06-20-2016 05:58 PM
It seems that with the option / SOLUTION in the RANDOM statement of PROC MIXED, I can create a table of estimated random coefficients, which I need. I have two questions:
1. How do I, if I can, save them as a dataset?
2. What do the results represent? Are they the estimated coefficients themselves or are they deviations from the mean (the fixed coefficient)? If they are deviations, how do I get the actual coefficient? Is it, for example, FIXED COEFFICIENT - DEVIATION, FIXED COEFFICIENT + DEVIATION, ...?
06-28-2016 01:50 PM
A little quick digging through the reference for the RANDOM statement:
1. ods output solutionr=solutionr; will put the values into a dataset
2. The results are EBLUPS for the random effects. A good way to think of them is as residuals of the fixed effects (y minus X * beta hat) times (G hat *Z prime *V hat-inverse), so they are scaled by a quadratic form.
For a quick review, look at Mixed Models Theory in the Details section of the MIXED documentation.