External validity: From do-calculus to transportability across populations 论文

摘要

Abstract. The generalizability of empirical findings to new environ-ments, settings or populations, often called “external validity, ” is es-sential in most scientific explorations. This paper treats a particular problem of generalizability, called “transportability”, defined as a li-cense 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 ” for ex-pressing knowledge about differences and commonalities between pop-ulations of interest and, using this representation, we reduce questions of transportability to symbolic derivations in the do-calculus. This re-duction yields graph-based procedures for deciding, prior to observing any data, whether causal effects in the target population can be in-ferred from experimental findings in the study population. When the answer is affirmative, the procedures identify what experimental and observational findings need be obtained from the two populations, and how they can be combined to ensure bias-free transport. Key words and phrases: experimental design, generalizability, causal effects, external validity.