Understanding heterogeneous preferences in random utility models: A latent class approach

A finite mixture approach to conditional logit models is developed in which latent classes are used to promote understanding of systematic heterogeneity. The model is applied to wilderness recreation in which a branded choice experiment involving choice of one park from a demand system was administered to a sample of recreationists. The basis of membership in the classes or segments in the sample involved attitudinal measures of motivations for taking a trip, as well as their stated preferences over wilderness park attributes. The econometric analysis suggested that four classes of people exist in the sample. Using the model to examine welfare measures of some hypothetical policy changes identified markedly different welfare effects than the standard single segment model, and provided insight into the differential impact of alternative policies.