Terms

Election glossary

22 terms that appear on these pages, explained in our own words. Polls first, then electoral law, then institutions.

Election poll: A representative survey of usually 1,000 to 2,500 eligible voters about their party preference. The raw answers are weighted by age, gender, region and education, with the undecided and non-voters removed. What gets published is the Sunday question; some pollsters additionally report an unweighted political mood. An election poll describes the fieldwork period, not a future election. Current election polls.
Election forecast: Strictly speaking, the projection on election night based on exit polls; in everyday use, any extrapolation of the polls to election day. Seats, coalitions and probabilities can be calculated from the polling average; how far such extrapolations have historically missed depends above all on the distance to the election. Forecasts for all upcoming elections.
Sunday question: The standard question in election research: “Which party would you vote for if the election were held next Sunday?” The answers are weighted and adjusted for the undecided; the result is the published poll. It describes the mood at the time of the survey, not the result of a future election. Current federal polls.
Projection and political mood: Some pollsters publish two figures: the raw political mood and a projection that factors in long-term party loyalties and voting behaviour. The projection is what gets published and what is shown here, because it is closer to the election result.
Margin of error: Every sample deviates randomly from the population. With 1,000 to 2,000 respondents the margin is around two to three points for large parties and about one point for small ones. A difference within this margin is no evidence of a change. On top of that come errors no statistic captures: who does not respond, who conceals their intention, how the weighting is done.
Polling average: Several polls together are more reliable than one. This portal takes the most recent poll from each pollster within a time window and averages them without weighting. Other providers weight by age or sample size; the differences are usually below half a point. Method.
House effect: A pollster measures a party systematically higher or lower than the others because method, weighting and sampling differ. The house effect says nothing about who is right. But it explains why a jump in the average sometimes comes down to nothing more than which pollster published last. House effects at federal level.
Sample size: The number of respondents. More respondents reduce random fluctuation, but only by the square root: four times as many respondents halve the error. Beyond about 1,000, a larger sample adds little; what matters then is the quality of the sample.
Telephone, online, face to face: Pollsters survey by telephone (randomly dialled landline and mobile numbers), online (panels of registered participants, weighted by population characteristics) or in person at home. No method is inherently better; what matters is how well the weighting matches the population. Pollster accuracy.
Electoral threshold (five-percent hurdle): A party only receives seats if it wins at least five percent of the votes statewide. The threshold is meant to prevent fragmentation. Exceptions: direct mandates (three at federal level, one or two in some states) and parties of national minorities such as the SSW in Schleswig-Holstein. The European election currently has no threshold.
Basic-mandate clause: A party that wins enough constituencies outright enters parliament according to its second-vote share even below five percent. At federal level that takes three direct mandates, in Berlin, Brandenburg, Hamburg and Schleswig-Holstein one, in Saxony two. In 2021 the Left owed its entry to the Bundestag to this rule.
First vote and second vote: The first vote elects a person in the constituency, the second vote a party list. What counts for the parties' strength in parliament is the second vote. That is why polls ask about the second vote. In Bavaria both votes count; in Bremen and Hamburg voters have several votes to distribute.
Sainte-Laguë/Schepers: The calculation method used to translate vote shares into seats, at federal level and in most states. Each party's votes are divided successively by 1, 3, 5, 7 and so on; the largest quotients win the seats. The method treats large and small parties neutrally.
Hare/Niemeyer: Another calculation method: each party first receives the whole-number quota of its seats, and the remaining seats go to the largest fractional remainders. Used in Berlin, Brandenburg, Hesse, Mecklenburg-Western Pomerania, Saxony-Anhalt and Thuringia. The result usually differs from Sainte-Laguë by at most one seat.
d'Hondt: The oldest method, dividing by 1, 2, 3, 4 and so on. It slightly favours large parties and in Germany is now used only in Saarland.
Overhang and compensatory seats: If a party wins more constituencies than it is entitled to by second votes, overhang seats arise. To keep the proportions right, the other parties receive compensatory seats and the parliament grows. The Bundestag abolished this in 2025 and stays at 630 seats. Most state parliaments still grow; North Rhine-Westphalia went to 195 instead of 181 seats in 2022. Example: North Rhine-Westphalia.
Coalition and majority: A government needs more than half the seats, so at least 316 out of 630. This portal shows all two- and three-party alliances that reach this threshold, without political assessment. Whether an alliance comes about is decided by the parties, not by the arithmetic.
Blocking minority: More than a third of the seats is enough to block constitutional amendments, because those require two thirds. In Thuringia in 2024 the AfD became the first party in a state parliament to reach this threshold.
Bundesrat: The chamber of the states with 69 votes, three to six per state depending on population. Each state votes as a bloc, which is why coalitions abstain when they disagree. For legislation requiring the states' consent, the federal government needs 35 votes. State elections therefore also shift the majorities at federal level. Votes in the Bundesrat.
Legislative term: The term of a parliament: four years in the Bundestag and in Bremen, five years in all other state parliaments and in the European Parliament. The election date falls within a legally defined window at the end of the term, which is why dates are often only fixed a year in advance. Election dates.
Voting age: In federal elections people are eligible to vote from 18; in the European election and in eight states already from 16: Baden-Württemberg, Berlin, Brandenburg, Bremen, Hamburg, Mecklenburg-Western Pomerania, North Rhine-Westphalia and Schleswig-Holstein.
Chancellor question: The question of whom respondents would directly elect as chancellor. It measures people, not parties, and is asked regularly by some pollsters. This portal does not show it yet, because the underlying database only contains the Sunday question.