The Prediction of Violence in Adult Offenders

Abstract

Using 88 studies from 1980 to 2006, a meta-analysis compares risk instruments and other psychological measures on their ability to predict general (primarily nonsexual) violence in adults. Little variation was found amongst the mean effect sizes of common actuarial or structured risk instruments (i.e., Historical, Clinical, and Risk Management Violence Risk Assessment Scheme; Level of Supervision Inventory—Revised; Violence Risk Assessment Guide; Statistical Information on Recidivism scale; and Psychopathy Checklist—Revised). Third-generation instruments, dynamic risk factors, and file review plus interview methods had the advantage in predicting violent recidivism. Second-generation instruments, static risk factors, and use of file review were the strongest predictors of institutional violence. Measures derived from criminological-related theories or research produced larger effect sizes than did those of less content relevance. Additional research on existing risk instruments is required to provide more precise point estimates, especially regarding the outcome of institutional violence.