Methodology

Sources

• Democracy Club — candidates and results for recent elections, including by-elections, from democracyclub.org.uk. Used under Democracy Club's open data licence, with attribution.
• Andrew Teale's Local Elections Archive Project — historic local election results by ward, used with attribution.
• The Elections Centre handbooks (Rallings & Thrasher) — historic local election results and council composition.
• House of Commons Library and the Electoral Commission — general election and by-election results.
• Council websites — declarations of results, used to fill and check gaps.

Every result keeps a record of the source it came from. When two sources disagree, the result is held for review rather than guessed.

Matching candidates across elections

Sources rarely share an ID for a person, so the same candidate can appear under slightly different names ("Jon Smith", "J. Smith", "Jonathan Smith"). Psephology.AI joins these records into one person using evidence: name compatibility, the same party, the same or neighbouring areas, and the rule that a person can only hold one seat at a time. Uncertain pairs are researched and either confirmed, rejected or left separate — we would rather show two pages than wrongly merge two people. Merges can be reversed.

Boundary changes

Wards are redrawn regularly. A ward keeps its own page for each set of boundaries, and where the change is known we link the old and new wards (split, merged, renamed or abolished) so results can be followed over time. Comparisons in the Insights reports only compare like with like, or say clearly when they cannot.

Definitions

Every measure on the site is calculated one way, defined here with a worked example from real results.

• Vote share — The proportion of the vote a party won in a contest.
• Change in share — How far a party's vote share moved since the same ward was last fought.
• Swing (two-party, Butler) — The movement between two named parties, averaged over both.
• Majority — The winning margin.
• Swing needed to gain — The swing a party needs, on a uniform swing, to win a ward it does not hold.
• Uncontested seat — A seat filled without a vote because there were no more candidates than seats.
• Candidates per seat / per ward — How crowded the ballot was.
• Contestation rate — How widely a party put up candidates.
• Winner's share — The average vote share of winning candidates.
• Effective number of parties — How many parties are really competing, adjusting for their size.
• Winner's bonus (seats vs votes) — How much first-past-the-post over- or under-rewards a party.
• Incumbency and re-election rate — How many sitting councillors stood again, and how many won.
• Women candidates and councillors — The share of candidates and winners who are women.
• Turnout (three methods) — The proportion of the electorate who voted.
• Council control and No Overall Control (NOC) — Whether one party holds more than half the seats on a council.
• Majority at risk — Whether a controlling party could lose its majority at this election.
• Psephology National Share Estimate — Our estimate of how Great Britain would have voted if everyone voted as they did in this year's local elections.
• By-election vote tracker — A running estimate of party support built from council by-elections.
• Notional result — An estimate of how the previous election would have gone on new boundaries.
• Index of change — How much a ward was redrawn.
• Split-ticket indicator — Signs that voters backed different parties on the same day, for example for council and mayor or across two council tiers.
• National Equivalent Vote — An estimate of each party's share if the whole of Great Britain had voted in that May's local elections, not just the councils that held elections.

Known gaps

• Turnout and electorate are missing for many older local elections.
• Coverage before 1973 is thinner, and some councils' historic results are incomplete.
• Some ward boundaries are still being matched to their outlines on the map.
• Northern Ireland local elections use a different voting system and are only partly covered.

See the coverage page for live counts. Spotted an error? Tell us.