The structure of role perception: A comparison of two methods of assessment

To test the stability of role perception multidimensional scaling analyses were applied to family role constructs (Mother, Father, etc.) based on data from 2 methods which differed significantly in terms of the types of judgments of stimulus dissimilarity which were required. In the present application demonstrating stability of role perception reduced essentially to the problem of showing the statistical (rather, of course, than the strict mathematical) equivalence of 2 presumably multinormal data matrices. The problem, however, was complicated also not only because of the existence of individual differences in perceptual style but also because of the possible interaction of these with the method used for the assessment of subject judgments. An attempt was made, therefore, for both a statistical comparison as well as pattern comparisons after the data were factored along the lines of Tucker-Messick.