03-04-2013 06:33 PM
I want to adjust p values (by Bonferroni) for six estimates/contrasts in proc mixed. Is there anyway to do that? It appears that the p values can only be adjusted in lsmeans.
03-05-2013 09:15 AM
I don't know what version of SAS you are using, but you should examine the LSMESTIMATE statement as an alternative. If you are on SAS 9.2, you may need to change to PROC GLIMMIX to have this available. The "More Like This" panel to the right will lead to other discussions that deal with this specifically.
03-05-2013 09:22 AM
It can be easily done with GLIMMIX. This procedure has many more options than found in MIXED. For instance:
estimate '1 vs 2' trt 1 -1 0 0,
'2 vs 3+4' trt 0 1 -.5 -.5,
'3 vs 4' trt 0 0 1 -1 / adjust=bon;
That is, you just separate each estimate with a comma, and then add an adjust option. I point out that Bonferroni is very conservative, especially with a large number of contrasts. You can't do this directly in MIXED.