Arshad,
> I was wondering if you could help explain
> a RFX puzzle that I have come across.
>
> I have two groups of (15) subjects performing
> a two condition box-car task (ie. task and rest),
> when I perform within group (spm) RFX analysis (task vs rest)
> I get significant effects at 0.05 corrected for each group.
OK. (Though, as you imply below, I assume that you get results in
different regions.)
> However, when I perform group comparisons for the same task
> I get very little effects surviving. I am puzzled as the each group
> RFX is different yet when I perform a between group comparison very
> little is revealed? Why does that happen in this context?
Just because you see a lack of significance in one area doesn't mean that
area isn't activated. You have found different *thresholded* results
in each group, but that doesn't mean there is evidence of a difference
between the groups (see the multi-month SPM thread by Brett et al. on
the limitations of thresholding). By doing the correct analysis (direct
statistical subtraction) you have found there is a lack of evidence for
a difference, suggesting that there may be sub-threshold activations
in on group, or that there is simply insufficient sensitivity to detect
a difference.
Hope this helps.
-Tom
-- Thomas Nichols -------------------- Department of Biostatistics
http://www.sph.umich.edu/~nichols University of Michigan
[log in to unmask] 1420 Washington Heights
-------------------------------------- Ann Arbor, MI 48109-2029
|