NCJ Number
80449
Journal
British Journal of Criminology Volume: 21 Issue: 4 Dated: (October 1981) Pages: 366-370
Date Published
1981
Length
5 pages
Annotation
Methodology and findings are presented from a British study designed to produce predictor variables that can be used to identify juveniles who need secure placement.
Abstract
Data were provided by 2 groups of 262 delinquent boys each matched by age and time in care prior to admission to the unit. The sample drawn from open community homes with education was termed 'general' delinquents, and the group drawn from special (secure) units was termed 'special.' Data were collected on 76 items thought to be operative prior to the boy's admission. Cluster analysis reduced this information to 39 items which provided 11 dependent variables with reliabilities around the .75 level. The reduced number of variables were entered into a stepwise multiple regression analysis, using group membership as a dichotomous criterion measure, enabling accurate assessment of the individual contribution made by each predictor variable. Variables related to care experience were removed, so that prediction could be based on data relating to items identifiable prior to long and variable residential placements. The four predictor variables that together explained about 34 percent of the variance between groups were personal maladjustment, delinquent behavior, domestic abuse, and destructive history. Those in the 'special' group obtained higher mean scores on each of these variables. The offenses of such a person are likely to be numerous and varied, including violent offenses and drug abuse. While there are dangers inherent in total reliance on a predictive instrument, the results outlined are useful in the necessary decisions about juvenile placements in the care system. Tabular data and six references are provided.