Exit and public opinion polls under scrutiny
What role do public opinion polls play during an election race? Can they influence the way we vote? And, more importantly, is the law that forbids the publication of polls for 15 days before an election helpful or harmful?
These questions have gained poignancy after a public opinion poll published alongside exit polls at 7 p.m. on the June 17 general election by Public Issue — the company used by Kathimerini and Skai TV — gave the Coalition of the Radical Left (SYRIZA) the lead over New Democracy, while the exit polls gave ND only a slim 0.5 percent victory. Once the votes were counted, however, the conservative party was ahead of the leftists by nearly 3 percentage points.
On this page, the head of Public Issue, Yiannis Mavris, analyzes how the figures put SYRIZA first and how such a lead was reversed in 15 days. The chief of polling company GPO (which works with Mega TV), Takis Theodorikakos, also provides his own explanation for the final outcome of the general election.
It is worth noting that Public Issue did not conduct an exit poll, but estimated that SYRIZA would have a narrow 0.5 percent lead over New Democracy based on a public opinion poll that could not be published because of the 15-day ban. In contrast, at 7 p.m., when the polls closed on Sunday, June 17, all the other polling companies foresaw that ND would have the 0.5 percent lead over SYRIZA, based on actual votes counted during interviews with voters outside polling stations.
The 2.5 percent discrepancy between the final result and the exit polls, not to mention the 3.5 percent discrepancy between Public Issue?s opinion poll prior to the election and the final outcome, is cause for concern.
The two experts argue that their miscalculations most probably occurred because they overlooked the fact that New Democracy?s upward trajectory would cross SYRIZA?s downward course in the crucial final days leading up to the general election, creating the difference on the day of the polls.
Yiannis Mavris*: Two weeks of rapid developments
The outcome of Greece?s general election on June 17 was by no means cut-and-dried. Quite the opposite: How the race would end was up for debate until the last minute. The momentum of the pre-election period was greater than any recorded in this country, even that leading up to the May 6 polls, while rapid developments in the two weeks before the polls played a definitive role in the outcome. It is evident that New Democracy?s eventual victory does not automatically mean that it was ahead of SYRIZA in the period prior to June 17.
After the May 6 election, a wave of voters moved toward SYRIZA from different directions in classic fashion: first coming in strongly at 31.5 percent, then ebbing to 26.9 percent. The trend continued until May 30, after which SYRIZA published its party program on June 1. It dropped for the next two-and-a-half weeks. The reasons for this decline are most probably to be found in the fear caused by a) the party?s program, b) ND?s public assault against the leftists and c) the pressure put on voters by foreign players and the rumors that Greece could be kicked out of the eurozone.
Illustrating this, 5 percent of former SYRIZA voters voted for ND ?in order to keep Greece in the euro? and 2.5 percent to ?avert chaos and anarchy.? Add to this the increased level of abstention, and SYRIZA lost a total 4.5 percent to eventually end up with 26.9 percent of the vote.
At the same time, ND saw its popularity rise in the final 15 days leading to the polls due to the phenomenon known as tactical voting. Ten percent of its voters said they cast their ballot for the conservatives simply to stop SYRIZA from winning. This wave (around 3 percent of the electorate) came mostly from the former ranks of Democratic Left and PASOK, which explains why public opinion polls before the elections had overestimated the performance of these two parties. We have not yet investigated whether this trend was a late swing — people who changed their mind at the last minute — or whether there were voters who concealed the fact that they would be voting for or had voted for ND, which would have been impossible to gauge in a public opinion or exit poll. This is something that can only be explored once the election is over.
Public Issue?s last estimate before the 15-day ban went into force — giving SYRIZA 31.5 percent and ND 25.5 percent — reflected the prevailing trends at that particular time rather than predicting the electoral outcome.
Meanwhile, SYRIZA?s support shrank during the two-week ban. The first unpublished survey (held from June 1-7) showed a big drop of 2.5 percent, though SYRIZA was still ahead of ND, while the next (June 11-14), showed the difference between the two parties shrinking to 0.5 percent, with SYRIZA in the lead at 28 percent to ND?s 27.5 percent.
Unfortunately, because of the ban, these studies could not be published, leaving the false impression that Public Issue had predicted a 6 percent victory for SYRIZA, when in fact it presented the prevailing trend for three days prior to the election.
Takis Theodorikakos**: Greater sense of responsibility needed
Given that we all now agree the ban on the publication of public opinion polls in the fortnight before elections allows rumor-mongering and political propaganda, which damage the reliability of information, let us examine the factors that shaped the behavior of Greek voters.
In the May 6 election, the dilemma was about being pro- or anti-memorandum, with voter support for New Democracy and PASOK dropping to an all-time low, while SYRIZA and other ?anti-memorandum? parties enjoyed unprecedented backing.
In the first three to four days after the May 6 election, polls that were not published showed SYRIZA continuing its impressive upward trajectory, and ND and PASOK falling even lower. The events that defined further developments and opinion polls were the stances of the three parties over the exploratory mandates for the formation of a government, where SYRIZA was blamed for leading Greece to a new election, and the subsequent announcements by ND leader Antonis Samaras and SYRIZA?s Alexis Tsipras of their parties? economic policy.
ND?s plans for the economy were met with support by 47 percent of people polled, compared to SYRIZA?s 36 percent. It is clear that SYRIZA?s platform was meant to draw voters from other far-left parties, and from the Communists (KKE) in particular, rather than to appeal to a new audience.
Another important moment was when Dora Bakoyannis joined forces with New Democracy, as 82.5 percent of center-right voters approved of the move. Meanwhile, the new dilemmas of ?euro vs drachma? and ?renegotiating vs abolition of the memorandum? put additional pressure on the electorate.
In the three weeks leading up to June 17, these elements gave ND a steady 2-4 percent lead over SYRIZA, with 12 percent of people polled saying they would decide who to vote for at the last minute.
Broadcasting the results of public opinion polls obviously influences the public and this influence corresponds to the prestige of the media publishing the polls and the companies conducting them. Polls are meant to explore to the attitudes, criteria and factors that determine the decisions of millions of people, and they are clearly not meant to influence the outcome of society?s decision.
It has, moreover, been abundantly proven that however many opinion polls are published, if their conclusions are incongruous with the prevailing reality, they can do nothing to reverse it. It is also obvious that taken alone, opinion polls are not prediction tools. In public dialogue, however, these estimates are often talked about as though they reflected the election result, even by serious analysts, thus influencing voters.
The experience gleaned from the last two elections, and June 17?s in particular, should be used to generate a greater sense of responsibility in terms of how polls are used and more respect for the scientific authorities that are consistently reliable. The reliability of such polls is especially important today, and so we must all — media, journalists, analysts and pollsters — remain respectful to the people, and be realistic and objective.
* Yiannis Mavris is the president and chief executive officer of Public Issue.
** Takis Theodorikakos is a political analyst and president of GPO.