[R] Generalized Estimating Functions

Thomas Lumley tlumley at u.washington.edu
Mon May 13 18:39:18 CEST 2002

On Sun, 12 May 2002, Frederico Zanqueta Poleto wrote:

>   Hi,
> I'm trying to fit a marginal model via GEE but I'm getting strange
> results and few problems.
> If I set the working correlation as exchangeable I'm getting the same
> fitting when I set as independent. Comparing to SAS results it shouldn't
> happen.
> If I try to use another working correlation (like AR-M or stat_M_dep), R
> just exits without giving any error message.

We probably need an example.  The `OME' example in help(gee) gives
different results using exchangeable and independence working
correlations, so it isn't a universal problem.

> Another doubt is how R estimates the scale parameter for Poisson? I get
> 1.36 in R, 1.25 in SAS (estimation by the square root of DEVIANCE/DOF)
> and 1.17 (estimation by the square root of Pearson's Chi-Square/DOF).
> I'm using R 1.4.1 but all the problems apply for S-Plus 2000 too.

This is because 1.36=1.17^2, I should think.  The disperson parameter from
gee() is a variance scale factor not a standard deviation scale factor.


Thomas Lumley			Asst. Professor, Biostatistics
tlumley at u.washington.edu	University of Washington, Seattle

r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list