Recent Posts

Polling cross-breaks show the Tories doing especially badly in their 'heartlands'. For example, YouGov show the biggest Tory vote-losses and Labour making greatest progress with ABC1 voters and in the South of England (below). 

That does not mean that the LDs won't benefit from the collapse in the Tories' vote, too. They took many more seats in 1997 than in 1992 on a reduced vote share, and I would expect a similar effect if the Tory vote falls sharply, as expected. 25-30 seats looks likely to me. However, they are not doing particularly well in what were their strongest demographic groups at the last election. 

YouGov Cross-breaks (last 6 polls) - comparisons to YG analysis of GE2019

South England 
Lab 40% (+17)
Con 28% (-27)
LD 13% (-4)

ABC1 Voters
Lab 49% (+16)
Con 21% (-22)
LD 11% (-3)

yougov.co.uk/politics/articles/26925-how-britain-voted-2019-general-election

(For a bit of added context - In GE1997, the Conservatives were still around 8 points ahead of Labour in the South of England in what is currently their previous worst result. )

So, further evidence that the 2016 Leave voters who now support Labour are untypical of Leave voters as a whole: exactly the point Sir John Curtice made in a YouTube clip I posted last week. 

The Conservatives still have a fair-sized lead with Leave voters - 43% to 27% per a running average of YouGov polls - but this needs to be seen in the context of the 74/14 lead they had at GE2019. So that's a 22% swing. But this is fuelled largely by taking disillusioned 2016Leave supporters rather than those who still believe in it. 

Re Reform UK

It has been a consistent feature of by-elections and local elections in the past year that RefUK always underperform relative to their poll rating. They took just 3.4% in Somerton and Froom, 3.7% in both Mid Beds, 3.7% in Selby and 5.3% in Tamworth - all well below their prevailing polling level of around 7%. In the English local elections, they took an average of 6% in the very small number of places where they stood - which must have been where they thought they had their best chances. 

Because of this, I'd suggest that their true level of support is probably about 4%. This is also the figure that Survation (who have a very good track record) have been showing for them all of this year; it's lower then others because they simply ask 'which party would you vote for?', without any prompting. I'm inclined to believe Survation's figures, which typically show Con 29%, Ref 4%, as opposed to the current average of all polls, which is Con 26%, Ref 8%.

(Survation also show lower Labour leads than others, but this is due to the higher Con figure that their non-prompting produces; their Labour figures are a little bit higher than the average, too.)

"Bandwagon rolls, Tory MPs defect in droves to Reform.

Reform wins next GE, Farage PM."

That sounds far-fetched to me, as it would depend on Reform expanding its appeal beyond those who are pro-Brexit - or perhaps in Brexit itself somehow regaining popularity.

Taking the past 5 YouGov's, Reform are taking 18% of Con 2019, but just 1% of Lab2019 and 1% of LD2019. From the 2016 Referendum itself, Reform are taking 20% of Leave voters, 1% of Remain voters, and a very low share (1%?) of those who did not vote.

Another stat from YouGov which I think is relevant here: only around 33% now believe that Leaving the EU was 'right'. It is no coincidence that this matches closely to the combined Con+Ref VI, because voting intentions are now aligned more closely than ever with views on Brexit. I would expect the Tories to sweep up nearly all of that 33% pro-Brexit support at the General Election, but if it goes to Reform UK instead the result would be the same.

Interesting take from John Curtice's on "Labour voters and attitudes to Brexit"

www.youtube.com/watch?v=keR5Zfk9Eto

"...the structuring of the Labour Party vote in particular, with reference to attitudes towards Brexit is almost exactly the same now as it was 4 years ago...the Labour Party, relatively speaking has gained ground among those people who voted Leave in 2016...however, a chunk of those people who voted Leave in 2016...and voted Conservative in 2019 no longer believe in Brexit...so the Leave voters that labour have got are not typical Leave voters.."

As you may have read here 5 weeks ago (see bottom of this page):

ukpollingreport2.proboards.com/user/48/recent

In broad terms, around 70% of all those who voted Leave in 2016 still think their decision was 'Right' (with around 20% 'Wrong' and 10% DK). Within the current Lab VI as a whole, just 11% of their support thinks Brexit was 'Right', despite around 22% of it comprising 2016 Leave voters. This means that at most 50% of 2023Lab/2016Leave voters think their 2016 decision was 'Right' (and it could be a bit lower). It also means that among Leavers-who-don't-intend-to-vote-Labour, 75% still think their decision was right.



So Labour's partial recovery among 2016 Leave voters involves taking a high proportion of those who have changed their mind on Brexit. And because of this, Labour's recovery with 2016 Leave voters does not mean that the political landscape is not still shaped by (current) attitudes to Brexit.

Thank you for highlighting this - particularly as Survation rank as one of the most reliable pollsters. Their policy of not prompting by party names explains why they get realistic-looking figures for Ref and Green.

This was actually an MRP poll, and showed Labour getting 426 seats, and the Tories only around 150. And they seem to have done another recently, with very similar results, although I havn't seen these publicised.

www.facebook.com/survation/?locale=en_GB

Survation
10 November at 09:25

'We are over one year into Rishi Sunak’s premiership and approximately one year out from the general election. Throughout September and October, Survation conducted a series of large sample polls on behalf of Greenpeace, 38 Degrees, and Lodestone Communications. We explored the results using multilevel regression and poststratification analysis (explainer here) to produce constituency-level estimates for voting intentions. The average Labour lead over the Conservatives was 17 points....'

Lab 425
Con 147
LD 22
SNP 32

My recollection Of GE2017 on the old board is the opposite to yours.

It was the most unexpected outcome of any General Election in living memory, and provoked a huge number of comments, both the during the campaign and in the aftermath. And from the point of view of polling, we had a huge range of data, from ICM still putting the Conservatives about 12 points ahead to Survation's 1% Con-lead, and their correct prediction of Theresa May needing to do a deal with the DUP.  On UKPR1, it was the election of Dr Mibbles, of YouGov's first MRP, and of the aparently outlandish results that showed. Many were still expecting May to get close to a 100-seat majority, right up to the exit poll.

Sadly, the comments section of UKPR1 is no more but there is still plenty of good, unbiased analysis of GE2017 available.

yougov.co.uk/politics/articles/18384-how-britain-voted-2017-general-election

www.ipsos.com/en-uk/how-britain-voted-2017-election

ukandeu.ac.uk/the-2017-general-election-not-that-close-after-all/

OK.

I may have misunderstood you, but to clarify: if you had been able to Vote Labour in a battleground constituency in 2017, can you now confirm that you would have done so?


I think that true answer is that you would not have done so, nor have you supported any campaign to make your own vote relevant (such as a change to the electoral system) but happy to be corrected.  [Edit - per the below, I think this is correct]

The fact that 'Stats for Lefties' puts such trust in The Times speaks for itself. 



Meanwhile, here's a reminder of their reporting on 'polling'' of British Muslims.
Which do we believe, Savanta's recent poll with Labour on 64%, or an open access poll with them on 5%?



Stats for Lefties 🇵🇸🏳️‍⚧️

@leftiestats

🚨 BREAKING: Amongst Muslims, Labour plunges to 5% (-66pts).



🟥 2019 ~ 71%

🟥 2023 ~ 5%



Labour have lost 93% of their Muslim voters.



Via Muslim Census, 17-26 October

Re the SNP's position in polls post-Rutherglen by Election.

We have only had 3 Scotland Westminster polls after 5 Oct - two ties and a YouGov with a 6-point Labour lead. The latter looks on the high side, given that YG's running average in cross-breaks for Scotland since early October is a narrow SNP lead (SNP36, Lab 34, Con 13).

Putting these 3 polls, plus YG's running average together gives an overall average of

SNP 34% (-11)
Lab 35% (+16)
Con 18% (-7)
(comparison to GE2019)

This compares to an average SNP lead of around 4 points in the 10 polls before Rutherglen.

I would note that the figures above almost certainly overstates the Tories' level of support, due to the 23% that R&W show for them. I would trust YG better for this at around 15%. Even at 18%, they appear to have lost a greater proportion of their 2019 vote than the SNP has. This is likely to be significant in Con/SNP contests.

As for the SNP/Lab battleground: it is very difficult to predict or project seat numbers. Any small movement would move a significant proportion of seats in either direction. A few things to note:

1. The SNP are likely to be able to squeeze a point or two from SGP.
2. More than half of Labour's gain in vote share has come from Con and LD rather than SNP.
3. Movements between Con/Lab/LDs in Scotland are fairly similar for those for GB as a whole. To me, it seems reasonable to except a bit of Con recovery from Lab in Scotland just as in England. The big difference is that Con are not losing so many supporters to Ref in Scotland, but they appear to be losing slightly more to Lab.

en.wikipedia.org/wiki/Opinion_polling_for_the_next_United_Kingdom_general_election#Scotland

Right now, I'd guess seat totals of something like SNP 30, Lab 20, LD 4, Con 3 for the next GE, but those SNP and Lab figures could move up/down by 15 seats in either direction.

The Green VI has risen, but only fairly slowly. It's averaging 7% in the 12 most recent polls, compared to 6% for the equivalent 6 months ago (the start of May) and about 5% from early Nov 2022.

As for where it's coming from - YouGov has them taking 2.5% of Con2019, 7% of Lab2019, and 6% of LD2019 - per the average of their last 6 polls.So Labour are losing a bit over 2 points from their 2019 vote to Green, but more than compensating by taking votes from Con and LD.

The figures above can be compared to their equivalents from Oct/Nov 2022 when the Greens were taking 1.5% of Con2019, 5% of Lab2019 and 4% of LD2019. So there has not really been any change in the pattern of where the Greens are picking up votes. 

I would have thought it would be fairly obvious where my political sympathies lie, but for the record I have voted Labour at all General Elections and European Elections from 1987 onwards, with the exception of 2019 when I voted tactically for the LDs (Con 46, LD 42, Lab 12 in my constituency) and I of course voted Remain in 2016. I also voted for AV in 2011. My political campaigning has been minimal.

I was right about you, wasn't I? 

These figures are not based on the 2019 General Election but on current regional polling data. 

You are right that the apparent target for Labour moved considerably from GE2017 (after which it was a 7% lead) to 12% post GE2019. In that respect, these figures suggest that the political landscape might essentially be moving back to around where it was in 2017. This is what John Curtice argued in May, basing his findings on the 2023 Local Elections (although others think it will move further)

www.whatukthinks.org/eu/2023/05/12/is-brexit-over-lessons-from-the-electoral-geography-of-the-local-elections/

Back in the post-1997, 2001 and 2005 periods, it appeared that Labour could achieve a majority with parity (or even slightly behind) in the popular vote, as their vote was more efficiently spread than the Conservatives'. And as recently as post-2010, Labour needed just a 2-3% lead, whereas the Tories then needed to be just over 7% ahead to get a majority.  It seems to be almost universally accepted now (barring a few BBC commentators) that the apparent target is going to move considerably at the next GE, just as it has done for both Con and Lab at previous General Elections. A shift of 6-7 points, for either the Con or Lab target, has several recent precedents.

[ To Add : ... this touches on the same issue I covered last month regarding a couple of MRPs from Survation and Savanta, both of which showed the SNP on 30+ seats, so Labour maybe 15-ish seats in Scotland, but where the rate of overall Lab-gains implied that a lead of just over 4% was needed for a majority. It was on page 71 of the last thread (14 Oct) if anyone is interested ]

The fluctuation in a 'Rejoin/Stay Out' poll of polls, but is now largely dependent on which pollster has reported most recently. There has been some significant movement towards 'rejoin' earlier this year, which may now have slowed. However, there is now a large difference in 'house effects' on this question. Most companies' figures are now relatively steady, giving us these averages by pollster over the past 3 months or so:

BMG 53/47
Deltapoll 56/44
Omnesis/WeThink 59/41 (but a bit variable)
YouGov 60/40

www.whatukthinks.org/eu/questions/should-the-united-kingdom-join-the-european-union-or-stay-out-of-the-european-union/?removed=removed&pollster%5B%5D=yougov


TO ADD: Clearly, some pollsters must be getting this wrong: it's a question of which one(s) it is. My instinct is to trust YouGov, particularly since they have a very long-established panel so should have the best data on their respondents' 2016 vote. Any 'false recall' of how people voted in 2016 would skew the headline results. I would also note that the Online pollsters (such as YouGov) actually did not do so badly in 2016, as their polls were typically 50/50. It was the phone polls which predicted a Remain win - in some cases by around 10 points. 

Another point on YouGov is that their Brexit-tracker (the 'Hindsight' question) shows only very slow movement: it is nothing like as volatile as Voting Intention. Because of this, I think it is likely that most movements from one poll to another on EU membership are just 'noise'. 

[Another addendum - a 59/41 lead for Rejoin per R&W - this seems close to their 'norm of around 60/40 ]
redfieldandwiltonstrategies.com/joining-or-staying-out-of-the-eu-referendum-voting-intention-21-october-2023/