Dear Daniel,
>Dear Karl, Dear SPM users
>
>We recently asked you for advice concerning our prepulse inhibition
>(PPI)-H2O-PET study.
>May we bother you again with some basic questions:
>
>This PET study includes 9 scans/3conditions per subject. So far we have
>measured two
>goups of subjects (a and b). The group a and b have been stimulated using a
>different number but identical startling acoustic pulses and prepulses
>respectively per scan.
>
>We wish to address the question whether the different number of stimuli used
>between group a and b lead to different activation patterns.
>So far we tried to use a multigroup design and the following contrasts to
>assess the differences between group a and b for the pulse alone conditions:
>
>1a 2a 3a 4a 5a 6a 7a 8a 9a 1b 2b 3b 4b 5b 6b 7b 8b 9b
>NS NS NS PP PP PP P P P NS NS NS PP PP PP P P P
>0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 -1 - 1 - 1
>
>
>similarly, for the differences base on prepulse plus pulse (PP) we used:
>
>1a 2a 3a 4a 5a 6a 7a 8a 9a 1b 2b 3b 4b 5b 6b 7b 8b 9b
>NS NS NS PP PP PP P P P NS NS NS PP PP PP P P P
>0 0 0 1 1 1 0 0 0 0 0 0 -1 -1 -1 0 0 0
>
>However, the program did not accept this contrast settings!!!
>
>What might be the appropriate contrast settings to investigate the effect of
>the different numbers of pulses and prepulses used between the two groups
>???
Your problem is that your parameters are not uniquely estimable, meaning
that the task effects (as you have defined them in your contrasts) are
confounded with the group effect. What you want is a task by group
interaction. Not knowing the details of your experiment I will assume that
NS denotes some form of "baseline" condition identical across the two
groups. The conrast you are interested in would then be
1a 2a 3a 4a 5a 6a 7a 8a 9a 1b 2b 3b 4b 5b 6b 7b 8b 9b
NS NS NS PP PP PP P P P NS NS NS PP PP PP P P P
-1 -1 -1 0 0 0 1 1 1 1 1 1 0 0 0 -1 -1 -1
Please note also the strictly speaking you should model this as a random
effects study. There are a lot of recent postings on the mailbase which you
may want to consult regarding this.
>
>Could it make sens to use a design where we treat the two groups (a and b)
>as one
>group of 18 scans ti investigate this effects?
>
>for example for the pulse condition:
>1a 2a 3a 4a 5a 6a 7a 8a 9a 10b 11b 12b 13b 14b 15b 16b 17b 18b
>NS NS NS PP PP PP P P P NS NS NS PP PP PP P P P
>0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 -1 -1 -1
>
No, this latter approach does not make sense. You would then confound your
condition effects with your subject effects.
Good luck Jesper
Jesper Andersson
Wellcome Dept. of Cognitive Neurology
12 Queen Square
London WC1N 3BG
phone: 44 171 833 7484
fax: 44 171 813 1420
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|