Much Ado About Nothing: RDD and the Incumbency Advantage

View Researcher's Other Codes

Disclaimer: The provided code links for this paper are external links. Science Nest has no responsibility for the accuracy, legality or content of these links. Also, by downloading this code(s), you agree to comply with the terms of use as set out by the author(s) of the code(s).

Authors Robert S. Erikson, Kelly Rader
Journal/Conference Name POLITICAL ANALYSIS
Paper Category
Paper Abstract An influential paper by Caughey and Sekhon (2011a) suggests that the outcomes of very close US House elections in the postwar eramay not be as-if random, thus calling into question this application of regression discontinuity for causal inference. We show that while incumbent party candidates are more likely to win close House elections, those who win are no di erent on observable characteristics from those who lose. Further, all di erences in observable characteristics between barely winning Democrats and barely winning Republicans vanish conditional on which party is the incumbent. Any source of a special incumbent party advantage in close elections must be due to variables that cannot be observed. This finding supports the conclusion of Eggers et al. (2015) that Caughey and Sekhon’s discovery of lopsided wins by incumbents in close races is a mere statistical fluke. Regression discontinuity (RD) is a useful tool for causal inference in natural settings. The idea is simple. When a treatment is assigned based on a threshold, cases barely below the threshold are approximately identical in all respects to those barely on the other side of the threshold. Thus, untreated cases near the threshold can serve as counterfactuals for treated cases near the threshold, and researchers can estimate a local average treatment e ect with minimal assumptions. RD has an important place in the analysis of two-candidate elections since very close elections on each side of the 50 percent vote threshold are similar in all aspects except for a virtual coin flip for who wins. If its assumptions are met, RD allows causal inference regarding the consequences of a party winning versus losing on its possible incumbency advantage in later elections (e.g., Lee 2008; Erikson and Titiunik 2015) or outcomes of elections for other o ices (Folke and Snyder 2012; Erikson, Folke, and Snyder 2015). Yet the validity of RD for causal inference has come under challenge. An influential and awardwinning article,1 Caughey and Sekhon (2011a), sounds a warning about the validity of the RD design when applied to US House elections in the postwar era. This article (herea er CS) notes an interesting oddity in the distribution of outcomes for very close elections—elections within a margin of 0.5 percentage points of vote for the incumbent party. The purpose of this letter is to show that, contrary to the impression onemight get from a first readingof CS, there arenoobservabledi erences in covariates betweenbarelywinningandbarely losing incumbent party candidates that could account for the lopsided incumbent party win rate. Additionally, any di erences in observable variables between elections barely won by Democrats and those barely lost by Democrats vanish conditional on which party is the incumbent party.2 This fact rules out incumbent party prowess at exploiting variables that are measured here as the source of the CS incumbent party win rate oddity. Authors’ note: The authors would like to thank three anonymous reviewers for their helpful comments. All replication materials are available at the Political Analysis Dataverse (Erikson and Rader 2016, doi:10.7910/DVN/567RS6). 1 Caughey and Sekhon (2011a) was the recipient of the 2012 Warren Miller prize for the best paper in Political Analysis published in the previous year. 2 Elections barely won by Democrats by a two-party vote are, by definition, elections that are barely lost by Republicans, and vice versa.
Date of publication 2017
Code Programming Language R

Copyright Researcher 2022