[R-sig-ME] PQL and crossed designs

[Sophia Kyriakou]

Dear List,

I need to fit a generalized linear model with crossed random effects using
penalized quasi-likelihood. I realize that this has troubled many people so
far, and I was wondering if anyone has figured out how to use MASS::glmmPQL
in order to fit a model e.g. on the salamander mating data.

Following
http://rstudio-pubs-static.s3.amazonaws.com/38992_0a51aebca767491690f2b97e05c5aa05.html

I tried

library(hglm)
data(salamander)

salamander$dummy <- factor(rep(1, nrow(salamander)))
salamanderG <- groupedData(Mate ~ 1 | dummy, salamander)
glmmPQL(Mate ~ 0 + TypeF:TypeM, data = salamanderG, family = binomial,
random = pdBlocked(list(pdIdent(~Male-1), pdIdent(~Female- 1))))

but as expected I get the error

Error in getGroups.data.frame(dataMix, groups) : invalid formula for groups

Any help is much appreciated.

Thanks,
Sophia

	[[alternative HTML version deleted]]