I am using Proc Genmod (logistic regression) to analyze a longitudinal data set with 540 time points. My question is simple: Should I focus my attention on the analysis results reported in the GEE table or the initial parameter estimates table?
My reading of the documentation leads me to believe that the GEE table is appropriate because of the autocorrelation problem arising from repeated measurement, and that the initial parameter estimates table does not correct for this autocorrelation. Am I correct?
