The idea that the weather affects turnout is a persistent one. You can understand it — if you’re umming and ahhing over whether or not to vote, and then the heavens open, you might decide against it. But the evidence that it actually does is pretty weak.

One widely cited 2007 study in the US found that an inch of rain suppresses voter turnout by 1%. That is a lot of rain — the average monthly rainfall in London in December is 58mm, just over two inches. But they did find a result, and say that it may have been a large enough effect to influence the 2000 election, when Florida was extremely rainy. Other studies have found similar small impacts in the Netherlands and Spain.

That finding doesn’t appear to replicate everywhere else. A big Swedish study found that rain had no discernible effect on turnout. It may be an artefact of different voting systems: for instance, the Swedish and American voting systems are very different — not least in that Swedes vote on a Sunday and Americans vote on a Tuesday, meaning that Americans are more likely to be rushing to fit it in around work.

And neither system is exactly like the British one. But Chris Hanretty, a political scientist at Royal Holloway University, looked at correlations between rain levels and voter turnout during the European referendum in 2016, and found no statistically significant link. Also, if you eyeball this graph of turnout by season, it doesn’t seem obvious that winter/autumn elections have lower turnout than spring/summer ones.

But *even if there is a link*, it’s very unclear how the turnout suppression will actually affect things. Let’s imagine that the one inch of rain = 1% reduced turnout claim is precisely accurate. And let’s say, for the sake of argument, that there actually will be an inch of rain in London today. (Which is unlikely. In reality, at the time of writing, according to this endearingly nerdy hobbyist in NW3 with a rain gauge on their roof, there has been about 1mm of rain, or about 1/25th of an inch.)

Then let’s say that the 1% of voters who don’t vote when it’s raining are much more likely to vote Labour. Let’s say they’re 60% Labour, 40% Tory, compared to the rest of voters who are 40% Labour 60% Tory.

(Please note: Chris Hanretty also did a Twitter thread along these lines the other day; go and have a read. I’ve used my own, simpler numbers.)

So imagine a constituency of 10,000 voters, all of whom vote every time apart from that 1% who don’t like the rain. 9,900 of them would turn out in the rain; you’d get a Tory win of 5,940 to 3960.

If it *didn’t* rain, and all 10,000 turned out, you’d get a Tory win of 5,980 to 4,020. Under these super-idealised situations, assuming that the strongest possible weather effect is real, assuming that the constituency is in the middle of a Biblical storm such as only comes once or twice a year, and assuming an implausibly large difference between the voting habits of people who own an umbrella and people who don’t, we can make the rain have an effect of a bit less than half of 1% on the result.

That’s not nothing. In 2017 there were about 20 seats in which that could have theoretically changed the result. But, again, this is incredibly idealised. The real effects would be much weaker; I suspect that they would be indistinguishable from zero.

The polls suggest Labour are going to lose, probably badly. That may not turn out to be true, but going out and calling upon the rain gods to either come or go (depending on your political preferences) probably won’t help very much. And it turns out you can’t trust The West Wing for psephological information, that’s what really hurts.

---

Tom Chivers is a science writer. His second book, How to Read Numbers, is out now.

TomChivers

---