LAUSR

ABSTRACT The study of nuclear proliferation has recently undergone a renaissance, one element of which is the widespread application of a particular statistical technique known as survival modeling. But survival models are often misapplied and this misapplication has consequences for our understanding of nuclear proliferation. Scholars of nuclear proliferation consistently fail to account for two methodological challenges present in survival modeling: selecting an appropriate distribution and meeting the proportional hazards assumption. We show by example how accounting for these challenges alters key results and has a bearing on current debates in the field. We endorse the continued, judicious application of methodologically sound survival models and conclude by highlighting three possible theoretical and methodological paths forward.