On Mon, 29 Jun 2009 14:16:49 +0000, Marco Sperduti
<[log in to unmask]> wrote:
>Dear SPM users,
>
>I have some questions about PPIs:
>
>1) When I extract the time course with the VOI button I'm asked if I want to
adjust it with a contrast. Which is the difference between adjusting or not? If
I decide to adjust, which contrast should I use?
Adjust for "effects of interest". It's probably an F-contrast created by SPM
itself. By adjusting for effects of interest, you're _removing_ effects of no
interest from the timecourse. See e.g.
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?
A2=SPM;OEA31Q;20040406173202%2B0100;ind04
>2) In my experiment I have more than one session. By now I'm extracting the
time course for each session and calculating the PPI separately for each
session, then I use the parameters in each PPI as regressors for a GLM with
multiples sessions. Is that correct?
For some strange reason the issue of multiple sessions and PPI is somewhat
obscure. What I've done in the past, if I recall correctly, is instead of three
regressors and one session, used 3N regressors and N sessions. (The "three"
is, per the SPM way of doing PPI, the psycho and physio regressors, and their
interaction (the PPI).)
>3) Once the single subject analysis is completed, how can I proceed for the
group analysis?
Usual SPM two-stage (heirarchical) way: test the PPI at the subject level,
and then take that to the group level. To test the PPI at the subject level,
put a +1 contrast weight for the PPI regressor for every session. NB: be
careful if there are different numbers of sessions per subject; in that case you
should take an average. (Otherwise subjects with more sessions will be
artificially weighted more in the group averages.)
>Thank's a lot,
>
>Marco
>
|