D Ma wrote;
>>I am running a xtgee model like this:
xtgee y x1 x2 x3 ... ,family(nb) link(log) vce(robust) nolog
Since there are lots of zeros in the dependent count varaible y, I want to use a zero-inflated model. Nonetheless, the xtgee command in stata doesn't seem to have that option. Do you have any suggestions?<<
The point estimates from zinb won't be all THAT different from a GEE estimate. Use cluster(myidvar) for the VCE estimate and you should get fairly close to what you would have gotten with a GEE. That said, do you have any reason to believe that you need a zero inflation model? "Lots of zeros" can be totally consistent with a low rate NB variable.
Jay
*
* 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/
