I think Vurbal hits the nail on the head: given a given loudspeaker system - meaning that the efficiency of the driver, stiffness of the cone, dampening in the box, rigidity of the box, etc, are all held constant - then Ohm's law and all that other stuff will tell you how changes to the input will affect changes in the output. And that's precisely *why* it's expressed in db's - we don't know the actual absolute value, but we can still safely talk about relative values. So for your headphones, if you've got an amp design that produces X in your headphones, you (in theory) could tweak it to produce, say, 2X. But in real life you can't, of course, because that amp is a proprietary product, a black box to you, so you can't actually tweak it yourself.
As far as factoring in those values above, I think it's largely voodoo. There are rules of thumb, even things that pretend to be scientific and formulaic, but as far as I'm aware that's just formalizing those rules of thumb. They're as useful as high school physics formulas that assume point-masses and zero friction.
I also agree with the sentiment here that the final impression of the sound is determined largely by the music's mastering. It seems clear to me that the perception of the sound volume can be very decoupled from the actual sound energy being output. And generally speaking, it's that cleaner sound, even at higher "sound pressure", isn't perceived as loud, but when there's audible distortion, that quickly drives up the perception of loudness.