This week, three polling firms — Norstat, Kantar Emor and Turu-uuringute AS — published party support ratings. The biggest controversy stemmed from Turu-uuringute's claim that the Center Party had surged into a clear lead in the polls, 5 percentage points ahead of Isamaa. A day later, however, a poll by Norstat and the Institute of Societal Studies claimed that Isamaa had instead widened its lead over the Centre Party, with a five-point advantage in Isamaa's favor. A Kantar Emor survey published by ERR on Friday found that while Isamaa still maintained a lead over the Center Party, it was so narrow that, taking the statistical margin of error into account, support for the two parties was essentially equal.

Such seemingly large differences and opposing trends naturally create confusion and distrust — especially when someone's preferred party drops in the ratings or a rival's support rises — but the situation is not actually so dramatic. The reason is that poll results are usually presented as aggregate figures without margins of error. Once those margins are taken into account, the differences may no longer appear so large.

For example, in the Kantar Emor poll, support for Isamaa stood at 23.5 percent with a margin of error of ±2.2 percentage points. This means that Isamaa's actual support in the May survey ranged from 21.3 to 25.7 percent. That is not far from either Turu-uuringute's 20 percent result or Norstat's 26.6 percent.

The same applies to support for the Center Party. In the Kantar Emor survey, it stood at 20.6 percent with a margin of error of ±2.1 percentage points — meaning actual support ranged from 18.5 to 22.7 percent. Taking the margin of error into account, the Center Party's support could therefore even have been higher than Isamaa's (22.7 percent versus 21.3 percent). Kantar Emor research expert Aivar Voog concluded that the gap in support between Isamaa and the Center Party is not statistically significant.

Differences in methodology

When comparing different poll results, it is important to consider that surveys are conducted using somewhat different methodologies. These differences include sample size and polling period, which in turn may be influenced by current events at the time. Polling firms or their clients may also differ in how they weight the results.

At the request of the Institute of Societal Studies, Norstat surveys at least 1,000 Estonian citizens aged 18 and older over the course of a week. The company uses both telephone and online surveys, with telephone interviews accounting for 85 percent and online responses 15 percent. The published results focus on aggregate data from the previous four weeks. In the combined rating, the sample size reaches 4,000, in which case the maximum statistical margin of error does not exceed 1.55 percentage points. For weekly results, where the sample size is 1,000, the maximum margin of error does not exceed 3.1 percentage points.

Kantar Emor surveys 1,500 Estonian citizens aged 18 and older once a month. Two-thirds of respondents are surveyed online and one-third through telephone interviews. Unlike Norstat, which focuses primarily on telephone polling, Emor places greater emphasis on online respondents. With this sample size, the maximum possible margin of error is ±2.2 percentage points. The polling period is generally 10 days — in the May survey, for example, from May 11 to 20.

Turu-uuringute AS most recently surveyed 903 Estonian citizens aged 18 and older. Similar to Emor, one-third of responses were collected by telephone and two-thirds online. Turu-uuringute's polling period is generally one week — in May, from May 7 to 13 — meaning the survey concluded a week earlier than Emor's. As a result, the poll results were influenced by different current political events.

--

Follow ERR News on Facebook and X and never miss an update!