A Comparison of Alternative Methods of Parole Prediction

Abstract

F OR a number of years there has been speculation concerning the relative merits of employing objective factors or responses reflecting attitudes as a basis for predicting parole outcome and selecting men for parole. In 1936, Ferris F. Laune, sociologist-actuary at the Stateville Branch of the Illinois State Penitentiary, published a study in which attitudes as revealed by responses to a questionnaire were used in combination with a number of objective factors to predict parole success or failure.' At the time the study was published the inmates tested had not been released on parole and, consequently, the predictive accuracy of this approach could not be evaluated in the light of actual parole outcome. Since that time, however, a large number of the men who answered the Laune questionnaire have been released on parole, and the parole outcomes are known. Scores on a parole experience table based solely on objective factors also are available for these men. The purpose of the present paper is to compare the relative accuracy of these two methods as a basis for predicting behavior on parole. There has been a tendency in the parole field to confuse prediction with selection. Parole prediction requires a categorical statement of future parole success or failure for each man. Such predictions may be based, for example, on psychiatric diagnosis, personal hunch, or score on an experience table. The objective is to correctly classify men as expected successes or failures on parole, and the appropriate measure of success in prediction lies in the degree to which the errors of classification are minimized.