Evaluating Market Segmentation Studies: The Properties of R²

Virtually all of the empirical studies on market segmentation can be characterized as unsuccessful in terms of R². That is, the set of independent variables used have not explained much of the variance in purchasing behavior. R²'s of 0.01 and 0,02 are not uncommon. In this article we show that R² is an inappropriate statistic to use in evaluating most of these studies. We show that good models with good sets of variables can yield low (and even zero!) R²'s. In this context “good” variables are ones that explain the average purchasing behavior of a family. The previous studies on market segmentation should be re-evaluated in light of the results presented here.