The Search for Construct Validity of Assessment Centers: Does the Ease of Evaluation of Dimensions Matter?

Abstract

The purpose of the present study was to investigate the effect of ease of evaluation of dimensions on the construct validity of a selection assessment center conducted in 1993. High ease of evaluation dimensions, operationalized as the greatest proportion of highly diagnostic behaviors, were expected to demonstrate greater construct and criterion related validity. Multitrait-multimethod analysis and confirmatory factor analysis results indicated that high ease of evaluation dimensions demonstrated greater convergent and discriminant validity than low ease of evaluation dimensions. Contrary to predictions, however, there was little difference in the criterion related validity of the high versus low ease of evaluation dimensions. Moreover, the entire assessment center yielded extremely low predictive validity using both dimension and exercise scores as predictors. The implications of the findings from this study are discussed.