Hi Bob,
Thanks for the reply. I think after reading another post of yours that the issue may pertain to smoothness estimates of covariates. With the current state of where things are at, do you recommend 3dttest++ -clustsim or using 3dFWHMX -acf on the group level stats (i.e. Deter ended stats Output of 3dMVM if we need it for interactions or residuals from 3dttest++ if that model suffices) along with 3dClustsim for publishing? My concern is that with two groups of 20 and adding age, gender and Moca as covariates that I am losing a lot of power, especially if I'm in the category of FPR of 1-2% of covariates (from your Clustsim post).
Additionally, I have some surface data (i.e. Cortical thickness etc) which I processed through 3dMVM due to the interactions of the model which are more accurate.
1) Is there any stats tool which directly samples surface data (from my understanding 3dttest++ -clustsim does not work with surfaces just yet)?
2) If not, is there any surface based smoothness estimation programs through AFNI that use the acf model such as slow_clustsim, or have these not been updated yet from the pure Gaussian model?
Thanks,
Ajay