Dear Alec,
these two contrasts are mathematically identical.
The 2. contrast B(L1>L2)>A(L1>L2) is nothing else as
( B(L1) - B(L2) ) - ( A(L1) - A(L2) )
= B(L1) - B(L2) - A(L1) + A(L2)
= A(L2) - A(L1) - B(L2) + B(L1)
= ( A(L2) - A(L1) ) - ( B(L2) - B(L1) ) , which is your 3. contrast
So, to come out of this dimlema you can, for example, use masking.
Good luck,
Karsten
--
-------------------------------------------------------------
Karsten Specht, PhD
Department of Biological and Medical Psychology
Bergen fMRI group
University of Bergen
Jonas Lies vei 91
5009 Bergen
Norway
Tel.: +47-555-86279
Fax: +47-555-89872
[log in to unmask]
http://fmri.uib.no/
-----Ursprüngliche Nachricht-----
Von: "Alec Sproten" <[log in to unmask]>
Gesendet: 19.sep.2011 14:34:15
An: [log in to unmask]
Betreff: [SPM] Different contrasts, same result?
>Dear SPM experts,
>
>we run an experiment with two groups of subjects (A and B) and with a task with two levels (L1 and L2).
>We are interested in the following contrasts:
>1. A(L1>L2)>B(L1>L2) (thus a t-test for A over B with the brain activation of L1 over L2 in both groups)
>2. B(L1>L2)>A(L1>L2)
>3. A(L2>L1)>B(L2>L1)
>4. B(L2>L1)>A(L2>L1)
>
>To get these contrasts, we did in the first level analysis the differential contrasts L1>L2 and L2>L1 in each group (so we can exclude within-subject variability consideration in the 2nd level analysis).
>In the second level we then made the comparison of these contrasts for the two groups. We specified the design with 2 factors:
>
>Task (within subject) - Independence: No. Variability: Equal.
>Group (between subject) - Independence: Yes. Variability: Unequal.
>
>Now to my question: when we looked at the results of contrast 2 and 3, we got aware that we got exactly the same result for the two contrasts. Can you please explain me how this comes, and how we should set the contrasts to avoid this effect?
>We are not sure what we should do but we only speculate that we need to change the definitions of the Task factor regarding the independence or the variability.
>
>
>Thank you very much in advance,
>Alec
|