Macroeconomists often can't run experiments, and some kinds of "macro-sociological" theories can't be tested either. That puts them in a fundamentally different epistemological class.