The Case for Using Statistics to Study Nuclear Security