(iPoster) Polling Place Floors and Voter Turnout: Evidence from South Korea

• In Event: Poster Sessions: Law and Courts & Elections and Voter Behavior
  In Poster Session: Poster Session: Elections and Voter Behavior

Abstract

Voting is a core democratic act, yet it is carried out through concrete administrative procedures and physical spaces that impose nontrivial burdens on citizens. While a large literature shows that turnout is sensitive to external voting costs—such as distance, waiting time, and procedural complexity—far less is known about how the internal spatial organization of polling places shapes participation. This paper examines one of the most basic yet overlooked internal features: the floor level at which voting takes place within a building.

Using comprehensive polling-station–level data from all South Korean legislative and presidential elections since 2012, I analyze whether turnout differs systematically between polling places located on the ground floor and those placed on higher floors. South Korea offers an ideal institutional context for this analysis. Election law mandates that polling stations be located on the first floor whenever possible, with higher floors used only when ground-floor space is unavailable. As a result, floor-level variation is limited, administratively generated, and largely detached from political strategy. Moreover, polling place locations—including building floors—are finalized and communicated to voters well in advance of election day, allowing floor level to function as a predictable and salient condition rather than an unexpected obstacle.

The empirical strategy exploits within–neighborhood variation by comparing multiple polling stations operating simultaneously within the same town (eup–myeon–dong) and election. Two-way and interacted fixed-effects models isolate comparisons between polling stations exposed to identical political, socioeconomic, and administrative environments, differing only in their internal spatial placement. Across increasingly restrictive specifications, polling stations located above the first floor consistently exhibit lower turnout. In the most conservative model—conditioning on town-by-election fixed effects and polling-station characteristics—non–ground-floor locations are associated with a 0.15–0.40 percentage point reduction in turnout.

A series of robustness checks reinforces this finding. First, exploiting cases in which the same building served as a polling place across elections but with different floor assignments yields similar effects, ruling out time-invariant building characteristics. Second, disaggregating floors reveals a clear non-linearity: turnout declines modestly on the second floor but drops more sharply on the third floor or above, consistent with a mechanism of increasing perceived burden rather than binary accessibility.

Although the estimated effects are small at the individual polling-station level, they are substantively meaningful when aggregated across thousands of locations and repeated elections. The findings suggest that even minor features of internal spatial design can subtly but systematically shape political participation. More broadly, the paper contributes to urban and electoral politics by demonstrating that the micro-geography of public service delivery—inside buildings, not just across neighborhoods—matters for democratic engagement.

Author

• Yesung Jeon, Yonsei University Graduate School