Hi Mark,
Thanks for the advice! I used the same data to test one variable and found select regions of significance (as expected), however I see the extended areas of significance when I test the same data against a subscore of the original variable. Therefore I do not suspect the issue to lie in the pipeline. I have demeaned the data (but also tried not demeaning). I have used "randomise -i GM_mod_merg_s3 -o fslvbm -d design.mat -t design.con -m GM_mask -5000 -T -V" with a GLM, as such:
EV1 EV2
Group Score
1 9
1 5
1 12
Any ideas?
Thanks,
Casey
|