Hello!
I'm working with a flexible factorial mixed-design model with 2 groups and 2 within-subject main effects, making a 2x2x2 anova.
I fell in the list with the advise to a similar case to mine, saying that mixed-design models of this sort are not valid, because SPM uses only one error term and one set of degrees of freedom. It's advised that we break the model into simpler models testing one effect at time.
I did it.
Now I observed significant activation values for both within-subject main effects tested separately. Now I want to model their interaction term, but I'm not sure how to do it in a separate model?
Anyone can help me?
Thanxx!
|