It would be fun to do a direct test. Find a large rigid vessel of known volume and couple it to the RB along with a small syringe. Rapidly inject and withdraw a few cc of air and watch the pressure change recorded vs calculated.
You can see the delta-P effect of just raising and lowering the RB by say 1 meter (~ 12 pascals/meter = 48000 counts). Unfortunately the dynamics of the sensor itself and entrained air plus filtering make it hard to be precise. But it certainly is in the neighborhood of 4000 counts/pascal at low frequencies.
or 2500 micro-pascals/count.
But, at full bandwidth, resolution is really only about 1/4 pascal because of the relatively large amount of self-noise (1000 counts) Rboom self-noise
For me, the raw data should be scaled by dividing by 100 right at the very start. Then the self noise would be 10 counts. But that’s another story…