[R-sig-ME] Question regarding glmer in lme4

[Ben Bolker]

Ben Bolker <bbolker at ...> writes:

> > On Fri, Nov 4, 2011 at 5:05 PM, Asa Johannesen <bsaj at ...> wrote:

> > > I'm a PhD student using a binomial glmer model and I added an
> observation level random effect to account for > overdispersion
> (seeing as I can't use a quasi-family) and now my supervisor is
> asking me if I have a > reference saying this is a good way of doing
> that. I only have the R-help file example, but that seems a bit >
> slim. Could you perhaps point me in the right direction?  > > > >

Oops, this got truncated somehow.

The second bullet point in http://glmm.wikidot.com/faq#overdispersion
lists references for this point, covering both the Poisson 
(lognormal-Poisson) and binomial (logitnormal-binomial) cases
[on a slight tangent, I wonder why there seems to be so much more
attention given to overdispersion in Poisson than in binomial cases.
Presumably it's because the frequency with which people seem to
analyze discrete data is more or less ungrouped Bernoulli >> Poisson >>
binomial with N>1 ...].

If anyone knows of any additional references, feel free to send
them to me -- or, even better, add them to the wiki yourself!