[R] Complements about heteroscedasticity

Thomas Lumley thomas at biostat.washington.edu
Sat Dec 9 19:01:36 CET 2000

On Sat, 9 Dec 2000 E97249 at edu.essec.fr wrote:

> Hi all,
> A couple of days ago, I posted a message regarding the support for
> heteroscedasticity testing in R. In fact, there are several tests and
> methods to assess whether the residuals of a model are heteroscedastic
> or not. The one for which I was claiming information can be found in
> William Greene's "Econometric Analysis", chapter 12. It's about
> White's *estimator*, not test, and it consists in replacing the beta's
> variance's traditionnal estimator by a new one calculated with the
> least squares residuals (instead of sigma square).
> From there, a new regression is performed that supposedly corrects the
> coefficents.

Code for a collection of these "sandwich" variance estimators for
generalised linear models can be found at
The functions are intended for various sorts of correlated response data,
but independent data can be analysed as a special case. I don't have code
for the re-estimation, though.


Thomas Lumley
Assistant Professor, Biostatistics
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