One other loser in elections: survey companies

It is an unchanging topic in Turkey after every election, comparing the actual results with the predictions of pre-election public opinion polls. However, the role of the survey agencies has never triggered such a major debate in any recent election before as they did in the presidential elections held last Sunday, Aug. 10.

This time, one other aspect is added to the classic debate of how close the opinion polls were to the actual results; such claims as to whether or not these research companies were manipulation tools to influence election results were also discussed. A widespread belief is that many voters did not bother to go to the polls, affected by the survey results, assuming the results were already determined. At least, this is the main opposition party’s official argument. It would not be a mistake to say this debate will branch out and grow in the coming days.

Difference between findings and results

First of all, we have to highlight this reality: The difference between the figures reached in the majority of the pre-election opinion poll surveys and the results of the elections is quite high. Among survey companies, there were some that exaggerated Recep Tayyip Erdoğan’s votes up to the range of 57 and 58 percent. There are also those that had very close predictions and in that way, which were correct in their estimations, but they were few.

One of the companies the polls of which were mistaken about was KONDA, the one Republican People’s Party (CHP) leader Kemal Kılıçdaroğlu openly accused of staging a “perception operation” in his interview with Aslı Aydıntaşbaş from daily Milliyet. The executive of the company Tarhan Erdem publicized their projections exactly three days before the elections, on Thursday, Aug. 7, and said Erdoğan would receive 57 percent, Ekmeleddin İhsanoğlu 34 percent and Selahattin Demirtaş 9 percent; whereas last Sunday, after votes were counted, Erdoğan received 51.79 percent, İhsanoğlu 38.44 percent and Demirtaş 9.76 percent.

Erdem, the other day, had an interview with daily Birgün, saying there was a “very serious difference” between their findings and the results; and he took this with “sadness and surprise.” According to Erdem, technically the difference between the findings and the results should not be more than the margin of error. They had announced their margin of error in their survey as plus or minus 2.6 percent, Erdem said, adding, “However, the difference we encountered is minus 9.1. There is no scientific explanation to this.”

Acceptable margin of error

Among the companies that had the closest estimates, Metropoll is noteworthy. This survey company predicted 50.5 percent for Erdoğan, 40.2 percent for İhsanoğlu and 9.2 percent for Demirtaş.

Executive of Metropoll Professor Özer Sencar, in an interview with daily Taraf, said they held three separate election surveys last term and in none of them Erdoğan’s votes exceeded 47 percent, without distributing the undecided votes. “The rate of the undecided was not 10 percent. For this reason, it would have been impossible to estimate that Erdoğan would receive 57 percent,” he said. Sencar claimed all of the predictions that went below or above the acceptable 2 percent margin of error were “on purpose,” and they were “perception operations.”

As you can see, the results are also causing a serious debate among the survey companies themselves.

Shadow of the perception

Meanwhile, the fact that the Justice and Development Party (AK Party) very strongly processed the argument, based on opinion polls, that Erdoğan was leading all along the campaign has played a significant role in this perception taking root in the society. For example, in his statement to news channel NTV on Wednesday evening before the elections, Erdoğan said, “In the surveys conducted this week, Erdoğan has 56.7 percent of the votes, İhsanoğlu 34.9 percent. There is a fall. Demirtaş has 9.4 percent of the votes.”

Erdoğan has shown his votes high, his chief rival İhsanoğlu’s votes low; he remained within a logical margin of error only concerning Demirtaş.

As a result, the presidential elections were conducted in a psychological environment shaped by a dominant perception; voter behavior was inevitably influenced by this perception. More research in the coming days will demonstrate the extent of this effect. Interestingly, this perception has also caused lethargy in a portion of AK Party voters.

As a result, the shadow of this perception has fallen over the election process, the most direct participation tool of democracy.

Ban on surveys neglected

There is also a legal side to the issue. Election Law Number 298 definitely bans the publication of surveys. The Supreme Election Board (YSK), before every election, reminds us of this ban with a circular. However, neither survey companies, nor the media abide by this ban.

Also, the ethical perspective of the issue is not any less important. The ethical principles the survey companies should apply should be thoroughly scrutinized as well as objective scientific criteria they should observe, in the light of the thought-provoking picture that has emerged. When a disclaimer or an apology is expected from the media when a mistake is made, then isn’t it our right to expect similar acts also from the survey companies which have mislead the society?