Testing the predictive validity of the Personality Assessment Inventory (PAI) in relation to inmate misconduct and violence.

Abstract

The Personality Assessment Inventory (PAI) has been widely employed in correctional settings as a screening tool to assess inmates' risk for committing various types of institutional misconduct. Evaluations have generally found the PAI scales Antisocial Features (ANT), Aggression (AGG), and the Violence Potential Index (VPI) to be modestly related to institutional misbehavior, thus supporting its construct validity. The current study provides the most comprehensive examination of the predictive and incremental validity of the PAI and its subscales among a large sample of imprisoned offenders to date. In particular, the size of the sample (n = 15,546) and follow-up period (mean time at risk of 2.2 years) allowed for the disaggregation of institutional misconduct by levels of seriousness and separate examinations by conviction offense and criminal history variables. The 3 scales most strongly related to general rule infractions were ANT, AGG, and the VPI. After controlling for age at intake, violent conviction history, prior violent arrests, and time at risk, the PAI scales were shown to add incremental validity to the classification of 4 types of disciplinary infractions ranging from 2 to 4 percentage points. The study also explored the relationship of the PAI's response bias scales to institutional misconduct. (PsycINFO Database Record