A recent suggestion by the Market Research Society, the pollsters' professional body, that polls might have underestimated the Tories' share of the vote for decades drew sharp responses from polling organisations at the time.
ICM's review, covering the first half of this year, says clear 10 and 8 per cent leads for the Conservatives over Labour shown by their May and June polls are signs that there may not have been a systematic Labour bias before the election. 'Usually, such sudden movements in party fortunes follow some dramatic development,' the report says. 'In this case the only event was the election itself.'
This, as ICM accepts, still leaves a number of possible reasons why poll predictions - virtually all suggested a hung parliament with Labour the largest party - were so wide of the mark.
ICM's analysis of recall interviews with people polled during the election campaign attributes 37 per cent of the discrepancy to failure to vote, late vote-switching, late deciders and 'secret' voters - those unwilling to tell pollsters how they would vote.
Jennie Beck, an ICM researcher, said the organisation was studying poll selection criteria. Most pollsters use age, class, sex and working status to construct representative samples but factors such as housing tenure and newspaper readership might also be significant.
The organisation is also attempting to identify the kinds of people who refuse either to reveal their voting intentions or to answer any questions at all. It found a disproportionately high percentage of secret voters were Tory supporters. ICM suggested they were unwilling to vote for higher taxes, but not keen to admit it.
Dislike of Labour's tax plans is the reason offered by the review for vote-switching on polling day, a theme hammered home by Michael Portillo, Chief Secretary to the Treasury, in a pamphlet, The Economics of John Smith.
It said Mr Smith had 'showed no understanding that the impact on the spending of the middle to high earners . . . would not have been balanced by the tiny improvements in the incomes of the less well paid, nor by the higher level of public spending.'Reuse content