Dear David,
> According to Jonathan: "you should be fine to use a different threshold for the initial results—it's the statistics that are output from the SVC button that are the relevant ones." Do it means "100% necessary to use FWE .05 and then use small volume correction" mentioned by Kailyn? it if is, I wanted to ask a further question about such.
There are two issues involved here. The first is how to define the
volume used in a small volume correction; the second is what
statistical threshold to use within that volume.
There are a number of ways to define the volume, including: a sphere
around a peak MNI coordinate from a previous study, an anatomical
region of interest, or a cluster from an independent contrast. The
critical thing is that the way you define your small volume must be
independent of the data you are testing. Let's say that you are
interested in the angular gyrus in your current study. In data from a
previous study, you have a cluster in the angular gyrus, but only at p
< .05 uncorrected. Although that wouldn't have really been strong
evidence in the previous study (because it's not corrected), it's
still fine to use this as a region of interest or small volume—you now
have a strong a priori hypothesis that this region may be involved.
The point is that it is not necessary to use any particular threshold
to define a small volume, only that you have a good motivation for
choosing it, and that it is independent of the contrast you are
evaluating within the small volume.
> Suppose we perform "result" procedure of SPM and chose uncorrected p value initially.
> Then we do some SVC at some peak coordinates, we could get some result table.
> Within the table, I wondered about the actual meaning of the column labelled as P FWE-corr over peak level since we still had another column labelled as P uncor there?
Once you perform the SVC, the corrected p values reflect the corrected
p value within whatever volume you specified. The uncorrected p
values are the same as the whole brain, and just reflect the result of
a single t or F test (not corrected). So, when doing an SVC, you have
the same choices regarding multiple comparison correction as you do at
the whole brain level—just corrected within the volume you've
specified (rather than across the whole brain).
Put another way, regardless of how you define your small volume, it's
still necessary to control for false positives within that volume.
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|
