Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

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

Re: st: RE: RE: predicted probabilities of interaction term

From   Richard Williams <>
To, "" <>
Subject   Re: st: RE: RE: predicted probabilities of interaction term
Date   Wed, 18 Jul 2012 12:26:05 -0500

At 10:49 AM 7/18/2012, Valle, Giuseppina wrote:
Thank you for your help -- I was unaware that I needed to use the subpop command when looking at the gender differences with the svy command. In terms of the postestimation commands, I have not located any syntax available that will produced predicted probabilities of the interaction terms with a mlogit. I have 4 values for "race" and 3 values for "relig." I assume that using prtab along with specified values of these two variables would not produce the estimates I am looking for, as that is the only command I have used for predicted probabilities.

If you want to get predicted values for individual cases use the -predict- command. If, on the other hand, you want to get predicted probabilities for certain combinations of values, check out the margins command. e.g. do something like

webuse nhanes2f,clear
svy: mlogit health age
margins race#sex, predict(outcome(1)) atmeans
margins race#sex, predict(outcome(2)) atmeans
margins race#sex, predict(outcome(3)) atmeans
margins race#sex, predict(outcome(4)) atmeans
margins race#sex, predict(outcome(5)) atmeans

I believe that would more or less clone what -prtab- does, although I personally might drop the -atmeans- option.

The margins command is powerful but complicated (and also a little tedious with multiple outcome commands like mlogit). The current issue of the Stata Journal has an article on it, and there is also a handout at


Giuseppina Valle, M.S.
Department of Sociology
Doctoral Candidate
Florida State University
526 Bellamy Building
Tallahassee, FL 32306-2240


From: [] on behalf of William Buchanan []
Sent: Tuesday, July 17, 2012 5:11 PM
Subject: st: RE: predicted probabilities of interaction term

Hi Giuseppina,

One suggestion that I have would be to use the -subpop()- option of the
-svy:- prefix.  This is something that surfaces on the Statalist regularly,
although it may not be an issue in your specific case.  Either way, the help
file for -svy- includes the following line:

"Warning:  Using if or in restrictions will often not produce correct
variance estimates for
      subpopulations.  To compute estimates for subpopulations, use the
subpop() option."

With regards to your other question, I would look at -help mlogit
postestimation- to get an idea of what is already available in Stata to
calculate the predicted probabilities of the outcome after a multinomial
logit model.


-----Original Message-----
[] On Behalf Of Valle, Giuseppina
Sent: Tuesday, July 17, 2012 1:38 PM
Subject: st: predicted probabilities of interaction term

I was wondering how to run predicted probabilities for a multinomial
regression with an interaction term.

Here is the syntax I am using:

svy: mlogit timing pledge sexed i.race i.relig i.race#i.relig ///
        wave4age stepfam singlemo singleda other HSdeg somecoll ///
        colldeg misseduc lowmidinc highmidinc highinc missinc ///
        midwest south northeast if gender==0,  ///
        rrr baseoutcome(2) nolog

I would appreciate any suggestions. Thanks!

Giuseppina Valle

*   For searches and help try:

*   For searches and help try:

*   For searches and help try:

Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index