Robert Bozick--
I can't tell you how to properly deal with the survey design issue,
but I can suggest a workaround: -help bs- will show you how to
bootstrap your estimates, and -help bsample- will get you started on
doing a manual bootstrap, which might be required to get unique
identifiers for individuals, and to include weights in each call to
-xtreg, mle-. You might also investigate -ssc desc gsample- for more
options. All of these support a cluster() option. If you can post a
copy (or subsample) of your data on the web somewhere, and provide a
link, and the command you want to run, someone might be motivated to
write some code to address your specific problem.
Note also that if you want to try a fixed effects model, you might
want to -ssc install xtivreg2- which allows pweights and the cluster()
option (these two get you most of the way to -svy- results), and
provides some good diagnostics.
See also
http://www.stata.com/help.cgi?program_properties#svy
On 11/20/06, ROBERT BOZICK <[email protected]> wrote:
Thanks Nick. I definitely need to upgrade my software!
Does anyone else know how to properly deal with the survey design issue when estimating these types of models or this capability (svy commands or otherwise) not available when using STATA?
*
* 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/
