Hi there,
my 2 rather subjective cents on this:
> It seems that a majority of papers publishing ROI analyses does not
> apply correction for multiple comparisons.
My gut feeling is that it used to be unacceptable a while ago, then one
needed to justify it thoroughly, and now it has become, as you say,
pretty much commonplace. Not sure I like this as...
> The rationale behind it, I suppose, being that ROI analyses correspond
> to planned comparisons and therefore are exempt from usual corrections
> applied to post-hoc tests.
... I have a lurking suspicion that maybe, jut maybe, some of the
"prior" hypotheses were generated following inspecting the results. It
is also worthy to remember that in some disorders (say, schizophrenia)
literally every brain region has been described or suspected to be
involved in the pathophysiology of the disorder, allowing to generate
prior hypotheses for every part of the brain. This somehow defies the
purpose of meaningfully restricting the search volume, in my eyes.
Don't get me wrong: I fully agree that the whole-brain correction for
multiple comparisons tends to be strict, and maybe too strict in many
cases, and restricting the search volume is therefore not a bad idea as
such (have done it myself). However, I believe whole-brain analyses
should still be our aim, and if our methods (in statistics, sequence
programming, data processing...) are not sensitive enough (or our groups
not large enough) to find significant activation then maybe there is
none - or we need to improve our methods (or increase power). To instead
effectively lower the threshold is, in my humble opinion, a slippery
slope. But that may be just me :)
Best,
Marko
--
=====================================================================
Marko Wilke (Dr.med./M.D.)
[log in to unmask]
Universitäts-Kinderklinik University Children's Hospital
Abt. III (Neuropädiatrie) Dept. III (Pediatric neurology)
Hoppe-Seyler-Str. 1, D - 72076 Tübingen
Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473
=====================================================================
|