I'm not too sure how to analyze the following situation in Feat. I am trying
to combine runs in a subject where each run is not an exact repeat in terms
of EVs:
A subject does 6 runs of a study that has 3 conditions (A, B, C),
but only 2 per run i.e.
run 1: A vs B
run 2: A vs C
run 3: B vs C
runs 4, 5, 6 repeat 1, 2, and 3
In addition, the EVs for a given condition are not the same across
runs (they are self-paced).
What would be the best way to deal with this? possibilities i thought
of was to concatenate data from all runs and fix EVs appropriately, or
analyzing each run separately, but then how do i combine the data? i'm
thinking by weighing particular contrasts in FLAME, but not sure how
exactly. Note that because of the pair-wise presentation of conditions in
the runs, the COPE numbers will never match, which
really is the source of the problem and led me to use lower level copes as
inputs as describes below.
Here are two ways I tried, but i am not sure what is more robust/elegant:
so for the first level, I basically analyzed each run separately so that for
each run I have 2 EVs and the contrasts are
1 0
0 1
1 -1
-1 1
which gives copes 1 through 4 for each run.
now for higher level analysis I entered all 24 copes in that order and all
were set to group1, then 9EVs:
1 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0
0 0 0 0 1 0 0 0 0
1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 1 0 0
0 0 1 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 1
1 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0
0 0 0 0 1 0 0 0 0
1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 1 0 0
0 0 1 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 1
and the contrasts simply:
1 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0
0 0 0 0 1 0 0 0 0
0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 1
The other option was to just have in the second order analysis, EVs for just
the main conditions vs rest (i.e. cope1 and
cope2 from each first level analysis). Again, all in the same group:
1 0 0
0 1 0
1 0 0
0 0 1
0 0 1
0 1 0
1 0 0
0 1 0
1 0 0
0 0 1
0 0 1
0 1 0
then the contrasts would be:
1 0 0
0 1 0
0 0 1
1 -1 0
-1 1 0
1 0 -1
-1 0 1
0 1 -1
0 -1 1
I feel this second way is better just intuitively, but it seems I'm going at
it the wrong way altogether.
Any tips appreciated!
thx in advance!
cherif
|