ABSTRACT
This chapter presents an overview of the mixed logit model, highlights key concepts for researchers and practitioners venturing into this modeling area. It provides an overview of initial mixed logit applications to both transportation and aviation specifically. The chapter describes identification rules for mixed logits, many of which evolved out of earlier work done in the context of probit models. The mixed logit model is able to relax several assumptions inherent in the generalized nested logit (GNL) and Network Generalized Extreme Value (NetGEV) models, that is, it is able to incorporate random taste variation, correlation across observations, and heteroscedasticity. Halton draws are commonly used to generate support points for mixed logit models. However, it should be noted that when estimating high-dimensional mixed logit models, alternative variance-reduction techniques need to be investigated because Halton draws generated with large prime numbers can be highly correlated with each other.
