Non-comprehensive thoughts on the polls #GE2017

By Nicholas Wong

Haven’t had the time to dig through every individual pollster and their methodologies but here are a few thoughts:

A close race?

The polls/models that seem to favour Labour the most (YouGov, Survation) seem to be the ones that estimate a higher turnout amongst young voters (who overwhelming support Labour over the Tories). The young-old turnout gap is usually substantial, so there’s reason to be (very) sceptical of a hung Parliament or Labour majority.

And it wouldn’t be the first time someone’s claimed they could win on the back of galvanising the never-voted-before bloc. Even if this youth wave happens, there’s the question of distribution — are they just running up majorities in safe Labour seats, or are they converting marginal seats?

But voter behaviour and turnout rates are also an ever-shifting target; when pollsters tweak their methodology, it’s not just to account for flaws made in the past, it’s also in anticipation of how they think conditions are now. Youth turnout in the referendum last year was significantly higher than expected, for example. If the Brexit vote’s your baseline, you might overestimate turnout this time. If 2015’s your baseline, you might be lowballing.

*There’s also a Qriously poll that actually puts Labour ahead of the Conservatives (the first to do so). But there are serious reasons to be sceptical of this poll.

What the polls agree on

What the polls do agree on is that the Tory-Labour gap has shrunk significantly — a consistent trend that just about every pollster has found. They also agree that Labour has yet to overtake the Conservatives in the polls — suggesting perhaps that the race has narrowed but not gotten neck-and-neck.

Unfortunately, this doesn’t say much. When you start a campaign with a 20-point lead, there’s not a lot of room to go up, and plenty to go down. There’s also a difference between what voters think in the day-to-day outside election time, and what they think during a campaign, where candidates and parties are generally under much greater scrutiny. It’s why so many US presidential candidates look so great on paper before primaries and the general election, only to find themselves in a much tighter race once people begin paying attention.

Of course, it surely didn’t help that Theresa May has run what pundits have called a weak campaign so far. She wasn’t a major figure in the Brexit referendum, she kept her Remain support low-key, and she avoided a drawn-out Conservative leadership contest; all of this kept her profile relatively clean for voters in the first few months of her premiership, but a general election campaign has necessarily meant spotlight — and it hasn’t helped.

Corbyn, on the other hand, has had his weaknesses litigated extensively in public since unexpectedly winning the Labour leadership in 2015. Given his own party in constant rebellion against him, his prioritisation of questions over quips at PMQs and his distinctly unpresidential style, it’s not surprising that opinion polls rated him poorly — so it’s also not surprising that he’s had nowhere to go but up in a general election.

Pundits have also noted that Corbyn looks much more at home on the campaign trail than across the dispatch box during PMQs — meeting voters rather than landing the best quips. And given the gap between the popularity of some of Corbyn’s policies and Corbyn the man, an election that pits his manifesto against the Tories’ probably hasn’t hurt.

Lastly, I’m sure there’s a bit of a reversion to the mean going on here, where some voters come back to the fold of their traditional party now that crunch time is here (as opposed to flirting with other options when they don’t have to cast a vote).

Polling errors and Shy Tories

Polling errors are a given. Two key questions, then: how big are the errors and who do they favour? Will the Tories do unexpectedly better again, like in 2015?

Tories have tended to beat their polling more than Labour , where normally you would expect polling errors to even out over time and have an equal chance of favouring either side. The popular explanation for this is the ‘Shy Tory Factor’ — the idea that Tory votes are underrepresented in polls because voters are reluctant to admit they vote Conservative.

Two points to make here. First, while the Shy Tory phenomenon is real enough, the explanation for it — voters misreporting their intentions — isn’t necessarily (or even likely) true. A post-2015 election study found that unrepresentative samples were the main reason the polls underestimated Conservative support in 2015. In other words, pollsters just weren’t surveying enough Tory voters.

Second, just because the Tories have noticeably beat their polls in the past, doesn’t make it a rule: it cannot be assumed they will always do better than the polls say. If a methodological flaw is identified, pollsters usually try to fix it. Tories also did much better than predicted in 1992; in response, pollsters began factoring in how respondents voted in previous elections to account for ‘shy’ voters. If Shy Tories were a thing, it’d probably be fixed by now.

This means that if polls keep lowballing Tory support, it’s either for a reason that’s hard to fix (e.g. making samples more representative) or that whichever methodological flaws present in particular years happen to favour the Conservatives.

Either way, this should caution us against mentally adding 3–4 points for the Tories to polls we see. Pollsters may or may not have managed to fix their failings from 2015, but they’ve certainly tried — which is why individuals polls and forecasts now suggest everything from a hung parliament to a Tory landslide.

“Pundits are always wrong”/ “The underdogs will win”

Conversely, it would be hasty (and wishful thinking for Labour supporters) to assume that Labour will benefit from a systematic polling error as part of some sort of underdog narrative.

The Brexit-Trump one-two punch was a good lesson in critically evaluating conventional political wisdom. But some pundits have the wrong takeaway — take for example the big worry that Marine Le Pen would win the French presidential election, despite Emmanuel Macron’s crushing lead in the polls. He won by nearly two to one against Le Pen.

It’s also worth noting, again, that while Trump and Brexit confounded political expectations, the polls weren’t oblivious to them. Per FiveThirtyEight:

“In the U.K. last year, pundits and punters were irrationally confident of a “Remain” victory in the Brexit vote, even though polls showed it only barely ahead of “Leave.” In the U.S., they ignored how much the race had tightened in the final weeks of the campaign and data that showed Trump would likely do better in the Electoral College than the popular vote. In the French presidential election last month, the conventional wisdom was irrationally worried about a Marine Le Pen victory even though she trailed Emmanuel Macron by 20 to 25 percentage points. In fact, it was Macron who beat his polls, winning by 32 points.”

If Labour pulls off an upset, then, it won’t be because of some law of political punditry. It’ll be because the polls are wrong, and very significantly so. And that’s not a given.


Given all the uncertainties here, meaningful analyses will probably have to wait until the actual results are out. We cannot fully evaluate Labour’s surge until we know where the dust has settled — did they increase their vote share, or their vote totals? Did they win or lose seats, and where? What was youth 18–24 turnout like?

If you want a prediction for which government you’ll wake up to on Friday, just stick with the exit polls — they’re your best bet.

Nicholas is a law graduate, and is currently pursuing a Master’s degree in international public policy.








Leave a Reply

Your email address will not be published. Required fields are marked *