Statistical Methods for Causal Inference

Substantive questions in empirical scientific and policy research are often causal. This class will introduce students to both statistical theory and practice of causal inference. As theoretical frameworks, we will discuss potential outcomes, causal graphs, randomization and model-based inference, causal mediation, and sufficient component causes. We will cover various methodological tools including randomized experiments, matching, inverse probability weighting, instrumental variable approaches, dynamic causal models, sensitivity analysis, statistical methods for mediation and interaction. We will analyze the strengths and weaknesses of these methods. The course will draw upon examples from social sciences, public health, and other disciplines. The instructor will illustrate application of the approaches using R/SAS/STATA software. Students will be evaluated and will deepen the understanding of the statistical principles underlying the approaches as well as their application in homework assignments, a take home midterm, and final take home practicum.