Hi,
while analysing fmri-data I encountered a question regarding the procedure of FEATs. I have an event-related, within-subject design with 3 runs per subject. So far i found out, that i have to make the analysis in three steps: first-level for each subject and run indiviually, than second-level: fexid-effects analysis to combine the runs for each subject, and at last third-level: group-analysis. I found a tutorial in which the second-level was done in one step in FEAT. If i would do it that way it would look like that:
1 0 0 0 0 0 0 … Subj1, Run1
1 0 0 0 0 0 0 … Subj1, Run2
1 0 0 0 0 0 0 … Subj1, Run3
0 1 0 0 0 0 0 … Subj2, Run1
0 1 0 0 0 0 0 … Subj2, Run2
0 1 0 0 0 0 0 … Subj2, Run3
0 0 1 0 0 0 0 … Subj3, Run1
0 0 1 0 0 0 0 … Subj3, Run2
0 0 1 0 0 0 0 … Subj3, Run3
...
However the tutorial was for a 2-group comparison. So my question is, if I it is legitimate to do my second-level as well in one step or if it would be better to run the second-level for each subject individually.
I already looked up on tutorials and mails in this forum but could not find any design that resembled mine.
Many thanks in advance and kind regards,
Hannes.
|