Psychopathy and female offenders

Psychopathy - characterised by "a persistent disregard for social norms and conventions; impulsivity, unreliability, and irresponsibility; lack of empathy, remorse and emotional depth; and failure to maintain enduring attachments to people, principles, or goals" (Hare, 1991, p. 45 as cited in Hemphill, Hare, & Wong, 1998) has been well-researched and accepted amongst researchers and clinicians as a risk factor predicting general, violent and sexual recidivism in male offenders (Hemphill, Hare, & Wong, 1998; Salekin, Rogers, Ustad, & Sewell, 1998). The Psychopathy Checklist - Revised (PCL-R), a measure of criminal psychopathy, combining equally personality or affective-interpersonal traits (Factor 1) and behavioural or antisocial lifestyle criteria (Factor 2), based on the 16 core traits of psychopathy identified by Cleckley, has been researched with almost exclusively male Caucasian offenders (Vitale, Smith, Brinkley, & Newman, 2002). However, Cleckley (1981, original work published in 1941) described two interesting case studies of female psychopaths, Roberta and Anna, in The Mask of Insanity.

Hemphill, Hare and Wong's review (1998) of the literature on the PCL-R concluded that "the PCL-R should be considered a primary instrument for guiding clinical appraisals of criminal recidivism and dangerousness" (p. 160) although it was "not designed to predict criminal behavior or to assess risk for violence" (Hare, 2003, p. 145). It has been shown that psychopaths are likely to recidivate within the first year upon their release (Hart, Kropp, & Hare, 1988 and Serin, Peters, & Barbaree, 1991 as cited in Salekin, Rogers, Ustad, & Sewell, 1998), to recidivate violently and to commit instrumental violent crimes (Hare, Clark, Grann, & Thornton, 2000 as cited in Richards, Casey, & Lucente, 2003). About 15% of female offenders would meet the PCL-R-criteria for psychopathy (Hemphill, Strachan, & Hare, 1999 as cited in Raine & Sanmart??n, 2001) and their rate of re-offending is higher than for other female offenders. Evidence is present for an overall lower prevalence rate of psychopathy in female offenders than in male offenders (Jackson, Rogers, Neumann, & Lambert, 2002; Salekin, Rogers, Ustad, & Sewell, 1998). This fact often leads to a debate on lowering the cut-off scores for female offenders. Some have found evidence that a lower PCL-R cut-off score (≥ 25 rather than ≥ 30) discriminates well between psychopaths and non-psychopaths (Jackson, Rogers, Neumann, & Lambert, 2002).

The PCL-R - in contrast to many actuarial risk scales - includes the contribution of personality traits to the persistence and stability of criminal behaviour and recidivism (Hemphill, Hare, & Wong, 1998). The latter authors suggest examining the influence of moderator variables, such as gender and race, in order to strengthen the link between the PCL-R and recidivism. Others have noted the cross-cultural variability in mean PCL-R scores and in the prevalence of psychopathy (Cooke, 1996 as cited in Hemphill, Hare, & Wong, 1998).

Salekin, Rogers, Ustad and Sewell (1998) comment that "if psychopathy measures are able to show adequate predictive validity in women, then their use in forensic settings would be warranted and could contribute substantially to dangerousness/risk assessments. On the other hand, if their predictive validity is only modest, then this information is also important given that inaccurate classifications/predictions could lead to serious ramifications for the examinee (e.g., not being given the opportunity for parole)" (p. 113). The authors used different psychopathy instruments (PCL-R, the Anti-Social Scale of the Personality Assessment Inventory and of the Personality Disorder Examination) with female offenders and concluded that psychopathy only moderately predicts recidivism in female offenders. In addition, the classification accuracy for psychopathy as a predictor of recidivism was modest to poor (i.e., it resulted in several false positives and false negatives). Further analyses revealed that personality criteria (Factor 1), with verbal aggressivity, "most appropriately predict recidivism in females" (Salekin, Rogers, Ustad, & Sewell, 1998, p. 124). However, the behavioural criteria of the different instruments did not predict recidivism. In short, a classification of psychopathy characterised by behavioural symptoms added little to the prediction of recidivism in this (small) sample of female offenders. The authors comment that caution is needed when using psychopathy measures with females. This is underscored by Jackson, Rogers, Neumann and Lambert (2002). The limitations of Salekin's study have been highlighted by Cale and Lilienfeld (2002), who conducted additional analyses which "do not support the claim that the factor structure of psychopathy differs markedly in males and females" (p. 1189). Richards, Casey and Lucente (2003) provide strong evidence for the hypothesis that particularly Factor 1 is related to increased risk for recidivism in incarcerated female substance abusers after treatment and release in the community. Loucks and Zamble (2000) found that in contrast to common findings for male offenders, "Factor-1 scores are as closely related to criminal behavior [in serious female offenders] as Factor-2 scores" (p. 34) whilst "Factor 2 (lifestyle) was more closely associated with violence than was Factor 1 (personality)" (p. 23).

Other research has found the PCL-R predicts recidivism in samples of female offenders (Loucks, 1995 and Zaparniuk & Paris, 1995 as cited in Hemphill, Hare, & Wong, 1998). However, methodological issues have been raised (Vitale, Smith, Brinkley, & Newman, 2002) such as the lower prevalence of psychopathy in female samples (see also Jackson, Rogers, Neumann, & Lambert, 2002), the different factor structure of psychopathy in male and female offenders (see also Salekin, Rogers, Ustad, & Sewell, 1998) and the generalisation of the PCL-R to a variety of samples, across race for instance. Another distinction between female and male psychopaths is co-morbidity, such as higher rates of suicidal behaviour, and disorders such as depression and anxiety in females (Mulder, 1994 as cited in Salekin, Rogers, Ustad, & Sewell, 1998).

Vitale, Smith, Brinkley and Newman (2002) examined 528 adult, nonpsychotic incarcerated women, one half Caucasian and one half African American. Their study confirmed the reliability and validity of the PCL-R in female offenders. They noted a relation between PCL-R scores and anxiety, negative affectivity and low intelligence, suggesting that "these factors may contribute strongly to the PCL-R scores of female offenders and may … lead to the misclassification of "neurotic" or "secondary" psychopathic women" (p. 223). The anxiety reflects the woman's negative reactions to the consequences of her antisocial behaviour rather than a personality characteristic. The researchers comment that the paucity of high PCL-R scores may reflect the lower base rate of psychopathy in women than in men or that the PCL-R items not adequately capture the construct as it is expressed in female populations. Importantly, except for some minor differences between the two racial groups, "there was relatively little evidence for the presence of race differences" (Vitale, Smith, Brinkley, & Newman, 2002, p. 225). The authors also note that the results do not explain the aetiology of psychopathy in women. Finally, they acknowledge that future studies need to address further the PCL-R structure across gender (see also Grann, 2000; Salekin, Rogers, Ustad, & Sewell, 1998). Richards, Casey, & Lucente (2003, as cited in Nicholls, Ogloff, & Douglas, 2004) "found that psychopathy scores were strongly related to recidivism following community release of female inmates" (p. 131). They found that psychopathy scores of female inmates were significantly associated with institutional rule violations, treatment non-compliance, avoidance of urine analysis for illegal drug use and general and violent prison misconduct. This is supported by Loucks and Zamble's (2000) findings that psychopathy predicts general offending, violent behaviour and prison misconduct in serious female offenders.

Concluding, some evidence is present that the PCL-R and PCL:SV show a relationship with crime and violence in women (Nicholls, Ogloff, & Douglas, 2004). Although it is best to use total PCL-R scores (rather than factor scores) with female offenders, clinicians should be "circumspect in utilizing psychopathy for risk assessment in female offender and clinical populations" (Jackson, Rogers, Neumann, & Lambert, 2002, p. 702).

Other research has indicated that the underlying dimensions of psychopathy for female offenders differ substantially from the two-factor model originally proposed by Hare in 1991 (Hare, 2003). Jackson, Rogers, Neumann, & Lambert (2002) attempted to test three psychopathy models (Hare's (1991), Salekin, Rogers and Sewell's (1997) and Cooke and Michie's (2001) three-factor model) using a cut-off score of ≥25 with 119 female inmates. Results question the application of the two-factor models. Cooke and Michie's three-factor model - arrogant and deceitful interpersonal style (ADI), deficient affective experience (DAE) and impulsive and irresponsible behaviour style (IIB) - appears to capture more efficiently the underlying dimensions of psychopathy in female offenders. "The most salient dimension appears to be the lack of emotional range and empathy" (Jackson, Rogers, Neumann, & Lambert, 2002, p. 701), with callousness, lack of remorse and shallow affect as the prominent items. Grann (2000) reports on a similar (European) study, examining specific PCL-R items with respect to gender. Although the majority of the PCL-R items did not show gender differences, some statistically significant gender differences emerged. Callous/lack of empathy and juvenile delinquency were considered "male items" and promiscuous sexual behaviour a "female item", discriminating best between male and female offenders. "These differences may reflect prevailing gender stereotypes" (Grann, 2000, p. 148) which may operate on different levels (in the questions asked and not asked in clinical interview, in what is reported in the files and what not and in the instrument itself). Testing a four-factor model of psychopathy which "affords to examine how the interpersonal, affective, and behavioral factors are associated with the antisocial aspects of psychopathy" (Vitacco, Neumann, & Jackson, 2005, p. 473) is the latest research update. The authors found support for both Cooke and Michie's three-factor model and the four-factor model "in relation to ethnicity, gender, and estimated IQ, as well as violence and other aggression" (p. 473). Kennealy, Hicks and Patrick (2005) report further support for the validity and reliability of Hare's two-factor and four-facet models of the PCL-R in female populations. Rogers, Salekin, Hill, Sewell, Murdock and Neumann (2000) examined the relationship between subcriteria and the PCL:Screening Version criteria. In terms of the criterion-related validity of PCL:SV subcriteria the researchers found that "fraud artist or con man", "no capacity for guilt - no conscience" and "little emotion in regard to actions" (subcriteria from Factor 1) "emerged as potential risk factors" with female offenders (p. 11) and "often physically abusive", "outbursts are shortlived", "no realistic long-term plans or commitments", "lived day-to-day" and "not thinking of the future" were considered potential protective factors. For a detailed review of sex differences in psychopathy and antisocial personality disorder, see Cale and Lilienfeld (2002). These authors document extensively that psychopathy and antisocial personality disorder are more prevalent in males than in females. They also comment that "there is no compelling evidence to support the claim that psychopathy or ASPD criteria should be tailored specifically to assessing either male or female adults" (Cale & Lilienfeld, 2002, p. 1198). Hare (2003) stated that standard PCL-R scores "have much the same meaning, with respect to the construct of psychopathy, in several different groups and settings" (p. 75) including African-American offenders, white female offenders and male forensic psychiatric offenders.

Research with the Historical Clinical Risk Scale (HCR-20), which includes psychopathy as an item, and female offenders is rare. Strand and Belfrage (2001) concluded few significant sex differences between violent mentally disordered men and women existed on the items, subscale and total scores of the HCR-20 and PCL:SV, except for some differences on specific items such as lower scores in "previous violence" for females. They committed less serious violence than the males but the frequency did not differ. Other interesting findings included female inpatients committing significantly more in-patient violence directed towards staff or self than the male inpatients, a majority of women were diagnosed with borderline personality disorder and a strong correlation between self-destructive behaviour and in-patient violence for females. They conclude that the HCR-20 is a useful tool when assessing in-patient violence for both mentally disordered males and females but one has to take into account that the nature of the violence differs between the sexes: mentally disordered men compared to in-patient women display more severe violence in the community or towards other patients.

Nicholls, Ogloff and Douglas (2004) found contrasting results: women had significantly lower scores on the subscales and total score on the HCR-20 than men at admission, but had similar scores upon discharge. The authors remind the reader that "while the reliability and validity of many existing measures have demonstrated utility in assessing violence risk with men, their utility with women is uncertain and requires further examination before the measures should be recommended for clinical use with female populations" (p. 132) and in particular with serious mentally ill women multiple tools for risk assessment need to be used. The researchers examined the predictive validity of the PCL:SV, HCR-20 and VSC (Violence Screening Checklist) in male and female psychiatric patients. They conclude that "psychopathy has a moderate to strong relationship with institutional and community violence among serious mentally ill women, fairly consistent with what we see in men" (p. 150) but not with physical violence and "that the HCR-20 and PCL:SV have moderate to strong predictive accuracy with [serious mentally ill] women" (p. 152). However, they recommend cautious use with these tools if applied on women, under the conditions the clinician is clear about its limitations, is up to date with literature on female aggression and does not rely on existing cut-off scores.

Concluding, in spite of some empirical support for the reliability and validity of the PCL-R in female offenders, using psychopathy for risk assessment in female offenders should be circumspect because of methodological issues (lower prevalence of psychopathy in women, different factor structure, generalisation of the PCL-R to a variety of samples and co-morbidity). The HCR-20 has been suggested as a useful tool when assessing inpatient violence for both mentally disordered males and females. In general researchers recommend cautious use with violence risk assessment tools if applied on women.