NCJ Number
196226
Journal
Criminal Justice and Behavior Volume: 29 Issue: 4 Dated: August 2002 Pages: 397-426
Date Published
August 2002
Length
30 pages
Annotation
This article challenges the claim that the Psychopathy Checklist-Revised (PCL-R) is the "unparalleled" measure of offender risk prediction.
Abstract
Five years ago, the PCL-R was ordained, with a few minor caveats, as the "unparalleled...measure for making risk assessments with white male inmates" (Salekin, Roger, & Sewell, 1996). The Salekin et al. meta-analysis included both predictive and postdictive study designs, generating 29 effect sizes of the PCL-R's predictive validity with general and violent recidivism and institutional violence. A broader literature review of the PCL-R shows that the results of the meta-analyses have been consistent regarding the PCL-R; i.e., it is a reasonably good predictor of recidivism, with a mean effect size in the .20 to .30 range. This article updates the number of PCL-R and LSI-R (Level of Service Inventory-Revised) predictive validities and refines the Gendreau et al. (1999) meta-analyses by examining the data set for outliers and the effect of base rates and by applying fail-safe statistics to determine the number of effect sizes necessary to overturn the findings. The article concludes that the level of Service Inventory-Revised notably surpasses the PCL-R in predicting general and violent recidivism, albeit only modestly in the case of the latter. In addition, other problematic issues regarding the PCL-R are outlined. The authors suggest that a more useful role for psychopathy in offender risk assessment may be in terms of the responsivity dimension in case management. Finally, the authors suggest further research directions that will aid in knowledge cumulation regarding the general utility of offender risk measures. 3 tables, appended details of studies included in the meta-analysis, 16 notes, and 57 references