NCJ Number
92902
Date Published
1983
Length
123 pages
Annotation
Using conventional statistical methodology, this paper develops a formula for identifying offenders having the highest incidence of criminal behavior as reflected in arrests for serious crimes, and this formula is extended to account for the type and amount of crime associated with the arrests of career offenders; after predicting the amount and kinds of crime that might be prevented by incarcerating habitual offenders, these predictions are used to simulate the effect that a career criminal program might have on crime and the Federal criminal justice system under alternative assumptions about how the program might operate.
Abstract
The first section of the presentation provides a definition of recidivism and explains the approach used in its measurement, followed by a discussion of the need for recognizing that future crime varies according to seriousness as well as the consistency with which offenders commit the same type of crime. The section that considers the prediction of recidivism discusses the statistical tools used, the accuracy of the estimates, and findings pertaining to the estimates of the probability of rearrest and predictors of the length of time free prior to rearrest. In a simulation exercise on extending and applying the results of the statistical analyses, attention is given to decision rules for selecting career criminals, using arrest records vs. conviction histories, and the impact of the prototypical decision rules. Other simulation exercises are applied to estimating the number of career criminals and sentencing career criminals. The appendixes contain a technical procedure for estimating recidivism and a bibliography of 60 listings.