Hi there,
I used a 2 (A) *2 *(B) *2 (C) within subject design. I was hoping to model the subject effect as well as is done with the flexible factorial design.
It looks like modelling the subject effect increases sensitivity in some cases, so I'm keen to try this before reporting that the full factorial design, where we specified all the factors as non-independent, resulted in a null effect for one (key) comparison.
However clearly the flexible factorial design is limited to 2-way ANOVAs. I thought of three possibilities:
(1) Is there a way of implementing a subject*A*B*C design in SPM12? - e.g. by 'pretending' I have a single factor with 8 levels?
(2) Alternatively, could I use a full factorial design, but add covariates to model the subject effects?
(3) Or alternatively still, I could use two flexible-factorial designs for each half of my experiment (e.g. a subject *A*B in level 1 of C, and a subject *A*B in level 1 of C), and compare them in a second step?
Thank you for any advice!
Deborah
|