Thanks!
The comments and the information provided were extremely helpful to me.
According to DRAFT r-sig-mixed-models FAQ, "do not compare lmer models
with the corresponding lm fits, or glmer/glm; the log-likelihoods are
not commensurate". The problem is not the link function but rather the
different additive terms.
Maybe, I made a mistake in explaining my problem
I want to compare Model A whit model B in order to evaluate the
significance of the random effects.
A<-lmer(Ymat~stu+spi+(1|ber),data=data,family="binomial")
B<-glm(Ymat~stu+spi,family="binomial",data=data)
I'm looking for an alternative way.
I made these questions to r-sig-mixed-models at r-project.org as you recommended.
Thanks again for your time!
Gabriela