I have conducted initial analyses and planned contrasts on my datasets using the Proc Mixed repeated measures procedure for 2 reasons: 1) there are both between and within effects in my data, 2) the datasets are uneven and when using the repeated function in GLM missing values are not considered.
Having said that, I was asked by a reviewer to run post hoc analyses with a Bonferroni adjustment. In reading the stats manual, I believe it is possible to perform this using Proc Mixed but I wanted to clarify that this is the case.
Additionally, it appears that IF I use Proc Mixed I have to run LSMEANS with a Bonferroni adjustment..... can anyone out there clarify this for me before I run the stats and interpret them (possibly incorrectly)? Thank you so much for your help.
I have attached a copy of my code if that is helpful.
Your code seems fine & lsmeans is the only option for running Bonferroni correction with proc Mixed. With GLM you would have had both the Means & Lsmeans options but not with proc Mixed.
Thank you for your prompt reply. I appreciate your taking the time to answer my question. One more question, do you think that GLM would be a better analysis??
If your model consists of random effects then you should use only mixed/glimmix & possibly genmod. If it is a fixed effects model then GLM is a good choice.
Reply With Quote
Post Hoc analyses using PROC MIXED
Posting Permissions