140.664.01 CAUSAL INFERENCE IN MEDICINE AND PUBLIC HEALTH

Presents an overview of methods for estimating causal effects: how to answer the question of “What is the effect of A on B?” Includes discussion of randomized designs, but with more emphasis on alternative designs for when randomization is infeasible: matching methods, propensity scores, regression discontinuity, and instrumental variables. Methods are motivated by examples from the health sciences, particularly mental health and community or school-level interventions.

Upon successful completion of this course, students will 1) Understand causal problems as potential interventions, through the framework of potential outcomes and assignment mechanisms, 2) Understand the spectrum of designs for both randomized and non-randomized studies, 3) Identify the situations for which non-randomized designs are most appropriate, and 4) Understand and be able to apply methods for estimating causal effects, including propensity score techniques, instrumental variables (“encouragement designs”), and regression discontinuity. The focus will be on learning how to critically review research that claims to estimate causal effects with non-experimental data. Students will also understand complications encountered in causal studies, including missing data, noncompliance, and hidden bias.