A Case for Regression Discontinuity Design in Impact Evaluations

• In Event: 18.025 - Innovations in Educational Evaluation Methods

Abstract

Using an RD design provides statistically robust estimates while allowing researchers a different causal estimation tool to be used in educational environments where an RCT may not be feasible. Results from External Evaluation of the i3 Scale-Up of Reading Recovery show that impact estimates were remarkably similar between a randomized control trial (RCT) and a regression discontinuity (RD) design. This paper discusses RD design advantages including congruence with school practices for assigning interventions, alleviation of ethical concerns, flexible and adaptable design methodology, and the ability to monitor the effectiveness of various types of interventions in the classroom. This presents opportunities for schools, districts, and states to be more proactive in evaluating the causal impacts of interventions for low-achieving students.

Authors

• Henry May, University of Delaware

• Akisha Jones, University of Delaware

• Aly Blakeney, University of Delaware