As the first post notes, the probability maths behind the assumptions behind this training is fundamentally flawed.
There are numerous multiple failure cases which are not considered for certification as they are demonstrated to be extremely improbable. Say rudder jam plus single engine failure, or multiple independent cause engine failures.
But once some of the events have happened, it is no longer extremely improbable that the remaining events will happen; all that is left is the residual probability of the remaining circumstances.
So if engine failure is 10e-5 and throttle jam is 10e-5, then combined failure is 10e-10 prior to departure - a perfectly acceptable probability for an assumed catastrophic top event.
But once one or other has occurred, all that is left is the residual 10e-5. No longer adequate for the top event, indicating that immediate action to mitigate the elevated risk is prudent.
Never mind the fact that you might be looking at a common cause scenario: what if the lever jammed due to FOD? How likely is it that there was FOD close to the second lever also, in which case there may be increased risk of a dual jam.