By Steven Rosenfeld
In the 2020 presidential election, 66 million Americans voted with a mailed-out ballot after most states loosened restrictions on qualifications to vote by mail to make voting safer in the pandemic. Another 36 million people voted in person at an early voting site before Election Day after many states expanded this option.
Together, more than 56 million voters cast a ballot in a different way than in 2016, which was “extraordinary,” as one recent scholarly study said. North Carolina’s increase in using mailed-out ballots, alone, was fivefold. Georgia’s was sixfold. Wisconsin’s was fifteenfold.
The presidential election set a turnout record and has since led to a record number of election administration bills in state legislatures, some preserving last fall’s expanded voting options and others rolling back those choices. Those state-by-state fights have led to some of the highest-profile voting rights battles since the early 1960s. Both parties are claiming that their vision for political representation faces existential threats.
In recent weeks, an influential voice, election scholars, who rely on “observed facts and data,” have begun to weigh in on what was most important, less so, and not at all important with helping 159 million Americans vote last fall. Their findings, while preliminary and sometimes contradictory, provide an important counterpoint to the partisan claims in the state-by-state voting rights battles.
An examination of a half-dozen academic draft papers, policy institute reports, and scholarly articles finds some consensus on which voting options boosted 2020’s turnout. These options include mailing every registered voter in a state or county a ballot and allowing voters to register or update their registration information and then vote. But there were also conflicting data and assessments over specific voting regimes and rules that were suspended to help voters get a ballot into their hands.
Last year, 29 states and the District of Columbia passed 79 laws to institute a grab bag of options centered around using mailed-out ballots and early in-person voting. While lawmakers were concerned about making voting safer, these steps also made voting more convenient and accessible by cutting bureaucracy and extending deadlines.
In many respects, 2020 was an unprecedented and successful experiment in making voting more convenient. While broad findings about these trends are coming in—including some work that has been misreported in the media—further research will explore what options were embraced by voting blocs with historically low turnout rates, such as communities of color and younger voters.
“We are at the very beginning of a period in which academic researchers can study the 2020 election,” said Rutgers University’s Lorraine Minnite, a political scientist who has studied voter turnout issues for more than a decade. “What’s coming out right now is preliminary. A fuller picture has yet to emerge, which is why you are seeing such disparate research designs and findings.”
Contours of Voting in 2020
Nationally, 2020 saw the highest presidential election turnout in 116 years. Almost nobody expected that result last spring, when state and local election officials pivoted to more flexible methods of casting ballots that protected the health of all involved. Notably, 159.7 million Americans or 67 percent of registered voters cast ballots, 23.8 million more voters than in 2016.
The biggest-picture contours of voting options and turnout in 2020’s general election—voting last fall—was the “America Goes to the Polls” report from Nonprofit VOTE and the U.S. Elections Project. Nonprofit VOTE works to increase participation. The U.S. Elections Project, founded by the University of Florida’s Michael McDonald, created a national repository of early voting and vote-by-mail data.
In 2020, 45 percent of Americans voted with mailed-out ballots, they reported. Twenty-five percent of Americans voted early at an in-person voting site. The final 30 percent voted in person on Election Day. In 2016, in contrast, 21 percent of the presidential electorate voted with a mailed-out ballot and 60 percent voted at an Election Day poll. The highest and lowest turnout states in 2020 reflected different regimens—from the starting line of voter registration to the finish line of getting and casting a ballot.
“All of the top 10 turnout states either sent all their voters a mail ballot, have same-day registration that allows voters to register or update their registration when they vote, or both,” the Nonprofit VOTE/U.S. Elections Project report said. “Eight of the bottom 10 turnout states cut off voter registration four weeks before the election or required an excuse [on a separate application] to use a mail ballot.”
Other policy reports and draft academic papers offered more detail on where there was a consensus about what voting options had the biggest impact on 2020’s voter turnout. There was agreement that the states offering a same-day voter registration and voting option, which 24 states and the District of Columbia did in 2020, boosted turnout by 5 percent, as noted in the Nonprofit VOTE/U.S. Elections Project report.
There also was an affirmation of prior research that mailing a ballot to every registered voter in a state or county boosts turnout. In 2020, the increase, reported by the Public Policy Institute of California (PPIC), averaged 3.9 percent nationwide in the 10 states and the District of Columbia that mailed registered voters a ballot. Of those, the six states and the District of Columbia that did this for the first time in 2020 saw voter turnout increase 4.6 percent.
PPIC’s top finding—based on a working academic paper—was that mailing every Californian voter a ballot led to a 12 percent increase in statewide turnout compared to 2016. Their paper noted that California did other things to encourage turnout, such as offering more choices on how and where to return ballots, adding the ability for voters to track their ballot’s whereabouts and launching an extensive public education effort about the new voting regimen.
Another working paper, from a team at the University of California, Berkeley; Stanford University; and the University of Washington, noted that its Colorado research, conducted prior to 2020’s presidential election, found turnout increases among several low-propensity voting cohorts:
Young voters (16.6% increase in turnout),
Blue-collar voters (10.0%),
Voters without a high school diploma (9.6%),
and all low-income voters (8.1%).
The turnout increase among Democrats and Republicans was 8 percent, but it was 12 percent among independents.
Inconsistent Data, Methods, Findings
But there were differences in preliminary research over other absentee voting regimes and their more specific rules—bureaucratic details that are now being removed or revived by some state legislatures. In general, there are four absentee voting regimes, according to the U.S. Election Assistance Commission:
- Every voter receives a mailed-out ballot;
- Voters must first apply and satisfy an excuse to get a ballot;
- Voters must apply for a ballot but no excuse is required;
- Voters can sign up once to be on a permanent absentee list, which typically is for seniors and people with disabilities.
The fine print of these absentee voting regimes, many of which were suspended in 2020, is now being reinstated or toughened in Republican-majority legislatures. These reforms include stricter voter ID requirements and shorter filing timetables.
In most states, registered voters must apply to receive a mailed-out ballot. In 2020, 15 states mailed voters an application, as opposed to leaving that task to voters. Fourteen states suspended their requirement that voters had to satisfy a predetermined “excuse” that they could not vote on Election Day, such as age, infirmity or travel. Four states kept their excuse requirement.
Consider the excuse requirement that 14 states suspended. A Stanford University team found that suspending the excuse led to a 0.8 percent increase in turnout. But the PPIC report found that suspending the excuse led to a 2.7 percent drop in turnout nationwide.
This apparent conflict is an example of where different methodologies yield different results. PPIC’s report offered no explanation for the turnout drop other than a footnote, which academics said was common in preliminary papers. In a follow-up phone call, Eric McGhee, a coauthor of the PPIC report, said that his team’s research and Stanford’s were different. As for the 2.7 percent drop, he said that states made many changes in the voting rules to help voters—some of which voters embraced; others that they apparently did not.
“If some jurisdictions were anticipating a bad outcome [with accommodating voters] and they adopted a particular reform to get ahead of that, and the reform had no effect on it, that could make it appear it was causing a decline,” he said. PPIC’s methodology did not allow his team to delve into this more nuanced scenario, he said.
On another absentee balloting issue, PPIC reported that directly sending voters an application to receive a ballot by mail led to a 1.7 percent turnout increase, which it called “modest.” That turnout increase might seem small to the public or “modest” to cautious scholars, but it was larger than Joe Biden’s presidential victory margin in the states of Georgia (0.23 percent), Arizona (0.30 percent) and Wisconsin (0.63 percent).
Another area where there was disagreement concerned whether voters choose their method of voting based more on convenience—what was more accessible—or more based on coronavirus fears. In 2020, 29 states and the District of Columbia changed their laws to allow people to vote by mail or early due to public health concerns, as noted in an April article in the Journal of Democracy coauthored by Nathaniel Persily, a Stanford Law School professor, and Charles Stewart III, an MIT political scientist who oversees a research lab known for its parsing of U.S. Census data of voters. (In 2020, they created the Stanford-MIT Healthy Elections Project, a resource for election officials.)
The Nonprofit VOTE/U.S. Elections Project team said that 66 percent of voters chose what was “most convenient” when deciding when and where to vote, while 24 percent cited “concerns about coronavirus” as a major influence on their decision. It drew on a Pew Research Center survey of about 12,000 voters in November 2020 asking why those individuals chose to vote early or via a mailed-out ballot. On the other hand, the Persily-Stewart article cited post-election census data and found the opposite response.
“Postelection responses to the SPAE [Survey of the Performance of American Elections] describe the reasons behind the shift to mail balloting,” they wrote. “Overall, 59 percent of respondents who stated that they were very worried about family members catching covid reported having voted by mail, compared to 28 percent who said they had no covid worries.”
Blinders or Not?
Academics have long played an important role in shaping election law and voting rules. Their facts and findings are a counterweight to partisan arguments. But insightful work is not always ready when lawmakers are reforming voting laws. Moreover, the push by some academics and the press to report findings that defy conventional wisdom can lead to premature, if not mistaken, reporting with high political stakes.
A recent episode of high-profile press coverage offers a cautionary tale about the perils of overclaiming about preliminary research and omitting important contexts, and other factors that affect the topic at hand, such as whether turnout alone is the best metric of what contributes to a more representative electorate. In this case, the focus was on which voting options and their associated bureaucracy did or did not boost overall turnout.
In an April 4 analysis, Nate Cohn, a New York Times data journalist and analyst of political trends, sparked a storm in election circles when he wrote that Democrats and Republicans were both mistaken “about whether making it easier or harder to vote, especially by mail, has a significant effect on turnout or electoral outcomes.” He continued, “The evidence suggests it does not.”
Cohn’s blanket assertion mostly relied on a preliminary paper that focused on one bureaucratic hurdle in one version of voting with mailed-out ballots, in 2020. The Stanford study that Cohn drew on mostly looked at Texas, but said that its findings were applicable to 14 statesthat suspended their rule that voters declare why they cannot vote at the polls.
The researchers reported that removing the excuse, alone, boosted turnout by 0.8 percent. But they concluded that increase could be “statistical noise” that did not prove anything about making voting easier and boosting turnout. Cohn’s conclusion, pinned on the preliminary study and older, pre-2020 research—before 45 percent of the electorate voted with mailed-out ballots—struck several nerves.
“The idea that making voting easier *won’t* improve turnout is one of political science’s worst takes,” immediately tweeted Charlotte Hill, a University of California, Berkeley, PhD candidate and coauthor of a 2020 working paper finding otherwise under Colorado’s universal vote-by-mail system. “And to be clear, many political scientists don’t buy it.”
Cohn did not discuss other versions of mail-based voting where 2020 turnout went up—such as the 10 states that mailed every voter a ballot to minimize the health risks. But he cited older research—from before voting with mailed-out ballots more than doubled nationally in 2020—that found, as he wrote, that “[a]lmost everyone who cares enough to vote will brave the inconveniences of in-person voting to do so.”
That assertion offended advocates who applauded election officials’ extraordinary efforts to expand voting options. And it was seen as immoral by organizers who strove to help millions of voters who used these options for the first time, despite ex-President Donald Trump’s attacks on 2020’s expanded options and on the voters using them.
“Voters are more than just numbers on a @Nate_Cohn spreadsheet,” tweeted Fair Fight, a Georgia-based group founded by Democrat Stacey Abrams. “They are people. Implying [that] 12-hour lines are not that bad because voters will find a way to make up for lost wages or they’ll vote after they faint is cruel and racist. Turnout would be even higher if not for barriers.”
Cohn, notably, had defenders. Rice University’s Robert Stein, who has studied elections for decades and worked in Texas and elsewhere to expand 2020’s voting options, said that Cohn’s report stayed within the boundaries of the academic research that he cited.
“Nate Cohn did not write what I will call ‘fake news,’” Stein said. “He wrote the right article based on much of the literature on convenience voting… What Nate was writing about was one form of vote by mail, and that is excuse or no-excuse mail-in voting.”
But Cohn’s critics countered that voting in 2020 was so different from prior presidential elections—with 56 million people casting ballots in a new way for the first time—that it was premature to overly rely on new research or on pre-pandemic literature. (The 56 million figure is based on U.S. Elections Project data from 2016 and 2020, and its recent report comparing turnout differences for early and mail voting.)
“The assumption of a continuity in the research findings from studies of absentee voting and the like in the past and this election could be incorrect, certainly at the margins,” said Minnite. “But it is at the margins that elections are won or lost.”
What Doesn’t Turnout Measure?
Voter turnout is the “most basic measure of the success of an election,” Persily and Stewart noted in the Journal of Democracy. But in 2020 that metric “does not inventory the ways in which the Trump administration, allied election administrators and outside groups undermined the execution of the election,” another data scientist commented in his private newsletter. Nor does turnout, alone, address another core issue: if specific voting options helped historically infrequent or low-propensity voters.
“There is an assumption that the more people are participating, then the closer we have gotten to the goal of the electorate being the same as the overall population,” PPIC’s McGhee said. “I think it is true that turnout level and the representativeness [are] certainly likely to be correlated with each other, but they are not the same thing. We have to be really careful about that. And the follow-on question of ‘how does this voting method affect that representativeness?’ is a super important one.”
The equity, or representativeness question, is what researchers like McGhee are now delving into next as they keep studying 2020’s voting options and voter turnout. But in the meantime, other data, including from nonacademics such as political data firms, about the impacts of certain voting options is filtering into 2021’s political fights.
Harris County, Texas, where Houston is located and an election jurisdiction larger than 25 states, operated eight 24-hour voting centers last fall to accommodate voters who could not leave their work or family obligations. It was one of many innovations to make voting more accessible, including drive-through voting sites and mailing voters absentee ballot applications.
A New York Times report in late April said that a majority of those who used late-night voting were people of color, and that the voting hours expansion was targeted by a bill in Texas’ GOP legislature to be barred from future elections. The Times linked to a tweet thread from the Texas Civil Rights Project, which said “56% of voters who voted during late-night hours were Black, Hispanic, or Asian. … Data proves these options offered by @HarrisVotes were popular with voters AND made voting more accessible for everyone.”
How reliable is this claim, which is drawn from data from TargetSmart, a Democratic political data firm, according to an attribution on the Texas Civil Rights Project’s tweets? Their raw data is one source among many that is being compared and studied by academics, said Stein, whose students are working on papers about Harris County’s voting using data from TargetSmart and other data firms. But that research will not be done before Texas’ legislature likely passes its 2021 voting reforms, he said.
The 2020 election shows that if politicians give voters more accessible options, some of those options will be used. But beyond the broad trends, the scale and impact may not be quickly known as they pertain to deregulating specific prior mail-based and early voting regimes, and newer accessible options.
That absence of definitive research has not stopped partisan Republicans bent on rolling back last fall’s array of voting options. But it has given a few Republicans pause and led to some of the most draconian proposed rollbacks to be deleted from bills. Many GOP legislators know that their party’s impulse could backfire, as their base, not just Democrats, took advantage of various voting options.
“Once you give voters opportunities to do things, you can’t pull them back,” Stein said. “Their concern is not whether Democrats will vote with [reinstated] voter ID or [more limited] early voting, but whether their voters will show up.”