Studies of nuclear proliferation share five serious problems. First, nuclear programs' initiation and completion dates are ambiguous and difficult to code, but findings are rarely subjected to sufficient robustness tests using alternative codings. Second, independent variables overlook important...