The logit models have been quite useful in analyzing the discrete dependent variable. Specifically, multinomial logit(MNL) and conditional logit(CL) models have been the workhorses in analyzing discrete choice data with polychotomous nominal categories. However, there are some limitation in these models which restrict the applicability of these models. Three well-known limitations in these models are the assumptions of homogeneous preference among individuals, independence of irrelevant alternatives(IIA) and no correlation across time and individuals. One of the recent advances in categorical data analysis is mixed logit(MXL) model, or random parameter model. MXL obviates the well-known limitations of standard logit models by allowing heterogeneous preferences, unrestricted substitution patterns, and correlation in unobserved factors over time. The purpose of this paper is to introduce MXL and its applications.
Keywords: Mixed Logit, Multinomial Logit, Conditional Logit