Hi,
I conducted within subject analyses and coded for contrasts examining
effects of interest & am now conducting a random effects within group
analysis using SPM2. I would like to exclude activations from one contrast
(C1-task difficulty) from the results of another contrast of interest (C2-
retrieval); since I believe some of C2-retrieval effects may be due to
task difficulty.
To do this I have exclusively masked out activations of random effects
results for C1 by saving this contrast result image, set at a liberal
threshold of .05 uncorrected k=0, and then exclusively masking for
C1.img , when examining the results of C2 at p<001 uncorrected. However,
in reading the SPM list, I noted that it was recommended that the better
method would be to explicitly mask for C1 when running the GLM of C2,
however, wouldn?t this reduce the # of voxels assessed and inflate the
significance of unmasked voxels?
Which option is best if I simply want to exclude activations from C1 when
examining the significance of C2 effects; exclusive masking at the results
assessment stage or explicitly masking at the run-GLM stage?
Thanks for your help.
Sincerely,
Natasha
|