I'm not good at quantifying this kind of thing, which may be why my teams don't do better--too many decisions by gut.
However, there is the factor that a so-so number 4 or 5 starter (say Garrett Stephenson or Jorge Sosa in 2003 terms) makes an above-average long reliever.
With pitching scarcity, I will sometimes hold a starter like this (Sosa is a better example because he's young and may improve) and use him as a long-reliever on the road and a starter at home. BY careful timing, you can get a lot of mileage out of these guys. Most teams don't have the luxury of qualtiy starters one through five, and this kind of starter can fill a void by coming in in blowouts, saving innings for the "real" starters.

This is just another feature that distorts any attempt to quantify the relationship, in my view.

Dave N