A number of scholars have highlighted the role of employers in shaping the development of the welfare state. Yet the results of this research have often been ambiguous or disputed because of insufficient attention to theoretical, conceptual, and methodological problems in the study of political influence. This article considers three of these problems in turn: the failure to distinguish and investigate multiple mechanisms of exercising influence, the misspecification of preferences, and the inference of influence from ex post correlation between actor (...) preferences and outcomes. We demonstrate the importance of each through a reexamination of the early development of the American welfare state. The striking feature we suggest is neither business dominance nor weakness but marked variation in influence over time and across institutional settings. (shrink)