afni.nimh.nih.gov

Hi all,

In our analysis, we ran the 3dconvolve on our dataset and when we were looking at our contrast of interest in 3dttest+, we included a covariate of trait measure scores. As we decrease the p value, the cluster of interest remain, however, they do decrease in voxel size. My question is that when a covariate is used, is there any statistical reasoning that would allow us to reasonably use a lower threshold?

Thanks in advance!