Dear List:
We have completed a PET experiment using a 2x2 (A,B) factorial design in
SPM99. The A main effect produced statistically significant activations but
the B main effect did not. The interaction effects also did not produce
significant activations. A question I have is whether any areas of the brain
are activated in both levels of A, regardless of the levels of B. I've looked
in the archives but have not found anything specifically related to this
question. I thought that a conjunction analysis involving the two levels of A
(i.e., conjuction of 1 1 -1 -1 and -1 -1 1 1) might answer this question, but
SPM simply tested the first contrast. I tried reversing the orthogonlization
order and SPM99 returned an error but tested the 2nd contrast. I read through
the '97 paper by Price and Friston on conjunction analysis and I'm not sure
that the analyses above are valid given the lack of baseline conditions in the
factorial design we used. Is there a way to statistically test whether there
are any common areas activated over the two levels of A?
Thanks,
Warren Darling
|