Turkey’s election results prove it – we need to rethink political polls

When it comes to elections, common sense should always prevail over polling results, writes Borzou Daragahi

Like it or not, polls have an impact (Copyright 2023 The Associated Press. All rights reserved)

In the run-up to Turkey’s presidential elections an overwhelming number of analysts, journalists and politicians predicted President Recep Tayyip Erdogan would fare poorly against challenger Kemal Kilicdaroglu, who was slated to win the vote in the first round.

There were analytical reasons for the widespread conclusion – a failing economy, record-high inflation and a flawed rescue and recovery response to the recent earthquake disaster. But by and large, the assumption was based on multiple polls that showed Kilicdaroglu ahead. Even Erdogan’s camp seemed at times perturbed by the polling figures.

It need not have been. After nearly clinching a majority in the first round, Erdogan won the decisive 28 May second round of the election with 52 per cent of the vote. On Saturday, Erdogan was sworn in for another five-year term as president while envoys from 80 countries, including dozens of prime ministers and presidents, looked on.

It was not the first time polls got a major election wrong. Polls got Brexit wrong. Pollsters famously failed to predict Donald Trump’s 2016 victory over Hillary Clinton. But that was within America’s famously fragmented and complex electoral system.

More recently, polls in Greece predicted close competition in general elections pitting the centre-right camp of Prime Minister Kyriakos Mitsotakis against challenger Alexis Tsipras left-wing coalition and several smaller parties. Mitsotakis, with 41 per cent of the vote, retained a parliamentary majority while Tsipras’ mustered only 20 per cent of the popular vote.

Like it or not, polls have an impact. They influence voters’ decisions on whether to stay home or head to the polls. They contribute to decisions by political candidates and parties on how and where to concentrate their efforts. They factor into decisions by editors and producers to assign a story or television news segment. They influence calculations by investors and foreign governments.

Experts in the polling industry have been discussing for years why polls are getting elections so wrong. Among the factors is the increasing weaponisation of polling results.

In Turkey, for example, experts say many of the polling agencies are linked strongly with either the opposition or the government and devise methodologies that subtly produce outcomes that favour their side. Polls conducted in Hungary and several other highly polarised nations are also unreliable for the same reason. Political parties and organisations will often conduct a bunch of polls, but only release those they believe will help their cause.

Turkey and several other countries have large diaspora populations that are excluded from polls. Some 6.5 million still politically engaged Turks live abroad and voted this year in droves, largely for Erdogan. Polls ahead of Moldova’s milestone 2020 general elections also got it wrong for failing to account for the hundreds of thousands of pro-European and anti-Russian voters living abroad who helped elect Maia Sandu, the country’s first woman as president. Pollsters in France inexplicably fail to poll the three million or so voters residing in the country’s 13 overseas territories.

Changing times have produced other methodological deficiencies, explains Maxime Stembach, a journalist at Europe Elects and a data analyst at France’s Science Po. Polls conducted by phone undercount younger voters while polls conducted online undercount older voters.

“You have to use statistical methods to compensate,” says Stembach. “If young people account for 50 per cent of the population and 30 per cent of respondents, you have to use mathematical models to make up the difference. Still, this is not perfect methodology.”

Voters have also changed. Decades ago, families identified with political parties or tendencies. Nowadays new parties and offshoots pop up all the time. Power has become diffused, with national leaders slowly losing their influence to corporations and local leaders. That makes voters more fickle, and unpredictable.

“There has been quite a transformation of the electorate,” says Stembach.  “The number of people who decide who they will vote for in the very last few days of the campaign is growing in most parts of the world. Same with people who abstain. An increasing share of people don’t know if they will vote or not vote on election day.”

Crucially, many of the polls conducted in Turkey ahead of the presidential elections don’t account for the respondents who refuse to answer or say they are not sure how they will vote. They simply divvy up those respondents among candidates based on the proportion of those who reply.

I am intuiting here, but I also think voters these days favour candidates who inspire them without regard to their interests, and often are shy to admit that to family members, friends or strangers contacting them for a poll.

There was a time when people voted for largely empty suits in the model of Emmanuel Macron, Joseph Biden or other technocrats because they promised prosperity. But no one trusts the suits anymore when they promise to restore the economy, and pollsters are largely viewed as part of the same resented establishment.

Those who support Erdogan, Trump, Brazel’s Javier Bolsonaro, Hungary’s Viktor Orban and other such right-wing populists vote for them because they are entertaining. They get the mob riled up. They have a way with words that appeals to the same largely, but not exclusively male demographic that would lie about their voting intentions. Let’s call it Pepe the frog factor, in reference to the icon used by Trump supporters to denote a certain type of mischievous, angry right-wing mindset.

Polls conducted during less politically volatile times may have more value in assessing the mood or direction of the electorate. One such poll conducted in 2020, foreshadowed the recent election results in Turkey. That poll showed half of Turks consider themselves religious and nationalist, the same brew that has helped Erdogan to victories since 2015. Only a quarter identified with the secular tradition of Mustafa Kemal Ataturk, the founder of modern Turkey and of Kilicdaroglu’s Republican People’s Party. Less than 15 per cent identified as liberals or leftists.

When it comes to elections, common sense should always prevail over polling results. Hindsight is always 20/20, but there were sceptics who warned that Erdogan was well positioned to win, and that polls were getting it wrong. Erdogan had all the traditional advantages of incumbency, as well as control of the media and state resources.

A speech or campaign promise by a politician ahead of a vote might be dismissed as ignorable noise. But to many, a poll conducted by a seasoned firm or respected university has the imprimatur of science and fact. Perhaps journalists should stop featuring polls so prominently, and voters should just dismiss them as just noise.