NCJ Number
97992
Date Published
1985
Length
58 pages
Annotation
Levels of association between self-report assessments of attitudes/personality and criminal recidivism were examined as a function of variations in research design, the nature of underlying constructs, and systematic errors in the measurement of predictor and criterion variables.
Abstract
Two samples of young adult probationers serving sentences of at least 6 months studied over a 30-month period from 1974 to 1976 were chosen to participate in the study. Assessment procedures included interviews and paper and pencil questionnaire battery tests measuring criminal sentiments, alienation, conventional success orientation, personal distress, empathy, socialization, and criminal personality. The battery of questions included such factors as scholastic maladjustment and family dissension. Extreme group validities provided inaccurate estimates of the relative predictive criterion validity of the seven sets of predictor variables. The predictive validity of procriminal sentiments and psychopathy were relative to assessments of the more distal constructs of sentimental ties to convention, personal distress, and empathy within two samples of young adult probationers. Substantial and cumulative increases in validity were associated with improved sampling of the predictor and criterion domains. Sampling of predictors was improved by introducing retests and the use of multimethod-multitrait predictors. Sampling of the criterion was improved through extended followup and multimethod assessment. The findings are inconsistent with a number of the 'post hoc' arguments made against dispositional constructs in criminology and provide a systematic example of the importance of general social psychological principles in the prediction of criminal behavior. Tabular data and 62 references are provided. (Author abstract modified.)