I am comparing different methods of estimating water use from irrigated land. I have 2 factors: Irrigation Type (a,b,c,d--these are irrigation polygons representing the various methods selected randomly in a GIS--however they are intact groups and were not assigned "treatments" at random) and METHOD (1,2,3) and the response variable (RESP--water use). I initially used PROC GLM in SAS for the analysis. Since I am comparing the same method for each polygon is it possible that the errors are correlated and I should be using PROC MIXED with a repeated measures statement ? Please advise. Thank you.