R help - Apr 2010 - computation of dispersion parameter in quasi-poisson glm

Hi list,

can anybody point me to the trick how glm is computing the dispersion
parameter in quasi-poisson regression, eg.
glm(...,family="quasipoisson")?

Thanks &regards, Sven

On Fri, 9 Apr 2010, Sven Garbade wrote:
> Hi list,
>
> can anybody point me to the trick how glm is computing the dispersion
> parameter in quasi-poisson regression, eg.
> glm(...,family="quasipoisson")?
It's the sum of squared Pearson residuals divided by the residual degrees 
of freedom. For example:

example("glm")
fm <- glm(counts ~ outcome + treatment, family = quasipoisson)
summary(fm)
sum(residuals(fm, type = "pearson")^2)/df.residual(fm)

hth,
Z
> Thanks &regards, Sven
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

On Fri, 9 Apr 2010, Sven Garbade wrote:
> Hi list,
>
> can anybody point me to the trick how glm is computing the dispersion
> parameter in quasi-poisson regression, eg.
> glm(...,family="quasipoisson")?
It isn't.  glm() does not need (and does not compute) the dispersion 
parameter.

summary.glm will ... list it and the code is in the top few lines.
There is no 'trick': this is as described in MASS pp 186, 208 (which 
also explains some of the pitfalls).
> Thanks &regards, Sven
-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595