Seriously, what the hell just happened?
For anyone not living in the UK, I'm talking about our recent election which took pretty much everyone by surprise. All the major newspapers and polls have spent the last few months proclaiming that Britain's right-wing Conservative party were due a landslide victory. This is more or less the only reason that Theresa May bothered to call an election in the first place- with polls reading two thirds of votes would go to the Conservatives, she figured that she could only gain by calling a snap election. Just to give you an idea of how wrong she was, and how wrong the polls got it, here's the headline from the Independent on election day:
'Final reckoning: May on track for sweeping win'1
and here's the headline from the same newspaper the day after:
'Black and blue: May's election gamble backfires'2
This isn't the first time polls have been vastly out either, with Donald Trump being predicted to lose confidently in the US 2016 election, and the Brexit referendum being viewed as a definite remain. So seriously, what happened? Quite a bit of money goes into polling the public about votes so you'd think the that this would get reliable results. YouGov for instance, one the the UK's largest polling companies, reported a turnover of £88,200,000 last year☀. You'd think when there's that much money at stake, companies would have found a way of making polls extremely accurate by now.
Maybe that's a bit naive though. After all, polling by its very nature is based on asking people questions, which is generally a bad idea. People are wildly wrong on most things, a recent survey by Ispos MORI showed that, as Britons, on average:
- We guess that the immigrant population is twice as large as it is.
- We think that there are almost 12 million less overweight and obese adults than there are
- We think that there are around twice as many non-religious people as there are.
So maybe we just suck at answering questions, and anything based on that methodology is always going to fail. But you'd think this wouldn't be such an issue with asking people there intentions or thoughts. As Evans writes, there's no real difference between asking someone if something is true, and if they think something is true֏. So if it's not down to us just being wrong generally, why does everyone keep getting it so wrong?
Part of it might be down to our overly complicated voting system- in the UK votes are dividing amongst more-or-less equally sized constituencies and then whichever party has the most constituencies wins the election. This leads to some pretty crazy rifts between votes and seats, and makes election results even harder to predict⅋. For instance Labour had 2% less votes this election, but ended up with 9% less seats4. So when even knowing the popular vote doesn't guarantee the election result, you can see why it gets difficult. Still, even at just predicting the popular vote, the polls were pretty far out.
Maybe the main reason that polls get it so wrong, is that they suck at predicting voter turnout. Evans Witt covers this in an NPR article: "We know polls do a poor job with emotion/enthusiasm/commitment"5. Getting people to stop in the street and answer questions can be hard, so polls tend to be short, and lack detail. People can tell you who they would vote for, but if you want to get an idea of exactly how likely they are to go out and vote for them, you're probably talking about a much more detailed survey that a lot of people don't have time for. This might go some way to explaining why two candidates who defied polls recently (Jeremy Corbyn and Donald Trump) were on fairly extreme sides of the political spectrum (Corbyn on the far left, and Trump on the far right). People with strong opinions are likely to be a lot more motivated to go out and vote than those who favour more middle-of-the-road candidates. It's hard to imagine people getting fired up at a Clinton rally in the same way as they do at a Trump one.
It's probably also true to say that polls are still pretty good even if they're not quite great. France's recent election in which centrist Macron comfortably defeated far-right Le Penn was universally predicted by polls. And many of the recent votes that have gone against expectations, like the US Presidential Election, UK General Election, and UK Referendum Vote, have all been extremely close calls. Trump won the election but not the popular vote. The Conservatives won the most seats, but not by enough to govern. The UK voted to leave Europe, but by a majority of 2%. It's easy to be down on polls, but all in all, they're fairly accurate, just not really accurate. Which, for predicting the future, is maybe not quite so bad. After all, weather predictions are notoriously inaccurate for more than a few days ahead6, and the weather is a much better understood phenomenon that the human psyche. Like Jung says, 'the great decisions of human life have as a rule far more to do with the instincts and other mysterious unconscious factors than with conscious will and well-meaning reasonableness'☮. Plus, it's not true that all the polls were wrong, there was one experimental poll by YouGov that managed to accurately predict the UK general election, but that was mainly ignored in the run-up to the election7.
So, do believe the hype, but just not very much. Wait, is that right?
1. The Independent, June 8 2017
2. The Independent, June 9 2017
☀. YouGov's financial investors report for the 2015/2016 year.
3. Ipsos Mori, Perils in Perception (2015)
֏. 'If someone asks me “Do you think there is going to be a third world war?”, I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question “Will there be a third world war?"'
G Evans, The Varieties of Reference (1982).
⅋. The US presidential election, also has a similar system. Which is why Donald Trump one, despite Hillary Clinton gaining more votes.
4. You can see the full breakdown of the vote share on BBC news.
5. Danielle Kurtzleben, 4 Possible Reasons The Polls Got It So Wrong This Year (2016)
6. Minitab, Weather Forecasts: Just How Reliable Are They?
☮. Jung, Modern Man in Search of a Soul (1933)
7. New Scientist, How YouGov's experimental poll correctly called the UK election (2016)


