If you are able to estimate the covariance matrix in some way, then you can convert your covariance matrix to a correlation matrix and use the corr(fixed) option in -xtgee-. Otherwise, even if you don't estimate the covariance matrix, if you just specify corr(ind), with the robust option the resulting estimates and standard error would be consistent estimates, only less efficient than if you had the correct correlation structure specified.
Perhaps others could comment on how to go about estimating the covariance matrix?
Tim
-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Jisheng Cui
Sent: 20 January 2009 06:24
To: [email protected]
Subject: st: additional cluster effect in GEE
I wonder whether someone is aware how to adjust for an additional cluster effect, such as schools, in GEE analysis. This cluster effect is different from the panel (or time) effect.
Jisheng (James).
*
* 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/
*
* 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/
