Any of a number of procedures for the fitting of a selected geometric model to a body of multivariate data. In the original form advanced by Warren S. Torgerson, multidimensional scaling dealt with judgments about stimuli, with its major aim being the derivation of psychological dimensions presumed to underlie the judgments, and the determination of scale values for stimuli along the derived dimensions. Classical multidimensional scaling was a significant methodological advance in the characterization of complex stimulus domains. If forced to rely exclusively on unidimensional scaling methods, a researcher could characterize a multidimensional stimulus domain only by first scaling the stimuli on individual dimensions and later assembling the scaled dimensions in a postulated model. Even then, the model would often be incomplete because of insufficient foresight in choice of dimensions, or inability to determine the relations between, and relative importance of, the chosen constituent dimensions.