The idea of selective release

We interviewed 203 juvenile inmates housed in a juvenile corrections facility in New Orleans, using a revised version of the Rand instrument developed by Chaiken and Chaiken. The purpose was to demonstrate how prediction scales, used in the past to identify high-rate offenders for selective incapacitation, actually may be more suitable for identifying low-rate offenders for selective release. Our results show that the prediction scale performed worse at identifying high-rate offenders but better at identifying low-rate offenders. We discuss the policy implications of our findings.