Bellwether counties suddenly became meaningless in this year's election

Wall Street Journal article: There were 19 bellwether counties that had reliably voted for the eventual U.S. presidential victor in ten consecutive elections, dating back to 1980.

But this time around, Trump won 18 out of 19 of them - yet Biden won the overall presidential election. This shows how far polarization is going - increasingly, there are fewer and fewer purple counties in America - you are either blue or red now.

I don’t think it’s necessarily a sign of polarization. Demographics change. There’s no reason to assume that bellwethers will always stay bellwethers.

I’d be curious to know how many “bellwether counties” that had been consistently right since 1976 got it wrong in 2016. And then again 1972-2012. 1968-2008. And so on.

What I’m getting at is… maybe “bellwether counties” change all the time? I neither have nor desire to have access to WSJ. So if the article addresses that, then I guess please quote it. As it stands, I guess add me to the “demographics change” camp. This election was polarizing, sure, but “bellwether counties” (I keep putting it in quotes because I am not convinced there really is such a thing, at least not to the extent of being able to predict future performance based on past performance) are hardly the best evidence of that.

I suspect they didn’t “suddenly” become meaningless, and they were always that way. People like to see patterns in things, and the number of elections is a pretty small sample size.

If elections were totally random (the results from a county didn’t correlate to the national results at all) any county you look at would have a 1 in 1024 chance of having predicted the last 10 elections correctly in a row. Wikipedia tells me there are ~3000 counties in the US, so just as a baseline in a hypothetical uncorrelated situation it wouldn’t be unusual to see around 3 “bellwether” counties that would no doubt attribute some meaning to that status.

Obviously elections aren’t completely random, which is likely why there are 19 and not 3, but it does suggest to me that there is likely a lot of noise and silliness in these “bellwether” claims, in particular in assigning any specific meaning to them.

It’s easy to attribute meaning to patterns when your sample size is small. See Electoral Precedent.

I did that same calculation, but didn’t want to take the time to throw in all the exclaimers about “not really random” and then go on to explain that too. So, bravo!

Modhat: Moved to P&E from IMHO

This is, and I Am Not Making This Up, an actual key piece of evidence put forward by an actual attorney who actually works for the actual President of the United States in an actual motion to the actual U.S. Supreme Court.

“The fact that nearly half of the country believes the election was stolen should come as no surprise. President Trump prevailed on nearly every historical indicia of success in presidential elections…He won 18 of the country’s 19 so-called “bellwether” counties—counties whose vote, historically, almost always goes for the candidate who wins the election…”

I don’t know; I think this is surprising. If they really were just random, and just the selection of the coin flips that got lucky, then we’d have expected to see about half of them go for Trump this year. The fact that 18 out of 19 did sure seems to me like it indicates something. What? I don’t know.

[Voiceover: It doesn’t mean fraud]

What you’re basically seeing is a long term switch away from the Democratic Party in rural and ruralish counties and a turning toward the Republicans (and especially toward Trump).

I can’t find the exact figures but it’s something like this:

Bill Clinton won about 1700 counties
Barack Obama won about 900 counties
Joe Biden won fewer than 500 counties.

Rural counties in general have moved toward the Rs. Exceptions have been the Black Belt in states like LA and GA, Indian reservations in places like SD and AZ, heavily Mexican American counties in the SW, and a small number of largely white counties in places like VT.

I think if you look at those bellwether counties you’ll see a bunch of small population rural and ruralish counties, mostly white, which were fairly evenly divided politically for many years. They were Democratic enough to go for Clinton in the nineties and even Obama some time afterward. But the bulk of them have shifted rightward and aren’t currently going to support a Dem for president. So it’s not really a weird coincidence or mysterious in any way—it’s part of a national trend in which the formerly swingiest counties are no longer swingy, and so it’s not surprising that so many of them moved out of the bellwether category at the same time.