Assessor type, number of distinguishable dimension categories, and assessment centre construct validity

Overall previous research has found assessment centres to have low construct validity. Criticism of the analytical procedures used in such research have stimulated a re-evaluation of this issue. In the present study we have examined two variables whose influence on assessment centre construct validity has been neglected. These variables are assessor type (or profession, i.e. psychologist or manager) and the number of distinguishable dimension categories required to adequately represent the ratings. Using Marsh's (1989) approach to confirmatory factor analysis for the assessment centre ratings, we have found that two different models fit the psychologist and manager data. While all the predetermined dimensions were reflected in the psychologists' ratings (i.e. high construct validity), only two broad dimension categories: performance skills and interpersonal skills, accounted for the managers' ratings. Theoretical and practical implications of the findings are discussed.