New Methodological Developments for Inference in the Regression-Discontinuity Design

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

This award funds the development of new methods for the statistical analysis of social science data. The project focuses on 'regression discontinuity' (RD) models, which are very widely used in practice. This kind of statistical model is used to estimate the effect of a treatment on an outcome when unobserved factors may threaten the validity of the statistical analysis. While RD models are widely used to analyze data, many important methodological and statistical features of the models are not well understood. This project develops novel methods for the analysis of RD models and applies the results to concrete empirical problems in Economics and other social sciences. The research develops new methodological and practical tools for estimation, inference and falsification of RD designs. This includes a method that accounts for the fact that RD estimators are sensitive to the choice of bandwidth, a method that is valid in finite samples, the analysis of RDs with multiple cutoffs and the development of new tools for spatial RD designs. Broader impacts of this research will come from the use of these new methods to provide better insight for economic policy.