Philosophical accounts of causation have traditionally been framed as attempts to analyze the concept of a cause. In recent years, however, a number of philosophers have proposed instead that causation be empirically reduced to some relation uncovered by the natural sciences: e.g., a relation of energy transfer. This paper argues that the project of empirical analysis lacks a clearly defined methodology, leaving it uncertain how such views are to be evaluated. It proposes several possible accounts of empirical analysis and argues that the most promising approach would treat it as a contingent identity discovered by identifying the relation (or relations) that most nearly approximate the inferential role of causal concepts in a psychological theory of causal judgment.