Hello SPM users,
I have a question pertaining to disparate group sizes. I want to
compare two groups, one numbering 12 and the other 60. Would it be
valid to compare these groups using "Population main effect: 2
conditions, 1 scan/condition (paired-t test)"?
It seems as though statistical problems could arise, given the
following crude hypothetical: 10 people in the small group
demonstrate activity in the amygdala at intensity of 10000, while 50
people in the large group demonstrate activity in the same area with
an intensity of 5000. Both subsets are 5/6 of the group size, and the
intensity of amygdalar activation in the small group is double that of
the large group. However, it seems that SPM might assign the larger
group with significant activation in the amygdala, as the higher N
imbues the group with more statistical power. Does SPM account for
this problem?
Thanks!
Aaron
