UK electoral reform simulator: methodology
What this is. Every tab takes the real votes cast at the 2024 General Election, constituency by constituency (House of Commons Library), and counts them again under a different electoral system. The votes were cast under first-past-the-post, so each party’s vote is carried over as it was; people would vote differently under another system, with more parties standing and less tactical voting. Read every result as “the same 2024 votes, counted another way”, not as a forecast.
The combined-constituency tabs. Freestyle and the ~3, ~4.5, ~5, ~7 and ~9 combined tabs group whole 2024 constituencies into larger ones that elect several members. Every group sits wholly inside one English region or one nation. “~3 combined” has 216 groups of (mostly) 3 constituencies, each electing 3. “~4.5 combined” cuts the Commons to 435 seats: 145 groups of 3–6 constituencies, each electing 3. “~5 combined” has 120 groups of about 5–6 (3 or 5 seats each), “~7 combined” 93 groups of about 7 (5 or 7 seats) and “~9 combined” 73 groups of 6–10 (5, 7 or 9 seats). Freestyle’s 650-seat house uses 141 county-and-borough groups of 3–7 constituencies; its 450-seat House of Constituents has 104 groups of 4, 6 or 7 constituencies electing 3, 4 or 5. In the smaller houses the seats are shared between the four nations, and then between the English regions, by Adams’ method on the December 2025 electoral registers, with a floor that keeps Scotland, Wales and Northern Ireland from being under-represented.
The default count: Droop quota and largest remainder. Each group adds up its constituencies’ votes by party. A party wins one seat for every full Droop quota it has (the group’s votes ÷ (seats + 1), plus one), and any seats left go to the largest remainders. Nothing transfers, so this is a party-list count, not a ranked one.
The Simulated STV toggle. Turns the count into a single transferable vote. Each party’s vote is treated as a disciplined bloc running a full slate: a bloc elects a member for every full quota, and while seats remain the smallest bloc is excluded and its votes pass to the parties still standing, by a transfer table. Every ballot is assumed to number every box, so nothing exhausts. Independents, the Speaker and single minor-party candidates can win one seat at most; their surplus passes on. The transfer table: Conservative, Labour, Lib Dem, Reform and Green voters split between those five parties as in YouGov’s July 2024 ranking survey (8–9 July, 2,085 adults); Scotland and Wales add the SNP and Plaid shares from YouGov’s small Scottish and Welsh subsamples (indicative only); Northern Ireland uses the party-to-party transfers of the 2022 Assembly count (NI Assembly Research Service). The shares to the SNP, Plaid, smaller parties and independents elsewhere are Earth Elects’ own estimates, as are the transfers from independents who campaigned on Gaza (guided by Policy Exchange’s 2026 poll of Muslim voters). The preferential tabs and toggles count with the same table, so the two never disagree.
The compensation toggle. Where a party didn’t stand in every constituency of a group, it is given, in each seat it skipped, 75% of its average vote in the seats it did contest, and every party is then scaled back to the real turnout. Independents, the Speaker and the pooled Other aren’t boosted, and the Speaker’s seat doesn’t count as one the main parties skipped. It combines with Simulated STV, adjusting the votes first.
Who is counted on their own. The parties differ by nation: in England Labour, Conservative, Reform UK, Lib Dem and Green; Scotland adds the SNP and Wales Plaid Cymru; Northern Ireland has Sinn Féin, the DUP, Alliance, the UUP, the SDLP and the TUV. The Workers Party of Britain is counted everywhere it stood. Every independent and minor-party candidate is taken out of “Other” and counted in their own right wherever the Commons Library’s 2024 candidate list adds up exactly to the seat’s Other vote, so a candidate like People Before Profit’s in Belfast West or the Ashfield Independents’ can win a seat on their own vote. Where it doesn’t add up, the candidates stay pooled as “Other / minor”: that pool counts towards the quota and shows on every chart, but it can’t win a seat, because it is several unconnected candidates and not one, and the seat goes to the next in line.
Independents on the national bars. Independents who stood on Gaza (Jeremy Corbyn, Shockat Adam, Adnan Hussain, Ayoub Khan, Iqbal Mohamed, Leanne Mohamad and others), with the two minor-party candidates who ran on the same campaign, form one pink “Gaza independents & parties” block; other independents are one grey “Independents” block; the Speaker, Lindsay Hoyle, is shown on his own.
Three bars, three things. The seat simulation bar is the count under the system you picked. The real 2024 vote share bar is the votes actually cast, with no counting method applied: what an exactly proportional result would look like. The real 2024 result bar is the seats each party really won under first-past-the-post.
The other tabs. MMP, Parallel, Preferential 650, the Netherlands-style and the country- and region-wide tabs explain their own method in the summary at the top of each tab.
Caveats
Tactical voting. 2024 was fought under first-past-the-post, which rewards voting for whoever can beat the candidate you like least rather than for your first choice. So the totals every tab starts from are partly a read of what people thought would work under FPTP, not of what they wanted. Labour’s 33.7% is the clearest case: some of it was “get the Tories out” voting, so its genuine first-preference support was probably somewhat lower, and other parties’ higher.
Newer population data than the boundaries. The seat shares between the nations and English regions use the December 2025 registers, the newest available, but the 650 constituencies and their votes are the 2024 ones, drawn on older population figures. Two or three years of movement should only change results at the margins. The same 2025 figures decided how many old constituencies were merged into each new one on the combined tabs, so places that had outgrown their old boundaries had more merged in: a deliberate choice, not a neutral default. Adams’ method and the floor for Scotland, Wales and Northern Ireland are a choice too, not an objective standard: both give the smaller nations more seats than strict population would, and someone else might pick another method or no floor at all.
What a single FPTP ballot can’t show. The transfers behind every preferential and STV count are researched estimates, not real transfer data, and the 75% compensation rule is an assumption about how a party would have polled where it didn’t stand. The House of Nations and Regions assumes people vote as they did for the Commons the same day, when a real upper-house election could see different turnout and choices. MMP, Parallel (MMM) and both top-up features assume people vote the same for their local candidate and their party list, when real mixed-member systems show plenty of split-ticket voting.