COMPARISON OF MULTIVARIATE GROUPING TECHNIQUES COMMONLY USED WITH PROFILE DATA

A number of multivariate procedures have been used to group subjects from profile data. Three of the more widely used techniques are (1) factor analysis with oblique or orthogonal rotation; (2) distance measures followed by hierarchical grouping; and (3) correlational cluster analysis. In attempting to contrast and compare the efficacy of these techniques two samples of 150 subjects each were separated into three groups based upon psychological, behavioral and statistical criteria. The resulting subjects from this selection process were grouped using each of the three multivariate procedures. Comparisons were made between procedures on the basis of correct placement, incorrect placement and residual subjects. It appeared that the correlational cluster procedure was the more efficient with no differences being found between the factor analytic or hierarchical grouping approaches. © 1969, Taylor & Francis Group, LLC. All rights reserved.