Multiple Discriminant Prediction of Delinquency and School Dropouts

Abstract

PREDICTION studies of socially disvalued behaviors among male adolescents often neglect either crossvalidation procedures or the influence of incarceration upon the S’s test performance (Glueck and Glueck, 1950; Gough and Peterson, 1952; Sarbin and Jones, 1955). A further limitation of these studies is that they often deal only with extreme deviants (incarcerated repeated offenders). To be sure, these studies have provided valuable evidence concerning variables associated with deviancy. The present study departs from the typical ex post facto study in the following ways. (1) The predictor data were collected prior to the acting out of the criterion behavior.2 (2) A simultaneous double crossvalidation was carried out. (3) Two levels of deviancy were investigated. (4) The data from three criterion groups were sub-jected to a Multiple Discriminant Analysis, which provides more information than two independent regression analyses. The three-group criterion was &dquo;normals,&dquo; dropouts, and de-linquents (see Method for definitions). Predictor variables were se-lected on the basis of a major theme evident in a host of studies, that delinquents are more impulsive than are nondelinquents (Mc-