You're right, the fixed "room gain" addition is by far the fuzziest part of the whole calculation and can be considered wrong.
Room gain can still factor into efficiency ratings, as some manufacturers seem to design for more roll-off anechoic (and can often spec a higher sensitivity as a result, under the assumption that most rooms will fill it in just right) and some for more flat anechoic (with lower spec'd sensitivity and perhaps a bit of a in-room boost in the low end, which could arguably be more "musically pleasing").
Mainly I'd meant to point out that on-paper efficiency numbers seem to be not-too-infrequently "fudged" (or at least highly inconsistent). Jim Smith (author of "Get Better Sound") has posted that 1 model of Zu speakers spec'd at 98dB (though perhaps only a claimed ~97dB/Watt if they were 12 ohms and the spec'd number was sensitivity) measured 3dB less loud than the 96dB/Watt Tannoy Canterbury in his setup, same input. A 4dB discrepancy is very substantial.
Edited by mulveling - 11/4/12 at 11:16am