Thank you Ziad and Rick!
I think I understand why the standard deviation from independent t test was different from 3dttest++ now.
When we run the two sample t test, the beta value in that ROI is a single number with a standard deviation of zero.
When we run 3dttest++, it is voxel-wise test, the beta value in that ROI is a mean of lots of voxels with a standard deviation not equals zero.
That’s why the mean is always the same, but the standard deviation is not the same.
I will try to run it with a single voxel to see if it is the same.
Thank you again Ziad and Rick!
Kai