It was too close to call right up until polling day, when Alexis Tsipras won a resounding victory.
Greek voters became the latest electorate to blindside pollsters on Sunday when they put the SYRIZA leader back into power by a margin forecast by no one. Following surprise results in the U.K. general election in May and in Indonesia a year earlier, the Greek outcome adds to growing evidence that political polling is in crisis.
That has ramifications far beyond the political arena. Surveys of voter intentions move financial markets – think the pound and Scotland’s independence referendum one year ago – at a time when they have rarely been less reliable. As pollsters go back to the drawing board to figure out where they went wrong, their work has never been under more scrutiny.
“One of the biggest challenges we face is predicting whether people will actually vote,” said Ben Page, the chief executive officer of British pollster Ipsos Mori. A widespread decline in loyalty to the established political forces makes matters even more complex, Page said. “Every successive generation has become less likely to always vote for the same party.”
SYRIZA, an acronym for the Coalition of the Radical Left, is a case in point. After breaking the 40-year dominance of Greece’s two main parties in January, SYRIZA took 35.5 percent of the vote on Sunday versus 28.1 percent for New Democracy. A poll of polls compiled by Bloomberg averaging the findings of 16 Greek pollsters on the eve of the election gave Tsipras’s party about 29 percent, with one in ten respondents still undecided.
Voter apathy might be the biggest wildcard. At 56 percent, turnout in Greece was the lowest since at least the 1990s. Others just don’t get reached, particularly as younger voters opt for mobile phones over landlines, while the prevalence of online questions might skew the sample polled.
The final tally in Greece showed that polling companies accurately predicted support for New Democracy and for some smaller blocs including the Communist Party. Yet they failed to forecast the performance of either SYRIZA or of Independent Greeks, the two parties that formed a repeat coalition. The polls overestimated support for The River, which was mooted as a potential SYRIZA coalition partner, and the SYRIZA splinter group Popular Unity, which failed to cross the 3 percent threshold to win parliamentary seats.
“We may need to change the practices that we have been using for the past 25 years,” said Costas Panagopoulos, head of Greek pollster Alco, whose last survey had SYRIZA taking 27.4 percent. “It seems that a part of society thinks that opinion surveys are tools of the establishment and refuses to participate.”
With a political environment in “flux,” voters switch allegiances if they vote at all and often lie to pollsters, according to George Arapoglou, general manager of Greek pollster Pulse RC. Panos Kammenos, the Independent Greeks leader who served as defense minister in Tsipras’s last coalition, derided the polls as a “huge fraud.”
The Greek surprise came on the back of a more spectacular failure in the UK, where Prime Minister David Cameron’s Conservatives defied predictions of a hung parliament to win a surprise majority. May’s election was the most intensively polled ballot in British history, yet no published poll came close to predicting the result.
A yearlong inquiry, funded by the polling companies, is trying to establish what went awry. One pollster found their sample, for example, over-represented disabled people, and under-represented those with tumble-dryers.
“There’s a perennial clash between news value and what makes good polling,” said John Curtice, professor of politics at Strathclyde University in Glasgow. “If you want representative samples, it takes time. People want things instantly. It’s just a more and more difficult job to do.”