The Predictive Accuracy of Early Intervention System Identification Criteria

• In Event: Early Intervention Systems: The State of the Art

Abstract

Police early intervention (EI) systems implicitly rest on the prediction that an officer who meets the identification criteria would, in the absence of an intervention, continue to engage in problematic conduct. Predictions of this kind are notoriously difficult to make. Most EI systems rest on “time and numbers” identification criteria that identify officers who exceed a specified numerical threshold on any criterion during a specified time frame. Using the most commonly-used time frames and numerical thresholds for EI systems, we empirically estimate the accuracy of their predictions, and that of alternative identification criteria, using data from several police agencies.

Authors

• Christopher Harris, University of Massachusetts – Lowell

• Robert E Worden, University at Albany, SUNY

• Kelly J Becker, The John F. Finn Institute