Did Covid-19 deaths tip the midterms Blue?
Mortality is up in Red counties and small demographic shifts are noticeable in competitive states.
Note: This article was written and published before the midterm elections. It looks like Covid might have indeed made a difference in some areas, but it’ll tough to know. Here’s the article as it was, the day of the elections—before any votes had been cast. -JF
In early 2020, Covid-19 was killing droves of people in big cities. Because cities are more likely to be Democratic strongholds, by the end of the first year of the pandemic, the national electorate was actually a little bit more Republican than usual.
However, since Covid-19 vaccines were authorized (and pandemic mitigation was tossed out the window in many areas), things have changed. Counties that voted for Biden are more likely to have higher vaccination rates. Since early 2021, cumulative mortality in Red states has increased steadily, surpassing coastal and urban Blue enclaves. Going into the 2022 midterm elections, states in the South account for 44% of the excess mortality in the US during the Covid-19 pandemic, despite comprising just 39% of the US population.
Could these differences tip the balance of close elections? Possibly. I wanted to see if excess mortality could find meaningful changes in the electorate—changes which might not be showing up in polls.
To start, I had to know whether pollsters already might have already corrected for these effects. I asked poll master Nate Silver whether the polls leading into the elections already take these unexpected demographic shifts into account. I wanted to know whether polls reflect the population that is alive to answer surveys (meaning that a de facto correction had already been made), or whether pollsters take the samples they obtain and then make corrections so that the readouts reflect the supposed ages and party affiliations of the voters in a region. For example, if 10% of survey responses came from seniors, but seniors make up 20% of voters in said region, would a pollster double the weight of the answers obtained from seniors in the poll? According to Silver, most would. If most pollsters are relying on 2018 midterm turnouts or 2020 Census data in creating their poll results, as Silver believes they are, the polls of “likely voters” we’ve been seeing may be wrong to some degree in many areas.
To see whether this could make a noticeable difference on an election, I ran some numbers, using Pennsylvania as a use-case.
Essentially, I set up a small study. I took 4 counties in Pennsylvania: 1 medium sized predominantly Democratic county (Delaware county) and 3 smaller predominantly Republican counties (Mercer, Union, and York counties) that add up to approximately the same population as Delaware county. These counties were similarly but oppositely partisan. Delaware went 63% for Biden in 2020, while Mercer, Union, and York went 63% for Trump. With my fairly evenly matched Blue and Red zones identified, I next calculated how much excess mortality had occurred from March 1st, 2020 through September 1st, 2022.
All told, among residents old enough to vote, since March 2020 there have been around 3,800 more deaths in these 4 counties than there otherwise would have been, but for the pandemic. Most (67%) of those excess deaths occurred in the 3 Red counties (despite being 51.8% of the population of our group of counties). Pre-pandemic, Blue-blooded Delaware county accounted for 47% of the mortality among the 4 chosen counties. During the pandemic, that figure fell to 38%.
Whether these differences might tip an election depends on a few things. First, how close is the election and at what jurisdictional level? Local elections are unlikely to be changed by these differences because these areas are so politically lopsided. A few hundred (or thousand) votes won’t turn a local 63%-37% race into a nail-biter. But these counties do contribute to state-wide elections and in tight races, slightly smaller turnouts (whether due to voter interest or voter aliveness) from Blue or Red areas could amount to important differences.
But it’s hard to know just how much of a difference excess deaths will make. For starters, even though there were more deaths in Red areas, we don’t know if those deaths were evenly distributed among the local partisan population. To be “conservative,” I assumed that excess mortality was equally divvied up among the electorate, meaning that 63% of excess mortality would have occurred among 2020 Democratic voters in Delaware county, and 63% of excess mortality would have occurred among 2020 Republican voters in Mercer, Union, and York counties. With those assumptions, the difference in the electorate would amount to a net change of just 330 more Democratic voters than otherwise expected, out of 922,000 voting-aged adults in the 4 counties.
That sounds small. And it is. But it could be significant for two reasons. First, if this experiment were repeated in other parts of the state, the total effect could quickly add up to many thousands of votes, even in areas with less stark partisan divides. In the past, many statewide elections have come down to margins that are just a few hundred votes, let alone a few thousands. So, the effect of excess mortality could indeed make a difference in a close contest. Moreover, many deep Red areas often voted 85-90% for Trump, and so the effects from those regions might be larger than the signal we are capturing using the 4 counties I selected. Also, as mentioned, I assumed that within each county, the excess mortality was distributed in proportion to the usual partisan breakdown. However, given what we know about vaccination rates by party affiliation, it’s likely that there were fewer deaths among Democrats in all 4 counties, and more deaths among Republicans. Even a 5-10% gap in vaccination rates could manifest as an electorate with many thousands more Democrats and many thousands fewer Republicans than pollsters currently anticipate.
I expect that there will be a few exceedingly tight races somewhere in the US today. Once these races declare themselves, we’ll run the numbers to see whether the pandemic was in responsible for unexpected changes in electorate in those regions.
Since the beginning of the pandemic, I’ve often wondered if politicians downplaying Covid might be shooting themselves in the feet. After all, wouldn’t a politician want to keep their constituents alive long enough to vote for them again? Today, we’ll find out whether pandemic mortality killed any political careers.
❓💡🗣️ What are your questions? Comments? Join the conversation below.
Follow me on Twitter, Instagram, and on Facebook and help me share accurate frontline medical information!