Optimal Design for Regression Discontinuity Studies With Clustering

• In Event: Statistics in the Field: Innovative Applications and Investigations

Abstract

An important concern when planning research studies in populations with a hierarchical structure is to determine the optimal sample size at each level. The goal is to minimize the variance of the estimator of the treatment effect subject to a budget constraint. The optimal design problem has been solved for randomized experiments with two levels of clustering (Raudenbush, 1997; Raudenbush and Liu, 2000). The current work extends this earlier work by presenting results for the optimal design of regression discontinuity studies with two levels of clustering.

Authors

• Chris H Rhoads, University of Connecticut

• Charles Dye, University of Connecticut