Split-plot repeated measures with proc mixed

[palolix]

Hi!

I want to analyse my data, which correspond to a split-plot design as repeated measures (date), but I am not sure if my model is correct.  The samples were only taken in two dates; 18.04.12 and 09.06.12.

proc mixed data=one;

class blk treatment cultivar date;

model y=

treatment|cultivar|date/ddfm=kenwardroger;

random blk blk*treatment;

repeated date;

lsmeans treatment|cultivar|date/adjust=simulate cl;

run;

Thanks a lot!

Caroline

[SteveDenham]

That will work.

Are you sure about making all of the comparisons in the lsmeans statement?  Will it be meaningful to compare, say Treatment1/Cultivar A in April to Treatment3/Cultivar B in June?  Think about what the overall tests tell you, and then set up the lsmeans (or Lsmestimates) to deal only with the comparisons of interest, before applying a multiple comparison adjustment.

Steve Denham