Just thinking out loud here, and trying to do that scientifically.
Here are the two parts that actually differ, to help think through it:
The balanced receiver end, is same for both:
In the coax cable, the signal and shield are separate at the source, and therefore from the perspective of the balanced receiver, "lo" and "sh" have to be soldered. Also, on the source end "shield" is often at the ground potential. And on the receiver end, pin 1 is grounded - therefore pin 2 is also.
On the shielded twisted pair (or quad), they show the shield and one conductor soldered at the source. Then at the receiver, you have 3 distinct pins [with only pin 3 grounded, in the receiver].
Overall, it's true pin 2 is also grounded (because they are connected, on the source side), however I think it has to factor in the length of the cable and the assumption that the noise is entering somewhere during that cable length.
That could result in the noise being on pin 2 but not the double-grounded shield, and therefore the balanced circuit able to remove it - but since we're not dealing with a balanced differential driver or inverted signal on pin 2, I'm not yet understanding the removal or rejection mechanism fully.
And actually - maybe I'm thinking of that wrong anyway. There's only one op amp in the receiver and I think actual balanced receivers require 3, with two of them doing that noise removal by inverting and combining...
But regardless I'm not seeing how the noise is attenuated. Has to do with pin 2 not being grounded...