The big front page news overnight is that pollsters YouGov are predicting a hung parliament. YouGov are respected pollsters and their new model is apparently based on polling 50,000 voters. Guido has recommended betting on a Tory outright win. Should punters change their bets?
Let’s get some perspective. YouGov also gave us an eve of election poll in 2015 predicting a hung parliament; with the Tories on 288, Labour 268.
I was at a party that election night with YouGov’s boss Stephan Shakespeare. When the Exit poll flashed up on the screens at 10pm predicting a Tory outright victory – Shakespeare turned ashen.
The actual result had the Tories on 331 and Labour on 232. To his credit, Stephan said they had got it badly wrong and they were going to rethink their model. This new model is based on 50,000 interviews over the course of a week, which they profile against the general population to make it representative.
Their central prediction now is that the Tories will lose 30 seats and Labour gain similar, putting us firmly in hung parliament territory.
Other pollsters such as ICM are still predicting a big win for the Tories with a 100 seat majority.
All my experience, including anecdotal feedback from Labour MPs, says that the Tories are set to win. Labour have campaigned better than expected, Theresa May’s manifesto launch was a failure and she has not had a good campaign.That might be enough to stop a Tory landslide, but it won’t be enough to see them lose seats.
YouGov may be right about the final vote-share narrowing. Their model is unlikely to take account of campaign methods. Political parties also do polling that they don’t publish. This tells them where they are in contention and could win or lose a seat. The parties then pour resources into those seats which distorts the national picture captured in most polls. That is why elections in safe seats are relatively quiet and elections in marginal seats are a blizzard of leaflets and mailshots coming through the door.
YouGov have recent form for this. They predicted the national vote share for Donald Trump and Hillary Clinton pretty accurately. Clinton did get more votes than Trump – in the wrong places. Their last poll predicted an electoral college landslide for the former First Lady by 317 to 221.
If you are going to bet on YouGov’s prediction, bear this in mind. You can get 5/1 against ‘No Overall Majority’ with Paddy Power.
I’m sticking with a Conservative majority, even though the price has shifted from 1/10 to 1/7 now. In fact, I’m going to add to my position on it happening before polling day.