dear all,
When comparing groups with a random-effects analysis, must the contrasts
which are taken to the second level be the same for both groups? That is,
if one has three conditions, A, B, and N, can one take the A-N contrast
for Group 1 and the B-N contrast for Group 2 and perform a valid 2nd-level
comparison with a 2-group t-test?
The idea behind this is that I'm interested in the activity related to a
particular emotion (e.g., positive emotion) and there are 2 groups which
differ in the materials that will elicit this emotion -- group 1 has this
emotion in response to the condition A stimuli but not condition B, and
vice versa for group 2. N is a control condition that elicits neutral
emotion.
I'd like to: 1) characterize the emotion - neutral activations within each
group separately (no problem here) and then 2) compare the emotion -
neutral activations (those related specifically to this emotion) between
groups directly. The idea is to characterize the group differences in the
neural response associated with this emotion. This would seem to be
addressed by the analysis above, but perhaps someone can suggest a better
analysis? Thanks,
Stephan Hamann
Dept. of Psychology
Emory University
Atlanta, GA 30322
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|