The closer we get to the elections, the more frequently political parties' support ratings are reported in the media. ETV news show "AK. Nädal" posed the question of how these ratings are created and who is polled, and whether they also affect the election campaigning and parties' activities.
Party support ratings conducted by, for instance, pollsters Kantar Emor, who have 32 years of experience in market research, are now being ordered on a weekly basis, compared with a monthly one at other times.
Aivar Voog, Kantar Emor research manager, told "AK. Nädal" that: "We make use of both online interviews and telephone interviews, though in the case of party surveys, the proportion is two-thirds online, one-third over the phone."
"In the case of political party surveys, the sample size is at least 1,500 respondents, i.e. 1,000 online and 500 by telephone," he continued.
In order to obtain a proportionally representative sample, Kantar Emor says it uses the distribution of citizens of voting age as obtained from Statistics Estonia.
Online surveys are conducted on the basis of a pre-selected contact base, he added.
"We started recruiting these people as early as 20 years ago. At that time, we primarily used face-to-face interviews, in the respondents' homes, while in the course of this we asked people if they were willing to take part in our online interviews as well; over the course of a couple of decades, we have received quite a large number of them," Voog went on.
"We also use telephone interviews for this, in order to recruit more people to the contact database. Currently, this database contains approximately 30,000 contacts, from which we make a random selection," he added.
Meanwhile, Norstat has been conducting political party surveys on a weekly basis since early 2019.
A combined method is also used to collect data, but interviews over the phone pick up a larger share, while the company conducts these at random.
Norstat CEO Evelin Pae said: "We're more of a believer in phone interviews. The respondent profile online has slightly different values and views than those who respond over the phone. For this reason, we still lean toward phone interviews, because it's given us the most accurate results so far, on what the opinion of Estonian residents is."
Norstat's call center is located in Valga, on the Latvian border, while random numbers are used for telephone interviews, which helps to ensure that answers are unique. The weekly sample size comes to a thousand people, though this is aggregated over a four-week period.
"The results are based on the cumulative data of the current four weeks, i.e. the sample is actually 4,000 respondents, which is already a significantly larger sample. Since certain fluctuations or randomness can occur on a weekly basis, the cumulative view of four weeks reduces this tendency towards randomness, somewhat," Pae said.
Pae added that some people anticipated being contacted by Norstat.
"We have also seen comments on social media where people have that they have not yet been able to take part in our party surveys. I hope that we will reach them one day. It is this randomly generated sample of numbers that gives us the chance of actually reaching everyone," she went on.
As to how much of an influence the results of surveys conducted by companies like Norstat, Kantar Emor and Turu-uuringute, Ott Lumi said this was: "Undoubtedly very big."
"Political parties constantly monitor the ratings and then try to react. They analyze the situation and try to make corrections accordingly; for example, they discuss whether the right people for them are in the picture and so on. Politics is certainly increasingly becoming ratings-based," Lumi went on.
On the other hand, some politicians have been known to buck that trend by stating that ratings do not play any major role, as the real truth gets revealed only at the elections.
Political communication expert Annika Arras said: "They still have to take a certain stance."
"And no one wants to over-analyze support publicly, also due to the fact that it reflects the past, and you don't know what will happen tomorrow and how the voter will might act. Mistakes are easy to make here, and any mistake can lead to repercussions, so I understand very well why the parties are careful," Arras went on.
Parties with experienced political actors in their campaign team are often most successful in the ratings.
During campaign season, it is not wise to make major changes, but try to target those voters who are not yet sure of their feelings, Arras added.
She said: "The changes in the last phase of a campaign can primarily result from the fact that the voters who have not yet decided will move towards doing so, while those who are split between one political party or another, will act."
"They still have doubts, and what steps or reactions the political parties will bring out in the very last weeks may be decisive," she added.
In the case of an unexpectedly poor ratings, it is not so important to react to the changes as to try to explain the results to the public, she said.
"When a surprisingly bad rating arises, usually there is a 'crisis meeting' with the party chair the next morning, where the topic of discussion is not first and foremost what is happening, but rather the pressing question is how we comment on this [publicly]. This is normal routine," Arras said.
Party ratings are largely reliable, she added.
"I mostly trust Estonian research companies and companies which conduct sociological research. I believe that this reflects what is really happening in society and what the current mood is. It is certainly not a final truth, but it is surely an important indicator."
In addition to differing combinations of phone, online and face-to-face polling, other differences between the major pollsters' methodologies include different age limits – while they will only poll Estonian citizens of age 18 and upward ahead of a Riigikogu election, as these are the only people who can vote, Kantar Emor caps its respondents' age at 86, while different approaches to weighting survey groups by socio-economic indicators, and different margins of error, are also employed.
ERR News' elections page is here.
The Riigikogu election takes place on March 5, while the advance voting period starts February 27.
Editor: Andrew Whyte, Marko Tooming