DS-UA 9201 Causal Inference

Term: Fall 2022
Instructor: Dr. Judith Abécassis
Level: Undergraduate

Topics

Introduction to Causation, Identifiability assumptions;

Randomized experiments, Treatment effect heterogeneity, Stratification strategies;

Observational experiments, Causal graphs, Weighting and Matching;

Regression-based methods: From coefficient estimation to g-computation, Doubly robust estimation;

Uncertainty in causal estimation: Asymptotic theoretical variance estimation, Empirical approaches to variance estimation;

Variable selection for adjustment, Instrumental variables;

Regression Discontinuity Designs, Difference in differences;

Sensitivity analysis, Causal mechanisms: Causal mediation analysis, Dynamic treatment regimes; Causal discovery;

Description

Modern statistics and machine learning focus on discovering significant associations in the data and leverage them to provide reliable predictions. However, in many areas (medical applications, or to inform political decision for instance), the practitioner is interested in a more complex question which is to estimate the effect of a particular action on an outcome of interest. This course will focus on how to reason about causality and make causal determinations using experimental or observational data. It will begin by introducing the counterfactual framework of causal inference and then discuss a variety of estimators to make inferences about causal relationships from the data. For each approach, we will discuss the necessary assumptions that a researcher needs to make, how to assess whether these assumptions are reasonable, and finally how to interpret the quantity being estimated. Finally, we will consider more complex settings to illustrate the variety of causal inference potential applications. The course will focus on the theoretical concepts of causal inference, but an important part of the materials and exercises will be dedicated to practical application with real datasets. The course will introduce scientific programming using the Python programming language.