[R] lm() variance covariance matrix of coefficients.

Peter Dalgaard p.dalgaard at biostat.ku.dk
Sat Jun 3 01:34:48 CEST 2006


Rolf Turner <rolf at erdos.math.unb.ca> writes:

> Peter Dalgaard wrote:
> 
> > Rolf Turner <rolf at erdos.math.unb.ca> writes:
> > 
> > > summary(object)$cov.unscaled
> > 
> > You need to multiply that with sigma. However, vcov(object) is easier. 
> 
> 	Well, I thought unscaled meant unscaled --- the plain
> 	unvarnished covariance matrix!  I figure that multiplying
> 	the *covariance* matrix by something would be scaling
> 	it.  Silly me.

Think (quasi-)binomial glm() and things become clearer. Unscaled
corresponds to a scale factor of 1.
 
> 	Also:
> 
> 	(a) Shouldn't that be ``multiply by sigma^2'' rather
> 	than by sigma?

Yup

> 	(b) Wouldn't it be helpful to have a pointer (``see also'')
> 	to vcov() in the help on summary.lm()?

Well, it *is* in ?lm ...

-- 
   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907



More information about the R-help mailing list