MULTIDIMENSIONAL SCALING WITH A PRIORI DIMENSIONS

W aern Y. Multidimensional scaling with a priori dimensions. Scand. J. Psychol., 1972, 13 , 178–189.—This paper presents a method of multidimensional scaling by direct similarity estimates. The method is presented as an alternative to other multidimensional scaling methods, when hypotheses can be made concerning the dimensions underlying the perceptual variation. The method is used with five different sets of stimuli and compared, to an as to dimensions assumption-free component analysis, based on a vector model. The two methods resulted in similar configurations in most cases. One difference was found in a colour material, where the assumed dimensions could be regarded not to be orthogonal to each other. Here the a priori scaling revealed the assumed configuration as projected on the oblique axes whereas the component analysis could not give this information.