Iowa County, a county in (unsurprisingly) Iowa, had as many blue votes as red votes.
Obama: 4,173
McCain: 4,173
"Other": 125
I thought this would be a rare event until I messed around with a few numbers. If you assume that voters are independent and equally likely to vote for one of two candidates, the chance of an 8,000 person county being equally divided is almost 1%. With fewer people in a county, the odds increase. Assuming 100 counties with fewer than 8,000 people, one expects at least one county in the country to evenly split the vote.
Those are some pretty big assumptions, though. The odds are very sensitive to how split the voters are, so if one candidate has odds of 51% and the other 49%, then the probability for such a county drops to around 0.1%. If you assume 52% odds for one candidate, then we're down to 0.001%. I used the handy binomial calculator from Vassar to crunch the numbers. Also, I do not think that county size and vote color are independent; I would guess that small counties tend to vote red.
A quick, non-exhaustive search shows there were a few other counties that came close:
KY's Bath County: 24
MN's Murray County: 25
MN's Aitkin County: 10
MO's Washington County: 9 (EDIT: added later)
I couldn't find systematic county data to look at, so I might have missed counties that were closer or even tied. The data above (having a few counties in the sub-25 range) suggests around a 20% chance of a tie during an election in which a democrat is slightly ahead.
7 comments:
Iowa even has its own countries now, huh?
Sorry. Couldn't help it.
You try typing "county" a ho-jillion times in a post and we'll see how often you type country.
I have trouble with typing other words, but these are embarrassing and will not be shared.
Blogging is all about sharing the embarrassing with the world.
I had a comment about the votes but I am too lazy to do stats to back it up -- mostly I'm wondering about odd/even populated counties which means also wondering about third-parties. Hm.
I'm not really seeing any down sides. Do tell.
I'm making a poster for an upcoming seminar, and am realizing how much more fun life would be if I was a graphic artist.
Re: odd / even, I hadn't thought of it. I'd guess since you always have a few wackos voting for a third party, you have a 50% chance of having an even number of red and blue votes, so you'd want to halve the odds or double the expectation.
re third party wackos -- yeah, that can vary the parity, that's why i said thinking about it leads to thinking about 3rd parties but if your third party wackos are more likely to be drawn from one of the 2 major parties than the other, that off-centers your vote distribution bell curve and brings you back into the tiny percents again.
It occurs to me now (either I am smarter or less smart first thing in the morning) that clearly the real answer is that when you say a county has 8k people you really mean they cast 8k votes. (Which probably means they have more like 13k people.) Which means that just halving the odds does make sense.
It is true that a third party candidate would shift the votes one way or the other. You could probably come up with a function that takes into account the percentage of votes that go from either party to the third (fourth fifth sixth) candidate.
Post a Comment