Since we published our map of the unofficial Election Day results three weeks ago, many people have made the same observation: the pattern of areas that voted for President Trump seems to match the current map of COVID-19 positivity rates. Here's a typical tweet, from Andy Newman at the NY Times:
I was also asked about this on the Brian Lehrer Show the following week, and my initial reply was that I wasn't sure. It seemed to me that there were plenty of areas at the time that had high positivity rates but which had voted Democratic, and other areas, especially upstate, which had voted Republican but which still had relatively low rates. Still, the question stayed with me and I decided to dig into the data to see what it said. (TLDR: the association is weak, if it exists at all. Read on for details!)
To get the data (using the Board of Elections' unofficial results, which still don't include absentee votes) in the right format, I had to map every election district in the city to a ZIP code, and then look up the positivity rate. This is inherently an imprecise process, because some election districts overlap with multiple ZIP codes. For simplicity, I mapped the center of each election district to a point and then looked up the ZIP code for that point. (Steven Romalewski at CUNY's Center for Urban Research, suggested this approach.)
Then I pulled the positivity rates from the city's Department of Health dataset for three date ranges: the week of the election, this week, and the total period since April. Finally, I created a scatterplot:
What does it say? Look at that black line—the trend shows that election districts that voted more heavily for President-elect Biden have lower positivity rates this week. If you switch the date range to the week of the election (November 1st), you'll see the trend holds, though it is more shallow, because overall positivity rates were lower.
Before we draw any conclusions, let's also look at each borough. Click a borough in the legend on the upper right. Here's where it gets interesting. Though the trend for the whole city shows higher positivity rates for Republican election districts for the week of November 17th -- and that still seems to be true when looking just at Brooklyn or Staten Island -- that relationship does not hold elsewhere. In the Bronx and Queens, for instance, the trend is basically flat—no real association between positivity rates and voting. In Manhattan, the trend is actually reversed: areas with higher Democratic voting seem to have higher positivity rates.
Statisticians love to say that correlation isn't causation, but here we don't even have a clear correlation between political voting and positivity that holds across the boroughs. To me, that signals there may be some link, it's weak, and the patterns we're seeing in the map can better be explained by other factors.
For instance, timing. Staten Island might have high positivity rates this month simply because it wasn't hit as hard during the first wave, not because of its underlying political divisions. Brooklyn's positivity rate might more easily be explained by social distancing factors—areas where people can't or won't social distance have higher rates.
To make the point more clear, let's look at two maps. The first one divides the city into four areas, by political preference and positivity:
Notice in particular the light red areas in Queens: These are places that voted Republican but have lower than average positivity rates. Now look at the dark blue: Those are Democratic areas with high-positivity rates. If political preference was the driving factor here, those areas shouldn't exist. But they do, which tells us that other factors besides political preference are more important.
Or look at this map, which shows total positivity over the whole course of the epidemic:
The pattern here shows COVID-19 has hit hard both Democratic and Republican areas of the city outside the richest parts of Manhattan, Queens, and Brooklyn.
So, while it might be satisfying to some Democrats to look at the voting results map, and blame the Republicans in Borough Park or Staten Island for their current COVID situation, remember, plenty of people who voted Democratic were affected too, both recently and over the course of the pandemic.