There’s an increasing sense of certainty among pollsters, bankers and pundits that there’ll be no brexit from the eu on june 23. The betting money is going in one direction; there’s a 78 percent implied likelihood of a remain victory on betfair, a betting website. And sterling leapt over a point against the dollar thursday on the release of an opinion poll showing a comfortable 17-point lead for “remainers.”
Might i humbly suggest a degree of overconfidence here? With odds at nearly 5-1, a bet on brexit is starting to look cheap. I’m not saying it will happen—it may still be unlikely, but it’s not 5-1 unlikely. Here’s why.
First of all, the polling has been all over the place. The leading polling organizations, which are sometimes accused of “herding” their forecasts together to provide mutual cover, are producing wildly divergent numbers as to the state of the race. In particular, there has been an unsettling divide between polls conducted over the telephone, which have been showing comfortable leads of over 10 points for remain, and online, which have the race much closer, only a narrow lead for remain or nearer neck-and-neck.
There’s been much beard-stroking and pontificating as to the reason for this divide. A theory has gained traction among parts of the political community that the phone polls are closer to the mark because social liberals are somehow harder to get hold of online, but there’s precious little evidence for this. In truth, everyone has been a bit stumped by this distinction that has never appeared so strongly in the past.
Then, on friday, a colleague at yougov, andy morris, published an analysis that showed, in fact, the samples being reached by telephone pollsters consistently contain too many people who are educated to degree level or higher—and we know that those people are much more likely than less educated people to want to remain in the eu. Andy reweighted some recent phone polls to correct this error and suddenly they are much more in line with the close race the online polls have been showing all along.
So, if you are looking for an accurate measure of what the whole uk population says about brexit today, it looks like the race is still close. It may well move in the coming weeks (the scottish referendum saw a dramatic move back to the status quo in the final two weeks), but for now, it’s close.
Then there’s the issue of turnout. We can get a good sense of how the different groups in society are breaking on the question of in/out, but what is much harder to predict is who will actually bother to turn up and vote. Most normal people do not feel passionately on the topic, and with the vote taking place on a midsummer’s day, whole swathes of the country will likely find more interesting things to do than vote in the referendum.
The two most important factors affecting levels of turnout are age (older people vote much, much more reliably than younger people) and social class (more educated, affluent people vote in greater numbers than less educated, poorer people). Meanwhile, older people are dramatically more in favor of brexit (by as much as 70/30 in some groups) and young people overwhelmingly back staying in the eu (by as much as 87 percent in some groups), so there is a big potential advantage to the brexiters if there is a big age gap in voting. On the other hand, the less well-educated and affluent you are the more likely you are to support brexit. So, if all those less-educated people the telephone pollsters were missing from their samples don’t show up to vote anyway, the result could still be the big remain victory they are showing.
We have modeled these competing effects using data from the last 5,000 yougov respondents, and have for people to play about with turnout levels themselves. What it shows clearly is how different levels of turnout among different groups can make all the difference in a close race.
It’s impossible to predict the level of turnout as every referendum is different. An astonishing 85 percent of scots turned out to vote in the independence referendum in 2014, but a less impressive 41 percent of people turned out to vote in the referendum on moving to an alternative vote system in 2011.
So, if class turns out to be the big factor in this referendum, and that better educated people, even the kids among them, turn out in droves to support staying in the eu, while less engaged people stay at home, we would expect a big victory for the remain campaign. But if the more traditional age gap holds true, and old people of all social classes turn out and express their frustration with a vote for brexit, while the college kids are too hungover or distracted to make the effort, britain could be in for a surprise result that takes it out of the eu.
Turnout is the great unknowable factor in this election, and is the great last hope of the brexiters