The UK’s messiest election ever?

What do we know about the next election? In terms of the final outcome, not very much. One thing we do know is that voters feel angry, frustrated and let down by politics and politicians of all stripes. As such, the next general election looks set to be the messiest, and most difficult to call, yet.

I ignore the week-to-week noise of small movements in voting intention polls (and so should you), but you would have to be living under a rock for the last 18 months to not notice the scale of the public’s rebellion against our political system and the ‘rules’ that used to govern it. Voters on all sides are frustrated and looking for change. It would have been unthinkable two years ago, but Reform has now led every single voting intention British Polling Council poll for almost a year, with a newly insurgent Green party battling for second place. The recent Gorton and Denton by-election had Greens and Reform in first and second place, with the combined vote share for Labour and the Conservatives at just 27 per cent. Indeed, it looks like Reform and Greens will also top the tables of the upcoming set of local elections on May 7th.

With a political map this fractured, trying to predict the parliamentary outcome of an election … can be a fool’s errand

Join Britain’s most civilised publication.

Challenge the consensus. Access rigorous analysis.

Archive article

Don't worry. You can continue reading by subscribing to get full access.

Subscribe

Already a member? Log in.

Premium article

Don't worry. You can continue reading by subscribing to get full access.

Subscribe

Already a member? Log in.

Subscribe Now

However, even the Greens and Reform are struggling to persuade voters to support them in huge numbers. Trust in politicians and the political class is at an all time low, and (for now) it looks unlikely that any party will consistently poll above 30 per cent, with 5 political parties polling within just 10 per cent of each other. We may even see further fracturing still with independent candidates likely to do well and a prospect of the right splitting further in the form of Rupert Lowe’s party, Restore.

With a political map this fractured, trying to predict the parliamentary outcome of an election tomorrow, let alone in 3 years time, can be a fool’s errand. But that hasn’t stopped people trying. We have already had several MRPs based on voters’ current behaviours.

We see seat projections in newspaper headlines and across social media, even in the middle of election cycles. Most of these are MRPs. MRP (multilevel regression and post-stratification) is a modelling technique that combines a large poll with other sources of information, primarily census data and past elections in the UK, to estimate local levels of support. Combining the model and polling then gives a result for each seat and can project a voting intention for each seat. In recent elections in the UK, MRPs have been an accurate predictor for the ultimate outcome of the election, and provide intrigue (and headlines) for journalists and politicians alike, who can look at (hypothetically) who will lose their seat to who. Several newspapers earlier this week ran with various headlines gleefully reporting that one such model would see more than half of Keir Starmer’s cabinet lose their seats.

In the 2024 general election campaign, the polling industrial complex astonishingly produced more than 20 MRPs and nearly 150 voting intention polls in just 6 weeks. However, even in 2024 the increasing complexity of our politics muddied the waters. Whilst these models all correctly called the ultimate outcome of a large Labour majority (it would have actually been difficult not to have done), there were already huge discrepancies when it came to what happened in individual seats. Projected seat outcomes for Labour during the campaign ranged from 430 to 516 (final result 412), the range for the Conservatives were from 56 seats to 180 (final result 121). The lowest Reform prediction was 0, the highest was 18 (final, 5). One firm even released two MRPs at the same time that showed significantly different results.

This is not to say that MRPs are not useful or should be disregarded entirely. Indeed, MRPs were often more accurate than the voting intention polls at the time, nearly all of which massively over inflated Labour’s lead. Much of the time pollsters give considered variations based on turnout and other factors, that give a sense of how volatile projected seat outcomes can be, although too often such nuances are completely ignored by everyone else.

However, they should not be treated as gospel. Predicting election results will increasingly necessitate pulling together lots of different data points, including models, polls and qualitative research, to try and get a more accurate picture of the public mood.

This will be even more important in 2029, which looks set to be even messier still, and an even tougher election to call. A 5 way (6 way in Scotland and Wales) fight means it will be difficult for the public, pollsters and pundits alike to work out who will win, or even who the best placed challenger will be in constituencies across the country. Already ahead of the locals political parties are citing different models with vastly different predictions in an attempt to squeeze out their opponents. One such model released in the past week showed 148 seats in Britain being won on less than 30 per cent of the vote. In 2024, there were just 7. Ayr, Carrick and Cumnock in Scotland were predicted to win with just 21% of the vote, with Labour, the Conservatives, SNP and Reform all competing with just 1 per cent between them. These circumstances will not only make the final tally incredibly unpredictable, but are likely to make any future government more unstable, with a record number of MPs sitting on ever-more precarious majorities and facing challenges on many fronts.

Voting intention trends are meaningful, and seat projections are useful are a useful part of the picture. But much of the noise coming from them for the next general election at the moment, is just that, noise. Rather than agonising over who may be up or down in a single constituency, at this point in an election cycle it is better to look at the bigger picture and trends in public opinion, including why voters are increasingly turning their backs on politics as we knew it in the first place.