I am rereading your question as: "When do circumstances become different enough that evidence from one situation doesn't apply to the other situation?" and this sounds like the fundamental question of reference class tennis, which I believe does not have a good answer.