I am running a repeated measures anova using proc glm. I am not having a problem with the anova itself, but when I do a tukey test at the end, it is giving me pairwise comparisons broken down by each trial (month). What I need is adjusted pairwise comparisons for the entire experiment eg) pairwise experiments between each treatment with all trials (months) combined.
This is what I have entered that is giving me the tukey comparisons broken down my month:
The output gives me pairwise comparisons between treatments (habitat) for each month (m1 - m2 - m3 ), but I need p values for pairwise comparisons between treatments for the entire experiment (all three months).
I usually achieve this using SPSS, as it computes the pairwise comparisons for combined data. Although it only computes such comparisons in one direction only and needs some tweaks for pairwise comparisons at both horizontal and vertical dimensions.
My own problem is to force SPSS to do pairwise comparisons between all the cells one by one (in which I fail), and I'm glad there is such a solution to it.