To adapt Winston Churchill’s comment on democracy, opinion polls are the worst means of finding out what voters think – except for all the others. They are not perfect, for even the best-conducted poll is subject to sampling error. But they are far better than any other way of following the ebbs and flows of an election campaign; and YouGov is breaking new ground in the current campaign by polling fresh samples of the electorate every day for the Sunday Times and the Sun.
Already our polls have shown which events in recent weeks have changed the public mood, and which have left voters unmoved. In February, Gordon Brown’s television interview with Piers Morgan briefly lifted his, and Labour’s, standing; and the Conservatives’ recent pledge to cancel much of next year’s planned increase in national insurance gave a clear boost to David Cameron.
On the other hand, voters shrugged off stories about Brown’s temper, Lord Ashcroft’s non-dom tax status and the willingness of ex-Cabinet ministers to make money from their experience and contacts. In each case the public’s reaction has been: so what?
Abolishing the laws of probability?
We know this because each day YouGov is polling at least 1,400 members of its almost 300,000-strong internet panel. Five years ago, our final election poll for the Sunday Times predicted the national vote share precisely – Labour 36%, Conservatives 33%, Lib Dems 23%. In the 2008 London elections, YouGov’s final survey forecast that Boris Johnson would defeat Ken Livingstone by 53-47% - which is exactly what happened, to the chagrin of other polling companies that showed Livingstone ahead. And our prediction of the result of last June’s Euro-election came closer to the final result than any other company. (For those who care about these things, we also foretold Will Young’s victory in the first Pop Idol, and Joe McElderry’s triumph in last year’s X Factor.)
That does not mean we have abolished the laws of probability. Sampling error remains a risk. Nineteen times out of twenty our figures should be accurate for each party to within two percentage points; but one poll in twenty is likely to be less accurate. And it’s vital to remember that more pollsters than ever before, employing a wider range of methods, are taking the pulse of the electorate. This election will not only determine the future course of Britain: it may also tell us which polling methods work best.
Pending the result, experienced poll-watchers know not to put too much weight into a single figure in a single poll. To obtain a full, rounded picture of what is happening, it is better to look at a range of questions and a range of polls.
Changes and blips: the advantages of daily polling
That is another reason why YouGov is polling daily. When we show a sudden change, we cannot be absolutely certain at first whether it reveals (a) a sampling freak, (b) a blip – a real but short lived change or (c) a lasting change – or some mixture of the three. The ones that matter are changes that last beyond a single 24-hour new cycle.
Only daily polls allow us to detect these. With weekly polls, it’s impossible to separate such changes from blips and anomalies until it’s too late. For example, last week’s YouGov/Sunday Times poll, showing a ten-point Conservative lead, the highest since January, could have been dismissed as a rogue poll, had it not extended a trend detected earlier in the week, with the Tory lead growing steadily from four points to eight points in the days following Alistair Darling’s Budget.
A saga in three-and-a-half acts
So: taking our polls since the beginning of the year, what has been happening? The saga has had three-and-a-half acts so far. In Act I, the Tories entered election year with a lead of around ten points- much where they stood during the second half of 2009. Act II opened with the news, in late January, that Britain’s economy had started to grow again. As the media began to talk less of recession and more of recovery, the gap between the [arties started to narrow, and the Tories’ lost their advantage as the party thought best able to run the economy.
Act III started a little over two weeks ago, on Budget Day. Immediate public reaction was mixed, bit it grew more hostile as the days passed. With their promise to cancel the National Insurance increase for millions of workers, the Conservatives scored a palpable hit; although the endorsement of a number of business leader may well have been equally important in persuading voters that the Tories could be trusted to run the economy effectively.
Act IV began last week, with the start of the campaign, and so far the Tories appear to be maintaining the lead they established in Act III.
What could happen to change the voters’ minds – either to ensure a working overall majority for Cameron, or to bring Brown back into the race? The nature of the first three acts suggests that the biggest influence by far is likely to be the economy. Almost six out of ten electors say it is one of the top issues influencing their vote – compared with barely a quarter at the same stage of the last election.
This means that one of the most important moments in the current campaign will be 9.30am on Friday April 23. This is when official figures for the state of Britain’s economy in the first quarter of 2010 will be unveiled. Further signs of growth will cheers Labour – but evidence that we are sliding back into recession with add credence to Tory attacks on the Government’s record.
Televised debates: gaffes and punches
The other possible game-changer will be the leader’s television debates. It is possible that Brown, Cameron or Nick Clegg will land a lethal punch on one of their rivals. But the likelier cause of an upset is a gaffe that gets replayed in new bulletins and mocked by the press. Otherwise, the biggest opportunity lies with Clegg, to exploit media exposure on equal terms with Brown and Cameron. Will the Liberal Democrats gain votes as a results – and, if so, at whose expense?
Like other pollsters, YouGov will seek answers to that question, and two others that will matter between now and May 6. First, will the Tories do better in their target seats than in the rest of Britain? In recent weeks, YouGov, Populus and Ipsos-MORI have all found swings up to two points higher than the national average in Labour-held marginals. If this pattern is repeated on polling day, then this will offset some of the pro-Labour bias in Britain’s political geography. My guesstimate is that the Tories will need a 3-4% lead in the popular vote to become the largest party, and a 7-8% lead to secure an overall majority.
The second question is whether Labour can expect a “traditional” swing back to the government as election day approaches. Probably not: the “tradition” is a myth. (The best-remembered example, 1992, is deceptive. True, the Tories retained power after the polls pointed to defeat; but the polls were wrong throughout. There is no evidence of movement during the course of the campaign.) If Labour is to seize victory from the claws of defeat, it must work hard to bring it about, not assume it will happen naturally.
This commentary first appeared in the Sunday Times on April 11