Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 107 - SPEED: Statistical Methods, Computing, and Applications Part 1
Type: Contributed
Date/Time: Monday, August 8, 2022 : 8:30 AM to 10:20 PM
Sponsor: Section on Nonparametric Statistics
Abstract #322424
Title: A Comparison of Regression Discontinuity Effect Estimation for Small Samples
Author(s): Daryl Swartzentruber* and Eloise E Kaizar
Companies: The Ohio State University and The Ohio State University
Keywords: Regression ; discontinuity; nonparametric; education; small ; sample
Abstract:

Regression discontinuity (RD) designs are popular quasi-experimental studies in which treatment assignment depends on whether the value of a running variable exceeds a cutoff. Much of the current research in RD designs comes from the field of economics, where large sample sizes are common. Typical treatment effect estimation in these settings involves fitting nonparametric regressions on either side of the cutoff and subtracting the limit of the mean outcome from the left from the corresponding limit to the right. RD designs are also used in educational applications such as examining the effect of supports for schools or students that fail to meet certain benchmarks. In such applications sample sizes can be relatively small or there may be sparsity around the cutoff. We propose a size metric that better captures the amount of information available for RD estimation and present results from a simulation study comparing the small sample performance of popular RD estimation techniques. We also apply these techniques to a data set on school accountability scores in the state of Indiana.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program