The party, on 52 per cent, is one point lower than in last week's poll, with the Conservatives down three on 27 per cent. Too much should not be read into these changes, which could partly reflect the shift in the Liberal Democrat figure, up from 10 to 14 per cent - more in line with other polls.
A poll by Opinion Research Business, a breakaway from Gallup, for yesterday's London Evening Standard showed a similar Labour lead of 24 points. The trend in polls is significant since the Wirral South by-election two weeks ago. Comparing three post-Wirral polls with the same polls before the by-election, Labour's lead widened by an average of five points.
With only seven weeks to go, a Tory party official sounded a faintly despairing note: "The longer we delay the election, the more the gap will close."
John Major must hope that the polls are as wrong as they were last time - and more. In 1992, the Tories won by a margin of 7.6 percentage points, rather than the one-point average Labour lead shown by the final opinion polls. But it seems unlikely that they will be as far out this time.
The debate among pollsters over the past five years has been occasionally bad-tempered and partly inconclusive.
Most pollsters accepted that they would have to overhaul their methods, with Nick Sparrow, of ICM, leading the way with instant research into the causes of the failure and experiments with new techniques. Bob Worcester, of MORI, was most defensive, but MORI has also changed its methods.
It is now generally agreed that there were four causes of the polls' error in 1992, although no one can say how important each one was. One has definitely been solved; the other three have been tackled in various ways, but no one will know until 2 May how successful these changes have been.
1.Out-of-date census-type information. All polls adjust their findings to make up for the fact that they have interviewed too few men, too many pensioners, too few rich people, and so on. This is called weighting. In 1992, the pollsters thought the electorate was more working class than it actually was. NOP's unweighted figures, for example, were more accurate than the figures they published.
2. Late swing. Some voters changed their minds in the last few days last time, or on election day. The main movements were from Labour to Liberal Democrats, and from Lib Dems to Tories. This time all pollsters are likely to call back on the telephone, the day before polling day, to people they have already interviewed to check if they have switched, which will give clues to any late swing.
3. Stay-at-homes. Tory voters were more likely to turn out last time: a factor hard to separate from a late swing. All the pollsters can do this time is ask people how certain they are to vote, although all the evidence is that it is the Tory vote, rather than Labour's, that is "soft". Our poll today finds 62 per cent of Labour voters "certain to vote", against 48 per cent of Tories and 35 per cent of Lib Dems.
4. Shy Tories. This is the really tricky problem: in 1992, Tory voters suddenly did not want to take part in opinion polls or, if they did, tell pollsters how they would vote. We might guess that, with Margaret Thatcher gone, there was no counterbalance to the fact that Tory voting had come to be seen as selfish or uncaring. With the Tories now breaking records for unpopularity, it is assumed that this "conspiracy of silence" still exists.
It is difficult to adjust for this, especially for those who refuse to take part in opinion polls. But the evidence from large random surveys conducted over long periods is that roughly equal numbers of people remember voting Labour and Tory at the last general election (people tend to revise their memories according to their present views). This might be a way of checking the reliability of polls.
In our poll today, Labour has a one-point lead in how people say they voted in 1992.
Liberal Democrat 14%
The Independent/Harris Poll results.
Harris Research interviewed 1,040 adults face-to-face in their homes between 7 February and 10 March.Reuse content