Does the Convergent Validity of the PAI Antisocial Features Scale Depend on Offender Response Style?

ABSTRACT

Researchers have recently questioned the utility of the response style indicators included on many self-report measures of personality and psychopathology. We examined whether the size of convergent validity coefficients for Personality Assessment Inventory (PAI) Antisocial Features (ANT) scores depends on PAI validity scale scores. Using PAI and Psychopathy Checklist–Revised (PCL–R) scores from 477 offenders evaluated for civil commitment as sexually violent predators, we found that PAI Positive Impression (PIM), Negative Impression (NIM), Malingering Index (MAL), Defensiveness Index (DEF), and Infrequency (INF) scores moderated the association between ANT and PCL–R scores. The association between ANT and PCL–R scores decreased as offenders overstated psychopathology (i.e., higher NIM or MAL scores) or exhibited increasing disengagement (i.e., higher INF scores). However, the association between ANT and PCL–R scores increased as offenders engaged in defensive reporting (i.e., higher PIM or DEF scores). The interaction effects were most common for ANT–E (Egocentricity), and to a lesser extent ANT–A (Antisocial Behaviors). PAI discriminant function validity indexes did not exhibit moderating effects on ANT and PCL–R scores. There was no evidence of validity scale suppression effects. These findings provide support for the potential role of some PAI response style measures for ANT scale interpretation in forensic settings.

Acknowledgment

The research contained in this document was coordinated in part by the Texas Department of Criminal Justice (587-AR09). The contents of this document reflect the views of the authors and do not necessarily reflect the views or policies of the Texas Department of Criminal Justice.

Notes

¹ In addition, the correlations between ANT and PCL–R scores in this study (see later) fell squarely within the range expected based on prior research, suggesting that PAI scores did not have an undue influence on PCL–R scoring. In other words, there was no evidence that the correlations between PCL–R and ANT scores were any larger than they are when PCL–R evaluators are blind to PAI scores.

² We also examined effects for facet scores, but did not include them here because they were substantively similar to the results from the factor-level analyses (results available from the first author).