A lot has been said about the polls in the run up to yesterday’s UK election. “The polls got it wrong” they said. Jeremy Hardy joked that pollsters now command about as much professional respect as homeopaths!
Well, after last night, I think the polling industry can hold its head high. The joint Ipsos MORI/GfK exit poll was on the money.
At 10pm last night, it predicted 314 Conservative seats and 266 Labour. The current BBC estimate as I write is 319 and 261. Just five seats out. A pretty small margin of error by anyone’s measure.
This backs up the argument I made after the 2015 election. Exit polls will always trump hypothetical polls that ask people to imagine how they’d vote if there was an election tomorrow. Exit polls are much closer to the behaviour they’re trying to measure. And this means they’re more accurate.
So, what about all the other polls over the past six weeks?
They showed some quite divergent results. The overall trend – that Labour was surging – proved correct.
No doubt there will be isolated polls highlighted as “getting it right”. But the body of evidence from the poll of polls showed a small Tory lead. Something played out in the popular vote.
But for me, the real picture is that the whole industry got better at talking about its polls.
The margin of error was there for all to see in the final Ipsos MORI poll.
Several pollsters explained how they manipulate survey results with turnout models. These generally assumed the young were less likely to vote – a fair assumption based on past elections. It looks like the young did vote in great numbers. And the models I saw factoring this in showed the Tories short of the overall majority. (I’m sure it was Yougov but can’t find a link at the moment.)
There will be lots more air time devoted to this election over the next few days. But it won’t be about how the polls got it wrong.