Hi all again,
I realize I wasn't necessarily clear when I wrote this last night. I
have ONE behavioral measure per subject. Hence, at the first level of
analysis, I run a GLM and obtain a single contrast map for each subject for
the condition of interest. Afterwards, at the second level of analysis, I
run a simple regression across the subjects with the each subject's
behavioral value making up the regressor. Since I have run this separately
for two groups, I would like to compare the groups to see where they differ
in their regressions on a voxel-wise basis, and I would like to implement
this in SPM2. I hope this was clearer... Thanks!
Dani
Daniel Simmonds
Developmental Cognitive Neurology
Kennedy Krieger Institute
[log in to unmask]
On Thu, 4 Jan 2007 03:53:21 +0000, Daniel Simmonds
<[log in to unmask]> wrote:
>Hi all,
> We have run separate simple regression analyses for two groups (control
>and patient) with a single behavioral variable as the regressor. We
>obtained 4 contrasts - the positive correlation [1] and the negative
>correlation [-1] for each group. As the next step, we would like to compare
>the groups, to see where the correlations differ between the groups, on a
>voxel-wise basis. We would like to see voxels in which controls have
>steeper correlations than patients, and vice versa (afterwards, we would
>perform post-hoc analyses to characterize whether the correlations were
>"more positive," "less negative," etc...). How can I implement this
>comparison of correlations between groups in SPM2? Thanks so much!
>Dani
>
>Daniel Simmonds
>Developmental Cognitive Neurology
>Kennedy Krieger Institute
>[log in to unmask]
>========================================================================
|