Pre-Election Polls: A Great Treasure

It was an exciting political race in the polls before the state elections in Saxony-Anhalt: the former election winner CDU only received approval of 22 percent instead of 37, while the right-wing extremist AfD was over 40. Three parties - SPD, Greens and BSW - scratched the 5 percent hurdle. Would they all fail to get into parliament? Then the AfD would even have an absolute majority of the state parliament seats and could govern alone. Hidden beneath the boring rows of numbers lay a dramatic spectacle.

Political election polls, especially the so-called Sunday question, are now part of everyday political reporting. So much so that many people are very annoyed: “Abolish the Sunday question,” comments a correspondent on Deutschlandradio, for example, listing arguments that are often used. The result of the Sunday questions is not as clear as it often seems; The error tolerances are not discussed enough, as are the survey methods; many people treat the polls as election predictions. And the polls may influence voters and may even be responsible for the rise of the right-wing extremist AfD.

All of that is correct. But the points are part of the complexity of political surveys and are also discussed by the makers themselves. They discuss the weaknesses of one or the other methodology and identify error tolerance. Statisticians choose a method and thus their strengths and weaknesses. A sample of a few thousand people cannot possibly replace a complete survey, but a survey with a margin of error of 3 percent is better than no information at all and cheaper than surveying 60 million eligible voters.

Those who know more can act more informed. The fact that knowledge guides people's behavior is a strength, not a weakness. It therefore makes sense to treat political polls like the weather report, which only determines what you wear. You should still go outside instead of hoping for sunshine inside.

The essence of political surveys works like scientific magic: if you follow precise recipes, then just a few hundred people can provide information about many millions. The fact that thousands of people in Germany are regularly asked about their political thoughts grounds politicians and gives more opinions the opportunity to be discussed in public.

What if there hadn't been a Sunday question in recent years? In the state elections in 2026, everyone would have been surprised that right-wing extremists won an election in Germany. And would the CDU really have accepted that after decades of government it had surprisingly lost half of its voters?

The many institutes that regularly conduct political surveys in Germany are a luxury: because repetition ensures scientific results. If several institutions, with different surveys and their own, secret weighting methods, deliver similar results, it is very likely that they are close to reality. This is a great treasure that makes for wiser political debate.

In fact, many people often try to squeeze more out of the polls than they give. They hope for small, short-term movements in the election polls, which often dominate headlines, but could also simply be measurement errors. The poll numbers also do not save us from the uncertainty that we do not know before the election how the election will turn out. No amount of calculation can change that.

Not even that of the time. They recently simulated 1,000 possible seat distributions in the state parliament of Saxony-Anhalt from election surveys and found that a majority for the CDU, SPD, Greens and Left was most likely. The calculations were not much more meaningful than the election surveys themselves. And even if this time-future was very likely, it did not become reality.

Election polls are often a self-fulfilling but self-preventing prophecy: they give people knowledge about the society around them and thus the opportunity to act differently. The surveys in Saxony-Anhalt probably influenced the election in many ways: Significantly more people went to the polls than in previous elections, the SPD and Greens polled well above the 5 percent limit and the BSW could also have received borrowed votes from the AfD.

This is exactly how we can use surveys in a meaningful, useful way: by letting them guide our actions in the real world. And by leading the media and journalists to interesting further research questions. Instead of searching the surveys for more meaning, it makes more sense to supplement them with more knowledge: Is the AfD even ready to take on government responsibility? What does the election campaign by the Greens, SPD and BSW look like locally?

Surveys are the beginning, not the end, of action. In Saxony-Anhalt, the survey results triggered a campaign for tactical voting. To many people, the campaign seemed disreputable, a call to dishonest behavior. But voting is always a tactic, because no party represents a person's own opinion in its entirety. And every vote is an attempt to maximize one's influence. In Saxony-Anhalt, this action meant that the election result had several significant deviations from the surveys.

In Mecklenburg-Western Pomerania, the election surveys raise similar questions to those in Saxony-Anhalt. The CDU, Greens and BSW are dangerously close to the 5 percent limit here. So is it worth it for SPD or Left voters to give their votes to Greens or even CDU politicians in order to strengthen the democratic camp? The right-wing extremist AfD has the most popularity - does it therefore need every vote or could its voters be given more coalition options by voting for the BSW? The surveys will not be able to provide the answers to these questions. These are questions for neighbors, for friends, for strangers. They can be found out in the world.