Dear SPM-group:
I have a group of 10 subjects who I predicted would show amygdala
activation in an emotion task based on prior studies. I originally planned
to conduct a random-effects analysis, but on checking individual subjects
I found that 6 showed primarily left amygdala activation, 3 showed
primarily right amygdala activation, and one weakly activated both left
and right amygdala (this was a blocked fMRI study with an implicit
baseline of fixation).
>From this individual-subjects analysis it seems that the assumptions
underlying the random effects model are not valid for these data, there
is clear bimodality.
On the other hand, it would seem that there should be some way to answer
the question, "do people typically activate structure X either on the
right or the left" in SPM. Or, is a fixed-effects approach my only
recourse, perhaps with the added individual-subjects analysis to show that
each subject in each subgroup (left amygdala and right amygdala group)
shows the effect individually?
As always, all help much appreciated,
Stephan Hamann
Dept. of Psychology
Emory University
Atlanta, GA 30322
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|