Hi Brad,
On Fri, 29 Jul 2005, Brad Goodyear wrote:
> > A couple more quick questions.
> > You mentioned that number of sessions is a consideration. If you have
> > only 2 sessions, can you even do a mixed effects at the second level
> > in a repeated measures when all subjects are included in the analysis?
> > You do get a map, but seems sparse for some subjects even at p=0.05
> > uncorrected.
For the purposes of combining across sessions for each subject, this is
right, such as in the 3-stage example shown at:
http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html#MultiSessionMultiSubject
Alternatively, if you wanted to be able to investigate the cross-session
results for each subject, yes, run FE separately cross-session for each
subject (but this doesn't then form part of the full analysis in the
previous paragraph).
> > Also, if I'm interested in how combining sessions improves
> > quantification of results, and say I want to compare results obtained
> > combining 4 runs with that obtained using only the first run, is it
> > wrong to use a mixed effects when combining the 4 runs because it uses
> > the cross-session variance in its estimation? Should I compare 1 run
> > with '4 runs mixed' and also with '4 runs fixed'?
Indeed - again, if you want to ask that comparison wrt each subject
separately, then yes, just compare each separate session with the FE
across sessions. You may also want to ask this question wrt the
cross-subject analysis - in which case compare the 3-stage analysis with a
simple two-stage that just feeds one session from each subject into a
second-level cross-subject ME analysis.
Cheers, Steve.
--
Stephen M. Smith DPhil
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|