The Causal Inference Playbook: Advanced Methods Every Data Scientist Should Master

A deep dive into six advanced causal inference methods for data scientists who already know the basics. Using a job training program as a running case study, it covers: Doubly Robust Estimation (AIPW) for protection against model misspecification, Instrumental Variables (2SLS) for unmeasured confounding, Regression Discontinuity for quasi-experimental designs, modern Difference-in-Differences addressing staggered adoption problems (Callaway & Sant'Anna), Heterogeneous Treatment Effects via Causal Forests using econml, and Sensitivity Analysis using the Cinelli-Hazlett framework. Each method includes Python code and a discussion of assumptions and limitations. A decision flowchart helps practitioners choose the right method for their situation.