I am looking back into some data that I had previously analyzed and
trying to test it in a different way. I know that there is a
relatively straightforward method for getting what I want and that
Steve Smith had explained during the UCLA training session last summer.
While it seemed obvious at the time, my mind is now again blanking on
how to go about it.
My research design is as follows:
3 subject groups --- Rep, Dem, Nov (6 subjects per group)
7 question categories -- A, C, L, M, R, T, U
3 scans per subject
In my first level (individual scans), I have created number of
contrasts, including:
C versus A, L, M, R, T, U
L versus A, C, M, R, T, U
My second level just takes the mean activation of the three scans per
each subject.
My third level creates contrasts based upon the groups:
(Dem & Rep) versus Nov
What I want to do now is an analysis that treats the C (conservative
threat) questions for the Republicans the same as it treats the L
(liberal threat) questions for the Democrats. So, I could do an
analysis such as (L for Dem & C for Rep) versus (L for Rep & C for
Dem). Basically, I'm trying to find a good way of seeing if there is a
common activation for these two threat stimuli, even though the
perception of threat will be reversed for the two groups.
I vaguely recalled that this could be done in the second level
analysis. Thanks in advance for your help.
Darren
|