Comparison of several procedures for generating J-coefficients

Abstract

J-coefficients were examined as estimators of criterion-related validities for the occupation of machinist in a medium-sized international manufacturer of precision automotive accessories. Test and job performance components of the J-coefficient were estimated from supervisory, incumbent, coworker, policy capturing, importance, and test expert ratings. The results indicated that combinations of estimates from several sources generated J-coefficients that reproduced the pattern and mean values of the validities. Supervisory and incumbent sources were superior to test expert source for estimating test components, whereas co-worker, policy capturing, and importance sources were equally effective for estimating job performance components. The implications of the J-coefficient are discussed for personnel selection. (PsycInfo Database Record (c) 2025 APA, all rights reserved)