[R-sig-ME] MCMCglmer categorical priors

[Jarrod Hadfield]

Hi, 

In short, yes. The observation-level variance is not identifiable for
multinomial data with a single trial (e.g. binary) or ordinal data.
Information must come from the prior or from other data (as in
zero-altered models). Because of this I usually just fix the
observation-level variance to some value, usually 1. It would be equally
valid to fix it to some other value or even have a weaker prior. 

Cheers,

Jarrod.


On Tue, 2011-06-07 at 17:30 -0400, Lucas Kid wrote:
> Mixed modelers,
> 
> Is the advice concerning the priors in
> 
> https://stat.ethz.ch/pipermail/r-sig-mixed-models/2011q2/005800.html
> 
> specifically
> 
> prior=list(R=list(V=1, fix=1), G=list(G1=list(V=1, nu=0)))
> 
> applicable to any prediction of binary data (a single vector of Ts and Fs)?
> 
> Thank you in advance!
> 
> Luke
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> 



-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.