The construct validity of a Belgian assessment centre: A comparison of different models

This study investigates the construct validity of a Belgian assessment cent re through a comprehensive set of confirmatory factor analysis models. In p articular, the general confirmatory factor analysis approach as well as the correlated uniqueness approach are fitted to the same data. Results replic ate Sagie and Magnezy's (1997) finding that the correlated uniqueness model is appropriate to represent exercise effects in assessment centres, as a g ood fit and no estimation problems are obtained. In this model the dimensio ns explain 36% of variance, revealing evidence of convergent validity. Poss ibly, this is due to the careful design of this assessment centre. Evidence of discriminant validity, however, is not established, as the dimension fa ctors are highly correlated.