What the Makerfield by-election really tells us about Britain’s broken voting system

Andy Burnham is the most popular politician in Britain. According to YouGov, 35 per cent of people have a positive opinion of him — nearly twice that of Sir Keir Starmer. And yet, as the Makerfield by-election approaches on 18 June, nobody can confidently say he will win it.

That is not a story about Burnham, or about Starmer, or even about Labour. It is a story about a voting system that asks a country with five-party politics to express its preferences with a single, blunt “X”.

The maths of Makerfield

Makerfield has elected a Labour MP at every election since 1983. Labour’s Josh Simons won the seat in 2024 on 45.2 per cent, Reform UK on 31.8. In the local elections on 7 May 2026, Reform won every single ward.

Into this, Labour is sending its best player. Survation found that with Burnham as the candidate, Labour has a 67 per cent chance of winning. Without him, that chance is essentially zero. Sir John Curtice put it more bluntly: “less than 5 per cent” with anyone else.

Think about what that sentence means. The democratic outcome in a town of 76,000 voters depends not on what those voters believe, but on whether one particular celebrity politician is willing to drive up the M62.

Vote-splitting on both flanks

Makerfield matters because the splitting is happening on both sides at once.

On the right, Restore Britain, Rupert Lowe’s breakaway from Reform, is polling 5–6 per cent nationally; Find Out Now polling, commissioned by Restore itself, shows Reform’s vote dropping from 25 to 21 per cent when Restore is on the ballot. Add Advance UK, UKIP and the rump Conservative vote, and the right is now genuinely fractured — as the left has been for years.

On the left, the Greens are not standing aside. After winning Gorton and Denton in February with 41 per cent and a 33-point swing from Labour, Zack Polanski’s party is selecting a Makerfield candidate. Caroline Lucas has called on them to “put country before party” and give Burnham a clear run; the party’s answer, in effect, is that under first-past-the-post every party is forced to be selfish.

She is right, and so are they. That is the bug.

The plausible result on 18 June is a winner on something like 35–40 per cent of the vote, with 60–65 per cent of the constituency having voted for somebody else. We will call that a mandate. It is not one.

A system designed in 1884

Britain did not choose first past the post. Lord Salisbury did, in private, in 1884.

Negotiating with Gladstone over the price of expanding the working-class franchise, Salisbury insisted on carving England into hundreds of new single-member constituencies. On the question of how minority opinion would be represented, he wrote on the draft, in his own hand: “Minorities not to be directly represented.” The Redistribution of Seats Act 1885 made single-mark plurality the British default.

The system as we now use it dates from 1950 — younger than the NHS, and built on voting theory that was already a century out of date when Parliament settled on it.

We have learned a lot in 140 years. None of it is in our ballot box.

What Burnham’s bid actually proves

Look at the arithmetic. In a constituency that has voted Labour for forty years, with the country’s most popular politician on the ballot, the broadly progressive vote (Labour + Green + Lib Dem) is comfortably larger than the broadly right-wing vote (Reform + Restore + Conservative). And the seat can still go to a candidate the majority did not want.

That is not bad luck. It is what first-past-the-post does, by design, whenever there are more than two serious choices on offer.

Burnham himself has publicly backed proportional representation — a position he reached after winning the Manchester mayoralty under a preferential ballot that Conservatives later abolished. Sir Keir Starmer’s view, in effect, is that PR’s time has not come.

So Britain’s most prominent Labour advocate of electoral reform now has to win a single-mark-plurality seat, in order to lead a party whose leadership rejects his view, so that he might one day be in a position to do something about the system that, in the meantime, may well be what stops him getting there at all.

A grown-up next step

Disclosure first: I advise Citizen Sector, which is part of the coalition asking Parliament for this Commission, and I donate to Make Votes Matter. The argument that follows is one I have a stake in. It is also one I believe.

I am not asking readers to agree with me about which voting method should replace first-past-the-post. I have my views — approval voting for single-winner contests, proportional methods for multi-member ones — and reasonable people land in different places.

I am asking for something smaller, and more important: that Parliament establish an independent National Commission on Electoral Reform.

A Commission would set out the criteria a strong voting system should meet, evaluate how well our current one delivers, examine credible alternatives, run a real public conversation, and report back to Parliament with evidence-based recommendations. Parliament — not the Commission, and not me — would then decide.

Forecasts already show a hung parliament whose composition turns on how the right-wing vote splits rather than on what the country actually wants. We can either walk into that knowingly, with the evidence in front of us, or we can walk into it the way we walked into 2024 — by accident, and then complain about the result.

It has been 140 years since Salisbury settled the question for us in his drawing room. Andy Burnham’s bid for Makerfield is as good a moment as any to ask it properly, in public, this time.

If you agree, sign the petition.

The system isn’t broken. It’s working.

Tyseley & Hay Mills is not the system breaking down. It’s the system succeeding, exactly to the specification it was given. Every multi-candidate first-past-the-post race is a building completed to a 141-year-old blueprint that says: whichever candidate has the largest single pile of votes wins, regardless of how small that pile is, regardless of whether more voters preferred someone else, regardless of whether the winner has any meaningful support at all.

The blueprint works. The building is up. We have been complaining about the shape for several decades, but the architect would, if he could be reached for comment, point out — entirely fairly — that he built what he was asked to build.

The councillor in Tyseley is not the villain. He stood, he campaigned, he won under the rules he was given. He played the game in front of him. The system designed in 1884 to behave exactly this way is the villain. You can hold both thoughts at once. We’re going to.

The audit

For the 2026 cycle I built electionresults.uk — an audit of every council seat in England and Wales, comparing every winner’s vote share against the Droop quota: the mathematical floor that any standard proportional voting method (Single Transferable Vote, D’Hondt, Sainte-Laguë, the regional-list tier of Mixed-Member Proportional) requires to guarantee a seat. The quota is one over the number of seats plus one — 50% in single-member wards, down to 20% in four-seat wards.

Anyone whose vote share fell below the quota for their ward would not have been guaranteed that seat under any of the common proportional alternatives. That makes the count voting-method-neutral by construction. We are not asking you to choose between proportional voting methods. We are showing you how many wins, in May 2026, would have failed under all of them.

Across 136 councils, 2,943 ward races, and 5,031 seats:

Almost one council seat in three this May went to people who would not have won under any fair count.

That’s the headline. Now look up your own council. The site lets you search by postcode, by ward, by council name. Most readers, I suspect, will find at least one seat in their area in the bin.

Wakefield. Sutton. The two parties that prove the same point.

The aggregate is the story. The cases are the receipts.

In Wakefield, Reform UK won 44% of the vote and 92% of the seats — 58 of 63. Under proportional allocation they would have won about 30. That is the largest single-council distortion of the cycle. It happens to favour Reform. In 1983 it favoured the Conservatives. In 1997 it favoured Labour. In 2024 it favoured Labour again, and so on through every landslide of the modern era. The system rewards whichever party happens to have its vote efficiently distributed in a given moment. That is what makes the machinery arbitrary, not what makes it partisan.

In Sutton, the Liberal Democrats won 44% of the vote and 93% of the seats — 51 of 55. The Lib Dems are the party whose entire constitutional platform demands the abolition of first past the post and its replacement with proportional representation. They will be the first to tell you they shouldn’t have won 93% of those seats. They are also the people who happen to have benefited from one of the most disproportional council outcomes in England this year.

I want you to sit with the symmetry. Two opposite parties; one extreme outcome each; identical machinery; same week. Reform and the Liberal Democrats may agree on almost nothing else, but on the question of whether the Sutton and Wakefield results reflect what voters actually said, they would tell you the same thing: no.

If you needed permission to feel angry about this without feeling partisan, the Lib Dem result in Sutton is your permission slip. They didn’t ask for those seats. They are not entitled to them. They were handed to them by an accident of geography and a 141-year-old blueprint. That is a fact that ought to outrank tribal loyalty, and on this issue it does.

Each minority winner is a receipt

Think of every below-quota win as a receipt — a small withdrawal made against an account that was opened in private, in 1884, in a room at Lord Salisbury’s London residence in Arlington Street. Three men did the negotiating: Salisbury himself, William Gladstone, and Gladstone’s Cabinet emissary Sir Charles Dilke. They settled the British parliamentary system on single-member single-mark plurality, and quietly abolished the minority-representation clause from the 1867 Reform Act. The Redistribution of Seats Act 1885 followed. We did not vote for it. We were not consulted on it.

The grim joke is that Salisbury, weeks earlier, had publicly argued the opposite case in print. In The National Review in October 1884 he endorsed cumulative voting and warned that single-member equal districts would convert “the right to vote into a right to be out-voted.” The words are his. Then he walked into the room at Arlington Street and signed the country into the system he had just attacked.

Every receipt this May was a withdrawal against that account. 1,593 receipts in one cycle. Multiply across decades and the figure becomes ordinary. That’s the thing about institutional theft: at scale it stops looking like theft. It starts looking like the weather.

Scotland already closed the account

Walk three hundred miles north and the same election week looks completely different.

Scottish councils have used Single Transferable Vote since 2007. Same kinds of councils as in England, same political parties, same general media, same cost of living, same week of election count, same broad political culture. Different ballot. Voters in Scotland rank as many or as few candidates as they wish; their votes transfer when no longer needed; multi-member wards return three or four councillors instead of one.

In the 2026 Scottish council elections, 9.0% of council seats fell below quota. In England and Wales: 19.6% across the same multi-member ward sizes Scotland uses.

Same voters, same parties, same system architecture above and below — and half the rate of unrepresentative outcomes on the side that uses a richer ballot. The Scottish data is not from a thought experiment, an academic model, or a foreign country. It is from people in Glasgow and Edinburgh and Dumfries and Aberdeen casting their votes in the same election cycle as people in Wakefield and Sutton and Birmingham, on the only major piece of British electoral machinery that doesn’t trace back to Salisbury’s pencil note.

Northern Ireland has run STV since 1973. The Welsh Senedd shifted off the Additional Member System and onto closed-list proportional representation for the 2026 election. Every devolved or local UK polity that has been given a richer ballot has kept it. The only place in the UK that has not been given the choice is the place that matters most.

The five things you’ll be told this isn’t

Before the comments fill up, let me concede the strongest sub-claim on each side.

“Proportional representation breaks the constituency link.” Closed-list PR can — that’s why the Welsh Senedd’s 2026 closed lists are already controversial. But Single Transferable Vote, used in Northern Ireland for half a century and in Scotland for nearly two decades, strengthens the link: most voters end up with at least one local representative they actually voted for. The constituency-link defence only works against the worst PR variants.

“PR produces unstable, indecisive government.” Arend Lijphart’s 36-country study found the opposite: proportional democracies match or exceed majoritarian ones on stability and policy performance. Coalitions are the modal form of European parliamentary government, and Norway, Germany, Ireland, Denmark and the Netherlands have run more cohesive long-run policy than majoritarian Britain. The 2024 UK general election, in which 33.7% of the vote produced 63% of the seats, ought to be the final death of the “stable government” defence: that is not stability, that is a brittle artefact of vote distribution that will flip on a small swing.

“Ranked or list ballots are too complicated for British voters.” Northern Ireland has used STV since 1973, Scotland since 2007, the Republic of Ireland since 1922. Spoiled-ballot rates are comparable to plurality elections. The argument that British voters, uniquely, cannot handle a richer ballot is condescending and unsupported.

“PR rewards extremists.” Reform UK got 14.3% of the 2024 general election vote and five seats out of 650; under any common PR system, around ninety. The Greens got 6.7% and four seats; under PR, around forty. First past the post does not suppress fringes; it suppresses anyone whose voters aren’t geographically clustered. That is a different thing.

“The 2011 Alternative Vote referendum settled this.” It didn’t. The 2011 vote was on plain Alternative Vote — not even a proportional system — and the campaign was tied to Nick Clegg’s collapsing personal popularity. I am not a fan of ranked voting methods, and the complexity there is a fair critique, but it’s irrelevant to the question of proportional representation. Polling on actual proportional representation has moved by twenty-five points since then. As of 2025, YouGov has 49% of the public for PR and 26% against — a record gap, with majority support among voters of every major Great-Britain-wide party including Conservative and Reform.

What to do, if you’d like to do something

If you’ve read this far, the system has already lost the argument with you. The remaining question is what you do with the conclusion. Asks, in order of friction:

The lowest-friction action: sign the National Commission for Electoral Reform petition. It takes thirty seconds, it puts a signature against a number, and the number is the most useful single piece of pressure on the political class.

Look up your own council on electionresults.uk. Find the receipts. Share the worst one in your area. The case for reform gets made one specific ward at a time.

If you’re in or near the Labour Party, the political bottleneck on UK electoral reform is not the public, it is Labour’s leadership. The 2022 Labour Conference passed a PR motion that Keir Starmer immediately disowned; PR was absent from the 2024 manifesto, despite Labour going on to benefit from the most disproportional election in modern UK history. The pressure points that work:

• write to your MP citing the audit’s specific local results in your constituency

• join the Labour Campaign for Electoral Reform or Labour for a New Democracy

• raise a motion at your CLP for the next conference cycle

Labour can be moved on this. The 2022 vote proved it. It just needs to be moved by people inside the party.

If you’d like to learn how the alternatives actually work before you commit, proportional.uk is the most honest tour of the options I know how to write.

When does the rest of Britain un-choose?

In Tyseley & Hay Mills, the building works exactly as the architect intended. The receipts are still being printed. They will keep being printed at every cycle until somebody somewhere in Westminster writes a different note than the one Lord Salisbury wrote in private in 1884.

Scotland already wrote it. Northern Ireland wrote it half a century ago. Wales wrote a version of it last month. The political weather has moved. The polling has moved. The audit has the receipts.

It’s not the British public that needs convincing. It’s a Westminster political class that benefits, in alternating turns, from the same machine that elected a councillor on 20.5% of the vote.

We don’t have to keep paying the bill.

Sources for the 2026 audit, including methodology and per-council breakdowns, are at electionresults.uk/methodology.

Disclosure: I built electionresults.uk and proportional.uk, and I volunteer with Citizen Sector, and Make Votes Matter, the UK proportional-representation campaign.

If this essay was useful, the most useful thing you can do with it is forward it to one person who isn’t already convinced. The case for voting reform is made one reader at a time.

The St. Louis, MO Primary Municipal Election was held on March 4, 2025. The election included multiple races: Mayor, Comptroller, Alderman Ward 3, and Alderman Ward 11. While we have cast vote record data for all these races, this analysis focuses specifically on the mayoral contest, which provides the richest dataset for understanding voter behavior under approval voting.

In the Mayor’s Race, Cara Spencer (68.2%) and incumbent Tishaura O. Jones (33.2%) advanced to the general election, defeating Michael “Mike” Butler (24.9%) and Andrew Jones (13.6%). With 34,945 ballots cast and 48,908 total approvals, there was an average of 1.4 approvals per ballot. St. Louis runs a unified, non-partisan primary in which candidates may run and are listed without any party affiliation next to their name. The top two winners go head to head in the General Election. This is the sixth ever election in the United States using Approval Voting, and the first ever approval voting election where we have the Cast Vote Records from the election. These records reveal fascinating details on St. Louis voters’ preferences.

Why St. Louis Adopted Approval Voting

St. Louis moved to approval voting following the embarrassing 2017 mayoral primary, where Lyda Krewson won with just 32.0% of the vote against six Black candidates. This election starkly demonstrated the devastating effects of vote splitting under the traditional “Choose One” system — a white candidate prevailed despite most voters preferring other options, simply because the majority coalition was divided among multiple candidates. The 2017 result made clear that St. Louis needed a voting system that could reveal true voter preferences rather than rewarding vote splitting.

Understanding Cast Vote Records

We’re used to looking at voting results as the aggregates— for example, a candidate got 52% of the vote. The Cast Vote Record allows us to see how a voter voted across all of the races on their ballot. For the normally used “First Past the Post” voting system where you Choose One this doesn’t often provide a lot of insight, but Approval Voting elections are different.

Under Approval Voting you can pick all the candidates you like. Put another way, your ability to support a candidate in a race is completely independent of your ability to support other candidates. From a data perspective, each candidate is in their own race, and voters choose to either support them, or not. This is where Cast Vote Records can reveal very interesting details. CVRs can show us, ballot by ballot, people’s preferences and allow us to see which candidates voters thought were similar, and how voters were able to use the flexibility of approval voting to support multiple candidates.

Approval Distribution: How Voters Used the System

Our analysis of the Approval Distribution reveals that a third (32.8%) of voters supported more than one candidate, with a quarter (26.1%) approving exactly two candidates, 6% approving three candidates, and half a percent (0.4%) approving all four. When we break down the approvals by candidate, we find dramatic differences in voting patterns. While supporters of the front-runners showed more focused support (38.8% of Spencer supporters and 41.9% of Tishaura O. Jones supporters approved multiple candidates), supporters of non-front runners Michael “Mike” Butler and Andrew Jones were far more likely to express broader preferences, with 84.4% and 83.1% respectively voting for more than one candidate.

Debunking the Bullet Voting Myth

One of the great myths used against Approval Voting is that voters will “Bullet Vote” (only approve of one candidate). FairVote (the US organization; FairVote Canada and FairVote UK are unassociated) insists that in Approval Voting elections voters will not take advantage of the ability to support multiple candidates because of the chance that a less favored candidate will win because of their vote. In the worst case scenario, they hypothesize, this would devolve an approval voting election into a “Choose One” First Past the Post election, along with the horrific vote splitting we see in those elections.

FairVote defines bullet voting as:

Bullet voting: insincerely expressing a preference for only a single candidate to increase that candidate’s chance of victory. This strategy applies to any degree of insincere preference truncation, such as expressing a preference for two candidates when one sincerely prefers three.

— Fairvote.org - Comparing single-winner voting methods

The St. Louis data decisively refutes these concerns. Our analysis demonstrates that in real-world elections voters use approval voting to support all the candidates they like. While some voters genuinely preferred only one candidate (a perfectly valid choice), the system enabled nuanced preference expression where it mattered most. For supporters of Mike Butler and Andrew Jones - voters whose voices would have been completely silenced in a traditional primary - 84.4% and 83.1% respectively were able to express additional preferences. This wasn’t strategic voting or gaming the system; it was honest preference expression that approval voting enables.

The Expressiveness Question

RCV supporters claim that approval voting is less expressive than ranked systems because individual ballots don’t show fine-grained preferences between approved candidates. This intuition, while understandable, misses how approval voting achieves expressiveness at the aggregate level.

Consider a voter who approves Spencer and Butler but not Tishaura O. Jones or Andrew Jones. While their ballot doesn’t rank Spencer versus Butler, it does reveal four preference relationships: Spencer > Tishaura O. Jones, Spencer > Andrew Jones, Butler > Tishaura O. Jones, and Butler > Andrew Jones. Other voters make different approval choices—some might approve only Spencer, others might approve Spencer and Tishaura O. Jones, and still others might approve all except Tishaura O. Jones. These varied patterns collectively reveal the electorate’s preferences with remarkable precision.

This is where the law of large numbers becomes crucial. Individual voters who can’t differentiate between two candidates they both like statistically tend to have similar relative preferences to the broader electorate. Research by voting systems experts shows that approval voting consistently produces more accurate results—meaning outcomes that better satisfy the average voter—than even ranked choice voting with completely honest voters, because approval voting captures what matters most: which candidates voters genuinely support versus those they oppose.

Coalition Patterns: The Co-Approval Matrix

The Co-Approval matrix shows the fraction of voters who approved one candidate that also approved the other. This reveals voting patterns and candidate coalitions. The data shows fascinating cross-support patterns: 45.7% of Andrew Jones supporters also approved Mike Butler, while 25.1% of Butler supporters reciprocated that support. Both of these candidates’ supporters showed significant cross-approval with Cara Spencer (60.6% of Butler supporters approved Spencer, 67.1% of Andrew Jones supporters approved Spencer). In contrast, there was much less overlap between Tishaura O. Jones and the other candidates, with only 26.9% of Butler supporters approving Tishaura O. Jones and just 10.4% of Andrew Jones supporters approved of Tishaura O. Jones. This suggests Cara Spencer, the highest approved candidate, successfully positioned herself as a consensus candidate who could attract support from across the political spectrum.

Election-Wide Coalition Patterns

The election-wide co-approval matrix extends our analysis beyond the mayoral race to examine potential coalitions across all races on the ballot—Mayor, Comptroller, and the two Alderman contests. While this cross-race analysis reveals interesting political dynamics and voter behavior patterns, it’s important to note that these correlations are more academically fascinating than practically relevant to approval voting’s performance. The matrix shows how voters who supported certain mayoral candidates tended to align with specific candidates in other races, offering insights into the broader political coalitions and ideological groupings within St. Louis. However, unlike the within-race co-approval patterns that directly demonstrate approval voting’s ability to reveal voter preferences and candidate positioning, these cross-race correlations primarily illuminate the existing political landscape rather than the voting method’s effectiveness.

Opposition Voting: The “Anyone But” Analysis

The “Anyone But …” analysis examines the 2,200 voters (6.3% of all ballots) who marked every candidate except one. These ballots represent the strongest form of negative voting possible under approval voting - essentially saying “I’ll support anyone except this person.” The data reveals stark opposition patterns: 64.7% of these voters specifically excluded incumbent Mayor Tishaura O. Jones, while 26.7% excluded Andrew Jones, 6.3% excluded Cara Spencer, and only 2.3% excluded Mike Butler. This suggests Tishaura O. Jones inspired the most ardent opposition, while Butler was viewed least unfavorably—even by voters who didn’t actively support him.

Comparing Approval Voting vs Ranked Choice Voting

Having the full cast vote records from St. Louis creates a unique opportunity to compare voting systems in practice. The site that hosts this analysis, approval.vote, is a sister site to ranked.vote, which provides similar detailed results for ranked choice voting elections.

The contrast in data complexity is striking. Approval voting produces clean, interpretable metrics like our Approval Distribution charts and Co-Approval matrices that clearly show voter coalitions and preferences. Ranked choice voting, by comparison, generates far more complex data patterns. The best visualization for RCV is the Sankey diagram—see the 2021 NYC Mayoral election as a prime example, where 13 candidates required eight elimination rounds to reach a conclusion.

Instead of our straightforward approval metrics, RCV analysis relies on Pairwise Preferences, First Alternates, and Final Vote by First Choice breakdowns. This complexity isn’t just academic—it reflects how much harder RCV makes it for voters to understand outcomes and for analysts to extract meaningful insights about voter preferences, which ultimately reprsents itself in mistrust in RCV elections.

For a deeper dive into these system differences, read this detailed comparison between RCV and Approval Voting.

The Verdict: Approval Voting Works as Intended

The St. Louis data conclusively demonstrates that approval voting functions exactly as designed. Far from the “bullet voting” dystopia predicted by critics, we see a nuanced electorate that used the full flexibility of the system: one-third of voters supported multiple candidates, supporters of lesser-known candidates overwhelmingly expressed broader preferences, and clear coalition patterns emerged that would be invisible under traditional voting.

Perhaps most importantly, the system revealed genuine voter sentiment that traditional voting would have obscured. Butler and Andrew Jones supporters didn’t “waste” their votes - they expressed their full preferences while still participating meaningfully in the broader electoral choice. Spencer’s cross-coalition appeal and Tishaura O. Jones’s polarizing but still successful campaign both tell stories that “Choose One” voting simply cannot capture.

On April 8th, 2025 St. Louis had its General Election. Cara Spencer (64.15%) defeated Tishaura O. Jones (35.85%), demonstrating consistency between an Approval Voting election and a “Choose One” election with only two candidates.

The St. Louis experience demonstrates how better voting methods create a virtuous cycle of democratic improvement. Approval Voting made elections genuinely competitive, forcing representatives to work harder to earn and keep their positions. When politicians must appeal to broader coalitions rather than narrow bases, they craft more thoughtful policies that serve wider constituencies. These better laws and governance decisions compound over time, transforming St. Louis from a city known for dysfunction into one with responsive, accountable leadership. The result isn’t just better politics—it’s better outcomes for residents in everything from city services to economic development.

This transformation couldn’t have happened without the vital work of voting reform organizations: the Missouri based Show Me Integrity, and the national Center for Election Science. Both organizations remain critically underfunded despite their outsized impact. If this analysis resonated with you, consider supporting their work. Changing how we vote matters far more than winning any individual election—it’s the foundation upon which all other democratic improvements rest. For those who donate to causes, there’s no investment with higher leverage for positive change.

Special thanks to the IT team at the St. Louis City Board of Election Commissioners for making this data available.