Yes but in this case we're talking about returning to the office, which doesn't have the exact same dynamics.

In the case of the return to office dynamic, you're specifically filtering out high-quality workers who also prefer to be remote, and the results are retention of all low/medium quality workers regardless of remote/office preference, and high quality workers who don't mind or have some reason that keeps them in the office. You don't filter out low-quality workers whatsoever because they'll at least go to the office to keep their job assuming they don't have options (or else they'd be higher quality workers or maybe they just leave the work force).