Are the Tories really facing their worst general election result since 1997?

Lowest share of the vote since the 19th century and a wipeout of seats in line with New Labour’s rise... is this the beginning of the end of the Conservatives as Britain’s dominant political party, asks Sean O’Grady

There are still imponderables... eg what if Rishi Sunak is ousted between now and the general election? (AP)

A fresh, in-depth and fairly innovative poll conducted by YouGov suggests the market research company’s new MRP model finds Keir Starmer would win a 120-seat majority, were the general election to be held tomorrow. The Tories would be left with barely more seats than they had in the New Labour landslide in 1997, and the lowest ever vote share for the Conservatives in a British general election since the time of the Great Reform Act of 1832 and the dawn of modern democratic politics.

Could it be that bad?

Yes. The previous nadirs for a Tory electoral performance (in share of vote) were recorded by John Major (1997, 30.1 per cent), Ted Heath (October 1974, 35.8 per cent), Winston Churchill (1945, 35.2 per cent) and you have to go all the way back to Robert Peel in 1834 and the Duke of Wellington in 1832, both on 29.2 per cent, to see how badly Rishi Sunak's prospective 26 per cent will look in a long historical perceptive.

What does the poll say?

It’s indeed mostly bad news for the government. If you were an optimistic Conservative you’d point to the number of “don’t knows”, the Labour poll rating registering below 40 per cent, the (relatively) low lead of 13.5 per cent, and the fact that Reform UK’s popularity is in single figures and that they won’t have any seats and will be thus unable to establish a bridgehead inside parliament.

Labour, on 39.5 per cent, would in fact be slightly down on Jeremy Corbyn’s score in 2017 (40 per cent) – not such a positive endorsement. The Greens seem to getting an exceptional boost in this poll, at 7.5 per cent some way ahead of their recent polling and by-election showings, probably on the back of Starmer’s attitude to the war in Gaza. The collapse in SNP support is also a welcome windfall for Labour.

On the other hand... the poll implies a swing of 13 per cent to Labour and against the Conservatives, which would exceed that achieved by Tony Blair in 1997 or Clement Attlee in 1945. The Tories will lose all of their red-wall seats, large swathes of the blue wall, and overall see their representation in the Commons more than halved. The prospective Labour majority, potentially, could propel them to a second term and a hold on power well into the 2030s. The comparatively healthy showing for the Liberal Democrats and Reform UK, plus a large cohort of abstentions, suggests a perfect storm for the Tories – a lethal multi-pronged attack on the 2019 baseline set by Boris Johnson.

What are the figures?

In terms of seats in Great Britain (ie excluding the 18 MPs from Northern Ireland):

Labour: 385 seats (+183 from 2019)

Conservatives: 169 (-196)

Liberal Democrats: 48 (+37)

SNP: 25 (-23)

Plaid Cymru: 3 (-1)

Green: 1 (no change)

Reform UK: 0 (no change).

The projected national votes would be:

Lab: 39.5 per cent

Con: 26 per cent

Lib Dems: 12.5 per cent

Reform: 9 per cent

Green: 7.5 per cent

SNP: 3 per cent

Plaid: 0.5 per cent

Others: 2 per cent.

What’s going on?

It’s fairly straightforward. As Patrick English, YouGov’s director of political analytics, explains it succinctly: “Labour are winning voters off Conservatives; Conservatives are also losing voters to Reform and (to a lesser extent) Lib Dems; Labour themselves are losing voters to Greens (particularly in their stronger 2019 seats).” Indeed, the Greens are on the cusp of winning a second seat in parliament, Bristol Central, which they would take from Labour frontbencher Thangam Debbonaire.

What does it mean for the Tory leadership?

Surprisingly little, aside from the shock effect. It just so happens that most of the principal contenders and players are all in very safe seats, and would survive the cull – Kemi Badenoch, Suella Braverman and James Cleverly would probably be among the walking wounded. Michael Gove, Jacob Rees-Mogg, Liz Truss and Rishi Sunak would survive, but Sunak might not hang around. David Cameron, of course, doesn’t have to worry about elections any more.

The new parliamentary party, though much smaller, probably wouldn’t have shifted very far from its present balance and ideological composition – proportionately the left and right would be in roughly the same relative strength as now. However, the grassroots would want a greater share in policy and the choice of leader, and the party would also be faced with difficult questions about a potential realignment of the right involving Nigel Farage and Richard Tice’s Reform UK.

But election night would yield another record, the largest number of felled cabinet and senior figures in modern British history – including Jeremy Hunt (chancellor), Penny Mordaunt (Commons leader), David TC Davies (Welsh secretary), Victoria Prentis (attorney general), Alex Chalk (justice secretary), Johnny Mercer (veterans minister) and Lee Anderson (deputy chair). The post-mortem would be as disputed as the one after 1997, and a lurch to the right, again as happened after 1997, looks inevitable. A civil war will ensue.

Why is this poll so special?

It’s got a much bigger sample and it’s more concerned with what happens in swing seats in the coming general election, rather than conducting a mere opinion poll and then, fairly crudely, converting that into a Commons seat projection under the vagaries of the first past the post system.

What’s an MRP poll anyway?

Expensive, because the sample size is so big (14,000 here, more than 10 times the usual one), and because of the work that goes into the modelling that turns the raw data into projections for marginal parliamentary seats. Thus it is much more focused on the political battleground and tries to get a better hold on the prospective voting trends of particular types of voters in particular types of seats than the usual polls do. Thus, say, take the sub-sample of female professionals in their late twenties or early thirties, married with two kids, who voted Remain in 2016 and then Tory in 2019, living in a classic Con-Lib Dem home counties marginal. This can only be reasonably discerned using much larger samples and extra data.

The demographic modelling is based on standard multilevel regression with poststratification (MRP) techniques often used in the analysis of election results, combining, for example, census data, economic figures (eg interest rates/percentage of renters in a seat) and deeper breakdowns of electoral results – psephology, as it’s called.

YouGov says that the technique has been “benchmarked” to correctly estimate the 2019 general election to within a couple of seats of each party’s actual performance in that election, and that a similar approach successfully predicted elections as recently as Spain in July 2023. As a “common sense” check, it seems YouGov’s work is in line with current conventional polls, as well as actual local election and by-election results.

Could it be wrong?

Of course, in at least three respects. The MRP might be inherently faulty, based on dodgy base data or flawed assumptions, for example on prospective turnout, tactical voting or all those factors. Also, like all polls, it can only be a statement of current opinion, and things can change. What if interest rates came down dramatically? What if Nigel Farage returned to politics full-time? What if Sunak was ousted? Or Ed Davey quit? Or plane loads of refugees start taking off for Rwanda? There may still be about nine or 10 months until the next election – and the electorate has become more volatile in recent decades. All to play for.