Weighted Empirical Adaptive Variance Estimators for Correlated Data Regression
These are standard error estimators for generalised linear models
fitted to correlated data that extend the model-robust sandwich
estimators used by Huber for independent data and Liang and Zeger for
longitudinal data. All of these estimators are described in
Lumley T., Heagerty P.J. (1999) ``Weighted Empirical Adaptive
Variance Estimators for Correlated Data Regression'' JRSS B 61:2;459-479
The most complete version is a tarred and gzipped R package. It can be used in S-PLUS as well
by unpacking, sourcing the file R/glmweave.R and copying the
S help files from .Data/.Help to your .Data/.Help
directory. Windows users may prefer the zipped version. I don't know how to set
this up as an S-PLUS for Windows library -- if you do I would be
grateful to be told -- but it has all the components.
The functions (with links to HTML documentation) are
The functions contain no compiled code, so they are easy to install on a variety of systems.
I also have Stata code for the Newey-West estimator for generalised
linear models (glmnewey.ado,glmnewey.hlp), but not yet for the WEAVE
estimators.
Back to my work page
Find out about R.