This NIJ Update describes the development, features, and uses of a risk-assessment tool that shows promise in increasing the efficiency and effectiveness of probation supervision by matching the resource allocation for each probationer to his/her level of risk for reoffending.
Development of the risk-assessment tool began when leaders from the Philadelphia Adult Probation and Parole Department (APPD) contacted the University of Pennsylvania's Jerry Lee Center for Criminology requesting its help in developing an evidence-based means of tailoring its officers' caseloads to the risk levels of their probationers. This cooperative effort was funded by the National Institute of Justice (NIJ). Using a statistical approach called "random forest modeling," the tool developed considers the nonlinear effects of a large number of variables in interaction with one another in determining the risk level (low, moderate, or high) for reoffending over a period of 2 years. In 10 to 15 seconds, the tool performs a risk assessment of each new probationer. This report describes how the flexibility of this tool can be tailored to the resources and databases of each jurisdiction. The determination of an acceptable error rate is also discussed, given that no prediction tool is perfect. In describing the outcome of this project, one of its developers commented, "Using random forest modeling gave us the best science available to identify the most dangerous offenders. It has ensured that we are preserving resources and that the people who are subject to the policy decisions based on those risk assessments are being treated in a fair and consistent way."