I am searching for the best covariance structure between levels of a fixed effect (called "year" in the below statements). Each level of the fixed effect has several observations and, then, I cannot model the R matrix with the repeated statement.
A general overview of the program statements is:
proc mixed covtest data=datos;
class plot treatment tree year orientation;
model p= plot treatment year plot*treatment plot*orientation plot*year treatment*year/ddfm=satterth;
random tree (plot)/ subject=plot type=sp(exp) (x y);
I would apreciate very much any suggestion.
Thank you in advance