Dear all,
We are preparing to analyze data from a neuroimaging treatment study as part of a clinical trial and we propose to use a 5 (GROUP: CBT +FLX responders, CBT +FLX non-responders, CBT+ PLB responders, CBT +PLB non-responders, controls) x 2 (time 1, time 2) factorial design in SPM8. We have a patient sample size of n=140 and 30 controls. Assuming we get approximately 20 non-responders in each treatment arm, will a factorial design be powered enough to detect any significant differences in our neuroimaging data? Is there a better model to account for smaller sample sizes in each cell?
Many thanks,
Rebecca
