Introduction to Choice Models

StatWizards Academy – Introduction to Choice Models

Welcome to StatWizards Academy, a place offering free tutorials on subjects we cover. Like a growing number of our academic colleagues, we are committed to supporting world-wide education, and this is our initial contribution. We hope the materials on this and future pages along with our free product demos will expand the use of discrete-choice models, Fader-Hardie forecasting models and future tools we will continue to add to our portfolio.

This page covers the general topic of discrete-choice models. We have assembled a host of materials from around the web into this one place and hope you find them useful.


We start with our own basic, comparatively non-technical, and slightly dated introduction to choice modeling which appears to the right. This tutorial draws on a past U.C. Berkeley lecture on the subject. To anyone unfamiliar with choice modeling, this is a good place to begin.


Of the many lectures available on the subject, especially on the web, no list would be complete without Dan McFadden's Nobel prize lecture, given in honor of his economics prize in 2000 for developing early discrete-choice methods. It's a good introduction to the exceptional usefulness of this family of techniques.

If you can handle a more math-intensive set of lectures, we recommend Moshe Ben-Akiva's two introductory M.I.T. lectures on discrete-choice analysis. The first lecture introduces the concept of choice analysis and discusses the random-utility model; the second covers model specification, estimation and forecasting.


In a world of the Internet, it would be folly to try to list all the introductory papers available on this subject. Instead, Google the term "discrete-choice introduction" and choose from the list. For a quick start on the subject, we recommend Sawtooth Software's technical paper covering a wide array of conjoint techniques.


A number of excellent texts on discrete-choice models exist, covering the topic with differing breadth and depth. The following are listed by author and include links to each book on

Ben-Akiva, Moshe, Steven R. Lerman, Discrete Choice Analysis: Theory and Application to Travel Demand. Cambridge, Massachusetts: MIT Press, 1985. ISBN-13: 978-0262022170. A good, though math-intensive introduction to discrete-choice analysis. Because this and the 1985 Train text were published before modern advances in discrete-choice theory, their coverage is necessarily limited to the state of the art prevailing at that time.

Louviere, Jordan and others, Stated Choice Methods: Analysis and Applications. Cambridge, England: Cambridge University Press, 2000. ISBN-13: 978-0521788304. A text of astonishing breadth and clarity, covering topics in experimental design, modeling and forecasting and including numerous case studies.

Train, Kenneth, Qualitative Choice Analysis: Theory, Econometrics, and an Application to Automobile Demand. Cambridge, Massachusetts: MIT Press, 1985. ISBN-13: 978-0262200554. A clear, well-written introduction to discrete-choice theory that covers simple and nested logit.

Discrete Choice Models with Simulation. Cambridge, England: Cambridge University Press, (2nd edition) 2009. ISBN: 0-521-01715-7. An update to Dr. Train's prior text, this book introduces the concept of simulated maximum likelihood, the technique that renders mixed-logit models possible. This text is also available online and is used in Dr. Train's online course, described below.

Online Courses

As distance learning grows in popularity, a number of online courses taught by excellent lecturers are available. We present them in ascending order by cost.

Discrete-Choice Models with Simulation, University of California, Berkeley. Kenneth Train videotaped his course at U.C. Berkeley and makes it available online for free. Dr. Train is a superb lecturer, and we recommend this course highly.

Discrete Choice Modeling and Conjoint Analysis: Taught by the first-rate lecturer Tony Babinec, this course emphasizes practical applications of discrete-choice models. Cost: approximately $500. Academic price: approximately $400.

Introduction to Latent Class Modeling: Statistical Innovations. Dr. Jay Magidson with Dr. Jeroen Vermunt developed Latent GOLD® Choice, the touchstone for latent-class choice modeling. This course introduces the concept and demonstrates the use of the software.

On-Site Courses

Those who would like a more in-depth exposure to discrete-choice topics can sign up for on-campus courses. Here is a partial list.

Discrete Choice Analysis—Predicting Demand and Market Shares: M.I.T. Short Programs. One of the seminal thinkers behind discrete-choice theory, Moshe Ben-Akiva, offers a short course at M.I.T.'s campus on modern applied discrete-choice modeling. Highly recommended by participants.

Found in Topics