Hi AFNI Community,
I'm running into a funny problem with ClustSim. As I understand, things are changing with the way AFNI runs ClustSim.
In the past, our lab has used 3dFWHMx and 3dclustSim to calculate the cluster extent threshold. I was recently told to add the -ACF option, to yield a more accurate value.
However, I also came across the new -clustsim option in 3dttest++, which seems like it will save a lot of time and scripts. I tried this option out, and noticed that it's giving me much larger cluster extent thresholds than I get with 3dFWHMx and 3dclustSim (with the -ACF option). For example, at the whole brain level, the -clustsim option in 3dttest++ tells me my clusters need to be larger than 51 voxels to reach significance at p < .05. But when I run the same sample through 3dFWHMx and 3dclustSim, I'm told my clusters need to contain 19 voxels to reach significance at the same level.
Can anyone explain why this might be? Should I trust one method of deriving the cluster extent threshold over the other?
Thanks for the help.
Phil
