NCJ Number
109924
Date Published
1987
Length
52 pages
Annotation
This study examined the ability of a 7-item scale to predict recidivism among chronic offenders and investigated the information loss resulting from the use of recorded arrest rates as opposed to self-report information.
Abstract
Followup arrest histories were analyzed for 2 samples: 2,700 men committed to the California Youth Authority between 1966 and 1971, and 200 California prison inmates serving time for burglary or robbery who had been released from that commitment at least 2 years before the study. For both samples, the scale was only about half as accurate in predicting followup arrest rates as it was in predicting retrospective self-reported offense rates. The highest relative improvement over chance achieved with the scale was 24 percent; alternative versions of the scale proved to be no more accurate. Even with this poor predictive accuracy, the average arrest rates experienced by predicted high and low arrest rate offenders in the first sample differed by no more than a factor of two. The poor predictive accuracy and modest differences in average arrest rates of the groups categorized by the scale do not appear to justify the large differences in sentence lengths for offenders in different risk categories that would be necessary to achieve selective incapacitation effects. Moreover, little or no apparent correlation exists between individual self-reported offense rates and individual arrest rates: there appears to be an inverse relationship between self-reported offense rate and probability of arrest. 33 tables and 21 references.