As the novel coronavirus pandemic besieges the United States, more and more observers are suggesting that November’s votes should be cast by mail — allowing the least possible in-person contact, reducing health risks to both voters and poll workers.

But shifting millions of voters to mail balloting would bring other risks. First, voting by mail includes many steps between someone’s request for an absentee ballot and delivering that ballot to be counted — gaps that ballots can fall through. Second, voters can make mistakes marking the ballot that they may not be able to correct.

If states do significantly expand their voting by mail, will they be able to reduce these risks?

The growth of voting by mail

Voting by mail has grown steadily for two decades, increasing from 8 percent in 1992 to 21 percent in 2016, according to U.S. Census Bureau data, mostly in a few states. Five states — Colorado, Oregon, Washington, Hawaii and Utah — now mail all registered voters a ballot before every election, with California moving in that direction. Thirty-three states and the District of Columbia allow “no excuse” absentee mail-in voting, which means anyone can vote by mail, for any reason. The remaining 13 limit voting by mail to people with excuses such as travel or illness.

In only seven states in 2016 did a majority of voters submit ballots by mail, not in person. In most states, fewer than 10 percent cast ballots by mail. That means that, for most states, suddenly doubling or tripling their mail-in balloting would be expensive and would require significant planning and new logistical systems.

Leaks in the vote-by-mail system

Voting by mail is different from voting in person. In most states, a voter must request an absentee ballot. A ballot is mailed to the voter, who fills it out and returns it by mail. Once delivered to the local election office, the ballot is verified — often by matching the ballot’s signature to the voter’s signature on file — and then counted.

Each step has hazards. One of us, Charles Stewart, after analyzing data reported by the U.S. Election Assistance Commission and in public opinion studies, estimated that in 2008, 3.9 million requested ballots were never received; 2.9 million ballots mailed to voters were never returned; and 800,000 returned ballots were rejected.

Not all these gaps were failures. Some people may have requested mail ballots and then voted in person. But that’s not true of others; nearly 1 million ballots were rejected, mainly because of mismatched or missing signatures or ballots arriving too late.

People who vote by mail are significantly more likely to make mistakes than those who vote in person, as Stewart and collaborators found in a study of California election results from 1990 to 2010. Of course, people who vote in person may make mistakes — but they can more easily ask for help in correcting their ballots.

How can these problems be prevented?

State officials who manage elections with the highest mail-voting rates know how to handle these risks. They’ve identified qualified vendors for printing ballots and envelopes, partnered with the U.S. Postal Service to deliver them, developed tracking systems so voters can see where their ballots are in the delivery stream, and found ways to review signatures and count ballots that reduce the number of rejected ballots. Doing this took decades, not months.

People new to voting by mail are especially prone to making technical mistakes — say, forgetting to sign the ballot envelope — that can lead to rejection. Studies of Florida and California show young people and minority voters are more likely to have their absentee ballots rejected. Official data have shown significant variability among counties within states in rejection rates. Whether these differences are due to variations in actual mistakes or the strictness of officials has not been studied academically, as far as we know.

Some vote-by-mail states allow people to drop off ballots in person. In the 2016 election, most voters in Colorado, Oregon and Washington who received mailed ballots returned them either to a vote center or a secure drop-off location. Voting by mail may be a misleading term; voting at home may be more accurate.

Mail is just a system for distributing ballots; the return system can be up to the voter.

Many at-home voters return their ballots at the last moment — which makes it harder to process them and to report results on election night, which can undermine voters’ confidence in the election. But receiving, verifying and counting the large number of ballots takes more staff and equipment. As another of us, Robert Stein, found, if election officials ask voters to return ballots earlier, there is some evidence that voters will do so. However, allowing someone other than family or friends to return at-home ballots is controversial, teetering on the line between ease of voting and lax security.

Can it be done by November?

At-home voting, when fully developed, may cost less, as states save on polling places and poll workers. But shifting to a new voting system would involve many upfront costs. One advocacy group suggested that a transition to a “vote-at-home” system in Michigan would cost $38 million to fund mailings, dropboxes, voter education and other infrastructure. Funding this may be especially difficult during the coronavirus crisis, with state and local revenue down and expenses up.

To make such a mass change possible, some states would have to change their laws to allow widespread at-home voting. Some state legislatures are not meeting; all have other immediate priorities. Partisanship may get in the way. And states will have to choose between taking emergency action just to weather the crisis or deliberately making lasting policy changes. Either way, policymakers will have to see whether at-home voting requires changing existing rules for, say, voter identification and proof of residency.

Many voters who might otherwise not have the opportunity will be happy to cast their ballots at home this November. But rushing to put it into place nationwide would surely bring some unpleasant and unintended consequences.

Barry Burden (@bcburden) is a professor of political science and director of the Elections Research Center at the University of Wisconsin at Madison.

Robert M. Stein is the Lena Gohlman Fox Professor of Political Science and fellow at the James A. Baker III Institute for Public Policy at Rice University.

Read more: