Well, good for Clallam County (if you have no idea where that is, it’s on the Olympic peninsula in Washington state. Port Angeles? Nirvana? Does that help?). They voted for Biden, and now have a 40 year winning streak in ‘predicting’ the outcome of presidential elections.
Clallam County is on quite the winning streak — one that no other county in the United States can claim.
According to the latest count on Monday from the Clallam County Auditor’s office, 50.43% of voters cast their ballot for former Vice President Joe Biden and 46.72% voted for President Donald Trump.
That means the county has picked the correct presidential winner in every race since 1980.
Going into this election cycle, Clallam County was just one of 19 counties nationwide to hold that voting record.
With just over 50% of the county’s votes going to Biden, it is now the only county in the country to have kept up such a winning streak.
The county residents don’t know how they are so good at this. I have an explanation, though: it’s chance.
40 years is just ten presidential elections. Because of the way our system works, it’s a binary choice, basically — so like a coin flip. The likelihood of calling 10 coin flips in a row is 1 in 1024. There are more than 1000 counties in the country, so I’d guess that the chance of one of them having a ten election streak is pretty good. Also the fact that it’s not really a 50:50 chance — one outcome proves to be slightly more likely after the fact — and that Clallam has a nice demographic mix, with the slightly urban Port Angeles plus all the conservative loggers and fishers, and it’s not surprising at all.
Then, of course, look what xkcd posted:
None of that will prevent Washington state journalists from descending on Clallam county in 2024.
Charly says
Yeah, one does not need to be a statistic whiz to come to see that this is just a chance.
Reginald Selkirk says
Trump may have been the winner in 2016, but he certainly wasn’t correct.
JoeBuddha says
Reminds me of a bet I’ve always wanted to try on the statistically challenged. I could bet you I can toss a fair coin in a fair way and get five heads in a row. The secret is to start with fifty (to hedge your bets). 32 is too close.
jenorafeuer says
There’s an old scam along those lines:
Find ten thousand people (actually 10240). Tell half of them the market will go up next week, half of them the market will go down.
Whichever way it went, stop talking to the people you sent the wrong prediction to, and to the remaining 5120 people, do the same thing: tell half the people the market will go up, half that it will go down.
By the end of ten weeks, you have a group of 10 people who have seen you predict the market correctly ten times in a row. Now you start convincing them to ‘invest’ money with you…
unclefrogy says
that was a long buildup to the punch line hahahahah!
Sean Boyd says
A minor quibble, PZ: Nirvana got started in Aberdeen, which is Grays Harbor County and in particular isn’t on the Olympic Peninsula.
I’m actually a bit surprised that any WA county not containing the major urban centers skewed towards Biden. This state gets conservative real quick once you leave the Seattle to Olympia corridor.
nomadiq says
There is always an XKCD cartoon…
davidc1 says
I loved the way you amuricans didn’t let a little thing like a civil war mess with the 4 year term rule .
Were the Southern Rebels allowed to vote in 1864?
davidc1 says
I should have added” defeated former” to Southern rebels .
kaleberg says
It’s definitely chance, but we do have a magic billboard that predicts the next crash. Back in 1999, it was pushing tech stocks just in time for the tech crash. Back in 2006, it was pushing real estate investment just in time for the real estate crash. I’ll have to drive by and see what is going to hit us next. (There is an actual mechanism for this billboard though. As investors unload, they need suckers, and even as a scam progresses in major population centers, we have a good supply of uninformed folks out here.
chesapeake says
The Southern rebels were not defeated until 1865.
Erp says
There are apparently over 3,000 counties in the US. Some states like Kentucky have really small counties and others really large (I, for instance, live in a county whose population is about the same as Nebraska or West Virginia and whose land size is a bit bigger than Rhode Island and it isn’t even the biggest county in the state by population or area).
Erp says
I note also this is a winning streak which is likely to be a bit longer than a losing streak. I do wonder which county is the worst at choosing the winner of a presidential election?
gijoel says
Isn’t there a scam where they mail out different horse racing results to a bunch of different people and over a couple of weeks. Until they zero down to a couple of lucky marks and demand payment for their predictive systems.
Helge says
For quite a while, 44 years, Utah could boast of always voting the way the nation voted. That ended with Kennedy…
Mississippi’s streak in 1848 was just 20 years.
Ridana says
My favorite in the xkcd comic was “No Dem. incumbent without combat experience has beaten someone whose first name is worth more in Scrabble…until Bill beat Bob.”
davidc1 says
@11 I thought it ended in 64? Were they allowed to vote in 1870?
Anton Mates says
3,141+ counties and county-equivalents in the US, as per Erp @12. If the chance of a 10-election winning streak is 1/1024, there’s a 95.35% probability that at least one county will achieve that winning streak. Actually, we should expect several of these to exist.
However, the chance of a 10-election winning streak is actually quite a bit lower than 1/1024 for most counties, because each county’s election choices over time are not independent from one another. Solidly conservative or liberal counties are going to vote for one party almost every time, so they have almost no chance of a 10-year winning streak because neither party actually wins 10 times in a row at the national level. It’s only a perfect 50/50 “battleground” county that has a winning streak chance as high as 1/1024.
This is an example of statistical overdispersion–you’d need something like a beta-binomial distribution to model it. I haven’t the time, but I suspect it works out to match PZ’s conclusion. There shouldn’t be a lot of counties with 10-year winning streaks, but it’s not surprising that there’s at least one, and that one is probably split fairly evenly between Republican and Democratic voters.