[R] influence.measures for glm

John Fox jfox at mcmaster.ca
Thu May 2 22:35:37 CEST 2002

Dear Thomas and Robert,

At 09:21 AM 5/2/2002 -0700, Thomas Lumley wrote:
>On Thu, 2 May 2002, Robert Espesser wrote:
> > From the mail archives, it is not clear for me in what extend
> > influence.measures is  correct for
> > a glm model  ( family=binomial, for my problem).  I'm aware of the
> > existence of
> > glm.diag and glm.diag.plot, but is there somewhere a function like
> > influence.measures for a glm model , from which it is very easy to get a
> > list
> > of the influential observations.
> >
>I think all the influence measures for lm also work for glm. Certainly
>lm.influence(), which underlies them, does.  Most of these measures are
>defined for glms in terms of the weighted linear model fit by the last
>iteration of IWLS.

If I remember correctly, some common influence statistics (and other 
casewise diagnostics), such as Cook's distances and studentized residuals, 
need a little tweaking to get the usual generalizations to GLMs. Again, if 
I remember right, most of the information needed is available from 


John Fox
Department of Sociology
McMaster University
Hamilton, Ontario, Canada L8S 4M4
email: jfox at mcmaster.ca
phone: 905-525-9140x23604
web: www.socsci.mcmaster.ca/jfox

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