Hi Rick,
I ran 2-way anova (3dANOVA2; n = 40) and found a main effect that is too huge for me to separate it into reasonable clusters.
Attached picture shows a main effect (F-stat) at an uncorrected p = 0.0001. When I run 3dclust on this threshold (using AlphaSim results of the minimum cluster size/volume), the whole blob turns out as one cluster connected throughout the brain. I had to raise the threshold to be at uncorrected p = 0.00000001 in order to see this huge blob separate into two clusters (one in the back and the other in the front); however, these two clusters are still huge in their respective volume size (e.g., 160,000 vmm3), encompassing a whole variety of regions (3dExtrema attests to this).
The goal is to have more distinct regional clusters so I can extract their values per experimental condition and run further analyses with these values. But right now I'm at loss with this huge cluster. I could keep increasing the threshold but that's at the risk of losing some regions that I'm interested in, and I'm not even sure if it's statistically correct to threshold the F-stat at the level of 1e-10.
Please advise what I could do.
Thank you so much,
Hyejeen