Thanks Geng,

What does this imply for "good practice" in presenting groupwise statistical maps? If I show/use a common thresholding for both maps-- both for visual display, and for consideration of significance, I don't want to erroneously imply lack of significant activation in the group that has a slightly lower n, even though on average, every subject activates similarly on an individual basis.

Are you saying that since -mean in 3danova uses pooled variance, that displaying the t-test subbrick of the group-wise 3dAnova mean for each group separately will not penalize the group with the lower n?

The essense of my question is, that in any textbook table of critical values for T and F statistics, the significance level (p value goes down) of a given T or F statistic increasesas the n increases. How might I work around this in AFNI?

Jim