ABSTRACT
Tests of general mental ability (GMA) and cognitively loaded skill and knowledge measures, such as work samples, are the selection procedures most predictive of performance in complex work. GMA tests are at least as predictive of work performance as other more job-specific tests in the cognitive domain and other assessments outside the cognitive domain, such as personality. Certainly many jobs require some minimum level of specific job knowledge to ensure a capacity to perform the work. The chapter suggests that additional research and discussion is necessary to clarify and reach professional consensus regarding two competing perspectives about the meaning of differential prediction. One perspective views differential prediction as an extension of validation requiring the criterion to be theoretically construct-relevant to the predictor. The other views differential prediction as analysis of valued outcomes requiring the criterion to be an encompassing measurement of valued outcomes created by employees.
