rdd-design

What problem does it solve?

This Skill helps you estimate causal effects from observational data using Regression Discontinuity Design (RDD) by leveraging a cutoff in a running variable to approximate a randomized experiment.

Core Features & Use Cases

• Sharp RDD estimation: Compute the local treatment effect at the cutoff using bandwidth-weighted local polynomial regression, including standard errors, confidence intervals, and interpretation.
• Fuzzy RDD (local IV) estimation: Estimate LATE by combining reduced-form and first-stage jumps at the cutoff, with uncertainty via a delta-method standard error.
• Validity checks and robustness: Run McCrary-style manipulation (density continuity) tests, test covariate balance via separate RDDs on predetermined covariates, and assess bandwidth sensitivity to gauge estimator stability.
• Visualization: Produce cutoff plots with binned means and local fits for transparent reporting.

Quick Start

Use the RDD skill to estimate the treatment effect at a known threshold from your outcome array y, running variable array x, and cutoff value c by asking your agent to run sharp and fuzzy RDD plus manipulation and bandwidth sensitivity checks on your dataset.