Hybrid choice models : principles and recent progress incorporating social influence and nonlinear utility functions

Hybrid choice models have been developed as an extension of discrete choice models, particularly multinomial logit models, in an attempt to include attitudinal variables. The quintessence of hybrid choice models is that a model of attitude formation is estimated and the estimated attitudes are added to the commonly used set of attributes in discrete choice models: attributes of the choice alternatives and socio-demographic variables. The most commonly applied model is based on linear specifications, both for the attitude model and the utility function. In this review paper, we discuss the principles underlying the hybrid choice model, summarize the specifications used in previous applications of the model and then continue discussing recent progress that added social influence to the model specification and replaced the linear specification of the utility function with a nonlinear function