New Developments in Regression Discontinuity Designs: Covariates Adjustment and Coverage Optimal Inference
Year: 2024
Type: report
Abstract: This project proposes a generic and versatile covariate adjustment method that directly incorporates covariate balancing conditions for the regression discontinuity (RD) design and related natural experiment settings. The new empirical balancing method reweights the standard local polynomial estimator using balancing weights that minimize a specified distance (e.g., Kullback-Leibler) from uniform weights while satisfying covariate balance constraints in finite samples. The method applies to RD-related settings where traditional regression-based adjustment cannot be naturally extended or requires stronger conditions for consistency. It also improves efficiency by reducing asymptotic variance, being equivalent to an empirical likelihood framework with over-identifying moment restrictions from covariate balance.
Authors: 余 征飛, 0000003835, 40774758
Institution: University of Tsukuba
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
SDG: Reduced inequalities
Open Access status: green
OpenAlex ID: W7208106399