USING COMPUTERS TO MAKE JUDGMENTS - CORRELATIONS AMONG PREDICTORS ANDTHE COMPARISON OF LINEAR AND CONFIGURAL RULES

Statistical prediction is an important method for predicting and descr ibing human behavior. Though linear rules are generally recommended fo r prediction tasks, configural rules can do well. Their success seems to be related, in part, to whether correlations among predictors are n egative. One may wonder how frequently predictors are negatively corre lated in real-life settings and whether the addition of interaction te rms leads to a meaningful improvement in prediction in such situations . This article addresses the above questions in the context of the dia gnosis of schizophrenia. Symptom ratings sewed as predictors, diagnose s served as criterion scores. Negative correlations were found among t he predictors. A configural model made more accurate estimates of the likelihood of schizophrenia than did unit weight and differential weig ht linear models.