Audit

How did I do in my assessment of the general election? Time to take stock.

What I got right

For months I’d said that I was suspicious of the polls. And there was indeed a serious polling miss. On 3 July 2024, the Sky News polling average was:

Lab 41

Con 23

Reform 15

Lib Dems 11

Green 7

The actual UK-wide result in vote share was:

Lab 34

Con 24

Reform 14

Lib Dems 12

Greens 7

Every single pollster had overestimated Labour’s vote share, some by a big margin. Many had seriously underestimated the Conservative vote share. Some serious soul-searching needs to be undertaken by the polling industry.

There’s a suggestion, and it does seem to have something in it, that Labour suffered a late swing away from it at the very end. That does not necessarily let pollsters off the hook. Several purported to have measures in place to deal with how the undecideds might break. They were among the very worst-performing pollsters of all.

Nor am I particularly impressed by the MRPs. Some of them came quite close to the actual result and have been doing victory laps. But since they did so off implied vote shares that were substantially different from those actually seen and far greater tactical voting than they had imagined, they seem simply to have benefited from two major errors coincidentally cancelling each other out. That seems to me that they have two problems rather than one.

To be clear, I have huge respect for pollsters. Their job is incredibly difficult. But they did not do collectively well this time and I fully expect them, with the professionalism that we see from them, to interrogate themselves starkly as to why.

I got the general vibe of the election right a long time ago. As long ago as December 2021 I wrote:

“No two elections are ever alike, but dynamics of the next election may well have some resemblances to the dynamics of 1997. At the previous election, the Conservatives successfully made the election about Labour’s (and specifically its leader’s) inadequacies, frightening centre ground voters into voting for them. In 1997, the election was about Conservative inadequacies, leading the centre ground voters to vote against them.”

Last year, in the wake of the local elections where some had noted that Labour had got an uninspiring vote share and were even speculating that it pointed to a hung Parliament, I wrote this:

“honestly, this stuff isn’t very complicated. In a first past the post system, if you get 26% of the vote, you lose. If you get 26% of the vote and everyone else despises you, you get obliterated.”

The Conservatives got 24% last week. They got beyond obliterated.

I got Reform more or less right. I correctly did not expect them to overtake the Tories or to hoover up dozens of seats. In fact, my pre-election expectation of them taking five seats was spot on. Still, they got a higher vote share than I had been expecting even a couple of months ago.

What I got wrong

If I got those two general points right, I completely misunderstood what the consequences of those points would be. In particular, I substantially overestimated how much of the anti-Conservative vote would devolve onto Labour.

My expectation was that Labour would hoover up these votes all over the place except where it was stark staring obvious to vote Lib Dem or Green. In fact, unlike 1997, voters and the political parties were far more canny. Without any formal agreement in place, Labour and the Lib Dems appear to have set up a silent conspiracy as to which seats they would target in any given area. And voters appear to have got the hint. In 1997, the Lib Dems rose from 18 to 46 seats. In 2024, on a far lower vote share, they rose from 8 seats to 72.

(This silent conspiracy looks set to continue and intensify. Labour and the Lib Dems are fishing in different pools. The Lib Dems’ first target seat for 2029 held by Labour is Labour’s 85th most vulnerable seat. Labour’s first target seat held by the Lib Dems is Labour’s 73rd target. They can continue to play nicely. The Conservatives should be very worried about a couple of dozen seats which would fall if progressive voters fall in more fully behind the main challenger. Seats like The Wrekin and North Cotswolds look very vulnerable indeed. There is no particular reason why the Conservatives can be confident that they have reached their nadir.)

There has been much talk about the rise of Reform. There hasn’t been anywhere near enough talk about the Lib Dems’ successes, which are potentially far more consequential. They have successfully established themselves as the main party for affluent areas. It’s not at all obvious how the Conservatives are going to dislodge them now.

I had underestimated how solid the Green vote would be. I expected that when push came to a shove, the Greens would be squeezed in marginals. That did not happen in the way that I anticipated. For example, in Bromley & Biggin Hill the Conservative margin of victory was 302. The Greens tallied 2,583 and saved their deposit. I wonder how many of those Green voters are comfortable with that result.

I certainly didn’t expect them to win two rural seats. They appear to have been beneficiaries of the same anti-Conservative sentiment that sorted itself so tactically between Labour and the Lib Dems. However, they will be unable to participate again in a silent conspiracy. Every single seat they finished second in is Labour held. There’s a left-wing battle brewing for next time.

I’d also seriously underestimated the strength of left-of-Labour feeling generally. I’d thought Jeremy Corbyn would lose his seat, and he took it comfortably. The slew of independents who won or nearly won seats was a jolt to me. Those attributing this only to Gaza are, I think, mistaken. Equally, when push comes to a shove in a tighter election I suspect that such voters might well prefer what they see as a flawed Labour government to a Conservative government. Something to watch.

Summary

I did well in how I saw things shaping up. I was one of the first to see that Labour were going to win and were going to win big. I also did well to see that the polling could not be fully relied upon.

But I didn’t do particularly well with the detail. Something to work on. Something to think about. Good. I like challenges.