A party gains seats while losing votes, a familiar constituency is recorded as a gain, and a national share changes partly because a rival did not contest the same places. Historical election figures can be accurate individually while still being misleading when placed side by side.
Establish a comparable baseline before describing change. Election date, geography, party classification and the measure being compared all belong in the calculation.
Keep election results separate from current representation
A general election result records the candidates elected on that occasion. The Commons later changes through vacancies, by-elections, changes of party and other events.
Comparing the next election with the previous election’s original seat total answers one question. Comparing it with the party tally immediately before dissolution answers another. Label the baseline instead of mixing the two.
The Commons Library’s result datasets are based on verified declarations and record corrections. For reproducible analysis, retain the dataset reference date as well as the election date.
Check whether the place is the same
Boundary changes can redraw a constituency while leaving its name familiar. An old-seat result and a new-seat result need not represent the same communities.
Notional previous results estimate a comparable old-election baseline on the new geography. They should be marked as estimates, while the actual election results remain observed counts.
The Commons Library’s boundary maps help identify the change. For a national comparison, also distinguish Great Britain from the whole United Kingdom and explain any excluded seats.
Votes, shares and seats answer different questions
Raw votes show how many ballots a party received. Vote shares show its fraction of the stated valid-vote total. Seats show local victories under the electoral rules.
A party’s share can rise while its vote count falls when turnout declines. Its seats can rise disproportionately when gains occur in close constituencies, as explained in vote share versus seats.
In an illustrative sequence, a party wins 100 seats with 30% and then 150 with 32%. The two-point vote change and 50-seat gain are both valid descriptions. They do not establish a general conversion rule of 25 seats per point.
Party labels and candidate coverage
Parties can change names, merge, withdraw candidates or make electoral arrangements. Independent candidates can be grouped differently across datasets. Specify the classifications used rather than assume every label represents an unchanged organisation.
A national percentage also reflects where candidates stood. Comparing performance in the seats contested at both elections can answer a different question from comparing the whole national electorate.
Neither approach should quietly replace the other. For a comparison of party reach, candidate coverage is relevant; for a comparison of local preference, an unmatched set of contests can complicate the inference.
What can totals tell us about switching?
A fall for Party A and a rise for Party B do not prove that the same voters moved directly between them. Some may have moved to other parties, some abstained and new voters may have entered.
British Election Study panel research follows individual responses over time, supplying evidence which aggregate totals cannot. Such evidence still needs its own sampling and weighting assessment.
The same caution applies to claims about tactical voting: a useful-looking redistribution is not proof of the motive behind it.
A comparison readers can reproduce
State the elections, geographical coverage, source and baseline. Show the measure and its denominator, then explain material boundary or classification changes.
Our historical results archive provides starting points. Add swing or majority calculations where they answer a specific question, and distinguish your interpretation from the verified figures which support it.
Sources and further reading
- House of Commons Library: result files and dataset corrections
- House of Commons Library: comparable constituency geography
- British Election Study: tracing individual voting changes
Reviewed 7 October 2026. Worked examples are illustrative unless an election or poll is identified.