Abstract: | This paper considers the general problem of analyzing data for job similarities/differences. Cluster analysis and univariate analysis of variance, which are recent suggestions for attacking this problem, are briefly reviewed. The suggestion made in this paper is to use multivariate analysis of variance, accompanied by a multivariate extension of the well known proportion of variance index, ω2. Discriminant analysis and related techniques are suggested to provide information regarding specific hypotheses. The potential users are provided with the references to well known computer packages that allow all the analyses to be performed easily, rapidly, and accurately on their own data. Appropriate interpretations of each result are also indicated, and illustrated with an example. |