Hello again Steve, When analyzing some dependent variables, where I only need 2 levels of the factor 'cultivar' (which has 3 levels; Nordstern, Sang and Beretta), so the number of observation that sas use are 27 out of 40, I only get the LSM but not the estimates. The problem is that in my Type 3 Test of Fixed Effects Table I only get the 'nem' and 'cultivar' main effects but not the interaction nem*cultivar (which I suppose is not estimable), so I think that is why I am not getting the estimates. When writing the estimates statements I think I should only write he main factors but not the interaction: lsmestimate nem cultivar 'label' [1, 1 2][-1, 2 2], instead of nem*cultivar 'label', but if I just write nem cultivar 'label', sas tells me that is expecting a * (nem*cultivar), so it does not work either. Here is my code: proc mixed data=one; where cultivar="Nordstern" or cultivar="Sang"; class blk nem cultivar; model ahreave= nem cultivar nem*cultivar/ddfm=kr; random blk blk*nem; lsmeans nem*cultivar; lsmestimate nem*cultivar 'nem1 vs nem2 | cultivar=Norstern' [1, 1 2][-1, 2 2], 'nem1 vs nem3 | cultivar=Sang' [1, 1 3][-1, 3 3], 'cultivarNord vs cultivarSang | nem=1' [1, 1 2][-1, 1 3]; run; Is there a way to write the lsmestimate without the interaction nem*cultivar (since this is not estimable)? Do you think this is the reason why I am not getting the estimates results? I would greatly appreciate if you could give me some light. Thank you very much Steve! Caroline
... View more
