Reevaluation of Preference Distribution Analysis

Abstract

Many articles have discussed the value of interpreting preference data for a distribution of buyer preferences arrayed on a product attribute or dimension [1-3, 5, 6, 9]. Products are seen as ranging from light suds to heavy suds, weak chocolate flavor to strong chocolate flavor, etc. The premise is that differences in buyers' preference patterns reflect these underlying dimensions. Kuehn and Day [9] discussed this approach with a technique called preference distribution analysis. Benson [1] has developed another technique that uses this same dimensional concept. Recently, both Benson and Day [1, 2, 5, 6] presented empirical data demonstrating the usefulness of these techniques over simple preference testing. Here, the arguments on the popularity fallacy and the majority fallacy certainly raise issue with the usual interpretation of simple preference data, especially considering market segmentation strategies. This article's purpose is not to question the concept of this dimensional approach but rather to question the data collection procedure used to develop the preference distribution. The argument is that the paired comparison procedure, as currently used, can lead to misinterpretation of the underlying preference distribution. This bias can be illustrated by referring to Day's JMR article on the development of a preference distribution for chocolate ice cream [5].