6.6 Generalized and Mixed?
So far in this chapter we have dealt with with two extensions to the General Linear Model: the Generalized Linear Model for non-continuous outcome variables, and the Linear Mixed Model when there is clustering in the data. You might be asking the question: what if I have both a non-continuous outcome variable, and clustering.
That’s no problem: you need to use both extensions at the same time, making a Generalized Linear Mixed Model. You fit a Generalized Linear Mixed Model with lmerTest
exactly as we have done so far this session. Instead of lmer()
, the function you need is glmer()
, and the additional thing that you need to put in the function call is the specification of a family of models (e.g. binomial
, Poisson
), exactly as we did with function glm()
earlier in the chapter. So the call will look something like: s1 <-glmer(y ~ x + (1|participant), family=binomial, data=data)
.