Nonetheless, you need careful statistical analysis of critically assessed data. And, ideally, causal models to guide the choice of which statistics to look at. Single cases don't refute a theory, whether the theory is anthropogenic global warming or "more guns, less crime." In Aristotle's terms, they're rhetoric (based on enthymeme and striking examples) rather than demonstration (based on syllogisms and inductive generalizations).

Note that I am not attempting to argue for or against either theory in saying this. My concern is only with method.