Correct p-values multiple testing mixed models


New Member
I am trying to figure out how to correct for multiple testing when having several mixed models.
I am using R, nlme package to make the models.
I have run 14 mixed models for 14 pre-defined continuous output variables. 1 model is the primary outcome and 13 models are the secondary outcomes.
It should be noted that these outcome variables are somewhat related (heart function measurements) so the models are not completely unrelated.

Each person have been tested 4 times (t0,t1,t2,t3) and time is therefore used as fixed effect.
So I have tested t0 vs. t1, t0 vs t2, t0 vs. t3 in each model. I have no other fixed effects.
I have run F-statistics for each model and if the overall effect of time was not significant I have not used the individual p-values between time points.
Because of that I end up only using 9 of the full models (including the one with primary outcome).

My question is: ow do I adjust for the multiple comparisons: 3 time tests for each fully used model x 9 models + 14 F-statistics for overall effect = 41 tests (for example for Holm-Bonferroni correction)
or how would you do it?
Last edited: