Thanks much to both Arne and Austin. Very helpful!
-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of Arne Risa Hole
Sent: Wednesday, December 12, 2007 5:36 PM
To: [email protected]
Subject: Re: st: RE: Multinomial logit panel model
Alternatively to Austin's suggestion you could run a multinomial logit
with random intercepts, see section 9.3 in the -gllamm- manual for an
example. You can also estimate this type of model using the -mixlogit-
module which is available from SSC, type -ssc describe mixlogit- and
follow the instructions to install. A paper describing the module was
published in the most recent version of the Stata Journal (7-3), you
can download a pre-publication draft version from my website:
<http://www-users.york.ac.uk/~ah522/home.htm>. See in particular the
second set of examples in section 4 which use the same dataset as in
the -gllamm- manual.
The advantage of using -mixlogit- is that it uses simulation rather
than quadrature so it is faster than -gllamm- in most cases.
Hope this helps.
Arne
On 12/12/2007, Austin Nichols <[email protected]> wrote:
> Bradley J <[email protected]>:
>
> Does that mean you have 4800 obs on 200 indiv? Are those 24 repeated
> measures supposed to be independent measures of the same quantity?
>
> I'm thinking you could just run mlogit and cluster on the 200 indiv.
> If the X is identical across repeated obs, and the Y does not vary
> much, the clustering will be near perfect, and you will get
> arbitrarily close to an mlogit using the collapsed data of 200 obs.
>
> On Dec 12, 2007 4:48 PM, Anderson, Bradley J <[email protected]> wrote:
> > Apologies if I'm asking a naive question. But I have done some searching and I'm not finding answers. Is software available to estimate a multinomial logit model with panel data? I'm thinking this can maybe be done in GLLAMM. Are there alternatives. Also, I'm being asked to estimate models and I'm not sure we have sufficient data. The response variable has 3 categories, sample size is 200, and there are 24 repeated measurements of the response variable. Any thoughts on how to approach this would be much appreciated.
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
