A generalization is not the same as a universal claim, which can be disproved with a single exception.

A generalization is about the shape of some statistical cloud. We have to show that it's not shaped that way to disprove it. So find enough exceptions and we're there. But how many are enough?

It's usually easiest (imho) to disprove these by showing a different generalization that is true, but conflicts with the given one.HN