RACE, GENDER, AND GUIDELINES-BASED DECISION-MAKING

Risk dimensions used in guidelines systems have been implicated as con tributing to racial (and gender) disproportionalities in America's pri son and jail populations. Developers of some systems dealt with invidi ous risk predictors by purposely ignoring them, resulting in misspecif ication while not eliminating their effects. Eliminating all variables correlated with suspect factors would greatly attenuate power renderi ng practical decision-making took useless. This article shows that one risk-prediction device forming the basis of an operational guidelines system is, in fact, correlated with race and gender but that the appr oach suggested by some critics does not overcome this: Control factors will remain correlated with the final model even after second-order p olicy controls are implemented. Further although the suggested approac h is agnostic with respect to the native of policy controls, these wil l have considerable practical importance. Illustration is provided. Un biased models can be estimated with little appreciable loss of predict ive utility, and this is demonstrated.