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: Estimating MNL model

From   Maarten Buis <>
Subject   Re: st: Estimating MNL model
Date   Fri, 11 Oct 2013 21:00:59 +0200

A multinomial logit model can be estimated in Stata using -mlogit-,
see -help mlogit-. You could have found that out yourself by typing in
Stata -findit multinomial logit-.

I assume that with "specific variables" you mean variables that
influence the probability of only one outcome category. This is
conceptually problematic: If the probability of one outcome changes,
than the probabilities of one or more other outcome categories has to
change as well, otherwise the probabilities do not add up to 1. A
multinomial logit solves that by focussing on the ratio of the
probability of one outcome catogary and a fixed baseline category. So,
if you constrain a variable to influence only one variable (using
-constraint-, see: -help constraint-), any change in the probability
of that outcome category is compensated by a change in the probablity
of only the base category. I suspect that this is not what you want,
but it is hard to imagine another solution that would be generally
reasonable. It may be that your specific problem allows for a
solution, but since you did not tell us any of the specifics we cannot
help you with that.

Hope this helps,

On Fri, Oct 11, 2013 at 6:23 PM, Nabin Kafle <> wrote:
> How can I estimate MNL moedl in stata with alternate specific and
> generic variables. What would be the command to describe them and run
> the model. Thanks in advance
> --
> Nabin Kafle
> *
> *   For searches and help try:
> *
> *
> *

Maarten L. Buis
Reichpietschufer 50
10785 Berlin
*   For searches and help try:

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