Whose Vote Is Lost by Mail? Evidence from Philadelphia in the 2020 General Election

Abstract

The 2020 elections raised the salience of procedural frictions that prevent people who intend to vote by mail from casting a counted ballot. Here, we focus on the incidence of such lost mail votes in Philadelphia County during the 2020 general election. There were heightened concerns about lost mail votes in that election because a judicial ruling issued just before the election stated that mail ballots returned outside of a secrecy envelope—commonly referred to as “naked ballots”—would not count. Our combination of observational and experimental analyses show who was at the greatest risk of casting a lost mail vote and how voter education efforts reduced their incidence. We find that voters from racial/ethnic minority backgrounds were more likely to return a naked ballot than White voters, partly due to language barriers. Younger voters were less likely to return a naked ballot than older voters, a finding which may be linked to widespread communication efforts seeking to educate citizens about how to successfully cast a mail ballot, often via social media. Our analysis of a randomized experiment exploring the effects of a postcard sent to almost 18,000 mail-ballot recipients further demonstrates how communication can reduce the risk of lost mail votes, especially by getting people to return their mail ballots more quickly.