Our voting intention methodology

Our headline voting intention comes from an online sample of 2000+ adults in Great Britain, weighted to be representative of the population (voters and nonvoters in 2024) and then narrowed down to those who say they are likely to vote – with undecided voters given a party wherever we can reasonably tell which way they are leaning.

1. Weighting. The sample is weighted by region, age and sex (interlocked), whether people have a degree, their ethnicity and how they voted at the 2024 general election, including the half or so of adults who either did not vote or can’t remember whether they did, so that the sample is representative of the GB population.
2. A “squeeze”. Anyone who tells us they don’t know how they would vote is asked which party they would pick if they were forced to choose. Those who choose a party get added to our headline calculation
3. Undecided voters. Those still undecided after the squeeze are given the party that people most like them are likely to support, using a model based on their 2024 vote, demographics, EU referendum vote, region and housing tenure, that is built afresh from each poll
4. Turnout. Only those who put their likelihood of voting at 8 or more out of 10 are included in the final result, or 9 or more for those who didn’t vote in 2024. This is because whether someone voted in the last election is one of the best predictors of whether they will vote in a future one

What each step does to the different parties

Our weighting, squeeze and turnout models all have small effects on each party’s vote share. In this week’s poll (6-9 November 2026),  Labour is three points higher in the raw figure than the headline, the Conservatives are two points higher, and Reform’s vote share changes by less than two points between the raw data and the headline.

Meanwhile the Liberal Democrats, the Green Party and Restore Britain’s vote shares are slightly higher in the headline results than the raw figures.

Of all the stages in our voting intention calculation, weighting is the one with the greatest effect on vote shares. This week it takes four points off the Conservatives, two off Labour and one off Reform, and adds around two points each to the Liberal Democrats, the Green Party and Restore Britain, while the squeeze and the allocation of undecided voters move no party by more than half a point in either direction. The turnout filter then adds two points to the Conservatives and takes a point or so off the Greens and Labour, since Conservative supporters are more likely to say they are certain to vote – although at this time last year the same filter added one and a half points to Reform, which suggests it tends to follow whichever party’s supporters are most enthusiastic at the time.

What’s changed since last year?

Comparing this week’s poll with the same one a year ago, the methodology is be doing much the same job each time – every party’s headline has moved in the same direction as its raw figure, with Reform down nine points in the raw data and eleven in the headline, and the Conservatives up by five or six points on both – in other words, the changes to each party’s standing in the past year is mostly a reflection of what respondents themselves are telling us.

Factoring in less-likely voters is never going to be a reliable boost for a given party – their preferences have changed in the past year, much like the rest of the public’s. In the lead up to the local elections in 2026, Reform polled at 41% among those who did not vote in 2024, significantly ahead of Labour. We see in our data that this lead has shrunk significantly, with 2024 nonvoters now more evenly split between Reform, Labour and the Tories.