Discrete Choice Models

Discrete Choice Models analyze decisions where individuals choose one option from a finite set (e.g., car vs. bus). These models are based on utility maximization, where each choice is made to maximize satisfaction, represented as:

Uij=βXij+ϵij 

Here, Uij​ is the utility, Xij​ represents observable factors, and ϵij​ accounts for unobservable influences. Choices are modeled probabilistically, with the likelihood of choosing an option based on its utility.

Types of Models

1. Logit Model: Assumes Gumbel-distributed errors; simple and widely used.
2. Probit Model: Assumes normally distributed errors, suitable for correlated choices.
3. Multinomial Logit (MNL): Handles multiple alternatives but assumes independence of irrelevant alternatives (IIA).
4. Nested Logit: Groups similar options into “nests” to relax the IIA assumption.
5. Mixed Logit: Allows random preference variations, offering greater flexibility.