Don't Pass Go

Abstract

Data for two samples of competent defendants were collected from hospital records in order to determine group differences for those who maintain competency to stand trial (nonrecidivists) versus those who do not maintain competency (recidivists). Stepwise discriminant analysis derived a model with two significant variables for predicting recidivism: medication change and severity of charges. Decreases in medication after being determined competent and a maximum possible penalty of greater than 40 years were significant in predicting recidivism. The discriminant model resulted in 63% correct classifications for the derivation sample and 68% correct classifications for the crossvalidation sample. Effectiveness of the predictive model and implications for assessment and treatment are discussed.