Statistics 590C - Intro to Causal Inference

Seeking answers to questions of causality is a fundamental part of the scientific process and the advancement of human knowledge. Answers to causal questions are imperative to supporting decision-making in areas such as healthcare and public policy. This course provides an introduction to causal inference in statistics. We will introduce the potential outcomes framework to causal modeling, and use it to study core causal models including randomized experiments, backdoor adjustment, instrumental variables, difference-in-difference, regression discontinuity, and mediation analysis. Directed Acyclic Graphs will be introduced as an alternative approach to causal reasoning and as a tool for assessing conditional independence assumptions. In each model, we will study the causal and observed data structures, the causal estimand(s) of interest, the causal identification conditions and the associated graphical model, the identification result, statistical inference for the identified parameter, and R code implementing statistical inference. We will use case studies from the real scientific literature to illustrate each model. In the final project, students will apply the techniques learned in class to conduct causal analysis for a scientific question of their choosing and write up their results.

STATISTC 515, 516, and 525