New Methods & Techniques

Clear, accessible explanations of new research, methods, and tools in causal inference and experimentation. We break down complex ideas without losing rigor.

Honest DiD: Sensitivity Analysis for Pre-Trend Violations Rambachan and Roth (2023) and the HonestDiD Package

Read more →

Two-Way Fixed Effects with Heterogeneous Treatment Effects: The de Chaisemartin-D'Haultfoeuille (2020) Critique and the DID_M Estimator

Read more →

Synthetic Difference-in-Differences: Arkhangelsky et al. (2021)

Read more →

Regression Discontinuity Design: Sharp, Fuzzy, and the Calonico-Cattaneo-Titiunik Bandwidth

Read more →

Double Machine Learning: Debiased Estimation with High-Dimensional Controls

Read more →

Callaway–Sant’Anna DiD: Staggered Adoption and Group-Time ATTs

Read more →

Synthetic Difference-in-Differences: Arkhangelsky et al. (2021)

Read more →

Regression Discontinuity Design: Sharp, Fuzzy, and the CCT Bandwidth

Read more →

Honest DiD: Sensitivity Analysis for Pre-Trend Violations Rambachan and Roth (2023) and the HonestDiD Package

Two-Way Fixed Effects with Heterogeneous Treatment Effects: The de Chaisemartin-D'Haultfoeuille (2020) Critique and the DID_M Estimator

Synthetic Difference-in-Differences: Arkhangelsky et al. (2021)

Regression Discontinuity Design: Sharp, Fuzzy, and the Calonico-Cattaneo-Titiunik Bandwidth

Synthetic Difference-in-Differences: Arkhangelsky et al. (2021)

A hybrid of synthetic control and DiD that inherits the strengths of both. We unpack the SDiD estimator and what makes it tick.

Regression Discontinuity Design: Sharp, Fuzzy, and the CCT Bandwidth

How do you turn an arbitrary threshold into a credible experiment? A deep dive into RDD—from the basic intuition to optimal bandwidth selection.

Double Machine Learning: Debiased Estimation with High-Dimensional Controls

When your controls are high-dimensional, standard OLS breaks down. Double Machine Learning uses cross-fitting and Neyman orthogonality to recover valid causal estimates while letting machine learning handle the nuisance functions.

Callaway–Sant’Anna DiD: Staggered Adoption and Group-Time ATTs

When units adopt treatment at different times, the standard two-way fixed effects estimator produces biased results. Callaway and Sant'Anna's group-time ATT framework is now the standard fix—explained from first principles.