ABSTRACT
There is much work in the social sciences on discrete choice models. Among those, the multinomial logit is the most common model used for analyzing survey data when the number of choices is greater than 2. In many cases, however, the underlying assumptions leading to the traditional maximum likelihood estimator (MLE) for the logit model are inconsistent with the perceived process that generated the observed data. One of these cases is the random parameter (RP) logit model (also known as “mixed logit” – see Revelt and Train 1998) that can be viewed as a variant of the multinomial choice model. In this chapter we formulate an Information-Theoretic (IT) estimator for the RP mixed logit model. Our estimator is easy to use and is computationally much less demanding than its competitors — the simulated likelihood class of estimators.
