I have been digging into election data recently. Specifically, the MIT Election Lab has a great dataset with the presidential election results of each county since 2000. I paired that data with a 2013 county classification scheme from the NCHS which sorts counties into groups based on how large they are and how close they are to a large metro.
I was interested in where the shifts were from the 2012 to the 2016 election. Where did Democrats lose (and gain) ground? Of course, when you lose about 2 million votes in the popular vote margin, as the Democratic candidate did from 2012 to 2016, you are doing more losing of ground than gaining of ground.
The first thing I saw in the data was that smaller counties, already a strong area for Republicans, became even redder in 2016. Take all counties with less than 250,000 people. In 2012, Romney won a large 58% of the two-party vote from these counties. In 2016, Trump won an even larger 64% of the two-party vote from these counties. Just looking at raw vote margin, we saw a 3.6 million popular vote swing overall towards the Republican candidate. That is more than the overall popular vote swing of 2.1 million votes!
Remember that margins are important, even in places that are already pretty red or blue. At least in modern elections, if the Democrat can not lose too badly in rural America, they can make up the deficit in more urban areas and win the state overall (of course, this depends on the relative urban/rural makeup of a state). If they give up more ground than usual outside of large cities, they have an even tougher hill to climb.
Take the very important 2020 swing state of Pennsylvania as an example. Let’s divide the state into two parts, the very urban counties of Philadelphia and Allegheny (where Philadelphia and Pittsburgh reside, respectively), and the entire rest of the state.
In 2012, Obama won Philadelphia County and Allegheny County by a combined 583 thousand votes. He lost the rest of the state by a wide 273 thousand votes, but that means he still maintained a 310 thousand vote winning margin.
Now, let’s look at 2016. Clinton actually won Philadelphia County and Allegeny County by roughly the same 583 thousand votes. So far, so good. But she lost the rest of the state by an even larger 627 thousand votes, meaning she lost the state overall by 44 thousand votes! The areas outside Philadelphia and Pittsburgh were her undoing.
Ok, I should mention that this is a bit of an oversimplification. Dividing the state into just urban centers and everything else glosses over the suburbs, where Democrats made real progress in 2016. We can see on the image above shades of blue and purple in Montgomery, Chester, and Delaware counties, indicating a shift toward the Democrat.
Still, Pennsylvania echoes some of the trends we saw overall in the country in 2016. Clinton actually improved a bit on Obama’s 2012 overall performance in counties classified by the NCHS study as “Large Central Metros”, but was undone by losses in the smaller counties.
The fidelity of this type of analysis can be improved if you go to a smaller level than county (such as precincts), but I think the larger point still stands. Democrats lost real ground in rural America in 2016, and it remains to be seen if they will improve on their performance there in 2020.