Thaks for your response. I Tried it, but It doesn't work. I got the message: ==== 604 Contrast 'time0*traj3 vs time1*traj3' NOTE: SCL source line. 605 time0*c1traj3- time*c1traj3 1 -1; ----- 22 ERROR 22-322: Syntax error, expecting one of the following: a numeric constant, a datetime constant. 606 weight meanpond; 607 run; 608 title1; ====

Llorenç Badiella <badiella@MAT.UAB.ES> wrote: > > Hi all, > > Try the Contrast option in Proc Mixed. > > Hope it helps. > > Llorenç Badiella > Servei d'Estadística UAB > > En/Na barrere Bendia ha escrit: > > > Hi SAS-L freinds there. > > I'm using proc mixed and I wante to test the difference between two > > parameters (B1 and B2) of two independents dummies variables: time0*c1traj1 > > and time*c1traj1 ( and many others: time0*c1traj(k) time*c1traj(k), > > k=1,2....10). I tried: > > estmate 'time0 vs time' time0*c1traj1 -time*c1traj1 1 1 . But It didn't work. > > I tried other thing, but I had a message like this: the variables must be in > > model. > > Please, have you any ideas? > > > > proc mixed data=un noclprint covtest METHOD=reml covtest; > > class family children; > > model prosocia= time sex agecent transit*time > > > > time0*c1traj1 time0*c1traj2 time0*c1traj3 time0*c1traj4 > > > > time*c1traj1 time*c1traj2 time*c1traj3 time*c1traj4 > > > > /************************************/ > > / solution notest ddfm=bw outp=ponde; > > random intercept time /subject=family type=un gcorr ; > > random sex_enfa /subject=family ; > > random intercept /subject=children(family ) ; > > estimate 'time0_traj1 vs time_traj1 > > time0*c1traj1-time0*c1traj1 1 -1; > > weight meanpond; > > run; > > > > I would be grateful if you could help me. > > Thanks a lot. > > ===================== >