An Examination of Triarchic Psychopathy Constructs in Female Offenders

Abstract

This study sought to expand scientific knowledge on psychopathic personality traits in female offenders by evaluating the relationship between MMPI-2-RF triarchic scales and self-reported external variables in a sample of 205 female offenders. Results indicated that boldness was inversely related to internalizing dysfunction, including suicidal behavior; psychosis; youth conduct problems; problems stemming from alcohol use; and a history of outpatient mental health treatment. Meanness was positively related to internalizing dysfunction as well as youth conduct problems, anger, prison disciplinary reports, and psychosis. Disinhibition was associated with a history of abuse in childhood, suicidal behavior, internalizing dysfunction, problems associated with alcohol and drug use, family history of mental illness, prison disciplinary reports for violence, number of previous criminal charges, and anger. Consistent with views of psychopathy as a configural condition, interactive effects of boldness with disinhibition and meanness were observed for multiple key external variables (e.g., conduct problems, substance use, non-suicidal self-injurious behavior). This study provides further evidence for the triarchic conceptualization of psychopathy in female offenders and lends additional support for the validity of MMPI-2-RF triarchic psychopathy scales.

An Examination of Triarchic Psychopathy Constructs in Female Offenders

Psychopathy is regarded as a compendium of interrelated pathological personality traits adversely impacting interpersonal, emotional, and behavioral functioning. Individuals with psychopathic personality traits generally evince high degrees of callous, manipulative, socially domineering, and disinhibited behavior with deficits in empathy and negative emotionality (Patrick, Fowles, Krueger, 2009). A dearth of information exists about these traits in women, with a specific gap in findings from incarcerated female samples (Nicholls & Petrila, 2005). The current study aims to address this shortcoming and extend prior research by examining external correlates of the triarchic model of psychopathy in a sample of female offenders.

Psychopathy in Female Populations

Much of the available research on the occurrence of psychopathic traits in women has examined possible gender differences in the nomological network of psychopathy. Differential prevalence rates for male and female offenders have been reported, with incarcerated females showing slightly lower rates of psychopathy using conventional cutoffs (10–23%; Salekin, Rogers, Ustad, & Sewell, 1998; Vitale, Smith, Brinkley, & Newman, 2002), relative to male inmates (15–30%; Salekin et al., 1998). Some researchers have suggested that gender biases in the diagnosis and measurement of psychopathy are relevant to consider in reference to these base-rate findings. Forouzan and Cooke (2005) described how gender differences in symptomatology are reported for many personality disorders, and as psychopathy is often understood as a disorder of personality functioning, these biases may be expected to extend to this construct as well. Nicholls and Petrila (2005) reported that females engage in less crime and violence than men, and as a result, instruments designed to measure psychopathy that rely heavily on specific criminal behaviors will likely produce lower scores for women and therefore be under representative. Additionally, Nicholls and Petrila (2005) also suggested that differential gender socialization may result in equivalent levels of psychopathy being expressed differently in males and females (e.g., males may be more likely to exhibit behavioral features of psychopathy, whereas females may display more personality-based characteristics).

Research exploring typical presentations of psychopathic personality traits for females and males finds that females are more likely to exhibit stronger interpersonal skills, utilize different means of manipulation (sexuality vs. physical strength), and show more relational aggression (Kreis & Cooke, 2011). Differential symptomatic features such as manipulativeness and emotional instability (related to meanness and disinhibition respectively; Drislane, Patrick, & Arsal, 2014; Patrick et al., 2009) were also more likely to be reported for females in this study, suggesting that females with psychopathic traits may be more subtle and utilize relational skills for purposes of exploitation and dominance (Kreis & Cooke, 2011). Additional differences at the genetic level have been reported, including findings from a larger study on etiological configurations of psychopathic personality traits and associations with internalizing and externalizing pathology suggesting that fearless-dominance (reflecting boldness) relates to externalizing in males and impulsive-antisociality (associated with disinhibition) is associated with internalizing in females (Blonigen, Hicks, Krueger, Patrick, Iacono, 2005). Taken together, this highlights the limited generalizability of results from male samples and underscores the need for further study of psychopathic traits in females. Specifically, as recent findings indicated that the annual growth rate of females under correctional supervision was twice that of males (Glaze & Kaeble, 2014), research exploring the triarchic psychopathy model in incarcerated females is of significance.

Of the studies examining psychopathic personality traits in female offenders, variables such as trauma history, suicidality, substance abuse, emotion processing and expression, anger, aggression, and violence are of particular relevance and often under examined (Nicholls & Petrila, 2005; Rogstad & Rogers, 2008; Sellbom, Donnelly, Rock, Phillips, & Ben-Porath, 2017; Verona, Hicks, & Patrick, 2005). Specifically, history of trauma has been established as relating to psychological distress and thus has become an increasingly important aspect of understanding the mental health and criminal behavior of female offenders (Lynch, Fritch, & Heath, 2012). Trauma exposure of incarcerated women often includes instances of interpersonal victimization, such as domestic violence, sexual assault, and childhood physical and sexual abuse (DeHart, Lynch, Belknap, Dass-Brailsford, & Green, 2014). Female inmates report more traumatic experiences than males, with women indicating histories of victimization at alarmingly high rates and often chronicling multiple instances of abuse and violence (DeHart et al., 2014; Green, Miranda, Daroowalla, & Siddique, 2005; Lynch et al., 2012). Furthermore, research suggests a relationship between prior history of physical and sexual abuse and psychopathic traits within female samples (Hicks, Vaidyanathan, & Patrick, 2010; Verona et al., 2005), making the examination of these traits within an incarcerated female population especially pertinent. The rapidly escalating rates of imprisonment along with increased prevalence of psychopathology and trauma experiences in females emphasizes the importance of understanding and assessing various constructs relevant to psychopathy in these samples, rather than generalizing findings from male offenders to female populations.

Triarchic Psychopathy Model

Patrick and colleagues (2009) proposed the triarchic psychopathy model to reconcile historical perspectives of the construct along with contemporary debates regarding its scope and boundaries. This conceptualization hypothesizes that psychopathy reflects three distinct trait-based phenotypic domains: boldness, meanness, and disinhibition (Patrick et al., 2009). The triarchic model links psychopathy to both adaptive and maladaptive personality features and provides a foundation for reconciling heterogeneity among conceptions of psychopathic personality.

The construct of boldness is comprised of self-assurance and self-efficacy in stressful and potentially dangerous situations (Patrick et al., 2009). Individuals with high levels of boldness tend to be dominant and assertive in their interpersonal and social interactions. They seek novelty and adventure and remain relatively calm and unaffected in dangerous or fearful situations (Hall et al., 2014).

Extant research suggests positive associations between boldness and criterion measures of narcissism, sensation seeking, and behavioral activation system (BAS) drive (Brislin, Drislane, Smith, Edens, & Patrick, 2015; Sellbom & Phillips, 2013) and inverse associations with negative affect, depression, and suicidal desire (Brislin et al., 2015; Harrop et al., 2017). A comparison of boldness with normal-range personality characteristics demonstrates positive associations with extraversion, openness, social potency, well-being, achievement and stress reaction (Brislin et al., 2015; Patrick & Drislane, 2015), and inverse associations with facets of personality such as neuroticism, anxiety, and antagonism (Brislin et al., 2015; Patrick & Drislane, 2015). Additionally, the boldness domain has exhibited neurobehavioral associations with threat sensitivity (related to amygdala activity) and decreased startle response to aversive stimuli (Benning, Patrick, & Iacono, 2005).

Meanness consists of deficient empathy, exploitativeness, and cruelty (Patrick et al., 2009). It is related to contempt for affective interactions and intimate relationships as well as the absence of compassion toward others’ suffering (Hall et al., 2014). Individuals with high levels of this trait demonstrate feelings of superiority and a cynical worldview, which are associated with the tendency to view personal and social interactions as a zero sum game and thus, are competitive in their exchanges with others. In addition, these individuals may derive certain pleasures from the destructive and harmful impact of their actions (Hall et al., 2014).

Meanness is related to expected criterion variables such as aggressive forms of externalizing behavior, alienation, antagonism, absence of remorse, superficiality of emotion, external expression of anger, and detachment (Brislin et al., 2015; Patrick & Drislane, 2015). Meanness is negatively associated with agreeableness, conscientiousness, and control (Brislin et al., 2015; Patrick & Drislane, 2015). Research on neurobehavioral correlates of meanness suggests reduced emotional reactivity to stress and amygdala response to fearful expressions (Brislin et al., 2018; Patrick & Drislane, 2015).

Disinhibition reflects impulsive and irresponsible behaviors and disconstraint (Patrick et al., 2009). Individuals with elevated levels of disinhibition are often characterized as making rash decisions without consideration of the consequences of their behavior. They appear to be motivated by the possibility of transient rewards and frequently engage in problematic substance use and violations of established rules or laws (Patrick et al., 2009).

Associations with criterion measures of sensation seeking, BAS drive, antisocial personality disorder (ASPD), substance use disorders (SUDs), anxiety, and depression have been found (Patrick & Drislane, 2015; Sellbom & Phillips, 2013), as well as positive relations with neuroticism, alienation, stress reaction, capability for suicide, anger expression, and aggression (Brislin et al., 2015; Harrop et al., 2017; Patrick & Drislane, 2015). Moreover, disinhibition exhibits negative associations with conscientiousness, agreeableness, and control (Patrick & Drislane, 2015). Disinhibition is hypothesized to be associated with neurobehavioral dimensions reflecting poor inhibitory control, relating to deficits in fronto-cortical regions (Patrick & Drislane, 2015).

Recently, Sellbom and colleagues (2016) utilized items from the Minnesota Multiphasic Personality Inventory −2 -Restructured Form (MMPI-2-RF; Ben-Porath & Tellegen, 2008/2011) to construct MMPI-2-RF triarchic scales using data from offender and non-offender samples with both male and female participants. Use of the MMPI-2-RF scales for measurement of psychopathic personality traits is especially advantageous as the MMPI-2 (Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 1989; Butcher, Graham, Ben-Porath, Tellegen & Dahlstrom, 2001), from which the items of the MMPI-2-RF are derived, is one of the most widely utilized inventories of personality and psychopathology within clinical/forensic populations (Archer, Buffington-Vollum, Stredny, & Handel, 2006). The MMPI-2-RF triarchic scales show expected construct validity in female correctional inmates and male and female university students (Sellbom et al., 2016). Many of these findings were replicated and extended in Kutchen et al. (2017). However, as there are limits to the generalizability of results from studies of psychopathic traits in differing populations (e.g., Colins, Bijttebier, Broekaert, & Andershed, 2014); further validation of the triarchic domains in samples with disproportionately lower rates of representation is warranted. Specifically, few studies have explored the triarchic model within a female correctional sample and even fewer have included variables uniquely relevant to the understanding of psychopathy in this population (e.g., abuse and trauma history; Nicholls & Petrila, 2005; Verona et al.,2005).

The Current Study

The purpose of the present study was to expand scientific knowledge on psychopathic personality traits in female offenders by evaluating the relationship between MMPI-2-RF triarchic scales and external variables in a sample of female offenders. The available research on incarcerated women suggests that variables such as history of abuse, psychopathological symptoms, and factors related to offender behavior (e.g. expression of anger and aggression) may provide further elucidation of the triarchic domains within this population (Gottfried, Anestis, Dillon, & Carbonell, 2016; Nicholls & Petrila, 2005).

We hypothesized that the MMPI-2-RF Boldness scale would show an inverse relationship to several variables of interest. As the traits associated with the MMPI-2-RF Boldness scale relate to emotional resilience and lack of psychopathology (Sellbom et al., 2016), we expected to find an inverse association with symptoms of internalizing psychopathology (e.g. mood and anxiety symptoms, PTSD), suicidal behavior, family history of mental illness, mental health treatment, and anger experience and expression (Brislin et al., 2015; Drislane et al., 2014; Harrop et al., 2017; Sellbom, 2015; Venables et al., 2015). We hypothesized the MMPI-2-RF Meanness scale to be positively associated with violence and the verbal or physical expression of anger and negatively correlated with the ability to hold in or control anger (Brislin et al., 2015; Patrick et al., 2009). Additionally, we predicted the MMPI-2-RF Disinhibition scale would show a positive relationship with externalizing dysfunction (e.g., substance abuse, aggression, violence), past suicide attempt(s), suicidal ideation, and non-suicidal self-injury, and trauma and abuse (Venables & Patrick, 2012; Verona et al., 2005). More specifically, we anticipated a positive association between MMPI-2-RF Disinhibition and scales of conduct disorder, drinking symptoms, tolerance to alcohol, alcohol withdrawal, alcohol abuse, drug withdrawal, and drug abuse. Given that contemporary views of psychopathy suggest it is a configural condition involving the intersection of multiple trait dispositions (cf. Lilienfeld, 2013), we further tested for the interactive (synergistic) effect of triarchic dimension in predicting clinical criteria. As the analyses of the interactions were novel and exploratory in nature, no a priori hypotheses were made.

Methods

Participants and Procedures

Study participants were 316 female inmates from a state correctional facility with security classifications ranging from minimum to maximum security. One hundred and one participants were excluded from the analyses based on the below exclusionary criteria, resulting in a sample size of 215. The excluded participants did not significantly differ on level of education or age. Excluded participants were more likely to identify with an ethnic or racial minority group χ2(3, N=312) =22.85, p<.001, ϕ=.27. The mean age of the sample was 33.89 (SD = 8.15, range = 19–64). The sample demographics were consistent with previously reported studies of female state prisoners (Snell & Morton, 1994). See Table 1 for a summary of demographic data for included participants.

Table 1.

Demographic and MMPI-2-RF triarchic scale data in the final valid sample

	N	Range	Mean (SD)	α
Age (years)	215	19–64	33.89 (8.15)
Education (years)	174	6–16	10.78 (1.72)
# of Past Offenses	174	0–48	7.51 (7.90)
MMPI-2-RF Meanness	215	1–23	11.97 (4.79)	.81
MMPI-2-RF Disinhibition	215	1–12	6.13 (2.71)	.64
MMPI-2-RF Boldness	215	1–20	11.93 (3.28)	.66
Race/Ethnicity	N	Percentage
African American	97	45.8%
Caucasian	107	50.5%
Hispanic/Latino	7	3.3%
Native American	1	0.5%
High School Diploma (N=185)
Yes	75	40.5%
No	110	59.5%
GED (N=171)
Yes	37	21.6%
No	134	78.4%
Violent Offense (N=185)
Yes	104	56.2%
No	81	43.8%

Self-report validity was assessed using relevant scales from the MMPI-2-RF. Based on Ben-Porath & Tellegen (2008/2011), protocols which met the following criteria were excluded from the final dataset: Cannot Say (CNS-r) raw scores of 18 or greater missing responses, Variable Response Inconsistency-revised (VRIN-r) Scale T-scores of 80 or above, True Response Inconsistency-revised (TRIN-r) T-scores of 80 or above, Infrequent Responses (F-r) scale T-scores =120, and Infrequent Psychopathology Responses (Fp-r) T-scores ≥100. Following the standard validity scale interpretive strategy recommended by Ben-Porath & Tellegen (2008/2011), 13 profiles were excluded due to invalid CNS-r scores. After this exclusion, 44 were excluded due to high VRIN-r scores, followed by excluding 32 due to TRIN-r, six were excluded due to high F-r scores, and six were excluded due to high Fp-r scores.

Potential participants were randomly selected from the institutional roster and invited to participate in the present study. Participants provided informed and written consent, and study procedures were approved by both the correctional institution and a university’s Institutional Review Board. Study measures were administered in groups of five to 20 women.

Measures

MMPI-2-RF.

The Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI–2–RF; Ben-Porath & Tellegen, 2008/2011; Tellegen & Ben-Porath, 2008/2011) is a 338-item true/false instrument designed to assess for psychopathology and individual personality traits. The MMPI-2-RF consists of six sets of scales: Validity, Higher-Order, Restructured Clinical, Specific Problems, Interest, and a revised version of the Personality Psychopathology Five (PSY-5; Harkness et al., 2014).

The current sample was administered the 567-item Minnesota Multiphasic Personality Inventory-2 (MMPI-2; Butcher et al., 1989). MMPI-2 responses were rescored and transformed to MMPI-2-RF scores. The validity of this strategy has been demonstrated in prior studies (e.g., Tarescavage, Alosco, Ben-Porath, Wood, & Luna-Jones, 2015; Tellegen & Ben-Porath, 2008/2011).Triarchic psychopathy scores were computed from MMPI-2-RF items that have been shown to provide effective estimates of triarchic psychopathy constructs as described by Sellbom and colleagues (2016), including Boldness, Meanness, and Disinhibition scores (see Sellbom et al., 2016, for more details). Table 1 includes more information about these scales.

Quickview Social History Questionnaire (Quickview: Giannetti, 1987/1994).

The Quickview Social History Questionnaire is a 235 multiple-choice questionnaire that contains items regarding the social and clinical history of participants. The scales computed for the purpose of addressing the primary hypotheses of the present study are described below.

Physical, sexual, and emotional abuse.

Abuse was assessed separately by perpetrator and type (e.g., physical, sexual). Table 2 includes information about each abuse variable.

Table 2.

Correlation and regression analyses of abuse and family history of mental illness variables

Scale/Variable			RF-Meanness		RF-Disinhibition		RF-Boldness
	Range (α)	Mean (SD)/ %	r	ORb IRRc	r	ORb IRRc	r	ORb IRRc	R2(ab) χ2(c)	N
Abuse variables
Abuse by mother	0–15 (.84)	1.9 (2.56)	.09	1.00c	.20**	1.11 c**	-.04	.99c	9.61c	206
Abuse by father	0–16 (.87)	1.14 (2.28)	.03	.98 c	.21**	1.19 c**	.13	.94 c*	25.39c	206
Total abuse by parent	0–26 (.88)	3.07 (3.95)	.08	.99c	.25**	1.14 c**	-.10	.97 c	18.21c	206
Childhood sexual abuse	0–1	48.7%	.01	.95b	.18*	1.20b*	-.14	.92b	.03b	195
Adulthood rape	0–1	56.1%	.04	1.00b	.09	1.07b	-.13	.92b	.02b	198
Abuse by partner	0–1	21.7%	.05	.09b	-.03	-.07b	-.10	-.11b	.02b	203
Family history of mental illness
Mother	0–8 (.65)	.94 (1.41)	-.08	.94c**	.18**	1.18c**	-.04	.99c	15.57c	204
Father	0–6 (.49)	.58 (.93)	.06	.95c	.36**	1.30c**	-.13	.96c	29.84c	204
Biological sibling	0–8 (.74)	.96 (1.60)	-.05	.97c	.04	1.06c	-.12	.95c	5.86c	204
Family composite	0–14 (.70)	2.48 (3.18)	-.04	.95c*	.21**	1.16c**	-.11	.96c	19.63c	204

Note:

^*p<.05;

^**p<.01;

^a=linear regression,

^b=logistic regression,

^c=negative binominal regression

Suicide attempts, ideation, and self-injury.

Items of the Quickview Social History that pertained to past suicide attempts (dichotomous y/n), suicidal ideation in childhood or as an adult, and non-suicidal self-injury (NSSI) were included for analysis. Table 3 provides information about each of these variables and the suicide composite variables which was comprised of standardized (z-scores) of childhood suicidal ideation, adult suicidal ideation, number of suicide attempts, and NSSI.

Table 3.

Correlation and regression analyses of psychological dysfunction, crime, violence, and treatment

Scale/variable		RF-Meanness		RF-Disinhibition		RF-Boldness
	Range (α)/ Mean (SD)/%	r	βa ORb IRRc	r	βa ORb IRRc	r	βa ORb IRRc	R2(ab) χ2(c)	N
Externalizing dysfunction
Conduct Disorder	0–10 (.70)/1.78 (1.86)	.22**	.10a	.43**	.51a**	-.15*	.06a	.32a	191
Alcohol composite	-.64–3.21 (.86)/−.03 (.83)	-.06	-.19a**	.21**	.30a**	-.20**	-.18a**	.11a	161
Drugs composite	-.84–2.49 (.72)/−.06 (.82)	.03	-.15a*	.31**	.38a**	-.13	-.11a	.13a	131
Externalizing composite	-.89–3.37 (.64)/−.01 (.76)	.12	-.12a	.45**	.51a**	-.12	-.10a	.23a	113
Drinking sxs	0–6 (.67)/.78 (1.25)	-.01	.98c	.16*	1.11 c*	-.23**	.91c**	16.29c	204
Alcohol tolerance	40%	.01	.96b	.19*	1.21b*	-.15	.92b	.06b	161
Alcohol withdrawal	0–11(.91)/1.00 (2.33)	-.09	.91 c**	.18**	1.24 c**	-.23**	.88 c**	54.15c	204
Alcohol abuse	0–7 (.78)/1.25 (1.82)	-.09	.93 c**	.17*	1.17 c**	-.09	.97 c	17.86c	204
Num. of drugs	0–9 (.88)/ 2.64 (2.76)	-.08	.94 c**	.24**	1.18 c**	-.10	.98 c	20.04c	204
Drug tolerance	61.8%	.13	1.03b	.17	1.11b	-.18*	.90b	.06b	131
Drug withdrawal	0–10 (.89)/ .99 (2.16)	.01	.94 c*	.26**	1.33 c**	-.10	.91c	43.66c	204
Drug abuse	0–8 (.88)/ 2.11 (2.68)	-.01	.93 c**	.32**	1.26 c**	-.08	1.00c	34.14c	204
Crime/Violence
Num. of discp. reports	0–63 / 4.93 (9.82)	.21**	1.09 c**	.11	1.02 c	.01	1.00c	24.83c	165
Any violent discp. Reports	13%	.23**	1.14b*	.17*	1.05b	-.05	.95b	.06b	153
Num. of past charges	0–48 / 7.51 (7.90)	.16*	1.00c	.30**	1.12 c**	.03	1.02c	15.99c	174
Violent offending	56.2%	.05	1.04b	-.05	.93b	-.04	.97b	.01b	185
Anger(STAXI)
Expression In	6–30 / 17.95 (4.89)	.36**	.35a**	.25**	.07a	-.37**	-.39a**	.29a	212
Expression Out	7–30 / 15.39 (4.37)	.47**	.32a**	.45**	.30a**	.10	.09a	.30 a	212
Control In	5–32 / 23.78 (5.55)	-.34**	-.18a**	-.44**	-.36a**	.05	.05a	.22 a	212
Control Out	10–32 / 23.01 (5.19)	-.39**	-.24a**	-.42**	-.30a**	-.05	-.05a	.22 a	211
Internalizing dysfunction
Depression	0–13 (.89) / 2.59 (3.39)	.05	.1.00c	.13	1.07c	-.17*	.94c*	10.83c	204
Phobias (y/n)	71.1% (y)	.20**	1.08b*	.25**	1.10b	-.22**	.90b*	.08b	205
Obsessions	46.6%	.38**	1.20b**	.28**	1.10b	-.21**	.84b**	.21b	174
Compulsions	43.2%	.22**	1.06b	.23**	1.14b	.01	1.01b	.07b	176
PTSD symptoms	0–7 (.86) / 2.15 (2.35)	.06	1.00c	.17*	1.07c	-.18*	.95c*	8.63c	191
Panic (y/n)	52.1%	-.01	.97 b	.12	1.11b	-.21**	.89b*	.06b	163
Child disordered eating	0–5 (.51) / 0.49 (.89)	.01	.98c	.11	1.10c	.01	1.00c	3.42c	203
Adult disordered eating	0–4 (.65) / 0.54 (.95)	-.01	.98c	.06	1.05c	-.01	1.00c	1.04c	205
Other psychopathology
Psychotic symptoms	0–10 (.68) 1.03 (1.65)	.17*	1.05c*	.16*	1.04c	-.15*	.93c**	15.83c	205
Hypomania	0–9 (.76) / 1.94 (2.22)	.19**	1.02c	.24**	1.09c*	.09	1.04c	13.46c	205
Suicidal behavior
Past sui attempts (y/n)	42.7% (y)	-.12	.99b	.11	1.20b*	-26**	.86b**	.08b	192
NSSI	17.7%	.12	1.05b	.17*	1.13b	-.24**	.83b**	.08 b	191
Suicide composite	-1.5–7.68 (.76)/−.04 (2.22)	.09	.03a	.19**	.17a*	-.33**	-.33a**	.14 a	187
Mental health treatment
Inpatient treatment	15.7%	.11	1.07b	.08	1.03b	-.07	.94b	.02 b	185
Outpatient treatment	33.2%	.02	1.00b	.05	1.03b	-.15*	.91b*	.03 b	195
Composite of treatment	0–2 (.80) / 0.51 (.71)	.07	.06a	.07	.03a	-.15*	-.15a	.03 a	177
Drug treatment	0–2 (.58) / 0.43 (.69)	.07	.94c*	.29**	1.21c**	.02	1.04c	12.11c	204

Note:

^*p<.05;

^**p<.01;

^a=linear regression,

^b=logistic regression,

^c=negative binominal regression

Self-reported psychopathology.

Variables were created from items of the Quickview Social History that pertained to a history of DSM-IV internalizing psychopathology, including symptoms of fear, anxiety, and mood disorders. Table 3 includes information for the depression symptoms variable, hypomania/mania symptoms variable, psychotic symptoms variable, phobia (dichotomous y/n), posttraumatic stress symptoms variable, presence/absence of ever experiencing obsessive and compulsive symptoms, panic (dichotomous y/n), adult disordered eating symptoms variable, and the child disordered eating symptoms variable. The presence of disordered eating during childhood was significantly correlated with having disordered eating in adulthood (r=.49, p<.01).

Other count data variables were created from items of the Quickview Social History that pertained to a history of DSM externalizing psychopathology, including antisocial symptoms and substance use disorders. Table 3 includes information for the conduct disorder symptoms variable. Drug and alcohol use was assessed in numerous variables. Specifically, Table 3 includes information about the drinking symptoms (sxs) variable (related to problematic alcohol use), tolerance to alcohol (dichotomous y/n), alcohol withdrawal symptoms variable, alcohol abuse symptoms variable, drug symptoms variable (which assessed symptoms of problematic use of illicit substances across nine substances), drug tolerance (dichotomous y/n), drug withdrawal symptoms variable, and drug abuse symptoms variable. Composite variables of the drug and alcohol scales were created, respectively. The alcohol composite variable included the standardized (z-scores) of: the alcohol abuse symptoms variable, alcohol tolerance and withdrawal symptoms, and the problematic drinking symptoms variable. The drug variable included the standardized (z-scores) of: drug tolerance and withdrawal and drug abuse problems. The externalizing dysfunction composite variable included the standardized (z-scores) of the drug and alcohol composite variables and the conduct problems variable. Information for the composite variables is also presented in Table 3.

Family history of mental illness.

Variables were created from items of the Quickview Social History related to familial history of psychopathology (e.g., substance use, nervousness, depressed, attempted suicide). Information about the mother symptoms variable, father’s history of mental health problems, biological sibling symptoms variable, and the in family composite variable can be found in Table 2.

Health and mental health treatment.

Items from the Quickview Social History pertaining to a history of seeking mental health treatment were also evaluated. Information about inpatient, outpatient, the composite of psychiatric treatment, drug treatment, alcohol treatment, and composite of drug/alcohol treatment is located in Table 3.

State-Trait Anger Expression Inventory-2 (STAXI-2; Spielberger, 1999).

The STAXI-2 was administered as a measure of anger. The STAXI-2 is a 57-item self-report measure, which assesses anger expression in (AX-I; referring to the extent one holds in or suppresses angry feelings), anger expression out (AX-O; anger expressed verbally or physically towards people and property), anger control in (AC-I; ability to control angry feelings by calming one’s self down), and anger control out (AC-O; ability to control angry feelings and inhibit their outward expression). The STAXI-2 has demonstrated adequate reliability and validity (Spielberger, 1999). Information about the STAXI-2 scores in the current sample are located in Table 3.

Violence.

Offenses were dichotomized into violent or nonviolent offense classifications. These were identified using the Florida Department of Corrections (FDOC; 2008) Offense Category List. The FDOC uses the same descriptors in the prison records and therefore classification of violent and nonviolent offenses was obtained from prison records. Violent crimes included murder/manslaughter, sexual assault or battery, robbery, violent personal offenses, resisting arrest with assault, and burglary with assault. Nonviolent crimes included burglary, theft, forgery, fraud, and drug offenses. Information about violence is located in Table 3.

Data Analysis

Unless otherwise noted, Pearson correlations were computed between MMPI-2-RF triarchic scales and criterion measures to evaluate bivariate associations. Binary variables were examined using point-biserial correlations to include childhood sexual abuse, rape in adulthood, abuse by partner, alcohol and drug tolerance, violent offending, phobias, panic, past suicide attempts, NSSI, and inpatient and outpatient treatment. Linear (for continuous dependent variables), logistic (for dichotomous dependent variables), and negative binominal (for count variables in which the variance exceeded the mean) regression analyses were computed to examine the unique (reported as standardized beta coefficients [B], odds ratios [OR], or incidence risk ratios [IRR] ) and joint prediction (reported as model R2 or chi-square ) of MMPI-2-RF estimated Boldness, Meanness, and Disinhibition TriPM scores in relation to self-reported histories of physical and sexual abuse, internalizing and externalizing psychopathology, suicidal tendencies, and anger.

Next, to test for the interactive (i.e., non-additive or synergistic) effect of triarchic psychopathy traits in predicting clinical criteria, we computed hierarchical regression models separately for testing boldness X meanness and boldness X disinhibition effects. In the first step of each model either boldness and meanness or boldness and disinhibition were entered. To test for the incremental contribution of the interactive effect of triarchic traits, the second step of the model included an interaction term that was created by multiplying mean-centered MMPI-2-RF-Tri variables (i.e., MMPI-2-RF Boldness X MMPI-2-RF Disinhibition and MMPI-2-RF Boldness X MMPI-2-RF Meanness). We did not test for the interactive effect of MMPI-2-RF Meanness X MMPI-2-RF Disinhibition due to the high correlation between these variables (r = .47; cf. Patrick & Drislane, 2015). Two parallel sets of hierarchical regression analyses were conducted to examine the interaction of triarchic scales with all criterion variables and, where applicable, only included tests in predicting composite scales (i.e., substance use and suicidality in aggregate, rather than separately for each drug, alcohol, and suicidal behavior scale). Only significant interactions are reported in the text and figures.

Results

Interrelations amongst Psychopathy Scores

MMPI-2-RF Boldness, Meanness, and Disinhibition scores were compared to one another. Consistent with previous research (cf. Patrick & Drislane, 2015), MMPI-2-RF Meanness and MMPI-2-RF Disinhibition scores were significantly correlated with one another (r = .47, p < .001), while MMPI-2-RF Meanness and MMPI-2-RF Disinhibition were not significantly correlated with MMPI-2-RF Boldness (r = .05, p =.48 and r = −.05, p =.50, respectively).

History of Sexual and Physical Abuse

Table 2 presents correlations and results from negative binomial regression analyses examining associations between psychopathy scores and differing forms of abuse in childhood and adulthood. A history of abuse was largely unrelated to both MMPI-2-RF Meanness and MMPI-2-RF Boldness scores. MMPI-2-RF Disinhibition scores were significantly positively related to being abused by one’s parents or caregivers and sexual abuse in childhood.

Familial History of Mental Illness

Table 2 also presents correlations and standardized coefficients from negative binomial regression analyses between psychopathy scores and variables reflecting reported family history of mental illness. MMPI-2-RF Boldness scores were negatively related to a reported family history of mental illness. MMPI-2-RF Meanness scores were negatively related to a history of mental illness in the mother and the family composite variable. MMPI-2-RF Disinhibition scores were positively related to a history of mental illness in either the mother or father, as well as the family composite variable of mental health history.

Suicide attempts, ideation, and self-harm

Table 3 presents correlations and results from regression analyses examining associations between psychopathy scores and suicidal ideation, attempts, and self-injury. In regard to suicidal behavior, MMPI-2-RF Meanness scores were not significantly related to any of the suicide or self-injury scales. MMPI-2-RF Disinhibition scores were significantly positively related to number of previous suicide attempts. MMPI-2-RF Boldness scores were significantly negatively associated with previous suicide attempts, a history of engaging in NSSI, and the suicide composite variable.

Self-Reported Psychopathology

Table 3 presents results from correlation and regression analyses examining associations between psychopathy scores and self-reported psychopathology.

Externalizing dysfunction.

Scores on the MMPI-2-RF Meanness scale were positively associated with childhood conduct problems and negatively associated with the drug and alcohol composite variables, alcohol abuse, alcohol and drug withdrawal symptoms, number of drugs used, and symptoms of drug abuse. MMPI-2-RF Disinhibition scores were positively associated with conduct problems, both the drug and alcohol composite variables, the externalizing composite variable, symptoms commonly associated with alcohol use, alcohol tolerance and withdrawal symptoms, alcohol abuse, number of drugs used, drug withdrawal symptoms, and symptoms of drug abuse. Scores on the MMPI-2-RF Boldness scale were negatively associated with conduct problems, the alcohol composite variable, symptoms commonly associated with alcohol use, and alcohol withdrawal symptoms.

Internalizing/other dysfunction.

MMPI-2-RF Meanness scores were positively associated with the presence of phobias, obsessions, compulsions, symptoms of psychosis, and hypomania. MMPI-2-RF Disinhibition scores were positively correlated with phobias, obsessions, compulsions, PTSD symptoms, symptoms of psychosis, and hypomania. Conversely, scores on the MMPI-2-RF Boldness scale were negatively associated with depressive symptomology, phobias, obsessions, PTSD symptoms, panic, and symptoms of psychosis. Childhood and adult disordered eating symptoms were unrelated to the MMPI-2-RF-based triarchic scales.

Mental Health Treatment

Table 3 presents correlations and results from linear regression, logistic regression, and negative binomial regression analyses between psychopathy scores and the scales related to mental health treatment. Most of the psychopathy scores were unrelated to mental health treatment. MMPI-2-RF Meanness and MMPI-2-RF Disinhibition scores were positively associated with a history of substance use treatment while the MMPI-2-RF Boldness scale was negatively associated with both a history of obtaining outpatient psychiatric treatment and the composite variable of psychiatric treatment.

Anger

Table 3 presents correlations and results from linear regression analyses amongst psychopathy scores and aspects of anger measured by the STAXI-2. Both MMPI-2-RF Meanness and MMPI-2-RF Disinhibition scores were strongly positively associated with anger expression in and anger expression out and strongly negatively associated with anger control in and anger control out. MMPI-2-RF Boldness scores were only negatively associated with anger expression in.

Violence and Crime

Table 3 presents correlations and results from logistic regression and negative binomial regression analyses between psychopathy scores and the measures of violence, disciplinary reports within the institution, and previous offending. MMPI-2-RF Meanness scores were positively associated with the number of disciplinary reports received, violent disciplinary reports, and number of previous charges. MMPI-2-RF Disinhibition scores were positively associated with violent disciplinary reports and number of previous charges. The MMPI-2-RF Boldness scale was largely unrelated to these measures.

Examining interactive effects between triarchic scales and clinical criterion.

(Only significant interactions are presented in figures).

Conduct problems

For conduct problems, a significant MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect was observed (β=.19, ΔR2 = .017, p = .029). The figure (Graph A) depicts the direction of this effect, and simple slope tests indicate a significant effect of MMPI-2-RF Boldness at high levels of MMPI-2-RF Disinhibition (β = .21, p = .018), but not at low levels (β=−.04, p=.609). A corresponding MMPI-2-RF Boldness X MMPI-2-RF Meanness effect in the prediction of conduct symptoms was not significant. (ΔR2 < .001, p = .975).

Paternal history of mental illness

For father history of mental illness, a significant MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect was observed (β=−.013, ΔR2 = .018, p =.039). The figure (Graph B) depicts the direction of this effect and simple slope tests indicate a significant effect of MMPI-2-RF Boldness at high levels of MMPI-2-RF Disinhibition (β = −.26, p = .006), but not at low levels (β=−.01, p=.921). A corresponding MMPI-2-RF Boldness X MMPI-2-RF Meanness effect in the prediction of father history of mental illness was not significant. (ΔR2 =.008, p = .108).

Alcohol composite

For the alcohol composite scale, a significant MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect was observed (β=−.011, ΔR2 = .017, p =.05). The figure (Graph C) depicts the direction of this effect and simple slope tests indicate a significant effect of MMPI-2-RF Boldness at high levels of MMPI-2-RF Disinhibition (β = −.33, p = .001), but not at low levels (β=.09, p=.335). A corresponding MMPI-2-RF Boldness X MMPI-2-RF Meanness effect in the prediction of the alcohol composite scale was not significant. (ΔR2 ≤.001, p = .792).

Depressive symptoms

For depressive symptoms, a significant MMPI-2-RF Boldness X MMPI-2-RF Meanness effect was observed (β =.032, ΔR2 = .023, p = .027). The figure (Graph D) depicts the direction of this effect, and simple slopes tests indicate a significant effect of MMPI-2-RF Boldness at low levels of MMPI-2-RF Meanness (β = −.31, p = .001), but not at high levels (β=−.01, p=.948). A corresponding MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect in the prediction of depressive symptoms was not significant. (ΔR2 = .013, p = .102).

Drug composite

For the drug composite scale, a significant MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect was observed (β=−.015, ΔR2 = .033, p =.006). The figure (Graph E) depicts the direction of this effect and simple slope tests indicate a significant effect of MMPI-2-RF Boldness at high levels of MMPI-2-RF Disinhibition (β = −.31, p = .001), but not at low levels (β=.04, p=.681). A corresponding MMPI-2-RF Boldness X MMPI-2-RF Meanness effect in the prediction of the drug composite scale was not significant. (ΔR2 = .001, p = .308).

STAXI-2 anger control out

For the STAXI −2 AC-O, a significant MMPI-2-RF Boldness X MMPI-2-RF Meanness effect was observed (β =.060, ΔR2 = .033, p = .004). The figure (Graph F) depicts the direction of this effect, and simple slopes tests indicate a significant effect of MMPI-2-RF Boldness at low levels of MMPI-2-RF Meanness (β = −.20, p = .022), but not at high levels (β=.17, p=.068). A corresponding MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect in the prediction of STAXI −2 AC-O was not significant. (ΔR2 = .004, p = .342).

Non-suicidal self-injury

For NSSI, a significant MMPI-2-RF Boldness X MMPI-2-RF Disinhibition effect was observed (β=−.009, ΔR2 = .045, p = .002). The figure (Graph G) depicts the direction of this effect, and simple slope tests indicate a significant effect of MMPI-2-RF Boldness at high levels of MMPI-2-RF Disinhibition (β = −.45, p < .001), but not at low levels (β=−.06, p=.524). Additionally for NSSI, a significant MMPI-2-RF Boldness X MMPI-2-RF Meanness effect was observed (β =−.004, ΔR2 = .021, p =.041). The figure (Graph H) depicts the direction of this effect and simple slope tests indicate a significant effect of MMPI-2-RF Boldness at high levels of MMPI-2-RF Meanness (β = −.40, p ≤ .001), but not at low levels (β=−.12, p=.201).

Discussion

The current study sought to expand the scientific knowledge on psychopathic personality traits in female offenders by evaluating the relationship between MMPI-2-RF triarchic scales and external correlates in a sample of female offenders. As previously described, while the rate of female incarceration has increased rapidly, much of the literature on psychopathic personality traits using correctional samples has focused on males. Thus, we aimed to expand scientific knowledge on psychopathic personality traits in female offenders by evaluating the relationship between MMPI-2-RF triarchic scales, both individually and concurrently, and relevant external correlates in a sample of female offenders.

By and large, associations between MMPI-2-RF triarchic scales and external correlates were consistent with hypothesized expectations, as triarchic psychopathy constructs evinced differential relationships with criterion variables in theoretically consistent directions. This was particularly the case for boldness and disinhibition with disinhibition showing unique relations to maladaptive criteria and boldness demonstrating a generally more adaptive profile of associations. Disinhibition, the integration of poor impulse control and negative affect (Patrick et al., 2009), was positively related to childhood abuse history, parental psychopathology, indicators of suicidality, internalizing and externalizing psychopathology, violent disciplinary reports, number of previous criminal charges, and anger expression and inversely related to anger control. These associations are consistent with hypotheses and prior research (e.g., Krueger, Markon, Patrick, Benning, & Kramer, 2007; Sica et al., 2015; Venables et al., 2015; Venables, Hall, & Patrick, 2014). As boldness encompasses ostensibly adaptive traits such as stress resiliency, self-assurance, an ability to remain calm under pressure, and venturesomeness (Patrick et al., 2009), prior research finds that boldness is inversely related to indicators of internalizing psychopathology and other maladaptive traits (e.g., Brislin et al., 2015; Harrop et al., 2017; Patrick & Drislane, 2015). Results in our sample of female offenders, wherein boldness was negatively related or unrelated to all external correlates (e.g., internalizing psychopathology, suicide, history of mental health treatment, substance use), were thus consistent with hypotheses based on both the theoretical underpinnings of the triarchic model of psychopathy and prior research findings on boldness and related constructs (e.g., Brislin et al., 2015; Fanti, Kyranides, Drislane, Colins, & Andershed, 2016; Sellbom, 2015; Sellbom et al., 2017).

Results involving meanness were less consistent with expectations. Meanness references a constellation of attributes including low empathy, callousness, lack of close attachments, and antagonism (Patrick et al, 2009). Consistent with this conceptualization and our empirically-based hypotheses (e.g., Brislin et al., 2015; Kyranides, Fanti, Sikki, & Patrick, 2017; Patrick & Drislane, 2015; Venables et al., 2014), we found meanness to be positively related to anger expression and history of conduct disorder symptoms, inversely related to anger control, and positively related to the number of disciplinary reports (including violent reports) the inmate incurred and number of previous criminal charges. Unexpectedly, meanness was also positively related to a number of psychopathology variables, including having at least one phobia (although see Fanti et al., 2016), obsessions, compulsions, psychosis, and hypomania. Meanness was also related to history of substance use treatment and inversely related to a maternal family history of mental illness and the family composite variable. Findings on the relationship between psychopathology, substance use, and meanness traits were unexpected but may be indicative of true gender differences in the expression of meanness traits. For example, borderline personality disorder, a disorder which includes amongst its features extreme internalizing and externalizing dysfunction and labile emotionality (including callousness at times), has been suggested as a female variant of psychopathy (Sprague, Javdani, Sadeh, Newman, & Verona, 2012). The current pattern of results, in which meanness was related to anxiety, psychosis, and hypomania, may be reflective of the presence of such a variant in this sample of female offenders. Still, given the unexpected nature of these findings, they warrant replication in other samples.

A considerable strength of the current study is the use of a female offending sample, and the inclusion of external correlates thought to be of particular importance to this population (e.g., trauma history). Female offenders are underrepresented in psychopathology and forensic research, yet rates of incarcerated women are rapidly escalating (Glaze & Kaeble, 2014) and the incidence of psychopathology in female offenders exceeds that of males (Drapalski, Youman, Stuewig, & Tangney, 2009; James & Glaze, 2006). Prior research noting the high rates of trauma and interpersonal victimization in incarcerated women (deHart et al., 2014; Green et al., 2005; Lynch et al., 2012), and linking history of abuse to psychopathy (Verona et al., 2005; Weiler & Widom, 1996), suggest that history of abuse and trauma are of particular importance to consider in this sample. Indeed, in the current study, approximately half the sample reported a history of childhood sexual abuse (48.7%) and/or rape as an adult (56.1%). One-fifth (21.7%) of the sample reported a history of abuse by a partner. In addition to these startling rates, our findings further show a unique, differential relationship amongst triarchic constructs in regard to abuse and trauma, such that disinhibition is positively related to childhood trauma while boldness and meanness are not. This is consistent with prior work finding childhood physical and sexual abuse to be related to the lifestyle-antisocial behavior facets of the PCL-R in both female offenders and male sexual offenders (Graham, Kimonis, Wasserman, & Kline, 2012; Verona et al., 2005) and contributes to our understanding of potential developmental pathways for the traits and behaviors associated with disinhibition (see for example Blair, Peschardt, Budhani, Mitchell, & Pine, 2006).

The inclusion of boldness in the conceptualization of psychopathy has been the source of some controversy. The debate centers on results similar to those presented here, finding that boldness consistently appears to be adaptive to some extent, which is in contrast to the argument that the construct of psychopathy is purportedly maladaptive (e.g., Miller & Lynam, 2012). Yet others argue that psychopathy is a configuration of both adaptive and maladaptive traits (Lilienfeld, 2013; Lilienfeld, Patrick, Benning, Berg, Sellbom, & Edens, 2012), consistent with many historical conceptualizations such as Cleckley (1941/1976) and with prototypicality ratings by mental health professionals (Berg, Lilienfeld, & Sellbom, 2017). Univariate findings in the current study are consistent with the latter conceptualization and serve to further extend prior work by finding similar adaptive associations in an understudied population. Given this purported configural arrangement of both adaptive and maladaptive psychopathic traits, it is crucial for research to examine interactions amongst psychopathic traits in predicting relevant criteria. Indeed, when examining the impact of concurrent triarchic domains in predicting external correlates, we observed several significant interaction effects and specifically a buffering effect of boldness for many clinical problems. Boldness appears to operate as a resilience factor for some forms of psychopathology and potentially drug and alcohol problems (however, see Hicks, Iacono & McGue, 2014), but not directly acting out (e.g., conduct problems) where high levels of boldness could lead to increased delinquency and early behavioral problems. The greatest levels of substance use (both drug and alcohol composites) were reported by those who scored low on boldness but high on disinhibition. Likewise, paternal history of mental illness was highest for participants scoring highly on disinhibition and low on boldness. The highest levels of NSSI were endorsed by those scoring low on boldness but high on meanness. The ability to control angry feelings and inhibit their outward expression as measured by the STAXI −2 was greatest for participants scoring highly on the boldness but low on the meanness scale. Additionally, depressive symptoms were observed in individuals who scored high on boldness and low on meanness.

These findings speak to the role that boldness plays in buffering the expression of psychopathology associated with meanness and disinhibition at a zero-order level. In the case of substance use, for example, those with high boldness and high disinhibition had relatively low levels of substance use. Only disinhibited female offenders without concurrent boldness traits exhibited high levels of substance use. This is consistent with previous research noting a potential protective effect of boldness and similar traits on the development of psychopathology in at-risk individuals (e.g., Anestis, Harrop, Green, & Anestis, 2017; Anestis, Anestis, & Preston, 2018; Sellbom, 2015; Venables et al., 2015). Interestingly, in this sample, the buffering effect of boldness was not observed in the case of childhood antisocial behaviors. While an interactive effect was found between boldness and disinhibition in predicting conduct problems, the interaction was such that those scoring concurrently high on both disinhibition and boldness reported the greatest number of childhood antisocial behaviors. This finding has important implications for the aforementioned debates regarding the role of boldness (or fearless dominance) in psychopathy by providing evidence that it does not merely index of positive adjustment and is unrelated to deviancy (cf. Miller & Lynam, 2012).

Results from the present study and other recent investigations of MMPI-2-RF-Triarchic scales (Kutchen et al., 2017; Sellbom et al., 2016) indicate a consistent pattern of convergent and discriminant validity with personality and clinical criterion measures as observed in alternative operationalizations of triarchic constructs. As the MMPI-2-RF is widely used in clinical and research contexts, it is an ideal assessment instrument for further developing and refining self-reported biobehavioral measures such as those reflected by triarchic psychopathy constructs. Given shifts in the field towards incorporating neuroscience methods and measures in the assessment of mental health and illness, such as the National Institute of Mental Health’s Research Domain Criteria (RDoC) initiative (Kozak & Cuthbert, 2016), it will be valuable in future research to further validate and refine MMPI-2-RF-based triarchic psychopathy scales using neurobiological criteria. The MMPI-2-RF’s rich item-pool and history in psychological assessment make it particularly well-suited for identifying neurobehavioral traits and examining their validity in large-existing datasets.

Future research would benefit from examination of clinical problems assessed via interview or incorporating other objective information (e.g., review of institutional files) in contrast to the current study’s use of self-report scales. Additionally, studies employing larger sample sizes for examining the interactive/configural effects of triarchic dimensions in predicting important clinical criteria is warranted as our sample size may have limited the exploration of these effects in the current study. Additionally, using samples of female psychiatric patients would strengthen findings regarding the relationships between MMPI-2- RF-based triarchic psychopathy scales and psychopathology. Finally, utilizing the MMPI-2-RF-TriPM scales along with the administration of another measure of psychopathy would add to the literature on construct validity of using the MMPI-2-RF-TriPM scales.

Despite these limitations, we believe these findings represent a noteworthy contribution that both extends our understanding of psychopathic traits within an important and largely understudied population and contributes to the growing body of evidence supporting the validity of the MMPI-2-RF triarchic scales. We found that the MMPI-2-RF triarchic scales have theoretically expected associations with external correlates in a female offender sample, with a few notable exceptions that may speak to gender-differentiated expressions of psychopathic traits. Importantly, in this sample, triarchic psychopathy traits displayed a synergistic effect in predicting outcomes, the findings of which are relevant to ongoing debates regarding boldness. Overall, these data contribute to our growing understanding of the interactive nature of psychopathic traits and their unique manifestations in female offenders.

Figure 1.

Interactive effects between triarchic scales and clinical criterion