Judged Similarity and Brand-Switching Data as Similarity Measures

This article suggests how the transition matrix, which is central to many stochastic models in marketing, can be used as a similarity matrix for input to multidimensional scaling or cluster analysis in place of the more widely used judged similarity measures.

Abstract

This article discusses a procedure for building a perceptual map of a market based on revealed preference data. Specifically, it suggests how the transition matrix, which is central to many stochastic models in marketing [2, 5, 7, 8, 14, 15], can be used as a similarity matrix for input to multidimensional scaling or cluster analysis in place of the more widely used judged similarity measures [3, 4, 18]. This article thus attempts to demonstrate a conceptual link between stochastic models of market share and individual choice models.