It never fails to amaze me how ridiculously trusting people are of the results of using a TRAINING simulator to investigate the behaviour of an aircraft operated outside of the training envelope.
I agree. The sim may be a great tool for researching things like how a pilot will recognize and respond to an unusual attitude but using it to predict what the plane will actually do at unusual attitudes and airspeeds is a bit of a stretch. Still, sim results are often included in accident reports with little caveat.