Resources

In this paper, authors match a high-quality, random sample of the U.S. population to multiple lists revealing that at least 11% of the adult citizenry is not on a voter list. An additional 12% is mislisted (i.e., not living at their recorded address).

In this paper, authors use snapshots of voter registration files (VRF) over time and machine learning models to test the effectiveness of unsupervised anomaly detection methods in detecting VRF modifications. They find that statistical models comparing administrative districts within a short time span and non-negative matrix factorization are most effective for surfacing anomalous events for review.

Even before the 2020 election, this reseach finds that voter turnout across the states is consistently higher in every general election over the past decade in states with greater shares of overall ballots cast by mail. Drawing on turnout data from the 2012-2020 Current Population Survey (CPS) and the Cooperative Election Study (CES), authors find states with greater usage of mail voting experience higher overall voter turnout.

Research finds adopting VBM increases turnout because it reduces the physical costs of voting for all voters and mitigates the information costs of voting conditional on the types of voters and salience of elections.

All mail voting in Colorado had a positive overall turnout effect of approximately 8 percentage points—translating into an additional 900,000 ballots being cast between 2014 and 2018.

Focusing on natural experiments in Texas and Indiana, this research finds that 65-year-olds turned out at nearly the same rate as 64-year-olds, despite the fact that only 65-year-olds could vote absentee without an excuse in these states. Being just old enough to vote no-excuse absentee did not substantially increase Democratic turnout relative to Republican turnout.

Philadelphia officials randomly sent 46,960 Philadelphia registrants postcards encouraging them to apply to vote by mail in the lead-up to the June 2020 primary election. While the intervention increased the likelihood a registrant cast a mail ballot by 0.4 percentage points —or 3%—many of these additional mail ballots counted only because a last-minute policy intervention allowed most mail ballots postmarked by Election Day to count.

The research finds no evidence that voting by mail increases the risk of voter fraud overall. Between 2016 and 2019, RBM (VBM) states reported similar fraud rates to non-RBM (non-VBM) states. Moreover, it is estimated Washington would have reported 73 more cases of fraud between 2011 and 2019 had it not introduced its VBM law.

Using individual-level voter data from a 2020 Washington State election, the research shows that voters are more likely to use the nearest drop box to their residence relative to other drop boxes. In Washington’s 2020 August primary, 52% of drop box voters in our data used their nearest drop box.

With Maine as a case study, the research shows that, in the past four general elections, over 10% of vote-by-mail ballots arrived at local elections offices either on Election Day itself or one day earlier. Moreover, of the vote-by-mail ballots most vulnerable to postal delivery disruptions, a greater share of them were cast by unaffiliated voters and Democrats than by Republicans.

This analysis of the Florida general elections of 2016, 2018, and 2020 shows that voters inexperienced with mail voting disproportionately submit ballots that end up rejected due to (1) late arrival at elections offices or (2) signature defects on return envelopes. Inexperienced mail voters are up to three times more likely to have their ballots rejected compared to experienced mail voters.

This analysis shows that the implementation of vote by mail causes a significant decrease in voter confidence in Washington and Colorado. However, this decrease appears to be temporary, disappearing after only a single election cycle.