Causal Inference

Overview

In this course, you will learn how to uncover cause-and-effect relationships in observational data—an essential skill for driving business decisions, policy evaluations, and scientific insights. You’ll explore a powerful suite of causal inference methods designed for time-series and panel data, including Interrupted Time Series, Difference-in-Differences, Event Study, Synthetic Control, and Regression Discontinuity models. Each lesson features hands-on Python exercises to build your technical fluency, and you’ll apply what you’ve learned in a final project that demonstrates your ability to estimate and validate causal effects. By the end of the course, you’ll be equipped to translate complex observational data into actionable insights that support evidence-based decision-making.

Syllabus

Foundations of Causal Inference and Interrupted Time Series

Explore causal inference, differentiate causation from correlation, design DAGs, and use Interrupted Time Series to estimate treatment effects when experiments are not feasible.

Difference-in-Differences and Validating Causal Assumptions

Estimate treatment effects using Difference-in-Differences, validate model assumptions with pre-trend checks, and strengthen causal claims through placebo testing.

Event Studies and Time-Varying Causal Effects

Explore event studies to analyze treatment impacts over time. Learn to model dynamic effects in staggered rollouts and interpret time-varying causal relationships effectively.

Synthetic Controls and Advanced Bayesian Alternatives

Learn Synthetic Control methods to build counterfactuals from control units. Implement in Python, then explore advanced alternatives like BSTS in R.

Regression Discontinuity and Communicating Causal Results

Use Regression Discontinuity for threshold-based treatments. Apply sensitivity tests and learn to communicate non-experimental findings with clarity, rigor, and stakeholder relevance.

Touchdowns and Trendlines: Estimating the Effect of Super Bowl Ads

Estimate the impact of a large sporting event ad campaign on branded search interest for a mid-sized company. Apply causal inference methods, validate assumptions, and select the best model.