And that's an axiomatic error, not statistical. They hold it as an axiom that they should think in terms of percentage instead of individual outcomes. Statistics can't fix it, because you can't question axioms with mathematics alone.