Instructor: Youjin Lee
Textbook: Hernán, M.A. and Robins, J.M. (2020). Causal Inference: What If. Chapman and Hall/CRC. ISBN 978-1-420-07617-2.
Description: Many studies in public health and social science confuse association with causation without careful examination of the conditions under which they can (partially) be understood causally. This can lead to ineffective—and sometimes harmful—public policies. This course provided guidance on when we can connect what we observe in the real world to causality and how we can read and analyze causal effects from empirical evidence. To that end, the course helped develop causal reasoning, providing a methodological background of causal inference that may be used to perform valid practices as a public health researcher. The course sought to foster an ability to separate causal problems from statistical problems, as well as to identify the assumptions required to identify causal effects and estimate them from observational data in diverse settings. The course first examined the differences between randomized designs and observations studies, as well as some challenges in causal effect estimation with observational studies from a missing data perspective. It then went over useful causal inference methods for evaluating policy effects from different types of empirical evidence and various advanced topics, including mediation analysis and time-varying treatments.
Most everyday decisions and public policies are based on questions about the potential effect of a certain cause. Does working from home due to the pandemic cause increase in the productivity of students? Do gun control laws cause more or fewer crimes? Answering such questions requires causal reasoning, which is often not achieved by naïve comparison of observed phenomena—i.e., comparing outcomes before and after the pandemic without taking into account other factors that could affect them.
Assignments:
Fall ‘22