United Kingdom General Election: July 4, 2024

[YL]

If I run* should I do it as an independent or for Reform? I think independents have more chance of a victory than attaching a minor party label to your ballot line. Plus if you run for a party they might tell you that you have to run in some absolute hole in Sunderland or Southampton rather than in your local seat. However, Reform did ask me to apply for the local elections (I didn't). They're not a serious outfit though. The other issue is that I know my local MPs and don't really want to stand against them. Plus my local Labour candidate is kinda fit.

*Quite unlikely I actually go through with it but I've always wanted to do it.

Harry, have you been selected for anywhere yet? You should definitely run again.

Reform UK apparently already have candidates in both Filton & Bradley Stoke and Thornbury & Yate.  Bristol Central is still available

(Lists of selected candidates, maintained by Owain Sutton of Vote UK)

[YL]

https://twitter.com/YouGov/status/1744656534208909536

As the graph helpfully included there shows, the vote share changes are within the normal range of fluctuations in YouGov polls, so I wouldn't read much into those, but that should hardly be much comfort for the Government.

If you look at the trend lines on the plot on Wikipedia's article on the polls, then it's fair to say that for some time Labour have been in the low to mid 40s, the Tories in the mid to high 20s (more like mid at the moment) and the Lib Dems around 10% or so, with not much change.  There has been a definite recent increase in the reported Reform shares, and while I am sceptical that the polls are reporting their share correctly for the reasons given by Eadmund, I suspect they genuinely have picked up a bit of support in the last couple of months.  (How Wellingborough fits the pattern will be interesting to see, but they've said they're not standing in Kingswood.)

[YL]

The main players in 1992 were I think more charming than those of 2015. And both major parties have more reasons to not want to dwell on 2015. I'd say the highly inaccurate exit polls also help in terms of making it particularly iconic.

Tbf the exit poll wasn't actually that bad - it did get the general idea of Conservatives going to form government, SNP landslide in Scotland and Labour doing worse than expected. By far the opinion polls were a lot worse than the exit poll during 2015.

I think the inaccurate exit poll referred to here is the 1992 one.

[YL]

Anyway, regarding 1992, here is Wikipedia's plot of the polls before that election:

(from here).

So, yes, Labour developed a big lead in the reported polls by early 1990, roughly comparable to the lead now (though with both parties higher). But it was fairly short-lived and it evaporated completely in November 1990, when there was a rather significant change in the UK political scene. After that the Tories actually took the lead for a few months; Labour did re-take the lead in the run-up to the election, but they were never more than a handful of points ahead. The coming election is unlikely to be "just like 1992 again" because in its analogue to November 1990 the Tories chose Liz Truss as their new leader.

Of course, there was also the polling error, but a polling error of comparable size on current polls wouldn't be keeping Sunak in Downing Street.

There was of course a UK election where a party with a substantial lead and expectations of a landslide blew it in the campaign and ended up in a hung Parliament, but the Tories aren't so keen on talking about that one for some reason.  (It was mentioned in a recent Guardian article about Labour trying to dampen expectations, though.)

[YL]

George Osborne claims he has inside info the election is being planned for November 14th.

https://www.theguardian.com/politics/2024/jan/11/general-election-likely-to-take-place-on-14-november-says-george-osborne

Not sure why it's being so dragged out, if this is true. The result will essentially be a done deal at this point. Whatever keeps the Tories in power for whatever time they have left, I guess.

It's what governments who think they are going to lose tend to do. I suppose it's a combination of trying to enjoy the trappings of power for as long as possible and hoping that something will turn up.

[YL]

Note that the sample size here is around 14,000. This is rather lower than the sample sizes YouGov used for their big MRPs in 2017 and 2019, and more like the sample size used for other pollsters' MRPs. While it does actually seem to me to have fewer oddities than many MRPs I've seen, at that sample size it's not going to be able to actually pick up constituency level effects.

In particular, its sample size in areas where Plaid are a genuine factor is likely to be very small indeed, so I would take its figures for them with a massive pinch of salt. For similar reasons, I would be a bit sceptical that the Greens are really on the verge of taking the lead in Bristol Central, though I wouldn't be astonished if that turned out to be right.

[YL]

The "official" Rallings and Thrasher notionals have been published this morning. There's a spreadsheet here with full details by constituency and the Guardian has a map where you can find individual constituencies.

For those familiar with elections in countries which make full detail of results available rather than just publishing a constituency level aggregate, remember the disclaimer that these are based on modelling of how individual wards would have voted using local election results, not actual polling district or ward level results from 2019. In one case, Wolverhampton South East, we actually know that the "official notional" is wrong, because those two councils did make ward level results available and these show that it was narrowly Conservative.

[YL]

YouGov have some polling in various demographic groups; see the Twitter thread starting below for details.

https://twitter.com/YouGov/status/1752281504728707076

- the usual ridiculous age divide
- something of an education divide, though note that this seems to be mostly among older voters
- weak gender gap
- little effect of "class" (though of course they mean the dubious ABCDE system)

Caveat: I don't know how well weighted these samples are. Caveat 2: as usual I suspect Reform UK are over-polled (but note how like Tories their supporters look).

[YL]

How much worse than 1997 can the result be expected to be for the Tories in terms of seat count, given depolarization since then? Is something like "Liberal Democrats become the Official Opposition" a pipe dream or a realistic prospect?

I lean towards "pipe dream", unfortunately. There are enough constituencies that I really find it very hard to see the Tories losing, even in their current state, that I struggle to see them falling low enough to levels where the Lib Dems might overtake them. I suppose if Reform UK really did split their vote heavily enough then Labour or the Lib Dems might come through the middle and win those seats on a low vote share, but I don't really think that's likely.

(Also, there's a lot of skepticism over the 'Reform UK' surge being real. Is there a time when we'll know whether it is or isn't?

The main reason many of us are sceptical about Reform UK's polling figures is that they aren't performing at a level consistent with them in real elections, especially Westminster by-elections, and we haven't seen a convincing explanation for why not. We have three more of those coming up, and one of those in particular (Wellingborough) looks like an opportunity for them to show that actually they can perform. If that doesn't happen, and their polling figures don't fall back again as the General Election approaches, then the first we'll know of whether they're real or not is the exit poll (to the extent that it tells us about that) and then the first results.

Also, are most 'Reform UK' voters individuals who despise the Tories but can't actually envision themselves voting for a party not on the right, or are they against-all voters who might well end up casting a vote for Labour or the Lib Dems to turf the Sunak government?)

A bit of both, I think, but I did see some polling saying that Labour specifically were very unpopular with people expressing a Reform UK voting intention, which suggests more the former. (But the Tories are unpopular with them too.)

[YL]

Get your pinches of salt ready…

Constituency polls, carried out by Survation for Plaid Cymru

Ynys Môn

Plaid 39
Lab 27
Con 26
Reform 4
Lib Dem 1

Caerfyrddin

Plaid 30
Con 24
Lab 24
Ind Edwards 10
Lib Dem 4
Reform 4

[YL]

Get your pinches of salt ready…

Constituency polls, carried out by Survation for Plaid Cymru

Ynys Môn

Plaid 39
Lab 27
Con 26
Reform 4
Lib Dem 1

Caerfyrddin

Plaid 30
Con 24
Lab 24
Ind Edwards 10
Lib Dem 4
Reform 4

How does that compare with the results in those seats in the 2019 GE?

Ynys Môn was Con 35%, Lab 30%, Plaid 29%, Brexit Party 6%.

Caerfyrddin's notional result is Con 39%, Plaid 31%, Lab 25%, Brexit Party 4%, Lib Dem 1%. This may be a little misleading as the two outgoing constituencies it takes territory from have different dynamics.

[YL]

These polls have a sample size of like, 240 voters, right? Feels pretty flimsy, although it'll make for a nice bar chart on the candidate leaflet.

What's the Welsh for "only the Lib Dems Plaid can win here"?

Tables are here for Caerfyrddin and here for Ynys Môn. The sample sizes quoted on the front pages are 520 and 507 respectively, which are not I think unusual for constituency polls; the lower figures being reported are what's left after don't knows, won't votes and so on are removed.

The Ynys Môn one has breakdowns according to how long voters have lived in the constituency and Welsh language ability. Sample sizes are of course very small for many of these categories.

Both polls prompted for certain named candidates: Plaid, Labour and Tory in Ynys Môn and Plaid, Labour and Ind Edwards for Caerfyrddin. ("Ind Edwards" is Jonathan Edwards, the sitting MP for Carmarthen East & Dinefwr, elected as Plaid.) Owain Sutton's spreadsheet of candidate selections shows some other names in both cases.

[YL]

Some of Survation's constituency polls at the last GE weren't bad at all, and they called the Hartlepool byelection right as well (though they wrongly had the Tories winning Batley and Spen, which was IIRC their most recent individual seat effort up to now)

But their samples then, whilst not large, were somewhat bigger than these two.

They also polled Wakefield (not great: they overestimated both the top two, especially Labour) and Mid Beds (not bad).

[YL]

What's the shortest time span legally possible to hold a GE? Can it be scheduled within a month?

It's 25 working days from dissolution of Parliament to polling day, and normally the dissolution would be announced a few days in advance. So just over a month.

[YL]

Some of Survation's constituency polls at the last GE weren't bad at all, and they called the Hartlepool byelection right as well (though they wrongly had the Tories winning Batley and Spen, which was IIRC their most recent individual seat effort up to now)

But their samples then, whilst not large, were somewhat bigger than these two.

They also polled Wakefield (not great: they overestimated both the top two, especially Labour) and Mid Beds (not bad).

Ah yes, forgot it was them who did Mid Beds as well as Opinium.

The orchestrated LibDem rubbishing of those polls really was something to behold

You mean the idea that a reasonably reputable polling company would conduct a poll of Flitwick and call it a poll of a whole constituency? I never got to the bottom of where that one came from.

[YL]

And today’s prize for bad reporting of polls goes to those who have taken a Survation poll of British Muslims and used its figures with undecided voters included for a comparison with the last General Election.

[YL]

Pinch of salt time again...

We have another MRP, by FindOutNowUK and Electoral Calculus for the Mirror (link)

Lab 452 (inc. Speaker, it seems)
Con 80
Lib Dem 53
SNP 40
Plaid 4
Green 2
Ashfield Independents 1

Curiously, in spite of showing such a massive landslide it has Labour losing two seats: Bristol Central to the Greens and Sheffield Hallam to the Lib Dems. Both Labour and the Lib Dems hold all their by-election gains, at least if you take the main successor seat in each case. (Tiverton & Minehead it has as Labour!) In Scotland it has a complete Tory wipeout, but in Wales it has them holding Brecon, Radnor & Cwm Tawe.

[YL]

Remember though, and it is very important that people from countries where better data exists are aware of this, all demographic data on GB elections is pretty sketchy and a lot of it is actively bad.

I know that and while I agree you should take data like this with a huuuge grain of salt it´s still pretty interesting.

It's interesting, but unfortunately having played with it a bit I don't find it credible. Region seems to change too much, as does the "urban"/"suburban" distinction. (Which one is an area two miles from the centre of a city full of semi detached houses?) And it comes up with Reform UK on 12% for my demographics, which I find very hard to believe.

[YL]

Usual caveats apply, but we have a constituency poll of Godalming & Ash (Jeremy Hunt's new seat) by Survation for 38 Degrees:

Lib Dem 35% (+1)
Con 29% (-24)
Lab 23% (+15)
Reform UK 8% (+8)
Green 3% (+1)
(changes from 2019 notional)

I imagine bar charts are being made out of this already.

[YL]

So who do Gen Zers in the UK who are from extremely wealthy aristocratic families who went to places like Eton and hang out in Sloan Square all vote for these days?

People like that will be fairly Tory regardless of their age group, but outside of certain parts of central London there really aren't very many of them.

I would imagine that your typical young Tory voter is someone from a Tory family who follows the family voting habit and doesn't see enough of a reason to change. There are surely also a few libertarian ideologues out there, or are they Reform or trying to infiltrate the Lib Dems again?

[YL]

It strikes me how little headway the Lib Dems have been making, still polling below the 2019 result. For the Tories to fall sub-100 seats you’d surely need to see the Lib Dems breaking 16%.

The thing is that they will probably do badly nearly everywhere they are not targeting, so the national share may be misleading and what's really important is how they are doing in their target seats. If that Godalming & Ash poll we got is both accurate and typical, then the Tory slump will be enough on its own in places like that, and with Labour clearly third (and who would think Labour could win there anyway?) it shouldn't be too hard to squeeze the Labour vote a bit to seal the deal.

The main effect of their current weakness in national vote share is that in some more optimistic targets (Maidenhead is an example) models are currently showing them behind Labour, and while they will still no doubt draw bar charts based on the last General Election result it may be harder to get the squeeze going. And that may lead to some Tory holds on a low vote share, or if the Tory performance is catastrophic enough to some very weird Labour gains. (Electoral Calculus has Labour winning Chichester!) I think that if the Tories really are as low as 20% then FPTP won't be kind to them.

[YL]

Labour entered the 1997 campaign on 55% of the vote - in what universe?

There were actually quite a few polls putting them on about that in March and early April 1997; the highest was 58% from Harris on 17 March. It should be noted, however, that the Tories' vote share didn't increase very much and the one they got in the election was if anything slightly below what the polls had been giving them in January and February and indeed in much of the campaign.

(I'm not convinced this is very relevant, anyway. Opinion polling has changed a lot.)

[YL]

Sunak has just said that there will not be a General Election on 2 May.

[YL]

There are some interesting Reform UK candidates out there. Here's a thread on one of them:

https://twitter.com/snigskitchen/status/1770542877250105823

[YL]

I know MRP polls are supposed to be viewed in the macro, rather than the granular, but there’s some *weird* stuff in the Survation projection:

Labour winning the Tory-Lib Dem marginals of Hazel Grove, as well as Woking for one. They’re also winning Frome and East Somerset, the prototypical Lib Dem tactical voting seat.

The difficulty is that on a national and basically all demographic level, the Labour vote is up a lot while the Lib Dem vote is static. There is no scientifically rigorous way to model which constituencies will see tactical voting for the Lib Dems and which will see a credible Lib Dem vote fall away in Labour’s favour. They could stick some arbitrary tactical voting on the numbers eg; Labour vote will rise a lot less where the Lib Dems are 2nd, but it would clearly still be arbitrary and wrong in many places as well. There’s similarly a long-standing issue with MRPs (some more than others), where the vote for parties is flattened and it’s particularly striking for the Labour vote in Tory-Lib Dem marginals. A lot of this is simply a reflection of the underlying Labour vote (both natural Labour voters and people who are defaulting to them nationally as the non-Tory option) which will be squeezed by the Lib Dems come election day.

The Electoral Calculus MRP a few weeks back was a bit more plausible in that respect, but even it had Labour ahead in Chichester, for example. I’d quite like to see one of those YouGov ones with a proper MRP sample size, but they’ve only done them during the actual campaign.