By Bagehot
LAST year’s general election was not a happy experience for British pollsters. Throughout the short campaign, they overwhelmingly claimed the race was very tight. The press dutifully reported this consensus. “Well hung”, ran a Sun headline; “It couldn’t be closer”, asserted the Guardian; it was “neck-and-neck”, I wrote for The Economist. Nonsense, it turned out: on May 7th the country gave the Conservatives their first majority for 23 years.
How had the pollsters got it so wrong? Several explanations have since emerged. The first: there were more “Shy Tories” than had been anticipated. This term, which first arose after another surprise Conservative triumph, in 1992, refers to voters who feel slightly embarrassed at voting for such an untrendy party, so do not admit to pollsters (or perhaps even themselves, until confronted by the ballot paper) that they trust it more than the alternatives.
The second theory is that there were too many online polls. These are cheaper and easier than phone polling—so popular with story-hungry newspapers—and are more likely to elicit a “don’t know” response (talking to an actual person, people feel under more pressure to commit to one side). This can obscure an instinctive inclination towards loss aversion and caution.
The third theory is that pollsters had not sufficiently corrected for the pro-Labour bias of those voters easiest to reach. The sort of younger, more politically active Britons prone to take online polls tended to be left-leaning. Meanwhile Tory voters tended to be busier—out at work or occupied with children—so trickier to pin down over the phone.
What unites these three theories is the observation that certain Tory-inclined voters, for structural or conscious reasons, were political introverts for the purposes of the polling. Which prompts the question: could something similar be happening in the current EU referendum campaign? The recent days have brought some evidence suggesting so; with the introverts, this time, being Remain voters.
Last night, for example, NatCen, a social research body, published an experimental poll designed to avoid the flaws in conventional methods. It used new means: rather than inviting people to volunteer, the pollsters picked respondents at random to curb self-selection bias. Voters who did not respond to initial contact online received follow-up phone calls, to ensure that not just the easiest-to-reach were being polled. Projected propensity to vote based on demographic data—not always the same thing as reported propensity to vote—was factored in. Though the polling was carried out during a period (late May and early June) in which Leave appeared to be storming ahead, it puts Remain on 53% and Leave on 47%.
If, as this indicates, some of the polling over the past months has overstated support for Leave, that is borne out by a study released on June 17th by BMG Research. This suggests that pro-Brexit voters, like Labour supporters in last year’s election, are easier to reach. Among voters who responded to pollsters’ first call, Remain had a lead of 1.1%. Among those who required a second call, it was 5.6%.
One more straw in the wind: the overall trajectory of the polls. In the final week of the campaign there has been a clear, if not overwhelming, tilt towards Remain. The Economist’s poll-of-polls now puts it ahead for the first time since May 23rd. Among voters “certain” to turn out, a poll by ORB this morning has Remain on 53% (up five points) and Leave on 46% (down three). Perhaps most encouraging for the anti-Brexit campaign: YouGov’s polling has seen a sudden jump in the proportion of voters who think Brexit would leave “you personally” worse off.
What these could show is that the “undecided” and “Leave” columns of previous polls contained lurking “Brintroverts”: voters who over the past months would default to a fashionably “common sense” Eurosceptic answer, perhaps based on glimpsed tabloid headlines, when put on the spot by pollsters but now, as polling day nears, are engaging with the choice and breaking towards remain. It is easy for commentators to assume that ordinary folk have, like them, obsessed over every twist and turn of the campaign—and thus to put too much store by polling conducted weeks or months before the actual vote. It may be that warnings, like that by Barack Obama on his visit to London in April, which did not register immediately in the polls, did lodge in voters’ minds and are now coming to the fore.
To be sure, a Leave vote on Thursday is still eminently possible. Remain’s lead in our poll-of-polls is only one point, with 11% of the electorate still undecided. In today’s encouraging ORB poll its seven-point lead falls to two points once all voters (rather than just those certain to cast their ballots) are included. Moreover, for all that they can try to correct for the errors that so embarrassed them last May, pollsters are in uncharted territory. The EU referendum is not a general election; it is just the third nationwide plebiscite Britain has held. Perhaps there are also some “Shy Leavers” out there: well-educated or young folk who do not like to admit they are siding with Nigel Farage. There are many other big, hard-to-predict factors, like differential turnout (will younger voter participate in sufficient numbers?), to take into account. Still, the Brintroverts give Remain campaigners tentative grounds for optimism.
