Maria Fallon Romo grew up in a family that talked politics at the dinner table. At 53, the North Dakota special education teacher has been an active voter for about 30 years and, until recently, had never experienced a problem casting her ballot. In 2018, she decided to cast a mail-in ballot for the first time, thinking it would be more convenient than her usual in-person vote.
Two years later, Romo learned that her 2018 vote never counted. An election official with no formal criteria to follow decided that Romo’s signature on her absentee ballot envelope didn’t line up with the signature on her absentee ballot application form — and her vote was rejected. “It was very shocking and very disheartening,” she told The Intercept and Type Investigations. “Just like a lot of other Americans, I think when I vote that my vote counts.”
Romo has lived with multiple sclerosis for over 20 years, which can make it difficult for her to write neatly and consistently. North Dakota, like many states, does not require special training on signature matching or instruct officials to consider disability, age, language proficiency, or any number of other factors that could make a person’s signature look different under different circumstances. By law, officials didn’t even have to tell Romo that her vote was rejected. For North Dakota’s June primary, she said, she has saved a screenshot of her signature on her ballot request form, joking that she can practice matching it exactly on her ballot.
This year, as politicians, voting rights advocates, and elections experts call for a sharp increase in the use of mail-in ballots to prevent the spread of Covid-19, many more voters could find, like Romo, that their votes don’t count. That’s because significantly expanding vote-by-mail also expands the universe of ballots that get scrutinized by officials, and past experience shows that more scrutiny means more ballots get discarded, sometimes for good reason but sometimes by mistake.
“Just by the sheer volume — and just assuming there’s a constant error rate — we’re going to get a significant increase in just the number, the absolute number of rejections,” said University of Florida political scientist Michael McDonald, who specializes in U.S. elections. “There are going to be many thousands, if not millions, of people who are going to be having problems with their mail ballots.”
Inconsistent signature matching rules are just one risk that accompanies voting by mail in the United States. Unlike regular ballots cast at a polling location, ballots cast by mail are inspected before being counted. They can be rejected by election officials for a number of reasons: lateness, errors on the attached form, lack of signature, and a lack of witnesses, among others. Voters who might otherwise have asked a poll worker for help might make mistakes on the ballot they fill out at home, causing their vote not to count.
According to a new calculation by The Intercept and Type Investigations, over 950,000 votes were rejected by officials in the last presidential election, based on data released by the U.S. Election Assistance Commission. This includes more than 300,000 domestic mail-in ballots and more than 600,000 provisional ballots cast at the polls. The risk of rejection varies widely by state. In the 2016 election, the mail-in ballot rejection rate in Georgia was roughly 30 times higher than the mail-in rejection rate in Wisconsin. This figure includes both domestic absentee ballots and ballots from military personnel and citizens abroad.
The five states that currently hold all-mail elections — Colorado, Hawaii, Oregon, Utah, and Washington — didn’t count between 0.7 and 0.9 percent of mail-in ballots received. New York, on the other hand, which received far fewer mail-in ballots relative to its voter turnout, didn’t count ballots it received at a rate up to almost eight times higher than those other states.
Some of the most populous states in the country also had the highest ratio of rejected ballots compared to the size of their voting population, the analysis by The Intercept and Type Investigations of rejected ballots in 2016 across all relevant ballot types — provisional, absentee, and overseas military and civilian — found. A recent lawsuit in New York, where 2016 vote-by-mail ballots were rejected at nearly five times the median rate for the country as a whole, shows some of the pitfalls that may await.
The hurdles that accompany vote-by-mail logistics are not insurmountable. The states that vote entirely by mail have carried out elections for years without major problems. Research shows that states where vote-by-mail has expanded have seen a modest increase in voter turnout, and voting rights advocates are unanimous about the benefits of giving voters more options to cast their ballot.But with potentially millions voting by mail for the first time, especially amid the coronavirus pandemic, jurisdictions accustomed to a trickle of mail-in ballots will get a flood. Over half of U.S. states received fewer than 10 percent of their 2018 votes by mail, including populous states like New York and Texas, according to an analysis by the Brennan Center for Justice. Several of the country’s largest jurisdictions will be among the many trying to quickly scale up their vote-by-mail infrastructure to emulate systems that took other states, like Washington, years or even decades to create. Voters who don’t receive absentee ballots in time — like many in Georgia last week — may still show up in person, meaning that the existing in-person infrastructure cannot be abandoned or aggressively consolidated. As the New York Times pointed out, administrators will, in essence, have to conduct two simultaneous elections.
Reducing the risk of valid votes being thrown out will require heroic efforts on the part of election administrators, who will be tasked with reviewing far more votes than ever before. Already, teams of experts and advocates have formed to safeguard the process, but they will need to work quickly, and receive political support, to be successful.
Denise Roberts headed to her Tioga County, New York, polling place early on November 8, 2016, to avoid the lines. But her name wasn’t listed in the poll book. It didn’t make sense, she recalled in an interview with The Intercept and Type Investigations. She had moved to the area several years ago and voted at the same fire station just 12 months before. She even had a paper card from the county Board of Elections confirming that she was registered at her current address.
The poll workers told Roberts that her registration was inactive, and she would have to vote using an affidavit ballot, known in other states as a provisional ballot. Rather than being counted automatically, the ballot would, like Romo’s, be scrutinized after Election Day by elections officials, who would determine whether her vote was legitimate.
Roberts wondered if the fact that she was a black woman in a 95 percent white, rural county had come into play. “It was humiliating,” she said. “I didn’t believe what had just happened.” Then the same thing happened to Roberts’s daughter-in-law Angela, who lived with her. Despite being properly registered at the address where they lived, the Robertses had no confidence that their affidavit ballots would be counted. Their suspicions were correct.
The Robertses were later deposed in a voting rights lawsuit that John Powers of the Lawyers’ Committee for Civil Rights Under Law, who represented the plaintiffs, called a “canary in the coal mine” for efforts to expand vote-by-mail in the 2020 election. It revolved around the state’s practice of moving people from an “active” registration list to an “inactive” registration list if certain trigger events occurred. One of those events, crucially, is the return of a single piece of election mail as undeliverable.
At the time, the names of “inactive” registrants in New York weren’t sent to the polling place, and these voters were required to cast provisional ballots. This year, they will be provided to poll workers, per a federal court order, but they will still be required to vote provisionally. That means their votes can be rejected due to errors on the ballot or, occasionally, for no clear reason whatsoever. That was the case with the Robertses, whom officials ultimately admitted had been wrongly disenfranchised due to administrative error. The state’s apology was little comfort for Denise Roberts. “I never expected to experience that my voting rights would be taken away by somebody making a decision,” she said.
Marc Meredith, a University of Pennsylvania political scientist who was an expert witness in the case, calculated at the time that New York had a higher rate of rejected provisional ballots compared to the total number of votes than any other state in the U.S. In addition to tens of thousands of people who relocated within the state, in 2016 roughly 45,000 people cast provisional ballots with their registered addresses, suggesting that they had very likely been improperly deemed “inactive” by the state. Tens of thousands of erroneously deactivated voters were thus funneled into an administrative process in which their votes had a good chance of being discarded. In the 2016 general election, only about 51 percent of the provisional ballots cast in New York were ultimately counted.
Relying on undeliverable mail as a way to maintain accurate address lists can be highly problematic. “We have uncovered, over the course of time, significant issues with the consistency of the information that we get from the post office vis-à-vis who’s at this location and who is not at the location,” testified New York City Board of Elections Executive Director Michael Ryan, the top election official in the city. He told lawyers in the case that quotidian problems can have disastrous results, especially in urban areas: If the center lock is broken in the bank of mailboxes of a large apartment building, for instance, the postal worker won’t be able to leave the mail there, causing everyone in the building to become “inactive” unless they go to the post office in person to collect their mail within 14 days.
“Quite frankly, it has caused us to have little faith in the overall reliability of the quality of information that we get from the post office,” Ryan said, adding that the moves to inactive status disproportionately affect people who live in multiunit buildings. Ryan complained that to avoid this “unintended consequence” of New York’s registration laws, he tries not to send mailings unless they are critical. Missing virtually any piece of mail — even a letter confirming a voter has been moved back to “active” status — can cause a voter to be kicked off the “active” list if the post office cannot deliver it. In her opinion, Judge Alison Nathan found “the central proxy that the State uses to determine whether a voter has moved has serious problems with its reliability.”
New York’s mail delivery problems bode poorly for the state’s ability to ramp up vote-by-mail in time for the November election, experts said. Some voters won’t receive mail-in ballots to which they are entitled, and other ballots will likely end up in the wrong voters’ hands.
The same is true for other states unaccustomed to regularly interacting with voters that way, said Charles Stewart III, an elections expert at the Massachusetts Institute of Technology. “The vote-by-mail states are doing a lot of mail business with their voters; they’re sending a ballot every election, they’re sending voter information guides every election, they’re interacting with their votes a lot by mail. And one consequence of that is that their voter rolls are very clean,” he said. “Whereas the purpose of the voter roll in New York is not to do a lot of mail transactions with voters, that’s not why it was constructed. And so you’re suddenly putting a burden on the voter list that it wasn’t designed to bear.”
Mail-in and provisional ballots can be rejected for legitimate reasons, such as a would-be voter lacking eligibility. They can also be rejected for more contentious reasons, many of which involve human error: failing to sign the ballot, failing to have a witness where required, or making a mistake on the form that goes alongside the ballot.
The most common reason in 2016 was the one that invalidated Maria Fallon Romo’s ballot: a perceived mismatch in the signature between the ballot and the voter registration, according to a report by the Election Assistance Commission. In many states, the criteria used to reject those ballots are poorly defined and left to the discretion of untrained election workers.
It is a fraught effort: Signatures can vary depending on age, health, disability, writing utensil, and even noise and stress levels. Though signature matching helps election officials confirm the identity of someone casting a ballot from home, rather than from a polling place, election experts said the effectiveness — and accuracy — of signature checks vary widely by state. Some states, like Colorado, have detailed guidance on how to examine a voter’s signature, based on scientific principles of forensic document examination. Other states do not. Research has shown that laypeople are three and a half times more likely than trained forensic document examiners to incorrectly identify signatures as nonmatching. Laypeople are also more likely to reject authentic signatures than to make the opposite error and accept forged signatures.
Research has also shown that voters of color and language-minority voters are more likely to have their mail-in ballots rejected by elections officials. Voters whose native language uses non-Latin characters display greater signature variability, which officials might interpret as attempted forgery. Voters of color might also have less information about how to cast a valid ballot; a 2014 study found that election officials were less likely to answer Latinx voters’ questions compared to white voters’ questions.
- In the 2016 general election, only about 51 percent of the provisional ballots cast in New York were ultimately counted.