Hi all, I'm trying to run a proc glm using 3 classes (Phylum, Island Neighborhood). However, neighborhood has a large # of levels while island and phylum have relatively few:

Class Levels Phylum 2 Island 6 Neighborhood 28

Although I can analyze the main effects of phylum, island, and neighborhood separately, when I try to use them all in the same model I run out of degrees of freedom for both phylum and neighborhood. Is there any way to get around this, or should I just be content with looking at each one separately? Proc glms using just island and phylum, or island and neighborhood work okay (although I run into the same problem when testing for an interaction effect for island and neighborhood).

Many thanks for any help you can provide,

~Heather Norton hln109@psu.edu