> obviously something is not quite right--got any inkling of what that might be?
I'm baffled by this too, and one thing I've noticed is that, most of the voxels statistically significant at group level are not really significant at individual level. For example, the contrast between the two condition at voxel of (i, j, k) = (65, 47, 36) is significant with both t and Z at group level, but the t-statistic for the contrast is NOT significant for most subjects.
> in general the data look quite good in the sense of each condition vs. rest
> looking very much like it does in similar published studies
Group or individual level?
Gang