This research addresses the dual aims of selection research: the understanding and prediction of job performance. Two areas of research regarding criterion construct validity are examined and a research model is developed in an attempt to integrate this literature. This research model formalizes suggestions made by James (1973) and sets forth different levels, referents, and methods for criterion validation. A series of hypotheses regarding the interrelationships among alternative job performance measures and the relationships between criteria and predictors are presented. A longitudinal study was conducted to test this Job Performance Model in a field setting. Five measures of job performance and six ability tests for performance prediction were examined for sewing machine operators in a garment manufacturing plant. Data analyses indicated: High convergent validity among multiple methods of job performance measurement when the level of specificity was matched; Low to insignificant predictability of the alternative job performance criteria; and Differential prediction of job performance, depending on the method and referent for performance evaluation. lt was concluded that measurement characteristics of job performance criteria represent boundary conditions for subsequent prediction. The model presented here has merit for addressing the interrelationships among multiple performance criteria as well as the relationships between criteria and performance predictors. / Ph. D. / incomplete_metadata