You are right, Gang. I was looking at the ANOVA with a fifth condition. The one with 4 conditions has 48 degrees of freedom for the t tests, as you say. I tried running the paired t tests, and the results are very similar (same regions show up), just the significance levels are not as good, obviously. Thanks for your input.
-MikeGang Chen wrote:
> Mike,
>
> For the ANOVA, did you run your analysis with 3dANOVA or
> 3dANOVA2? Sounds like you were using the former, but 3dANOVA2
> would be more appropriate, in which case the
t tests of
> those contrasts should have 48 degrees of freedom, not 64.
>
> You could also run 3dttest with its -paired option to get those
> contrast tests of interest. The difference with those of
> 3dANOVA2 should be small unless there are some irregularities
> in some of the conditions.
>
> Gang