Photographer: Harold Cunningham/Getty Images

Pollsters Flying Blind on Brexit Can Take Solace From Swiss

  • Switzerland’s citizens vote in plebiscites four times a year
  • Pollsters have erred on immigration, ban of minarets

It’s not just British polls that get it wrong. Even in Switzerland, where plebiscites have shaped national politics for over a century, the electoral tea leaves can be hard to read.

Brexit Watch: The pound, the polls, and the probability of Brexit, all in one place

With Britons gearing up to decide whether to quit the European Union -- only the third U.K.-wide referendum since 1973 -- there’s often doubt about whether opinion surveys are credible. Notable snafus in the past have included failing to predict Prime Minister David Cameron’s Conservatives would win a majority in last year’s general election.

The handful of pollsters in Switzerland, where as many as 12 popular initiatives are put to a national ballot each year, share their U.K. counterparts’ predicament. While they have the advantage of historical data stretching back to 1891, they have to glean information in a linguistically and culturally diverse country, making it difficult to gauge overarching national trends.

“Polling for elections isn’t so hard, but referendums are much harder to conduct surveys for, and it’s harder still if you don’t have a point of reference in the recent past,” Claude Longchamp said by phone from Bern, where he heads pollster gfs.bern. “If we have a historical benchmark we can ask people about, we can weight the results. And in the U.K. they don’t have that.”

For a QuickTake explainer on the U.K. referendum, click here.

In Switzerland, a national vote on any topic can be triggered by collecting 100,000 signatures from among the country’s German, French, Italian and Romansh-speaking citizens. Past subjects have ranged from reining in executive pay to expelling foreigners convicted of serious crimes to cutting down on EU immigration. Votes are typically held four times a year and there are also regional referendums in the country that is home to more than 8 million people.

Gfs.bern conducts its polls by telephone, with a sample size of typically 1,000 to 1,500 people. By contrast, the Zurich-based research institute Sotomo favors online surveys. They include questions on how a respondent has voted in the past, which director Michael Hermann contends is a more accurate way of weighting results than simply controlling for variables such as age, gender or place of residence. Still, there are “no magic potions,” he said.

Voters unwillingness to admit to certain views may have been what went awry in 2009 on an initiative to ban the construction of new minarets. While gfs.bern’s survey suggested the measure would fail, with a “yes” vote share of just 37 percent, it passed with support at 58 percent.

Cagey Respondents

“In Switzerland, with questions of European relations or immigration, people tend to be cagey about what they think -- it’s possible that people don’t want to voice their politically incorrect opinions in public,” Hermann said. “With an online survey, you don’t have the effect of social desirability, because you don’t have to speak to someone -- you’re anonymous, you can express whatever you think.”

Online surveying proved more effective than the telephone in 2014, when voters backed quotas on immigration from the EU. Gfs.bern’s last poll before the vote showed opponents ahead.

Gfs.bern’s Longchamp, a regular fixture on national television when plebiscites are held, remains unconvinced.

“I say beware of online questionnaires,” he said. “People have to motivate themselves to participate.”

Sometimes nobody gets it right: In the case of Switzerland’s June 6 vote on a measure designed to improve public services, which would have meant pay cuts for top executives of state-owned companies, both newspaper 20 Minuten’s online poll of 17,495 people and gfs.bern’s phone survey said the result was too close to call. The proposal wound up failing by a margin of roughly two thirds.

For a QuickTake explainer on Switzerland’s people power, click here.

“Having historical reference points helps you put things into context,” said Georg Lutz, a professor of political science at the University of Lausanne. “But that doesn’t mean you have a precise estimate.”

In Britain, telephone-based surveys have tended to put “Remain” in the lead, while online questionnaires generally showed the two sides tied. There’s been a shift recently, with polls this week giving “Leave” the lead. Bookmakers, by contrast, see about a 60 percent chance of voters rejecting Brexit.

With pro-Brexit groups in favor of reducing immigration, at play are some of the same concerns that led the Swiss to opt for EU quotas despite warnings that it might undermine the economy.

“In the case of Brexit, where even representatives of established political parties advocate it, it’s hard to know if people skewing their poll answers due to concerns about social desirability plays a key role,” said Thomas Widmer, professor of political science at the University of Zurich. “I find it hard to imagine that lots of people are too scared to say they support leaving.”

Before it's here, it's on the Bloomberg Terminal.