Dear Swann
> -----Original Message-----
> From: Swann Pichon [mailto:[log in to unmask]]
> Sent: Thursday, October 04, 2007 11:19 AM
> To: [log in to unmask]
> Subject: Setting up a PPI model with factorial designs > 2*2
>
> Dear Darren Gitelman,
>
> It has often been said that 2*2 factorial designs were the
> most appropriate ones to run PPI's analysis.
> I was however wondering whether more complex factorial
> designs could pose a problem.
> For example, In a 2*3 design comprising 2 factors F_A(2
> levels) & F_B(3 levels), we could imagine that the
> interaction between the psychological factor F_A and the
> timeseries of R modulates the activity of another region only
> in the context of (F_B1 versus F_B2) but not in the context
> of (FB_1 versus FB_3).
> How could these differential effects be dissociated with a
> simple model (PPI.ppi, PPI.Y, PPI.P) ?
I think the simple answer is that they can't using PPI. This is something
that DCM should be able to do, but I'm not sure the modulation of activity
using the interaction you describe is fully implemented yet.
Darren
> In advance, thank you very much for your kind help,
>
> Swann Pichon
>
> --
> Cognitive Neuroscience Ph.D. Student
> LPPA - CNRS UMR 7152 - College de France
> 11 Place Marcelin Berthelot - 75231 Paris cedex 5 TEL : +33
> (0)1 44 27 13 87 GSM : +33 (0)6 26 43 83 61 FAX : +33 (0)1 44 27 13 82
>
|