Hi. Good question....and not very easy!
The problem is how to validly make the two models comparable. If the two
models have different average duration values then they will have
different average regressor height, and you have to think hard about
whether the resulting PEs are then comparable.
So - _probably_ the best thing to do would be to rescale the durations in
one of the models so that the average value is the same as the average
duration value of the other model. I think that to first order this then
makes the betas comparable.
To actually make the comparison, you _could_ try putting both EVs into the
same analysis (i.e. into the same model) but if they are very similar then
the difference in fitting between them will get driven by the noise, and
I'm not sure how interpretable that would be. But it might work ok. Or you
can do as you suggest and fit them separately and do comparison stats
outside of FEAT.
Good luck! Steve.
On Tue, 20 Jul 2004, Amande Pauls wrote:
> I have a question about comparing two different models used for data
> I have used reaction times in the duration column in my 3 column event
> file (it is an event-related design) in one case, and the duration of
> the visual stimulation in the other case. The two models bring out
> different areas and I was wondering whether there is a way of
> quantifying this beyond just reporting different areas in a table. Can I
> do a paired t-test for example? The copes are derived from the same
> datasets, but were analyzed in different ways.
> Thank you for any help.
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