Causal Transportability with Limited Experiments

Most scientific explorations are concerned with generalizing empirical findings to new environments, settings, or populations, a problem that the machine learning literature labeled "transfer learning." Our talk focuses on a particular type of generalizability, called “transportability”, defined as a license to transfer causal effects learned in experimental studies to a new population, in which only observational studies can be conducted. We introduce a formal representation called “selection diagrams”.