State and local officials across the country are making difficult decisions about how to enable citizens to vote without jeopardizing their health. One widely discussed approach involves allowing more people to vote by mail, lowering the risk of spreading the novel coronavirus.

Most Americans support that option for this November’s election. And for good reason: Those of us who study mail voting agree that it has little effect on election results because it has a marginal impact on turnout and doesn’t give either party an advantage.

But voting by mail increases the number of ballots that are rejected — and not counted in the final tally. And ballots from younger, minority and first-time voters are most likely to be thrown out. Here’s how we know.

Mail ballots are rejected at a much higher rate than in-person ballots.

According to the U.S. Election Assistance Commission, of the more than 140 million votes cast in the 2016 general election, 23.7 percent were via mail. Of the roughly 33.2 million mail ballots that election administrators received and tabulated, approximately 1 percent weren’t counted. Reasons for rejection include “the signature on the ballot not matching the signature on the state’s records,” “the ballot not having a signature,” a “problem with return envelope,” or “missing the deadline.” By contrast, a third fewer ballots cast in person were rejected in 2016.

Even in states where voters have had several years’ experience in voting by mail, such as in Washington, Oregon and Colorado, mail ballots get rejected. In the 2016 election, 0.81 percent of Colorado’s mail ballots were rejected; in Oregon, 0.86 percent; and in Washington, 0.90 percent. And that doesn’t include the hundreds of thousands of ballots mailed to voters that were returned as undeliverable.

Here’s how we did our research

To conduct our research, we looked at mail ballots cast in Georgia during the 2018 midterm elections. Georgia was once covered by Section 5 of the Voting Rights Act, which required the state to clear in advance any election rule changes with the federal government. The Supreme Court invalidated Section 5 in 2013 in Shelby v. Holder. Although Georgia has allowed anyone to vote by absentee ballot without an excuse since 2006, it is a different process from the three states that prepared for years to move to all-mail elections. Georgia does not typically have large numbers of mail voters.

In our study, we use Georgia’s statewide voter file, which has information on each individual voter’s race, age, gender and registration status. We linked the individual-level data with administrative data on mail ballots cast during the 2018 midterm elections, which has information on when a voter’s ballot was received by an election official and whether it was counted or not. In the 2018 midterms, about 266,000 registered voters in Georgia cast a mail ballot. Election officials rejected more than 7,000 of them, or nearly 3 percent, either because of an error on the return envelope or because they arrived after Election Day.

Ballots from younger voters, people of color and first-time voters were the most likely to be thrown out

A disproportionate number of younger voters, people of color and first-time voters had their mail ballots rejected for those two reasons. We account for a battery of individual-level details available in the Georgia voter file and mail ballot data, using a two-stage modeling approach. The first stage identifies who is more likely out of all of Georgia’s registered voters to cast a mail ballot. The second stage models who is more likely to cast a mail ballot that is rejected. This approach allows us to account for the fact that voters who choose to cast mail ballots may be different than those who choose to vote in person. Using Georgia’s mail ballot data, we are also able to find why the ballot was rejected.

Our findings are clear. The figure below illustrates, for example, that black voters across Georgia’s 159 counties are disproportionately more likely to have their ballots rejected than white voters. If white and black voters had their mail ballots rejected at the same rates, all the counties would align on the 45-degree line. They do not. There are several outliers, including Polk, Taylor, Atkinson and Glynn, the same county in which Ahmaud Arbery was fatally shot while running in a predominantly white neighborhood.

Our study shows that this pattern holds for other racial/ethnic minority groups, too, as well as first-time and young voters, especially those between the ages of 18 and 22.

Why are ballots rejected at such disparate rates?

It’s not entirely clear. On the one hand, there may be reasons some groups of voters are more likely to make mistakes when voting by mail. Many new voters — for example young and first-time registrants — may be unfamiliar with how to vote by mail, or not know how much postage is required or how far in advance to return the ballot to have it arrive on time. Mail voters don’t have the benefit of interacting face-to-face with poll workers who might be able to help them navigate any difficulties.

On the other hand, that can’t fully explain why racial and ethnic minorities are less successful at filling out and returning a mail ballot. But these differences do track with research on voting by mail in Florida, as well as other research showing that racial and ethnic minorities tend to receive lower-quality services from election officials, at polling places and in other bureaucratic agencies. This might also be the result of gaps in voter education programs or inconsistencies in reviewing standards across Georgia counties.

Mail voting may be necessary to ensure public health, but without much-needed protections — prepaid postage, voter education outreach by local election officials, an opportunity to address a missing or mismatched signature, helping voters by picking up and returning mail ballots, and counting ballots postmarked (rather than received) by Election Day — it could disenfranchise some voters.

Enrijeta Shino (@enrijetashino) is an assistant professor of political science at the University of North Florida.

Mara Suttmann-Lea (@Mara_Suttmann) is an assistant professor of American politics in the department of government and international relations at Connecticut College.

Daniel A. Smith (@electionsmith) is a professor and chair of the department of political science at the University of Florida.