Doctors and demographers recently noticed another tragic example of how polarization shapes America: The pandemic has killed more people in the nation’s Republican enclaves than its Democratic strongholds. They explain the gap by pointing to Republican resistance to vaccines and the GOP’s more cavalier approach to combating the virus in general.
Those findings suggest many more Republicans — tens of thousands, if not hundreds of thousands — have died of COVID-19 than Democrats, leading some to wonder with some morbidity what the political impact will be. Will Democrats, facing the normal midterm election headwinds plus high inflation, do surprisingly well in 2022 for the simple, sad fact that there are fewer Republicans?
Or, to put it another way: Can we expect this partisan mortality shift to show up in the polling data?
A partisan divide
In January, the Pew Research Center found that 33 percent of Republicans had not received a vaccine, compared to 10 percent of Democrats. Another Pew survey that month showed a widening mask gap, with Republicans less likely than Democrats (39 percent vs. 79 percent) to say they wore masks inside stores most or all of the time.
The disparate approaches have come with a partisan death divide. GOP-run states that lifted lockdowns sooner had higher excess death rates than blue states, an article in the Journal of the American Medical Association showed. Florida and Georgia had more than 200 deaths per 100,000, while New York had 112 per 100,000, New Jersey 73 per 100,000 and Massachusetts 50 per 100,000. “Between August and December 2021, Florida experienced more than triple the number of excess deaths (29, 252) as New York (8,786), despite both states having similar population counts (21.7 million and 19.3 million, respectively),” Steven H. Woolf wrote.
The Pew Research Center similarly found that more Americans died in counties that supported Trump than those that backed Biden. Comparing the 20 percent of Americans each living in counties that Trump or Biden took by the highest margins in 2020, Pew found the reddest places suffered nearly 70,000 more deaths from COVID-19 since the pandemic began. And overall, the COVID-19 death rate in all counties Trump won was 326 per 100,000, higher than 258 per 100,000 for Biden.
The mortality disparity between the parties is large enough to potentially swing an election — the 2016 election was decided by just under 80,000 votes spread among Michigan, Pennsylvania and Wisconsin. So, could it also be enough to throw off the midterm polls?
“It’s a fair question,” said Berwood Yost, director of the Center for Opinion Research at Franklin and Marshall College, which polls Pennsylvania regularly.
Pennsylvania’s rural, red counties have been losing population for a while now — well before the pandemic — Yost said, while its cities have been growing. “Those smaller rural communities lost sizable population,” Yost said. “But, at the same time, they gained voters.”
Even though deep blue Philadelphia and its suburbs have been growing, the registration figures there haven’t moved much. “Folks in these other areas are just more active and engaged,” Yost said. “And their partisanship has shifted toward Republicans.”
“Despite a population decline of 5.4 percent, voter registration has only declined 2.3 percent in micropolitan areas and has actually increased by 1.5 percent in non core counties, despite a population loss of about 4.5 percent,” Yost wrote recently. “At the same time, the most urban counties, despite a modest population increase of 2.0 percent, [have] seen a decline of 5.0 percent in registered voters.”
Recent census data for 2020 suggests that the pandemic might have at least temporarily reversed the longstanding trend of large metro areas growing while smaller towns shrink. Some of the changes are attributable to domestic migration — big city dwellers seeking more space and cheaper rent in smaller cities and suburbs. It’s unclear whether that trend will continue as the pandemic subsides and employers encourage workers to come back to the office.
Will the deaths show up in the data?
All the pollsters CQ Roll Call spoke with for this report said that COVID-19 deaths, even though they’ve fallen disproportionately on Republicans, shouldn’t affect their surveys.
“All adults, regardless of voter registration status, are interviewed because we utilize recent Census data of the 18-year-old and over population for weighting purposes. Weighting is a statistical adjustment of the data. Gender, age, education, race, and region are the demographics that are weighted to reflect Census information,” Doug Schwartz, director of the Quinnipiac University Poll, said in an email. “This process permits not only obtaining full coverage of the target population, but also ensures that subsets within the population are accurately represented within our results.”
The Gallup Poll doesn’t adjust their samples for party identification, said Jeff Jones, a senior editor.
Instead, they use census data, particularly the annual American Community Survey, to make sure their samples look like America in terms of race, gender, region and educational attainment.
Party identification shifts more rapidly than party registration. Voters will quickly align or distance themselves from a party depending on current events.
“Last year, you would have been in a lot of trouble if you weighted by party because at the beginning of the year the country was a lot more Democratic. At the end of the year, they’re a lot more Republican,” Jones said, adding that Gallup stopped election polling a few years ago.
Last year began with the Jan. 6 attack on the Capitol, which lowered the GOP’s esteem in voters’ eyes, Jones said, but ended with COVID-19 variants spreading and rising inflation, which hurt Democrats.
Yost’s Pennsylvania polls pull from state voter registration rolls. Yost adjusts his polls by party registration, which as recently as 2008 favored Democrats by about a million but now is nearly evenly split.
“It seems like those Republican areas have been harder hit [by COVID-19] and that might have an effect,” Yost said. “But, demographically, people in more urban areas just don’t get out and vote in the same ways.”
Registration also tends to lag behind larger shifts in voting patterns. Many union Democrats in rust belt towns started voting Republican right after the local factory closed, but they’ve been far slower to update their party registration.
The latest Franklin and Marshall poll surveyed 223 registered Democrats, 200 registered Republicans and 67 independents — a 45 percent, 41 percent, 14 percent split. But when asked for what party they identify with, the respondents split 50 percent Republican, 40 percent Democrat and 6 percent independent, while the rest declined to say.
County election boards regularly prune their voter lists of the deceased, so if there are significant changes to the electorate’s composition caused by COVID-19, they should show up in the registration data, Yost said. Pennsylvania removed 84,577 deceased voters from the rolls in 2020.
The Keystone State also removes inactive voters (after first sending a notice and giving them a chance to stay on the rolls), which helps catch the death certificates that never made it to the county clerks’ offices and unconfirmed moves out of state. In 2020, Pennsylvania removed 157,690 inactive voters.
There will still be delays and mistakes — no matter what, the data pollsters use will be at least a little old. But that’s just a limitation of polling, Yost said.
“Polls are blunt instruments, not precision instruments,” he said. “They’re useful for understanding context.”