The multi-dimensional nature of job performance has led to the use of the multitrait-multirater method for analyzing performance evaluation data (Lawler, 1967). The requirements of convergent and discriminant validity of that approach were seldom satisfied in empirical studies. Arguments were made against the simple notion of convergent validity and for the separation of the analysis of ratings coming from raters at different organizational levels (cf. Borman, 1974). The Present paper discusses the multitrait-multirater method from a facet analytic perspective. It is proposed that performance evaluation variable can be constructed by a mapping with two facets: a rater's facet and a trait's facet. The former is specified as an ordered facet and the latter as a non-ordered facet. The resulting predicted structure is a cylindrex. Data from four different studies are analyzed and basically support the proposed framework. The regularity of the pattern of the variables' intercorrelations across different rater groups is remarkable. Different aspects of the cylindrex structure are discussed as alternatives for the requirements of the multitrait-multirater method.
No takes yet. Share an insight, caveat, or question.
Shapira et al. (1989) studied this question.