The Polletariat: is poll fever starting already?

Cameron blamed his ulcers on them and Farage loves a good one. As opinion polls dominate the political story, Andrew Neather analyses the prediction business
Poll dancers: in this race, all parties will be watching the polls more nervously than ever before
Andrew Neather22 October 2014

These are strange times in British politics. Last week Douglas Carswell took his seat in the Commons as a Ukip MP. In the wake of his huge by-election victory, one national poll gave Ukip 25 per cent of the vote. But whoever you vote for, the pollsters win: YouGov recently reported a 12 per cent increase in adjusted operating profit to £7.7 million in the year to July; the0 company’s annual sales are now £67.4 million.

On the night of the Scottish referendum last month, YouGov correctly predicted the result — though 11 days earlier it sent shockwaves through Westminster by putting the Yes campaign ahead for the first time, 51 to 49 per cent. From then until 18 September the polls were the big story. “I want to find these polling companies and I want to sue them for my stomach ulcers because of what they put me through,” David Cameron later complained to former New York mayor Michael Bloomberg.

Commentators agree that May 7’s general election is the hardest to call of recent times. Is poll fever starting already?

Next year will be different from 2010 in part because of the sheer volume of polls. The rise of internet polling has dramatically cut the cost of such surveys. In June 2009, for example, there were just six national opinion polls published; now there are 40 or more a month, with 10 organisations polling.

Since three months before the last election YouGov has published a daily poll; there are also regular ones from Populus, ComRes, Opinium and Survation. (Northern Ireland, with a different party system for its 18 Commons seats, is excluded.)

The polling is more detailed too. Fewer than a third of the 650 Commons seats change hands, and these marginal seats follow national trends only up to a point, in part because the majority are two-horse races. Only detailed local polling can show whether key marginals will buck the national trend or not.

“Marginal polling is hugely important,” says Ipsos MORI’s Ben Page. Yet at least until 2010, little was conducted: to poll in a string of constituencies with a decent sample size in each — around 1,000 respondents — is very expensive. But now it is more common than ever, with regular polls in key constituencies by Survation and by billionaire Tory peer Lord Ashcroft.

Ashcroft in particular has transformed our knowledge of what is going on in corners such as North Cornwall and Sutton & Cheam, covered in his report last month of 21 key Lib-Dem battlegrounds (lordashcroftpolls.com). Though he is a dedicated Conservative, many of Ashcroft’s recent findings have not made comfortable reading for fellow Tories.

Any poll is a snapshot rather than a prediction — but still we want them to predict the result. Should we? Most pollsters are commercial organisations that stand or fall by the accuracy of their predictions. Still, accurate sampling is fraught with difficulty.

“Are the people who are prepared to talk to a pollster for five minutes the same as the ones who will vote?” asks the University of Southampton’s Will Jennings. People who are over-polled, especially those on polling panels, can become unrepresentative: just by having to think about it, they become more engaged than the average voter. And up to 25 per cent of votes are now cast by mail, often weeks before polling day.

The polling industry has suffered embarrassments, notably its failure to predict John Major’s 1992 election victory, thanks to the phenomenon of “shy” Tory voters. Pollsters refined their weighting systems to take account of this “desirability bias” — though the 1997 and 2001 elections saw many over-compensate, overstating Labour’s lead.

Such weighting is a crucial part of the way raw polling figures are processed, whether to allow for demographic imbalances — for instance in gender — or to ensure a politically balanced sample. In particular, some pollsters now weight results according to people’s likelihood of voting. Those who haven’t voted before are much less likely to get around to doing so — a factor that hit the Scottish Yes campaign’s result as 16- and 17 year-olds failed to turn out in predicted numbers. So ICM, for example, discounts by 50 per cent the opinions of those who did not vote last time.

Not all polls are public: the most crucial polling is arguably done privately by the parties to test their campaigning. Campaign insiders sometimes boast that their polling shows them doing better than published polls. But YouGov’s Anthony Wells writes: “Polling conducted for political parties should be treated with a medium-sized ocean of salt until you’ve seen the tables with your own eyes.”

YouGov’s Peter Kellner adds: “You can be sure that [Tory campaign mastermind] Lynton Crosby will be conducting a lot of research and testing messages and attack lines.”

And the big unknowns in 2015? First, the showing of the Scottish National Party will be important for Labour. If the SNP continues its referendum rout of Labour’s working-class urban strongholds, that could cost Ed Miliband’s party seats.

Second, where will 2010 Lib-Dem voters go? There has been little movement of voters between Labour and the Tories. The main shift has been away from the Lib-Dems: pollsters agree that at least 25-30 per cent of these voters moved to Labour, though a recent Ashcroft poll suggested that some may be drifting back. The Lib-Dems will probably pick up from current record lows of around seven per cent. But it is hard to see them repeating the boost given them by “Cleggmania” after the 2010 leaders’ TV debates (how long ago that seems).

Most volatile of all is the Ukip vote, on which could well hinge the overall result. “People say they’re going to vote Ukip, but are they?” asks Ben Page. Ukip voters tend to be angry: they are more likely to tell pollsters they are certain to vote. There is also fierce debate over the effect of prompting people with Ukip’s name, which tends to give a higher Ukip result. A majority of pollsters overstated the party’s share ahead of this year’s European elections, where it won 27 per cent of the vote.

In 2010 Ukip’s vote shrank to 3.1 per cent from the 16.5 per cent it won in the 2009 European elections. Now polling 15 per cent-plus, that share will doubtless fall next May, though it is hard to see it doing so by as much, especially if Nigel Farage gets a spot in leaders’ TV debates.

“If Ukip stay at 10 per cent or more, that’s very bad for the Conservatives,” says Kellner.

Polls can get it wrong. In the recent Heywood and Middleton by-election two polls badly understated the Ukip vote — which Lord Ashcroft put down to voters deciding only at the last moment. And all polls have to be read as part of a trend: “You question the one that’s the odd one out,” says Mike Smithson of the influential politicalbetting.com blog.

But in this race, all parties will be watching the polls more nervously than ever before.

WHO'S ASKING THE QUESTIONS?

YouGov

Founded in 2000 by Stephan Shakespeare and Nadhim Zahawi, YouGov has been the pioneer in the UK of internet polling: now almost all pollsters use such techniques. Made a famously accurate prediction of the 2008 London mayoral race and of the 2012 US presidential election.

Ipsos MORI

The second-biggest pollster, formed by the takeover by Ipsos of Sir Bob Worcester’s MORI in 2005. Now led by Ben Page, a veteran of both corporate and political surveys. Made one of the most accurate predictions in 2010.

Survation

Survation is one of the new kids on the block, founded in 2007. “Really bright people, commercially very sharp,” comments Mike Smithson. Biggest success to date: predicted the Scottish referendum’s result with its last phone poll.

ICM

Founded in 1989, ICM is widely regarded as the “gold standard” in polling, in large part because of its swift adoption of weighting in the aftermath of the 1992 debacle. Did especially well in the 1997 election, with forecasts for Labour’s victory far more modest than the other pollsters.

Opinium

Founded 2007, Opinium uses all-online polling. Biggest success: correctly predicting result of 2012 London mayoral race.

Create a FREE account to continue reading

eros

Registration is a free and easy way to support our journalism.

Join our community where you can: comment on stories; sign up to newsletters; enter competitions and access content on our app.

Your email address

Must be at least 6 characters, include an upper and lower case character and a number

You must be at least 18 years old to create an account

* Required fields

Already have an account? SIGN IN

By clicking Create Account you confirm that your data has been entered correctly and you have read and agree to our Terms of use , Cookie policy and Privacy policy .

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged in

MORE ABOUT