A Multiple Models Approach to Assessing Recidivism Risk: Implications for Judicial Decision Making

Lynette Chow-Martin, The Pennsylvania State University
Eric Silver, The Pennsylvania State University

Public protection is a primary focal concern of judges. Thus, sentencing decisions are based, in part, on assessments of the likelihood of future criminal behavior. Yet, judges seldom use actuarial (or statistical) prediction tools in their work. This reluctance is due largely, to concerns about predictive accuracy. In this paper, we describe a newly developed, multiple models approach to recidivism prediction. We show how combining the predictions from a series of classification three models, enhances our ability to classify cases into groups that vary along a spectrum of risk. Given the judicial concern with public protection, we believe that our results justify renewed attention to the potential uses of actuarial tools within the context of judicial decision-making.

(Return to Program Resources)

Updated 05/20/2006