Not so Fast on Fragmentation: Why local party choices aren’t nearly as fragmented as the national polls
The fragmentation of party preferences has been the central story of this electoral cycle so far. After winning a parliamentary majority on a historically low vote share in 2024, Labour's support fell precipitously under Starmer. While recovering a little under Burnham, this is not yet close to the level in 2024. The Conservatives, meanwhile, have failed to capitalise on Labour’s collapse, and challenger parties have reachedvote shares that would once have seemed remarkable: Reform UK leading the field at around 27% for most of 2025 and the beginning of 2026, and the Greens and Liberal Democrats holding steady in double figures.
The result is a fragmented national landscape unlike anything British politics has ever experienced: four or five parties polling simultaneously above ten percent, and a frontrunner whose lead largely reflects the sheer degree to which the vote has scattered. On the face of it, this forebodes ‘chaotic’ electoral outcomes, as splintered party preferences collide with an electoral system designed for two or three competing parties. It raises the prospect of seats won on minority vote shares, tighter multi-party contests, and previously safe outcomes becoming vulnerable to very small swings.
This upheaval, however, will depend on the extent to which the national fragmentation is also replicated at the local level. So far, this is not what we find. Evidence from by-elections and local elections shows that fragmentation has increased – both nationally and locally – but unevenly across local contexts, with only a minority of seats truly seeing multi-party races emerge.
This makes a great deal of sense. A famous 'law' of political science, ‘Duverger’s Law’, named after the French political scientist Maurice Duverger, predicts exactly this: under a first-past-the-post system like ours, local contests settle into races between the largest two-party options, however many are competing nationally. Why? One reason is geography: parties do best where their kind of voter lives. But it is also about how voters behave. Duverger identified that in majoritarian electoral systems, voters have a strong incentive to work out which candidates can realistically win in their own constituency. Supporting anyone else risks wasting their vote. Over time, support therefore concentrates on the two candidates best placed to win locally – the most ‘viable’ - even among voters who would prefer a third, fourth, fifth (and so on) party.
In Britain, this choice now plays out within two broad camps or ‘blocs’: Labour, the Lib Dems and the Greens on one side, the Conservatives, Reform UK and Restore Britain on the other. Voters who change their minds have mainly, recently, been moving within their own camp rather than crossing over. So, the real choice in most places is between the strongest party in someone’s own bloc and the strongest in the other. Which two those are varies hugely from place to place, and that is why the national picture looks so much messier than the local one.
What do we find?
Let’s look at the election results that help us separate the national polls from local-level election results.
We use a measure of the ‘effective number of parties’ to calculate the meaningful fragmentation level in results, which counts how many parties are really in play by weighting each party by its share of the vote.1Where the vote is more evenly split between a greater number of parties, the effective number of parties is higher.
The red line in Graph 1 tracks how fragmented national polls have been since the 1980s, with the 'effective number of parties' in local and national elections measuring, on average, the number of parties gaining a meaningful amount of support in the polls. The blue and green box plots show the spread of fragmentation across general election constituencies (green) and local election wards (blue) in each election since 1983, with the middle dots indicating the median level of fragmentation at the seat level at that election.
Graph 1 – Fragmentation at the national and seat level over time
The red line for polls takes a smoothed average. Beneath that, in pink, are the observed values, which show, in 2019, a spike in fragmentation coinciding with the May 2019 European Parliament elections, which very clearly disappeared, until after the 2024 general election.
The rise in fragmentation at the national level seen since 2024 is unprecedented. Until now, national fragmentation had hovered between 2.5 and 3.5 ‘effective parties’, reflecting a landscape in which support was spread, to varying degrees, across the Conservatives, Labour, the Liberal Democrats and nationalist parties in Scotland and Wales. By contrast, in this electoral cycle fragmentation has entered uncharted territory: in scale, now spreading over an average of 5.3 parties, and in duration, having risen consistently for over two years.
At the seat level, however, fragmentation in general election and local election seats tends to be significantly lower than national fragmentation. With few exceptions, over 80% of local races are less fragmented than the overall picture in national polls at the time. This is also true for general election seats in 2024, and local election wards in 2025 and 2026. On top of this, fragmentation at the seat level has not increased as rapidly as the fragmentation in national polls. On the contrary, after surging in 2024, seat-level fragmentation appears to be stabilising, or plateauing. Despite the 2025 and 2026 local elections taking place in largely different subsets of seats, their fragmentation distributions at the ward level are almost identical.
Additionally, a greater number of vote winning parties does not necessarily mean three, four or five-party races. Fragmentation can just as easily show up as ‘reserve’ votes –support for parties still far from contention – as it can produce genuine multi-party marginals.
This is shown in Graph 2, which plots the number of parties in contention (defined as within 20 percentage points of the winner) in local election wards in May 2026. Even where three or four parties are picking up a meaningful vote share, the contest itself is still most often a one- or two-party race. National fragmentation, in other words, has not yet dissolved local two-party-ism, just as Duverger’s Law predicts.
None of this makes fragmentation harmless. A jumble of local two-way fights, each between a different pair of parties, still potentially adds up to councils and a parliament where nobody has a majority. May's results showed exactly that. But the disorder is a particular kind. It is not that voters everywhere now have five serious options. It is that the two serious options are different from place to place, and are often no longer the two we (or voters) are used to.
That makes the real fight an internal one. Within each camp, parties are competing to be seen as the one that can win locally. That perception is what turns sympathetic voters into actual votes, and it is now up for grabs in a way it has not been for decades. Every result feeds the next: a strong second place tells voters in similar seats who to back next time.
It is why a single contest can matter far beyond the seat itself. The upcoming by-election in Holborn and St Pancras, for instance, is not primarily a fight between left and right. It is more a fight between two parties of the left. Whichever of Labour and the Greens wins will have shown voters well beyond north London which of them is worth backing based on their electoral viability. For a century, Labour and the Conservatives were the automatic answer to the question voters keep asking: who around here can actually win? They cannot currently count on that anymore.
Authored by: Jane Green, Professorial Fellow at Nuffield and Director of the Nuffield Politics Research Centre; Marta Miori, UK Politics Impact Research Fellow.