Quarterly Journal of Political Science > Vol 11 > Issue 1

The Mythical Swing Voter

Andrew Gelman, Columbia University, USA, gelman@stat.columbia.edu , Sharad Goel, Stanford University, USA, scgoel@stanford.edu , Douglas Rivers, Stanford University, USA, rivers@stanford.edu , David Rothschild, Microsoft Research, USA, davidmr@microsoft.com
Suggested Citation
Andrew Gelman, Sharad Goel, Douglas Rivers and David Rothschild (2016), "The Mythical Swing Voter", Quarterly Journal of Political Science: Vol. 11: No. 1, pp 103-130. http://dx.doi.org/10.1561/100.00015031

Publication Date: 18 Apr 2016
© 2016 A. Gelman, S. Goel, D. Rivers and D. Rothschild
Elections,  Public opinion,  Bayesian models
Electionsswing votersmultilevel regression and poststratification


Download article
In this article:
1. Introduction 
2. Evidence from Empirical Studies 
3. Discussion 
A. Methods and Materials 


Most surveys conducted during the 2012 U.S. presidential campaign showed large swings in support for the Democratic and Republican candidates, especially before and after the first presidential debate. Using a combination of traditional cross-sectional surveys, a unique panel survey (in terms of scale, frequency, and source), and a high response rate panel, we find that daily sample composition varied more in response to campaign events than did vote intentions. Multilevel regression and post-stratification (MRP) is used to correct for this selection bias. Demographic post-stratification, similar to that used in most academic and media polls, is inadequate, but the addition of attitudinal variables (party identification, ideological self-placement, and past vote) appears to make selection ignorable in our data. We conclude that vote swings in 2012 were mostly sample artifacts and that real swings were quite small. While this account is at odds with most contemporaneous analyses, it better corresponds with our understanding of partisan polarization in modern American politics.