Hi AFNI Experts!
I have some statistical questions, which I came across while going through the 3dANOVA manual.
I have a set of male subjects, who all underwent the same experiment.
The experiment was set up as a mixed design consisting of rapid event related "blocks" and pure baseline periods in between. Two tasks had to be performed during these event related blocks: a memory task and an easy cognitive task (Labeling). Visual stimuli with 3 valences (pleasant, unpleasant, neutral) have been used for both tasks.
The scientific background for my statistical questions is briefly following:
Some brain regions have repeatedly shown to be activated dependend onthe valence of stimuli (e.g. amygdala during unpleasant and pleasant in comparison to neutral stimuli). Other brain region like BA 9 have been shown to be activated during working memory tasks in comparison to simple cognitive tasks as labeling.
I am interested in figuring out, how emotional and cognitive circuits interact.
Eg. if BA9 activation is enhanced by the presence of a pleasant stimulus in comparison to unpleasant or neutral stimuli.
My data are normalized to % signal change and AUCs has been used as input.
I specified 2 factors:
Factor A: Valence consisting of 3 levels (Pleasant, unpleasant, neutral)
Factor B: Cognitive state consisting of 2 levels (working memory and labeling)
At this timepoint only a within subjects analysis is planed.
Following six briks have been chosen for the "dset" input for 3dANOVA2:
P1: pleasant under WM
P2: pleasant under LA
U1: unpleasant under WM
U1: unpleasant under LA
N1: neutral under WM
N2: neutral under LA
I used a fixed effects model.
Well, after thinking about all, I am not quite sure, if this is the best way to explore this anymore. Is it reasonable to use a random effects model? And why? If yes, how can I calculate contrasts since this options are not available within this model. Are my factors correctly set up? And how do I do post hoc tests in AFNI? Finally, does the output of 3dANOVA need a correction for multiple comparisons?
Sorry for bugging you with all these questions. Any comment would be very helpful.
Thanks,
Lukas