Removing Perceptual Distortions in Product Space Analysis

Abstract

Removing distortion in perceptual judgments has been the subject of several recent research studies. The authors examine the issue of identifying and removing perceptual distortions in product space analysis. The purpose of the article is to show how a simple data transformation can purge the influence of the common general factor which typically surfaces when compositional approaches to building product spaces are employed. The method, which relies on a two-step double-centering transformation, eliminates irrelevant sources of variation. It is fundamentally different from the previously recommended practices in that it does not require the researcher either to (1) throw away the first principal component or (2) arbitrarily select attributes in order to define an affect index which then is used to replace the original simple correlations with partial correlations controlling for the influence of affect captured by the index.