Discriminability in multidimensional performance evaluations

Abstract

A series of behavioral expectation scale (BES) applications were analyzed in an attempt to point out an appropriate number of dimensions to be included in such studies. Data from 4 independent samples, representing 3 different occupations, and incorporating a total of 436 multidimensional evaluations were subjected to factor analysis. Results reflected the lack of unique information contributed when the number of dimensions exceeds 9. The problem of lack of dimension independence was discussed in terms of theory and application to multidimensional performance evaluation. Suggestions are advanced for limiting the number of dimensions as a potential solution to information redundancy.