From 1992 … Why the UK pollsters got it wrong

May 8, 2015

Pollsters failed dismally to predict the strength of support for Prime Minister David Cameron’s Conservatives in this week’s national election — just as happened in 1992 when John Major scored an unexpectedly decisive victory. Here’s Reuters’ take on why the polling industry got it so wrong, 23 years ago.

Why the pollsters got it wrong

By Maggie Fox

10 April 1992

LONDON, April 10, Reuter – Pollsters, egg dripping from their faces on Friday after failing to predict the Conservative Party’s election victory, blamed a last-minute change of heart by voters for the surprise.

Opinion polls taken up to the day of the election had predicted a very close race. Even exit polls, taken as voters leave polling booths, showed the race as neck-and-neck.

Prime Minister John Major, accompanied by wife Norma (L), acknowledges the applause following his sp..Pollsters admitted they were confounded by the unexpectedly strong Conservative showing.

“It’s a very big embarrassment, no doubt about that,” said National Opinion Polls political research director Nick Moon. “One possibility was that we were polling in the wrong places.”

Alan Terry, research executive for the MORI opinion poll, said voters may have been frightened by opposition Labour Party proposals to raise taxes by as much as 19 per cent.

“When they actually stood there with their ballot paper and their pencil, tax rose up in their minds,” he said.

Patrick Dunleavy, professor of government at the London School of Economics, agreed this was possible.

“It could have been a last-minute shift,” he said. “It’s possible that previous polls showing Labour leads were right.”

But he also pointed to Conservative claims that support for their party was underestimated throughout the campaign.

“I think we need to look seriously at where the Conservative support was,” he said.

Pollsters were conducting post-mortems, meeting with one another on Friday and checking back with those surveyed.

“We are currently carrying out two surveys, going back to people we surveyed before, finding out whether they voted, whether they changed their minds and when,” said Brian Gosschalk, director of political research at MORI.

Dunleavy said the polls had correctly indicated a trend away from Labour support. Polls earlier in the month had shown a Labour lead of several points.

But this could not explain the failure of exit polls to predict the Conservative victory.

Exit polls taken for two newspapers and for Sky television predicted Labour would be the biggest party but without a majority. The British Broadcasting Corporation had the Conservatives winning, but not with an overall majority.

Only a Harris exit poll for Independent Television News came close. It predicted the Conservatives would win 41 per cent of the vote and Labour 37 per cent. The Conservatives ended up with 42.2 per cent versus 35.2 per cent for Labour.

The most accurate poll was one of the most unorthodox. The Muffin Poll, conducted at a bakery in Manchester in northern England, kept its record for 100 per cent accuracy, held since 1964.

Muffins are dyed blue for sale to Conservative voters, red for Labour and yellow for the Liberal Democrats. The poll showed the Conservatives ahead throughout the campaign.

“It goes to show when the customers get their teeth into an election it is far more accurate than just stopping them in the street,” said baker Sam Ward.

No comments so far

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/