The central idea of -gllamm- is to integrate out the random effects.
If you don't have latent variables in your model, and all variables
are observed, -gllamm- is not so useful -- in fact it has difficulties
when you try to restrict variances to be zero. With this setting, you
are in the realm of econometric simultaneous equation models estimable
with say -reg3-. However the latter does not support -svy- features
(and in fact -_robust- estimation, which is the principal mechanics
behind -svy-). I think the question of robust standard errors in
-reg3- came up a month or so on statalist though. You may have hope
with a combination of -reg3-, -bs4rw- (by Jeff Pitblado) and
-bsweights- (me) for bootstrap standard errors.
On 3/21/09, MAY BAYDOUN <[email protected]> wrote:
> Dear Statalisters,
>
> I was wondering if one can use gllamm to carry out structural equations modeling (without latent variables, just measured ones) and accounting for sample weights and design complexity (PSU, stratum etc.). If only weights can be accounted for that can be sufficient, but if all can be accounted for that would be better. Is there a way to combine svy: with gllamm? Thanks for your help on this,
>
> Sincerely yours,
>
> May
>
>
> May Baydoun, PhD in Epidemiology (UNC-Chapel Hill) Staff Scientist, National Institute on Aging, NIH/IRP, Biomedical Research Center, Baltimore, MD
>
>
>
> *
> * 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/
>
--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
* 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/