--- yan yu <[email protected]> wrote:
> I have a problem with using streg for weighted data. Unlike with
> unweighted data, streg with unshared gamma frailty for weighted data
> does not report a likelihood ratio test for unobserved
> heterogeneity. I used the code [pw=sampleweight] in stset to
> incorporate sample weights. I tried to do the likelihood ratio test
> manually, by running models with and without the frailty
> parameter. But this seemed not to work: The loglikelihoods from streg
> are substantially different between when frailty(gammma) is included
> (ll=-464), and when frailty is not included (ll=448). With only 1
> degree of freedom, the difference in ll seems to be too big,
> considering that the theta estimate is .45 with a standard error of
> .48.
>
> Do you have any idea about what is going on? Could you suggest some
> appropriate test for heterogeneity?
The problem is you can't do a likelihood ratio test with robust, and
pweights imply robust. You can find more on this here:
http://www.stata.com/support/faqs/stat/lrtest.html
Normally you would do a Wald test instead, but I remember vaguely that
there were problems with a Wald test on variances in random effects
models, and frailty models are pretty similar... So I am stuck on this
one.
-- Maarten
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
___________________________________________________________
Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for
your free account today http://uk.rd.yahoo.com/evt=44106/*http://uk.docs.yahoo.com/mail/winter07.html
*
* 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/
