> Back to the fMRI data. Imagine a purely within-subject 3x3-ANOVA, which should be reasonable nowadays. E.g. something like "face" (happy, sad, fearful), and "sex" (male, female, morph). Maybe I have specific hypotheses, but maybe I do not (at least for some levels, e.g. concerning "morph"). In the latter case, I would run F-tests for "face", "sex" and the interaction. Imagine I get some clusters surpassing an otherwise defined voxel-size threshold. What should I do then?
(1) If you get an interaction, you should exclude those voxels from
the effects of face and sex.
(2) You can use a mask to constrain the analysis within face or sex or
the interaction, but do not use SVC. SVC adjusts for a smaller search
region.
Imagine, hypothetically, that I said any cluster with a p-value of .2
was significant for a main effect. Now I look in those clusters and
finds clusters that are p<0.05 for individual tests because I use
small-volume correction. You've biased yourself towards finding a
group difference. While the case is true that a significant F-test
means that you will have at least 1 significant pst-hoc t-test, it
does not mean that you have any more than one significant post-hoc
t-test. Thus, the reduction in search volume and using SVC can lead to
false positives.
>
> Or should I run lots of t-tests right from the beginning? I would already have to conduct 12 one-sided tests for "face" and "sex". And to ensure that the results make sense, I would have to check all the interactions as well.
>>>> Tests for face can be considered independent of sex. Correct for the number of tests in each one if you want. However, this isn't done as there is already a correction for the number of voxels, which is much higher.
|