EVALUATI0N OF DIFFERENT APPROACHES FOR MARKET SEGMENTATION AND A PROPOSAL

One of the most popular subject in marketing area, today, is Market Segmentation. During the last few years, multivariate statistical techniques, such as nonmetric multidimensinol scaling, discriminant analysis, canonical correlation, and cluster analysis or numerical taxonomy, have been applied to a wide variety of business areas, especially behavioral sciences and marketing. Form consumer behavior theory we know that sets of variables, namely a set of buying behavior variables, a set of socioeconomic variables, and a set of personality characteristic variables should be taken into account in market segmentation. This complexity of market segmentation forces the marketing analyst to use multivariate statistical techniques for studying market segmentation. However it is relatively easy to apply these multidimensional statistical techniques by the available conv puter programs, such as BMDO series, to market segmentation problem, we have to keep at least two important points in our minds. The first point is that ali of these multivariate techniques are proximities or approximations. And the second point is that the available tests to test the results of the multivariate analysis are not so powerful and in some techniques there are no tests available at ali. In this article we like to take. a real life market segmentation problem, evaluate different approaches, and propose an original approach to solve the problem in terms of two criteria or points given above.