenstein et al., 2009). Although both fixed-effect and random-effects estimates are presented, only random-effects estimates will be discussed.

Between-study variability was assessed using both the Q and I² statistics. While the Q statistic provides a measure of the significance of between-study variability, the I² statistic provides an indication of the magnitude of this variability. The Q statistic is distributed as a chi-square with k − 1 degrees of freedom (k being the number of studies; Hedges & Olkin, 1985). I² is presented as a percentage with 25, 50, and 75 indicating small, medium, and large proportions of variability, respectively (Huedo-Medina, Sánchez-Meca, Marín-Martínez, & Botella, 2006). When Q was significant, the presence of outliers was considered by examining both the size of the individual effect and the weight it contributed to the overall estimate. A study was considered an outlier if the effect was the largest or smallest of all the effects included in the aggregated estimate and if the effect contributed a large weight to the estimate. In addition, a study was only considered an outlier if, after removing the study, the overall Q value was reduced by half.

In the presence of significant between-study variability, one possible explanation is that other variables may moderate the relationship between the variables. Although significant moderators can be found in the absence of variability across studies (i.e., non-significant Q), these analyses have lower power. For this reason, moderators were only explored on effect sizes that demonstrated a significant Q value or if the magnitude of the variability (as measured by I²) exceeded .50. It should be noted that moderators are only one possible explanation for why effect sizes may vary significantly across the sample of studies. Moderators included study demographics (e.g., country of origin, outcome measure) and sample characteristics (e.g., age, sample type). For categorical moderators, between-level Q was calculated. This entails calculating the overall Q statistic and subtracting the Q value produced by each individual level of the moderator variable. If the between-level Q is significant, the moderator accounts for a significant portion of the overall variability. Each level of the moderator variable required at least three effect sizes to be analyzed. For continuous moderators (e.g., age, ethnicity), a random-effects meta-regression analysis (unrestricted maximum likelihood [ML] method) was conducted to assess the relationship between the effect size and the moderator.

We were also interested in comparing the effect sizes across factors, facets, and scale type (e.g., differences between facets for each type of scale); however, these variables violate the assumption of independence necessary to conduct proper moderator analyses. This means that the same individuals cannot be present in more than one category. For this reason, the 84% CIs were calculated based on the random-effects results to compare significance between effect sizes that violated the assumption of independence as suggested by Tryon (2001). When comparing across different effect sizes (e.g., Factor 1 vs. Factor 2), non-overlapping 84% CIs represent a significant difference at the .05 level, whereas non-overlapping 95% CIs represent a significant difference at the .01 level (Tryon, 2001). All analyses were conducted using comprehensive meta-analysis (CMA; Borenstein, Hedges, Higgins, & Rothstein, 2005).

Results

In total, 53 studies (86% peer reviewed) reporting on 55 unique samples were coded (N = 8,753). The median year of publication was 2009 (range = 1987-2012) and the majority of studies originated from the United States (71%). When examining the samples, the average sample size (using the largest sample associated with an effect size) was 159 (range = 33-679). The majority of the samples were adult only (56.4%; 38.2% juvenile only; 5.5% contained both). The mean age of the samples was 23.0 years (range = 9.3-46.6 years) and the majority were also male only (53.7%; 14.8% female only; 31.5% both; 1.8% not reported). Approximately 40% of the samples were from the general offender population and another 46.4% were from the general community population (the remaining samples were either psychiatric or a mixed sample). Finally, 39 of the 55 samples (71%) reported on ethnicity. On average, minorities made up 40.7% of the total ethnicity across the samples. When examining how the violent outcome was operationalized, 47% of samples used a self-report measure of instrumental or reactive violence, while the remaining 53% used a more objective method of measurement (e.g., independent raters, informant reports).

Psychopathy and Violence

Overall Psychopathy

The relationship between psychopathy overall and instrumental violence can be seen in Table 3. Psychopathy (across all scale types) showed a moderate, significant correlation with instrumental violence indicating that as psychopathy scores increased, ratings of instrumental violence increased (r = .36, 95% CI = [.30, .41]). This pattern of results was consistent for the total, factor, and facet scores. When examining the 84% CIs (necessary when comparing effect sizes that violate the assumption of independence), the relationship between psychopathy and instrumental violence was significantly larger for the Interpersonal facet (r = .42, 84% CI = [.39, .45]), followed by the Lifestyle facet (r = .38, 84% CI = [.35, .41]) and the Affective facet (r = .35, 84% CI = [.32, .38]). The Antisocial facet (r = .29, 84% CI = [.24, .34]) was the least related to instrumental violence. It is important to note that most of the effect sizes showed significant and considerable variability.

Table 3:

Relationship Between Psychopathy and Instrumental Violence

	Fixed		Random
Risk Factor	Mean r	95% CI	Mean r	95% CI	84% CI	Q	I²	k	n
Psychopathy (overall)	.36	[.34, .38]	.36	[.30, .41]	[.35, .37]	392.40***	86.75	53	8542
Total scores	.36	[.34, .38]	.33	[.27, .39]	[.31, .35]	237.68***	82.75	42	6521
Factor 1	.23	[.19, .27]	.30	[.20, .39]	[.27, .33]	103.72***	80.72	21	2291
Factor 2	.24	[.20, .28]	.29	[.19, .39]	[.26, .32]	107.05***	82.25	20	2217
Interpersonal	.44	[.40, .48]	.42	[.28, .54]	[.39, .45]	148.58***	89.90	16	1603
Affective	.38	[.34, .41]	.35	[.25, .44]	[.32, .38]	102.50***	81.46	20	2141
Lifestyle	.41	[.37, .45]	.38	[.25, .50]	[.35, .41]	116.96***	87.18	16	1603
Antisocial	.28	[.21, .34]	.29	[.18, .39]	[.24, .34]	18.70*	51.88	10	775

Note. CI = confidence interval.

p < .05. ***p < .001.

Table 4 presents the same findings for reactive violence. Similar to the findings for instrumental violence, the relationship between psychopathy and reactive violence was moderate and significant (r = .35, 95% CI = [.29, .41]). Unlike the results for instrumental violence, Factor 2 was more related to reactive violence (r = .38, 84% CI = [.35, .41]) compared with Factor 1 (r = .30, 84% CI = [.27, .33]). Also, the relationship between psychopathy and reactive violence was stronger for the Lifestyle facet (r = .44, 84% CI = [.41, .47]) compared with the other psychopathy facets. Considerable between-study variability was evident for all effect sizes with the exception of the Affective facet.

Table 4:

Relationship Between Psychopathy and Reactive Violence

	Fixed		Random
Risk Factor	Mean r	95% CI	Mean r	95% CI	84% CI	Q	I²	k	n
Psychopathy (overall)	.35	[.33, .37]	.35	[.29, .41]	[.34, .36]	283.14***	86.23	40	7,394
Total scores	.33	[.31, .36]	.33	[.26, .41]	[.31, .35]	205.00***	87.80	26	5,130
Factor 1	.34	[.29, .38]	.30	[.19, .40]	[.27, .33]	43.56***	74.75	12	1,577
Factor 2	.40	[.36, .44]	.38	[.26, .48]	[.35, .41]	62.29***	80.73	13	1,651
Interpersonal	.42	[.37, .46]	.36	[.19, .52]	[.33, .39	130.78***	90.83	13	1,262
Affective	.31	[.26, .35]	.30	[.24, .36]	[.27, .33]	24.31	34.18	17	1,800
Lifestyle	.49	[.45, .53]	.44	[.32, .56]	[.41, .47]	78.11***	84.64	13	1,262
Antisocial	.30	[.21, .39]	.31	[.12, .47]	[.25, .37]	22.17**	72.94	7	434

Note. CI = confidence interval.

p < .01. ***p < .001.

Type of Scale

To reduce some of the between-study variability, analyses were divided by type of scale. Table 5 demonstrates these results for instrumental violence. All indices of psychopathy remained significantly related to instrumental violence; however, the relationship was significantly smaller for clinical rating scales (r = .29, 84% CI = [.26, .32]) compared with informant and self-report scales (r = .50, 84% CI = [.48, .52], and r = .37, 84% CI = [.35, .39], respectively). For clinical rating scales, the relationship between psychopathy and instrumental violence was consistently small across total, factor, and facet scores. When examining informant measures, Studies 31 and 36 were identified as outliers for certain variables and were subsequently removed (these studies were also removed from all moderator analyses). Unlike clinical rating scales, the Interpersonal (r = .62, 84% CI = [.57, .66]) and Lifestyle facets (r = .55, 84% CI = [.50, .60]) were significantly more related to instrumental violence compared with the Affective facet (r = .44, 84% CI = [.39, .49]) for informant scales.

Table 5:

Relationship Between Psychopathy and Instrumental Violence Divided by Type of Scale

	Fixed		Random
Type of Scale	Mean r	95% CI	Mean r	95% CI	84% CI	Q	I²	k	n
Clinical	.27	[.23, .32]	.29	[.23, .34]	[.26, .32]	28.00	28.58	21	1,504
Total scores	.28	[.23, .32]	.30	[.22, .37]	[.27, .33]	40.49**	53.07	20	1,451
Factor 1	.31	[.24, .38]	.32	[.22, .40]	[.27, .37]	13.97	35.58	10	665
Factor 2	.23	[.15, .30]	.24	[.14, .34]	[.19, .29]	15.73	42.80	10	665
Interpersonal	.24	[.17, .31]	.26	[.15, .37]	[.21, .31]	20.29*	55.65	10	775
Affective	.18	[.10, .24]	.22	[.09, .34]	[.17, .27]	24.00**	62.50	10	775
Lifestyle	.19	[.12, .26]	.22	[.12, .31]	[.17, .27]	15.22	40.85	10	775
Antisocial	.28	[.21, .34]	.29	[.18, .39]	[.24, .34]	18.70*	51.88	10	775
Informant	.41	[.38, .44]	.44	[.31, .55]	[.42, .46]	176.54***	94.34	11	3,321
Minus 31, 31.01	.52	[.48, .55]	.50	[.39, .59]	[.48, .52]	62.81***	87.26	9	2,163
Total scores	.37	[.34, .40]	.33	[.15, .49]	[.31, .35]	124.61***	95.99	6	2,756
Minus 31, 31.01	.49	[.45, .53]	.42	[.24, .57]	[.39, .45]	40.84***	92.65	4	1,598
Interpersonal	.69	[.64, .73]	.68	[.54, .78]	[.65, .71]	19.20***	84.38	4	517
Minus 36	.62	[.55, .68]	.62	[.55, .68]	[.57, .66]	1.05	0.00	3	357
Affective	.45	[.38, .51]	.44	[.34, .54]	[.39, .49]	8.29	51.77	5	565
Lifestyle	.61	[.56, .66]	.60	[.47, .70]	[.56, .64]	12.62**	76.23	4	517
Minus 36	.56	[.48, .62]	.55	[.44, .65]	[.50, .60]	4.05	50.63	3	357
Self-report	.40	[.37, .42]	.37	[.29, .44]	[.35, .39	224.09***	88.84	26	5,075
Total scores	.44	[.41, .46]	.38	[.30, .44]	[.36, .40]	98.05***	80.62	20	3,616
Factor 1	.20	[.15, .25]	.28	[.13, .42]	[.25, .31]	83.09***	87.96	11	1,626
Minus 33	.33	[.28, .39]	.31	[.18, .43]	[.27, .35]	38.94***	76.89	10	947
Factor 2	.24	[.20, .29]	.33	[.16, .48]	[.30, .36]	91.19***	90.13	10	1,552
Affective	.50	[.44, .55]	.46	[.34, .56]	[.42, .50]	12.94*	69.10	5	801
Minus 43	.43	[.35, .51]	.42	[.30, .53]	[.36, .47]	6.39	53.08	4	454

Note. CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Two studies were identified as outliers for self-report scales and were removed (i.e., Studies 33 and 43). Subsequently, there was no statistical difference between Factor 1 (r = .31, 84% CI = [.27, .35]) and Factor 2 (r = .33, 84% CI = [.30, 36]) for the relationship between self-reported psychopathy and instrumental violence. Only the Affective facet could be examined and showed a moderate relationship with instrumental violence (r = .42, 84% CI = [.36, .47]). Although between-study variability was reduced by examining the effect sizes by scale type and by removing outlying studies, significant variability was still evident overall for informant and self-report scales.

Table 6 presents the effect sizes for reactive violence divided by type of scale. Consistent with the results for instrumental violence, all effect sizes were significant; however, the relationship between psychopathy and reactive violence was smaller for clinical rating scales (r = .24, 84% CI = [.19, .29], after removing Study 41), followed by self-report scales (r = .34, 84% CI = [.32, .36]), and largest for informant scales (r = .43, 84% CI [.41, .45]). For clinical rating scales, the relationship between psychopathy and reactive violence was small and significant across total, factor, and facet scores, which was congruent with the results for instrumental violence. For informant scales, after removing outliers, the Interpersonal (r = .59, 84% CI [.54, .64]) and Lifestyle facets (r = .59, 84% CI = [.54, .64]) were significantly more related to reactive violence compared with the Affective facet (r = .30, 84% CI = [.23, .36]). Finally, for self-report scales, the relationship between psychopathy and reactive violence was significantly larger for Factor 2 (r = .42, 84% CI = [.39, .45]) compared with Factor 1 (r = .34, 84% CI = [.31, .37]). Significant between-study variability remained for informant and self-report scales.

Table 6:

Relationship Between Psychopathy and Reactive Violence Divided by Type of Scale

	Fixed		Random
Type of Scale	Mean r	95% CI	Mean r	95% CI	84% CI	Q	I²	k	n
Clinical	.28	[.21, .35]	.30	[.17, .42]	[.25, .35]	31.51***	68.27	11	695
Minus 41	.24	[.16, .31]	.24	[.16, .31]	[.19, .29]	8.87	0.00	10	657
Total scores	.33	[.25, .40]	.36	[.19, .50]	[.31, .41]	30.99***	74.19	9	529
Minus 41	.28	[.19, .36]	.28	[.17, .39]	[.22, .34]	10.94	36.02	8	491
Factor 1	.18	[.06, .30]	.19	[.04, .33]	[.10, .27]	4.36	31.18	4	249
Factor 2	.27	[.15, .38]	.27	[.01, .49]	[.19, .35]	12.06**	75.11	4	249
Interpersonal	.14	[.05, .24]	.14	[.05, .24]	[.07, .21]	3.91	0.00	7	434
Affective	.20	[.11, .29]	.20	[.11, .29]	[.13, .26]	4.20	0.00	7	434
Lifestyle	.29	[.20, .37]	.29	[.20, .37]	[.23, .35]	5.76	0.00	7	434
Antisocial	.30	[.21, .39]	.31	[.12, .47]	[.25, .37]	22.17**	72.94	7	434
Informant	.40	[.37, .43]	.43	[.28, .56]	[.41, .45]	208.57***	95.69	10	3,283
Total scores	.36	[.33, .39]	.34	[.11, .53]	[.32, .36]	160.73***	97.51	5	2,718
Minus 2, 2.01	.16	[.11, .21]	.16	[.11, .21]	[.12, .20]	0.82	0.00	3	1,492
Interpersonal	.66	[.60, .70]	.65	[.49, .76]	[.61, .68]	20.43***	85.31	4	517
Minus 36	.59	[.51, .65]	.59	[.46, .69]	[.54, .64]	5.26	61.95	3	357
Affective	.38	[.30, .44]	.36	[.24, .47]	[.31, .41]	9.90*	59.60	5	565
Minus 36	.30	[.21, .39]	.30	[.21, .39]	[.23, .36]	0.30	0.00	4	405
Lifestyle	.65	[.60, .70]	.64	[.52, .74]	[.60, .68]	13.99***	78.55	4	517
Minus 36	.59	[.52, .65]	.59	[.52, .65]	[.54, .64]	1.02	0.00	3	357
Self-report	.37	[.35, .40]	.34	[.28, .40]	[.32, .36]	98.99***	78.79	22	4,698
Total scores	.38	[.35, .41]	.34	[.26, .42]	[.32, .36]	83.32***	83.20	15	3,165
Factor 1	.37	[.32, .41]	.34	[.22, .46]	[.31, .37]	31.16***	77.53	8	1,328
Factor 2	.42	[.37, .46]	.42	[.26, .53]	[.39, .45]	44.39***	81.98	9	1,402

Note. CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Moderators

The results have demonstrated that, overall, psychopathy is equally related to both instrumental and reactive violence. There is some evidence that Factor 2 may be more important in explaining reactive violence, while the Interpersonal facet may be more important in explaining instrumental violence; however, these results were not consistent across scale types. Although separating the analyses by scale type and removing outlying studies reduced the between-study variability (this was especially evident for clinical rating scales), considerable variability remained. To further explore this variability, several moderator analyses were conducted. It is important to note that at least three effect sizes for each level of categorical moderators were required for the moderator to be analyzed. Tables 7 and 8 present the results for instrumental and reactive violence, respectively.

Table 7:

Moderator Analyses by Type of Scale (Instrumental Violence)

	Fixed		Random
	Mean r	95% CI	Mean r	95% CI	Q	I²	k	n	Q _btw
Clinical total score
Country	.28	[.23, .32]	.30	[.22, 37]	40.49**	53.07	20	1,451	4.99*
Us	.24	[.18, .30]	.27	[.17, .36]	22.15*	50.34	12	966
Other	.35	[.27, .43]	.34	[.22, .46]	13.34	47.55	8	485
Sample type	.28	[.23, .32]	.30	[.22, 37]	40.49**	53.07	20	1,451	3.66
Offender	.26	[.21, .31]	.28	[.20, .36]	35.90**	55.43	17	1,279
General/Psychiatric	.40	[.26, .52]	.40	[.26, .52]	0.93	0.00	3	172
Gender	.27	[.22, .32]	.27	[.20, .34]	39.93**	54.92	19	1,364	1.25
Male	.29	[.23, .36]	.34	[.22, .44]	27.44**	63.56	11	847
Female/both	.24	[.15, .32]	.22	[.12, .32]	11.24	37.72	8	517
Outcome reporting	.28	[.23, .32]	.30	[.22, 37]	40.49**	53.07	20	1,451	4.44*
Objective	.31	[.25, .36]	.34	[.25, .43]	31.07**	54.94	15	1,043
Self-reported	.19	[.10, .28]	.19	[.08, .30]	4.98	19.72	5	408
Outcome measure	.28	[.23, .32]	.30	[.22, 37]	40.49**	53.07	20	1,451	20.66***
Categorical	.39	[.32, .45]	.38	[.31, .46]	15.54	22.77	13	715
Validated scale/ Simulation design	.16	[.09, .23]	.16	[.09, .23]	4.29	0.00	7	736
Informant (no outlier)
Gender	.52	[.48, .55]	.50	[.39, .59]	62.81***	87.26	9	2,163	19.35***
Male	.44	[.40, .49]	.41	[.21, .57]	29.26***	89.74	4	1,110
Female/both	.58	[.54, .62]	.58	[.53, .62]	14.20**	71.84	5	1,053
Outcome reporting	.52	[.48, .55]	.50	[.39, .59]	62.81***	87.26	9	2,163	0.95
Informant/objective	.55	[.47, .62]	.36	[.15, .54]	23.46***	87.21	4	351
Self-reported	.51	[.48, .54]	.51	[.39, .62]	38.40***	89.58	5	1,812
Self-report
Country	.40	[.37, .42]	.37	[.29, .44]	224.09***	88.84	26	5,075	28.88***
USA	.37	[.34, .39]	.31	[.22, .40]	190.29***	90.02	20	1,403
Other	.54	[.48, .58]	.54	[.48, .58]	4.92	0.00	6	774
Peer reviewed	.40	[.37, .42]	.37	[.29, .44]	224.09***	88.84	26	5,075	0.02
Yes	.40	[.37, .42]	.37	[.28, .46]	201.31***	90.06	21	4,149
No	.39	[.33, .44]	.33	[.16, .49]	22.75***	82.42	5	926
Sample type	.40	[.37, .42]	.37	[.29, .44]	224.09***	88.84	26	5,075	4.47*
Offender	.32	[.25, .39]	.32	[.21, .42]	13.83	49.39	8	598
General/psychiatric	.40	[.38, .43]	.38	[.29, .47]	205.79***	91.74	18	4,477
Gender	.39	[.37, .42]	.35	[.27, .43]	222.22***	89.20	25	5,010	3.20
Male	.42	[.38, .45]	.34	[.24, .43]	71.60***	80.45	15	2,069
female/both	.38	[.34, .41]	.38	[.24, .51]	147.42***	93.90	10	2,941
Outcome reporting	.40	[.37, .42]	.37	[.29, .44]	224.09***	88.84	26	5,075	51.20***
Objective	.18	[.12, .25]	.18	[.12, .25]	4.51	0.00	10	829
Self-reported	.43	[.41, .46]	.45	[.36, .54]	168.38***	91.09	16	4,246
Outcome measure	.40	[.37, .42]	.37	[.29, .44]	224.09***	88.84	26	5,075	51.85***
Categorical	.23	[.10, .35]	.23	[.10, .35]	2.11	0.00	4	228
Validated scale	.43	[.41, .46]	.45	[.36, .54]	168.38***	91.09	16	4,246
Simulation design	.17	[.09, .24]	.17	[.09, .24]	1.75	0.00	6	601

Note. CI = confidence interval. Each level of the moderator required at least three effect sizes to be analyzed.

p < .05. **p < .01. ***p < .001.

Table 8:

Moderator Analyses by Type of Scale (Reactive Violence)

	Fixed		Random
	Mean r	95% CI	Mean r	95% CI	Q	I²	k	n	Q _btw
Informant
Gender	.40	[.37, .43]	.43	[.28, .56]	208.57***	95.69	10	3,283	10.54**
Male	.35	[.30, .39]	.36	[.14, .55]	68.07***	95.59	4	1,634
Female/both	.44	[.40, .48]	.50	[.28, .66]	129.96***	96.15	6	1,649
Outcome reporting	.40	[.37, .43]	.43	[.28, .56]	208.57***	95.69	10	3,283	25.28***
Informant	.60	[.52, .67]	.54	[.27, .73]	14.87***	86.55	3	313
Self-reported	.37	[.34, .40]	.39	[.21, .54]	168.42***	96.44	7	2,970
Self-report
Country	.37	[.35, .40]	.34	[.28, .40]	98.99***	78.79	22	4,698	0.41
USA	.38	[.35, .40]	.32	[.24, .39]	78.57***	82.18	15	3,842
Other	.35	[.29, .41]	.40	[.28, .51]	20.02**	70.03	7	856
Peer reviewed	.37	[.35, .40]	.34	[.28, .40]	98.99***	78.79	22	4,698	43.41***
Yes	.42	[.38, .44]	.39	[.34, .44]	50.20***	68.12	17	3,772
No	.19	[.13, .25]	.17	[.09, .26]	5.39	25.75	5	926
Sample type	.37	[.35, .40]	.34	[.28, .40]	98.99***	78.79	22	4,698	2.43
Offender	.31	[.22, .39]	.31	[.14, .46]	16.12**	68.98	6	447
General/psychiatric	.38	[.35, .40]	.35	[.28, .42]	80.44***	81.35	16	4,251
Gender	.37	[.35, .40]	.34	[.28, .40]	97.85***	79.56	21	4,633	6.72**
Male	.42	[.38, .46]	.36	[.26, .45]	36.92***	75.62	10	1,612
Female/both	.35	[.32, .38]	.33	[.24, .41]	54.21***	81.55	11	3,021
Outcome reporting	.37	[.35, .40]	.34	[.28, .40]	98.99***	78.79	22	4,698	33.22***
Objective	.16	[.07, .24]	.16	[.05, .28]	11.05	45.69	7	565
Self-reported	.40	[.37, .42]	.40	[.34, .45]	54.72***	74.42	15	4,133
Outcome measure	.38	[.35, .40]	.37	[.32, .42]	88.39***	78.50	20	4,568	23.27***
Validated scale	.40	[.37, .42]	.40	[.34, .45]	54.72***	74.42	15	4,133
Simulation design	.17	[.08, .26]	.19	[.03, .34]	10.40*	61.54	5	435

Note. CI = confidence interval. Given that clinical scales had non-significant between-study variability for overall psychopathy and total scores, it was not analyzed for moderators. Each level of the moderator required at least three effect sizes to be analyzed.

p < .05. **p < .01. ***p < .001.

Clinical Scales

To ensure adequate power for moderator analyses, only effect sizes with significant Q values or I² values exceeding .50 were analyzed. For this reason, moderator analyses were conducted on the total scores for clinical scales. Between-study variability could not be explained by sample type or gender composition of the sample. Effect sizes were larger for studies conducted outside the United States. Clinical total scores were also moderated by outcome reporting with larger effect sizes in studies using an objective measure of type of violence as opposed to a self-reported measure. When the outcome was further examined, larger effect sizes were seen for studies utilizing a categorical definition of violence (based on the existing literature) compared with studies that utilized a validated, continuous scale or a simulation design.

When examining continuous moderators, clinical rating scores were moderated by ethnicity (k = 13), b = −.0061, SE = .0022, Z = −2.82, p = .005; effect sizes decreased as the percentage of ethnic minorities within the sample increased. In other words, effect sizes were smaller in samples with higher ethnic diversity. Age was not a significant moderator (k = 19), b = −.0011, SE = .0032, Z = −0.352, p = .725. Moderator analyses were not completed for reactive violence given that there was no significant between-study variability for psychopathy overall or for total scores and I² values did not exceed .50.

Informant Scales

Considerable variability was evident for informant scales overall. Given the three effect size rule, the only categorical moderators that could be assessed were gender and outcome reporting. Overall, the relationship between informant psychopathy scores and instrumental violence was significantly larger in samples containing both males and females. Outcome reporting was not significant. Informant scale scores were also significantly moderated by the mean age of the sample (k = 7), b = −.0540, SE = .0090, Z = −6.01, p < .001; for every decrease of one point in mean age, the relationship between instrumental violence and psychopathy increased. In other words, effect sizes were larger in younger samples. The relationship between psychopathy and instrumental violence as measured by informant scales was also moderated by ethnicity (k = 7), b = −.0037, SE = .0018, Z = −2.00, p = .046; effect sizes decreased as the percentage of ethnic minorities within the sample increased.

When examining the relationship between informant measures and reactive violence, effect sizes were larger in mixed gender samples but also when the outcome was objective (in this case provided by an informant) rather than self-reported. The relationship between informant-rated psychopathy and instrumental violence was moderated by the mean age of the samples (k = 6), b = −.0694, SE = .0093, Z = −7.43, p < .001 (larger effect sizes as the mean age decreased), but not by ethnicity (k = 6), b = .0012, SE = .0018, Z = 0.64, p = .540.

Self-Report Scales

For self-report scales, the relationship between psychopathy and instrumental violence was moderated by country and sample type with larger effect sizes for studies originating outside the United States and for non-offender samples. The relationship was not moderated by peer review status or by gender. As opposed to informant measures, the relationship between self-reported psychopathy and instrumental violence was larger when the outcome was also self-reported and when a validated scale was used. Finally, effect sizes were significantly moderated by the mean age of the sample (k = 24), b = −.0155, SE = .0024, Z = −6.55, p < .001, indicating that for every decrease of one point in mean age, the relationship between instrumental violence and self-reported psychopathy increased. Self-reported psychopathy scores were not significantly moderated by ethnicity (k = 21), b = −.0017, SE = .0009, Z = −1.93, p = .053.

For reactive violence, the relationship between self-reported psychopathy and reactive violence was significantly moderated by peer review status, sample type, gender, and both outcome moderators. Effect sizes were larger for peer-reviewed studies, general or psychiatric samples, male samples, and when the outcome was self-reported and measured using a validated scale. Consistent with the instrumental findings, the relationship between self-reported psychopathy and reactive violence increased as the mean age decreased (k = 20), b = −.0100, SE = .0030, Z = −3.21, p = .001. Ethnicity was not a significant moderator (k = 16), b = .0000, SE = .0010, Z = 0.10, p = .992.

Publication Bias

Given the limited number of unpublished sources available for the present study (n = 8), analyses assessing for publication bias were conducted. Preliminary inspection of funnel plots conducted for each outcome provided no evidence of publication bias for the overall results. Orwin’s (1983) fail safe N was also calculated, which would indicate the number of studies with a clinically unimportant finding (set at r = .10, for the present study) necessary to negate the present findings. For instrumental violence, the fail safe Ns for total, factor, and facet scores ranged from 19 to 138 studies needed to negate the overall findings. For reactive violence, the Ns ranged from 15 to 100. When the analyses were broken down by type of scale, fail safe Ns decreased to a range of 13 to 70 for instrumental violence and 2 to 53 for reactive violence.

Discussion

Past meta-analyses have established that psychopathy is related to violence, and furthermore, that this relationship is more pronounced for the SD characteristics as opposed to the IA characteristics of psychopathy (Kennealy et al., 2010; Walters, 2003a, 2003b). In addition, Kennealy and colleagues (2010) reported no significant interaction between IA and SD factors, indicating that although IA features predict violence significantly, future violence would be best predicted by looking at impulsivity and past criminal behaviors. Despite the superiority of SD in predicting violence, the question remained as to whether IA features were more important in predicting violence that was premeditated and callous (i.e., instrumental). The results of the present study indicated a moderate and significant relationship between psychopathy and both types of violence; higher psychopathy scores were related to higher rates of instrumental and reactive violence across studies. There was no evidence that psychopathy, overall, was more related to one form of violence over the other.

When examining the differential associations of factor and facet scores, the relationship between psychopathy and instrumental violence was stronger for the Interpersonal facet, while the relationship between psychopathy and reactive violence was stronger for Factor 2. The Lifestyle facet appeared important in explaining both types of violent outcomes. These findings, however, were inconsistent across scale type. For clinical rating scales, all factor and facet scores were equally and consistently related to both outcomes, while for informant measures, both the Interpersonal and Lifestyle facets were more related to both violent outcomes compared with the Affective facet. When assessing psychopathy through self-report measures, both factors were equally related to instrumental violence, while Factor 2 was more related to reactive violence. It is interesting to note that the effect sizes for both outcome measures were consistently moderate, indicating that other variables, besides psychopathy, are likely important when examining violence.

For both types of outcomes, the relationship between psychopathy and violence was smaller for clinical rating scales compared with informant and self-report scales, which is somewhat contrary to other meta-analyses examining psychopathy and violence in general (e.g., Asscher et al., 2011). It is possible that the interaction of another variable, sample type, was partly responsible for this finding. For self-report scales, the effect sizes were larger for non-offender samples compared with offender samples. Considering that the majority of samples included in the analyses for clinical rating scales were offenders, it is not surprising that the effect sizes were smaller compared with self-report scales. In addition, age was also a significant moderator indicating larger effect sizes in younger samples. The smaller effect sizes for clinical rating scales compared with informant measures may be explained by the fact that the majority of samples using clinical rating scales were adult only and all of the samples using informant measures were juvenile samples. This is also supported by the exclusive use of youth samples in the Asscher et al.’s (2011) study, which may explain the differing results.

Moderators

The current meta-analysis consistently identified significant between-study variability indicating the possibility that other factors may be contributing to the overall relationship between psychopathy and instrumental and reactive violence. After dividing the analyses by scale type, the between-study variability for clinical rating scales was substantially reduced. Contrary to some other meta-analyses on psychopathy and violence (e.g., Guy et al., 2005; Kennealy et al., 2010), country of origin was not a consistent moderator, and furthermore, studies outside the United States tended to show larger effect sizes. The mean age of the sample was a consistent moderator with stronger relationships between psychopathy and both outcomes in younger samples. This finding is consistent with Walters (2006) who also examined age as a moderator.

In past meta-analyses, gender and ethnicity have not always been consistent moderators. In the present study, although there was some evidence that the relationship between psychopathy and instrumental and reactive violence was larger for samples with both female and male participants, this was only true for informant measures. In instances where purely male samples could be contrasted against purely female samples, no significant differences were found. Ethnicity was also not a consistent moderator; however, consistent with Edens et al. (2006), smaller effect sizes were sometimes found for samples with larger ethnic heterogeneity.

The moderator analyses concerning sample type were also dependent on scale type. For example, sample type only significantly moderated the relationship between self-reported psychopathy and instrumental violence (i.e., stronger relationship for non-offender samples). It is important to note that given the distribution of sample types across scales, this moderator could not always be analyzed. Regardless, this finding can be partly explained by consistency in test creation and administration; self-report scales were developed on non-offender samples and therefore show larger relationships with outcomes such as violence when utilized with this population.

Similarly, moderator analyses demonstrated greater relationships between psychopathy and both outcomes when there was consistency between who was reporting on the psychopathic traits and who was reporting on the outcome measure (as evidenced by both the reporting strategy and the operationalization of the outcome measure). Self-reported psychopathy scores were more highly related to self-reported outcomes. All self-reported outcomes were based on validated self-report scales (e.g., the Reactive and Proactive Aggression Questionnaire [RPQ]; Raine et al., 2006). Similarly, for clinical rating scales, effect sizes were larger in studies where the researchers determined the criteria for evaluating type of violence (i.e., categorical distinction based on the research literature) instead of utilizing a self-report scale. There was also some evidence that the relationship between informants’ ratings of psychopathy were more related to informant outcomes, at least for reactive violence. These results demonstrate that the way in which psychopathy and the criterion measure are assessed can potentially inflate results.

Practical Implications

Although there are inconsistencies in the results across scale type, the results do indicate that psychopathy overall is not more related to instrumental violence compared with reactive violence. This finding has clear implications for how psychopathy is currently used in the criminal justice system. For example, in preventive detention hearings, it is more likely that an expert present a total score as opposed to factor scores in drawing conclusions concerning risk for violence and recidivism (Blais & Forth, in press). The present findings suggest that it would be inappropriate to assume that an offender is more likely to engage in calculated and callous violence based solely on his or her psychopathy score. As Camp, Skeem, Barchard, Lilienfeld, and Poythress (2013) caution, clinicians working within real-world settings should not assume a causal mechanism underlying psychopathy and “predatory” violence, especially in light of the current findings.

Limitations

Although the study presents several interesting findings, limitations should also be noted. First, a common criticism of meta-analyses is the supposed “file drawer problem”, which describes the fact that articles with significant findings and large effect sizes are more likely to be published and therefore more likely to be included in meta-analyses resulting in a biased sample (Rosenthal, 1979). Although the present search included unpublished sources such as dissertations, the majority of the studies were published. Moderator analyses of publication status (i.e., peer reviewed) were therefore not always possible. Analyses of the impact of this potential publication bias (funnel plots; Orwin’s fail safe N) did not indicate that this would have severely affected the overall results of the present study.

There was concern, however, once the analyses were broken down by scale type. Not only was there a small number of studies available for the analyses of facet scores (several analyses included less than five studies), but the fail safe Ns indicated that very few studies would be necessary to negate some of the findings. This was especially true for the reactive violence outcome. More studies are needed before firm conclusions can be drawn regarding the relationship between psychopathy facets and instrumental and reactive violence. Future studies should therefore extend the search criteria in an effort to include more sources with a focus on unpublished material.

A second limitation of the present study was that the analyses were based on the zero-order correlations presented in the studies. Recent research has started to explore whether Factor 1 of psychopathy is able to add to the prediction of violent outcomes after controlling for the effect of Factor 2 (e.g., Kennealy et al., 2010). Such analyses could not be conducted in the present study and incremental analyses should be considered in future studies. In addition, future studies should restrict the analyses to studies with truly prospective designs to protect against criterion contamination. This refers to the fact that psychopathy assessments often contain items pertaining to instrumental and reactive violence and the same information may be used in scoring both the items and the violent outcome. Such contamination can inflate the relationship between psychopathy and outcome measures in retrospective designs (Patrick & Zempolich, 1998; Swogger & Kosson, 2007).

Conclusion

Despite the study limitations, the results of this study call into question the assumption that psychopathy is more related to instrumental violence as opposed to reactive violence; our results demonstrate equal relationships between psychopathy and instrumental and reactive violence across scale types. Although further research is required when examining facet scores, the present findings do confirm that the relationship between psychopathy and instrumental and reactive violence is complex and may depend on how psychopathy and violence are operationalized and measured.

Footnotes

Acknowledgements

We would like to thank Kelly M. Babchishin for providing guidance on the statistical software and analyses.

Notes

Julie Blais is a doctoral candidate in forensic psychology at Carleton University. Her research interests include psychopathy, risk assessment, and risk communication. She has also completed projects on the study of special populations including mentally disordered offenders and Aboriginal offenders.

Elizabeth Solodukhin is a master’s student at the University of Ontario Institute of Technology. She completed her undergraduate work at Carleton University. Her research interests include deception detection and wrongful conviction.

Adelle Forth, PhD, is an associate professor at the Department of Psychology, Carleton University. She worked with Dr. Robert Hare to develop the Hare Psychopathy Checklist-Revised (PCL-R) and is the senior author of the Hare Psychopathy Checklist: Youth Version (PCL:YV). Adelle devotes her time to teaching and research on the assessment of psychopathy, the relationship between emotional and cognitive processes and psychopathy, victims of psychopathy, and violence risk assessment.

References

Andershed

Kerr

Stattin

Levander

(2002). Psychopathic traits in non-referred youths: A new assessment tool. In Blaauw

Sheridan

(Eds.), Psychopaths: Current international perspectives (pp. 131-158). The Hague, the Netherlands: Elsevier.

Asscher

J. J.

van Vugt

E. S.

Stams

G. J.

Dekovic

Eichelsheim

I. V.

Yousfi

(2011). The relationship between juvenile psychopathic traits, delinquency and (violent) recidivism: A meta-analysis. Journal of Child Psychology and Psychiatry, 52, 1134-1143. doi:10.1111/j.1469-7610.2011.02412.x

Babiak

Neumann

C. S.

Hare

R. D.

(2010). Corporate psychopathy: Talking the walk. Behavioral Sciences & the Law, 28, 174-193. doi:10.1002/bsl.925

*Barker

E. D.

(2003). Male and female reports of intimate couple aggression: The influence of method and social desirability (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (UMI No. 3116259)

*Barr

(2008). Extending psychopathy to incarcerated adolescent males: An investigation of construct equivalence (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (UMI No. 3304634)

*Barry

T. D.

Thompson

Barry

C. T.

Lochman

J. E.

Adler

Hill

(2007). The importance of narcissism in predicting proactive and reactive aggression in moderately to highly aggressive children. Aggressive Behavior, 33, 185-197. doi:10.1002/ab.20198

Benning

S. D.

Patrick

C. J.

Hicks

B. M.

Blonigen

D. M.

Krueger

R. F.

(2003). Factor structure of the Psychopathic Personality Inventory: Validity and implications for clinical assessment. Psychological Assessment, 15, 340-350. doi:10.1037/1040-3590.15.3.340

Berardino

S. D.

Meloy

J. R.

Sherman

Jacobs

(2005). Validation of the Psychopathic Personality Inventory on a female inmate sample. Behavioral Sciences & the Law, 23, 819-836. doi:10.1002/bsl.666

Berkowitz

(1993). Aggression: Its causes, consequences, and control. Philadelphia, PA: Temple University Press.

10.

*Bezdjian

Tuvblad

Raine

Baker

L. A.

(2011). The genetic and environmental covariation among psychopathic personality traits, and reactive and proactive aggression in childhood. Child Development, 82, 1267-1281. doi:10.1111/j.1467-8624.2011.01598.x

11.

*Bjørnebekk

(2007). Dispositions related to sensitivity in the neurological basis for activation of approach-avoidance motivation, antisocial attributes and individual differences in aggressive behavior. Social Behavior and Personality, 35, 1251-1264. doi:https://dx-doi-org.web.bisu.edu.cn/10.2224/sbp.2007.35.9.1251

12.

Blais

Forth

A. E.

(in press, 2013). Prosecution-retained versus court-appointed experts: Comparing and contrasting risk assessment reports in preventative detention hearings. Law and Human Behavior.

13.

*Bobadilla

Wampler

Taylor

(2012). Proactive and reactive aggression are associated with different physiological and personality profiles. Journal of Social & Clinical Psychology, 31, 458-487. doi:10.1521/jscp.2012.31.5.458

14.

Borenstein

Hedges

L. V.

Higgins

J. P. T.

Rothstein

H. R.

(2005). Comprehensive Meta-Analysis (Version 2) [Computer software]. Englewood, NJ: Biostat.

15.

Borenstein

Hedges

L. V.

Higgins

J. P. T.

Rothstein

H. R.

(2009). Introduction to meta-analysis. West Sussex, UK: Wiley.

16.

Brinkley

C. A.

Schmitt

W. A.

Smith

S. S.

Newman

J. P.

(2001). Construct validation of a self-report psychopathy scale: Does Levenson’s Self-Report Psychopathy Scale measure the same constructs as Hare’s Psychopathy Checklist-Revised? Personality and Individual Differences, 31, 1021-1038. doi:10.1016/S0191-8869(00)00178-1

17.

Bushman

B. J.

Anderson

C. A.

(2001). Is it time to pull the plug on the hostile versus instrumental aggression dichotomy? Psychological Review, 108, 273-279. doi:IO.I037//0033-295X.108.1.273

18.

Camp

J. P.

Skeem

J. L.

Barchard

Lilienfeld

S. O.

Poythress

N. G.

(2013). Psychopathic predators? Getting specific about the relation between psychopathy and violence. Journal of Consulting and Clinical Psychology, 81, 467-480. doi:10.1037/a0031349

19.

*Chase

K. A.

O’Leary

Heyman

R. E.

(2001). Categorizing partner-violent men within the reactive-proactive typology model. Journal of Consulting and Clinical Psychology, 69, 567-572. doi:10.1037//0022-006X.69.3.567

20.

*Cima

Raine

(2009). Distinct characteristics of psychopathy relate to different subtypes of aggression. Personality and Individual Differences, 47, 835-840. doi:10.1016/j.paid.2009.06.031

21.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

22.

Coid

Yang

Ullrich

Roberts

Hare

R. D.

(2009). Prevalence and correlates of psychopathic traits in the household population of Great Britain. International Journal of Law and Psychiatry, 32, 65-73. doi:10.1016/j.ijlp.2009.01.002

23.

Cooke

D. J.

Michie

(2001). Refining the construct of psychopathy: Towards a hierarchical model. Psychological Assessment, 13, 171-188. doi:10.1037/1040-3590.13.2.171

24.

*Cornell

D. G.

Warren

Hawk

Stafford

Oram

Pine

(1996). Psychopathy in instrumental and reactive violent offenders. Journal of Consulting and Clinical Psychology, 64, 783-790. doi:10.1037/0022-006X.64.4.783

25.

*Crawley

Martin

F. H.

(2006). Impulsive-aggression, antisocial behaviour and subclinical psychopathy: Preliminary findings from an undergraduate female sample. Psychiatry, Psychology, and Law, 13, 232-242. doi:10.1375/pplt.13.2.232

26.

*Davis

C. E.

(2010). The affective features of psychopathy in incarcerated women (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (UMI No. 3401774)

27.

*Declercq

Willemsen

Audenaert

Verhaeghe

(2012). Psychopathy and predatory violence in homicide, violent, and sexual offences: Factor and facet relations. Legal and Criminological Psychology, 17, 59-74. doi:10.1348/135532510X527722

28.

Edens

J. F.

Campbell

J. S.

Weir

M. J.

(2006). Youth psychopathy and criminal recidivism: A meta-analysis of the psychopathy checklist measures. Law and Human Behavior, 31, 53-75. doi:10.1007/s10979-006-9019-y

29.

Falkenbach

D. M.

Poythress

N. G.

Heide

K. M.

(2003). Psychopathic features in a juvenile diversion population: Reliability and predictive validity of two self-report measures. Behavioral Sciences & the Law, 21, 787-805. doi:10.1002/bsl.562

30.

*Fanti

K. A.

Frick

P. J.

Georgiou

(2009). Linking callous-unemotional traits to instrumental and non-instrumental forms of aggression. Journal of Psychopathology and Behavioral Assessment, 31, 285-298. doi:10.1007/s10862-008-9111-3

31.

*Fite

P. J.

Raine

Stouthamer-Loeber

Loeber

Pardini

D. A.

(2010). Reactive and proactive aggression in adolescent males: Examining differential outcomes 10 years later in early adulthood. Criminal Justice and Behavior, 37, 141-157. doi:10.1177/0093854809353051

32.

*Fite

P. J.

Stoppelbein

Greening

(2009). Proactive and reactive aggression in a child psychiatric inpatient population: Relations to psychopathic characteristics. Criminal Justice and Behavior, 36, 481-493. doi:10.1177/0093854809332706

33.

*Flight

J. I.

Forth

A. E.

(2007). Instrumentally violent youths: The roles of psychopathic traits, empathy, and attachment. Criminal Justice and Behavior, 34, 739-751. doi:10.1177/0093854807299462

34.

Forth

A. E.

Kosson

D. S.

Hare

R. D.

(2003). The Hare Psychopathy Checklist: Youth Version. North Tonawanda, NY: Multi-Health Systems.

35.

Frick

P. J.

Bodin

S. D.

Barry

C. T.

(2000). Psychopathic traits and conduct problems in community and clinic-referred samples of children: Further development of the Psychopathy Screening Device. Psychological Assessment, 12, 382-393. doi:10.1037/1040-3590.12.4.382

36.

Frick

P. J.

Hare

R. D.

(2001). Antisocial Process Screening Device. Toronto, ON: Multi-Health Systems.

37.

Glenn

A. L.

Raine

(2009). Psychopathy and instrumental aggression: Evolutionary, neurobiological, and legal perspectives. International Journal of Law and Psychiatry, 32, 253-258. doi:10.1016/j.ijlp.2009.04.002

38.

Guy

L. S.

Douglas

K. S.

Hendry

M. C.

(2010). The role of psychopathic personality disorder in violence risk assessments using the HCR-20. Journal of Personality Disorders, 24, 551-580. doi:10.1521/pedi.2010.24.5.551

39.

Guy

L. S.

Edens

J. F.

Anothony

Douglas

K. S.

(2005). Does psychopathy predict institutional misconduct among adults? A meta-analytic investigation. Journal of Consulting and Clinical Psychology, 73, 1056-1064. doi:10.1037/0022-006X.73.6.1056

40.

Hare

R. D.

(1991). Manual for the Revised Psychopathy Checklist (1st ed.). Toronto, ON: Multi-Health Systems.

41.

Hare

R. D.

(2003). The Hare Psychopathy Checklist-Revised (PCL-R) (2nd ed.). Toronto, ON: Multi-Health Systems.

42.

Harpur

T. J.

Hakstian

A. R.

Hare

R. D.

(1988). Factor structure of the Psychopathy Checklist. Journal of Consulting and Clinical Psychology, 56, 741-747. doi:10.1037/0022-006X.56.5.741

43.

Harris

G. T.

Rice

M. E.

Cormier

C. A.

(1991). Psychopathy and violent recidivism. Law and Human Behavior, 15, 625-637. doi:10.1007/BF01065856

44.

Hart

S. D.

Cox

D. N.

Hare

R. D.

(1995). The Hare Psychopathy Checklist: Screening Version (1st ed.). Toronto, ON: Multi-Health Systems.

45.

Hedges

L. V.

Olkin

(1985). Statistical methods for meta-analysis. San Diego, CA: Academic Press.

46.

*Hodges

Heilbrun

(2009). Psychopathy as a predictor of instrumental violence among civil psychiatric patients. International Journal of Forensic Mental Health, 8, 131-141. doi:10.1080/14999010903199373

47.

Huedo-Medina

T. B.

Sánchez-Meca

Marín-Martínez

Botella

(2006). Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychological Methods, 11, 193-206. doi:10.1037/1082-989X.11.2.193

48.

*Jones

D. N.

Paulhus

D. L.

(2010). Different provocations trigger aggression in narcissists and psychopaths. Social Psychological & Personality Science, 1, 12-18. doi:10.1177/1948550609347591

49.

Kennealy

P. J.

Skeem

J. L.

Walters

G. D.

Camp

(2010). Do core interpersonal and affective traits of PCL-R psychopathy interact with antisocial behavior and disinhibition to predict violence? Psychological Assessment, 22, 569-580. doi:10.1037/a0019618

50.

*Kerig

P. K.

Stellwagen

K. K.

(2010). Roles of callous-unemotional traits, narcissism, and Machiavellianism in childhood aggression. Journal of Psychopathology and Behavioral Assessment, 32, 343-352. doi:10.1007/s10862-009-9168-7

51.

Kosson

D. S.

Neumann

C. S.

Forth

A. E.

Salekin

R. T.

Hare

R. D.

Krischer

M. K.

Sevecke

(2013). Factor structure of the Hare Psychopathy Checklist: Youth Version (PCL: YV) in adolescent females. Psychological Assessment, 25, 71-83. doi:10.1037/a0028986

52.

*Laurell

Belfrage

Hellström

(2010). Facets on the psychopathy checklist screening version and instrumental violence in forensic psychiatric patients. Criminal Behavior and Mental Health, 20, 285-294. doi:10.1002/cbm.779

53.

*Lehmann

Ittel

(2012). Aggressive behavior and measurement of psychopathy in female inmates of German prisons—A preliminary study. International Journal of Law and Psychiatry, 35, 190-197. doi:10.1016/j.ijlp.2012.02.007

54.

*Leistico

A.-M. R.

Salekin

R. T.

(2003). Testing the reliability and validity of the Risk, Sophistication-Maturity, and Treatment Amenability Instrument (RST-i): An assessment tool for juvenile offenders. International Journal of Forensic Mental Health, 2, 101-117. doi:10.1080/14999013.2003.10471182

55.

Levenson

M. R.

Kiehl

K. A.

Fitzpatrick

C. M.

(1995). Assessing psychopathic attributes in a noninstitutionalized population. Journal of Personality and Social Psychology, 68, 151-158. doi:10.1037/0022-3514.68.1.151

56.

Lilienfeld

S. O.

(1994). Conceptual problems in the assessment of psychopathy. Clinical Psychology Review, 14, 17-38. doi:10.1016/0272-7358(94)90046-9

57.

Lilienfeld

S. O.

Andrews

B. P.

(1996). Development and preliminary validation of a self-report measure of psychopathic personality traits in noncriminal populations. Journal of Personality Assessment, 66, 488-524. doi:10.1207/s15327752jpa6603_3

58.

Lilienfeld

S. O.

Fowler

K. A.

(2006). The self-report assessment of psychopathy: Pitfalls, problems, and promises. In Patrick

(Ed.), Handbook of psychopathy (pp. 107-132). New York, NY: Guilford Press.

59.

Lynam

D. R.

(1997). Pursuing the psychopath: Capturing the fledgling psychopath in a nomological net. Journal of Abnormal Psychology, 106, 425-438. doi:10.1037/0021-843X.106.3.425

60.

*Marsee

M. A.

Frick

P. J.

(2007). Exploring the cognitive and emotional correlates to proactive and reactive aggression in a sample of detained girls. Journal of Abnormal Child Psychology, 35, 969-981. doi:10.1007/s10802-007-9147-y

61.

*McDermott

B. E.

Quanbeck

C. D.

Busse

Yastro

Scott

L. C.

(2008). The accuracy of risk assessment instruments in the prediction of impulsive versus predatory aggression. Behavioral Sciences & the Law, 26, 759-777. doi:10.1002/bsl.842

62.

Meloy

J. R.

(2006). Empirical basis and forensic application of affective and predatory violence. Australian and New Zealand Journal of Psychiatry, 40, 539-547. doi:10.1080/j.1440-1614.2006.01837.x

63.

Millon

(1993). The Millon Adolescent Clinical Inventory (MACI) manual. Minneapolis, MN: National Computer Systems.

64.

*Muñoz

L. C.

Frick

P. J.

Kimonis

E. R.

Aucoin

K. J.

(2008). Types of aggression, responsiveness to provocation, and callous-unemotional traits in detained adolescents. Journal of Abnormal Child Psychology, 36, 15-28. doi:10.1007/s10802-007-9137-0

65.

*Murrie

D. C.

Cornell

D. G.

Kaplan

McConville

Levy-Elkon

(2004). Psychopathy scores and violence among juvenile offenders: A multi-measure study. Behavioral Sciences & the Law, 22, 49-67. doi:10.1002/bsl.573

66.

Neal

T. M. S.

Sellbom

(2012). Examining the factor structure of the Hare Self-Report Psychopathy Scale. Journal of Personality Assessment, 94, 244-253. doi:10.1080/00223891.2011.648294

67.

Neumann

C. S.

Kosson

D. S.

Forth

A. E.

Hare

R. D.

(2006). Factor structure of the Hare Psychopathy Checklist: Youth Version (PCL: YV) in incarcerated adolescents. Psychological Assessment, 18, 142-154. doi:0.1037/1040-3590.18.2.142

68.

Neumann

C. S.

Kosson

D. S.

Salekin

R. T.

(2007). Exploratory and confirmatory factor analysis of the psychopathy construct: Methodological and conceptual issues. In Neumann

C. S.

Kosson

D. S.

Salekin

R. T.

(Eds.), The Psychopath: Theory, research, and practice (pp. 79-104). Mahwah, NJ: Lawrence Erlbaum.

69.

*Nouvion

S. O.

Cherek

D. R.

Lane

S. D.

Tcheremissine

O. V.

Lieving

L. M.

(2007). Human proactive aggression: Association with personality disorders and psychopathy. Aggressive Behavior, 33, 552-562. doi:10.1002/ab.20220

70.

Olver

M. E.

Stockdale

K. C.

Wormith

J. S.

(2009). Risk assessment with young offenders: A meta-analysis of three assessment measures. Criminal Justice and Behavior, 36, 329-353. doi:10.1177/0093854809331457

71.

Orwin

R. G.

(1983). A fail safe N for effect size in meta-analysis. Journal of Educational Statistics, 8, 157-159. doi:10.2307/1164923

72.

*Ostrov

J. M.

Houston

R. J.

(2008). The utility of forms and functions of aggression in emerging adulthood: Association with personality disorder symptomology. Journal of Youth Adolescence, 37, 1147-1158. doi:10.1007/s10964-008-9289-4

73.

Patrick

C. J.

Zempolich

K. A.

(1998). Emotion and aggression in the psychopathic personality. Aggression and Violent Behavior, 3, 303-338. doi:10.1016/S1359-1789(97)00003-7

74.

Paulhus

D. L.

Neumann

C. S.

Hare

R. D.

(in press). Manual for the Self-Report Psychopathy Scale. Toronto, ON: Multi-Health Systems.

75.

*Phillips

D. M.

(2003). Developmental pathways for children with disruptive behavior disorders (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (UMI No. 3122156)

76.

*Porter

Woodworth

(2007). “I’m sorry I did it . . . but he started it”: A comparison of the official and self-reported homicide descriptions of psychopaths and non-psychopaths. Law and Human Behavior, 31, 91-107. doi:10.1007/s10979-006-9033-0

77.

Poythress

N. G.

Edens

J. F.

Lilienfeld

S. O.

(1998). Criterion-related validity of the Psychopathic Personality Inventory in a prison sample. Psychological Assessment, 10, 426-430. doi:10.1037/1040-3590.10.4.426

78.

*Preston

L. A. D.

(1998). Psychopathy and its associations to the Five-Factor Model of normal personality (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (UMI No. 9923611)

79.

*Ragatz

L. L.

Anderson

R. J.

Fremouw

Schwartz

(2011). Criminal thinking patterns, aggression styles, and the psychopathic traits of late high school bullies and bully-victims. Aggressive Behavior, 37, 145-160. doi:10.1002/ab.20377

80.

*Raine

Dodge

Loeber

Gatzke-Kopp

Lynam

Reynolds

. . . Liu

(2006). The Reactive- Proactive Aggression Questionnaire: Differential correlates of reactive and proactive aggression in adolescent boys. Aggressive Behavior, 32, 159-171. doi:10.1002/ab.20115

81.

Ray

J. V.

Hall

Rivera-Hudson

Poythress

N. G.

Lilienfeld

S. O.

Morano

(2013). The relation between self-reported psychopathic traits and distorted response styles: A meta-analytic review. Personality Disorders: Theory, Research, and Treatment, 4, 1-14. doi:0.1037/a0026482

82.

Reidy

D. E.

Shelley-Tremblay

J. F.

Lilienfeld

S. O.

(2011). Psychopathy, reactive aggression, and precarious proclamations: A review of behavioral, cognitive, and biological research. Aggression and Violent Behavior, 16, 512-524. doi:10.1016/j.avb.2011.06.002

83.

*Reidy

D. E.

Zeichner

Martinez

M. A.

(2008). Effects of psychopathy traits on unprovoked aggression. Aggressive Behavior, 34, 319-328. doi:10.1002/ab.20238

84.

*Reidy

D. E.

Zeichner

Miller

J. D.

Martinez

M. A.

(2007). Psychopathy and aggression: Examining the role of psychopathy factors in predicting laboratory aggression under hostile and instrumental conditions. Journal of Research in Personality, 41, 1244-1251. doi:10.1016/j.jrp.2007.03.001

85.

*Reidy

D. E.

Zeichner

Seibert

L. A.

(2011). Unprovoked aggression: Effects of psychopathic traits and sadism. Journal of Personality, 79, 75-100. doi:10.1111/j.1467-6494.2010.00691.x

86.

Rosenthal

(1979). The “file drawer” problem and tolerance for null results. Psychological Bulletin, 86, 638-641. doi:10.1037/0033-2909.86.3.638

87.

Salekin

R. T.

Rogers

Sewell

K. W.

(1996). A review and meta-analysis of the Psychopathy Checklist and Psychopathy Checklist-Revised: Predictive validity of dangerousness. Clinical Psychology: Science and Practice, 3, 203-215. doi:10.1111/j.1468-2850.1996.tb00071.x

88.

*Seals

R. W.

Sharp

Michonski

J. D.

(2012). The relationship between the Youth Psychopathic Traits Inventory and psychopathology in a U.S. community sample of male youth. Journal of Personality Assessment, 94, 232-243. doi:10.1080/00223891.2011.650303

89.

Sellbom

(2011). Elaborating on the construct validity of the Levenson Self-Report Psychopathy Scale in incarcerated and non-incarcerated samples. Law and Human Behavior, 35, 440-451. doi:10.1007/s10979-010-9249-x

90.

*Serin

R. C.

(1991). Psychopathy and violence in criminals. Journal of Interpersonal Violence, 6, 423-431. doi:10.1177/088626091006004002

91.

*Stafford

Cornell

D. G.

(2003). Psychopathy scores predict adolescent inpatient aggression. Assessment, 10, 102-112. doi:10.1177/1073191102250341

92.

*Stanford

M. S.

Houston

R. J.

Baldridge

R. M.

(2008). Comparison of impulsive and premeditated perpetrators of intimate partner violence. Behavioral Sciences & the Law, 26, 709-722. doi:10.1002/bsl.808

93.

*Steadham

J. A.

(2012). Psychopathic and antisocial personality disorder traits as predictors of reactive and instrumental aggression (Master’s thesis). Available from ProQuest Dissertations and Theses database. (UMI No. 1521503)

94.

Swogger

M. T.

Kosson

D. S.

(2007). Identifying subtypes of criminal psychopaths: A replication and extension. Criminal Justice and Behavior, 34, 953-970. doi:10.1177/0093854807300758

95.

*Swogger

M. T.

Walsh

Houston

R. J.

Cashman-Brown

Conner

K. R.

(2010). Psychopathy and axis I psychiatric disorders among criminal offenders: Relationships to impulsive and proactive aggression. Aggressive Behavior, 36, 45-53. doi:10.1002/ab.20330

96.

*Taubner

White

L. O.

Zimmermann

Fonagy

Nolte

(2012, June). Mentalization moderates and mediates the link between psychopathy and aggressive behavior in male adolescents. A selection of posters described in the Journal of American Psychoanalytical Association, 60, 605-612. doi:10.1177/0003065112450148

97.

Tryon

W. W.

(2001). Evaluating statistical difference, equivalence, and indeterminacy using inferential confidence intervals: An integrated alternative method of conducting null hypothesis statistical tests. Psychological Methods, 6, 371-386. doi:10.1037/1082-989X.6.4.371

98.

*van Baardewijk

Vermeiren

Stegge

Doreleijers

(2011). Self-reported psychopathic traits in children: Their stability and concurrent and prospective association with conduct problems and aggression. Journal of Psychopathology and Behavioral Assessment, 33, 236-245. doi:10.1007/s10862-010-9215-4

99.

*Verschuere

Candel

Van Reenen

Korebrits

(2012). Validity of the modified Child Psychopathy Scale for juvenile justice center residents. Journal of Psychopathology and Behavioral Assessment, 34, 244-252. doi:10.1007/s10862-011-9272-3

100.

*Vitacco

M. J.

Gonsalves

J. T.

Smith

B. E. R.

Lishner

D. A.

(2012). Can standardized measures of risk predict inpatient violence? Combining static and dynamic variables to improve accuracy. Criminal Justice and Behavior, 39, 589-606. doi:10.1177/0093854812436786

101.

Vitacco

M. J.

Neumann

C. S.

Caldwell

M. F.

(2010). Predicting antisocial behavior in high-risk male adolescents: Contributions of psychopathy and instrumental violence. Criminal Justice and Behavior, 37, 833-846. doi:10.1177/0093854810371358

102.

*Vitacco

M. J.

Neumann

C. S.

Caldwell

M. F.

Leistico

A. M.

Van Rybroek

G. J.

(2006). Testing factor models of the Psychopathy Checklist: Youth Version and their association with instrumental aggression. Journal of Personality Assessment, 87, 74-83. doi:10.1207/s15327752jpa8701_06

103.

*Vitacco

M. J.

Van Rybroek

G. J.

Rogstad

J. E.

Yahr

L. E.

Tomony

J. D.

Saewert

(2009). Predicting short-term institutional aggression in forensic patients: A multi-trait method for understanding subtypes of aggression. Law and Human Behavior, 33, 308-319. doi:10.1007/s10989-008-9155-7

104.

*Walsh

Swogger

M. T.

Kosson

D. S.

(2009). Psychopathy and instrumental violence: Facet level relationships. Journal of Personality Disorders, 23, 416-424. doi:10.1521/pedi.2009.23.4.416

105.

Walters

G. D.

(2003a). Predicting criminal justice outcomes with the psychopathy checklist and lifestyle criminality screening form: A meta-analytic comparison. Behavioral Sciences & the Law, 21, 89-102. doi:10.1002/bsl.519

106.

Walters

G. D.

(2003b). Predicting institutional adjustment and recidivism with the Psychopathy Checklist factor scores: A meta-analysis. Law and Human Behavior, 27, 541-558. doi:10.1023/A:1025490207678

107.

Walters

G. D.

(2006). Risk-appraisal versus self-report in the prediction of criminal justice outcomes: A meta-analysis. Criminal Justice and Behavior, 33, 279-304. doi:10.1177/0093854805284409

108.

Weaver

C. M.

Meyer

R. G.

Van Nort

J. J.

Tristan

(2006). Two-, three-, and four-factor PCL-R models in applied sex offender risk assessments. Assessment, 13, 208-216. doi:10.1177/1073191106287116

109.

Weiler

B. L.

Widom

C. S.

(1996). Psychopathy and violent behaviour in abused and neglected young adults. Criminal Behavior and Mental Health, 6, 253-271. doi:10.1002/cbm.99

110.

*Williamson

Hare

R. D.

Wong

(1987). Violence: Criminal psychopaths and their victims. Canadian Journal of Behavioral Science, 19, 454-462. doi:10.1037/h0080003

111.

*Wirth

J. H.

(2010). Ostracism and aggression: The moderating influence of psychopathic traits (Doctoral dissertation). Available from ProQuest Dissertations and Theses database. (UMI No. 3444879)

112.

Wong

(1984). The criminal and institutional behaviors of psychopaths (Program Branch User Report, 1984-87). Ottawa, Ontario: Ministry of the Solicitor General of Canada.

113.

*Woodworth

Porter

(2002). In cold blood: Characteristics of criminal homicides as a function of psychopathy. Journal of Abnormal Psychology, 111, 436-444. doi:10.1037/0021-843X.111.3.436

114.

*Yeh

M. T.

Chen

Raine

Baker

L. A.

Jacobson

K. C.

(2011). Child psychopathic traits moderate relationships between parental affect and child aggression. Journal of the American Academy of Child & Adolescent Psychiatry, 50, 1054-1064. doi:10.1016/j.jaac.2011.06.013