Psychopathy and Machiavellianism - Wiley Online Library

5 downloads 0 Views 160KB Size Report
Jessica L. Maples-Keller,1 Nathan T. Carter,1 and. Donald R. Lynam2. 1University of Georgia. 2Purdue University. Abstract. A robust literature has emerged on ...
Journal of Personality 00:00, Month 2016

Psychopathy and Machiavellianism: A Distinction Without a Difference?

C 2016 Wiley Periodicals, Inc. V

DOI: 10.1111/jopy.12251

Joshua D. Miller,1 Courtland S. Hyatt,1 Jessica L. Maples-Keller,1 Nathan T. Carter,1 and Donald R. Lynam2 1 2

University of Georgia Purdue University

Abstract A robust literature has emerged on the Dark Triad (DT) of personality—Machiavellianism (MACH), psychopathy, and narcissism. Questions remain as to whether MACH and psychopathy are distinguishable and whether MACH’s empirical and theoretical networks are consistent. In Study 1 (N 5 393; MTurk research participants), factor analyses were used to compare two-factor (MACH and psychopathy combined 1 narcissism) and three-factor models, with both fitting the data equally well. In Studies 1 and 2 (N 5 341; undergraduate research participants), DT scores were examined in relation to a variety of external criteria, including self- and informant ratings of personality, adverse developmental experiences, and psychopathological symptoms/behaviors. In both studies, MACH and psychopathy manifested nearly identical empirical profiles and both were significantly related to disinhibitory traits thought to be antithetical to MACH. In Study 3 (N 5 36; expert raters), expert ratings of the Five-Factor Model traits prototypical of MACH were collected and compared with empirically derived profiles. Measures of MACH yielded profiles that were inconsistent with the prototypical expert-rated profile due to their positive relations with a broad spectrum of impulsivity-related traits. Ultimately, measures of psychopathy and MACH appear to be measuring the same construct, and MACH assessments fail to capture the construct as articulated in theoretical descriptions.

Since the original publication in 2002 (Paulhus & Williams), there has been a surge of publications on the Dark Triad (DT), comprising the simultaneous study of the three overlapping constructs of narcissism, psychopathy, and Machiavellianism.1 Narcissism is often viewed through the lens of a “self-regulatory model” (Morf & Rhodewalt, 2001) in which narcissistic individuals are continually striving for attention, admiration, and affirmation from others; from a trait perspective, narcissistic individuals tend to be grandiose, attention seeking, entitled, and domineering (Paulhus, 2001). Psychopathy, which has typically been studied in the context of clinical and forensic psychology, represents an amalgam of traits, including grandiosity, superficial charm, callousness, exploitativeness, impulsivity, and irresponsibility (e.g., Cleckley, 1941; Hare, 2003). Machiavellianism is a construct based on the writing of Niccolo Machiavelli and has been studied most commonly in the socialpersonality literature (McHoskey, Worzel, & Szyarto, 1998). Machiavellian individuals tend to be viewed as ambitious, strategic, capable of delaying gratification, manipulative, and amoral. Empirically, the DT constructs manifest moderate to large interrelations; O’Boyle, Forsyth, Banks, and McDaniel (2012) reported meta-analytically derived and corrected correlations of

.30 (Machiavellianism–narcissism), .51 (narcissism–psychopathy), and .59 (Machiavellianism–psychopathy). Similarly, research from a behavioral genetic approach (Vernon, Villani, Vickers, & Harris, 2008) demonstrated that the DT constructs overlap, in part, due to shared genetic features, particularly psychopathy’s relations with narcissism and Machiavellianism (genetic correlations of .48 and .66, respectively). All three manifested substantial genetic correlations with trait Agreeableness as well (genetic correlations ranging from –.42 to –.78), supporting its role as a foundational construct for the DT (e.g., Miller et al., 2010). The empirical literature on the DT has two main foci. The first focus involves an attempt to identify the core feature(s) that account for their overlap; candidate core features include low Agreeableness from the Big Five/Five-Factor Model (FFM; e.g., Jakobwitz & Egan, 2006; Miller et al., 2010; O’Boyle, Forsyth, Banks, Story, & White, 2015), low Honesty-Humility from Big Six perspectives (e.g., Lee & Ashton, 2014), or callousmanipulation (Jones & Figueredo, 2013). The second focus Correspondence concerning this article should be addressed to Joshua D. Miller, Department of Psychology, University of Georgia, 125 Baldwin Street, Athens, GA 30602. Email: [email protected].

2

attempts to delineate ways in which the DT constructs differ from one another in relation to other relevant variables, including personality traits outside of Agreeableness and HonestyHumility (e.g., Paulhus & Williams, 2002), interpersonal behavior (e.g., Rauthmann & Kolar, 2013), morality (e.g., Jonason, Strosser, Kroll, Duineveld, & Baruffi, 2015), employmentrelated behavior (e.g., DeShong, Grant, & Mullins-Sweatt, 2015), sexual strategies (e.g., Jonason, Lyons, & Blanchard, 2015), antisocial behavior (e.g., Pailing, Boon, & Egan, 2014), and financial behavior (e.g., Jones, 2013, 2014). However, the substantial relation between psychopathy and Machiavellianism poses problems for the DT literature. Concerns regarding the distinctiveness of psychopathy and Machiavellianism date back over 18 years to McHoskey et al. (1998), who argued that Machiavellianism, assessed with the most commonly used measure (MACH-IV; Christie & Geis, 1970), was a “global measure of psychopathy” (p. 205). There are several potential reasons for this overlap. The first is that the two constructs are conceptually indistinct. Glenn and Sellbom (2015) make this argument explicitly, noting that “the concept of Machiavellianism, or a manipulative and exploitative style, has been described as a central feature of psychopathy, starting with very early accounts” (p. 363), including Cleckley (1941). Many major assessments of psychopathy do include content related to Machiavellianism, such as Machiavellianism Egocentricity in the Psychopathic Personality Inventory-Revised (Lilienfeld & Widows, 2005); items such as “conning/manipulative,” “pathological lying,” and “glibness/superficial charm” in Hare’s Psychopathy Checklist-Revised (e.g., Hare, 2003); the Interpersonal Manipulation factor in Hare’s Self-Report Psychopathy Scale-III (e.g., Paulhus, Neumann, & Hare, in press); and a similar Manipulation scale in Lynam and colleagues’ (2011) Elemental Psychopathy Assessment. As proponents of Machiavellianism will note, however, Machiavellianism is thought to include aspects not seen in psychopathy, including a long-term, strategic focus, the ability to delay gratification, and average to good impulse control. For instance, Jones and Weiser (2014) cite research suggesting that Machiavellian individuals are “long term oriented . . . difficult to provoke into aggression . . . have intact executive functioning . . . and exert effort to resist cheating when short-term goals undermine long-term investments” (p. 20). This perspective would suggest that psychopathy and Machiavellianism overlap significantly in traits related to an antagonistic interpersonal approach, but they differ in regard to other central constructs, especially those related to impulse control. There are, however, problems with this suggestion. As currently measured, Machiavellianism and psychopathy share nearly identical Big Five/FFM profiles in which the meta-analytically derived correlations with the FFM do not differ significantly across the two DT constructs for any of the five domains. That is, both Machiavellianism and psychopathy are described primarily via negative correlations with Agreeableness and Conscientiousness (O’Boyle et al., 2015). Additionally, Vernon and colleagues (2008)

Miller, Hyatt, Maples-Keller, et al.

found substantial, negative genetic correlations with Conscientiousness for both Machiavellianism and psychopathy (rgs 5 –.43 and –.50, respectively). As such, empirical findings that indicate Machiavellian individuals describe themselves as impulsive, careless, undisciplined, lax, and unambitious run contrary to theoretical descriptions of Machiavellian individuals as being cautious, deliberate, and strategic. Overall, despite some differences in the theoretical descriptions of psychopathy and Machiavellianism, relatively few empirical differences have emerged. Importantly, many of the traits empirically associated with Machiavellianism appear to be antithetical to theoretical descriptions (e.g., impulsive, careless). Thus, the overlap may be the result of deficiencies in current measures of Machiavellianism that fail to assess the construct in a manner faithful to theoretical descriptions. In the following three studies, we address critical issues related to the degree to which psychopathy and Machiavellianism can be viewed as independent constructs versus a single construct, including whether extant measures of Machiavellianism align with expert conceptualizations. In Study 1, we administer multiple measures of each DT construct and examine their distinctiveness using confirmatory factor analyses, comparing a three-factor model in which each DT trait is modeled separately against a two-factor model in which psychopathy and Machiavellianism are modeled as a single factor. In addition to examining these basic factor models, we also compare the two- and three-factor models when accounting for method variance. We also examine the facet-level FFM personality profiles associated with the DT traits. While numerous studies have examined the domains of the FFM in relation to the DT, no studies have done so at the facet level examining all three DT constructs simultaneously. Of particular importance is whether psychopathy and Machiavellianism manifest similar relations with the impulsivity-related facets found across FFM domains (i.e., impulsiveness from Neuroticism, sensation seeking from Extraversion, self-discipline and deliberation from Conscientiousness; Whiteside & Lynam, 2001). We examine these DT–FFM profiles using composites of three DT measures and at the individual scale level for Machiavellianism. The latter test allows for an examination of whether the Machiavellianism–FFM profiles differ across any of the three most prominent operationalizations. In Study 2, we examine a broader and multisource nomological net for the DT and test the degree to which psychopathy and Machiavellianism manifest different relations with adverse developmental experiences, interpersonal attachment, self- and informant reports of FFM traits, and internalizing and externalizing symptoms/behaviors. This broader array of external criteria allows for a more comprehensive test of the ways in which Machiavellianism and psychopathy may differ. Importantly, the use of informant reports of the personality traits associated with these constructs provides a critical test of the theoretically incompatible link between Machiavellianism and low Conscientiousness by testing whether informants also view Machiavellian

Psychopathy and Machiavellianism

individuals as being rash, undisciplined, unambitious, and less competent. This is especially important as informant reports are based, in part, on behavioral observations of the participants and may be less prone to cognitive biases that may be associated with self-reports. In Study 3, we collect expert ratings of the FFM profile of the prototypical case of Machiavellianism. We then examine the degree of consensus across expert raters and compare the aggregated expert FFM profile of Machiavellianism with an expertgenerated FFM profile of psychopathy (Miller, Lynam, Widiger, & Leukefeld, 2001) and with the empirically derived FFM– Machiavellianism profiles reported in Study 1. These comparisons allow for an examination of the degree of conceptual overlap between Machiavellianism and psychopathy, as well as a comparison between the conceptual and empirical profiles of Machiavellianism.

3

Table 1 Interrelations Among DT Scales Psychopathy

Machiavellianism

Narcissism

SRP-III SD3 DD MACH-IV SD3 DD NPI-13 SD3 DD Psychopathy SRP-III SD3 DD Machiavellianism MACH-IV SD3 DD Narcissism NPI-13/NPI-40 SD3 DD

.82 .68

.65

.62 .60 .65

.57 .57 .58 .55 .62 .69

.63 .57

.62

.54 .40 .42

.52 .43 .42 .24 .43 .51

.37 .21 .33

.44 .43 .40 .29 .51 .57

.63 .48

.49

Note. SRP-III 5 Self-Report Psychopathy Scale–III; SD3 5 Short Dark Triad; DD 5 Dirty Dozen; NPI-13 5 Narcissistic Personality Inventory–13-item version; NPI-40 5 Narcissistic Personality Inventory–40-item version.

STUDY 1 Method Participants and Procedure. Participants included 393 adults (61% female; 85% Caucasian, 9% African American; Mage 5 35.4, SD 5 12.5) who participated via Amazon’s Mechanical Turk (MTurk) website. Individuals were compensated $1.00 for completion of the study. The various scales were given in a randomized order, and the study took, on average, 28 minutes to complete. Institutional review board approval was obtained for all aspects of the study. Measures. Dirty Dozen (DD). The DD (Jonason & Webster, 2010) is a 12-item measure with scales for narcissism (a 5 .89), psychopathy (a 5 .84), and Machiavellianism (a 5 .87). Short Dark Triad-3 (SD3). The SD3 (Jones & Paulhus, 2014) is a 27-item measure of narcissism (a 5 .78), psychopathy (a 5 .79), and Machiavellianism (a 5 .83). Narcissistic Personality Inventory-13 (NPI-13). The NPI-13 (Gentile et al., 2013) is a 13-item forced-choice selfreport measure of trait narcissism with subscales that measure Leadership/Authority, Grandiose Exhibitionism, and Entitlement/Exploitativeness. In the current study, the total score was used (a 5 .77). Self-Report Psychopathy Scale: Version III (SRPIII). The SRP-III (Paulhus et al., in press) is a 64-item measure of psychopathy with subscales of Interpersonal Manipulation, Callous Affect, Erratic Lifestyle, and Antisocial Behavior. In the current study, the total score was used (a 5 .94). (MACH-IV). The MACH-IV (Christie & Geis, 1970) is a 20-item measure of the personality trait of Machiavellianism (a 5 .83).

International Personality Item Pool NEO-120 (IPIP NEO-120). The IPIP-NEO-120 (Maples, Guan, Carter, & Miller, 2014) is a 120-item inventory of the FFM that measures the domains with 24 items each and facets with four items each. Alphas for the domains ranged from .85 (Openness) to .93 (Neuroticism, Conscientiousness). Alphas for the facets ranged from .69 (self-consciousness) to .93 (depression, deliberation), with a median of .80.

Results The correlations among the three sets of DT indicators are provided in Table 1. The median convergent correlations among the psychopathy, Machiavellianism, and narcissism scales were .68, .62, and .49, respectively.2 The median divergent correlations (i.e., with Machiavellianism and narcissism scales) for the psychopathy, Machiavellianism, and narcissism scales were .55, .56, and .43, respectively. To test whether Machiavellianism and psychopathy are redundant constructs, we estimated two basic confirmatory factor-analytic (CFA) models: (a) a three-factor model representing the traditional three-factor structure of the DT, and (b) a two-factor model that collapsed Machiavellianism and psychopathy into a single factor, along with the traditional narcissism factor. In addition to these models, we also estimated these models accounting for method variance (i.e., variability due to the trait-irrelevant idiosyncratic measurement artifacts) in order to rule out common method variance as a potential explanation for our findings. This was accomplished using the correlated traits correlated uniqueness (CTCU; see Lance, Noble, & Scullen, 2002) model approach. Both of the two CTCU models are the same, as their corresponding basic CFA described above, with the exception that the residuals for common measures are allowed to correlate (whereas they are set to zero in basic CFA). For both a two-factor model and a three-factor model, the

4

Miller, Hyatt, Maples-Keller, et al.

Table 2 Comparison of the Two- and Three-Factor Models for the Dark Triad Model Two factors Three factors CTCU two factors CTCU three factors

v2

df

TLI

CFI

SRMR

AIC

/M,P

/P,N

/M,N

251.00 211.14 132.51 97.94

26 24 17 15

.921 .928 .938 .949

.943 .952 .971 .979

.068 .066 .054 .052

1,637.53 1,601.66 1,537.37 1,506.46

— .891 — .897

.708 .694 .680 .672

— .677 — .649

Note. TLI 5 Tucker-Lewis Index; CFI 5 comparative fit index; SRMR 5 standardized root mean square residual; AIC 5 Akaike information criterion; CTCU 5 correlated traits correlated uniqueness; /M,P 5 interfactor correlation for Machiavellianism and psychopathy; /P,N 5 interfactor correlation for psychopathy and narcissism; /M,N 5 interfactor correlation for Machiavellianism and narcissism.

residuals associated with the DD scale were allowed to covary with one another. Similarly, residuals associated with the SD3 scale were allowed to covary with one another. Finally, the residuals associated with each of the measures unique to their specific trait (i.e., MACH-IV, SRP-III, and NPI-13) were allowed to covary. Results for the basic two- and three-factor CFA models suggested equivalent model-data fit for the two- and three-factor models. Although the three-factor model fit significantly better than the two-factor model according to the chi-square difference test, Dv2(2) 5 43.73, p < .001, this test statistic is known to result in high Type I error rates and is not recommended as an indication of model-data fit (see Joreskog & Sorbom, 1989). Therefore, we relied on alternative fit statistics for model comparisons, including the Tucker-Lewis index (TLI), the comparative fit index (CFI), and the standardized root mean residuals (SRMRs), according to commonly accepted guidelines (see Hu & Bentler, 1999). The TLI and CFI both compare the fit of the proposed model to the fit of the null model (i.e., no relation between the latent trait and indicators or interrelations between traits). Whereas TLI operationalizes the fit of these two models as the v2/df, the CFI considers (v2 – df). The SRMR is the average deviation of the correlation matrix predicted by model estimates from the observed correlation matrix. Finally, we consider the Akaike information criterion (AIC), which evaluates the likelihood of the model with a penalty for model complexity (i.e., number of parameters). Smaller values of the AIC indicate a better-fitting model. As shown in Table 2, both the two-factor and three-factor basic CFA models showed marginally good fit to the data, with TLIs and CFIs greater than .90, and SRMRs less than .08 (see Hu & Bentler, 1999). Importantly, both models were similar in their fit; the TLIs and CFIs for these models were different by .01, and the SRMRs were different by only .001, suggesting that the two models are equivalent. Although the AIC was generally smaller for the two-factor model, these differences were not large. Inspection of the percentage of variance explained by the latent variables revealed little difference between the two- and three-factor models. Whereas the two-factor model explained between 43% and 79% of the variance in the observed scores, the three-factor model explained between 43% and 83% of the variance. Taken together, these findings suggest there is little difference between the model proposing a differentiation between psychopathy and Machiavellianism and the model pro-

posing a common factor for these two traits. Similar findings were observed using the CTCU models; in fact, differences in fit were almost exactly the same as those observed for the basic CFA models. These findings suggest that common variance between methods did not substantially impact the basic CFAs. Further evidence to support the idea that Machiavellianism and psychopathy are redundant can be found in the interfactor correlations, /, of the three-factor model. The correlation between the Machiavellianism and psychopathy factors was .889, suggesting these variables share nearly 80% of their variance. Additionally, Machiavellianism and psychopathy showed nearly identical correlations with narcissism at .67 and .69, respectively. The CTCU variant of this model showed similar findings, with an interfactor correlation of .897 between Machiavellianism and psychopathy, and similar (though slightly smaller) correlations among other traits. Finally, factor loadings of Machiavellianism and psychopathy indicators showed roughly equivalent loadings across the two- and three-factor models, with differences ranging from absolute values of .058 to .014 (see Table 3). These differences were generally smaller than the standard errors of loadings, which ranged from .04 to .09 in the two-factor model and from .06 to .10 in the threefactor model. Differences in loadings were even smaller for the CTCU models, ranging from –.045 to .01, and were considerably lower than the standard errors of loadings in these models, which ranged from .07 to .12 in the three-factor CTCU model. The fact that the loadings for the indicators of Machiavellianism and psychopathy do not change across the two- and three-factor models provides evidence that the factor content is equivalent across the models, even when accounting for method variance. To examine further the similarity of the DT constructs, we created DT composites by summing z-scored values for the nine DT scales (three per DT construct). These composites were all significantly related: Machiavellianism–psychopathy (r 5 .78), Machiavellianism–narcissism (r 5 .58), psychopathy–narcissism (r 5 .58). We then correlated the three DT composites with the five domains and 30 facets of the FFM (see Table 4). Rather than focusing on statistically significant differences across the three DT scales with the 35 FFM scores (105 comparisons), we examined the absolute similarity of the DT’s FFM profiles using double-entry q-correlations (rICC; see McCrae, 2008, for a review). This approach assesses the absolute similarity in both the shape and elevations of the trait profiles and thus is a more stringent measure of agreement than one that focuses on shape

Psychopathy and Machiavellianism

5

Table 3 Comparison of Factor Loadings for the Two- and Three-Factor Models and Their Correlated Trait Correlated Uniqueness Model Variants Two-Factor Model Indicator MACH-IV SD3 Mach DD Mach SRP-III SD3 Psych DD Psych NPI-13 SD3 Narc DD Narc

Three-Factor Model

Psychopathy/Mach

Narcissism

Machiavellianism

Psychopathy

Narcissism

.708 (.721) .725 (.737) .774 (.720) .889 (.904) .859 (.881) .781 (.737) — — —

— — — — — — .816 (.826) .725 (.730) .656 (.631)

.750 (.756) .783 (.782) .802 (.749) — — — — — —

— — — .913 (.918) .881 (.891) .763 (.727) — — —

— — — — — — .816 (.827) .726 (.732) .655 (.625)

Dk –.042 –.058 –.028 –.014 –.022 .018 .000 –.001 .001

(–.035) (–.045) (–.029) (–.014) (–.010) (.010) (–.001) (–.002) (.006)

Note. Values in parentheses are for the correlated trait correlated uniqueness variant of the model. All loadings are standardized; Dk indicates the difference in factor loadings between the two models. Mach 5 Machiavellianism; SD3 5 Short Dark Triad; DD 5 Dirty Dozen; SRP-III 5 Self-Report Psychopathy Scale–III; Psych 5 psychopathy; NPI-13 5 Narcissistic Personality Inventory–13-item version; Narc 5 narcissism.

alone (e.g., simple r among the profiles, although this works well in most cases too; McCrae, 2008). The FFM profiles associated with Machiavellianism and psychopathy were essentially identical (rICC 5 .97), whereas both Machiavellianism and psychopathy manifested smaller profile similarities with narcissism (rICCs 5 .58 and .61, respectively). Next, we examined the correlations between the three individual Machiavellianism scales with the FFM domains and facets and compared their profiles with regard to absolute similarity (Table 5). These analyses were conducted to test whether the three Machiavellianism scales perform similarly or whether they manifest important differences, particularly with regard to theoretically relevant variables (e.g., Conscientiousness). In general, the three Machiavellianism scores manifested similar FFM profiles of low Agreeableness and Conscientiousness and high Neuroticism with absolute profile similarities (i.e., rICCs) of .84 (MACH-IV–SD3 Mach), .91 (MACH-IV–DD Mach), and .97 (SD3 Mach–DD Mach). Finally, we compared the FFM profiles of the three individual Machiavellianism scales to the FFM profile generated by the psychopathy composite in terms of their absolute similarity; all three Machiavellianism scales generated FFM profiles that were nearly identical to the profile created by the psychopathy composite, including MACH-IV (rICC 5 .91), SD3 Machiavellianism (rICC 5 .89), and DD Machiavellianism (rICC 5 .95).3

Discussion In the current study, multiple markers of the DT were used to examine their interrelations in a confirmatory factor-analytic approach and then tested in relation to the domains and facets of the FFM. A CFA of the three sets of DT indicators found that, although the three-factor model demonstrated a statistically significantly better fit than the two-factor model, both the two- and three-factor solutions demonstrated similar, mostly adequate fit to the data. In the three-factor model, the psychopathy and Machiavellianism factors were almost perfectly correlated with each other (r 5 .89) and demonstrated nearly identical correla-

tions with narcissism (rs 5 .67 and .69, respectively). The data indicate that, at the structural level, the two constructs, psychopathy and Machiavellianism, are barely distinguishable. Critically, the similarity of psychopathy and Machiavellianism was tested via their relations to underlying traits from the FFM, and Machiavellianism again proved to be virtually identical to psychopathy. The similarity found for the FFM facetbased profiles in Study 1 adds to the existing, much larger literature on the overlapping domain-level correlates of psychopathy and Machiavellianism (O’Boyle et al., 2015). The existing domain-level analyses suggest that both are strongly negatively correlated with Agreeableness and Conscientiousness, as well as moderately positively correlated with Neuroticism. The large, negative relation between Machiavellianism and Conscientiousness is particularly noteworthy and problematic given that it runs counter to theoretical accounts of Machiavellianism in which these individuals are said to put aside short-term goals and gratification in the pursuit of larger but delayed rewards. It is important to note that these findings are not driven by the nonimpulsive traits of Conscientiousness such as achievement striving, order, and dutifulness. The trait profiles of psychopathy and Machiavellianism are nearly identical, even when examining the more fine-grained facets; both are broadly related to impulsigenic traits such as impulsiveness, excitement seeking, selfdiscipline, and deliberation. In addition, this pattern of relations is not due to the use of a singular measure of Machiavellianism such as the MACH-IV, which has been used predominantly in DT studies, as all three measures manifested FFM profiles that were consistent with one another and that included positive relations with impulsivity-related traits. In sum, psychopathy and Machiavellianism yield identical personality profiles that are not contingent upon the specific measure of Machiavellianism used and include trait correlates that, although consistent with the extant literature (i.e., O’Boyle et al., 2015), run directly counter to the manner in which this construct is described. These findings are problematic for the literature on the DT in general and Machiavellianism more

6

Miller, Hyatt, Maples-Keller, et al.

Table 4 Correlations Between DT Composites and FFM Domains and Facets Psychopathy Neuroticism Anxiety Angry hostility Depression Self-consciousness Impulsiveness Vulnerability Extraversion Warmth Gregariousness Assertiveness Activity Excitement seeking Positive emotions Openness Fantasy Aesthetics Feelings Actions Ideas Values Agreeableness Trust Straightforwardness Altruism Compliance Modesty Tendermindedness Conscientiousness Competence Order Dutifulness Achievement striving Self-discipline Deliberation Profile similarities (rICCs) Machiavellianism Narcissism

Table 5 Correlations Between Individual Machiavellianism Scales and FFM Domains and Facets

Machiavellianism Narcissism

.26 .09 .39 .26 .02 .30 .10 .11 –.13 .23 .08 –.05 .47 –.19 –.09 .13 –.15 –.14 .06 –.20 –.01 –.80 –.34 –.76 –.56 –.69 –.36 –.42 –.51 –.29 –.31 –.61 –.29 –.31 –.51

.33 .23 .39 .30 .10 .33 .14 .06 –.17 .12 .07 .01 .36 –.18 .00 .17 –.06 .01 –.02 –.14 .04 –.71 –.41 –.77 –.46 –.56 –.29 –.29 –.44 –.24 –.25 –.52 –.22 –.32 –.45

.97* .61*

.58*

–.08 –.10 .09 –.08 –.22 .07 –.13 .50 .22 .47 .45 .23 .47 .20 .11 .25 .10 –.04 .16 .00 –.06 –.55 –.06 –.55 –.19 –.39 –.67 –.27 –.12 .06 –.08 –.30 .08 –.02 –.28

Note. DT 5 Dark Triad; FFM 5 Five-Factor Model. *p  .01. Correlations  .25 are boldfaced.

specifically. It is possible, however, that psychopathy and Machiavellianism differ on other aspects of their nomological networks outside of the basic traits that describe both (e.g., potential developmental precursors, externalizing behavioral correlates). Similarly, some of the overlap in trait profiles could be due to the reliance on mono-method (self-self) correlations between the DT constructs and the FFM. It will be useful to test whether informant-based FFM profiles of psychopathy and Machiavellianism also converge on a highly overlapping profile.

STUDY 2 In Study 2, we examine the DT constructs in relation to adverse developmental experiences (e.g., abuse, difficult parenting),

MACH-IV SD3 Mach DD Mach Psychopathy Neuroticism Anxiety Angry hostility Depression Self-consciousness Impulsiveness Vulnerability Extraversion Warmth Gregariousness Assertiveness Activity Excitement seeking Positive emotions Openness Fantasy Aesthetics Feelings Actions Ideas Values Agreeableness Trust Straightforwardness Altruism Compliance Modesty Tendermindedness Conscientiousness Competence Order Dutifulness Achievement striving Self-discipline Deliberation Profile similarities (rICCs) SD3 Mach DD Mach Psychopathy composite

.37 .25 .38 .33 .13 .35 .24 –.13 –.32 .01 –.05 –.12 .23 –.32 .05 .16 –.08 .06 .03 –.08 .11 –.64 –.51 –.60 –.50 –.47 –.15 –.27 –.47 –.37 –.28 –.50 –.27 –.36 –.38

.21 .17 .30 .20 .03 .20 .01 .15 –.07 .12 .16 .09 .34 –.05 –.03 .16 .00 –.02 –.07 –.15 –.02 –.57 –.26 –.62 –.34 –.45 –.29 –.27 –.28 –.09 –.15 –.35 –.11 –.23 –.33

.28 .18 .34 .25 .10 .30 .11 .13 –.04 .16 .08 .04 .36 –.10 –.03 .12 –.08 –.01 –.02 –.13 .00 –.62 –.28 –.76 –.35 –.51 –.30 –.22 –.39 –.18 –.22 –.49 –.18 –.23 –.46

.84* .91* .91*

.97* .89*

.95*

.26 .09 .39 .26 .02 .30 .10 .11 –.13 .23 .08 –.05 .47 –.19 –.09 .13 –.15 –.14 .06 –.20 –.01 –.80 –.34 –.76 –.56 –.69 –.36 –.42 –.51 –.29 –.31 –.61 –.29 –.31 –.51

Note. FFM 5 Five-Factor Model; SD3 5 Short Dark Triad; Mach 5 Machiavellianism; DD 5 Dirty Dozen. *p < .01. Correlations  .25 are boldfaced.

attachment styles, self- and informant-rated personality domains, and internalizing and externalizing outcomes (e.g., depression, hostility, substance use, antisocial behavior). Although we examine all three DT constructs, we focus most explicitly on the degree to which the empirical profiles for Machiavellianism and psychopathy differ across these important external criteria.

Method Participants and Procedures. Participants were 341 students (218 women and 121 men; 2 unknown); 271 participants were White, 35 were Asian, 29 were Black, and 9 were of

Psychopathy and Machiavellianism

Hispanic ethnicity. Mean age was 19.3 (SD 5 2.2). Participants were undergraduates recruited from a research participant pool at a large, public southeastern university who received research credit for their participation. A fixed administration order was used in this study. Institutional review board approval was obtained for all aspects of the study. After providing informed consent, participants completed questionnaires and provided email addresses for peers. Emails were sent to a peer identified by the participant as someone who knew the participant well to see if he or she was willing to serve as a peer informant in this study. Interested peers provided informed consent and then completed informant versions of the NEO Five-Factor Inventory (Costa & McCrae, 1992), Dirty Dozen (Jonason & Webster, 2010), and Short Dark Triad (Jones & Paulhus, 2014). We compared the participants for whom we received informant reports (n 5163 for FFM domains; n 5 178 for Dark Triad composites) with those we did not with regard to demographic variables and the self-reported DT traits to determine whether the groups were different in any meaningful way; no significant differences were found between the groups. Measures. NEO Five-Factor Inventory (NEO-FFI). The NEO-FFI (Costa & McCrae, 1992) is a 60-item measure of the FFM. Both participants and informants completed the NEO-FFI. Alphas for the self-report domains ranged from .71 (Openness) to .86 (Neuroticism, Conscientiousness) and from .70 (Openness) to .89 (Conscientiousness) for the informant reports. Self-Reported Measures. Self-Report Psychopathy Scale– III (SRP-III) The SRP-III (Paulhus et al., in press) is a 64-item self-report measure of psychopathy that yields a total score (a 5 .93) and four subscales: Interpersonal Manipulation, Callous Affect, Erratic Lifestyle, and Antisocial Behaviors. Only the total scale was used here. MACH-IV. The MACH-IV (Christie & Geis, 1970) is a 20-item measure of the personality trait of Machiavellianism (a 5 .81). Narcissistic Personality Inventory (NPI). The NPI (Raskin & Terry, 1988) is a 40-item forced-choice self-report measure of trait narcissism. In the current study, only the total score was used (a 5 .82). Experiences in Close Relationships-Revised (ECR-R). The ECR-R (Fraley, Waller, & Brennan, 2000) is a 36-item self-report measure of two adult attachment styles: avoidance (18 items; a 5 .93) and anxiety (19 items; a 5 .94). Child Abuse and Trauma Scale (CATS). The CATS (Saunders & Giolas, 1991) is a 38-item self-report measure of physical, verbal, emotional, and sexual abuse. In the current study, we used only the items included in the revised scales described by Poythress et al. (2006).

7

All 14 items were also used to create a total scale (a 5 .89). The total score was log-transformed in order to reduce problems with non-normality. Parenting Warmth and Monitoring Scale. This 24-item self-report scale (Lamborn, Mounts, Steinberg, & Dornbusch, 1991) measures the degree of warmth and parental supervision given to children. In the current study, questions pertaining to parental monitoring were asked for the time frame of 12th grade (warmth: a 5 .83; monitoring: a 5 .78). Psychological Control Scale (PCS). The PCS (Barber, 1996) is a 16-item self-report measure of the level of psychological control or intrusiveness asserted by one’s mother and father. In the current study, participants were asked about their parents’ behavior in this domain when they were a senior in high school (a 5 .85). Brief Symptom Inventory (BSI). The BSI (Derogatis & Melisaratos, 1983) is a 53-item measure of psychological symptoms experienced during the past week that includes nine specific symptom scales (e.g., Interpersonal Sensitivity, Anxiety, Hostility) and a global severity index (GSI). Alphas ranged from .65 (anxiety) to .96 (GSI), with a median of .76. Crime and Analogous Behavior Scale (CAB). The CAB (Miller & Lynam, 2003) is a self-report inventory that assesses a variety of externalizing behaviors, including substance use, antisocial behavior, gambling, sexual behavior, and intimate partner violence. A lifetime substance use variety count was created by giving participants a 1 for every substance they endorsed using (five items; e.g., marijuana). A lifetime antisocial behavior count was created by giving participants a 1 for every relevant act they endorsed (nine items; e.g., stealing). A lifetime gambling (GAMB) count was created by giving participants a 1 for every relevant act they endorsed (six items; e.g., played card or other games for money). Finally, one question was used to inquire about the number of lifetime sexual partners. The antisocial and number of sexual partners variables were logtransformed in order to reduce problems with non-normality. Informant-Completed Measures. Dirty Dozen (DD). The DD (Jonason & Webster, 2010) is a 12item measure with scales for narcissism (a 5 .89), psychopathy (a 5 .82), and Machiavellianism (a 5 .86). In the current study, informants completed the DD. Short Dark Triad (SD3). The SD3 (Jones & Paulhus, 2014) is a 27-item measure of narcissism (a 5 .63), psychopathy (a 5 .78), and Machiavellianism (a 5 .68); the current study used the original SD3 that was later modified in relatively minor ways before it was published. In the current study, informants completed the SD3. To create informant report scores for the Dark Triad, the psychopathy, narcissism, and Machiavellianism scores from the

8

Miller, Hyatt, Maples-Keller, et al.

Table 6 Dark Triad and Attachment, Adverse Developmental Experiences, and Traits

Attachment style Anxious Avoidant Abuse Parenting style Parental warmth Parental monitor Psychological control Traits Self-Report FFM Neuroticism Extraversion Openness Agreeableness Conscientiousness Peer Report FFM Neuroticism Extraversion Openness Agreeableness Conscientiousness Dark Triad Peer Report Psychopathy Machiavellianism Narcissism

Psychopathy

Machiavellianism

Narcissism

.13 .21 .24

.37 .23 .25

–.05 –.10 .03

–.27 –.42 .26

–.20 –.25 .29

–.07 –.11 .09

.13 –.20 .04 –.60 –.48

.35 –.33 –.01 –.61 –.43

–.17 .21 .03 –.27 –.01

–.07 –.16 .06 –.30 –.24

.19 –.32 .07 –.40 –.29

–.18 .16 .02 –.04 –.01

.29 .18 .07

.34 .33 .13

.06 .07 .17

Note. FFM 5 Five-Factor Model. Correlations  .25 are boldfaced.

DD and SD3 were z-scored and then summed to form composites. The scales from the two measures demonstrated moderate convergence: psychopathy (r 5 .46), narcissism (r 5 .41), and Machiavellianism (r 5 .51).

Results The correlations among the three self-reported DT scales ranged from .22 (Machiavellianism–narcissism) to .40 (narcissism–psychopathy) to .54 (Machiavellianism–psychopathy). We next correlated the three DT scales with an array of relevant criteria, including developmental experiences such as parenting and abuse, attachment style, and self- and informant-reported personality traits from the FFM, as well as peer-reported DT scores (see Table 6) and self-reported scores on internalizing symptoms and externalizing behaviors (see Table 7). Rather than focusing on statistically significant differences across the correlations manifested by the three DT scales with the 35 criteria (105 comparisons), we again quantified the absolute similarity of the DT’s profiles with the 35 criteria using rICCs. The empirical profiles associated with Machiavellianism and psychopathy were highly similar in absolute terms (rICC 5 .89), whereas both Machiavellianism and psychopathy manifested much smaller profile similarities with narcissism (rICCs 5 .03 and .27, respectively).

Table 7 Dark Triad’s Relations With Internalizing Symptoms and Externalizing Behaviors Psychopathy Machiavellianism Narcissism Internalizing Somatization Obsessive-compulsive Interpersonal sensitivity Depression Anxiety Hostility Phobic anxiety Paranoia Psychoticism GSI Externalizing Substance abuse Antisocial behavior Gambling Risky sex Poached another partner Been poached Sex partners

.15 .25 .28 .25 .22 .22 .35 .15 .24 .26

.33 .32 .34 .35 .33 .33 .35 .29 .37 .38

–.07 .02 .00 –.01 –.05 –.01 .08 –.08 –.04 –.02

.30 .29 .25

.18 .04 .06

.22 .06 .08

.34 .36 .27

.27 .31 .13

.23 .20 .07

Note. GSI 5 global severity index. Correlations  .25 are boldfaced.

Across 35 sets of correlations, Machiavellianism and psychopathy manifested relations with the various criteria that were quite similar; both were significantly negatively related to selfand informant-reported Agreeableness and Conscientiousness and to self- and informant reports of psychopathy and Machiavellianism, both were positively related to self-reported abuse and the receipt of negative parenting (i.e., low warmth and monitoring, high control), and both manifested small to moderate positive correlations with internalizing symptoms and externalizing behaviors. The largest differences were for anxious attachment and selfand peer-reported Neuroticism; Machiavellianism was more strongly related to these criteria than was psychopathy.

Discussion The data from Study 2 add additional evidence to the question of whether psychopathy and Machiavellianism are meaningfully distinct traits. First, using a broader and more comprehensive set of criteria, psychopathy and Machiavellianism again manifested highly similar empirical profiles across 35 criteria (rICC 5 .89), including similar relations with adverse developmental experiences (i.e., abuse, difficult/poorer parenting), attachment styles, personality, psychological distress, substance use, and riskier sexual behavior. Second, the substantial overlap in personality profiles associated with psychopathy and Machiavellianism was largely replicated when using informant reports. Most importantly, informants described psychopathic and Machiavellian participants as antagonistic and disinhibited—the latter trait description being discrepant from classic descriptions of Machiavellianism in which these individuals are thought to

Psychopathy and Machiavellianism

behave in a planful and deliberate manner, manipulating others for their own longer-term advantage. In sum, the current study adds important data demonstrating that psychopathy and Machiavellianism, for all intents and purposes, are redundant constructs, at least as currently measured. This is problematic in that the two are not distinct enough to treat as separable constructs and because the basic profile associated with Machiavellianism appears to be inconsistent with how the construct is described initially and in modern writings. A question remains as to whether the near-perfect overlap between the empirical profiles associated with psychopathy and Machiavellianism represents shortcomings in the measurement of the construct or true overlap in the conceptualization. In the former case, better measures of Machiavellianism might reveal larger differences with psychopathy. In the latter case, this is simply an instance of the jangle fallacy (Kelley, 1927), in which two nominally distinct constructs actually represent the same thing.

9

2006). A rating of 1 indicated that the prototypical Machiavellian individual would be “extremely low” on a given trait, and a rating of 5 indicated that the prototypical Machiavellian individual would be “extremely high” on a given trait. Raters participated by clicking on an embedded link; upon activating this link, the individual was asked a series of demographic questions and then completed ratings of the 30 FFM facets with regard to the prototypical Machiavellian individual. Participants. Responses were obtained from 43 raters; however, seven indicated that they had not published any articles on the DT, so their responses were not included, leaving a usable sample of 36 experts. On average, expert raters were 40.43 years old (SD 5 12.86), received their PhD 10.5 years ago (SD 5 11.00), and published 6.6 publications (SD 5 9.12) on the DT and related traits.

Results STUDY 3 From a theoretical perspective, one would expect some degree of overlap between psychopathy and Machiavellianism, as traits from the latter (e.g., manipulativeness) are included in descriptions of the former (e.g., Glenn & Sellbom, 2015). Empirically, however, data from the previous studies suggest that, as currently measured, psychopathy and Machiavellianism are effectively identical. The question that remains is whether this overlap is an accurate reflection of the two constructs or an artifact of their assessment. The latter is a possibility given that the empirical trait profile associated with Machiavellianism, which includes evidence of disinhibition across personality domains (e.g., impulsive urgency, excitement seeking, low self-discipline and deliberation), appears to be inconsistent with theoretical writings that emphasize normal to superior impulse control. In order to better understand the expected trait profile of Machiavellianism, we asked experts (i.e., those who have published multiple articles on the DT) to rate the prototypical Machiavellian individual on the 30 facets of the FFM. These data allow for the development of a consensus profile that quantitatively describes experts’ conceptualization of Machiavellianism that can be compared to a similar expert profile for psychopathy (i.e., Miller, Lynam, Widiger, & Leukefeld, 2001), as well as to the empirical profiles found in Study 1.

Method Procedure. Emails were sent to individuals (N 5 261; 28 emails bounced back, and we were not able to find alternative working emails) who had authored two or more peer-reviewed articles on the Dark Triad asking them to complete a brief survey in which they rated the prototypical Machiavellian individual on the 30 facets of the FFM using the Five-Factor Model Rating Form (Mullins-Sweatt, Jamerson, Samuel, Olson, & Widiger,

Before proceeding with the central analyses, we calculated three indices of agreement among raters in order to ensure that raters were working with similar conceptions. The first index, withingroup reliability (i.e., rwg), indexes the agreement among a set of judges rating a single target on a single item via proportional reduction in error variance (LeBreton, James, & Lindell, 2005); the average rwg across FFM items was .57 for the Machiavellian profile. The second index was an intraclass correlation coefficient calculated using a two-way random model based on absolute agreement; the ICC was .97. The final index was an average corrected item-total correlation (CITC) from an analysis that treated items as cases and raters as variables; thus, each individual rater’s profile was correlated with the average profile of the other raters. The CITC ranged from .31 to .90, with a mean of .68. In general, there was good agreement among raters on the FFM profile associated with the prototypical case of Machiavellianism. The expert-rated profile for Machiavellianism is reported in Table 8, along with the previously generated expert-rated profile for psychopathy (Miller et al., 2001); differences across these two profiles are quantified at the facet level via Cohen’s d effect sizes. Also presented are the Machiavellianism–FFM profiles from Study 1; the similarities of these profiles were examined in relation to the expert-rated profiles for Machiavellianism and psychopathy. In general, experts described the prototypical Machiavellian individual as being relatively uniformly low on traits from Agreeableness (e.g., straightforwardness, altruism, tendermindedness), low on two traits from Neuroticism (i.e., self-consciousness, vulnerability), and high on assertiveness from the domain of Extraversion. Overall, this profile was significantly correlated with the expert profile for psychopathy (r 5 .54). The primary differences between the expert ratings for Machiavellianism and psychopathy, as demonstrated by Cohen’s ds, were found for traits related to impulse control such that the prototypical Machiavellian individual was seen as being

10

Miller, Hyatt, Maples-Keller, et al.

Table 8 Expert Ratings of a Prototypical Case of Machiavellianism on the FFM

Neuroticism Anxiety Angry hostility Depression Self-consciousness Impulsiveness Vulnerability Extraversion Warmth Gregariousness Assertiveness Activity Excitement seeking Positive emotions Openness Fantasy Aesthetics Feelings Actions Ideas Values Agreeableness Trust Straightforwardness Altruism Compliance Modesty Tendermindedness Conscientiousness Competence Order Dutifulness Achievement striving Self-discipline Deliberation Profile similarities Psychopathy expert MACH-IV SD3 Mach DD Mach Mach composite

Mach Expert

Psychopathy Expert

M

SD

M

SD

d

MACH-IV

SD3 Mach

DD Mach

Mach Comp

2.39 3.28 2.94 1.92 2.08 1.92

1.10 1.00 0.98 1.05 0.94 0.87

1.47 3.87 1.40 1.07 4.53 1.47

0.52 0.64 0.51 0.26 0.74 0.52

1.07 20.70 1.97 1.11 22.90 0.63

.23 .39 .30 .10 .33 .14

.25 .38 .33 .13 .35 .24

.17 .30 .20 .03 .20 .01

.18 .34 .25 .10 .30 .11

2.06 3.39 4.14 3.78 2.81 2.72

1.07 0.99 0.87 0.93 1.01 0.74

1.73 3.67 4.47 3.67 4.73 2.53

1.10 0.62 0.52 0.98 0.46 0.92

0.30 20.34 20.46 0.12 22.45 0.23

–.17 .12 .07 .01 .36 –.18

–.32 .01 –.05 –.12 .23 –.32

–.07 .12 .16 .09 .34 –.05

–.04 .16 .08 .04 .36 –.10

2.28 2.77 3.31 2.94 2.78 3.03

1.09 0.84 1.12 0.92 1.02 0.91

3.07 2.33 1.80 4.27 3.53 2.87

0.88 0.62 0.86 0.59 1.10 0.99

20.80 0.60 1.51 21.72 20.70 0.17

.00 –.06 .01 –.02 –.14 .04

.05 –.08 .06 .03 –.08 .11

–.03 .00 –.02 –.07 –.15 –.02

–.03 –.08 –.01 –.02 –.13 .00

1.42 1.28 1.28 2.08 1.89 1.36

0.81 0.94 0.51 0.65 0.71 0.54

1.73 1.13 1.33 1.33 1.00 1.27

0.80 0.35 0.62 0.49 0.00 0.46

20.39 0.21 20.09 1.30 1.77 0.18

–.41 –.77 –.46 –.56 –.29 –.29

–.51 –.60 –.50 –.47 –.15 –.27

–.26 –.62 –.34 –.45 –.29 –.27

–.28 –.76 –.35 –.51 –.30 –.22

3.69 3.97 2.53 3.86 3.42 3.78

0.89 0.84 1.08 0.87 1.11 0.90

4.20 2.60 1.20 3.07 1.87 1.60

1.00 0.51 0.78 1.20 0.83 1.10

20.54 1.97 1.41 0.75 1.58 2.17

–.24 –.25 –.52 –.22 –.32 –.45

–.37 –.28 –.50 –.27 –.36 –.38

–.09 –.15 –.35 –.11 –.23 –.33

–.18 –.22 –.49 –.18 –.23 –.46

.54* .16 .39 .26 .27

Empirical FFM-Mach Profiles (Study1)

.39 .60* .54* .52*

Note. Expert ratings for Machiavellianism and psychopathy: 4 or higher 5 boldfaced; 2 or lower 5 underlined. Machiavellianism–FFM facets are from Study 1 (Tables (4 and 5)). Psychopathy expert ratings are taken from Miller, Lynam, Widiger, and Leukefeld (2001). Mach Comp 5 Machiavellianism composite from Study 1; FFM 5 Five-Factor Model; Mach 5 Machiavellianism; SD3 5 Short Dark Triad; DD 5 Dirty Dozen. Empirical profiles for the Machiavellianism composite and individual scales represent bivariate correlations with the FFM domains and traits. The similarity indices represent the correlation of the expert ratings (Machiavellianism, psychopathy) with each other (i.e., r 5.54) as well as with the correlational profiles reported in the last four columns of the table (i.e., these measure relative rather than absolute similarity, unlike the rICC reported earlier, because the expert ratings and correlational profiles are on different metrics).

lower in impulsiveness (d 5 22.90), sensation seeking (d 5 22.45), and Openness to actions (d 5 21.72) and higher in self-discipline (d 5 1.58), order (d 5 1.97), and dutifulness (d 5 1.41). Averaging across the six facets within each FFM domain, the largest differences between the two expert-rated profiles were for Conscientiousness (mean d 5 1.22), followed by Agreeableness (mean d 5 .50) and Extraversion (mean

d 5 –.43) such that the prototypical Machiavellian individual was seen as being substantially more conscientious, somewhat more agreeable, and somewhat less extraverted than the prototypical psychopathic individual. Profile analyses, in which the FFM–Machiavellianism correlations derived in Study 1 were examined in relation to the expert profiles of Machiavellianism and psychopathy,

Psychopathy and Machiavellianism

demonstrated that none of the three individual Machiavellianism measures or the overall Machiavellianism composite was very similar to the expert-rated Machiavellianism profile (rs ranged from .16 to .39). In every case, the empirical profiles for the individual Machiavellianism measures and the composite were more similar to the expert-rated psychopathy profile (rs ranged from .39 to .60) than to the expert profile for Machiavellianism.4

Discussion As expected, the expert-rated profile for Machiavellianism overlapped significantly with the expert-rated profile for psychopathy, particularly with regard to their shared undergirding by traits from the domain of Agreeableness/Antagonism. However, the expert-rated profiles differed substantially with regard to impulsivity-related traits, including impulsiveness (Neuroticism), excitement seeking (Extraversion), self-discipline (Conscientiousness), and deliberation (Conscientiousness). These differences are consistent with early and recent writings on Machiavellianism; for instance, Furnham, Richards, & Paulhus (2013) state that “although as malevolent as psychopaths, Machiavellians are more cautious and deliberate in their behavior: Hence, they do not act on temptation like psychopaths” (p. 208). However, these conceptual descriptions differ dramatically from the empirical findings in which Machiavellianism was consistently positively correlated with impulsivity-related traits (e.g., O’Boyle et al., 2015; Paulhus & Williams, 2002) and a fast and impulsive life history approach (e.g., McDonald, Donnellan, & Navarrete, 2012). In fact, the most commonly used measures of Machiavellianism yield personality profiles that are more similar to expert conceptualizations of psychopathy than they are to expert conceptualizations of Machiavellianism. The current results raise substantial concerns that the empirical literature on Machiavellianism, studied alone and as part of the DT, is of questionable validity, particularly in relation to certain core criteria and outcomes. Echoing McHoskey and colleagues (1998), we suggest that existing measures of Machiavellianism are functioning as proxy measures of psychopathy. Given the consensus among experts that Machiavellianism should be associated with strategic decision making, caution, deliberation, persistence, ambition, and organization, why then are scores on current inventories negatively associated with these traits? We believe part of the answer lies in the fact that Agreeableness and Conscientiousness are not orthogonal. In a large meta-analysis (k 5 212; total N 5 144,117) of the intercorrelations among the Big Five, Van der Linden, te Nijenhuis, and Bakker (2010) found that Agreeableness and Conscientiousness were substantially correlated at q 5 .43. Examinations of the hierarchical nature of personality (e.g., Digman, 1997; Markon, Krueger, & Watson, 2005) demonstrate that these two factors combine to form higher-level factors often called constraint/disinhibition in three-factor models and alpha in two-factor models (pairing here with Neuroticism). As a result, even though most Machiavellianism scales do not include content that is directly

11

relevant to impulse control problems, the assessment of low Agreeableness carries with it lower scores on Conscientiousness. Even when items appear to pull for high levels of Conscientiousness, they may not work in that fashion. For instance, the SD3 Machiavellianism scale includes an item that states, “You should wait for the right time to get back at people.” In terms of face validity, this item would appear to capture both low Agreeableness (i.e., seeking revenge) and high Conscientiousness (i.e., waiting for the right time). Empirically, however, this item is driven by the low Agreeableness component—so much so, in fact, that it captures low rather than high Conscientiousness. In Study 1, this item was substantially negatively related to both Agreeableness (r 5 –.54) and Conscientiousness (r 5 –.33). In addition to obliquely illustrating the difficulty inherent in double-barreled items, this example highlights the difficulty of creating items that simultaneously pull for low Agreeableness and high Conscientiousness—a state in opposition to how these trait domains typically behave. What is to be done? The first order of business is the development of better measures of Machiavellianism. Such measures need to capture the constraint present in the theoretical conceptualization of Machiavellianism, which should allow these measures to be distinguished from psychopathy. It is possible that such a measure may be created by writing better double-barreled items that can simultaneously assess low levels of Agreeableness and high levels of Conscientiousness (e.g., when I seek revenge, I do so planfully and with deliberation), but this approach is fraught. Double-barreled items are typically to be avoided, as they are often confusing and one is never sure to which barrel a respondent is responding (Clark & Watson, 1995). An alternative assessment approach is to assess these diverse aspects separately and combine them in the total score. Similar approaches have been used to assess the personality disorders and psychopathy (Lynam & Widiger, 2001; Miller, Bagby, Pilkonis, Reynolds, & Lynam, 2005; Miller et al., 2001). Specifically, expert ratings are used to identify the FFM traits that are most characteristic of a given construct, and individuals’ scores on those traits are then used to generate a score on the construct.

GENERAL DISCUSSION Since the Dark Triad was first described in 2002, a rapidly growing empirical literature has emerged (Furnham et al., 2013) in which psychopathy, narcissism, and Machiavellianism are measured simultaneously in relation to an array of potentially relevant criteria, including basic personality traits (O’Boyle et al., 2015), sexual strategies (McDonald et al., 2012), workplace behavior (O’Boyle et al., 2012), financial decision making (e.g., Jones, 2013), and antisocial tendencies (Pailing et al., 2014). Significant efforts have been aimed at describing the traits that account for the overlap among these constructs; contenders include Agreeableness from the FFM (Miller et al., 2010; Paulhus & Williams, 2002), Honesty-Humility from the

12

HEXACO model of personality (Lee & Ashton, 2005), and a narrower subset of traits found in these dimensions (i.e., callousmanipulation; Furnham et al., 2013). Much of the interest in this area, however, comes from searching for the ways in which the DT differs in relation to various phenomena, including outcomes such as mating, cheating, and antisocial behavior. Unfortunately, the current data suggest that this search may have gone astray with regard to Machiavellianism. Current assessments do not capture the theoretical aspects of Machiavellianism related to the presence of normal to above average impulse control and the ability to delay gratification that should distinguish Machiavellianism from psychopathy. Rather, as currently assessed, Machiavellian individuals are described as being impulsive, distractible, irresponsible, excitement seeking, rash, impersistent, unambitious, and disorganized—traits that experts believe characterize psychopathy but not Machiavellianism. We believe the current results cast substantial doubt on the validity of the extant empirical literature on Machiavellianism. We are in agreement with McHoskey and colleagues (1998) in suggesting that existing measures of Machiavellianism are actually measuring psychopathy and that the extant literature on Machiavellianism is better framed as an alternative literature on psychopathy. It is not that the existing literature is unimportant or irrelevant, but rather that it is more relevant to our understanding of psychopathy than Machiavellianism. Skeptics of this argument will likely point to the fact that the DT literature suggests that the two do not always work in the same way. We believe there are a few arguments that address this issue. First, psychopathy measures in general are not perfectly correlated with one another (e.g., Gaughan, Miller, Pryor, & Lynam, 2009), and thus some differences with regard to external correlates are to be expected. Second, the existing DT literature is characterized by two statistical/methodological problems that may, in part, explain divergences sometimes found between the two. The first represents the failure to test the DT’s relations with dependent variables against one another (e.g., test of dependent rs, test of semipartial correlations) so as to examine whether these differences (e.g., psychopathy’s correlation with outcome A vs. Machiavellianism’s correlation with outcome A) are statistically significant. A substantial number of articles on the DT fail to report the interrelations among the three constructs, however, and thus testing of this nature is not even possible after the fact. In much of this literature, the only significance testing that is done is whether each individual DT–outcome relation differs significantly from zero but not whether these relations differ from one another across the three DT scales. As such, researchers are unable to comment on whether the relations manifested by the DT constructs and various outcomes differ significantly from one another, only that they differ in their difference from zero. The second major problem deals with the multivariate approaches (e.g., partialing constructs from one another in order to examine their “unique” contributions) that are frequently advocated and used in the DT literature (e.g., Furnham et al., 2013). However, Lynam, Hoyle, and Newman (2006) note that there are important dangers (i.e., “perils of partialling”) inherent in this

Miller, Hyatt, Maples-Keller, et al.

approach in which substantial overlapping variance is removed (i.e., core of Agreeableness or Honesty-Humility) prior to examining the unique components of each construct. One issue, besides the problem of having another entire set of correlations from which to look for statistical significance (increasing the possibility of Type I error), is that one does not know what partialed constructs represent once substantial variance has been removed. For instance, what does Machiavellianism represent once the shared traits from psychopathy and narcissism have been removed? In Study 1, the correlation between latent Machiavellianism and latent psychopathy was .889 in the three-factor model; what can be left in Machiavellianism after removing almost 80% of its variation? While it is difficult to conceptualize what remains, one can be certain that it is not the same construct as the construct with which one began. Glenn and Sellbom (2015) examined this issue, for instance, and demonstrated that the DT does not contain very much unique personality content once the variance shared with psychopathy is removed. As such, in much of the DT literature, it is impossible to know what the partial correlations represent, especially for the most overlapping constructs—psychopathy and Machiavellianism. Moving forward, we believe further research on the DT should await the development of new or refined measures of Machiavellianism that successfully capture the underlying construct as it is conceived of by experts and theoretical writings.5 As currently assessed, Machiavellianism is effectively the same as psychopathy, and it differs substantially in important ways from experts’ conceptualizations, particularly with regard to critical traits related to the ability to behave in a manner that emphasizes long-term strategic thinking and behavior.

LIMITATIONS AND FUTURE DIRECTIONS The current studies are not without important limitations, of course, including the primary reliance on data collected using undergraduate and MTurk samples, as is the norm for this subject area. Future research might benefit from testing these issues in alternative populations (e.g., community samples, offender samples, occupational samples) that might have more diversity with regard to age, racial and ethnic status, and education, as well as levels of DT traits. In addition, while we used multiple measures of Machiavellianism (Study 1) and informants (Study 2), many of the relations were limited to mono-method approaches and should be supplemented with data collected from alternative perspectives. Finally, we did not examine the DT constructs in relation to all of the outcome variables that have been highlighted in the existing literature—such as financial (mis)behavior, decision making, aggression, and ego control—and the array of outcomes related to evolutionary perspectives that are commonly examined in this literature. We believe these data provide support for (a) the reinterpretation of existing data on Machiavellianism such that it is interpreted as informing the literature on psychopathy and (b) the

Psychopathy and Machiavellianism

need for careful thought and consideration of how to best measure Machiavellianism in a manner that is consistent with theoretical descriptions. Without such a development, continued research efforts aimed at understanding the core traits and mechanisms that link these three important and interrelated constructs as well as the outcomes that may be unique to each will be of limited validity and utility.

Declaration of Conflicting Interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding The authors received no financial support for the research, authorship, and/or publication of this article. Notes 1. According to PsycINFO, since the coining of the phrase Dark Triad, 138 peer-reviewed articles have been published with Dark Triad in the title. 2. Two items were removed from the SRP-III prior to the confirmatory factor-analytic models and before it was added to the psychopathy composite, as it shares those two items with the SD3 Psychopathy subscale. They were removed from the SRP-III rather than the SD3 simply because the former is substantially longer than the latter. 3. We also examined a reduced, item response theory (IRT)-based variant of the MACH-IV created by Rauthmann (2013; trimmed MACH), which produced an FFM profile that was very similar to the original MACH-IV (r 5 .96), SD3 Mach (r 5 .97), and DD Mach (r 5 .90). More importantly, its FFM profile still did not align with the expert profile (r 5 .25) due to its problematic negative correlations with Conscientiousness facets (e.g., deliberation: r 5 –.36). 4. Results were similar when MACH-IV subscales were used; that is, none of the subscales manifested more convergence with the expert profile than others, and none was substantially associated with this expert-rated profile (i.e., rs ranged from .14 to .18 for the three subscales). Similarly, no single MACH-IV subscale was substantially more strongly responsible for the overlap with psychopathy. In sum, there is no clear “driver” of the overlap between psychopathy and Machiavellianism, even when the narrower facets were used, and none of the MACH-IV subscales produces a trait profile that is more consistent with another with the expert-rated MACH profile. 5. In this article, we have assumed that Machiavellianism and psychopathy both exist and should be more differentiable than they are with current measures. An alternative hypothesis, raised by an anonymous reviewer, is that Machiavellianism could represent a folk concept that does not exist in nature in the same way that psychopathy or narcissism might be said to exist. We believe, however, that it is too early to make that determination until further efforts are made to assess Machiavellianism in a manner that is more faithful to its theoretical conception.

13

References Barber, B. K. (1996). Parental psychological control: Revisiting a neglected construct. Child Development, 67, 3296–3319. Clark, L. A., & Watson, D. (1995). Constructing validity: Basic issues in objective scale development. Psychological Assessment, 7, 309–319. Cleckley, H. (1941). The mask of sanity: An attempt to reinterpret the so-called psychopathic personality. Oxford, UK: Mosby. Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEOFFI) professional manual. Lutz, FL: Psychological Assessment Resources. Christie, R., & Geis, F. (1970) Studies in Machiavellianism. New York: Academic Press. Derogatis, L. R, & Melisaratos, N. (1983). The Brief Symptom Inventory: An introductory report. Psychological Medicine, 13, 595–605. DeShong, H. L., Grant, D. M., & Mullins-Sweatt, S. N. (2015). Comparing models of counterproductive workplace behaviors: The five-factor model and the Dark Triad. Personality and Individual Differences, 74, 55–60. Digman, J. M. (1997). Higher-order factors of the Big Five. Journal of Personality and Social Psychology, 73, 1246–1256. Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysis of self-report measures of adult attachment. Journal of Personality and Social Psychology, 78, 350–365. Furnham, A., Richards, S. C., & Paulhus, D. L. (2013). The Dark Triad of personality: A 10 year review. Social and Personality Psychology Compass, 7, 199–216. Gaughan, E. T., Miller, J. D., Pryor, L. R., & Lynam, D. R. (2009). Comparing two alternative models of general personality in the assessment of psychopathy: A test of the NEO PI-R and the MPQ. Journal of Personality, 77, 965–996. Gentile, B., Miller, J. D., Hoffman, B. J., Reidy, D. E., Zeichner, A., & Campbell, W. K. (2013). A test of two brief measures of grandiose narcissism: The Narcissistic Personality Inventory (NPI)-13 and NPI-16. Psychological Assessment, 25, 1120–1136. Glenn, A. L., & Sellbom, M. (2015). Theoretical and empirical concerns regarding the Dark Triad as a construct. Journal of Personality Disorders, 29, 360–377. Hare, R. D. (2003). The Hare Psychopathy Checklist-Revised (2nd ed.). Toronto, Canada: Multi-Health Systems. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Jakobwitz, S., & Egan, V. (2006). The Dark Triad and normal personality traits. Personality and Individual Differences, 40, 331–339. Jonason, P. K., Lyons, M., & Blanchard, A. (2015). Birds of a “bad” feather flock together: The Dark Triad and mate choice. Personality and Individual Differences, 78, 34–38. Jonason, P. K., Strosser, G. L., Kroll, C. H., Duineveld, J. J., & Baruffi, S. A. (2015). Valuing myself over others: The Dark Triad

14

traits and moral and social values. Personality and Individual Differences, 81, 102–106. Jonason, P. K., & Webster, G. D. (2010). The Dirty Dozen: A concise measure of the Dark Triad. Psychological Assessment, 22, 420–432. Jones, D. N. (2013). What’s mine is mine and what’s yours is mine: The Dark Triad and gambling with your neighbor’s money. Journal of Research in Personality, 47, 563–571. Jones, D. N. (2014). Risk in the face of retribution: Psychopathic individuals persist in financial misbehavior among the Dark Triad. Personality and Individual Differences, 67, 109–113. Jones, D. N., & Figueredo, A. J. (2013). The core of darkness: Uncovering the heart of the Dark Triad. European Journal of Personality, 27, 521–531. Jones, D. N., & Paulhus, D. L. (2014). Introducing the Short Dark Triad (SD3): A brief measure of dark personality traits. Assessment, 21, 28–41. Jones, D. N., & Weiser, D. A. (2014). Differential infidelity patterns among the Dark Triad. Personality and Individual Differences, 57, 20–24. Joreskog, K. G., & Sorbom, D. (1989). LISREL 7: A guide to the program and applications. Chicago: SPSS. Kelley, T. L. (1927). Interpretation of educational measurements. New York: World Book. Lamborn, S. D., Mounts, N. S., Steinberg, L., & Dornbusch, S. M. (1991). Patterns of competence and adjustment among adolescents from authoritative, authoritarian, indulgent, and neglectful families. Child Development, 62, 1049–1065. Lance, C. E., Noble, C. L., & Scullen, S. E. (2002). A critique of the correlated-trait correlated-method and correlated uniqueness models for multitrait-multimethod data. Psychological Methods, 7, 228–244. LeBreton, J. M., James, L. R., & Lindell, M. K. (2005). Recent issues regarding rWG, rWG, rWG (J), and rWG (J). Organizational Research Methods, 8, 128–138. Lee, K., & Ashton, M. C. (2005). Psychopathy, Machiavellianism, and narcissism in the five-factor model and the HEXACO model of personality structure. Personality and Individual Differences, 38, 1571–1582. Lee, K., & Ashton, M. C. (2014). The Dark Triad, the Big Five, and the HEXACO model. Personality and Individual Differences, 67, 2–5. Lilienfeld, S.O., & Widows, M. (2005). Professional manual for the Psychopathic Personality Inventory-Revised (PPI-R). Lutz, FL: Psychological Assessment Resources. Lynam, D. R., Gaughan, E. T., Miller, J. D., Miller, D. J., MullinsSweatt, S., & Widiger, T. A. (2011). Assessing the basic traits associated with psychopathy: Development and validation of the Elemental Psychopathy Assessment. Psychological Assessment, 23, 108–124. Lynam, D. R., Hoyle, R. H., & Newman, J. P. (2006). The perils of partialling: Cautionary tales from aggression and psychopathy. Assessment, 13, 328–341.

Miller, Hyatt, Maples-Keller, et al.

Lynam, D. R., & Widiger, T. (2001). Using the five factor model to represent the personality disorders: An expert consensus approach. Journal of Abnormal Psychology, 110, 401–412. Maples, J. L., Guan, A. L., Carter, N., & Miller, J. D. (2014). A test of the International Personality Item Pool representation of the Revised NEO Personality Inventory and development of a 120item IPIP-based measure of the five-factor model. Psychological Assessment, 26, 1070–1084. Markon, K. E., Krueger, R. F., & Watson, D. (2005). Delineating the structure of normal and abnormal personality: An integrative hierarchical approach. Journal of Personality and Social Psychology, 88, 139–157. McCrae, R. R. (2008). A note on some measures of profile agreement. Journal of Personality Assessment, 90, 105–109. McDonald, M. M., Donnellan, M. B., & Navarrete, C. D. (2012). A life history approach to understanding the Dark Triad. Personality and Individual Differences, 52, 601–605. McHoskey, J. W., Worzel, W., & Szyarto, C. (1998). Machiavellianism and psychopathy. Journal of Personality and Social Psychology, 74, 192–210. Miller, J. D., Bagby, R. M., Pilkonis, P. A., Reynolds, S. K., & Lynam, D. R. (2005). A simplified technique for scoring the DSM-IV personality disorders with the five-factor model. Assessment, 12, 404–415. Miller, J. D., Dir, A., Gentile, B., Wilson, L., Pryor, L. R., & Campbell, W. K. (2010). Searching for a vulnerable Dark Triad: Comparing factor 2 psychopathy, vulnerable narcissism, and borderline personality disorder. Journal of Personality, 78, 1529–1564. Miller, J. D., & Lynam, D. R. (2003). Psychopathy and the fivefactor model of personality: A replication and extension. Journal of Personality Assessment, 81, 168–178. Miller, J. D., Lynam, D., Widiger, T. A., & Leukefeld, C. (2001). Personality disorders as an extreme variant of common personality dimensions: Can the five-factor model represent psychopathy? Journal of Personality, 69, 253–276. Morf, C. C., & Rhodewalt, F. (2001). Unraveling the paradoxes of narcissism: A dynamic self-regulatory processing model. Psychological Inquiry, 12, 177–196. Mullins-Sweatt, S. N., Jamerson, J. E., Samuel, D. B., Olson, D. R., & Widiger, T. A. (2006). Psychometric properties of an abbreviated instrument of the five-factor model. Assessment, 13, 119–137. O’Boyle, E. H., Forsyth, D. R., Banks, G. C., & McDaniel, M. A. (2012). A meta-analysis of the Dark Triad and work behavior: A social exchange perspective. Journal of Applied Psychology, 97, 557–579. O’Boyle, E. H., Forsyth, D. R., Banks, G. C., Story, P. A., & White, C. D. (2015). A meta-analytic test of redundancy and relative importance of the Dark Triad and five-factor model of personality. Journal of Personality, 83, 644–664. Pailing, A., Boon, J., & Egan, V. (2014). Personality, the Dark Triad and violence. Personality and Individual Differences, 67, 81–86. Paulhus, D. L. (2001). Normal narcissism: Two minimalist accounts. Psychological Inquiry, 12, 228–230.

Psychopathy and Machiavellianism

Paulhus, D. L., Neumann, C. F., & Hare, R. D. (in press). Manual for the Self-Report Psychopathy Scale (SRP-III). Toronto, Canada: Multi-Health Systems. Paulhus, D. L., & Williams, K. M. (2002). The Dark Triad of personality: Narcissism, Machiavellianism and psychopathy. Journal of Research in Personality, 36, 556–563. Poythress, N. G., Skeem, J. L., & Lilienfeld, S. O. (2006). Associations among early abuse, dissociation, and psychopathy in an offender sample. Journal of Abnormal Psychology, 115, 288–297. Raskin, R., & Terry, H. (1988). A principle-components analysis of the Narcissistic Personality Inventory and further evidence of its construct validity. Journal of Personality and Social Psychology, 54, 890–902. Rauthmann, J. F. (2013). Investigating the MACH–IV with item response theory and proposing the Trimmed MACH*. Journal of Personality Assessment, 95, 388–397. Rauthmann, J. F., & Kolar, G. P. (2013). Positioning the Dark Triad in the interpersonal circumplex: The friendly-dominant narcissist,

15

hostile-submissive Machiavellian, and hostile-dominant psychopath? Personality and Individual Differences, 54, 622–627. Saunders, B., & Giolas, M. H. (1991). Dissociation and childhood trauma in psychologically disturbed adolescents. American Journal of Psychiatry, 148, 50–54. van der Linden, D., te Nijenhuis, J., & Bakker, A. B. (2010). The General Factor of Personality: A meta-analysis of Big Five intercorrelations and a criterion-related validity study. Journal of Research in Personality, 44, 315–327. Vernon, P. A., Villani, V. C., Vickers, L. C., & Harris, J. A. (2008). A behavioral genetic investigation of the Dark Triad and the Big 5. Personality and Individual Differences, 44, 445–452. Whiteside, S. P., & Lynam, D. R. (2001). The five factor model and impulsivity: Using a structural model of personality to understand impulsivity. Personality and Individual Differences, 30, 669– 689.