The Sun can fix the scale of what it describes as new Labour’s “record wipe-out” of the Conservatives with precision. It has determined that Blair’s already “massive majority” will increase by 48 seats to 227. Not by 50 or so seats to 230 or thereabouts – approximation is an unscientific vice. But by 48 seats to 227. The Tory party is on the “road to oblivion”, said the Sun on 2 May. “There is no doubt that on the findings of this poll, Labour would increase its majority,” said Bob Worcester, the Sun‘s pollster.
Note the faintly bathetic and rather sly qualification of “on the findings of this poll” – is Labour going to increase its majority or isn’t it? – and then consider the predictions of rival Mystic Megs. Gallup, for the Mirror, also polled at the end of April, and put the Labour majority at 250. The market researchers of ICM, working for the Guardian, were busy at the same time, and concluded that Blair would have a majority of 165.
A discrepancy of 85 seats is a rather wide “margin of error”. The polling companies cannot all be right. But they might all be wrong. They were all wrong in 1992, when the final polls gave Labour a lead of 1.4 percentage points. The Conservatives returned to power 7.6 points ahead of Labour. The blundering was not as bad in 1997, although MORI, Harris and NOP were as wrong then as they were in 1992, as John Curtice from Strathclyde University has pointed out. After the last general election, all the pollsters returned to their slapdash ways. No polling company antici-pated Labour’s defeat in the European elections or its failure to win a majority in the Welsh Assembly. In the contest for the London Assembly, Labour was well ahead of the Tories in the polls, but they were neck and neck in the ballot boxes.
Given this dismal record, why do newspapers and politicians bother with polls? What facts do they think they are revealing? The questions used to be urgent. The New York Times once had long debates on the morality of putting polls on the front page and dignifying them with a reporter’s byline. In 1992, John Birt, then director general of the BBC, showed he was wiser than many of his political correspondents when he ordered that the results of newspaper polls must never lead the news in the election campaign.
Now the New York Times regards polls with the same credulous gaze that the Romans once threw at the innards of chickens. The Today programme doesn’t merely report other people’s polls, but commissions its own and treats the results as the most important “news” of the morning. Since Labour assumed office in 1997, expenditure on polls has trebled in many ministries. Sensitive findings – on visitor reactions to the Dome, for instance – are never released to parliament. Polls are so important that they are official secrets.
The polling mania is a continuous affliction. But it is at its most preposterous in an election campaign. The rational mind would surely view predictions from organisations with a record of mistakes as superfluous. We are weeks away from being put out of our misery and knowing, once and for all, the composition of the next House of Commons. Yet even at 10pm on election night – when the first returns are less than two hours away – various Dimblebys on various networks will announce the conclusions of the exit polls collected by their minions at considerable expense. These polls will be discussed as if they are somehow on a par with real results from real constituencies.
Understanding the appeal of such quackery is not a simple matter. We must go through various states of mind, from the psephologically challenged via the politically manipulative to the psychologically flawed. A constant and unpleasant feature of my life as a hack is the moment, in desultory forward-planning meetings, when the editor snaps: “Anyone got any ideas?” Reporters slide down in their seats. Middle managers duck his glare and find an unexpected fascination in the pattern of the carpet tiles. The few stories that are stuttered out are old, boring or untrue – often all three at once. The editor wonders what on earth he has done to be stuck with a job lot of clueless deadbeats and admits defeat. “All right,” he sighs, “I suppose we’ll have to have a poll.”
Polls fill the intellectual vacuum. They have an air of authority. No wonder editors are prepared to pay for them – even at £8,000 a throw. Best of all, polls guarantee free publicity on the radio and in television newspaper reviews. Few like to spoil the party by saying that, whenever an election allows their accuracy to be audited, they are invariably wrong.
The gold standard for statisticians is the random sample. Lose randomness, and bias creeps in. Random samples, however, are costly. According to Roger Jowell, who organises the British Social Attitudes Survey, they are twice the price of the media’s polls. Addresses are picked at random. If the occupants are out, interviewers go back. If the occupants refuse to answer, they are asked to reconsider. Time is as important as money in these circumstances. You cannot have an instant random poll.
The media pollsters got the result of the 1992 election spectacularly wrong because they interviewed quotas of different types of citizen in the streets – X number of lower-middle-class women, Y number of upper-class men. If one lower-middle-class woman said she was too busy to answer questions and walked off, the interviewer tried another and another until they found one who would talk. The pro-Labour bias was in the difference between the characters of the talkers and the walkers. To generalise, but only slightly, Labour supporters tend to be kind and convivial people who are more than happy to help a stranger who stops them on the high street. Tories tend to be sociopaths with the soul of a rat who hate giving something for nothing. Hence the Labour poll lead and election defeat.
All the companies claim to have improved their techniques since 1992. All have failed to get the results of local, regional and European elections even half-right. If I were a chief executive spending a fortune on probing the tastes of wealthy (Tory) consumers, I might reflect on that failure for a moment.
Since the introduction of polling in the 1930s by George Gallup, the left has come up with an explanation for the love that the establishment has for error-strewn polls: far from providing a means for the governed to instruct their governors, goes the theory, polls help the elite keep control of the masses.
Popular demands to renationalise the railways, say, or to tax the plutocracy, can be ignored as long as pollsters can say what populist gestures on crime and asylum will allow the snobs to pose as the comrades of the mob. Doubters can be pointed to the support for polling among the powerful. Of itself, it gives the lie to the claim that polls are instruments for broadcasting the popular will. Peter Mandelson notoriously preferred polls to the parliaments and congresses of the age of representative democracy – which was, he said without regret, dying. Pollsters themselves don’t behave as if they are humble interpreters of the democratic will. Bob Worcester has become a pundit of gravitas and bottom. Lou Harris, of the eponymous firm, once boasted: “I elected one president, one prime minister, about 28 governors, and maybe close to 60 US senators.”
As with any plebiscite, control rests with the man who sets the questions. The supposedly sovereign citizen has little room for manoeuvre. If an opponent of Tony Blair were asked if he thought the PM was a “strong leader”, he might reply that Blair obviously dominated his party, but was feeble in his servility before the conservative interests. If he did say this, his complex views couldn’t be recorded.
Simplification shifts with ease into manipulation. “Push polls” allow special interests to present their prejudices as fact. If a politician wants to establish himself as the only serious candidate for his party’s nomination, polls showing him to be a “credible” leader can stop his rivals before their campaigns start and their policies are granted a hearing, as Gordon Brown knows to his cost.
American privatisers produced a series of polls for the compliant business press in the Nineties which “proved” that the young cared more about the future of soap operas. Journalists then concluded that the young had no objection to privatising social security. The tactic was no different from the scam pulled by the News of the World after its paedophile campaign produced assaults on the innocent. The paper justified itself with a MORI poll showing that 82 per cent of the population wanted a “Sarah’s law” to warn neighbours of the presence of child abusers. Only Alan Travis of the Guardian took the trouble to look on the MORI website. He found, buried in the small print, that 51 per cent of those questioned did not think that the News of the World was right to pursue its naming-and-shaming policy.
The scandals and blundering will not destroy polls. Even if the Labour majority is reduced to 100, the Gallups and the ICMs will plough on. The urge to believe in them reveals that many fellow citizens are the western European equivalent of the Siberian hunter-gatherers who prostrate themselves before their shamans. It would be funny, if so much damage on public debate wasn’t inflicted in the process.
Polls replace argument on what ought to happen with fantasy predictions on what will happen. The voice of the soothsayer drowns out the debates elections are supposed to encourage.