01-09-11: Weseixas
The 3 db gain is correct for free space ( anechoic) go ahead and measure your hi-fi system, with speakers l/r and with room gain it is more than 3 db, when you turn off one side from the listen position.
Weseixas, room effects are a separate issue from the contention that I was disagreeing with. The contention I was disagreeing with was that the spl resulting from two paralleled speakers would increase 3db due to doubling of the power supplied by the amplifier, and ANOTHER 3db due to doubling of total cone radiating area.
My contention in response was that total radiated acoustic power would be doubled, resulting in a 3db increase in spl (neglecting any inequalities in the effects of the room on the two speakers). The Stanford paper I cited directly supports that, based on the assumption of "no interference." The paper Frank linked to takes interference effects into account (as distinguished from room reflection effects -- as you may realize they are two different things), and states that on average there will be a 3db increase, with a range between 0 and 6db depending on the interference effects at the exact listening position and frequencies.
Your post that I quoted above refers to a 3db increase as being correct apart from room effects, so I think that we are now essentially converged on that issue.
As far as room effects are concerned, I would certainly agree that inequalities of those effects on the sound produced by the two speakers could change the 3db figure, but not always in the direction of making it larger. Consider the situation where positioning is such that one speaker receives a bass boost due to room effects that is significantly different than the bass boost on the other speaker. Turning off the speaker that is receiving the smaller of the two boosts would reduce spl at the affected frequencies by less than 3db (as averaged over the listening space to account for interference effects), while turning off the other speaker would reduce it by more than 3db (as averaged over the listening space to account for interference effects).
Regards,
-- Al