Recidivism.

The purpose of this book is to present a thorough analysis of the concept of recidivism. The first half of the book describes the role that different definitions of recidivism have played in evaluating correctional goals and programs and shows how improper policy conclusions have been based on studies that used inappropriate definitions of recidivism. It describes the many goals that the correctional system is called upon to achieve and specifies how recidivism is used in measuring their achievement. A taxonomy of recidivism definitions is proposed to make possible the comparison of recidivism statistics of two different studies. The second half of the book addresses the problem of analyzing data on recidivism. Deficiencies in the standard method of analysis are noted, and different methods and models that overcome these deficiencies are described. Selection of the “best” method and model is addressed by noting the inappropriateness of the standard selection criterion, the chi-square goodness-of-fit test. New selection criteria are developed, including a graphical test based on the model’s predictive ability. In addition, characteristics of the recidivism process are invoked to aid in the selection of an appropriate model. This model, based on the incomplete exponential distribution, is described and a number of examples of its use are presented. Appendixes containing tables, graphs, and computer programs (useful in estimating confidence regions) are provided, with examples of their use in analyzing correctional data.