The Nonmetric Multidimensional Approach Applied to Rank-Order Similarity Data

Abstract

The purpose of this study was to apply the nonmetric multidimensional approach to the analysis of rank order similarity data. Ss were nine students at Long Beach State College. All were volunteers. Ss were given the task of rank ordering, on the basis of similarity, nine color chips which varied in saturation and brightness. The method developed by Coombs was applied to these data, and a comparison was made with the results of Torgerson's data which were obtained by the method of complete triads. The present study shows that these methods yield comparable results. Coombs' method was also employed in the analysis of the psychological space of individual Ss. In this case, the results are not so definitive. For some individuals, the analysis yielded results highly correlated with the Munsell system, whereas for others, it did not. Several possible explanations are discussed.