Hello, I have a general question concerning GLMM. I would like to model Y as a function of X and Z, with a random effect of site because there are multiple measurements of Y at each site. However, Z is a characteristic of the site, not the individual sampling point, and there are not replicate sites with the same value of Z. Therefore it seems that Z is confounded with the site. If I add a random effect of site, how is the random effect of site partitioned from the fixed effect of Z?
I've seen examples that do just this (published in GLMM modeling books), which makes me think that I should be able to do it, too, but I would like to know, conceptually, how the model can partition effect between the random-effect of site ID and the fixed site characteristic.