Statistical Savants,
I am unsure about the following multiple comparisons question:
Let's say I've acquired whole-brain fMRI data on patients and
controls. I have behavioral variables on patients only, and run
correlations of five behavioral variables with global measures
extracted from each patient's whole-brain statistical results, such as
peak activation in the entire brain, or volume of activation in some
specified area etc.
In a separate ROI analysis (not a voxel analysis) on the same fmri
data I compare patients and controls for their activation in 5
previously chosen anatomic ROIs.
Should my multiple comparisons correction extend to all 10 p-values of
the five correlations and the five t-tests? I tend to think it should.
Can you mention a reference that will help clarify this situation?
Many Thanks in advance!
Jim
James N. Lee, Ph.D. (Biomed. Eng.), M.S. (Psychology)
Imaging and Neurosciences Center
729 Arapeen Drive
Salt Lake City, Utah 84108
801 581 4228
|