The regression discontinuity (RD) design identifies electoral effects but requires that relevant actors lack precise control over outcomes. Recent studies suggested this might not hold in large U.S. House elections, where incumbents often win close races. This paper examines whether such patterns exist elsewhere using data from the U.S. House across different periods and types of elections (statewide, state legislative, mayoral) plus national/local contests in nine other countries. No evidence was found for similar manipulation patterns in these diverse settings. The study also challenges explanations for incumbent success in close races within the U.S. system. Overall, this reinforces that RD assumptions are generally met across electoral contexts and provides guidance for researchers.
| On the Validity of the Regression Discontinuity Design for Estimating Electoral Effects: New Evidence from over 40,000 Close Races was authored by Andrew C. Eggers, Anthony Fowler, Jens Hainmueller, Andrew B. Hall and James M. Snyder, Jr.. It was published by Wiley in AJPS in 2015. |
