Dear FSL List,
I'm analysing an event related dataset consisting of 28 participants
with two sessions each.
I have a rather complicated factorial design, with three regressors,
two of which with two and one with three levels. Additionally, my
subjects are split into two groups (one between subject factor), so my
design becomes a
2*2*2*3 design.
After trying many different settings and models, I have come up with
the following approach.
1) Run 1 lower level feat full analysis separately for both runs
with all 28 subjects with 7 EVs (2+2+3), where each EV (3 column
design with the onset times for that category) is set to one to test
against baseline activation
2) Run 1st higher level fixed effects analysis using lower level feat
directories, with 56 inputs (2*28 subjects) and 28 EVs and 28
contrasts, such that each subject becomes an EV/Contrast (the manner
suggested in the feat manual)
3) Run 2nd higher level analysis using the output of (2)
Here is my question. Following the FSL FAQ #49, as I understand it, I
would have to select "load 3D COPE images" and load 7*28 cope images
as input. That's a lot of input ... Can I instead select "use lower
level feat directories" as input and load the cope1.feat through
cope7.feat directories containing th 28 lower level (subject merged
runs) copes?
Once the question of input has been resolved, I would then proceed to
again define my 7 EVs using the uploaded input and set up contrasts
where I check for interactions between the EVs
4) A 3rd higher level analysis using the output of the former, where
now only the between subject factor is tested
I hope I was able to explain my design and question in a
semi-understandable way. Any help and suggestions are greatly
appreciated!
Many thanks in advance and kind regards,
Laura
--
Laura Mega, stud. Dipl. Biology
Werner Reichardt
Centre for Integrative Neuroscience (CIN)
Paul-Ehrlich-Straße 17
72076 Tübingen
|