Fearless Dominance/Boldness Is Not Strongly Related to Externalizing Behaviors: An Item Response-Based Analysis

Abstract

There is substantial and ongoing debate regarding the centrality of Fearless Dominance/Boldness (FD/B) to psychopathic personality due, in part, to its generally weak relations with externalizing behaviors. In response to these findings, proponents of FD/B have offered two hypotheses. First, FD/B may have nonlinear associations with externalizing outcomes such that FD/B may lead to resilience at moderate levels, but an overabundance of FD/B will yield maladaptive behavioral outcomes. Second, FD/B may be related to antisocial outcomes when paired with high scores on other psychopathic traits such as self-centered impulsivity, meanness, or disinhibition. The current study tests these two possibilities using two large samples (Study 1: 787 undergraduates; Study 2: 596 Amazon’s Mechanical Turk participants). An item response theory scoring approach particularly sensitive to curvilinearity was used to maximize our ability to find a true curvilinear effect, if present. No evidence in favor of the curvilinearity hypothesis was found. Only a single significant interaction predicting substance use was observed between boldness and meanness. These findings contribute to a growing literature raising concerns regarding the relevance of FD/B to psychopathy.

Keywords

nonlinear curvilinearity antisocial PD IRT

Psychopathy is one of the most commonly examined and well-validated personality disorder constructs. However, considerable debate remains regarding its underlying structure and the centrality of certain components (e.g., Lilienfeld et al., 2012; Lynam & Miller, 2012; Miller & Lynam, 2012). One of the most widely debated elements of psychopathy is Fearless Dominance/Boldness (FD/B), which is characterized by resilience to stress and anxiety, fearlessness, and social influence. Formulations of the FD/B construct are represented in certain classic clinical descriptions of the disorder from the mid to late 20th century, in which psychopathic individuals are described as at once callous, egocentric, impulsive, and irresponsible while also seemingly well-mannered, charming, and agentic (e.g., Cleckley, 1976; Crego & Widiger, 2016; Miller et al., 2001). However, FD/B traits do not figure as prominently in other modern conceptualizations of psychopathy such as Hare’s model and measure, the Psychopathy Checklist (PCL; Hare, 1980; Psychopathy Checklist–Revised [PCL-R]; Hare, 2003), or its derivatives such as the Self-Report Psychopathy Scale (Paulhus et al., 2017) and Levenson Self-Report Psychopathy Scale (Levenson et al., 1995). These traits were excluded due to their lack of relations with other psychopathy-relevant traits and outcomes (Hare & Neumann, 2008; Miller & Lynam, 2012). For instance, a meta-analysis of the Psychopathic Personality Inventory (PPI; Lilienfeld & Andrews, 1996) showed that FD manifested null to small positive relations with antisocial behaviors that some consider the sine qua non of psychopathy-related outcomes (Miller & Lynam, 2012; cf. Skeem & Cooke, 2010). A recent meta-analytic review of the Triarchic Psychopathy Measure (TriPM; Patrick, 2010) revealed the same pattern for Boldness (Sleep et al., 2019).

Two hypotheses for the importance of FD/B have been proposed. First, FD/B may have nonlinear associations with externalizing outcomes such that FD/B may lead to resilience at moderate levels, but an overabundance of FD/B will yield maladaptive behavioral outcomes. Second, FD/B may be related to antisocial outcomes when paired with high scores on other psychopathic traits such as self-centered impulsivity (SCI), meanness, or disinhibition. The present study tests both hypotheses.

Measures of Fearless Dominance/Boldness

The PPI (Lilienfeld & Andrews, 1996), as well as its subsequent revision, the PPI–Revised¹ (PPI-R; Lilienfeld & Widows, 2005), is a widely used measure that treats FD/B as a major factor of psychopathy. PPI-based measures contain eight subscales that can be grouped into two higher order factors titled Fearless Dominance (PPI FD—comprising three subscales: Fearlessness, Stress Immunity, and Social Influence) and Self-Centered Impulsivity (PPI SCI—comprising four subscales: Machiavellian Egocentricity, Rebellious Nonconformity, Blame Externalization, and Carefree Nonplanfulness). An eighth subscale—Coldheartedness (PPI C)—is a separate scale that does not load on either of the two higher order factors.

More recent measures such as the TriPM (Patrick, 2010) and the Elemental Psychopathy Assessment (EPA; Lynam et al., 2011) have followed suit in including an FD-like construct as a component of psychopathy. The TriPM is based on Patrick et al.’s (2009) triarchic psychopathy model, which includes three central psychopathy traits—Boldness, Meanness, and Disinhibition. Empirically, TriPM Boldness overlaps very strongly with PPI FD (e.g., rs > .80). In general, PPI FD and TriPM Boldness appear to measure the same construct as both are mutually characterized by resilience to fear and anxiety, social dominance, low neuroticism, and high extraversion (e.g., Patrick et al., 2009) and are very highly correlated (meta-analytic r = .79, Sleep et al., 2019).

Controversy of Fearless Dominance/Boldness

Although the PPI-R and TriPM are widely used (e.g., Ross et al., 2009; Witt et al., 2009), empirical support for the FD/B construct is mixed, with a growing number of findings raising questions regarding FD/B’s relevance to psychopathy. While some researchers have suggested that FD/B serves to assess psychopathic traits represented by Factor 1 of the PCL and its subsequent revision, the PCL-R (Hare, 2003), they share only a small amount of variance. In fact, the variance shared between Factor 1 of the PCL-R and FD is smaller than that shared between Factor 1 and SCI (see Marcus et al., 2012; Miller & Lynam, 2012). Previous findings also indicate that FD/B bears null to weak positive relations with many other psychopathy measures and subscales, suggesting weaker convergent validity than other components of psychopathy (Miller & Lynam, 2012; cf. Murphy et al., 2016). For example, PPI FD shows null to weak relations with the other higher order factor, PPI SCI, which is thought to assess core psychopathy traits (Crego & Widiger, 2014; see Miller & Lynam, 2012, for a meta-analysis); the same is true for TriPM Boldness and its relations with Meanness (meta-analytic r = .16) and Disinhibition (meta-analytic r = −.05; Sleep et al., 2019). Such null findings run counter to theories that suggest low anxiety is the foundation of psychopathy (Lykken, 1995). They are, however, contrasted by the meta-analytic finding that PPI-FD has a moderate to strong association with Factors 1 and 2 of the Self-Report Psychopathy Scale (Marcus et al., 2012).

It is possible that previous meta-analyses on the relationship between FD/B and other components of psychopathy are misleading due to their reliance on the PCL-R, which puts little emphasis on boldness (Lilienfeld et al., 2016), although relations with other psychopathy measures are relatively small as well (Sleep et al., 2019). FD/B’s relations with non-PCL-based measures of psychopathy are stronger—but the magnitude of those effects is quite variable.

Second, there is limited empirical evidence indicating that PPI FD and TriPM Boldness converge with other psychopathy scales in relation to important correlates of psychopathy. For instance, FD generally manifests null to small relations with externalizing behaviors thought by many, but not all (Skeem & Cooke, 2010), to be characteristic of psychopathy such as aggression, substance use, and antisocial behavior (see Miller & Lynam, 2012). Similar findings have been reported regarding TriPM Boldness; a recent meta-analysis found relations for Boldness with indices of externalizing behaviors to range from .05 to .12 (Sleep et al., 2019).

Third, previous findings indicate that FD/B is more strongly related to adaptive outcomes than maladaptive ones. While the construct does show moderate convergence with an expert-generated prototype of Cleckley-based psychopathy from the perspective of the five-factor model (Lilienfeld et al., 2012; Miller & Lynam, 2012), it is associated with lower risk of psychopathology and greater emotional stability (e.g., Crego & Widiger, 2014; Miller & Lynam, 2012). FD/B is also strongly associated with low neuroticism/negative emotionality and high extraversion/positive emotionality, suggesting that it may be linked to such positive life outcomes as subjective well-being, self-esteem, physical health, and longevity, among others (Ozer & Benet-Martinez, 2006). Smith et al. (2013) have shown that FD/B traits are associated with acting heroically and altruistically in daily life—behaviors not typically associated with psychopathy. FD/B is also associated with qualities that underlie prosocial behavior including emotion recognition, empathy, and sociability, as well as prosocial behavior itself (Gatner et al., 2016). There is some evidence that FD/B is related to grandiose narcissism (e.g., Sellbom & Phillips, 2013) but these shared aspects are largely limited to the more adaptive, extraversion-related components of narcissism that do not involve the interpersonal antagonism characteristic of psychopathy and narcissism (Krusemark et al., 2018; Miller et al., 2020). This is consistent with narcissism and FD/B’s shared association with five-factor model extraversion, but divergent association with agreeableness (e.g., Crego & Widiger, 2014; Maples et al., 2014; O’Boyle et al., 2015). Finally, when the personality profile associated with boldness is presented without any psychopathy-related labeling or context, experts rated it as less relevant to psychopathy than Meanness or Disinhibition (Miller et al., 2016; cf. Berg et al., 2017). In fact, unlike other dimensions of psychopathy, the personality correlates of FD/B align strongly with expert ratings of “healthy personality” (Bleidorn et al., 2020).

There have been several attempts to explain FD/B’s null to limited relations with psychopathy-relevant outcomes such as antisocial behavior. The first is an interaction-based hypothesis that suggests that FD/B will be related to externalizing behaviors when paired with higher scores on other psychopathy components such as PPI SCI, meanness, or disinhibition. Lilienfeld et al. (2012) suggest that

fearless dominance . . . gives rise to the full clinical picture of psychopathy in the presence of elevated disinhibition, meanness, or both . . . there is no requirement that PPI-FD by itself should be associated with maladaptive functioning. (p. 332)

Although they note that these interactions have been difficult to detect (Lilienfeld et al., 2012). Empirical examinations of the interaction hypothesis have yielded mixed results, with a few studies noting FD × SCI interactions in relation to treatment failure (Rock et al., 2013), and sexually predatory attitudes (Marcus & Norris, 2014), while other studies have found no evidence of interactive effects (e.g., Gatner et al., 2016; Maples et al., 2014; Miller et al., 2016; see also Lilienfeld et al., 2012, for other examples). For example, across 42 outcomes, Vize et al. (2016) found evidence for only two FD by SCI interactions, and neither was in the anticipated direction. More recently, the interaction hypothesis was tested in a correctional sample (Weiss et al., 2019). Weiss et al. found no evidence of FD by SCI interactions for any of the four outcomes tested. Indeed, if FD/B interacts with the other components of psychopathy, some findings suggest that it may provide a protective effect. For instance, Gatner et al. (2016) demonstrated that at higher levels of TriPM Boldness: (a) Disinhibition and Meanness failed to exhibit stronger relations with negative outcomes and (b) meanness exhibited weaker relations with social aggression and impulsiveness.

A second possible explanation for why examinations of FD/B have failed to find expected relations with antisocial behavior and other externalizing behaviors was put forward by Blonigen (2013), who hypothesized that FD/B may bear curvilinear relations with maladaptive outcomes, such that optimal outcomes are shown in those with moderate standing on FD, whereas those with very high levels are disposed toward problematic behavior. Specifically, Blonigen (2013) stated,

[A] certain amount of boldness, fearlessness, confidence, and social dominance is likely to engender resilience in the face of adversity and success in a number of important life domains; however, an overabundance of such traits is likely to be expressed as narcissism, arrogance, recklessness, and risk-taking . . . (p. 88)

Unlike the interaction-based explanation, this curvilinear hypothesis regarding FD/B’s relations to externalizing behaviors has received little attention to date. Vize et al. (2016) tested the effect of FD/B on both concurrent and future behavior, and found that FD/B, measured by proxy in early adolescents, showed a curvilinear effect for three of 42 possible outcomes; importantly, none of these curvilinear effects were of the same form and only one was consistent with Blonigen’s hypothesis. Similarly, Gatner et al. (2016) tested TriPM Boldness’ curvilinear relation with several antisocial criteria, finding that Boldness exhibited a small incremental nonlinear effect when predicting TriPM Meanness and physical aggression. The authors interpreted their effect sizes as too small in magnitude to support Blonigen’s (2013) hypothesis. In general, these two studies cast doubt on the existence of important curvilinear relations among FD/B and externalizing outcomes, but both had limitations. First, a null finding in the Vize et al. study for curvilinear relations does not necessarily preclude the possibility that extreme levels of FD/B in adulthood (vs. early life) predict antisocial behavior. Second, the use of a proxy measure of FD limited the authors’ ability to test whether the lower order components of FD (e.g., stress immunity) demonstrate curvilinear relations with externalizing behavior. Given the multifaceted nature of FD, it is possible that curvilinear effects associated with lower order subcomponents may have been obscured in the higher order analysis. Last, both simulated and traditional analyses have indicated that traditional scoring methods are limited in their ability to detect curvilinearity at extreme ends of trait distributions (Carter et al., 2014, 2016, 2017). More recently, Weiss et al. (2019) evaluated the curvilinearity hypothesis on a correctional sample, finding no support for the hypothesis, but their analysis was limited to the PPI.

Study Aims

In the present study, we test both key hypotheses regarding how FD/B may relate to externalizing outcomes. We specifically examine whether FD/B shows an increasingly strong positive association with externalizing at high ends of the FD/B trait distribution (Blonigen, 2013), and whether FD/B interacts with high SCI, disinhibition, or meanness to predict externalizing behavior. We utilize the two most common measures of FD/B (i.e., PPI-R and TriPM) and test these hypotheses in a large undergraduate sample and a large Mechanical Turk sample, both of which are large enough to be reasonably well-powered for such analyses (see Results and Supplemental Power Analysis in supplemental materials [available online] for more details on the power analysis conducted). For the sake of comprehensiveness, we also test whether the non-FD/B-related scales of the PPI-R and TriPM’s (e.g., PPI-R Coldheartedness; TriPM Meanness) bear curvilinear relations with externalizing behaviors. By examining the subscales of the PPI-R’s FD factor, we can conduct a more granular test of where curvilinear relations may exist within the broader construct.

To investigate curvilinearity, we use an ideal point item response theory (IRT) model, namely, the generalized graded unfolding model (GGUM; Roberts et al., 2000), which is an approach toward reproducing response patterns for self-report measures of personality (e.g., Stark et al., 2006), attitudes (e.g., Carter & Dalal, 2010), and affect (LaPalme et al., 2017). The ideal point approach to scoring has been shown to more accurately recover curvilinear relationships in simulated (Carter et al., 2017) and empirical (Carter et al., 2014, 2016) data.

Method

Sample 1: Participants and Procedure

Sample 1 comprised 787 participants recruited from the research pool at a large Southern state university in the United States in exchange for research credit (53% men; M_age = 19.34; SD = 2.19; 83% Caucasian; 7% Asian; 6% African American). On signing informed consent, participants completed a packet of self-report questionnaires and were debriefed. Some of these data were used in previous studies on the factor structure of the EPA and the validity of brief psychopathy measures (Few et al., 2013; Miller et al., 2012), not for tests of curvilinearity. Institutional review board approval was provided for this study.

Sample 2: Participants and Procedure

Participants included 865 participants recruited from Amazon’s Mechanical Turk (MTurk) website. Research utilizing MTurk has shown that the MTurk population is a reliable source of data with more demographic diversity than most undergraduate populations (Chandler & Shapiro, 2016; Miller et al., 2017). To participate, participants had to be 18 years of age or older and reside in the United States. Participants were paid $2.00 for their participation. Of the 865 participants who completed informed consent, 269 participants were removed for failing one or both validity scales (see Measures section), for finishing the study in a time deemed invalid (>20 minutes), for having more than 25% missing data, or for random responding. The final data set consisted of 596 participants (63% female; 83% White; 10% Black; 8% Asian; 6% Hispanic; M_age = 37.04, SD = 11.75).² Some of these data were used in previous studies clarifying the nomological network of narcissism (Crowe et al., 2018; Miller, Lynam, Siedor, et al., 2018; Miller, Lynam, Vize, et al., 2018; Vize et al., 2017), not for studies on psychopathy in general and curvilinearity more specifically. Institutional review board approval was provided for this study.

Measures

Psychopathic Personality Inventory–Revised (PPI-R).³

The PPI-R (Lilienfeld & Widows, 2005) is a 154-item measure of psychopathy that provides scores for eight subscales, as well as a global psychopathy score and two psychopathy factor scores (PPI-R PPI FD, and PPI-R PPI SCI). The PPI-R was given in Sample 1. Alphas for the 8 PPI-R subscales range from .83 to .88 with a median of .85.

Triarchic Psychopathy Measure

The TriPM (Patrick et al., 2009) is a 58-item self-report measure of psychopathy that provides a global psychopathy score (α = .87) as well as three scales: Boldness (19 items; α = .87), Meanness (18 items; α = .89), and Disinhibition (20 items; α = .87). While there are typically 19 items in the TriPM Meanness scale, one item (“I sympathize with others’ problems”) was dropped from the scale due to misfit. See discussion in Scoring Methods. The TriPM was used in Sample 2.

Crime and Analogous Behavior Scale

The Crime and Analogous Behavior Scale (CAB; Miller & Lynam, 2003) is a self-report inventory that assesses a variety of externalizing behaviors. A substance use variety count was created by giving participants a “1” for every substance they endorsed using (six items; M = 1.96; SD = 1.43). An antisocial behavior count was created by giving participants a “1” for every antisocial act they endorsed (10 items; M = 0.90; SD = 1.03). The following acts were included in the antisocial behavior scale: theft (of more and less than $50 value), car theft, physical fighting, threat with a weapon, assault with a weapon, intentional injury of another person, burglary, and been arrested (for DUI [driving under the influence of alcohol] and something other than DUI). Because these represent counts of behaviors, special analytic approaches for counts were used. The full CAB was given in Sample 1, whereas a brief version was used in Sample 2. In Sample 2, a substance use variety count was created by giving participants a “1” for every substance they endorsed using (five items; M = 1.84; SD = 1.36). An antisocial behavior count was created by giving participants a “1” for every antisocial act they endorsed (10 items; M = 1.08; SD = 1.40).

Reactive and Proactive Aggression Questionnaire

The Reactive and Proactive Aggression Questionnaire (RPA; Raine et al., 2006) scale consists of 23 self-report items assessing two aggression scales: Proactive Aggression (12 items; M = 1.51, SD = 2.44) and Reactive Aggression (11 items; M = 6.83, SD = 3.66). Items are scored in the following manner: 0 (never), 1 (sometimes), and 2 (often). In Sample 2, the coefficient alphas for the proactive and reactive scales were both .82. Reactive and proactive aggression were substantially correlated, r = .55.

Validity Scales

Two validity scales from the EPA (Lynam et al., 2011) were used, the Infrequency Scale and the Too Good to be True Scale. Participants were removed from the analyses if they received a score of 4 or more on the Infrequency Scale or a score of 3 or more on the Too Good to be True Scale.

Data Analysis

Scoring Methods

In the current investigation, we utilize IRT scoring for all self-report measures, except those that are based on behavioral count data. The use of IRT-based scores is particularly important to the current investigation due to recent findings that they yield more accurate tests of curvilinearity (Carter et al., 2017). IRT models generally assume that participant’s responses to test items are influenced by characteristics of the person as well as the idiosyncratic characteristics of the measure administered. Responses are thought to be only partially determined by the level of the measured personality trait possessed by the respondent. IRT models also propose that responses are determined by the extremity of the item content, which is alternatively referred to as item difficulty, location, or popularity. Finally, IRT models consider that responses are also partially determined by the extent to which the item can differentiate between those with higher versus lower trait standing, referred to as discrimination (which can be thought of as similar to factor loadings). By accounting for item-level idiosyncrasies in test scoring, IRT scoring controls the Type I error in tests for curvilinearity caused by items whose content expresses more extreme levels of the trait than most persons being assessed, which Carter et al. (2017) showed is typical of commonly used personality measures.

Dominance Versus Ideal Point Models

Although IRT scoring generally increases the accuracy of tests for curvilinearity, Carter et al. (2016) pointed out that it is also crucial to utilize an IRT model that appropriately characterizes the item response process (i.e., the way respondents “use” the scale to respond to the item content). The majority of IRT models (as well as classical sum-scores and factor scores) assume a dominance response process, which implies that those with higher levels of a personality trait will always be more likely to endorse a positively keyed item. Although this assumption is sensible for measures of ability—in which respondents attempt to “dominate” the item, or answer correctly—and measures based on behavioral frequency, recent work has cast doubt on the validity of the dominance assumption for personality tests with Likert-type endorsement scales (e.g., LaPalme et al., 2017; Roberts et al., 1999; Stark et al., 2006).

Counter to the assumption of a dominance process, ideal point models assume that individuals are most likely to endorse items with content that matches their own personality trait level. Thus, under ideal point model assumptions, an individual can be less likely to fully endorse an item (i.e., “strongly agree”) both because the item is too extreme and because the item is not extreme enough. For example, a statement such as “I keep my room clean most of the time” may not be endorsed by persons low in conscientiousness because they do not keep their room clean, but may also not be endorsed by persons with very high conscientiousness because they keep their room clean all the time, not just “most of the time.”

There is debate regarding the necessity of ideal point models for personality constructs (Drasgow et al., 2010; c.f., Reise, 2010), but research has shown the ideal point assumption generally shows better fit to self-report personality measures (e.g., Stark et al., 2006). While overall scores within a sample may not show substantial differences across the two scoring approaches, assuming a dominance process when an ideal point process is more appropriate causes confusion in rank-order at the top end of the trait distribution (see Roberts et al., 1999). Thus, ideal point models have proven particularly useful for uncovering curvilinear relationships where they exist (Carter et al., 2014, 2016), and reducing false flags (i.e., Type I errors; Carter et al., 2017). However, as noted previously, the advantage of IRT scoring is predicated on the correct choice of an IRT model for scoring personality tests (Carter et al., 2016). Therefore, we consider the fit of both ideal point and dominance IRT models in the current investigation.

Model Estimation and Fit

The generalized partial credit model (GPCM; Muraki, 1992) assumes a dominance response process and is appropriate for polytomous Likert-type scores. The GGUM (Roberts et al., 2000) assumes an ideal point response process and is suited for polytomous Likert-type responses. The GPCM was estimated using the “mirt” package in R (Chalmers, 2012), and the GGUM was estimated using the GGUM2004 software program (Roberts et al., 2006). Both programs utilize marginal maximum likelihood estimation to determine item parameters and Expected a Posteriori scoring to determine persons’ trait levels.

See Supplemental Tables (available online) for all fit statistics used for model selection. Relative model-data fit for the GPCM (i.e., dominance model) and GGUM (i.e., ideal point model) was evaluated by comparing the Akaike information criterion (Akaike, 1973; Bozdogan, 1987). The GGUM showed superior model-data fit for all measures in Sample 1. GPCM showed superior model-data fit for all measures in Sample 2. The MODFIT program (Stark, 2007) was used to evaluate absolute model-data fit (i.e., χ²/degrees of freedom [df] ratios) of the GGUM (Sample 1) and GPCM (Sample 2). Evaluation of absolute model-data fit was necessary to verify that the data meets model assumptions of unidimensionality, which has been called into question for the TriPM Boldness domain (Shou et al., 2018). The χ²/df ratio compares the actual responses in the data with the responses predicted by the IRT model via the χ² statistic divided by the degrees of freedom for the test; ratios greater than three are generally considered to indicate model-data misfit.

In Sample 1, all PPI-R GGUM models showed ratios less than three. In Sample 2, GPCM TriPM Boldness and Disinhibition models showed acceptable model-data fit. As both the GGUM and GPCM assume unidimensionality, adequate model-data fit is indicative of an acceptable unidimensional structure for these traits. TriPM Meanness manifested unacceptable model-data fit as all ratios were greater than three. Further evaluation revealed that the misfit of the model was the result of multidimensionality. Once the single worst fitting item (“I sympathize with others’ problems”) was removed, model-data fit was appropriate with all ratios falling below three. All analyses were conducted using estimates from this 18-item version of the scale.

The GGUM was used to estimate all component scores in Sample 1. PPI-R domain-level scores were also generated in Sample 1. GGUM estimates of measured traits (θ) were averaged to generate higher order factor scores. The PPI-R SCI factor was estimated by averaging Blame Externalization, Carefree Nonplanfulness, Machiavellian Egocentricity, and Rebellious Nonconformity scores. The PPI-R FD factor was estimated by averaging Fearlessness, Social Influence, and Stress Immunity scores. The GPCM was used to estimate all TriPM scores for Sample 2 as well as reactive and proactive aggression scores.

In IRT, test information is conceptually similar to scale reliability and can be thought of as the scale’s measurement precision at a particular level of the latent trait, θ_j. Test information can be used to generate at test reliability function (TRF), which depicts measurement reliability at all levels of the latent trait. The TRF and observed distribution of latent psychopathy traits were examined as part of our evaluation of the samples’ appropriateness for testing the curvilinearity hypothesis. Histograms of latent traits and response distributions were also examined.

Regression Models

As count data (such as the CAB) often deviate from the assumptions of ordinary least squares (OLS) regression, alternative models were applied in the present analyses. Poisson regression is often applied to count data and assumes that the variance of data is congruent with the mean. When the variance of data deviates substantially from the mean, negative binomial regression is more appropriate. Negative binomial regression accounts for overdispersion (i.e., when variance exceeds the mean) by estimating an additional dispersion parameter (α) and applying more conservative tests of significance in proportion to the degree of dispersion and standard error (Atkins et al., 2013). For all regressions, the appropriateness of modeling with Poisson and negative binomial was evaluated by comparing model fit using Bayesian information criterion (BIC; Raftery, 1995), which includes a correction for the increased complexity of the negative binomial model. Poisson and negative binomial regressions were only considered for the CAB. For the noncount data outcomes (i.e., reactive and proactive aggression) standard OLS regression with robust standard errors was used.

To examine curvilinearity, all latent trait scores were standardized, and the polynomial term was calculated from the standardized value (Cohen et al., 2002). In the first step of all analyses standardized psychopathy scores (i.e., GGUM latent trait estimates) were entered as a predictor of externalizing behavior. In the second step, the squared value of the standardized psychopathy score was entered as an additional predictor and the change in model-data fit was evaluated to assess for curvilinearity.

The incremental contribution of the curvilinear effect for each model was evaluated using BIC (Raftery, 1995), and R² or Pseudo R² (McFadden, 1974). BIC was the primary fit index used when evaluating whether the curvilinear model indicated considerable improvement on the simpler linear model. Raftery’s (1995) suggestion of a BIC difference greater than two was used as positive evidence of a better fitting model. McFadden’s (1974) pseudo R² calculation was used for all Poisson and negative binomial models, which do not have a statistical equivalent to OLS R². Although the intended use of McFadden’s pseudo R² is similar to the OLS R² metric, its values tend to be much smaller and cannot be interpreted as variance accounted for by the model. McFadden R² values ranging from .2 to .4 are thought to be indicative of excellent model fit, comparable to OLS R² values of .7 to .9 (McFadden, 1979).

To examine interaction effects, the standardized latent trait estimates were entered as predictors of externalizing behavior along with the product term of the predictors. When a significant interaction term was identified, simple effects were evaluated at 1 standard deviation above and below the mean of the non-FD/B term. All potential domain level FD/B interaction effects were evaluated (i.e., FD × SCI; FD × Coldheartedness; Boldness × Meanness; Boldness × Disinhibition).

Results

IRT models estimate individual latent traits scores (θ). These scores are on a standardized scale with zero being the population mean of the latent trait and one being 1 standard deviation above the population mean of that trait. Histograms of all θ score distributions and outcome counts are included in the supplemental materials along with additional descriptive statistics of the raw and latent trait scores (Supplemental Table 5 available online) and graphs of all TRFs (Supplemental Figures 8–11 available online). In Sample 1, the average maximum latent trait score across PPI-R scales was 3.04 standard deviations above the population mean and the average range was 6.14 standard deviations wide. The smallest maximum latent trait score was 2.06 (Carefree Nonplanfulness; range = −3.68 to 2.06). The largest raw Carefree Nonplanfulness score was 45 out of a possible 57 and a total of 9 participants endorsed⁴ at least three quarters of the scale (i.e., 15 of 19 items) or more. The largest maximum latent trait score was 4.29 (Social Dominance; range = −4.36 to 4.29). The largest raw Social Dominance score was 52 out of a possible 54 and 202 participants endorsed at least three quarters of the scale (i.e., 14 of 18 items). Sample 1 TRFs revealed marginal reliabilities at 3 standard deviations above the population mean that ranged from .68 (Stress Immunity) to .88 (Carefree Nonplanfulness). In Sample 2, the average maximum TriPM latent trait score was 3.22 standard deviations above the population mean and the average range was 5.64 standard deviations wide. The smallest maximum latent trait score was 3.09 (Meanness; range = −1.76 to 3.09). The largest raw Meanness score was 65 out of a possible 72, and three participants endorsed at least three quarters of the scale (i.e., 14 of 18 items). The largest maximum latent trait score was 3.41 (Reactive aggression; range = −2.09 to 3.41). The largest raw reactive aggression score was 22 out of a possible 22, and 128 participants endorsed at least three quarters of the scale (i.e., 9 of 11 items). Sample 2 TRFs revealed marginal reliabilities at three standard deviations above the population mean that ranged from .70 (Boldness) to .92 (Meanness).

Psychopathy scale correlations are reported in Supplemental Table 1 (available online). Results of all curvilinear regression analyses are reported in Tables 1 to 4. Overdispersion was evaluated for all models predicting CAB externalizing outcomes by comparing the BIC fit of the Poisson model with the negative binomial model. Results indicated that Poisson regression was appropriate when using PPI-R scores to predict CAB outcomes in nearly all cases. Only the prediction of CAB antisocial behavior from PPI-R Social Influence required the use of negative binomial analyses. When TriPM subscales were used, Poisson was most appropriate for substance use, while negative binomial analysis was most appropriate for antisocial behavior. Model fit was also evaluated for the reactive and proactive aggression scales. Fit suggested that while OLS analyses were acceptable for the prediction of reactive aggression, negative binomial analyses were more appropriate for the prediction of proactive aggression.

Table 1.

PPI-R Substance Abuse Regression Analyses.

PPI-R facet	Parameter	Step 1				Step 2
PPI-R facet	Parameter	B	[95% CI]	BIC	Pseudo R²	B	[95% CI]	BIC	Pseudo R²
SCI factor	Intercept	.66**	[.61, .71]	2733.42	.028	.74**	[.67, .80]	2728.89	.038
	Linear	.19**	[.12, .26]			.17**	[.10, .25]
	Quadratic					−.13**	[−.22, −.05]
BE θ	Intercept	.67**	[.62, .72]	2760.29	.014	.71**	[.64, .77]	2762.82	.018
	Linear	.10**	[.05, .16]			.09**	[.03, .15]
	Quadratic					−.04*	[−.08, .00]
CN θ	Intercept	.66**	[.61, .72]	2746.72	.023	.71**	[.65, .78]	2748.44	.028
	Linear	.13**	[.08, .19]			.12**	[.07, .18]
	Quadratic					−.05*	[−.10, .00]
ME θ	Intercept	.67**	[.62, .72]	2756.38	.015	.77**	[.71, .82]	2719.04	.054
	Linear	.10**	[.05, .16]			.13**	[.06, .19]
	Quadratic					−.12**	[−.16, −.08]
RN θ	Intercept	.67**	[.62, .72]	2744.86	.018	.75**	[.69, .81]	2733.34	.034
	Linear	.12**	[.06, .17]			.11**	[.06, .17]
	Quadratic					−.09**	[−.14, −.04]
FD factor	Intercept	.66**	[.61, .71]	2739.94	.029	.69**	[.63, .74]	2744.08	.032
	Linear	.20**	[.13, .26]			.22**	[.14, .30]
	Quadratic					−.05	[−.10, .01]
F θ	Intercept	.66**	[.61, .71]	2725.01	.043	.65**	[.60, .71]	2731.63	.043
	Linear	.18**	[.13, .23]			.18**	[.13, .23]
	Quadratic					.00	[−.03, .03]
Social θ	Intercept	.67**	[.62, .72]	2762.15	.009	.69**	[.63, .75]	2766.85	.011
	Linear	.08**	[.03, .13]			.09**	[.03, .15]
	Quadratic					−.02	[−.05, .02]
Stress θ	Intercept	.67**	[.62, .72]	2767.04	.005	.68**	[.62, .74]	2773.32	.005
	Linear	.06*	[.01, .11]			.06*	[.01, .12]
	Quadratic					−.01	[−.05, .03]
CH θ	Intercept	.66**	[.61, .71]	2745.05	.028	.72**	[.66, .78]	2741.83	.037
	Linear	.15**	[.09, .20]			.14**	[.08, .20]
	Quadratic					−.06**	[−.12, −.01]

Note. PPI-R = Psychopathy Personality Inventory–Revised; BIC = Bayesian information criterion; SCI = Self-centered Impulsivity; CI = confidence interval; BE = Blame Externalization; CH = Coldheartedness; CN = Carefree Nonplanfulness; F = Fearlessness; FD = Fearless Dominance; ME = Machiavellian Egocentricity; RN = Rebellious Nonconformity; Social = Social Influence; Stress = Stress Immunity; Bold text indicates the better model. Pseudo R² value was calculated using McFadden’s (1979) formula.

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

Table 2.

PPI-R Antisocial Behavior Regression Analyses.

PPI-R Facet			Step 1				Step 2
PPI-R Facet		Parameter	B	[95% CI]	BIC	Pseudo-R²	B	[95% CI]	BIC	Pseudo R²
SCI factor		Intercept	−.13**	[−.21, −.05]	1994.35	.036	−.04	[−.15, .06]	1994.24	.043
		Linear	.31**	[.21, .41]			.30**	[.19, .41]
		Quadratic					−.16*	[−.29, −.03]
BE θ		Intercept	−.12**	[−.19, −.04]	2019.54	.018	−.05	[−.15, .04]	2021.33	.023
		Linear	.17**	[.09, .24]			.15**	[.07, .24]
		Quadratic					−.07*	[−.13, −.01]
CN θ		Intercept	−.12**	[−.19, −.04]	2016.91	.017	−.07	[−.17, .04]	2021.05	.020
		Linear	.16**	[.08, .24]			.15**	[.07, .24]
		Quadratic					−.05	[−.12, .02]
ME θ		Intercept	−.12**	[−.20, −.04]	2006.73	.027	−.01	[−.10, .07]	1985.64	.055
		Linear	.20**	[.12, .27]			.26**	[.17, .35]
		Quadratic					−.14**	[−.19, −.09]
RN θ		Intercept	−.12**	[−.20, −.04]	2007.22	.023	−.07	[−.16, .03]	2010.42	.027
		Linear	.19**	[.11, .26]			.19**	[.11, .27]
		Quadratic					−.06	[−.12, .00]
FD Factor		Intercept	−.13**	[−.21, −.04]	2001.94	.032	−.11*	[−.20, −.02]	2007.91	.033
		Linear	.29**	[.20, .38]			.31**	[.20, .43]
		Quadratic					−.04	[−.12, .05]
F θ		Intercept	−.14**	[−.22, −.06]	1980.28	.054	−.10*	[−.19, −.01]	1982.71	.058
		Linear	.28**	[.21, .35]			.31**	[.22, .39]
		Quadratic					−.05*	[−.11, .00]
Social θ		Intercept	−.10*	[−.18, −.02]	2029.46	.001	−.12**	[−.21, −.03]	2035.23	.002
Step 1 Disp.	4.91	Linear	.04	[−.03, .12]			.04	[−.03, .11]
Step 2 Disp.	4.95	Quadratic					.02	[−.02, .05]
Stress θ		Intercept	−.11**	[−.19, −.03]	2018.63	.016	−.11*	[−.2, −.02]	2025.24	.016
		Linear	.15**	[.08, .22]			.15**	[.07, .23]
		Quadratic					−.01	[−.05, .04]
CH θ		Intercept	−.13**	[−.21, −.05]	1994.40	.043	−.10*	[−.20, .00]	1999.43	.045
		Linear	.26**	[.18, .33]			.26**	[.18, .35]
		Quadratic					−.04	[−.11, .03]

Note. BIC = Bayesian information criterion; PPI-R = Psychopathy Personality Inventory–Revised; BE = Blame Externalization; CI = confidence interval; CH = Coldheartedness; CN = Carefree Nonplanfulness; Disp. = Dispersion; F = Fearlessness; ME = Machiavellian Egocentricity; RN = Rebellious Nonconformity; Social = Social Influence; Stress = Stress Immunity. The smaller BIC values (in bold text) indicate the better fitting model. Pseudo R² value was calculated using McFadden’s (1979) formula.

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

Table 3.

TriPM Externalizing Behavior Regression Analyses.

TriPM Facet		Parameter	Step 1				Step 2
TriPM Facet		Parameter	B	[95% CI]	BIC	Pseudo R²	B	[95% CI]	BIC	Pseudo R²
Substance use
Boldness θ		Intercept	.60**	[.53, .66]	2043.00	.001	.58**	[.51, .66]	2048.56	.002
		Linear	.02	[−.05, .09]			.02	[−.05, .08]
		Quadratic					.02	[−.03, .06]
Disinhibition θ		Intercept	.58**	[.52, .64]	2004.49	.049	.55**	[.48, .62]	2008.76	.051
		Linear	.19**	[.13, .26]			.19**	[.12, .25]
		Quadratic					.03	[−.01, .07]
Meanness θ		Intercept	.60**	[.53, .66]	2041.10	.003	.63**	[.54, .71]	2046.11	.005
		Linear	.05	[−.02, .11]			.05	[−.02, .11]
		Quadratic					−.03	[−.09, .03]
Antisocial behavior
Boldness θ		Intercept	.02	[−.09, .12]	1669.23	.004	−.04	[−.15, .10]	1673.69	.007
Step 1 Disp.	1.53	Linear	.08	[−.03, .18]			.07	[−.03, .16]
Step 2 Disp.	1.55	Quadratic					.04	[−.01, .09]
Disinhibition θ		Intercept	−.09	[−.20, .01]	1585.51	.133	−.13*	[−.24, −.01]	1590.55	.135
Step 1 Disp.	2.77	Linear	.48**	[.38, .58]			.46**	[.35, .56]
Step 2 Disp.	2.79	Quadratic					.04	[−.01, .09]
Meanness θ		Intercept	−.01	[−.12, .09]	1648.00	.038	−.04	[−.17, .10]	1654.07	.038
Step 1 Disp.	1.74	Linear	.25**	[.15, .35]			.24**	[.14, .34]
Step 2 Disp.	1.75	Quadratic					−.02	[−.06, .10]

Note. BIC = Bayesian information criterion; CI = confidence interval; TriPM = Triarchic Psychopathy Measure; Disp. = Dispersion. BIC values in bold text indicate a better fitting model based on Raftery’s (1995) recommended threshold of BIC differences greater than two. Pseudo R² value was calculated using McFadden’s (1974) formula.

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

Table 4.

TriPM Aggression Regression Analyses.

TriPM facet	Parameter	Step 1				Step 2
TriPM facet	Parameter	B	[95% CI]	BIC	R ²	B	[95% CI]	BIC	R ²
Proactive aggression
Boldness θ	Intercept	.00	[−.06, .06]	1410.75	.000	.000	[−.07, .07]	1417.14	.000
	Linear	−.01	[−.07, .05]			−.01	[−.07, .06]
	Quadratic					.000	[−.04, .04]
Disinhibition θ	Intercept	.00	[−.06, .06]	1268.83	.213	−.13**	[−.19, −.06]	1240.96	.257
	Linear	.36**	[.30, .42]			.38**	[.32, .44]
	Quadratic					.13**	[.08, .17]
Meanness θ	Intercept	.00	[−.06, .06]	1303.87	.165	−.14**	[−.21, −.06]	1281.02	.205
	Linear	.32**	[.25, .38]			.32**	[.26, .38]
	Quadratic					.14**	[.09, .19]
Reactive aggression
Boldness θ	Intercept	.00	[−.07, .07]	1575.98	.022	−.04	[−.12, .04]	1579.35	.027
	Linear	−.13**	[−.21, −.06]			−.14**	[−.22, −.07]
	Quadratic					.04	[.00, .08]
Disinhibition θ	Intercept	.00	[−.06, .07]	1448.48	.211	−.04	[−.12, .05]	1452.80	.214
	Linear	.42**	[.35, .48]			.42**	[.35, .49]
	Quadratic					.04	[−.02, .09]
Meanness θ	Intercept	.00	[−.07, .07]	1565.14	.040	−.07	[−.17, .03]	1567.27	.046
	Linear	.18**	[.11, .26]			.18**	[.11, .25]
	Quadratic					.07	[.00, .14]

Note. TriPM = Triarchic Psychopathy Measure; CI = confidence interval; BIC = Bayesian information criterion. BIC values in bold text indicate a better fitting model based on Raftery’s (1995) recommended threshold of BIC differences greater than two. R² values in negative binomial models are Pseudo R² values calculated using McFadden’s (1974) formula.

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

Note that both Poisson and negative binomial regression predict the natural log of the expected count variable. Unless otherwise indicated, all reported coefficients are reported in log units. To convert coefficients to count units they must be exponentiated (i.e., $e^{x}$ ). However, to do so changes their interpretations from additive to multiplicative (see Coxe et al., 2009, for review).

Evaluating Curvilinearity

Substance Use

Step 1 of all analyses evaluated only linear effects. All eight of the PPI-R facets and both factor score estimates showed statistically significant linear effects when predicting substance use. PPI-R standardized regression coefficients ranged in size from .06 (p < .05; Stress Immunity) to .18 (p < .01; Fearlessness). When TriPM scales were used to predict substance use, regression coefficients ranged in size from .02 (p = ns; Boldness) to .19 (p < .01; Disinhibition). In Step 2 of all regression analyses linear and quadratic terms were included. When predicting substance use, all PPI-R SCI scales (i.e., Blame Externalization, Carefree Nonplanfulness, Machiavellian Egocentricity, and Rebellious Nonconformity) and Coldheartedness yielded statistically significant quadratic effects ranging in magnitude from −.12 (p < .01; Machiavellian Egocentricity) to −.04 (p < .05; Blame Externalization). Of these five effects, BIC values indicate that only the quadratic effects of Machiavellian Egocentricity, Rebellious Nonconformity, and Coldheartedness represent meaningful improvements over the linear model. At the factor level, PPI-R SCI also yielded a statistically significant quadratic effect that improved model fit. No TriPM facets to show a meaningful quadratic effect for the prediction of substance use. All identified quadratic effects are negative in direction, which indicates a reduction in substance use at higher latent trait levels, contrary to the direction hypothesized (Blonigen, 2013; see Supplemental Figures 1–4 available online).

Antisocial Behavior

In Step 1 of analyses predicting antisocial behavior, seven of eight PPI-R subscales and two of three TriPM subscales showed statistically significant effects. PPI-R subscale coefficients ranged in size from .04 (p = ns; Social Influence) to .28 (p < .01; Fearlessness). TriPM subscales ranged from .08 (p = ns; Boldness) to .48 (p < .01; Disinhibition). In Step 2, PPI-R Fearlessness (−.05; p < .05), Machiavellian Egocentricity (−.14; p < .01), Blame Externalization (−.07; p < .05), and the SCI factor (−.16; p < .05) showed statistically significant quadratic effects. Of these, only Machiavellian Egocentricity meaningfully improved the fit of the model. Again, the directionality of the curve was the opposite of the hypothesized direction (see Supplemental Figure 5 available online).

Reactive and Proactive Aggression

In Step 1, all TriPM subscales were significant linear predictors of reactive aggression with standardized coefficients ranging from −.13 (p < .01; Boldness) to .42 (p < .01; Disinhibition). Only Boldness failed to predict proactive aggression with coefficients ranging from −.01 (p = ns; Boldness) to .36 (p < .01; Disinhibition). Significant quadratic effects were found when Meanness and Disinhibition were used to predict proactive aggression. These effects were consistent with Blonigen’s hypothesized direction, albeit for different psychopathy domains (see Supplemental Figures 6 and 7 available online).

Evaluating Interactions

No significant interactions were identified between FD and SCI or FD and Coldheartedness. An interaction between Boldness and Meanness was identified when predicting substance abuse (β₀ = .59, p < .001; β_B = .03, p = .389; β_M = .05, p = .087; β_B&x42;M = .09 p = .003). Boldness was positively associated with substance abuse, but only at high levels of Meanness (−1 SD Meanness: β_B = −.19, p = .003; +1 SD Meanness: β_B = .20, p = .005). An interaction between Meanness and Disinhibition was also identified when predicting proactive aggression (β₀ = −.07, p = .02; β_M = .17, p < .001; β_D = .28, p < .001; β_B&x42;M = .14 p < .001), such that high levels of meanness increased the association between disinhibition and proactive aggression (−1 SD Meanness: β_D = .22, p < .001; +1 SD Meanness: β_D = .57, p < .001). No other significant interactions were found.

Discussion

There have been two primary explanations offered for FD/B’s failure to relate to externalizing outcomes characteristic of psychopathy. The first hypothesis, put forward by Blonigen (2013), suggests that FD/B’s relevance to these maladaptive outcomes may only be seen at high levels of these traits. To date, this hypothesis has been tested three times with each finding no evidence to suggest a curvilinear relation with behavioral outcomes typical of psychopathy (Gatner et al., 2016; Vize et al., 2016, Weiss et al., 2019). Second, some researchers have argued that FD/B may be more relevant to these outcomes through synergistic interactions with antagonism/meanness and disinhibition in predicting maladaptive behavioral outcomes (Lilienfeld et al., 2012). However, these interactions do not materialize consistently (e.g., Gatner et al., 2016; Maples et al., 2014; Miller et al., 2016; Vize et al., 2016; cf. Marcus & Norris, 2014; Rock et al., 2013).

Effectively evaluating curvilinear effects at extreme ends of the trait continuum requires sufficient variability of psychopathic traits within the sample. Sample 1 included latent trait estimates more than 3 standard deviations greater than the population mean for all facets of FD. Each facet of FD also had over 100 individuals endorse at least three quarters of the item content. Sample 2 also included Boldness latent trait estimates exceeding 3 standard deviations beyond the population mean and over 100 individuals endorsed at least three quarters of the items. Equally important, the TRF suggested that individuals at extreme ends of all PPI-R and TriPM traits were reliably measured. Samples 1 and 2 therefore seem to include sufficient variance for providing a reasonable test of Blonigen’s (2013) curvilinearity hypothesis. There are fewer observations at the extreme ends of the PPI-R SCI facets in Sample 1, and Meanness and Disinhibition factors in Sample 2, but extreme responses were observed. Latent trait estimates greater than three were measured for three of the four PPI-R facets (the only exception being Carefree Nonplanfulness) and all TriPM facets. The amount of raw item endorsement was relatively small for Carefree Nonplanfulness in Sample 1 and Meanness and Disinhibition in Sample 2. This may have made some of the interaction hypotheses more difficult to identify. However, not all Sample 1 SCI factors were so limited. In Sample 1, Machiavellian Egocentricity and Rebellious Nonconformity included in excess of 30 individuals that endorsed at least three quarters of the scale.

Our analysis found no evidence for a curvilinear relation between FD/B and externalizing outcomes. However, it is noteworthy that in Sample 1, FD, and Fearlessness in particular, had linear effects comparable to the SCI domain. This is consistent with factor analyses which show that Fearlessness tends to load on both FD and SCI factors suggesting that, relative to other facets of the FD domain, it may contain more antagonism and disinhibition (both definitive features of SCI) content (Benning et al., 2003; Neumann et al., 2008; Weiss et al., 2019). This finding also highlights the importance of considering associations at the level of subscales rather than domain scores. One curvilinear effect was observed for Fearlessness in the prediction of antisocial behavior counts, but the effect was negative (indicating a reduction in antisocial behavior at higher ends of the spectrum) and small in magnitude. Additional quadratic effects were identified but these were observed for non-FD/B scales.

In line with the interaction hypothesis, a Boldness × Meanness interaction was found suggesting that Boldness, when paired with high Meanness, is positively associated with substance use. This was, however, the only statistically significant FD/B interaction observed in the two samples. These findings provide little evidence for the hypothesis that FD/B domains interact with other components of psychopathy to increase problematic externalizing behavior. These null effects are consistent with previous tests of the hypothesis (Gatner et al., 2016; Vize et al., 2016; Weiss et al., 2019). It is important to acknowledge that null findings are often difficult to interpret due to potential alternative explanations related to shortcomings in study design (e.g., small sample size, measurement error). While we cannot be certain that no such shortcomings have resulted in our failure to observe the hypothesized effects (see also the Limitations section below), unique aspects for of these analyses provide some support for the validity of our findings. The GGUM IRT model was utilized as it is the most precise statistical tool available for latent trait measurement. Furthermore, as verified by Monte Carlo simulations (see Supplemental Power Analysis in supplemental materials available online), our analyses were powered to detect relatively small effects.

Taking our results together with previous findings suggests that there is little evidence in favor of the interaction or curvilinearity hypotheses. Consistent with findings from relevant meta-analyses (Miller & Lynam, 2012; Sleep et al., 2019), our results suggest that FD/B manifest small positive linear associations with certain externalizing outcomes. For instance, the FD subscales were positively linearly related to substance use and antisocial behavior. No such linear associations were found for boldness.

PPI-R SCI subscales, PPI-R Coldheartedness, TriPM Disinhibition, and TriPM Meanness were also evaluated for curvilinear effects in the prediction of externalizing behavior. Statistically significant curvilinear relations were observed, but they were modest in size and negative in nearly all instances suggesting a reduction in, or perhaps more likely a leveling-off of externalizing behavior at high levels of the trait. Two of four PPI-R SCI scales (Machiavellian Egocentricity and Rebellious Nonconformity) and the PPI-R SCI factor revealed this pattern in the prediction of substance use in the undergraduate sample, as did Coldheartedness. Modest negative curvilinear effects were also found in the prediction of antisocial behavior, but only for PPI-R Machiavellian Egocentricity. These results are inconsistent with those of Weiss et al. (2019), who found modest positive curvilinear effects for Coldheartedness and Carefree Nonplanfulness in the prediction of Antisocial Behavior. Sample 2 revealed two curvilinear effects in the prediction of proactive aggression from TriPM Meanness and Disinhibition. Proactive aggression was more strongly associated with higher levels of Meanness and Disinhibition.

It is notable that in most instances, where curvilinear relations were found, the pattern of curvilinearity did not match the convex pattern that Blonigen (2013) hypothesized for FD/B domains. More often, the coefficient indicated a reduction in slope toward the extreme ends of the trait (see Supplemental Figures 1–7 available online). For example, externalizing behavior appears to decline with increasing levels of PPI Machiavellian Egocentricity.⁵ It is possible that the identified curvature is indicative of a plateau effect such that beyond certain levels of these traits there is no longer a meaningful increase in externalizing behaviors. However, because associations at the most extreme ends of the trait continuum are necessarily driven by a relative few, it is also possible that the reduction observed was an idiosyncratic sample-driven effect. Such an explanation is more likely for SCI factors as well as Meanness and Disinhibition, each of which had fewer extreme responses (see Supplemental Table 5 and Supplemental Histograms available online). It will be important for future research to address the degree to which these curvilinear relations replicate and what they mean for psychopathy’s relations with externalizing behaviors.

Regarding the interaction hypothesis, our results are consistent with previous null findings. Of the 18 interactions tested, only one effect consistent with predictions was identified. Meanness moderated the association between Boldness and substance abuse with only individuals high in Meanness showing a positive association between boldness and drug use. There was also an interaction between Meanness and Disinhibition such that individuals high in both traits had particularly high proactive aggression scores, but this interaction falls outside the scope of the FD/B interaction hypothesis.

Limitations

Some limitations of the current study and approach must be acknowledged. First, Sample 1 was composed of a relatively homogeneous undergraduate sample. Although samples of this nature are commonly employed in the study of psychopathy (e.g., Berg et al., 2015; Blagov et al., 2016), restriction of range could have affected the sizes of correlations found between psychopathy and criteria, particularly antisocial behavior and proactive aggression, which were both heavily skewed. The statistical methods used attempt to accommodate such distributions, but it is possible that stronger associations could be found in populations with higher base rates of such behaviors (e.g., forensic populations). Future examinations of the interaction hypothesis may require oversampling for externalizing behaviors.

Sample size may have been a related limitation. Although the study was well powered to detect most effects (see Supplemental Power Analysis available online), a larger sample would have allowed for additional data at the most extreme ends of the psychopathy spectrum. TriPM Disinhibition (d = −1.811) and TriPM Meanness (d = −1.16) means from our sample were more than 1 standard deviation below means from offender samples (Lilienfeld & Widows, 2005; Stanley et al., 2013). It is possible that interaction effects could be found in forensic samples with more extreme levels of such traits, although recent analyses limited to the PPI found no such interaction (Weiss et al., 2019). FD/B means from our samples adequately matched offender sample means suggesting that the range for the trait most central to the curvilinearity hypothesis was not particularly restricted.

The number of outcome measures should also be considered. As we are aware of few specific hypotheses regarding which specific antisocial behaviors FD/B is most likely to predict and given the sample size needed to assess the hypothesized effects, we chose to briefly measure a few prominent outcomes while emphasizing sample size. Some have argued that, given FD/B’s association with grandiose narcissism, it may predict antagonistic interpersonal behaviors observed among narcissistic individuals (e.g., entitlement, superiority, manipulativeness; Lilienfeld et al., 2012). It is possible that significant curvilinear or interaction effects could be found for FD/B were different outcome measures such as these collected.

Conclusion

These analyses used advanced statistical methods to evaluate two of the most prominent hypotheses for the role of FD/B in predicting externalizing outcomes. We found no evidence to support the curvilinearity hypothesis that FD/B traits are more strongly related to externalizing behaviors at higher levels. We also found little evidence to support the hypothesis that FD/B may interact with other components of psychopathy to result in externalizing behaviors. These hypotheses were tested in well-powered samples with common measures of FD/B.

No single study can provide definitive evidence of a null effect, but the absence of statistically significant findings in the present study are noteworthy. In this case, power simulations estimate 60% to 92% power to detect 22 boldness-related effects. Failure to find the hypothesized effects in the present paper contributes to a growing literature suggesting that FD/B may not be as relevant to psychopathy as some have proposed. If FD/B warrants a role in psychopathy, we believe it may be better understood as a peripheral feature, rather than an equal sized partner to antagonism/meanness and disinhibition, which are more strongly associated with the behavioral correlates that have made psychopathy the focus of such intense examination.

Supplemental Material

Supplemental_Material – Supplemental material for Fearless Dominance/Boldness Is Not Strongly Related to Externalizing Behaviors: An Item Response-Based Analysis

Supplemental material, Supplemental_Material for Fearless Dominance/Boldness Is Not Strongly Related to Externalizing Behaviors: An Item Response-Based Analysis by Michael L. Crowe, Brandon M. Weiss, Chelsea E. Sleep, Alexandra M. Harris, Nathan T. Carter, Donald R. Lynam and Joshua D. Miller in Assessment

Footnotes

Authors’ Note

Any opinions, findings, and conclusions expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Nathan T. Carter’s contribution was supported by the National Science Foundation under Grant SES-1561070.

Methodological Disclosure

We report how we determined our sample size, all data exclusions, and all manipulations.

ORCID iDs

Michael L. Crowe

Brandon M. Weiss

Joshua D. Miller

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Akaike

(1973). Maximum likelihood identification of Gaussian autoregressive moving average models. Biometrika, 60(2), 255-265. https://doi.org/10.1093/biomet/60.2.255

Atkins

D. C.

Baldwin

S. A.

Zheng

Gallop

R. J.

Neighbors

(2013). A tutorial on count regression and zero-altered count models for longitudinal substance use data. Psychology of Addictive Behaviors, 27(1), 166-177. https://doi.org/10.1037/a0029508

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(3), 340-350. https://doi.org/10.1037/1040-3590.15.3.340

Berg

J. M.

Hecht

L. K.

Latzman

R. D.

Lilienfeld

S. O.

(2015). Examining the correlates of the coldheartedness factor of the Psychopathic Personality Inventory–Revised. Psychological Assessment, 27(4), 1494-1499. https://doi.org/10.1037/pas0000129

Berg

J. M.

Lilienfeld

S. O.

Sellbom

(2017). The role of boldness in psychopathy: A study of academic and clinical perceptions. Personality Disorders: Theory, Research, and Treatment, 8(4), 319-328. https://doi.org/10.1037/per0000247

Blagov

P. S.

Patrick

C. J.

Oost

K. M.

Goodman

J. A.

Pugh

A. T.

(2016). Triarchic Psychopathy Measure: Validity in relation to normal-range traits, personality pathology, and psychological adjustment. Journal of Personality Disorders, 30(1), 71-81. https://doi.org/10.1521/pedi_2015_29_182

Bleidorn

Hopwood

C. J.

Ackerman

R A.

Witt

E. A.

Kandler

Riemann

Samuel

D. B.

Donnellan

M. B.

(2020). The healthy personality from a basic trait perspective. Journal of Personality and Social Psychology, 118(6), 1207. https://doi.org/10.1037/pspp0000231

Blonigen

D. M.

(2013). Is fearless dominance relevant to the construct of psychopathy? Reconciling the dual roles of theory and clinical utility. Personality Disorders: Theory, Research, and Treatment, 4(1), 87-88. https://doi.org/10.1037/a0027152

Bozdogan

(1987). Model selection and Akaike’s information criterion (AIC): The general theory and its analytical extensions. Psychometrika, 52(3), 345-370. https://doi.org/10.1007/BF02294361

10.

Carter

N. T.

Dalal

D. K

. (2010). An ideal point account of the JDI Work satisfaction scale. Personality and Individual Differences, 49(7), 743–748. https://doi.org/10.1016/j.paid.2010.06.019

11.

Carter

N. T.

Dalal

D. K.

Boyce

A. S.

O’Connell

M. S.

Kung

M.-C.

Delgado

K. M.

(2014). Uncovering curvilinear relationships between conscientiousness and job performance: How theoretically appropriate measurement makes an empirical difference. Journal of Applied Psychology, 99(4), 564-586. https://doi.org/10.1037/a0034688

12.

Carter

N. T.

Dalal

D. K.

Guan

LoPilato

A. C.

Withrow

S. A.

(2017). Item response theory scoring and the detection of curvilinear relationships. Psychological Methods, 22(1), 191-203. https://doi.org/10.1037/met0000101

13.

Carter

N. T.

Guan

Maples

J. L.

Williamson

R. L.

Miller

J. D.

(2016). The downsides of extreme conscientiousness for psychological well-being: The role of obsessive compulsive tendencies. Journal of Personality, 84(4), 510-522. https://doi.org/10.1111/jopy.12177

14.

Chalmers

R. P.

(2012). mirt: A multidimensional item response theory package for the R environment. Journal of Statistical Software, 48(6), 1-29. https://doi.org/10.18637/jss.v048.i06

15.

Chandler

Shapiro

(2016). Conducting clinical research using crowdsourced convenience samples. Annual Review of Clinical Psychology, 12, 53-81. https://doi.org/10.1146/annurev-clinpsy-021815-093623

16.

Cleckley

(1976). The mask of sanity: An attempt to clarify some issues about the so-called psychopathic personality (5th ed.). Mosby.

17.

Cohen

West

S. G.

Aiken

L. S.

(2002). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Routledge.

18.

Coxe

West

S. G.

Aiken

L. S.

(2009). The analysis of count data: A gentle introduction to Poisson regression and its alternatives. Journal of Personality Assessment, 91(2), 121-136. https://doi.org/10.1080/00223890802634175

19.

Crego

Widiger

T. A.

(2014). Psychopathy, DSM-5, and a caution. Personality Disorders: Theory, Research, and Treatment, 5(4), 335-347. https://doi.org/10.1037/per0000078

20.

Crego

Widiger

T. A.

(2016). Cleckley’s psychopaths: Revisited. Journal of Abnormal Psychology, 125(1), 75-87. https://doi.org/10.1037/abn0000130

21.

Crowe

M. L.

Sleep

C. E.

Carter

N. T.

Campbell

W. K.

Miller

J. D.

(2018). Self-esteem and narcissism: An item response theory analysis of curvilinearity. Personality and Individual Differences, 128, 16-20. https://doi.org/10.1016/j.paid.2018.02.016

22.

Drasgow

Chernyshenko

O. S.

Stark

(2010). 75 Years after Likert: Thurstone was right! Industrial and Organizational Psychology, 3(4), 465-476. https://doi.org/10.1111/j.1754-9434.2010.01273.x

23.

Few

L. R.

Miller

J. D.

Lynam

D. R.

(2013). An examination of the factor structure of the Elemental Psychopathy Assessment. Personality Disorders: Theory, Research, and Treatment, 4(3), 247-253. https://doi.org/10.1037/per0000016

24.

Gatner

D. T.

Douglas

K. S.

Hart

S. D.

(2016). Examining the incremental and interactive effects of boldness with meanness and disinhibition within the triarchic model of psychopathy. Personality Disorders: Theory, Research, and Treatment, 7(3), 259-268. https://doi.org/10.1037/per0000182

25.

Hare

R. D.

(1980). A research scale for the assessment of psychopathy in criminal populations. Personality and Individual Differences, 1(2), 111-119. https://doi.org/10.1016/0191-8869(80)90028-8

26.

Hare

R. D.

(2003). The psychopathy checklist revised. Multi-Health Systems.

27.

Hare

R. D.

Neumann

C. S.

(2008). Psychopathy as a clinical and empirical construct. Annual Review of Clinical Psychology, 4, 217-246. https://doi.org/10.1146/annurev.clinpsy.3.022806.091452

28.

Krusemark

E. A.

Campbell

W. K.

Crowe

M. L.

Miller

J. D.

(2018). Comparing self-report measures of grandiose narcissism, vulnerable narcissism, and narcissistic personality disorder in a male offender sample. Psychological Assessment, 30(7), 984-990. https://doi.org/10.1037/pas0000579

29.

LaPalme

Tay

Wang

(2017). A within-person examination of the ideal point response process. Psychological Assessment, 30(5), 567-581. https://doi.org/10.1037/pas0000499

30.

Levenson

M. R.

Kiehl

K. A.

Fitzpatrick

C. M.

(1995). Assessing psychopathic attributes in a noninstitutionalized population. Journal of Personality and Social Psychology, 68(1), 151-158. https://doi.org/10.1037/0022-3514.68.1.151

31.

Lilienfeld

S. O.

Andrews

B. P.

(1996). Development and preliminary validation of a self-report measure of psychopathic personality traits in noncriminal population. Journal of Personality Assessment, 66(3), 488-524. https://doi.org/10.1207/s15327752jpa6603_3

32.

Lilienfeld

S. O.

Patrick

C. J.

Benning

S. D.

Berg

Sellbom

Edens

J. F.

(2012). The role of fearless dominance in psychopathy: Confusions, controversies, and clarifications. Personality Disorders: Theory, Research, and Treatment, 3(3), 327-340. https://doi.org/10.1037/a0026987

33.

Lilienfeld

S. O.

Smith

S. F.

Sauvigné

K. C.

Patrick

C. J.

Drislane

L. E.

Latzman

R. D.

Krueger

R. F.

(2016). Is boldness relevant to psychopathic personality? Meta-analytic relations with non-Psychopathy Checklist-based measures of psychopathy. Psychological Assessment, 28(10), 1172-1185. https://doi.org/10.1037/pas0000244

34.

Lilienfeld

S. O.

Widows

M. R.

(2005). PPI-R professional manual. Psychological Assessment Resources.

35.

Lykken

D. T.

(1995). The antisocial personalities. Lawrence Erlbaum.

36.

Lynam

D. R.

Gaughan

Miller

J. D.

Miller

Mullins-Sweat

Widiger

T. A.

(2011). Assessing the basic traits associated with psychopathy: Development and validation of the Elemental Psychopathy Assessment. Psychological Assessment, 23(1), 108-124. https://doi.org/10.1037/a0021146

37.

Lynam

D. R.

Miller

J. D.

(2012). Fearless dominance and psychopathy: A response to Lilienfeld et al. Personality Disorders: Theory, Research, and Treatment, 3(3), 341-353. https://doi.org/10.1037/a0028296

38.

Maples

J. L.

Miller

J. D.

Fortune

MacKillop

Campbell

W. K.

Lynam

D. R.

Lance

C. E.

Goodie

A. S.

(2014). An examination of the correlates of fearless dominance and self-centered impulsivity among high-frequency gamblers. Journal of Personality Disorders, 28(3), 379-393. https://doi.org/10.1521/pedi_2013_27_125

39.

Marcus

D. K.

Fulton

J. J.

Edens

J. F.

(2012). “The two-factor model of psychopathic personality: Evidence from the Psychopathic Personality Inventory”: Correction to Marcus, Fulton, and Edens (2011). Personality Disorders: Theory, Research, and Treatment, 3(2), Article 195. https://doi.org/10.1037/a0027132

40.

Marcus

D. K.

Norris

A. L.

(2014). A new measure of attitudes toward sexually predatory tactics and its relation to the triarchic model of psychopathy. Journal of Personality Disorders, 28(2), 247-261. https://doi.org/10.1521/pedi_2013_27_118

41.

McFadden

(1974). Conditional logit analysis of qualitative choice behavior. In Zarembka

(Ed.), Frontiers in econometrics (pp. 105-142). Academic Press.

42.

McFadden

(1979). Quantitative methods for analyzing travel behaviour of individuals: Some recent developments. In Hensher

Stopher

(Eds.), Behavioural travel modelling (pp. 279-318). Croom Helm.

43.

Miller

J. D.

Crowe

M. L.

Weiss

Maples-Keller

J. L.

Lynam

D. R.

(2017). Using online, crowdsourcing platforms for data collection in personality disorder research: The example of Amazon’s Mechanical Turk. Personality Disorders: Theory, Research, and Treatment, 8(1), 26-34. https://doi.org/10.1037/per0000191

44.

Miller

J. D.

Few

L. R.

Seibert

L. A.

Watts

Zeichner

Lynam

D. R.

(2012). An examination of the Dirty Dozen measure of psychopathy: A cautionary tale about the costs of brief measures. Psychological Assessment, 24(4), 1048-1053. https://doi.org/10.1037/a0028583

45.

Miller

J. D.

Lamkin

Maples-Keller

J. L.

Lynam

D. R.

(2016). Viewing the triarchic model of psychopathy through general personality and expert-based lenses. Personality Disorders: Theory, Research, and Treatment, 7(3), 247-258. https://doi.org/10.1037/per0000155

46.

Miller

J. D.

Lynam

D. R.

(2003). Psychopathy and the five-factor model of personality: A replication and extension. Journal of Personality Assessment, 81(2), 168-178. https://doi.org/10.1207/S15327752JPA8102_08

47.

Miller

J. D.

Lynam

D. R.

(2012). An examination of the Psychopathic Personality Inventory’s nomological network: A meta-analytic review. Personality Disorders: Theory, Research, and Treatment, 3(3), 305-326. https://doi.org/10.1037/a0024567

48.

Miller

J. D.

Lynam

D. R.

Siedor

Crowe

Campbell

W. K.

(2018). Consensual lay profiles of narcissism and their connection to the Five-Factor Narcissism Inventory. Psychological Assessment, 30(1), 10-18. https://doi.org/10.1037/pas0000460

49.

Miller

J. D.

Lynam

D. R.

Vize

Crowe

Sleep

Maples-Keller

J. L.

Few

L. R.

Campbell

W. K.

(2018). Vulnerable narcissism is (mostly) a disorder of neuroticism. Journal of Personality, 86(2), 186-199. https://doi.org/10.1111/jopy.12303

50.

Miller

J. D.

Lynam

D. R.

Widiger

T. A.

Leukefeld

(2001). Personality disorders as extreme variants of common personality dimensions: Can the Five Factor Model adequately represent psychopathy? Journal of Personality, 69(2), 253-276. https://doi.org/10.1111/1467-6494.00144

51.

Miller

J. D.

Maples-Keller

J. L.

Lynam

D. R.

(2016). An examination of the three components of the Psychopathic Personality Inventory: Profile comparisons and tests of moderation. Psychological Assessment, 28(6), 692-701. https://doi.org/10.1037/pas0000221

52.

Miller

J. D.

Sleep

C. E.

Crowe

M. L.

Lynam

D. R.

(2020). Psychopathic boldness: Narcissism, self-esteem, or something in between? Personality and Individual Differences, 155, 1-5. https://doi.org/10.1016/j.paid.2019.109761

53.

Muraki

(1992). A generalized partial credit model: Application of an EM algorithm. Applied Psychological Measurement, 16(2), 159-176. https://doi.org/10.1177/014662169201600206

54.

Murphy

Lilienfeld

Skeem

Edens

J. F.

(2016). Are fearless dominance traits superfluous in operationalizing psychopathy? Incremental validity and sex differences. Psychological Assessment, 28(12), 1597-1607. https://doi.org/10.1037/pas0000288

55.

Neumann

C. S.

Malterer

M. B.

Newman

J. P.

(2008). Factor structure of the Psychopathic Personality Inventory (PPI): Findings from a large incarcerated sample. Psychological Assessment, 20(2), 169-174. https://doi.org/10.1037/1040-3590.20.2.169

56.

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(6), 644-664. https://doi.org/10.1111/jopy.12126

57.

Ozer

D. J.

Benet-Martinez

(2006). Personality and the prediction of consequential outcomes. Annual Review of Psychology, 57, 401-421. https://doi.org/10.1146/annurev.psych.57.102904.190127

58.

Patrick

C. J.

(2010). Operationalizing the triarchic conceptualization of psychopathy: Preliminary description of brief scales for assessment of boldness, meanness, and disinhibition [Test manual]. https://pdfs.semanticscholar.org/fb4c/c0fdae0840553500ee091579ae76a6477e8e.pdf?_ga=2.123029768.222663642.1581144234-491481082.1579597845

59.

Patrick

C. J.

Fowles

D. C.

Krueger

R. F.

(2009). Triarchic conceptualization of psychopathy: Developmental origins of disinhibition, boldness, and meanness. Developmental Psychopathology, 21(3), 913-938. https://doi.org/10.1017/S0954579409000492

60.

Paulhus

D. L.

Neumann

C. S.

Hare

R. D.

(2017). Self-Report Psychopathy Scale: Technical manual (4th ed.). Multi-Health Systems.

61.

Raftery

A. E.

(1995). Bayesian model selection in social research. Sociological Methodology, 25, 111-163. https://doi.org/10.2307/271063

62.

Raine

Dodge

Loeber

Gatzke-Kopp

Lynam

Reynolds

Stouthamer-Loeber

Liu

(2006). The Reactive–Proactive Aggression Questionnaire: Differential correlates of reactive and proactive aggression in adolescent boys. Aggressive Behavior, 32(2), 159-171. https://doi.org/10.1002/ab.20115

63.

Reise

S. P.

(2010). Thurstone might have been right about attitudes, but Drasgow, Chernyshenko, and Stark fail to make the case for personality. Industrial and Organizational Psychology, 3(4), 485-488. https://doi.org/10.1111/j.1754-9434.2010.01276.x

64.

Roberts

J. S.

Donoghue

J. R.

Laughlin

J. E.

(2000). A general item response theory model for unfolding unidimensional polytomous responses. Applied Psychological Measurement, 24(1), 3-32. https://doi.org/10.1177/01466216000241001

65.

Roberts

J. S.

Fang

H.-r.

Cui

Wang

(2006). GGUM2004: A Windows-based program to estimate parameters in the generalized graded unfolding model. Applied Psychological Measurement, 30(1), 64-65. https://doi.org/10.1177/0146621605280141

66.

Roberts

J. S.

Laughlin

J. E.

Wedell

D. H.

(1999). Validity issues in the Likert and Thurstone approaches to attitude measurement. Educational and Psychological Measurement, 59(2), 211-233. https://doi.org/10.1177/00131649921969811

67.

Rock

R. C.

Sellbom

Ben-Porath

Y. S.

Salekin

R. T.

(2013). Concurrent and predictive validity of psychopathy in a batterers’ intervention sample. Law and Human Behavior, 37(3), 145-154. https://doi.org/10.1037/lhb0000006

68.

Ross

S. R.

Benning

S. D.

Patrick

C. J.

Thompson

Thurston

(2009). Factors of the Psychopathic Personality Inventory: Criterion-related validity and relationship to the BIS/BAS and five-factor models of personality. Assessment, 16(1), 71-87. https://doi.org/10.1177/1073191108322207

69.

Sellbom

Phillips

T. R.

(2013). An examination of the triarchic conceptualization of psychopathy in incarcerated and nonincarcerated samples. Journal of Abnormal Psychology, 122(1), 208-214. https://doi.org/10.1037/a0029306

70.

Shou

Sellbom

(2018). Psychometric properties of the Triarchic Psychopathy Measure: An item response theory approach. Personality Disorders: Theory, Research, and Treatment, 9(3), 217-227.

71.

Skeem

J. L.

Cooke

D. J.

(2010). Is criminal behavior a central component of psychopathy? Conceptual directions for resolving the debate. Psychological Assessment, 22(2), 433-455. https://doi.org/10.1037/a0008512

72.

Sleep

C. E.

Weiss

Lynam

D. R.

Miller

J. D.

(2019). An examination of the Triarchic Model of Psychopathy’s nomological network: A meta-analytic review. Clinical Psychology Review, 71, 1-26. https://doi.org/10.1016/j.cpr.2019.04.005

73.

Smith

S. F.

Lilienfeld

S. O.

Coffey

Dabbs

J. M.

(2013). Are psychopaths and heroes twigs off the same branch? Evidence from college, community, and presidential samples. Journal of Research in Personality, 47(5), 634-646. https://doi.org/10.1016/j.jrp.2013.05.006

74.

Stanley

J. H.

Wygant

D. B.

Sellbom

(2013). Elaborating on the construct validity of the Triarchic Psychopathy Measure in a criminal offender sample. Journal of Personality Assessment, 95(4), 343-350. https://doi.org/10.1080/00223891.2012.735302

75.

Stark

(2007). MODFIT: Plot theoretical item response functions and examine the fit of dichotomous or polytomous IRT models to response data [Computer software]. University of Illinois at Urbana–Champaign.

76.

Stark

Chernyshenko

O. S.

Drasgow

Williams

B. A.

(2006). Examining assumptions about item responding in personality assessment: Should ideal point methods be considered for scale development and scoring? Journal of Applied Psychology, 91(1), 25-39. https://doi.org/10.1037/0021-9010.91.1.25

77.

Vize

C. E.

Collison

K. L.

Crowe

M. L.

Campbell

W. K.

Miller

J. D.

Lynam

D. R.

(2017). Using dominance analysis to decompose narcissism and its relation to aggression and externalizing outcomes. Assessment, 26(2), 260-270. https://doi.org/10.1177/1073191116685811

78.

Vize

C. E.

Lynam

D. R.

Lamkin

Miller

J. D.

Pardini

(2016). Identifying essential features of juvenile psychopathy in the prediction of later antisocial behavior: Is there an additive, synergistic, or curvilinear role for Fearless Dominance? Clinical Psychological Science, 4(3), 572-590. https://doi.org/10.1177/2167702615622384

79.

Weiss

Crowe

M. L.

Harris

Carter

N. T.

Lynam

Watts

A. L.

Lilienfeld

S. O.

Skeem

J. L.

Miller

J. D.

(2019). Examining hypothesized interactive and curvilinear relations between psychopathic traits and externalizing problems in an offender sample using item response-based analysis. Journal of Abnormal Psychology, 128(7), 689-699. https://doi.org/10.1037/abn0000462

80.

Witt

E. A.

Donnellan

M. B.

Blonigen

D. M.

Krueger

R. F.

Conger

R. D.

(2009). Assessment of Fearless Dominance and Impulsive Antisociality via normal personality measures: Convergent validity, criterion validity, and developmental change. Journal of Personality Assessment, 91(3), 265-276. https://doi.org/10.1080/00223890902794317

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

2.56 MB

Fearless Dominance/Boldness Is Not Strongly Related to Externalizing Behaviors: An Item Response-Based Analysis

Abstract

Keywords

Measures of Fearless Dominance/Boldness

Controversy of Fearless Dominance/Boldness

Study Aims

Method

Sample 1: Participants and Procedure

Sample 2: Participants and Procedure

Measures

Psychopathic Personality Inventory–Revised (PPI-R). 3

Triarchic Psychopathy Measure

Crime and Analogous Behavior Scale

Reactive and Proactive Aggression Questionnaire

Validity Scales

Data Analysis

Scoring Methods

Dominance Versus Ideal Point Models

Model Estimation and Fit

Regression Models

Results

Evaluating Curvilinearity

Substance Use

Antisocial Behavior

Reactive and Proactive Aggression

Evaluating Interactions

Discussion

Limitations

Conclusion

Supplemental Material

Supplemental_Material – Supplemental material for Fearless Dominance/Boldness Is Not Strongly Related to Externalizing Behaviors: An Item Response-Based Analysis

Footnotes

Authors’ Note

Declaration of Conflicting Interests

Funding

Methodological Disclosure

ORCID iDs

Supplemental Material

Notes

References

Supplementary Material

Psychopathic Personality Inventory–Revised (PPI-R).³