LONDON — BRITAIN'S public opinion pollsters are reexamining their role after getting the result of the April 9 general election completely wrong.
Amid calls to ban opinion polls before the next general election, the five main polling organizations have set up a special investigation unit. Its task will be to examine why, instead of predicting the Conservative Party's eventual victory, they all anticipated an election too close to call.
Robert Worcester, chairman of the Mori organization, which carried out surveys for the London Times, said large numbers of voters had made last-minute switches from the opposition Labour and Liberal Democrat parties to the ruling Conservatives.
Mr. Worcester denied claims by politicians and analysts that many voters had lied about their intentions.
The scale of the pollsters' error has jolted the leaders of a 300 million pound industry to such an extent that their collective organization, the Market Research Society, has ordered its director, Susan Stoessl, to carry out an urgent inquiry.
Some industry leaders worry that companies using market research to assess public reaction to their products will lose faith in the polls' reliability.
Of five polls published on voting day only Gallup put the Conservatives ahead, but their lead was well within the poll's margin of error. Of the other polls, NOP suggested a 3 point Labour lead, followed by Harris Research Center (2 points), Mori's 1 point, and ICM's dead-heat. All five polls were conducted within 72 hours of voting day.
In the end, the Conservatives under Prime Minister John Major gained a working parliamentary majority of 21 seats.
"Never before has every single poll strayed so far from reality," said Peter Kellner, an election analyst. He called for an "urgent post mortem" on the reason for the polls' fallacy.
Worcester, whose organization swung into action with a series of "reinterviews" immediately after the election result was known, turned up one of the first clues to what underlay the pollsters' collective error. Mori found that 11 percent of Liberal Democrats who, a week before polling day, said they would support their party, failed to do so on April 9. Among Labour Party supporters, 4 percent decided to back the Conservatives. A further 8 percent voted Liberal Democrat.
"No fewer than 8 percent of voters said they did not make up their minds until the final 24 hours of the election, and another 13 percent said that they decided during the final week," Worcester says.
Derek Conway, a Conservative member of Parliament who was reelected, has called for laws to ban opinion polls in the final two weeks of general election campaigns. "They distort the democratic process and ensure that rather than debating policies, the competing parties bend like reeds to perceived changes in voting intention," he says of the polls.
A similar theme was taken up by Lord Jakobovits, a former British chief rabbi. "Extrapolating what tens of millions are thinking from a tiny sample of opinions affronts human intelligence and negates true freedom of thought," he says.
Steven Barnett, director of the media group at the Henley Centre for Forecasting, agrees: A "relentless diet of statistics about alleged voting intentions" in the run-up to polling day, "undermined party moods, journalists' questions, and campaign strategies."
"Almost certainly the polls themselves determined how some people cast their votes," he says.
He called for greater responsibility by the media in the way polls are used, and "recognition of the inherent weaknesses in the results that are produced." Some say more work should be done in framing the questions put to people interviewed in opinion surveys.
Harris's Robert Waller speaks of a "qualitative difference" between what people say to canvassers and what they do when they enter a polling booth.
Labour's plan to put up taxes, Waller says, had hardly figured as a significant issue during preelection canvassing, but exit polls conducted on April 9 showed that taxes were one of the main reasons for a surge back to the Conservatives.