Dear Alfonso and CONN experts,

This is a follow up from a previous question.

I am having a hard time to interpret ROI to ROI because I have a data set which I know to be very noisy, C, but, nevertheless, arrives to ROI to ROI results. 

When I combine this noisy cohort with other 2 non noisy cohorts, A & B, I replicate results from the non noisy cohorts, but get completely different results from the noisy one, C.

Any idea on why this happens?

All best!

Renzo