EVALUATING MILITARY SELECTION AND CLASSIFICATION SYSTEMS IN THE MULTIPLE JOB CONTEXT

This article contrasts 2 research paradigms used in designing and eval uating test batteries and predictor composites for selection and class ification in the multiple job context. Selection and classification te st batteries such as the Armed Services Vocational Aptitude Battery ar e designed and evaluated primarily in terms of incremental predictive validity. An alternative research paradigm offered by differential ass ignment theory (DAT) uses a standardized mean predicted performance me asure as the index of efficiency. A review of the DAT research paradig m and studies applying it to U.S. Army test and performance data forms the focus for a comparison between the findings of predictive validit y and DAT research. We suggest that the DAT research paradigm is a mor e appropriate vehicle for evaluating test batteries and predictor comp osites used for assignment in the military context.