Joseph Weinberg’s piece highlights an important substantive and methodological question: how to analyze, theoretically and methodologically, differences in national policy autonomy among countries and across policy areas in the era of globalization or regional integration.  EU membership constrains the policy autonomy of member states, which can change the relationship between the explanatory variables of interest and the outcome variable.  As Weinberg argues, “While a particular set of independent variables may explain outcomes in sovereign countries, those same variables would have little explanatory power where decisions are made by a supranational body” (5). We agree wholeheartedly that, if membership in a supranational institution constrains certain policy outcomes, then researchers should account for that in their theoretical and empirical models.  We disagree, however, on the solution. In particular, we show how multi-level models have important advantages in modelling this phenomenon, compared to the split-sample regressions in his piece.