Abstract
A nonlinear, exemplar-based empirical modeling methodology was applied to the problem of classifying relative levels of recidivism risk in a population of released offenders from the Wisconsin Department of Corrections. Issues related to extracting relatively pure classes of exemplars from relatively ambiguous data are detailed. Risk was defined as the associative match to one of two exemplar groups; higher or lower risk offenders. The area under the Receiver Operating Characteristic (ROC) curve for 620 offenders examined in the initial subgroup was .94. Comparable results were found with a smaller validation sample of 408 offenders known to be higher risk. Implications of controlling for risk factor patterns are discussed.
Get full access to this article
View all access options for this article.
