Dear all, I am working with an unbalanced split-split plot design (date as a split-block factor) with 4 factors; "nem" =3 levels, "trt"=2 levels, "cultivar"=3 levels and date=4 levels. The following missing treatment combination for the nem*cultivar interaction are present ("."): nem cultivar 1 2 3 Nord x x . Sang x . x Beretta . x x Everything looks OK when getting the results of my estimates, nevertheless I get just a few significant p-values for most of them, and I know from my data that I should expect more significant results. Could it be that the problem is due to the split-plot model, where the different factors do not share the same error term as in a CRD or RCBD? Should I determine an appropiate error term for any particular estimate? Are maybe the many missing observations that I have yielding to non-valid results? In my estimates I want to compare the first date vs the last date (I have 4 dates in total, but for my first dependent variable all observations are missing for date 2, so that is why I am replacing date 4 with date 3 when comparing the first and the last date in the nonpositional syntax. In the positional syntax I omit date 2 when calculating the coefficients). Both syntax yield to the same results for all estimates, so this should be right. Here is my code, with only 2 of the estimates; Proc mixed data=one; class blk nem trt cultivar date; model lgsprCysts100= nem trt trt*nem cultivar cultivar*nem cultivar*trt nem*trt*cultivar date nem*date trt*date nem*trt*date cultivar*date nem*cultivar*date trt*cultivar*date nem*trt*cultivar*date; random blk*nem blk*nem*trt blk*nem*trt*cultivar blk*date blk*nem*date blk*trt*date blk*nem*trt*date blk*cultivar*date blk*nem*cultivar*date blk*trt*cultivar*date; lsmestimate nem*trt*cultivar*date 'Positional dateHafer2010 vs. dateSG2012 when nem=2 trt=1 & cultivar=Beretta' 0 0 0 0 0 0 0 0 0 0 0 0 1 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0, 'Nonpositional NTCD2111 - NTCD2114' [1, 2 1 1 1][-1, 2 1 1 3], 'Positional dateHafer2010 vs. dateSG2012 when nem=2 trt=2 & cultivar=Beretta' 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0, 'Nonpositional NTCD2211 - NTCD2214' [1, 2 2 1 1][-1, 2 2 1 3], 'Positional dateHafer2010 vs. dateSG2012 when nem=2 trt=1 & cultivar=Nord' 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 -1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0, 'Nonpositional NTCD2121 - NTCD2124' [1, 2 1 2 1][-1, 2 1 2 3], 'Positional dateHafer2010 vs. dateSG2012 when nem=2 trt=2 & cultivar=Nord' 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 -1 0 0 0 0 0 0 0 0 0 0 0 0, 'Nonpositional NTCD2221 - NTCD2224' [1, 2 2 2 1][-1, 2 2 2 3], run; I would greatly appreciate your help!!!! Thank you very much!! Caroline
... View more
