[R] glmer and glht

[Sibylle Stöckli]

Dear R colleagues,

The effect of habitat type on bee species number was tested using a glmer model with habitat as fixed effect (5 habitat types) and site and time as random terms.
The idea was now to conduct post hoc tests with glht (package "multcomp") for pairwise comparisons. The matrix was defined by symbolic description (linfct=c("habitat1=0", "habitat2=0"…). 
My question: I am now interested about the detailed statistic behind glht? As I was not applying ANOVA and mcp=Tukey, it is for sure not a Tukey test, but is it a LSD, Cuncan, Hochberg….and is the test suitable for an unbalanced data set (not every habitat occurs at every site?)

With best wishes
Sibylle






	[[alternative HTML version deleted]]