Thanks for the answers to my previous post,
I hope I am posting on the correct list now.
I managed so far to run the multinomial model with random effect with
the following command:
MCMCglmm(fixed=cbind(Apsy,Mygl,Crle,Crru,Miag,empty) ~
habitat:trait,random=~idh(trait):mesh,family="multinomial12",
data=dataA,rcov=~trait:units)
(where multiple responses are different species,
Habitat the explanatory
variable and Mesh the random effect)
The main question I am facing now is:
Why the multinomial model fit does not provide by default a parameter
estimate for each response category but a unique one for all of them?
I had to add interactions "trait:habitat" to get the same output as a
classical multinomial model... Therefore I wonder how the multinomial
model is implemented in this function... any idea?
The predict function is apparently not adapted for multinomial fit
either and need further improvements, right?
Best,
Amélie V
______________________________________________
R-help@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.
