Statalist


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

Re: st: xtmixed with random slope - STATA 11 vs STATA 10


From   rgutierrez@stata.com (Roberto G. Gutierrez, StataCorp)
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: xtmixed with random slope - STATA 11 vs STATA 10
Date   Thu, 29 Oct 2009 10:20:26 -0500

Ada Lo <loadaol@gmail.com> writes:

> I started using STATA SE/10 for Mac and then upgraded to STATA SE/11.  I was
> able to use xtmixed to fit a random intercept random slope model with
> unstructured covariance in STATA 10 but when I ran the exact same command in
> the same dataset in STATA 11, the iteration would not stop.  Basically, I
> stopped it when it reached 100... Does anyone know why and how to make it
> work in STATA 11?

> The basic model is - xt: xtmixed Memory age agesq i.retest i.dropout ||
> StudyID: i.retest, cov(unstructured) mle

> By the way, I do notice that the covariance structure could be the problem
> because when cov(independent) was specified, both STATA 10 and 11 could
> generate the model. However, because these are longitudinal data (3 repeated
> measurements over 8 years with roughly 490 participants) and the data are
> expected to correlate within individuals, I thought cov(unstructured) would
> be the best option given that autogressive covariance structure is not
> available in xtmixed...

Although very rare, changes in convergence paths on the same model fit from
version to version most likely result from refinements made to the optimizer.
If Ada can send me the data, I'd be glad to verify that this is indeed the
case and perhaps suggest a workaround.

--Bobby
rgutierrez@stata.com
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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