Forensic polling analysis shows how Boris Johnson is on course to win—and how he can be stopped

At the mid-way point of this fateful election campaign, my firm Datapraxis is publishing what we believe to be the first really authoritative estimates of the state of the race.

For anyone who worries about Boris Johnson unleashed, there is both good news and bad news. Using the largest polling sample available, and mapping the dynamics of different groups of voters on to every seat, we currently project a 48-seat Conservative majority.

As things stand, the Tories look set to rise from the 317 seats they won in 2017 to 349. Labour, meanwhile, could crash from 262 to 213, perilously close to the 209 it won in the wipeout of 1983. After all the hype and their early hopes, the Lib Dems will add just two net seats to their 2017 base of 12. By contrast, the SNP could sweep Scotland, rising from 35 to 49 of the 59 seats.

If this outcome comes to pass, it would be a chest-thumping victory for Johnson and the hard Brexiteer strategems of Dominic Cummings. They would have their mandate to “get Brexit done,” and be liberated to pursue their whims and fantasies in the new Parliament—perhaps even taking the UK out of Europe without a trade deal by the end of next year.

But there are two flies in the ointment for the Johnson gang. The first is that seven of their most prominent members could lose their seats. They include: Foreign Secretary Dominic Raab (Esher and Walton), former Conservative leader Iain Duncan Smith (Chingford and Woodford Green), the ever-seething Europhobe John Redwood (Wokingham), the “illiberal” Philip Davies (Shipley), former London mayoral candidate Zac Goldsmith (Richmond Park), and European Research Group Chair, Steve Baker (Wycombe), together with—at a stretch—the vainglorious but vulnerable Prime Minister himself (Uxbridge and South Ruislip).

The report we publish this weekend, “Seven Seats That Could Change The Election,” lays out how tactical voting and increases in voter registration and turnout could contribute to some or all of them being defeated, even if the Tories win a majority. Raab and Duncan Smith are in particular trouble, while Goldsmith seems almost certain to lose.

The second and more important fly in the ointment for the Tories is that similar dynamics are replicated in almost 100 other seats around the country. Two and a half weeks is a long time in an election campaign; and a rebel alliance may be forming from the bottom up.

I am guided by a bipartite maxim of the Italian Marxist Antonio Gramsci. His principle of “pessimism of the intellect” calls on us to confront the scale of the challenges before us. That is why I decided to secure unique access to the largest source of private polling data in this election—hundreds of thousands of responses to constituency-specific voting intention questions from YouGov’s research panel, probably the largest and most representative in the country. Our client list is confidential, but we are working for a number of leading progressive campaigns and candidates in this election, as well as using the data for our own independent research and publication.

We have been asking additional, deeper questions of this vast sample about who is decided and who might change their mind, tactical voting, postal votes, attitudes toward the leaders and local issues, and testing hundreds of messages during the campaign. We have modelled the results to build an unparalleled analysis of where the race stands—and where it might still head—using a technique called “multi-level regression with post-stratification,” or MRP for short. MRP marries insights from a large national polling database, and detailed information on the demographic mix and past political behaviour for every seat in the country. It can make sense of voters through a whole host different characteristics, which combine in literally millions of different ways, then maps which seats they live in, their likelihood to turn out, and the way they are moving over the course of the campaign.

Like most things, MRP can be done well or badly. In 2017, Lord Ashcroft commissioned one such study which erroneously predicted a Conservative landslide. But YouGov’s final MRP estimates, based in tens of thousands of fresh polling responses and a rigorous design, correctly predicted a hung parliament, and called the winner correctly in 93 per cent of seats—including Canterbury and Kensington, which almost no-one expected to fall to Labour.

Today tactical voting sites attempt to tap the same methods, and are sometimes getting it right; but some of them are making bad calls based on out-of-date polling and flawed modelling (for example by recommending a vote for the Liberal Democrats in Truro and Falmouth or Filton and Bradley Stoke, two seats where our data shows Labour as the main challengers).

We have access to private YouGov internal numbers as well as over 250,000 polling responses; but the numbers we are releasing are from Datapraxis’s own in-house modelling. All the information we are seeing suggests that the 2019 general election could in fact be 650 different elections in 650 seats. Unpopular leaders and fading party loyalties, combine with Remain/Leave affinities, a very different Scottish dimension and the growing role of independent third party campaigns and considerations of tactical voting to mean that uniform national swing is no longer a good guide. Politics has never been so volatile.

All this leads me to the second part of Gramsci’s maxim, “optimism of the will” and the readiness to act. This is not an election for anyone to sit out. It is a battle between a handful of 800-pound gorillas in CCHQ, and the nimble guerrillas of networked progressive Britain—not just the fiery Momentum or the imploding People’s Vote campaign, but ordinary citizens speaking their minds, campaigners making their own iPhone videos and passionate canvassers who are the unsung heroes of this campaign. We are many, they are few. And it is to help these disparate citizens pinpoint their efforts that we have built not an old-fashioned Rolls-Royce, but a Tesla-grade data operation for this election.

Apart from the political parties and donors, independent political campaigns such as FckBoris, tactical.vote, Campaign Together and More United are set to mobilise thousands of volunteers and raise funds to run social media advertising campaigns in marginal seats. Targeting is everything.

As Election 2019 unfolds, we'll be helping campaigns to decide where to do their targeting, and many of them will straining every sinew to ensure voters can make an informed choice.

All our data suggests that more people are considering voting tactically than in any previous election, and that “squeeze” dynamics in which anti-Conservative voters unite behind the strongest challenger could be very significant in the final two weeks. Large numbers of voters remain undecided.

The battlegrounds are clear. Conservatives are pressing Labour hard in many of its heartland seats where majorities voted Leave, in particular across the Midlands, the North, Wales and coastal towns. Many of these seats are not being sufficiently defended yet, although there are signs this may change. South of the Watford Gap, the Liberal Democrats have the best shot at beating the Tories in 30-40 more seats, but will need Labour voters to carry them over the line. Labour are also on the attack, particularly in the south; they are the main challengers in as many as 30 more metropolitan, diverse or struggling constituencies, including Chingford and Woodford Green, Chipping Barnet, Hendon and Hastings & Rye.

On this newly volatile landscape, new swing groups are emerging. Our analysis has identified groups such as “Young Instagram Progressives,” “Corbynsceptic Swing Voters,” “Anti-Tory Heartlands” and “Older Establishment Liberals” whose decisions will determine the final outcome. There seems almost no chance of a majority for Jeremy Corbyn’s Labour; but this may paradoxically buoy the chance of opposition forces in the coming weeks, just as it did in 2017.

If the dynamics analysed in our “Seven Seats” report play out, it will obviously give progressives something to cheer in the early hours of Friday 13^th December. Even a Conservative government without these leading Brexiteers would be very different in its character. But if the polls tighten or we see unprecedented tactical voting well beyond these seven seats, then we cannot rule out a hung parliament and—before long—a second Brexit referendum. May you live in interesting times.

Paul Hilder a British political strategist and CEO of Datapraxis

Prospect will be holding two special events during the build-up to polling day: Can we trust polling? And how do you actually run a campaign? Take a look at our events page for further information.