Dear Alfonso,

I ran an FC-MVPA followed by post-hoc analyses to identify predictors of treatment response, using baseline rs-fMRI data only. The sample includes 86 participants: approximately half received treatment A&B and the remainder received A&placebo, 40 were responders and 46 non-responders.

Given that our analyses of treatment efficacy did not show differences between treatment arms, I did not initially account for treatment allocation. Accordingly, the main contrast was responders (to either treatment) versus non-responders (to either treatment).

I have now received a comment asking to run another post-hoc analysis to demonstrate that treatment allocation didn't influence the results. Do you know what's the simplest way to address or demonstrate this?

I considered including treatment allocation (A&B vs A&placebo) as covariates, but I am unsure which contrast would best answer this question (I know I can always rerun all of my analyses controlling for treatment allocation, but I'm unsure how I'd compare these results to my previous ones, as I have many resulting clusters).

The model would include these covariates: Responders, Non-responders, Treatment A&B, Treatment A&placebo.

Many thanks for your help.

Best,