Pollsters blow it in British election. What lessons for the US?
When Conservatives won a clear majority in British elections this week, it stunned pollsters and pundits predicting a close race. Experts say the art and science of polling is changing, and this could impact next year鈥檚 US presidential race.
President Harry Truman, who won re-election in 1948, beating Republican Thomas Dewey, holds a copy of the Chicago Daily Tribune.
U.S. Information Agency
Britain had its 鈥淒ewey Defeats Truman鈥� moment this week.
Tory leader and incumbent Prime Minister David Cameron defied pre-election polls 鈥� which had predicted a very close race and the likely need for a coalition government 鈥� winning an outright majority.
Pollsters and pundits were stunned.
"Election results raise serious issues for all pollsters. We will look at our methods and have urged the British Polling Council to set up a review," said Populus, one of the main polling firms, on Twitter.
鈥淭here are likely to be variety of reasons behind the difference in the polls and the final outcome,鈥� Populus founder . 鈥淰ery late swing to the Conservatives, polling weightings, polling methodology and claimed propensity to vote will be just some of the factors that are likely to be discovered once an investigation is completed.鈥�
"It's a very, very big miss and at this stage we just don't know why,鈥� British political scientist Rob Ford told Reuters. 鈥淲e're shocked.鈥�
Prominent American political operatives were closely involved in the British election.
Obama campaign veteran David Axelrod聽advised big loser Ed Miliband and the Labor Party while another top Obama aide,聽Jim Messina, advised Prime Minister Cameron and the Conservatives. As the stunning results became clear, Messina tweeted: 鈥淭hings US&UK have in common: completely broken public polling & re-electing their strong leaders.鈥�
"In all my years as journalist & strategist, I've never seen as stark a failure of polling as in UK,鈥� Axelrod聽groused on Twitter. 鈥淗uge project ahead to unravel that.鈥�
For Messina, lessons learned from the polling debacle may be particularly relevant. He co-chairs Priorities USA Action, the outside group that鈥檚 raising money to support Hillary Rodham Clinton鈥檚 campaign. 聽
There are major differences in the two countries鈥� election systems, of course. The campaign period in the UK is far, far shorter, and the kingdom has more political parties winning (or losing) seats under a parliamentary system.
But both political systems rely heavily on polls 鈥� conducted for news organizations and privately for candidates 鈥� and in the US, one of the most respected political statisticians got it glaringly wrong too.
Nate Silver, formerly of the New York Times and now writing his FiveThirtyEight blog as part of his gig with ESPN, was聽聽272 seats for Conservatives and 271 for Labor 鈥� dead even.聽It takes 326 seats for a clear majority, and in the end Tories won big with 331 seats 鈥� 99 more than their main rival Labor.
How does Mr. Silver (who correctly predicted the outcome in the last three US presidential elections) explain this?
鈥淧erhaps it鈥檚 just been a run of bad luck. But there are聽lots of reasons to worry about the state of the polling industry,鈥� . 鈥淰oters are becoming harder to contact, especially on landline telephones. Online polls have become commonplace, but some eschew聽, historically the bedrock of polling methodology. And in the U.S., some pollsters have been caught聽withholding results聽when they differ from other surveys,聽鈥榟erding鈥� toward a false consensus about a race聽instead of behaving independently.鈥�
It鈥檚 something Silver has worried about for some time.
For one thing, he wrote last August, 鈥淩esponse rates to political polls are dismal鈥� 鈥� down to 10 percent from about 35 percent in the 1990s. Also, there are fewer top-quality polls (which are very expensive) as news organizations look for ways to trim budgets.
鈥淭hen there are the companies that have cheated in a much more explicit way: by fabricating data,鈥�
In his current analysis of the British election, Silver notes other recent elections where the polling was off:
鈥� The final polls聽showed a close result聽in the Scottish independence referendum, with the 鈥渘o鈥� side projected to win by just 2 to 3 percentage points. In fact, 鈥渘o鈥� won by almost 11 percentage points.
鈥� Although polls correctly implied that聽Republicans were favored to win the Senate in the 2014 U.S. midterms, they nevertheless significantly underestimated the GOP鈥檚 performance. Republicans鈥� margins over Democrats were about 4 points better than the polls in the average Senate race.
鈥� Pre-election polls聽badly underestimated Likud鈥檚 performance聽in the Israeli legislative elections earlier this year, projecting the party to win about 22 seats in the Knesset when it won 30.
鈥淚n fact,鈥� Silver writes, 鈥渋t鈥檚 become harder to find an election in which the polls did all that well.鈥�