Welcome to the Causal Inference with R – Experiments, the 2nd of 7 courses on causal inference concepts and methods created by Duke University with support from eBay, Inc. Designed to teach you about concepts, methods, and how to code in R with realistic data, this course focuses on experiment design, working with data from controlled and natural experiments, and dealing with noncompliance. We’ll stay away from dense statistical math and focus instead on higher level concepts that data scientists need to always consider when examining and making inferences about experimental data. The course instructors and creators are Dr. Matt Masten (Duke University), James Speckart (Duke), Brian Aronson (Duke), Dr. Tyler Ransom (University of Oklahoma), Dr. Bentley Coffey (University of South Carolina), and Alexandra Cooper (Duke).
We introduce how experiments help us find causal connections, and take a quick look at controlled experiments.
Randomized Experiments & Statistical Inference
These videos and exercises introduce randomized experiments, and explain why we need statistical inference in experiments.
Practice with Published Experiments - The Oregon Health Insurance Experiment
These videos and exercises offer practice in skills used in analyzing experimental data.
Common Issues in Experiments
These videos and exercises discuss common issues in carrying out experiments.
Dealing with Noncompliance in Experiments
These videos and exercises discuss the issue of noncompliance in experiments.
Long-term Average Treatment Effects
These videos and exercises discuss how to compute and interpret long-term causal effects.
These videos and exercises introduce natural experiments, sometimes called quasi-experiments.
Bounds Analysis & Putting It All Together
This chapter discusses how to form bounds on causal effects in the presence of experimental noncompliance, and we wrap up the course with a bigger exercise to review what we have learned.