The findings reinforce the suspicion that the MORI poll for yesterday's Times, which showed the gap between the two main parties closing by 12 points in a week, was more a statistical wobble than a political earthquake. Harris's interviews were done from last Friday to Monday, before MORI's, which were all done on Tuesday.
But Gallup, polling on Monday to Wednesday for the Telegraph, has recorded a two-point rise in Labour's lead over the last week.
The Independent/Harris poll contains more evidence of a hardening of the Tory vote, with more supporters (77 per cent) saying they are certain to vote than Labour supporters (74 per cent), but the difference is not significant.
The apparently dramatic drop in Labour's lead shown by MORI is partly because its latest finding is being compared with last week's unusually high Labour lead of 27 points. But polls carried out in a single day are notoriously variable, because they are vulnerable to one-off factors such as popular television events, which make people reluctant to speak to doorstep interviewers.
The other feature of the polls which is puzzling commentators is the consistently low Labour leads posted by ICM, although the trend in ICM's polls is also one of a slow narrowing of the gap. Part of the explanation for the difference between ICM and the rest is that its interviewers do two things before asking how people will vote. First, they ask people how likely they are to vote. "What we are attempting to do is to get people to think more clearly about whether they are going to vote and what they would actually do if they were in the polling station," says Nick Sparrow, ICM managing director. The company's research suggests that up to 6 per cent of those interviewed will happily say who they intend to vote for, while not actually intending to vote, either because they are not registered or because they are not motivated enough.
Secondly, ICM remind people of the names of the three main parties (four in Scotland and Wales), which boosts Liberal Democrat support at the expense of Labour. Next week Harris will change to a similar method by using a showcard listing the parties to ask how people intend to vote.
Mr Sparrow says the profile of the electorate used to weight his data is "exactly the same as the other pollsters" but believes the greater anonymity of telephone interviewers encourages Tory supporters to declare themselves. Although the gap may be narrowing slightly, the movement is slow and Labour is still so far ahead that comparisons with previous elections are difficult to make.
The average of the most recent polls from the five leading polling companies points to a Labour majority of 227 seats; our poll says 229. Yesterday's MORI poll, which showed Labour lead of 15 points, would still give it a 159-seat majority. This week's ICM poll, whose findings are most optimistic for the Conservatives, suggests a majority of 109.
Even if the polls collectively were as far out this time as they were last time, Mr Blair would have a majority of 135. In the last two general- election campaigns, the average leads shown by the opinion polls did not change significantly over the final three weeks of the campaign.
Harris Research interviewed 1,138 adults face-to-face in their homes between 4 and 7 April.
Poll position: How it all adds up
7 Mar 14 Mar 21 Mar 28 Mar 4 April Now
Labour 53 52 56 54 52 52%
Con 32 27 29 30 28 30%
Lib Dem 10 14 10 11 14 12%
Others 4 7 5 5 6 6%
The Independent/Harris pollReuse content