Many observers expect a record number of people to vote by mail in November rather than go to the polls in person and risk exposure to the novel coronavirus. However, mailed ballots are at higher risk of not being counted, and strains on the U.S. Postal Service could affect whether ballots are received in time. As a result, political analysts increasingly worry that this election might hinge on how voters cast their ballots.

Republican and Democratic leaders have openly speculated about how voting by mail and possible Postal Service disruptions might affect election results. President Trump has repeatedly tried to prevent states from expanding mail balloting and has rejected requests for more support for the Postal Service to handle an anticipated surge.

That’s become an important issue for Democratic leaders and campaign officials. House Speaker Nancy Pelosi (D-Calif.) has accused Trump of a “campaign to sabotage the election” by disrupting the Postal Service beforehand.

Here’s what the Democratic Party worries about: If more Democrats than Republicans vote by mail but mail ballots aren’t counted at the same rate as those cast in person, it might hurt Democrats’ chances of flipping key Senate seats and winning battleground states.

Are they right? Research on past elections — held before the pandemic and before Trump began criticizing voting by mail — found that voters from both major parties cast ballots by mail at similar rates. But this year’s circumstances are quite different. So is there a difference in how Democrats and Republicans intend to vote in November? To find out, we conducted a study. We found both that there is a gap — and that the gap is growing. Part of the gap can probably be attributed to Republicans’ distrust of expert evidence.

How do Americans want to vote in November?

We conducted three surveys, one each in April, June and August, through the polling firm Lucid, each with a sample size of more than 5,600 Americans weighted to be nationally representative on age, race/ethnicity, income and education. This means for each survey, we have about 2,500 Democrats and 2,200 Republicans; within those groups, we include those who said they leaned toward one party.

We find a significant, growing gap over how citizens want to cast a ballot. In April, 40 percent of Democrats said they would like to vote by mail, while 30 percent of Republicans indicated the same. By June, that gap had grown to 20 percentage points, with 45 percent of Democrats saying they’d vote by mail to 25 percent of Republicans. By August, half of all Democrats said they want to vote by mail this election, while only a quarter of Republicans said they would, for a gap of 25 percentage points.

We saw the same pattern in how respondents want others to vote. When asked in April whether they support sending mail ballots to any voter who requests one, 87 percent of Democrats and 64 percent of Republicans agreed, for a gap of 23 percentage points.

By August, this gap grew to 33 percentage points. Over 80 percent of Democrats supported this policy, while Republicans were essentially evenly divided between supporting and opposing absentee voting by request, which is in place in most states, including nearly every swing state in the presidential election.

Why do so many more Democrats than Republicans plan to vote by mail?

One clear explanation of the growing gap has been well documented: Partisans often take cues from their party’s elites, as scholarship has long found. When Trump criticizes voting by mail, as he frequently does, Republicans take note. When other Republican politicians reiterate Trump’s comments, the cue is made even more clear.

But we found another possible reason: Republicans and Democrats respond differently to information about the pandemic. When we showed respondents projections about how the pandemic would probably unfold, Democrats became more likely to want to vote by mail. Republicans did not.

Reading the projections had the biggest effect in April. As both the pandemic and the election campaigns unfolded, covid-19 projections mattered less. At the same time, the differences in Democrats’ and Republicans’ plans grew. This suggests that Americans’ opinions crystallized somewhat over the summer as those two groups’ views of the pandemic diverged.

That may be because Republicans are more likely to distrust experts than Democrats. Research conducted before the pandemic found that Republicans are less likely to view scientists as trustworthy sources of information. This appears to have spilled over into the issue of whether to vote by mail, contributing to the growing partisan gap.

Regardless of the reason, if Trump’s goal was to get more Republicans to vote in person, he will probably get his way in November. As a result, any hiccups in counting mail ballots would be most likely to hurt Democrats.

Mackenzie Lockhart (@LockhartM) is a PhD candidate in political science at the University of California at San Diego.

Seth J. Hill is an associate professor of political science at UC San Diego.

Jennifer Merolla (@MerollaJenn) is a professor of political science at UC Riverside.

Mindy Romero (@mindysromero) is an assistant professor in the University of Southern California Sol Price School of Public Policy and the founder and director of the Center for Inclusive Democracy at USC.

Thad Kousser (@ThadKousser) is a professor of political science and department chair at UC San Diego.