Revisiting prediction models in policing: Identifying high-risk offenders

Abstract

The use of prediction models for classifying offenders has been a common practice by the criminal justice system. Given the recent developments in criminal career research and continuing evidence that a small proportion of chronic offenders are responsible for the majority of crime, there is a continued need to identify high-risk offenders early on in their offending careers. The present study provides support for the accuracy of an innovative prediction instrument that was developed for identifying high-risk offenders in a rural county in a southern state. Offender risk classification was found to be associated with reoffending across different dimensions of assessment and the high-risk offenders had accumulated a greater mean number of arrests upon six-month follow-up when compared to the medium and low-risk offenders. Policy implications and directions for future research incorporating prediction models in policing are also discussed.