Dear Maria,
Well, you can set up separate models to look at task effects for group 1 and group 2, and if you're just interested in an interaction group x task, okay, why not ignore main effect group. But then there's no need for a 2x2 ANOVA model. You would take the con images reflecting the differential (task) contrast from group 1 and those from group 2 and contrast them with a two-sample t-test. To calculate effect sizes (Cohen's d) see this thread https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;6168c42b.1408 and Christian's reply, including a script (I think it's also included in the VBM8 toolbox).
Best
Helmut
|