Using Segmentation Approaches for Better Prediction and Understanding from Consumer Mode Choice Models

Abstract

The diagnostic and predictive efficacy of market segmentation and the relative power of two segmentation schemes (benefit and situational) are investigated by using a market share probabilistic choice model (LOGIT) as a dependent variable. The model relates consumer perceptions of several alternatives on various characteristics to discrete choice and is estimated first for the entire sample and then for each of the benefit and situational market segments. Empirical findings derived from data on consumers’ transportation preferences, perceptions, and choices in the San Francisco Bay area suggest that the models provide fairly accurate estimates of market share and that using the segmentation concept affords diagnostic and predictive advantages.