That amount of overdispersion would make the use of a poisson model
very questionable, and will very likely result in estimated standard
errors that are too low, hence the change in statistical significance
when you switch to quasipoisson.
On 06/03/07, Cristina Gomes <gomes at eva.mpg.de>
wrote:> Hi all,
> I was wondering if somebody could give me advice regarding the
> dispersion parameter in GLMM's. I'm a beginner in R and basically
in
> GLMM's. I've ran a GLMM with a poisson family and got really nice
> results that conform with theory, as well with results that I've
> obtained previously with other analysis and that others have obtained in
> similar studies. But the dispersion parameter is too large (25 compared
> to 1). So I ran a quasipoisson that allows for overdispersion and all of
> the significant results fell out. My question is, how much does
> overdispersion invalidate ones results when using a poisson family? What
> could the reason be for having non significant results with a
> quasipoisson distribution?
> If anybody could give me some help with this or/and recommend me
> literature that is somewhat comprehensible to lay people (i.e. non
> statisticians) that would be great.
> Thanks in advance,
> Cristina.
>
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>
--
================================David Barron
Said Business School
University of Oxford
Park End Street
Oxford OX1 1HP