Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: question on MLE d0 method for multilogit model with invariant regressors

From   Partha Deb <>
Subject   Re: st: question on MLE d0 method for multilogit model with invariant regressors
Date   Sun, 11 Apr 2010 22:50:34 -0400

Unless I've misunderstood, it seems that your first model specification is -mlogit- while the parameters of your second specification are unidentified. In the multinomial choice framework you describe, if variables vary across alternatives, the parameter must be constant (or at least not vary independently with each choice). That's the -clogit- .

Hope this helps.


Hey Sky wrote:
Hey, All

I try to do a mle d0 estimation for multilogit model with invariant regressors and have searched internet but no luck. I wish I can get some help here. any suggestion is appreciated.
here is the question:

given a period, the probability of a person choose choice k is:

pr(choice=k)=exp(x*beta_k)/(exp(x*beta_1) + exp(x*beta_2)+...)

that is, the regresors are invariant, x, but the parameters are variant when choose diffecrent choice, beta_1, beta_2, etc.
a little complicated case is when making the regressors and parameters both changes along with choices, i.e.

pr(choice=k)=exp(x_k*beta_k)/(exp(x_1*beta_1) + exp(x_2*beta_2)+...)

how to do it with the Stata MLE method? thanks for any clue you provide.
from Montreal

Make your browsing faster, safer, and easier with the new Internet Explorer® 8. Optimized for Yahoo! Get it Now for Free! at

*   For searches and help try:

Partha Deb
Professor of Economics
Hunter College
ph:  (212) 772-5435
fax: (212) 772-5398

Emancipate yourselves from mental slavery
None but ourselves can free our minds.
	- Bob Marley

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index