Are the polls right?
The polls paint a poor picture for First Past The Post. The most popular party right now is on 25 per cent. The third most popular party is on 21 per cent. General election predictions on these numbers are more risk than reward. But one would be forgiven for also thinking: surely not. Surely these numbers are abstract air-war noise? After all, they are not normal!
And yet, dear reader, it appears - abnormal though these numbers most certainly are - they are very much real.
Britain Predicts is being wheeled out by yours truly to gauge how our 650 parliamentary constituencies would vote in a general election. And I’ve developed its function to transpose those numbers onto local council wards - and compare them to by-elections.
The five council wards up last night as by-elections had two expected by my model (imperfect in execution, let’s not forget, and most certainly not water-tight in by-elections) to go Reform in a general election. Two were also expected to be held by Labour.
If the polls are as crazy as they look, then that is what you would expect.
So let’s compare.
What happened in Tendring (Clacton) went as you’d expect - a clean win for Reform as they take a Tory seat in Nigel Farage’s own backyard. (But note the under and over-performances of the Tory and Lib Dem candidates!)
What happened in Gillingham went as you’d expect. Labour hold on against a split right-wing vote.
What happened in Wokingham also went as you (or rather, the model) would expect: an easy Lib Dem hold in the face of Tory impotence.
But what happened in Rochester proper and Hyndburn, Lancashire, confounded the model - and by extension the polls.
Reform exceeded expectations in Rochester, topping the poll in a ward my model had Labour on course to win by six.
And in Baxenden ward, Hyndburn - a seat I’d figured for Reform, it was the Tories, not Reform, who romped home, with the defending Labour candidate relegated to a disappointing third place.
Compare and contrast the model numbers with the council by-election scores and - in aggregate - Britain Predicts did well enough. It got Reform close enough in Tendring, Gillingham, and Hyndburn, but understated them by 8pts in Rochester.
For the Tories, the clear miss for my model was Hyndburn - a seat, like Newcastle under Lyme, where activists aren’t deserting the locale in their droves for Reform. Andrew noticed this too in his by-election preview, and rightly called out my numbers for being bearish on the Greens. It appears Green appeal in Accrington is more confined to disaffected Muslim voters than anything else.
For Labour though, the model was bang on, across the board.
Which in and of itself is interesting. Council by-elections are low turnout hiccups of public opinion. They can be over-analysed, over-interpreted, and in closing this write-up, I have to wonder if I’m doing just that.
And yet, they are mirroring, relatively well, a nationwide forecast model built on three parties vying for a quarter of the vote.
Whatever the permanence of the current numbers, be in no doubt: the polls - they are very much real.
And now for something existential
Is Labour a working class party? It depends on how you look at it. Historically - tradtionally - speaking, not so much as it used to be. Labour’s support among council and social households is a world away from where it was in 1997. Accounting for the fact Labour did 8pts worse nationally in 2024 when compared to 97, that fact still has currency.
Labour doing 21pts worse among a demographic that - yes, takes up a smaller share of the voting pie than it used to - it has long been reliant on, feels a long time coming. And I wrote in further detail about that over at the New Statesman. What won it for Labour in ‘24 was not its traditional base, but its new one: the mortgage owners, the “steady as she goes” voters for whom the Toryism of Liz Truss in 2022 was an irrevocable awakening.
Thanks for reading
Have a great weekend. I’ve been Ben Walker. Be sure to subscribe and share. It’s a new world out there.