Regression Discontinuity Designs and Potential Outcomes

• Activity Number: 397
• Type: Invited
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): guido imbens*+
• Companies: Stanford University
• Keywords: regression discontinuity designs ; potential outcomes ; causality

Abstract

I discuss regression discontinuity designs from a potential outcome perspective. Looking at both sharp and fuzzy regression discontinuity designs we examine the critical assumptions for identifying causal effects and relate them to approaches based on unconfoundedness assumptions. I illustrate the methods using a data from a remedial summer program for high school students where only those students who scored below a threshold on a test were required to participate. Some students with score higher than the threshold participated in the program despite not being required to do so, and some students who score below the threshold nevertheless did not participate.