Multiattribute perceptual mapping with idiosyncratic brand and attribute sets

Abstract

This article proposes an extremely flexible procedure for perceptual mapping based on multiattribute ratings, such that the respondent freely generates sets of both brands and attributes. Therefore, the brands and attributes are known and relevant to each participant. Collecting and analyzing such idiosyncratic datasets can be challenging. Therefore, this study proposes a modification of generalized canonical correlation analysis to support the analysis of the complex data structure. The model results in a common perceptual map with subject-specific and overall fit measures. An experimental study compares the proposed procedure with alternative approaches using predetermined sets of brands and/or attributes. In the proposed procedure, brands are better known, attributes appear more relevant, and the respondent’s burden is lower. The positions of brands in the new perceptual map differ from those obtained when using fixed brand sets. Moreover, the new procedure typically yields positioning information on more brands. An empirical study on positioning of shoe stores illustrates our procedure and resulting insights. Finally, the authors discuss limitations, potential application areas, and directions for research.