Introduction to Causal Inference
This course presents a general framework for causal inference:
-
clear statement of the scientific question,
-
definition of the causal model and parameter of interest,
-
assessment of identifiability - that is linking the causal effect to a parameter estimable from the observed data distribution,
-
choice and implementation of estimators including parametric and semi-parametric methods,
-
interpretation of findings.
The estimation methods include G-computation, inverse probability of weighting (IPW), and targeted maximum likelihood estimation (TMLE) with Super Learner.
-
Developed with Dr. Maya Petersen of UC Berkeley
-
Winner of 2014 ASA’s Causality in Statistics Education Award - “individual or team that does the most to enhance the teaching and learning of causal inference in introductory statistics courses”
-
Course materials available at www.ucbbiostat.com

G-computation methods

Inverse probability weighting

Super Learner (ensemble method)

Targeted maximum likelihood estimation
(TMLE)