NCJ Number
89250
Date Published
1981
Length
255 pages
Annotation
This report and its appendixes describe progress to date in four areas of research involving mathematical models of criminal recidivism.
Abstract
One effort has focused on the estimation of the model's parameters and associated statements, using maximum likelihood and Bayesian procedures for the split population model and the mixed exponential and Weibull models. Another effort has concentrated on ways to choose appropriate models of recidivism from among candidate models. The research also covered the development of covariate models of recidivism. In these models the probabilities of recidivism of each member of the group is determined by parameters, based on the individual's unique characteristics. Additional research consisted of the critical analysis of pretest-posttest designs for evaluating juvenile delinquency programs. Further areas needing research are the properties of the joint density function, results under specific assumptions, and the effect of distributions of certain variables. Nine appendixes present detailed descriptions of each phase of the research. Data tables, figures, and bibliographies for individual papers are also supplied.