We noticed you're using an ad blocker. Democracy Pulse is free to use and relies on ads to keep running. Please consider disabling your ad blocker to support us. Thank you! 🙏

Glossary of Polling and Election Terms

Election coverage is full of jargon. This glossary contains original, plain-English definitions of more than 45 terms commonly used in opinion polling, electoral systems and government formation. It is intended as a reference companion to our in-depth guides.

A

Aggregator: A tool, model or publication that combines multiple polls into a single estimate. Good aggregators weight polls by recency, sample size, pollster track record, and known house effects rather than averaging them blindly.
Alternative Vote (AV): Also called Instant Runoff Voting. Voters rank candidates in single-member constituencies; the lowest-ranked candidate is eliminated and their preferences redistributed until somebody clears a majority. Used in the Australian House of Representatives.
Approval rating: A poll measure of the share of voters who say they approve of a leader's performance, distinct from vote intention. Approval is loosely correlated with vote intention but the two diverge sharply in mid-term periods and during crises.

B

Ballot order effect: The small but measurable tendency for candidates listed first on a ballot to receive a few additional votes purely because of their position. Many electoral authorities now randomise ballot order to neutralise the effect.
Bandwagon effect: The hypothesised tendency of voters to swing toward a candidate or party that polling shows to be winning. Empirical evidence for the effect is mixed and likely small in stable democracies.

C

Caretaker government: An interim government, often the previous administration, that continues to manage day-to-day affairs after an election but before a new government is formed. Caretaker conventions typically restrict major policy decisions during this period.
Coalition: A formal arrangement under which two or more parties share government, usually with a written agreement, divided cabinet posts and a joint legislative programme. The norm in most proportional-representation democracies.
Confidence and supply: An informal arrangement under which a smaller party agrees to support a minority government on confidence motions and budget votes only, retaining freedom on other legislation. Lighter-touch than a full coalition.
Cross-tabs: Tables breaking poll results down by demographic variables such as age, gender, region or past vote. Useful for spotting patterns but often based on smaller subsamples with much wider margins of error than the headline figure.

D

D'Hondt method: A highest-averages formula for converting votes into seats under proportional representation, used in many European democracies. Tends to favour larger parties slightly more than alternative methods like Sainte-Laguë.
Don't know: The share of poll respondents who say they have not yet decided how they will vote. How undecided voters are treated — included, excluded, or modelled — significantly affects the headline numbers.

E

Exit poll: A poll of voters as they leave polling stations on election day, designed to estimate the result before official counts are completed. Methodologically different from pre-election polls and subject to its own distinctive errors.

F

First Past the Post (FPTP): An electoral system in which the country is divided into single-member constituencies and the candidate with the most votes in each constituency wins. Used in the United Kingdom, the United States, Canada and India.
Formateur: The politician designated, usually by the head of state, to lead negotiations to form a government after an election. The role is most consequential in coalition systems.

G

Generic ballot: A poll question asking respondents which party they would support without naming specific candidates. Useful for measuring underlying party preference, especially in legislative races where individual contests vary in profile.
Grand coalition: A coalition between the two largest parties of a country, usually drawn from opposite political traditions, often formed to keep a third force out of government or to address national crises.

H

Herding: The tendency for late-cycle polls to cluster around a consensus number rather than spreading out as their methodology would normally produce, driven by professional reluctance to publish a divergent result.
House effect: A persistent difference between the numbers a particular polling firm produces and those of the rest of the industry, usually attributable to consistent methodological choices rather than partisan bias.
Hung parliament: A legislature in which no single party holds an outright majority of seats, requiring a coalition, confidence-and-supply deal or minority government to operate.

L

Likely voter screen: A set of questions and rules used to estimate which respondents are most likely to actually turn out and vote, used to filter or weight a poll. Different screens can move headline figures by several points.

M

Majority bonus: An electoral mechanism that awards extra seats to the largest party or coalition to manufacture a working majority, used in some forms of the Italian and Greek systems among others.
Margin of error: The published statistical uncertainty around a poll, usually expressed as ±x percentage points at a 95 percent confidence level. Captures only random sampling error and significantly understates total uncertainty.
Minority government: A government holding office without a stable majority in the legislature, building majorities on individual votes. Common in Scandinavia and other consensus democracies.
Mixed-Member Proportional (MMP): A two-vote electoral system that combines single-member constituencies with party-list top-up seats to produce broadly proportional outcomes. Used by Germany, New Zealand and others.
Mode effect: The systematic difference between poll results obtained through different modes of contact (phone, online, face-to-face). Mixing modes within a single survey complicates analysis but often improves coverage.

N

Non-response bias: The bias introduced when the people who agree to take a survey differ systematically from those who refuse. Increasingly important as response rates have collapsed in modern phone polling.

P

Panel: A pre-recruited group of respondents who agree to take occasional surveys, often in exchange for small rewards. The dominant mode of online polling, requiring careful quality control to remain representative.
Past-vote weighting: A weighting technique that adjusts a poll sample so that the share of respondents who say they voted for each party last time matches the actual previous result. Helps with non-response bias but assumes accurate recall.
Polling average: An aggregate figure produced by combining many recent polls, usually with adjustments for recency, sample size and pollster quality. Far more reliable than any individual poll.
Preference flow: Under preferential voting systems like AV or STV, the way that supporters of one candidate or party tend to allocate their lower preferences when their first choice is eliminated. A critical input for forecasting.
Primary vote: Under preferential voting, the share of first-preference votes a candidate or party receives, before any preferences are distributed. The two-party-preferred figure is then derived from this baseline.
Proportional Representation (PR): An umbrella term for electoral systems that allocate seats roughly in proportion to vote share, typically through party lists in multi-member districts. The dominant family of systems in continental Europe.
Push poll: A telemarketing technique disguised as polling that uses leading questions to influence rather than measure opinion. Not a legitimate form of research and disowned by the polling profession.

Q

Quota: Under STV, the number of votes a candidate must receive to be elected, calculated from total votes and number of seats. Once a candidate reaches the quota their surplus votes transfer to other candidates.

R

Refusal rate: The proportion of contacted potential respondents who decline to participate in a survey. High refusal rates raise the risk of non-response bias and require heavier weighting to correct.

S

Sample size: The number of respondents in a poll. Larger samples reduce random sampling error but at diminishing returns: doubling the sample only narrows the margin by about thirty percent.
Shy voter: A voter who supports a particular candidate or party but is reluctant to admit it to a pollster. Hypothesised cause of certain polling misses, though direct evidence is mixed and harder to detect than absence from the sample.
Single Transferable Vote (STV): A candidate-centred form of proportional representation in which voters rank candidates in multi-member constituencies and votes are transferred between candidates to fill all seats. Used in Ireland and Malta.

T

Threshold: Under proportional representation, the minimum share of the vote a party must receive to qualify for any seats, typically three to five percent. Designed to prevent extreme legislative fragmentation.
Turnout model: The set of rules or statistical model a pollster uses to estimate which respondents will actually vote. A common source of major polling errors when the actual electorate differs from the model's assumptions.
Two-party-preferred (2PP): Under Australian-style preferential voting, the projected vote share between the two leading candidates after all preferences have been distributed. Often a more meaningful headline figure than primary vote.

U

Undecided: See Don't know.

V

Voter intention: A poll question asking how respondents would vote if an election were held immediately. The most common single statistic published by political polls.

W

Wasted vote: A vote that does not contribute to electing a representative, common under FPTP for any vote cast for a losing candidate or for a winning candidate beyond the number needed to win. PR systems aim to minimise wasted votes.
Weighting: The post-collection adjustment of a poll sample so that under-represented groups count more heavily and over-represented groups count less, restoring demographic balance. Almost all published polls are weighted.

Keep reading

Each term here is treated in much greater depth in our guide collection. Start with How Election Polls Actually Work for the foundational concepts, then dive into Electoral Systems Explained to understand how votes turn into seats.