Regression analysis for determination of correct multiple

Hi,

I am looking for a simple idea how to decide between two valuations multiple using regression techniques, i.e. the outcome will tell me that e.g. EV/EBITDA multiple is more suitable than EV/FCFF for companies in that specific industry (defined by sample). I know that the choice of right multiple is usually more about common sense/practice, however, I was thinking of doing a linear regression (I know the relationship is probably non-linear) of change in EV against a change in EBITDA (1) and a change in FCFF (2) and than compare a statistical significance of both regressions (e.g. R2). What do you think? Any ideas?

Thanks.

Sam