The ER is a precise number that comes out of the optics design of the eyepiece (it's the distance from the back surface of the final lens (or perhaps it center in some cases) to the position of the entrance pupil.
Depending on the eyepiece design only it can depend on several parameters. Generally as the eyepiece focal length decreases the ER decreases too (so 10x bins have a lower ER than the 8x bins of the same design).
The best description of how the rays converge (afocally) through the exit pupil is in Wikipedia
http://en.wikipedia.org/wiki/Exit_pupil said:
The exit pupil can be visualized by focusing the instrument on a bright, nondescript field, and holding a white card up to the eyepiece. This projects a disc of light onto the card. By moving the card closer to or further away from the eyepiece, the disc of light will be minimized when the card is at the exit pupil, and the bright disc then shows the diameter of the pupil. A clear vial of milky fluid can also be used to visualize the light rays, which appear as an hourglass shape converging and diverging as they exit the eyepiece, with the smallest cross-section (the waist of the hourglass shape) representing the exit pupil.
The "blackouts" you see most of the time are just vignetting of this bundle of rays so the "range" is not a precise one (it depends on the angle of bundle and the size of your entrance pupil and the positioning of your eye on the axis (large exit pupils are preferred becaise they allow imprecise positioning of the binocular). Some blackouts ("kindney bean blackouts") appear to be due to curvature/spherical aberation in the exit pupil though I don't have any good reference for that but it seems to crop up in some designs making them less usable for "some" people (but I'm not sure how to define that population).
A part of the problem comes from a lot the manufactures just taking the measurement from the optical designer and quoting it in the specification. And not taking into account the thickness of the eyepiece or eyecup hardware. A classic exampe of this is the Swift 820 porro with a speced ER of 16mm (enough for eyeglasses you would think) but when you account for the hardware the usable ER is 12mm (measured by edz) which is not enough for a whole field for majority of eyeglass wearers.
Of course some do take account of this hardware and quote an (honest) effective ER but then it's difficult to tell who is doing what.
The other problem comes with the vertex distance of your glasses (the distance from the front of the cornea to the back side of the eyeglass lens). Though in opthalmic optics this is usually taken to be a "fixed number" of around 12mm if one makes the measurement (I've actually read a paper where they did that) they found a range of vertex distances from 6mm to 15mm (or more). This is where most of the problems come from: glasses worn down the nose, thick glasses (it matters!).
Add that to the shape of your face: how deep set are your eye's from the bridge.
You can see the number of variables mount up. So once again the only way to really know is to try it.
e.g. I recently got a Yosemite 8x30 with speced ER of 14mm. I can see pretty much the whole field in my close-fitting eyeglasses (and even only miss a very small amount in by bigger less close fitting glases). So I rather suspect Leupold are giving a "usable ER". Or I'm just getting lucky.
The only thing in the eye that could affect the position of the entrace pupil (which has to match up with the exit pupil of the optics) is if the curvature of the cornea, which itself is a lens which a large strength, is rather difference than they expect. As the cornea images the entrace pupil of the eye it moves it's apparent position from that actual physical position of the iris. But the iris itself is rather flat so ER doesn't change with diameter of the aperture in the iris.