Dear Mario,
> 1) Out of a group of several subjects (n = 17): what would you
> recommend to do with subjects (n = 4) who succeed on a cognitive (e.g.,
> memory) task but display very few significantly activated voxels, these
> voxels hardly revealing any temporal correlation with the experimental
> design (that is, they show essentially noise)?
>
> 2) Same question but this time with subjects succeeding on a task while
> showing no significant activation?
>
> 3) Would you keep these subjects for a group analysis? From a
> statistical viewpoint, can we consider that that these subjects are
> outliers? Is it legitimate to remove such subjects a posteriori (to
> include such subjects amounts to markedly influence the outcome of the
> statistical analyses, regardless of whether we use fixed effects model
> or random effects model)?
I don't think omitting subjects on this basis is a very good idea at all.
Your group analysis is testing the null hypothesis that there is no
significant activation produced by the cognitive task in the group of
subjects. So omitting subjects who show no activation is prejudging the
issue by assuming that there is a difference to be detected in the first
place! If the criteria on which you were planning to exclude subjects was
independent of the hypothesis being tested (and specified a priori), then
that would be a different matter. But that's not the case here, sorry.
You also can't assume that because the subjects show success on the memory
task that there *must* therefore be a difference in brain activity (that is
detectable using neuroimaging - I'm not a closet dualist!). Imagine a voxel
in which half the neurons fire in response to the memory task, and the other
half in response to the control task. The hypothetical voxel would show 'no
response' in imaging terms although being intimately involved in task
performance at a single neuron (and behavioural) level.
Sorry not to be more helpful...
Best wishes,
Geraint
--
Dr. Geraint Rees
Wellcome Advanced Fellow, Lecturer,
Division of Biology 139-74, Institute of Neurology,
California Institute of Technology, University College London,
Pasadena CA 91125 London WC1N 3BG
voice 626-395-2880 020-7833-7472
fax 626-796-8876 020-7813-1420
web http://www.klab.caltech.edu/~geraint
--
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|