Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: stata equivalent of proc mixed


From   Buzz Burhans <wsb2@cornell.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: stata equivalent of proc mixed
Date   Wed, 26 Nov 2003 09:33:15 -0500

As others have described, there a number of options for approaching mixed models in Stata, and in fact it takes a fair bit for some of us to work through the multiple options for working with such data in Stata. Perhaps the closest option to Proc MIxed, as David suggests, is -gllamm-, which is a great and flexible routine with some very nice diagnostics available through -gllapred- and -gllasym-. As I understand it however, one caution regarding use of =gllamm- for longtitudinal heirarchical data such as might be modeled with Proc Mixed is that -gllamm- does not provide the capability to model the residual covariance matrix structure, thus it is unable to capture the time related covariance of the residuals in cases where the fixed time effects are not strong enough to absorb the entire time associated correlation. -Gllamm- does provide a robust option, which allows one to relax the iid assumption for the residuals in such cases, however, the inferences about the model are then subtly changed also.

Buzz Burhans


At 10:28 PM 11/25/2003 -0600, you wrote:

Hi all,

Does Stata have an equivalent of SAS's proc mixed for time series data?

Thanks,
Jim Seward

Stata does implement certain models as others have mentioned (xtreg), and perhaps your problem is subsumed by those, but Stata does not have a powerful general mixed model like Proc Mixed. It will be a while, certainly more than a year, before anything appears in official Stata. It's on the list, according to tech-support. As for GLLAMM, it is slow, but that's an interesting option. If you don't have SAS, look to R. NLME/LME or LME4 in R will do most of what Proc Mixed will do. And it's free if you have the time to learn R. If you know both Stata and R, then you know the only two statistical packages with good programming languages. There is a useful text by Bates detailing NLME/LME but not yet LME4. Of course, if you can get my within what Stata offers, I envy you.

-Dave

*
* 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/
Buzz Burhans
wsb2@cornell.edu


*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index