Two Models for Predicting Recidivism

This study investigates the efficiency of two means of predicting the behaviour of prisoners when released from incarceration prior to the full completion of sentence: the clinical prediction of the parole boards, as opposed to statistical prediction by means of the discriminant analysis programme. The two main findings indicate that (1) the boards are in a state of psychological dissonance with respect to the two major aims of early release from prison—prevention of recidivism and probability of rehabilitation; and (2) when the dependent variable is recidivism after release, the statistical prediction is more efficient than that of clinical prediction. In view of these findings, the issue of the application of the statistical formula, considering its advantages and disadvantages, for the purposes of the required prediction is discussed.