Having done a high-altitude stalling module recently, I thought it was of great benefit despite the caveat that the aerodynamic simulation might not be exact.
It showed:
a) You had to agressively reduce the AoA by pitching down to a far greater degree than in any other flight phase.
b) The engines were next to useless, given spool-up times and the thrust generated near the aircraft ceiling, so it was more a glider-style recovery.
c) You were going to lose a LOT of altitude, no matter what.
d) If you tried to level out too early, before getting to c), it was quite possible to stall again.
I would argue that the fidelity of the simulator was unimportant in getting those basic principles across. After all, every LoC / jet upset is different, especially if it has been in part caused by something malfunctioning, be it software, hardware or wetware.