Hi Delia
> We have performed an fMRI experiment on two groups of subjects (P and
> N), each under
> two conditions (A and B). I would explore whether there are brain
> regions in
> P that show greater activation during A that during B (2x2 factorial
> design).
> Three simple questions:
> 1) Is it correct that I use Multi-group: conditions &covariates,
> giving the
> contrast 1 -1 -1 1, that is (PA-PB)-(NA-NB)?
>
Yes why not, you can use this kind of design. For this contrast, you are
looking at the interaction group*condition.
> 2) If I use the con.img for each individual, is it correct to tell that I
> perform a random effect analysis?
>
Correct
> 3) If I give the contrast 0 1 0 -1, then can I test statistically that
> activations of condition B in P are greater than those of condition B
> in N?
>
That's what this contrast does (setting condition A =0)
For further information on contrasts and factorial analysis, have look
at R. Henson and W. Penny ANOVA paper
http://www.fil.ion.ucl.ac.uk/~wpenny/publications/rik_anova.pdf
In addition if you are not sure of your 1st level (individual) matrices,
you can look at my SPM2 fMRI user guide
http://membres.lycos.fr/wamcyril/Documents/SPM2_eng.pdf
Best,
Cyril
|