Regression Discontinuity
On the Validity of the Regression Discontinuity Design for Estimating Electoral Effects: New Evidence from Over 40,000 Close Races
In Eggers et al. (2015), the regression discontinuity (RD) design is evaluated as a method for identifying electoral effects. RD is effective only when relevant actors cannot exert precise control over election outcomes. While prior research suggests such control may exist in close U.S. House races, this study examines whether similar patterns appear across other electoral contexts, including state and local races in the U.S. and elections in nine other countries. No evidence of systematic incumbent advantage is found in these cases, challenging earlier claims and reinforcing the validity of RD in a wide range of electoral settings. The study also provides best practices for RD researchers.
Assessing the External Validity of Election RD Estimates: An Investigation of the Incumbency Advantage
In Hainmueller, Hall, and Snyder Jr (2015), the regression discontinuity (RD) design is employed to estimate the incumbency advantage, providing local estimates for tightly contested elections. However, these estimates are limited to hypothetical 50-50 races, leaving uncertainty about their applicability in less competitive districts. This study introduces a novel method that expands the scope of RD estimates to elections where the winning candidate secured up to 57.5% of the two-party vote. The findings reveal that the incumbency advantage is consistent across both competitive and less competitive districts, broadening the external validity of RD estimates.
