For those that may have missed it, see the Neuroscience award here,
which is about what you can expect from a fishing expedition:
http://www.improbable.com/ig/winners/#ig2012
Unless you have a prior hypothesis (preferably one for which there is
documented evidence from before you looked at the results) about where
you expected to see differences, uncorrected results are little more
than noise.
Best regards,
-John
On 23 September 2012 20:09, Christophe Phillips <[log in to unmask]> wrote:
>
> Dear Gabor,
>
> your question seems to reappear every now and then...
> And the answer is typically this one:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1206&L=spm&P=R2293&1=spm&9=A&J=on&d=No+Match%3BMatch%3BMatches&z=4
> In short, no correction means that you have no idea if your result is
> genuine or bogus... unless you knew which voxel you were interested in.
>
> HTH,
> Chris
>
> ________________________________
> De: "Gabor Oederland" <[log in to unmask]>
> À: [log in to unmask]
> Envoyé: Samedi 22 Septembre 2012 15:32:10
> Objet: [SPM] Threshold .001 (uncorr) = fishing expedition?
>
>
> Dear SPM experts,
>
>
> in a recent thread
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;d9b3a0b5.1206 there was
> some discussion about an uncorrected voxel-threshold of .001.
>
> Now I wonder: Is .001 inacceptable in general or is this meant only when no
> cluster-threshold is used? That is, would it still be ok to report results
> of a whole-brain analysis with e.g. a voxel-threshold .001 uncorr and a
> FWE-corrected cluster-threshold of .05?
>
>
> Thanks in advance,
>
> Gabor
>
|