Hi Eric
This is your 2nd level analysis setup? Because if so, I am assuming
you didn't model the paired differences at the first level. I think
this would be the thing to do, assuming your contrast is task a - task
b, with both presented in a single scan. Then at the second level, you
need only model the mean. Outputs will be multiple copes (1 for each
contrast at the first level) which can be loaded into your group
analysis.
Does this make sense?
On Feb 9, 2010, at 2:20 PM, Eric Zalusky wrote:
> I am trying to run a second level analysis on a single subject with
> two scans.
> For the first level analysis, I used two contrasts 1 (for positive
> activation)
> and -1 (for negative activation).
>
> For the second level analysis I used the template from the website
> for a two-
> sample paired T-Test. My goal for this analysis was to find the
> difference in
> areas activated and deactivated between scan A and B.
>
> I loaded both feat directories and used the lower-level copes 1
> (positive) and
> 2 (negative). Fixed Effects was used. In the model set up I have 2
> EVs.
>
> Group EV1 EV2
> A 1 1 1
> B 1 -1 1
>
> C1 A-B 1 0
> C2 A-B -1 0
>
> The feat report for the second level gives me two lower-level
> contrasts:
> Positive: A-B and B-A.
> Negative A-B and B-A
>
> The problem I'm running into is that when you look at A-B for
> positive and B-A
> for negative they are exactly the same. This also applies to B-A for
> positive
> and A-B for negative.
>
> Have I correctly set up my second level analysis to look for a
> significant
> difference between A and B for a single subject? And do these
> results make
> sense?
>
> Thank you for your help,
> Eric
Jennifer Bramen, Ph.D.
Assistant Research Neuroscientist
David Geffen School of Medicine at UCLA
Developmental Cognitive Neuroimaging
635 Charles Young Drive South
Los Angeles, CA 90095-7332
Phone: (310)525-0865
Fax: (310)206-5518
Email: [log in to unmask]
Campus Mail Code 176919
|