> why do you recommend a one-sample t-test instead of a two-sample t-test?
This should indeed read two-sample t-test (when contrasting groups) and one-sample t-test (in case the two treatments are contrasted on single-subject level) or paired t-test (in case they are not) for tests within groups.
> Is there a way to code a contrast for J-Before vs O-Before? and also J-After vs O-After?
Yes, set up a "two-sample t-test" each.
> Is it statistically correct to use the cluster derived from the main effect of group in order to test the contrast J-After vs O-After and J-Before vs O-before in separate two sample t tests?
You have decided to run an ANOVA with a certain threshold and you obtained a sig. main effect group but no sig. interaction and no sig. treatment effect. I can't see how you would justify to take clusters with a main effect group to look at group differences seperated for pre and post / treatment 1 and treatment 2.
If you had a-priori hypotheses you could of course conduct planned comparisons right from the beginning (on whole-brain level or for a-priori regions), e.g. group 1 treatment 1 vs. group 2 treatment 1 and so on, but then there would be no need to test for main effects at all.
However, if you detected e.g. a group difference for group 1 treatment 1 vs. group 2 treatment 1 and no group difference for group 1 treatment 2 vs. group 2 treatment 2 you would still have to contrast these two contrasts to be able to conclude there's a sig. interaction.
Best
Helmut
|