Differential predictive validity of the Mortimer-Filkins Questionnaire was examined with respect to ethnicity in a sample of 4,633 convicted drink driving offenders in Mississippi. Linear-logistic regression analysis indicated the instrument is a stronger predictor of drink driving recidivism among Caucasians than African-Americans. Practical implications of a finding of differential validity for risk screening are discussed, and the need to evaluate other instruments with respect to differential validity is addressed.
Abstract