Incoming: Super Thursday results thread
UK Parliament by-election at Blackpool South
Metro mayoralties across England
London Assembly
Police and Crime Commissioners across England and Wales
Local council elections in parts of England
UK Parliament by-election at Blackpool South
Metro mayoralties across England
London Assembly
Police and Crime Commissioners across England and Wales
Local council elections in parts of England
You're going to hear a lot of spin from partisans, a lot of (unintentional) narrative skew based on which contests report earlier vs later, so I'll try and cut through that here.
Also, as usual when it's not a Westminster election, THERE IS NO EXIT POLL!!!
Also, as usual when it's not a Westminster election, THERE IS NO EXIT POLL!!!
They aren't wasting any time! This year a lot of attention is on metro mayor elections – London because it's London, and some others that the Conservatives are talking up because they're bucking the national trend, possibly by enough to hold on to them
x.com
x.com
This is the first time we've had a decent amount of polling for these contests outside of London. One thing that's already very clear – there a some sizeable personal votes for incumbents. This makes them variable/interesting, but NOT useful as barometers of anything else...
One possible exception to tis is the East Midlands (actually just Notts and Derbyshire) as that is new, so has no incumbent. This is a helpful Westminster voting intention benchmark – expect incumbents to outperform (quite dramatically in some cases)
x.com
x.com
This is also the first Super Thursday with voter ID rules. Expect to hear arguments, particularly if any of the "big" races are close. In any case, there has already been one notable casualty...
politico.eu
politico.eu
The Westminster by-election in Blackpool South hasn't had much fanfare because there is so little doubt about the result. But the swing will be worth watching, as will the Reform performance. For reference, UKIP got 17.3% in 2015. Result in a few hours
On the council elections – as always – gains and losses are a bad metric, because it depends which seats are up and how well each party did last time. So watch the PNS (Curtice/Fisher) and NEV (Rallings/Thrasher). Here's an explanation from @StephenDFisher
electionsetc.com
electionsetc.com
Some more background here from Rob Ford
Side note: Feels very odd not to be watching David Dimbleby or Huw Edwards on a night like this...
.@ChrisMasonBBC making the point about how bad the Hartlepool by-election 3 years ago was for Labour (Starmer considered resigning). Here's a good illustration of this:
x.com
x.com
Con to Lab swing in (heavily Leave) Sunderland running at 11%, with Reform +17
Greens gain Newcastle Elswick (39% Muslim)
4% Con to Lab swing so far across Newcastle as a whole (6/27)
9% swing in Broxborne (all in)
4% Con to Lab swing so far across Newcastle as a whole (6/27)
9% swing in Broxborne (all in)
Here are the top 10 wards up tonight by % Muslim (all ages, excluding those in councils with boundary changes)
From these swings you can infer that the PNS has Labour 11 or 12 points ahead at this stage. BBC understandably not officially publishing it yet
Caveats interpreting PNS/NEV historically:
(1) The % of the vote going to parties other than Lab and Con is much higher than it used to be
(2) The realignment (esp. age polarisation) might change the local/national relationship
(3) The anti-governing party bias may have lessened
(1) The % of the vote going to parties other than Lab and Con is much higher than it used to be
(2) The realignment (esp. age polarisation) might change the local/national relationship
(3) The anti-governing party bias may have lessened
Very much as polls have been suggesting...
x.com
x.com
This is only based on 4 wards, beware the ecological fallacy, etc, but again this fits with the polling of British Muslims. The Labour vote is down, but it’s not imploding (bear in mind that the starting point in these areas would have been *extremely* high)
In any case, The Truth Is the Greens are having a very good night so far
John Curtice makes the point that Labour's performance isn't of the order of the mid-90s. Which is true, but note that my caveats below still apply. PNS margin still running at 11-12 points with over 200 of 811 wards counted
x.com
x.com
The Conservatives are in danger of getting their worst locals result EVER in terms of Projected National Share. Currently tracking close to the 25% they got in 1995 (their worst since the measure began). Too early to call, but something to watch...
Par for a by-election this sort of seat would be high 30s, so this is a little bit below that. Second place thought to be close between Tories and Reform
x.com
x.com
How much coffee has Jonathan Gullis had? Talking faster than I can keep up with
Another military area, another huge drop in the Tory vote...
For those of you watching seat totals, this is the Beeb's current projection of where the numbers will end up
Some of the PCC contests are seeing chunky swings. Possibly the low turnout in those contests is the swingier voters (who would have gone blue last time) staying home. But hard to say without polling...
Swing of over 26% in Blackpool South!
Now this *IS* mid '90s-grade swing. Bigger than Selby & Ainsty
x.com
x.com
Funny how this stuff works. Analytically it makes no difference, but symbolically, and in terms of narratives, it matters quite a bit…
They got 25% in 1995. The swings from the overnight declarations would put them on about 24%. Still relatively early with 3/4 of wards to count, but an all-time low is a very real possibility…
Pretty much my view at this point. Mayoralties will be interesting (but overspun)
x.com
x.com
Houchen holds Tees Valley against a 17% swing. Labour source seems (in odd language) to be conceding West Midlands to Street (though that declaration is tomorrow). So the Conservatives have their mayoralties (plus Harlow council) to spin...
x.com
x.com
Houchen's margin of just over 12 points is wider than the average of polls, although the final one by YouGov was within the margin of error (well done)
Labour still looks like being 11-12 points ahead on the BBC's measure for the council elections, with about half of the key wards declared
x.com
x.com
Recap of the Blackpool South result. There's an interesting divergence between the local election results, which aren't quite of the order of the mid-90s, and Westminster by-elections, which very much are
x.com
x.com
Should I...
Labour vote becoming dramatically more efficient (albeit from a very low base). Another way to look at this is that Labour still has a huge lead *despite* having lost support to its left
There’s been a steady stream of this kind of talk. Fwiw it’s not unlike last time, which turned out to be correct in the sense that things were closer than had been expected
🚨There's evidence of tactical swing from the ward level results, with Labour and Lib Dems each advancing more against the Conservatives in their respective areas of strength – great insight from @whatukthinks (who spotted the same thing in the run up to 1997...)
The biggest polling fail I can think of is they Heywood and Middleton by-election in 2014, where two polls had Labour 19 points ahead but it actually went to a recount. This would be similar in magnitude but far worse, because it’s much easier to poll London than one constituency
This smallish swing is fascinating because Leicestershire (inc Leicester) includes former coalfields, fox hunting country, and a very Asian city…
Conservatives’ joint worst result ever (tied with 1995)
Labour’s share and lead aren’t exactly earth shattering - quite a contrast from Westminster by-elections…
Either (1) Money follows the rumours or (2) R…edacted
Several people seem to think scrapping PNS/NEV is a good idea. It’s not. Whatever their limitations, the inevitable alternative would be the media and SW1 focussing on gains and losses, which has all of the same problems AND is hugely dependant on who did well/badly 4 years ago🤦🏽♂️
Rallings and Thrasher NEV
London turnout 40.5%, doen a couple of points. Doesn’t seem like much of a pattern, so if the polls are miles off, it probably isn’t differential turnout
This exercise is to project yesterday’s voting patterns onto every ward in GB, which is fine. As is aggregating them up to Westminster constituencies. But it’s NOT a general election nowcast or forecast. IMO the NEV should be the story, Westminster seat numbers are just confusing
IMO this observation by John Curtice is getting less attention than it should - it probably does tell us something about the general election (and specifically the appetite for Lib-Lab tactical voting)
Yes and yes. But it would have been more helpful for him if it had shown, say, a majority of 30 (ie believable given polls)
This is a really important question...
Next to their mates from the Don’t Know Party
Sky News charts, including this one showing the change in the Labour vote in areas with a high percentage of Muslim residents
news.sky.com
news.sky.com
Also, even if it were a perfectly unbiased sample, it would still be a sample, so would have a margin of error. If it's close, you wouldn't be able to tell
x.com
x.com
10 was too high, mid single digits makes sense to me. Ironically one thing bringing the bar down is Labour losing votes to its left - they’re overwhelmingly in seats that are already red (often very red), so losing them makes its vote more efficient
x.com
x.com
Now this is interesting. Multiple caveats needed, and there is quite a bit variation here (as you'd expect) but only in one of these wards is the swing close to what polls had been indicating. If these are representative, London really could be close...
x.com
x.com
(But before anyone gets too excited, on this average swing Khan would still narrowly win...)
In case you’re wondering how people think they know before the counting begins…
London is so slow to count that it would have been quicker just to send the ballot boxes up to Sunderland
The results we do have - both from Inner London - suggest a swing of about 5% *to* Khan.
What about Outer London? The question *at counting area level* is whether there’s a difference in swing, not a difference in changes in turnout. We already know the latter effect is small
What about Outer London? The question *at counting area level* is whether there’s a difference in swing, not a difference in changes in turnout. We already know the latter effect is small
Quote of the day: "Late swing towards the opinion polls dare I say!" – @keiranpedley
@keiranpedley Khan wins South West London, which was blue last time
✅Game over. Khan re-elected.
✅Game over. Khan re-elected.
Much like last time, London is a lot closer than several of polls had it...
x.com
x.com
In the West Midlands, it's looking very close, much as expected x.com
Brent and Harrow sees a small swing to the Conservatives. Note that these two boroughs have the second largest and largest Hindu communities in London respectively. If this is what's at play, it's not a new trend, but note that this the London election since Sunak became PM...
In the West Mids, BBC reporting bundle checks at most of the counts. Sounds it's close...
Ealing and Hillingdon, the counting area containing Uxbridge, sees a 0.7% swing from Con to Lab. ULEZ looks like a factor, but not a decisive one this time
So Parker (Lab) has probably won... 3k votes would be about *half a percentage point*
Agree with most of this thread. It's worth adding that the Tory London campaign has still massively outperformed the national polling, and by more than in 2021. Could/should they have outperformed it even more? Arguably so...
x.com
x.com
.@jessphillips has told the BBC that it wasn't Labour that asked for bundle checks, so that is that
x.com
x.com
We only did one London poll, late last year, and that was at the tighter end of range at the time, so I’m not totally sure what’s happened since/what others have done. But the ethnic minority vote is both (1) A big difference from the rest of the country (2) Often badly polled
Recounts make sense when the margin is about half a point - of course you want to be 100% certain. But that’s still equivalent to about 250 votes in a Westminster constituency. Has that kind of margin on a first count ever been overturned? Would be almost unheard of
This. Even though London will (on average) represent a polling miss, the polls will *still* have been more accurate than the briefings
Some polls had Khan winning by over 20 points and he actually won by 11, so it was indeed “much much closer than some polls had suggested”. Not sure how *clear* that was last night, but it’s what happened
0.3 percentage points! This is one the polls did get right (all of them had it as a tossup)
Things that make no sense: Having a recount in one council area in the regional election that's just a straight popular vote. Surely you'd do recounts everywhere?
It does however make it even less likely that the first count gets overturned, because finding 2-3k miscounted votes just in Coventry is a much higher bar than doing so across the whole of the WMCA
.@nickeardleybbc points out that the Con to Lab swing in the West Mids PCC election is 4.5%, exactly the swing needed to produce a dead heat for the mayoralty
The Greens came within 70 votes of pipping the Lib Dems for 3rd place in London. Strong Super Thursday all round in them
By my (sleep deprived) maths, Labour needs a swing of about 4.5% in Sandwell to get the 11456 margin they need to win overall – exactly the swing they need across the region as a whole!
Things you don't tweet unless you're confident...
x.com
x.com
Results have been very bad for the Tories across the board. But in both London and Tees Valley (two completely different regions) they outperformed polls (in the West Mids polls were right). So it's possible that the national picture is a few points less bad for them than polls
One puzzle when it comes to historical comparisons is that Labour's PNS/NEV share and lead aren't anywhere near 1996 levels, but the Westminster by-election swings ARE very much are very much around those levels. I have a few theories re the locals:
However I think – and I have more experience with this than most – that it's too big of a stretch to point exclusively to the PNS/NEV and conclude that Labour is coming up short. The data as a whole is still consistent a big majority (and probably a landslide)
Yep, that's close...
x.com
x.com
Interesting analysis of the Reform vote vs MRP models
Tories were hit hardest in wards with more mortgaged homeowners. Look at the intercept!
Rallings & Thrasher as usual have written up their NEV analysis for the S Times. It's a bit more caveated than earlier coverage of it but they maintain that "The local elections, on the other hand, paint a more nuanced picture" (than Blackpool S/landslide)
thetimes.co.uk
thetimes.co.uk
Further explanation from Michael Thrasher. He and Colin Rallings are always to be taken seriously when it comes to analysing local elections. But IMO there are also reasons to think the locals vs GE relationship may have changed in recent years
x.com
x.com
One additional factor that's relevant right now is that Reform stood so few candidates in the locals...
This👇
x.com
x.com
Wow – that's some serious tactical swing
In Con/Lab wards:
LAB +9
LD no change
In Con/LD wards:
LAB +2
LD +6
spectator.co.uk
In Con/Lab wards:
LAB +9
LD no change
In Con/LD wards:
LAB +2
LD +6
spectator.co.uk
This is obviously bad news for the Conservatives because it makes their vote less efficient. Just bear in mind:
(1) This comparison is with 2021, when there wouldn’t have been so much anti-Tory tactical voting, but for Westminster it will be with 2019 when there would have been
(1) This comparison is with 2021, when there wouldn’t have been so much anti-Tory tactical voting, but for Westminster it will be with 2019 when there would have been
(2) In many areas, tactical voting will need to “catch up” with boundary changes. This will particularly affect Con-LD contests
This is useful analysis. I’d add that in 2015 when the NEV *did* foreshadow a significant polling miss, it wasn’t *just* the NEV - Labour’s performance in Westminster by-elections was also notably weak for a main opposition party. Now, it’s basically 1995, and consistently so
John Curtice warns against reading across 1-for-1 from local elections to a GE: "The divergence....has become more marked". Important factors include local/national ticket splitting, few Reform candidates in the locals, and Labour's recovery in Scotland
bbc.co.uk
bbc.co.uk
Where Reform stood, the Conservative vote dropped 8 points more than elsewhere, so that on its own takes you from a 7-9 point gap to something in the mid-teens, which is around the narrower end of the range of polls
Really helpful comparison of polls and real votes from Andy. Because of the large combined sample, these are all (in aggregate) statistically significant. Margin wise, a few points tighter than polls
x.com
x.com
Here's a similar comparison for 2013-15. The average net error (Con-Lab) was 4.7 points (vs 3.3 now) and in the 2015 GE it was 6.6 points*... But that's only a third of Labour's poll lead
*Because polls were wronger among lower-engagement types who only vote in general elections
*Because polls were wronger among lower-engagement types who only vote in general elections
And that's not to say that polls actually are wrong by that amount. It's just that Labour is currently far enough ahead that it IF it were a 2015 or 1970 (or even a 1992) size error, Labour would still win
If you take 8 points of the Conservatives and give it to Reform, you get 41-24, which is within the range of the polls. (These figures are England and Wales only, but the Lab-Con gap in Scottish Westminster polls is about the same size)
x.com
x.com
جاري تحميل الاقتراحات...