On Thu, 9 Apr 2009 12:01:54 -0400, Jeffrey West
<[log in to unmask]> wrote:
>hello
>
>i am attempting to limit the overall sessions required for my first level
analysis. basically, i have two groups of subjects, each with 10 participants,
totalling 20 subjects. for my tasks, i have 2 scans, one scan performed prior
to training administering an easy 1 task and an easy 2 task, also, a hard 1
task and a hard 2 task. this is the same pattern for both groups.
>
>i would like to put as many of the variables into the first level analysis as
possible, to eliminate the arguement that the pre and post or the easy and
hard tasks may have different baselines. so for example, for my first analysis, i
want to look at group 1 pre-training for the easy 1 and 2 and the hard 1 and 2
variables. i am using spm2 for these analyses. so for this first level analysis,
each subject will have 4 sessions (one for easy1, easy 2, hard 1, and hard2)
and that will give me 40 total sessions. this seems like a lot of seesions, one
for me to enter and two for the computer to analyze. i am ok with doing this
(as long as i dont have to start over), but it seems there should be a way to
combine e1 with e2 and h1 with h2 to take the sessions down to 20 total. i
can do this, but then i am stuck because i do not have vectors for e1
combined with e2 or h1 combined with h2. i only have vectors for them
separately. i would really appreciate any suggestions on where to go from
here.
>
>because in some of the other analyses i have planned, i would like to look at
subtracting group 1 (pre-training easy 1 and easy 2) from group 1 (post-
training easy 1 and easy 2), so each subject will have 4 sessions for a total
of 40 sessions again.
I don't think combining sessions is a good idea, unless you're very careful
about how you go about it. (There's been some posts on this list about this,
one pretty recently I believe.) Here, by "combining sessions" I assume you
mean taking multiple scanner runs (run = images gathered without the scanner
being off) and getting SPM to think they're one run.
If you're concerned about the labor needed to input the data, you might
consider other labor-saving devices (maybe you already have). First, you can
do batch scripting. Or you can store data (like vectors for onsets, user-
specified regressors) in variables in the matlab workspace, and then just type
the variable name in the textbox.
>
>thank you for your time.
>
>jef
>
>
>
|