Questions linger as new research suggests election was linked to rise in coronavirus cases.
Research on the effects Wisconsin’s spring election continues to emerge — and not all of it agrees.
A study released Monday by economists at the University of Wisconsin-Oshkosh and Ball State University suggests that in-person voting may have led to a “large” increase in the rate of positive coronavirus tests weeks later.
The report, which has not been peerreviewed, is at least the third to come out since the April 7 election and the first to conclude a positive link, giving the public a front-row seat on the messy, uncertain and sometimes lurching progress of science.
Epidemiologists and infectious disease experts in Wisconsin previously said the spring election did not lead to a feared spike in COVID-19 cases, though they warned that the effects may be hidden in the data and difficult to ever detect.
Investigations involving contact tracing by public health officials were similarly inconclusive. Last week, Milwaukee County epidemiologists said they found 26 county residents who may have been infected with coronavirus during in-person voting.
However, they said their attempts to prove the link were complicated by a lack of data and the fact that the election took place around Easter and Passover, which led to more people gathering in general.
The new study, led by economists Chris Cotti and Bryan Engelhardt at UW-Oshkosh, attempts to tease out the effects of the election on a broader scale by comparing the crowdedness of polling places in each county with the rate of positive coronavirus tests in the weeks afterward.
The researchers found that counties with higher voter density – in other words, a higher number of people who voted at each polling place – later saw a higher rate of positive coronavirus tests.
Specifically, a 10% increase in the
number of voters per polling place corresponded to a roughly 17% higher rate of positive coronavirus tests in that county two and three weeks later, after the coronavirus incubation period had passed.
“The findings suggest it may be prudent for policymakers and election clerks to either expand the number of polling places dramatically – maybe with longer hours or more early voting times – to try to keep the population density as low as possible,” Cotti said. “Or, just to have more mail-in options as well.”
The models found similar results even after removing Brown County, which experienced large outbreaks at meatpacking facilities shortly after the election, and Milwaukee County, which is so large it could be driving the statewide number.
The researchers also incorporated census data to control for demographic differences between counties, such as age and education levels. They also controlled for differences in social distancing habits by using detailed cellphone mobility data.
“We tried to account for as many factors as we could, and tried to be responsible with our analysis,” Cotti said. “There’s strong evidence of a good relationship here.”
Cotti said he and his co-authors will now solicit feedback on the study before submitting the paper for publication in a scientific journal.
“This analysis brings really important insight and does use proper methods. We were definitely going in a downward trend and then we started an upward trend. And it is quite possible we would have continued on the downward trend for a couple more days without the election.” Mustafa Hussein Health economist and social epidemiologist with the University of Wisconsin-Milwaukee
Differing viewpoints
Public health experts had varying views of the new study.
Nasia Safdar, an infectious disease expert with the University of Wisconsin-Madison and medical director of infection control at UW Health, said the study addresses an important question, but cannot eliminate the possibility that other activities during the same time period might have been the real cause of cases.
“They did a pretty careful assessment of traffic during the period of interest, but these challenges remain with these kinds of studies,” Safdar said. “It’s association, but not causation.”
Safdar said the best way to determine whether in-person voting directly led to infections would be to contact a large number of voters, trace their activities and contacts, and then see who tested positive in the weeks following the election.
An even better method would be to perform genetic sequencing on their viruses to see if they all came from a common source. However, these kinds of epidemiological studies are “labor-intensive,” Safdar said.
Mustafa Hussein, a health economist and social epidemiologist with the University of Wisconsin-Milwaukee, called the study “compelling.”
“This analysis brings really important insight and does use proper methods,” Hussein said. “We were definitely going in a downward trend and then we started an upward trend. And it is quite possible we would have continued on the downward trend for a couple more days without the election.”
Oguzhan Alagoz, an expert in infectious disease modeling at UW-Madison, said he thinks a slight bump in COVID-19 cases after the election may be attributable to in-person voting.
But Alagoz said the new study does not convince him the election had a significant role in driving the numbers, and that there are other factors that could have been responsible for the new cases.
“I agree in-person voting led to some increase, a slight increase, “he said.
The UW-Oshkosh and Ball State University study is at least the third to come out since the April 7 election. None has been peer-reviewed, and researchers have also disagreed over their quality.
One report released April 23 by a physician at Froedtert & the Medical College of Wisconsin compared the rate of new coronavirus cases in Wisconsin to the rate of new cases nationally and found no difference post-election.
However, some researchers criticized the study as too simplistic in its statistical methods.
A study released April 29 by epidemiologists at the University of Hong Kong and Stanford University used mathematical algorithms to reconstruct the trajectory of the epidemic in Wisconsin.
The study found “no detectable spike” on election day and concluded that in-person voting on April 7 appeared to be a “low-risk activity.”
Safdar said members of the public should be aware that pre-print studies, such as the one released Monday, have not been independently vetted for potential caveats.
“In terms of thinking about the rigor of results, you want to wait for that peer-reviewed study,” Safdar said.
State health officials are no longer asking questions about election activity on the forms used to track each positive coronavirus case.
On Friday, Wisconsin Department of Health Services spokesperson Jennifer Miller said the “final” tally of people who voted in person or worked at the polls and tested positive shortly afterward stands at 71.
“It is important to note that while these cases have voting or working at the polls in common, that doesn’t mean they couldn’t have been exposed somewhere else,” Miller said.