Teaching

Causal Inference with Directed Acyclic Graphs

This workshop provides an introduction to causal inference with directed acyclic graphs (DAG). DAGs combine techniques from graph theory with statistical probability concepts and offer powerful tools for causal reasoning. Originally developed in the computer science and artificial intelligence field, the approach gets more and more traction in other disciplines too, such as epidemiology, sociology, finance, and economics. DAGs allow to check the identifiably of causal effects from observed data and lay bare the assumptions necessary for causal inference by relying on intuitive graphical criteria, which do not require a lot of algebra. As an encompassing framework for causal inference, DAGs also help to understand the identification arguments behind other popular estimation techniques such as regression, matching, instrumental variables and selection models. This workshop has no particular prerequisites. However, a good working knowledge in econometrics and probability theory is advantageous. (Slides, R code)