02-20-2015 09:39 AM
I am looking to compare the variance of lab values across categories of individuals. There are three categories, and the lab values are a continuous variable. Each patient has multiple lab values (repeated measurements, order of measurements does not matter, different number of measurements per patient). A colleague suggested using a mixed effects model, such as this:
PROC GLIMMIX DATA=data;
MODEL lab =;
RANDOM _residual_ / subject=subject type=cs;
RANDOM intercept / subject=category type=un(1);
Essentially, applying an R-Side random effect at the individual level, and a G-Side random effect at the category level. I cannot find which output options will give me the variances for each category and the associated standard errors or test statistics. I am also wondering if there is another way to do this without such a complicated model that will be less smidifficult to explain to my co-investigators.
Any advice would be much appreciated.