How many 2020 Voters are not going to show up for the mid-terms?
…because they have since died of COVID-19.
There are now fewer than 60 days to go until the 2022 mid-term elections are held on November 8. The continuing drop in gasoline prices and by the earth-shattering Hobbs decision overturning Roe v. Wade have dented the expectations of a “red wave” at the polls. The COVID pandemic is receding into the background, overrun by news from Ukraine, monkey pox, and the FBI’s search of Mar-a-Lago to name a few.
Regrettably, people have neither stopped contracting COVID nor dying from it; an average of 314 Americans per day succumbed to the disease the week ending Sept. 7. At the end of August news hit that US life expectancy has dropped precipitously for the 2nd year in a row. The US has lagged other industrialized countries in life expectancy for some time and now sits behind such 3rd world countries as Peru, Morocco, and Tunisia.
This drop in life expectancy and the never ending casualties from COVID got me wondering about the potential impact on the mid-term elections. After all, older people are more likely to die of COVID and older people are more likely to vote. In this post we’ll use this correlation to estimate the number of Biden and Trump 2020 voters who have died from COVID-19.
Before the 2020 Election
In the early days of the COVID-19 pandemic it was the blue states who were hit the hardest. Urban states such as New York struggled with overflowing hospitals and morgues. There was a sense at the time that population density was a key factor in how hard an area would be impacted. That the pandemic was focused in the blue states may have impacted the Trump administration’s political calculus.
Yet by the time the 2020 election rolled around, the pandemic was impacting all the states, even rural ones. Approximately 229 thousand Americans had died by November 3, 2020 and with the exception of those who managed to vote by mail prior to their deaths, most of these people did not get to vote.
While there is nothing to suggest that COVID-19 has a political preference in terms of who it infects and kills, political ideology has regrettably driven widely varying public health policy and individual behavior. Blue states were more likely to shut down schools, require social distancing and masking, and have experienced higher vaccination rates since the approval of COVID-19 vaccines. As the pandemic wore on, infection and fatality rates began to grow faster in red states.
After the 2020 Election
Unfortunately there is no data on whether people who have died since the election voted or who they voted for. The CDC does track COVID-19 deaths by age, gender, race, and location but not by political affiliation or voting history.
Let’s compare the distribution of the total population, the rate of COVID-19 deaths post-2020 election, and propensity to vote by age group:
Note: we are purposely ignoring the 0–17 population in this post as they are ineligible to vote and also have experienced very low COVID-19 mortality.
Based on the above data we can compute the probability that a member of a particular age group voted in 2020 or has died of COVID-19 at some point during the pandemic.
The above charts are based on national data. Each state, and even each county or congressional district will have slightly different distributions but the data is not available at every level of granularity. The CDC reports on deaths by county but not by congressional district (which have also been redrawn since 2020). Fortunately vote totals are available both by county and congressional district so we’ll use county as the smallest geographical unit in our analysis.
How many 2020 voters have died?
To answer this question we need to focus on the 808k people who have died since the 2020 election. Using only age as variable, we can estimate the joint probability that someone who died was also a voter.
For a given county, the probability of a death being a voter is given by:
where n is the number of age groups and:
Thus:
After analyzing the data at the county level and aggregating back up to the entire US, we estimate that the total number of 2020 voters who have died is 573,060. This is 0.36% of 2020 voters. Conversely, the probability that a person who died after the 2020 election had voted in it is 0.71 or 71%.
Another way to look at this is that a bit more than half of the people who have died of COVID-19 in the US voted in the 2020 election.
If you voted, you most likely didn’t die of COVID, but if you died of COVID after the 2020 election, it’s highly likely you had voted.
County Level Mortality
The following chart shows the mortality rate by county since the 2020 election. The x-axis represents Trump’s share of the two-party vote. The size of each bubble represents that county’s population.
One can clearly see that blue counties were more impacted before the 2020 election but that since the election the trend has reversed with mortality increasing strongly with Trump’s share of the two-party vote.
If we look at the estimated mortality among 2020 voters the numbers are even starker:
Here the scale goes all way up to 1,400 COVID-19 deaths/100,000 voters. That’s up to a 1.4% mortality rate in those tiny counties at the top right of the chart.
The more interesting part of the chart is the middle where in evenly divided counties the mortality rate is just above 400/100k or 0.4% which is still quite large in the context of a tight election.
State-Level Impacts
We can roll-up the county level voter death estimates to the state level.
Those tiny black dots on the chart are the error bars representing the number of estimated voter deaths since 2020. They are almost invisible except for the states in the middle. Let’s zoom in:
Only in Arizona and Georgia has the estimated number of voters who have since died exceeded the margin of victory.
As one can see from the chart below, the correlation of estimated voter deaths with Trump’s 2020 margin is strong. West Virginia turns out to be the state that has lost the highest percentage of its 2020 voters to COVID-19 (almost 0.6%) followed by Oklahoma. Utah is a notable outlier as a red state (mortality well below the trend line) as is the blue state of New Mexico (mortality well above the trend line).
Conclusion
The estimated number of 2020 voters who have since died is very large overall — about the same as the total population of Wyoming. But the impact on the 2022 midterms and the 2024 general election is still very small compared to variations from other factors: natural death, children turning 18, migration between states, get-out-the-vote efforts, and voter suppression.
Remember, these estimates are just that, estimates. It’s not possible to determine an exact number of voter deaths without linking voter files to death records. The application of state-level aggregate measures at the county level is an important source of potential error in this analysis but county-level data on the distribution of population, votes, and COVID deaths by age range is not readily available.
Data and Code
The data and code behind this post are publicly available on GitHub as a Jupyter notebook.
