PAI Antisocial Features scale as proxy for psychopathy

October 31, 2011
More recently, research has been conducted regarding the validity of using the PAI as a measure of psychopathy (Douglas et al., 2007).  As stated above, using the PCL-R is the current standard in the field. However, due to the limitations of the instrument, there is an increased emphasis on using the PAI in place of the PCL-R (Douglas et al., 2007).

Salekin et al. (1997) observed scores on the PCL-R, PAI -ANT scale, and Personality Disorder Examination (PDE) were negatively correlated with the Warmth scale of the PAI. This indicates that individuals who score high on these scales are not particularly warm, empathetic, or interpersonally caring individuals and provides support for the construct validity of psychopathy.  Also, the psychopathy measures demonstrated convergent validity with two other PAI scales: the Dominance scale (PAI-ANT: r = .28) and the Mania scale (PCL-R: r = .31). These authors noted that this is an expected finding as dominance has been found to be highly associated with PCL-SV scores.


Similarly, Salekin et al. (1997) explained that previous researchers have linked mania to psychopathy and feel that the Psychopathic Deviate and Hypomania scales of the MMPI should be included in the assessment of psychopathy. In this spirit, Hare (2004) included items such as impulsivity and boredom/need for stimulation as part of the criteria for psychopathy. Morey (2007) included impulsivity and sensation seeking as components of psychopathy on the ANT scale of the PAI. Finally, the pattern of results strongly supported the relationship between psychopathy and aggression  (Salekin et al., 1997). That is, the PAI Aggression scale was highly related to PAI-ANT (r = .74) and A-PDE (r = .72) and moderately related to the PCL-R (r = .49). 


In addition, Edens et al. (2001) reviewed the psychometric properties of the PAI in forensic and correctional settings. They focused on the two personality scales, ANT and BOR within the PAI. The PAI was compared to the MMPI-2 to which accurately classified 81.8% of the Borderline patients and 77.3% of the student sample. Additionally, the ANT scale evidenced significant convergence with other measures of psychopathy such as the PCL-R (r = 0.53) and the Personality Disorder Examination Antisocial scale (r = 0.78). Similarly, these authors reported a strong correlation between ANT and the Psychopathic Personality Inventory (r = 0.68, p<0.01).


In the same review, Edens et al. (2001) assessed risk assessment by using the AGG scale contained in the PAI. Researchers found a significant correlation between AGG and recidivism (r = 0.29) over a 14-month follow-up. Furthermore, these authors added that both ANT and AGG may provide useful information regarding potential for violence. They reported that ANT was a significant correlate of institutional misbehavior among incarcerated sex offenders, ranging from 0.39 to 0.42.

Edens et al. (2002) also examined the utility of the PCL-R and the PAI in predicting institutional misbehavior among 92 incarcerated sex offenders. The PCL-R and PAI scores were gleaned from archival data. Results indicated that the PAI ANT scale (r = .39) performed comparably to the PCL-R (r = .35) in predicting the occurrence of institutional misbehavior. However, the measures were not accounting for the same variance in the prediction, suggesting some nonoverlap in the measurements. These authors suggested that the combined use of the PCL-R and PAI might lead to greater classification accuracy.


Finally, Douglas et al. (2007) researched the PAI’s ability to predict psychopathic personality features, as assessed by the PCL-R. The PAI and PCL-R were administered to two samples. The first sample consisted of 281 participants of incarcerated men in Canada. The second sample was comprised of 102 sex offenders from Texas. These researchers found support to the construct validity of the PAI as a proxy for the PCL-R, specifically between the PAI ANT scales and Factor 2 (range, r = .17 for ANT-E to r = .52 for ANT-A).  The ANT-A scale produced the largest correlation with the PCL-R total scale (r = .36). Furthermore, the PAI DOM scale was found to be significantly correlated with Factor 1 (r = .16). Douglas et al. concluded that the PAI would more appropriate to screen out individuals who likely do not have psychopathy, rather than be used to identify those that do have psychopathy. Additionally, the PAI may be better suited in detecting changes in psychopathic features over time.


Although several studies highlight the usefulness of the PAI in assessing psychopathy, its limitations cannot be ignored. Hart et al. (2003) noted caution against using self report inventories for assessing psychopathy, especially in forensic settings. These authors acknowledged that self report inventories yield useful information about other aspects of personality, test-taking attitudes, and for research purposes. However, these inventories are often poor measures of psychopathy in particular. Hart et al. outlined several reasons for this: they fail to assess and control for the effects of deceitfulness; the psychopathy related scales are significantly correlated with measures of emotional state at the time of assessment; because the scales mostly assess delinquent and antisocial behavior, they are all correlated with Factor 2 of the PCL-R and not Factor 1; and there is no literature supporting the ability of self report inventories to predict criminal behavior.


Return to Blog Main Page