by John Curtice, Stephen Fisher, and Patrick English, 7^th May 2026.

One of the highlights of the BBC’s coverage of local elections is the Projected National Share (PNS). This is an estimate of what the GB share of the vote would have been if (a) local elections had been held everywhere, (b) the outcome where there were no elections mirrored the pattern where there were, and (c) the principal parties contested all the seats.

Although there are devolved elections in Scotland and Wales today, the PNS is always based entirely on local election results in England. This means that the PNS this year will tell us about what would have happened if voters across the whole of Britain had local elections and the results mirrored what happened in the particular parts of England with local elections this year. It will not take account of the results of the entirely separate devolved elections in Scotland and Wales. To that extent, even though the PNS is expressed as a GB share of the vote, it is best thought of as an indicator of political developments in England.

A key virtue of the PNS is that it provides a summary statistic that is comparable across local election years irrespective of the particular mix of places in England that have local elections in any particular year, a mix that varies considerably from year to year. 

This year’s local elections, for example, take place primarily in urban England, including all of London. Consequently, they are places where Labour has hitherto at least been relatively strong. They are also places that on average voted for the UK to Remain in the EU at the 2016 referendum even though Britain as a whole voted to Leave. The PNS is designed, among other things, to remove this unrepresentative character.

In this blog we outline some of the challenges in estimating a PNS and explain how we are addressing them. 

In recent years, the PNS has been calculated as follows. First, we model the change in party support in a sample of ‘key’ wards that have been contested by all of the Conservatives, Labour and the Liberal Democrats in the current year and in a previous baseline year. This modelling generates an estimate of the change in party support for every local authority in Britain – both those with and those without a local council election – that takes into account the systematic variation in party performance in the local elections and the political character of each local authority. These estimates are then aggregated across all authorities to produce an estimated nationwide change in each party’s support. This estimated change is then applied to the PNS for a number of previous baseline years to generate a set of estimates for the current year. The final PNS is an average of these various estimates. This approach reduces the risk that the calculation is unduly influenced by possible error or distortion in the PNS estimate for a particular baseline year.

When local elections happen on the same day as a general election, and where the boundaries of local election wards are nested within individual or clusters of parliamentary constituencies, it is possible to estimate the PNS by comparing the general and local election results for the same geographical units. This provides is with a firm baseline for estimating the PNS in subsequent years. Alas for us poor psephologists, prime ministers have not been kind enough to call a general election on the same day as a local election since 2015. That means the baselines for recent PNS projections have themselves been previous projections. And so, over the last decade, to some extent the PNS has been based on projections from projections from projections.

Last year, however, in a modest number of places, we were able to calculate change in vote share between the 2024 general election and the 2025 local elections. That comparison was used to inform the 2025 PNS in conjunction with the council-level modelling approach discussed above.  

This year, we are fortunate that, despite differences in some places between ward and parliamentary constituency boundaries, there will be many more places where the results of the local elections can be aggregated to the (exact) boundaries of parliamentary constituencies. Moreover, in many instances all the component wards are being contested by all five main parties. That will enable us not only to calculate the average and overall change between the 2024 general election and 2026 local elections in these constituencies, but also build detailed statistical models of the variation in the pattern of change for all five parties across those constituencies. Those models can then be used to project what would have happened in other constituencies if they had had local elections this week. Adding up those projections across Britain will thus provide an estimate of the PNS that is built on the (potentially more secure) foundations of the 2024 general election results. Consequently, this year’s PNS will take significant account of the estimate produced by this approach.

Last year was the first at which the BBC published separate PNS figures for both Reform and the Greens. We explained then that, in the hypothetical ‘as if’ election whose outcome the PNS attempts to estimate, we were now, in effect, anticipating that all five parties would fight every seat in Britain, whereas previously the presumption had been that this would only be true of the Conservatives, Labour and the Liberal Democrats. 

The methodological difficulty in estimating a five-party PNS via comparing local election results this year’s with those in previous years is the patchy pattern of Green and Reform candidature in previous local elections that serve as a baseline year. Reform barely contested the local elections in 2021 or 2022. That means that in any comparison with those years, whatever share of the vote the party secures this time around will represent an improvement on zero. Even in the case of the Greens, the number of wards they fought previously is more limited than in the case of the Conservatives, Labour and the Liberal Democrats. Consequently, some of our ward level modelling this year will again focus wards fought by, at least, the Conservatives, Labour and the Liberal Democrats both this year and the baseline year, with the estimates for the Greens and Reform being derived from how the resulting Other vote has been distributed between Greens and Reform in the now many wards that both these parties have been contesting.

Meanwhile, with most of the local council results being declared on Friday, the PNS will not be published before late Friday afternoon. And while the current state of the polls means there is particularly intense interest in the message from the local ballot boxes this year, it should by now be clear that discerning that message will be far from straightforward. Any estimate of the party’s performances, including the PNS, will, in truth, be accompanied by more than the usual degree of uncertainty.

John Curtice is Professor of Politics at the University of Strathclyde.

Stephen Fisher is Professor of Political Sociology at the University of Oxford.

Patrick English is Head of Elections and Political and Social Data at YouGov.

Explanations of the PNS methodologies in previous years can be found at ElectionsEtc.com, for 2025, 2024, 2023, 2022, 2021, 2019, 2018, and earlier years.

Thanks to the BBC, Rob Ford, Lotte Hargrave, and Stuart Perrett for help with the data.