Dear Doug and Bruce
I have a couple of questions regarding multiple comparison correction.
Any help with this would be really appreciated!
I am conducting a cortical thickness analysis between 2 groups (patients vs controls).
My understanding is that first I should check for an age-cortical thickness interaction using DODS. If no interaction is found I can use DOSS.
My problem is that when I run the analysis for age-cortical thickness interaction and correct for multiple comparisons using FDR there is no significant age-cortical thickness interaction for left or right hem. So DOSS can be used to check for group differences in cortical thickness.
When I run the same analysis, this time using clusterwise correction, I find an age-cortical thickness interaction resulting in 1 cluster in the right hem. However, this cortical region is not present in the DOSS age independent cortical thickness difference between the patients and controls.
I am aware that FDR and clusterwise correction are doing 2 very different things but if FDR finds no difference between groups for age-cortical thickness interaction, would it be justified to stick with FDR if a difference was present using clusterwise correction?
Having looked into this, it appears to me that both FDR and clusterwise correction are equally valid, just different approaches.
I am really stuck on this point, as my understanding of this area is somewhat limited.
Thank you so much in advance
Best wishes
Marco