Hello,

I’m performing a repeated measures experiment with 3 factors, each of them with 2 levels. Each subject participated of 8 experimental sessions (all possible combinations of the three factors) in a random order. I was wondering what would be the proper way to analyze this data using the mixed proc. Should I use the SUBJECT option with the RANDOM statement (program 1) or the REPEATED statement (program 2)? Any suggestions on which covariance matrices should I test?

Thanks in advance,

Andre C. Ph.D. Student Roberto Vieira Laboratory of Sensory Physiology University of São Paulo www.fisio.icb.usp.br/~vinicius

PROGRAM 1: proc mixed data=DATA METHOD=REML CL ALPHA=.05 COVTEST; class SUBJ FACTOR1 FACTOR2 FACTOR3; model PSS = FACTOR1 FACTOR3 FACTOR2 FACTOR1*FACTOR3 FACTOR1*FACTOR2 FACTOR3*FACTOR2 FACTOR1*FACTOR3*FACTOR2 / HTYPE=3 DDFM=SATTERTH; random SUBJ / subject=SUBJ; lsmeans FACTOR1*FACTOR3*FACTOR2 / PDIFF adjust=TUKEY CL ALPHA=. 05; run;

PROGRAM 2 proc mixed data=data; class SUBJ FACTOR1 FACTOR2 FACTOR3; model PSS = FACTOR1 FACTOR3 FACTOR2 FACTOR1*FACTOR3 FACTOR1*FACTOR2 FACTOR3*FACTOR2 FACTOR1*FACTOR3*FACTOR2 / ddfm=satterth htype=3; repeated / sub=SUBJ type=un; lsmeans FACTOR1*FACTOR3*FACTOR2 / pdiff adjust=TUKEY; run;