Dear SPMers,
I would like to compare two groups of subjects ( 31 females and 50 males) using a flexible factorial design, with a 2 x 2 factorial design (two groups and two conditions), and a covariate (total GM volume). I have set up the design following the tutorial by Glascher and Gitelman (two groups of subjects, design 1) and have run the following contrasts:
main effect of condition:
[zeros( 1, 81 ) 1 -1 1 -1]
[zeros( 1, 81 ) -1 1 -1 1]
With this contrast the results make sense.
Then I have run the following:
interaction group x condition:
[zeros( 1, 81 ) 1 -1 -1 1]
[zeros( 1, 81 ) -1 1 1 -1]
Are these contrasts correct if I want to test group x condition interaction?
Finally I have tried to assess the main effect of group using the following contrasts:
[ones( 1, 31 )/31 -ones( 1, 50 )/50 0.5 0.5 -0.5 -0.5]
[-ones( 1, 31 )/31 -ones( 1, 50 )/50 -0.5 -0.5 0.5 0.5]
These contrasts only work when I completely remove the covariate (GM volume) from the design. I cannot figure out how to assess for differences between group while including the covariate in the model.
Thanks a lot for your time,
Vittoria
|