Best for Britain have today (27 November) released a new round of MRP data which shows that the Conservative Party could win up to 366 seats at the upcoming general election. However, their majority rests on them holding and gaining a significant number of increasingly marginal seats – meaning tactical voting will be decisive.

The seat-by-seat analysis of 39,476 British adults was carried out between 15 October and 24 November by Focaldata and is the first MRP poll to adjust for the impact of the Brexit Party standing candidates down in 317 Conservative held seats and 40 more non-Conservative held seats, as well as the Unite to Remain alliance.

Vist GetVoting.org to get your recommendation.

Without tactical voting, Labour, Lib Dems, SNP, Greens and Plaid Cymru would hold a combined 265 seats. This scenario would see the Conservative Party win up to 366 seats and a majority in Parliament, having been boosted by roughly three quarters (73.4%) of Brexit Party voters who no longer have a Brexit Party candidate in their seat.

However, while the data currently predicts a Conservative majority, many of the seats they are anticipated to hold or win are vulnerable. For instance the rise in Labour support, and fall in Conservative support, seen in the two recent polls (Kantar and ICM) would mean Labour gaining 24 seats.

Whereas at the beginning of the election campaign our October MRP revealed there were 131 seats where the margin of victory was less than 5,000 votes, there are now 165 seats in Great Britain which fall under this category – around three times the number of seats our data predicts that pro-EU parties would need in order to prevent a Johnson majority. 

Best for Britain have therefore identified 57 target seats where the chances of pro-EU voters successfully using their votes tactically to prevent a Conservative victory are highest. These include seats such as the Cities of London & Westminster, Pudsey and Gordon. According to our data, it would take less than 4,000 tactical votes in these seats to prevent the Conservative Party winning.

In 27 of these seats it would take less than 2,000 tactical votes to prevent a Conservative victory. In Keighley it would take just 29 Green and Lib Dem voters backing Labour to prevent the Conservative Party winning the seat.

Nationally, Best for Britain estimates it could take as little as 117,314 pro-EU voters using their vote tactically to prevent a Tory majority - representing less than 1% of those who voted in 2017.

The data, used to inform Best for Britain’s tactical voting recommendations at getvoting.org, is the second release by the Best for Britain campaign during this election period, with a final update due on 9 December - three days before polling day.

The campaign launched its tactical voting tool on 30 October at a press conference attended by former Liberal Democrat leader Vince Cable, Independent MP Dominic Grieve and Labour MP Anna McMorrin.

The latest recommendations have been released to coincide with the first wave of postal votes landing on doorsteps.

Key findings from the data

• Our data predicts that the Conservatives could win up to 366 seats, having been boosted by roughly three-quarters of the Brexit Party’s vote share where the Brexit Party have not fielded a candidate. Under this scenario, Labour would win 199 seats, Lib Dems would win 17 seats, SNP would win 44 seats, Plaid would win 4 seats and the Greens would win 1 seat
• However, if 4,000 or less pro-EU voters use their vote tactically in Best for Britain’s 57 target seats, the Conservatives would be reduced to 309 seats – a dozen short of a majority. Labour would win 244 seats, SNP would win 52 seats, Lib Dems would win 21 seats, Plaid would win 4 seats and the Greens would win 1 seat.
• Best for Britain estimates that the election could be swung against the Conservatives by as few as 117,314 pro-EU voters using their vote tactically.
• Despite the boost received by the Conservatives from the Brexit Party, there is still a substantial number of seats where tactical voting could be decisive. There are 165 seats where fewer than 5,000 tactical votes would be enough to prevent the Conservative Party winning the seat (compared to 131 at the beginning of the election campaign). In 27 of these it would take fewer than 2,000 tactical votes.

Commenting, Best for Britain CEO Naomi Smith said:

“Our data shows that tactical voting will be decisive at the upcoming election. 

“Even with the Brexit Party collapse, there are still lots of seats in play for remainers. 

“This is crucial as it means they could be won by pro-EU parties if voters hold their nose and vote for the party with the best shot of beating the Tories.

“Having updated our recommendations just as the first wave of postal votes land on doorsteps, we’re confident that this election is up for grabs. We can stop Boris Johnson, and stop Brexit.”

Commenting, Focaldata founder Justin Ibbetts said:

“There are 14 Conservative held seats where the margin of victory is lower than 2%. Clearly this election is on a knife-edge, where even small changes in the number of votes each party wins could swing key seats.”

Methodology

Focaldata is a polling company that specialises in using machine learning to generate more accurate results. In particular, it uses a technique called Multilevel Regression with Post-stratification (MRP) to map opinion poll data onto smaller geographic areas.

For this study Focaldata collected data from 39,476 respondents between 15th of October and the 24th of November 2019 using a range of online panel providers, including Panelbase, Opinium, Dynata, BMG, YouGov, and Focaldata’s own panel. Focaldata uses data from a range of suppliers to adjust for the ‘house effects’ between data providers.

Focaldata's MRP model uses a range of individual and constituency level variables, these include (but are not limited to) age, gender, education, VoteGE2017, VoteEURef, population density, % long term unemployed, % leave 2016, GE2017 vote share, Deprivation index and EU parliament 2019 vote share. A bayesian exploded logit model is used, which is fit using Hamiltonian Monte Carlo with the open-source software Stan. The models are trained on Google Cloud Platform.