Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: xtgee and corr(ind) vs corr(exch)


From   Jhilbe@aol.com
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: xtgee and corr(ind) vs corr(exch)
Date   Sat, 2 Nov 2002 16:13:32 EST

The xtgee command is the command for Generalized Estimating Equations (GEE). 
Built on the iterative estimating algorithm of GLM, or generalized linear 
models,  the GEE algorithm with the exchangeable correlation matrix updates 
the variance estimate at each iteration with leveled modified Pearson 
residuals. This in turn makes an adjustment in fitted values. The result is 
that one should expect the parameter estimates to be different when using the 
exchangeable correlation matrix against the indepentent correlation matrix 
(which is the same as simply modeling the data using the standard glm 
command. 

When using glm with a robust estimator, teh paramter estimates remain the 
same since the adjustment is made post hoc; i.e. one additional iteration is 
made, but with an adjustment made to the variance-covariance martix. The same 
is the case with GEE when using the robust variance estimate - the parameter 
estimates will not change. 

I trust that this helps clarify why you see different parameter estimates in 
the two models. 

Joe Hilbe



*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index