Voting by Mail in the U.S.

The MIT  Election Data + Science Lab  is dedicated to applying scientific principles to how elections are studied and administered, with the aim of improving the democratic experience for all U.S. voters. Our post today was written by John Curiel, a research scientist at the Lab.

Expanding the opportunities Americans have to vote by mail (VBM) has been proposed as the most important policy response to the COVID-19 pandemic, in order to keep elections healthy and secure in 2020. The following article and story maps are part of the attempt to visualize trends of vote by mail within the United States.

The number of Americans voting by mail (VBM) has gradually expanded over the past three decades, accounting for just seven percent of votes cast in 1992 and growing to approximately 24 percent in the midterm election of 2018.

Originally, mail balloting was only allowed for specific reasons, generally reserved for when a voter was away from home or infirm on Election Day. In the 1960s, California passed a law allowing voters to request an absentee ballot for any reason. Since then, states have adopted a number of policies that have affected how many people vote by mail. These policies as classified by the  National Conference on State Legislatures , going from most restrictive to least, are:

(1) excuse required absentee voting,

(2) no excuse required absentee voting,

(3) permanent absentee voting lists, and

(4) universal vote-by-mail. 

To date, the states with the least restrictive access to vote by mail are still almost entirely in the western United States.

Most states still require voters to apply proactively for an absentee ballot by mail, although the relative ease of applying varies across the states. Just as the most expansive vote-by-mail regulations are largely found in the western United States, these states also have some of the highest rates of voting by mail in the country.  We can explore and analyze vote by mail patterns by employing results from the Elections Administration and Voting Survey (EAVS) from the 2016 Presidential election, with details on the survey reported at the end.

Washington State was the first to adopt the vote-by-mail model (which is also referred to as “vote–at-home”). The state did not adopt the model all at once, though. In the 1992 presidential election, Washington permitted no excuse vote by mail; 17.5 % of voters in that election cast their ballots by mail. After that, the rate of VBM usage steadily increased until 2006, when the state permitted counties to declare and implement universal VBM policies. Following that change, Washington state effectively reached 100 percent vote-by mail-rates, a jump of nearly 30 percentage points.

In addition to the variability between states when it comes to VBM, there are also policy differences within states themselves. Even when there is a state requirement for voters to request a mail ballot, the use of absentee ballots can vary significantly by local jurisdiction.

Voting by mail of course comes with challenges. Most fundamentally, some steps in the in-person voting process are necessarily handled remotely in vote-by-mail, and that can cause problems. The first issue that can come up is whether a voter’s application for a ballot is be received by the central voting office. Subsequent issues that can arise follow a similar vein: whether the ballot mailed by the election office is received by the voter, or whether, when the voter mails that ballot back, it is received by the office.

As a percentage of all ballots transmitted for 2016, permanent absentee ballots see the greatest unreturned rate at 26.4%, followed by universal VBM states at 22.3%, followed by no excuse required states at 14.3%, and finally no excuse required at approximately 8.64%. These results are subject to variance between and within states. It should be noted that in states that automatically mail ballots to all registered voters, most non-returns reflect people who don't wish to vote in the election. Thus, non-returns from these states must be considered separately from those in states that require voters to make some sort of request for a mail ballot.

The percentage of ballots that are not returned for counting (as a percent of ballots mailed out) is an important measure of the number of absentee ballot applications that do not result in an actual vote being counted. It’s unclear why so many ballots that get mailed out to voters are never returned; we don’t have enough academic research on the problem to know for sure. Nonetheless, in 2016 approximately 20 percent of ballots that were mailed out were not returned, excluding universal VBM states.

The next issue that can arise for VBM systems is whether a ballot is accepted for counting. Most states require the voter to sign a form on the outside of the mailing envelope; that signature is then compared with a signature on file. Approximately 0.96% of the ballots mailed back for counting in 2016 were rejected, mostly because of signature problems—either because the signatures did not match those on file, or were missing altogether.

Prior to 2020, most states had fewer than 10 percent of ballots cast by mail. The recommendation that mail balloting be expanded for 2020 puts pressure on the states with little mail-ballot usage in the past to revamp their administrative procedures to process millions of new votes by mail. The recent Wisconsin primary shows how  substantially demand  for voting by mail can change in this COVID-19 era. The next work to be done is identify how the demand for VBM changes as public health hazards arise to voting in person, along with the supply potential and steps that the nation, states, and local governments will need to take so as to provide the electoral infrastructure so as to meet demand.

All data provided on vote by mail statistics for the story map slideshows are from the 2016 Election Administration and Voter Survey ( EAVS ) data. The methodology on how the VBM statistics are calculated can be found in the Elections Performance Index (EPI)  methodology report . All EAVS data are acquired from self reports by states and their localities.

John Curiel is a research scientist at the MIT Election Data + Science Lab. For more updates from the Lab, follow us on  Twitter  and sign up for our  quarterly newsletter .