I m proved Statistical Techniques for the Measurement of Recidivism

Abstract

In a recent paper in this journal, Kantrowitz discussed the use of cohort conditional-life tables in the analysis of correctional outcomes, and referred to some related work by Stollmack and Harris on failure-rate analysis. These two approaches are indeed closely related, but the latter does provide some important additional statistical power not otherwise available. These kinds of methodologies provide key insights on program performance and ultimately permit more appropriate statistical assessment of alternative correctional pro grams. The purpose of this paper then is to bring the criminal justice, and crime and delinquency research, communities up to date on developments in the application of failure-rate methods. The fundamental concept used in the approach is the modeling of program outcomes as failure patterns or rates over time. The method has been applied in a number of specific cases, includ ing to the data from Kantrowitz's California parole study. It is our view that the general adoption of this kind of methodology would lead to the development of standardized techniques of evaluation and more effective data-base management. The proper use of failure-rate methods per mits the handling of alternative definitions for program results and allows analysis of existing data in a more accurate and sensitive way. This should help to solve current problems of comparability and validity of program assessments, to identify significant parameters associated with recidivism, and to make more accurate population projections.