R help - Aug 2003 - leverage

Hi
Can anyone help with the technique of obtaining leverages from a 
conditional logistic regression model?  The code lm.influence does not seem 
to work for this data.
Thanks
Jane Murray

On Wed, 6 Aug 2003, jane murray wrote:
> Hi
> Can anyone help with the technique of obtaining leverages from a
> conditional logistic regression model?  The code lm.influence does not seem
> to work for this data.
If you use method="approximate" you can get delta-betas
(`influence'
rather than `leverage') with resid(model, "dfbeta")

For the exact conditional likelihood there isn't really a shortcut, but
if N is the number of observations:
  sapply(1:N, function(i) coef(update(model,subset=-i)))
is perfectly feasible.

	-thomas

>>>>> "jane" == jane murray <jmurray at
liverpool.ac.uk>
>>>>>     on Wed, 06 Aug 2003 09:16:48 +0100 writes:
    jane> Hi Can anyone help with the technique of obtaining
    jane> leverages from a conditional logistic regression
    jane> model?  The code lm.influence does not seem to work
    jane> for this data.  Thanks Jane Murray

But the generic  influence()  function {new in 1.7.0, from original
code of John Fox} has an S3 method, influence.glm(), that should work.
If not, that might indicate a bug we'd be interested to hear about.

Regards,

Martin Maechler <maechler at stat.math.ethz.ch>
http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><