Within limits, hanging the absorption off the wall puts the absorptive material in a location where the particle velocities are higher; hence, more absorption takes place. (At a rigid wall, the particle velocity is zero. With no motion, there is no kinetic energy to convert to heat.) This effect should mostly be noticed in the low frequencies. As frequency rises and the wavelengths of the sound waves start to approach the absorber thickness and/or spacing off the wall, you shouldn't detect a difference. (Note that the wavelength of a 100 Hz tone is roughly 3.4 meters, so we're still talking about small fractions of a wavelength in the bass region.)
I doubt you'd find much measurable difference between the 3 setups you described, except if you're measuring cost. I do think the 4" absorber mounted on the wall would be better than the other two options, barring impedance-matching issues, but "better" would probably be fairly slight. I'd like to see some real data to back up my hunch, though.