Dear all
I am fitting a generalised linear mixed model for the independent variable
presence/absence of a specie in some experimental plots. As I have repeated
measures of each plot during each season of several years and as several
plots can be grouped in bigger spatial units, I had thought my model should
be:
m1<-lmer(presence~dependent
variables+(1|year/season/zone/plot),family=binomial,verbose=TRUE)
Am I right until here? My problem is that some of the dependent variables
vary from one sampling occasion to the next one, but some other (plot
descriptors such as distances to several features or plot surface) are
constant all the time. Is it correct to include all these variables in the
same analysis? If not, can someone give me an alternative?
In case it would be correct, how can I correlate a constant variable for
each plot with some other variables that are not constant in each plot to
check for collinearity between independent variables?
Any help would be welcome
Jesus Ayanz
Phd Candidate
[[alternative HTML version deleted]]