What if polls prove to be wrong in Turkey too?
The victory of Prime Minister David Cameron’s Conservative Party in the U.K. elections was a shock not only to Labour and the Liberal Democrats, but to polling companies as well.
Up until the last day, even the last hour of the elections, they had estimated a neck-and-neck race between Labour and the Tories, placing the Scottish National Party (SNP) in a kingmaker position. The Hürriyet Daily News’ story based on agency reports also told us that too.
This is the second high-profile polling mistake in a row. The polling companies in the Israeli elections in March also proved badly wrong about the results. Their forecasts had focused on the rise of the newly founded Zionist Union, rather than a new victory by Prime Minister Binyamin Netanyahu’s Likud.
Analyzing the situation, Alberto Nardelli of The Guardian focused on three possibilities in his May 8 article, “How did the polls get it so wrong?” Either people were simply lying to pollsters, or they changed their minds dramatically literally when they got into the polling stations, or there is something old and wrong in the methodology of the polling companies.
Perhaps I can add one more factor. Is it possible that polling companies have started to see themselves as kingmakers who can manipulate the public opinion and the vote?
Polling companies are working hard nowadays in Turkey to make their final estimates, as there is less than a month left to the June 7 parliamentary elections. Their job is much more difficult than in previous elections for three main reasons.
It is not only the winning party and its vote percentage that they have to forecast. There is already general acceptance that the ruling Justice and Development Party (AK Parti) will come first, followed by the Republican People’s Party (CHP) and the Nationalist Movement Party (MHP). The first bet is whether the AK Parti will drop below 45, whether the CHP will get close to 30 percent from 26 percent, and whether the MHP will increase from 14 percent.
The second bet of the polling companies in Turkey is about whether the Kurdish problem-focused Peoples’ Democratic Party (HDP) will exceed the unfair 10 percent threshold and get into the parliament.
The third bet is related to this, as whether or not the HDP enters parliament will change the seat distribution entirely and have a direct effect on whether there will be a constitutional change after the polls to shift to a presidential system, as President Tayyip Erdoğan wants.
The polling companies in Turkey used to get fairly accurate results up until the 2011 election, when credible companies got it wrong but some new companies did rather well.
The current estimates fail to give convergent results. For example, forecasts for the AK Parti vary between 38 and 48 percent and for the HDP between 8 and 13 percent, with an incredible “undecided” constituency of around 10 percent. This is quite unusual so close to the elections.
It is possible that some people are not telling their true choice to pollsters, out of concerns whether that information could get into the hands of the state and cause them problems in future. It is also possible that some may change their minds at the last minute. But it seems necessary for all polling companies to stop seeing themselves as kingmakers and to find new methods, considering the sentimental behavior of voters along with rational motivations.