I implemented the new '-sdn1' option in 3dttest. When used with -base11, -sdn1 sets the standard deviation and data count for the missing set1. For example:

3dttest -base1 1.2 -sdn1 0.3 20 -set2 .....

This does a 2-sample pooled-variance t-test with the mythical set1 having a mean of 1.2 in each "voxel", and the standard deviation computed as 0.3 from 20 values (the 0.3 is not the standard error of the mean value 1.2, but is the standard deviation of the values that went into 1.2 -- hope that distinction is clear). If -sdn1 is NOT used, then the usual 1-sample t-test is performed, as before.

Let me know if this works for you, or if you find any problems with it. As always, AFNI users are "beta testers for life".

It is illegal to use -sdn1 with -unpooled or -paired. Don't even think of it, or terrible things might happen.

The 3dDeconvolve upgrade is still "under advisement".