Support for President Trump increased in 2020 in many of the U.S. counties that lost lives at the highest rate to COVID-19, according to an NPR analysis.

Of the 100 counties with the highest COVID-19 death rates per capita, 68 had a higher proportion of votes cast for Trump this cycle than they did in 2016. This includes both Republican-leaning counties and counties that supported Joe Biden.

"It's surprising to see that the president surged compared to 2016 in the areas that have been hit the hardest by the pandemic," says Thad Kousser, a political scientist at the University of California, San Diego.

But Kousser and several other experts stress that many things influence how people vote and that it's currently impossible to tease out what effect high death rates and the pandemic in general might have had on voting. Numerous other factors may have led voters in these places to support Trump.

Of the top 100 counties with the highest COVID-19 death rates, all but six had completed vote counts as of Thursday, when NPR ran this analysis. And the shift toward support for Trump was significant in some cases.

For example, the shift in the vote margin for Trump increased in Starr County, Texas, by 55 points. That's the largest shift toward the president in the U.S., though the county ultimately went for Biden; the county has seen 186 deaths and has a population of 65,000. And in Richmond County, N.Y., Trump's lead over Biden was 9 points higher than it was in 2016; that county saw 1,100 deaths.

Trump's support rose even in the county with the highest death rate per capita in the country: Gove County, Kansas, where there have been 18 deaths out of the town's 2,600 residents. There, Trump's tally increased from a 74-point win against Hillary Clinton in 2016 to a 76-point win against Biden.

Overall, of the top 25% of counties with the highest death rates, two-thirds saw increases in the share of votes cast for Trump.

Kousser notes that teasing out cause and effect with these results is difficult. It's not likely that high death rates caused voters to shift toward Trump, Kousser says, adding that it's "more likely that both of these trends are effects of something different about [these] counties."

For example, "they may have been simpatico with the president's approach to the pandemic," which led people to do things like be less likely to wear masks and impose lockdowns to slow the spread of the coronavirus, Kousser says.

Still, the correlation between high death rates and voting for Trump is worth studying, Kousser says: "[It] shows that the impact of the pandemic on this election was much more complicated than we thought."

The pandemic's impact on voters

It has been well documented that Republicans and Democrats hold extremely different views about the pandemic and Trump's response to it.

Though 41% of voters ranked the pandemic as the most important issue facing the country, according to the AP VoteCast survey conducted by The Associated Press, 73% of those who did so voted for Biden. The second-highest-ranked issue was the economy and jobs, with 28% of voters choosing that issue. Of those, 81% voted for Trump.

Some political scientists say this partisan divide is not surprising.

"What we've seen is over time that the biggest determinant of people's worries about COVID-19 and their attitudes about COVID-19 is not where they live or the COVID-19 deaths in their area, but it is their partisanship," says Shana Gadarian, an associate professor of political science at Syracuse University who has been surveying voters through the campaign. "We know partisanship matters a great deal for people's vote choice."

Americans she surveyed were only slightly more likely to support a ban on large public gatherings, for example, if someone in their family had been infected by the coronavirus. But they were far more likely to support such a ban if they identified as a Democrat, regardless of whether someone in their family had had COVID-19.

Gadarian stresses that "it's not the case that people are totally unresponsive to death and bad things happening in their area."

But "if you are someone who already trusts the president and you trust him to handle the crisis, then you are both not as concerned as Democrats are and you're more willing to trust that he is the person who can keep you safe and keep the country safe from COVID," she says.

Teasing out voter attitudes

Gadarian and others note that many factors influence how people vote. Without more detailed information about the counties where support increased even though the pandemic took a heavy toll, it's impossible to understand cause and effect, says Gary King, a Harvard University social scientist.

"We just don't know what the cause is," King said. "This is a just a snapshot."

It's possible support for Trump would have increased in those counties for reasons entirely unrelated to the pandemic, such as more campaigning or advertising by the president, King says.

"Maybe the people in those areas were Republicans and conservatives and Trump supporters to begin with, and they were increasing their support for Trump, entirely unrelated to the pandemic," King says.

It could also be that places that have strong support for the president also are the most likely to experience more deaths for other reasons, such as their populations having higher rates of risk factors such as older age, other health problems or inferior health care, he says.

It's possible the increased support for Trump among some voters stems from "the gains they believe they've gotten for the economy from the sacrifice they've made through, presumably, refusing to participate in masking and lockdowns," says Danielle Allen, a Harvard University political theorist.

"Having sacrificed, those communities may now be invested in that playbook," Allen says.

And, of course, it's possible that support for Trump would have increased even more in counties that supported him strongly if the pandemic had not hit them so hard.

"We just don't actually know, right?" King says.

But King and others say the effect of the pandemic on voting will be fodder for more in-depth research for years to come, given the historic nature of an election occurring in the midst of one of history's most consequential health emergencies.

"It's a very important research question," King says. "We absolutely want to track that down and figure that out."

Copyright 2020 NPR. To see more, visit https://www.npr.org.