Accommodating for Taste and Variance Heterogeneity in Discrete Choice†

Understanding and accommodating heterogeneity in variance (also referred to as heteroscedasticity) and taste has become a major area of research within discrete choice analysis. Both scale and taste heterogeneity can be specified as continuous or discrete, the latter can be associated with socio economic characteristics (i.e. observed heterogeneity) or it can be derived probabilistically (i.e. unobserved heterogeneity). Within the context of the Mixed Logit models, unobserved heterogeneity can be represented by a continuous function, a discrete mixture or using a combination of both. This paper uses data from two recreational site choice studies (one elicited through stated preference methods and one through revealed preference methods) to compare various model specifications for accommodating both scale and preference heterogeneity. Results show that model fit, welfare estimates and choice predictions are sensitive to the manner in which both types of heterogeneity are accommodated.