Psychometric Methods in Marketing Research: Part II, Multidimensional Scaling

Abstract

Recently, we presented some views about the history, growth, and future of psychometric techniques in marketing research (Carroll and Green 1995). Our Part I commentary focuses on conjoint analysis and related methods. In this concluding editorial, we discuss multidimensional scaling (MDS) in marketing, which goes back even earlier than conjoint analysis-to at least the early 1960s, following Shepard's pioneering papers on the nonmetric analysis of proximity data (Shepard 1962a, b). Space does not permit a discussion of the many classes of methods for the analysis of proximity and preference data often included in a broad definition of MDS. In particular, our purview does not cover correspondence analysis and cluster analysis.I Excellent reviews of these two areas have been prepared by Hoffman, De Leeuw, and Arjunji (1994) and Arabie and Hubert (1994), respectively. In addition to these publications, we recommend reading DeSarbo, Manrai, and Manrai's (1994) review of latent class MDS and their (1993) review of nonspatial tree models. Each review presents an insightful and comprehensive coverage of these specialized areas. Subsequent sections of this editorial discuss the history and maturation of MDS in marketing, including models and applications of individual differences models, constrained MDS, stochastic MDS modeling, normatively based MDS models for optimal product design, and scaling models developed for analysis of scanner data. We conclude with an appraisal of the state of practice of MDS in marketing, some of the problems associated with the gap between models and applications, and some suggestions for increasing the practical utility of MDS in marketing research.