LATENT CLASS ANALYSIS OF CHOICE BEHAVIOR

Abstract

The purpose of this paper is to propose a new model which combines latent class analysis and the multinomial logit model. Latent classes of this model are introduced to explain and predict choice behavior effectively. Simulation study demonstrates this model as good compromise to deal with individual differences. Finally, this model is applied to the analysis of the real data, and three meaningful latent classes were found to explain consumers' car selection.