[R-sig-ME] GLMM with binomial error and individual-level, random term

[Ben Bolker]

Highland Statistics Ltd <highstat at ...> writes:

> > Message: 3
> > Date: Sat, 26 Jan 2013 19:02:21 +0100 (CET)
> > From: v_coudrain at ...
> > Subject: [R-sig-ME] GLMM with binomial error and individual-level
> > 	random term

> > I performed a GLMM with binomial error and individual-level random
> term to account for overdispersion. I If I understood it correctly
> on http://glmm.wikidot.com/faq, > denominator df are not defined for
> such models and the significance of the parameters should be tested
> using Chi-square tests. Is this correct? In F-test, results are >
> generally reported by giving the numerator and denominator df, the F
> value and the p value. Hiw should I report the results of my model?
> Additionally I would like to > ask if somebody has relevant
> literature associated to the addition of an individual-level randorm
> term to account for overdispersion.

> Have a look at a paper from Dave Elston:

 [snip]  [also referenced from the http://glmm.wikidot.com/faq
page, along with other refs, as previously described]

> It is also in Chapter 2 in Zuur et al. (2012)...sorry for self-citing. I 
> think I would try a beta binomial GLMM in JAGS. I believe Ben has 
> written a package for beta binomial GLM. Not sure whether it can do 
> GLMM. I think gamlss can also do beta binomial GLMM...not sure. 

  There are a number of packages that can do beta-binomial GLM --
bbmle makes it fairly straightforward, although it's not specifically
for beta-binomial GLMs.  Probably also the VGAM and aod packages.

  glmmADMB doesn't do beta-binomial GLMMs, but could fairly
easily be extended to do so.  I'd be happy to accept high-quality
patches ... or a compelling reason to spend my time on it right
now ...

> We are 
> co-authoring a book with Joe Hilbe in which we have a 40 page chapter 
> where we compare binomial GLMM with the individual level random effect 
> and the beta binomial GLMM. ....... 'Beginner's Guide to GLM and GLMM 
> using R and JAGS'. Comes out in 2-3 months....sorry for 
> self-advertising...

  Seems reasonable if it answers the question.