|
|
Post by James E on Aug 22, 2023 17:58:22 GMT
A 25 point Labour lead with Deltapoll. It's their equal-largest lead so far this year, and their first 'Lab 50' since March. That's 8 points higher than the last Delta, but they have a habit of moving in the opposite direction to other pollsters. Lab 50% Con 25% LibDem 9% Green 7% Ref 4% SNP 3% UKIP 1% oth 1% Fieldwork 17-21 Aug. deltapoll.co.uk/wp-content/uploads/2023/08/Deltapoll-230821_trackers.pdf10-poll average of all polls now around 19%. |
|
|
|
Post by James E on Aug 22, 2023 17:33:45 GMT
Redfield & Wilton Strategies Labour leads by 25% in the Red Wall. Red Wall VI (6 August): Labour 53% (+4) Conservative 28% (–) Reform UK 7% (-1) Liberal Democrat 6% (-2) Green 4% (–) Plaid Cymru 1% (-1) Other 1% (–) Changes +/- 6 August That's a 17% swing on GE2019, so a couple of points more than the current average of 14-15, but these R&W 'Red Wall' polls have often produced 4-5 point movements, which have proved mostly to be just 'noise'. The norm seems to be that they vary around the prevailing Con to Lab swing, perhaps in recent months a little above, whereas previously a little below. en.wikipedia.org/wiki/Opinion_polling_for_the_next_United_Kingdom_general_election#Other_polling(scroll down a bit on the wiki link for their Red wall polls, fortnightly since June 2022) |
|
|
|
Post by James E on Aug 22, 2023 10:20:56 GMT
As a public service, Redwood and Wilton.... Labour 42% Conservative 27% Liberal Democrat 13% Reform UK 7% Green 6% Scottish National Party 3% Other 1% .... Remarkably similar figures are to be found in the latest YouGov (18 Aug): Lab 43% Con 28% LibDem 14% Ref 8% Green 7% other 1% So everyone one point higher, except the SNP, but that's to be expected as this is the South of England cross-break (n=733). The Tories had a 32-point lead (55/23) here at GE2019. And while this is on the high side, the norm with YouGov is still around a 20% swing in the South, compared to around 6% in London. The evidence remains overwhelming for something very different to a Uniform Swing. docs.cdn.yougov.com/kby5f0bevb/TheTimes_VI_Results_230818_W.pdfAlso, another record-equalling 36/64 lead for 'Wrong to Leave' in the Brexit Hindsight tracker - but that seems to be the 'new normal'. The recent average is 37/63. |
|
|
|
Post by James E on Aug 21, 2023 22:18:19 GMT
oldnat I'm using post-election analysis as my source for gender differences: YouGov for recent elections, and Ipsos Mori for those up to 2010. I guess that some people might not know how they voted the week before, but none are recorded. In the surveys I've looked at, the Con and Lab figures are generally around their overall vote share, with Lab a little above their share with women and Con a bit above theirs with men. But there is a definite pattern that other parties of the right (now Refuk, previously UKIP) have a very noticably male voter base.
|
|
|
|
Post by James E on Aug 21, 2023 18:33:30 GMT
Re the gender gap in British politics. A bit of checking suggests to me that the present pattern of women being more supportive of Labour has been around for close to 20 years. But it appears to have gone unnoticed. The last time that Labour did better with men than women was at GE2001, when it was 42/32 with men and 42/33 with women - per Ipsos Mori analysis, first link below. In 2005, Ipsos recorded a 34/34 tie with men and a Labour lead of 38/32 with women. So a 6-point gap there. This appears to have been the norm ever since. In 2010, the Tories led by 38/28 with men and 36/31 with women - a 5 point gap. 2015 was a slightly odd one, as the figures are distorted by UKIP faring much better with men than women (by 15% to 12%). As a result, both Con and Lab did better with women than men. Men voted by 37/29 for the Tories and women 38/33, so a 3-point difference - or 6 points on a 'Lab v Con+UKIP' comparison. Labour has fared 6 points better than the Conservatives with women than men per youGov's analysis of both GE2017 and GE2019. So the recent YouGov polling showing a 7-point gap should not be any great surprise. www.ipsos.com/en-uk/how-britain-voted-2001www.ipsos.com/en-uk/how-britain-voted-2005www.ipsos.com/en-uk/how-britain-voted-2010(further links - YG analysis of GEs 2015-2019 - in my post on this subject yesterday, page 40) |
|
|
|
Post by James E on Aug 21, 2023 18:01:56 GMT
leftieliberal "Even ignoring Lord Ashcroft's MRP polling, both YouGov (see my original comment) and Focaldata (fieldwork 27/11-10/12) were well out in 2019; although I have only seen the central prediction for Focaldata it was practically the same as YouGov's central prediction. YouGov sampled over 105,000 and Focaldata over 21,000 in 2019. If over 100,000 is not enough to produce reliable seat predictions, then it brings into question the point of MRP. Would it be better to do samples of 1000 in 100 constituencies (or 2000 in 50 constituencies)?" YouGov's polling for their 2019 MRP was not too bad, but still somewhat inaccurate. In vote shares, they had Con 2.1% low at 42.6% and Lab 0.8% high at 33.8% (they also had LDs 0.7% high). Because of this, they were out by a 1.45% swing, and the Tories won 21 seats with a less than 2.9% margin (so within such a swing). if you adjust for this relatively small polling error using the actual results, it turns a further 21 seats Conservatives, so that YouGov's central prediction of 339 seats for the Tories would become 360 seats. d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/uk714vkjy3/MRP_Tables_2019_Election_Public_Release%20(5).pdf So it was this polling error which was the main cause of the inaccuracy in YG's central estimate. Another way to look at this is that if they had got the vote shares absloutely right, YouGov would probably still have underestimated the Tories by 5 seats, and the SNP by 7 seats, while overestimating Labour by about 8, and the LDs by 4.
|
|
|
|
Post by James E on Aug 21, 2023 15:52:30 GMT
Mark Pack on MRP and whether swing in the GE is likely to be uniform or proportional. theweekinpolls.substack.com/p/mrp-what-it-is-and-why-it-may-or Mark Pack makes a good point that there were two MRP polls published just before the 2017 General Election. One, for Lord Ashcroft, was wildly wrong; the other by YouGov for The Times was very close to the actual result. Because confirmed predictions get more publicity more than wrong predictions, the story in the media was about how good the YouGov MRP model was (whereas they may have just been lucky in their choice of parameters). In their 2019 GE MRP poll (4-10/12/2019) YouGov got overall vote shares pretty close to the actual result, but their seat totals were well out (Con 311-367, Lab 206-256 both 95% confidence ranges) The actual result was Con 365, Lab 202 so just around the 95% confidence range limits. He also mentions Owen Winter's study of a particular MRP poll owenwinter.co.uk/2021/01/03/whats-up-with-the-sunday-times-mrp-projection/Again a MRP poll which produced some odd results when drilled down to constituency level, which Owen thinks may have been due to overfitting. The problem with MRP polls is that the methodology used to create them, while transparent at the highest level is quite opaque at the detailed level as Justin Ibbett says in response on Twitter Yep this is likely to be the case and is a known problem with MRP (over smoothing and regularisation, being too pulled to the last result). 2019 individual vote will drive the model (for obvious reasons). We can’t adjust those parameters values as they’re learned from the data
So now we are approaching another GE and we can expect more MRP polls which, by their nature, will be also affected by the 2019 vote. We need to be aware of this and treat MRP polls with caution. Thanks for that, LL. It is interesting to read Pack's views, but a little disappointing that he does not offer a conclusion on the question as to whether uniform Swing or Proportional Swing will prove more accurate. He does make a couple of mistakes, though. The 2017 Lord Ashcroft MRP was released on 12 May 2017, so all fieldwork must have predated that. The polling average around then was an 18% Conservative lead, and indeed a YouGov with fieldwork of 11-12 May also showed the Tories 18% ahead. Given that YG's final polls averaged around a 4% Consevative lead, it appears that there was a swing of around 7% to Labour between the Ashcroft MRP and election day. So it cannot be judged as 'wrong' any more than other polls from around that time, and for Pack to cite this as a failure of MRP shows a lack of attention to detail. lordashcroftpolls.com/2017/05/election-2017-ashcroft-model/en.wikipedia.org/wiki/Opinion_polling_for_the_2017_United_Kingdom_general_electionI would also dispute his (and Kellner's) claim that UNS has been accurate in previous GB General Elections. What we have seen in recent elections has been huge local variance in swings at seat level, as voting patterns have changed. I shared on the 'Will labour get an overall majority' thread the huge movements in the targets for a lab majority from 2005 to 2019, but I think these are worth repeating. Post-General Election implied target-leads for a Labour majority: 2005 -1% 2010 2% 2015 9% 2017 7% 2019 12% So this target has shifted hugely at recent elections - whereas if our elections were governed by UNS this figure would stay constant. And it is very likely to shift again at the next GE - probably to a much lower figure, as the 'brexit effect' wares off. The only historic election we have had with a really large swing (1997) saw a 'semi-proportionate' outcome, with the Conservatives faring noticably worse where they were defending. By my calculations, they lost caround 40-50 more seats in that election than UNS predicted, and this involved losing around 3% more of their vote share in their safe seats than UNS implied. He's right to say that we should treat at least some MRPs with caution, and in particular those with small samples. Less than 10,000 is not enough, and even with 50,000 (as per the YouGov pre-election MRPs) some very localised trends cannot be picked up. And there have been some truly awful MRPs in the past year-or-two: I recall one from Focaldata which showed disastrous results for the SNP and Plaid, as they had failed to calculate churn separately for Wales or Scotland. And there have been others which appear not to involve looking at the different dynamics of LibDem seats (or targets). |
|
|
|
Post by James E on Aug 21, 2023 10:26:42 GMT
@lululemonmustdobetter I keep an eye on the YouGov 'lilelyhood to vote' tables, and these support the idea of a low turnout. Overall, around 53% of respondents say that they are "certain to vote", compared to 58% in Summer 2018 - which is the nearest I can get to the middle of the 2017-19 Parliament*. So on that basis, we might expect turnout to be about 5 points down on 2019. Across age and social class cross-breaks, the fall in self-reported likelyhood to vote is fairly even. However, there are wide gaps (10-15%) between 2016 Remainers and Leavers, and also between 2019 Con and 2019 Lab (57% to 73% per the most recent YG). Comparing to summer 2018, Remainers say they are just as 'certain' now as then to vote (c70%), while Leavers are around 8-10% less 'certain to vote'. docs.cdn.yougov.com/nqg0mhiemd/TheTimes_VI_NetZero_230803_W.pdf*Self-reported likelyhood to vote rises when an election is imminent, so it's not yet appropriate to compare current polls to those near to GE2019. |
|
|
|
Post by James E on Aug 20, 2023 19:01:22 GMT
We will eventually rejoin for exactly the same reason we did the first time - the British economy desperately needs the EU markets. The advantage over 1/1/1973 may be that next time we will fully understand what we are joining. Not sure how long this will take - I suspect a lengthy period of growing alignment followed by it happening very quickly as the current generation of politicians disappear off the stage with their baggage. We will of course need to sign up to everything. Opinion has shifted steadily over the past two years, to the point that we now have a clear 60/40 majority (on average) in favour of rejoining. www.whatukthinks.org/eu/opinion-polls/poll-of-polls-uk-eu/There has not been a 60/40 majority either way for a decade, and back in 2013 it was the Leave side which had such a lead. And the Rejoin side is clearly winning the battle of hearts and minds among the electorate of the future. Using YouGov's most recent poll, while over 65s divide 36/64 for 'Stay out' (so exactly as they voted in the 2016 Referendum, those under 65 are 72/28 in favour of 'Rejoin'. That's a 20% swing since 2016, when they voted 52/28 for Remain. docs.cdn.yougov.com/7myo59zisk/Internal_Brexit_230809_W.pdfyougov.co.uk/topics/politics/articles-reports/2016/06/27/how-britain-voted |
|
|
|
Post by James E on Aug 19, 2023 21:14:49 GMT
"Labour has opened up a huge lead over the Tories among women voters, many of whom are turning to Keir Starmer’s party because they feel financially insecure after 13 years of Conservative rule, a new report says today. The analysis by Labour Together, a thinktank whose report is based on extensive polling and academic analysis, finds that the financial worries of huge numbers of women voters – particularly those aged under 50 – are persuading many to back Labour and reject the Conservatives. According to a poll of more than 5,000 people by YouGov, Labour now has a 28-point lead over the Tories among women, compared with a 21-point advantage among men.
Labour’s landslide victories under Tony Blair saw Labour win many more votes than the Tories among both women and men. However, in 1997, Labour was still winning more of its votes from men. Following the 2005 and 2010 elections, where the percentage of men and women voting Labour nearly equalised, David Cameron reopened the gender gap in 2015, with women representing 55% of all Tory voters. " Thanks for that, Steve, interesting article. But it's also true that the current gender gap with Labour faring better among women has now been a feature of British politics for several years. I noted on this site back in December 2021 a series of YouGov polls with "Labour faring better with women than men by 12, 8, 9, 12 and 9 points". ukpollingreport2.proboards.com/thread/14/dec-2021-lab-con-ldem?page=81YouGov's GE2019 analysis showed a 6-point gap between the Conservatives' lead with men (46/31) compared to women (44/35). yougov.co.uk/topics/politics/articles-reports/2019/12/17/how-britain-voted-2019-general-electionThere also was a 6-point gap in 2017 (45/39 with men and 43/43 with women). yougov.co.uk/topics/politics/articles-reports/2017/06/13/how-britain-voted-2017-general-electionAnd back in 2015, Labour under Ed Milliband fared 3 points better among women (33/38) than men (29/37). This rises to a 6-point gap on a 'left-right' axis, if you add in the fact that UKIP fared 3 points better among men and women. That fact alone exlains why the Tories actually got more votes from women 9though their lead with women was 3 % lower). yougov.co.uk/topics/politics/articles-reports/2015/06/08/general-election-2015-how-britain-really-votedAnd going even further back - I can't find any YouGov analysis of GE2010, but this from Ipsos mori shows the Conservatives with a 10% lead with men and a 5% lead with Women. www.ipsos.com/en-uk/how-britain-voted-2010The gender gap in 2005 was 6 points per Ipsos Morie (men 34/34, women 38/32), and you have not go back to 2001 for a UK election where the reverse applied ( men 42/32 Lab, Women 42/33) So the current pattern of women in the UK voting more for Loc parties has probably been around for 20 years . |
|
|
|
Post by James E on Aug 16, 2023 16:25:02 GMT
It has had Welsh captains such as Tony Lewis. It had a Scottish captain in Mike Denness, (born in Bellshill, North Lanarkshire)! It had a South African captain in Tony Greig (who qualified as he had a Scottish father) |
|
|
|
Post by James E on Aug 16, 2023 14:45:16 GMT
|
|
|
|
Post by James E on Aug 16, 2023 13:29:57 GMT
There is a something slightly odd about someone who doesn’t vote, continually analysing the previous Labour leader’s ratings etc etc (etc…) Hardly, it’s a polling site. And the 2017 election never got so much attention at the time. It’s not the only polling I look at, the other day it was the EU ref. It’s just that some seem to find discussion of Corbyn a bit vexing. Following our exchange on pre-referendum polling in 2016, I found what I think is the most authoritative source from that time, in the article linked below by John Curtice in late May 2016. www.whatukthinks.org/eu/wp-content/uploads/2016/05/Analysis-paper-6-The-Divergence-Between-Phone-and-Internet-Polls.pdfThe aim was to identify the most probable causes of the phone/internet poll divergence in EU Ref polling. At one point, this was as much as 10% in the overall average results. And the answer appears to have been that phone polls were finding too many graduate respondents, and that it was not understood at the time what a vital determining factor this was - indeed, polling companies did not then seek or weight by education level at all. But for our purposes, the important part is the comparative figures over the period from late summer 2015 to May 2016. And in this, the Internet polls showed no movement at all, averaging 50/50 in each of the three periods used by Prof Curtice. It was largely because of research such as this that those doing phone polls made changes to their methods. Ipsos Mori reported that their changed phone poll methodology moved their headline figures by 3 to 4 points from Remain to Leave. So while it is true that an average of all polls did put Remain ahead (av 54/46 in Summer 2015), this was entirely due to phone polls, and these moved significantly as a result of methodological changes. The campaigning and adverts on buses may have helped the Leave side to hold thier vote, but it looks most likely that they did not increase it. Internet Polls 1/9/15 to 18/2/16 Remain 50% Leave 50% Internet Polls 19/2/16 to 31/3/16 Remain 50% Leave 50% Internet polls 1/4/16 to 19/5/16 Remain 50% Leave 50% And both YouGov and Opinium averaged 50/50 in the period from 20 May to the referendum itself. So there was 'Polling Error' in that those internet polls understated Leave by 2 points, but like-for-like 'Polling movement' of zero. |
|
|
|
Post by James E on Aug 15, 2023 9:44:38 GMT
c-a-r-f-r-e-w"Regarding where May’s votes went, not sure, but I think James E might have said something about that previously?" This is what I said previously about GE2017 campaign movements (from first polls of the campaign to election day). The Tories lost about 3 points over the 2017 campaign, from 46.5% to 43.5%. May called the election on 18 April. These are the comparative figures for the 7 polls on 18-20 April 2017 compared to the actual result. Con 43.5% (down 3 from 46.5) Lab 41% ( up 15 from 26%) - But final pre-election polls averaged 36.5% for Lab and overstated UKIP LD 7.6% (down 3.4 from 11%) UKIP 1.9% (down by 6.4 from 8.3% av) Green 1.7 (down 1.3 from 3%) The Tories probably lost more than 3 points to Lab in the campaign, as their 3-point fall in VI coincided with a spectacular fall in the UKIP VI - from 8.3% in the early polls to a 1.9% share on election day. It isn't credible to suggest that three-quarters of those who had been backing UKIP just a few weeks earlier switched to Labour during the campaign. YouGov's final poll of the campaign showed just 18% of the UKIP2015 vote switching to Lab and 44% of it going to the Conservatives. en.wikipedia.org/wiki/Opinion_polling_for_the_2017_United_Kingdom_general_election [However, if Lab did win support from Con during the campaign as this suggests, then it may well have been a case of winning back deserters from 2015. The overall share of Con2015 who went to Lab in 2017 was put at around 9%, with a similar percentage going the other way] yougov.co.uk/topics/politics/articles-reports/2017/06/22/how-did-2015-voters-cast-their-ballot-2017-general
|
|
|
|
Post by James E on Aug 15, 2023 9:34:44 GMT
c-a-r-f-r-e-w " Some people might have quite rationally deduced that on balance, something like impact on GDP is for them outweighed by other things. " Indeed the damaging impact on gdp was outweighed for me by our freedoms and franchise being stolen by the brexitanians. Sure, that’s a valid issue to consider. I’m not keen on losing free movement myself. But many in the middle-class are relatively protected from some of the economic effects of EU and globalism more generally. (E.g. my partner did not have a queue of East Europeans competing with her when she got her headship). In the nearly 20 years I have served as a School Governor, I have not met a Head Teacher from Eastern Europe, nor even come across an applicant in cases where I have served on a HT selection panel, before and after we left the EU. (That's not to say that they wouldn't be welcome, but it just doesn't happen, in my experience). And we've touched before on the fact that Leave voters overwhelmingly did not expect negative economic consequences of Leaving. This is from YouGov 20 June 2016, with a sample of 1,643 leave voters. "Do you think Britain would be economically better or worse off if we left the European Union, or would it make no difference?" Worse Off 4%Better Off 48% No difference 38% Don't Know 10% [And, for the question "Do you think you personally would be financially better or worse off if Britain left the European Union, or would it make no difference?" - just 3% of those voting Leave answered "Worse Off", compared to 57% of those voting Remain. ] |
|
|
|
Post by James E on Aug 14, 2023 17:46:03 GMT
c-a-r-f-r-e-w Re your long post to me above (14:41) I won't quote this back, so as to try to keep this readable. To answer your points: 1. Clearly the 'lies on the bus' helped the Leave side to win. But that it not the same as claiming that Leave increased their support during the referendum. The detailed evidence I have linked above shows clearly that this was not the case. I have cited Jan 2016 as the start of the campaign, but this is not 'cherry picking'. The formal campaign, and the 'lies on the bus' came later, so if we use that, any gains by leave become less, even by the standards of the graph you have linked. 2. The advantage of looking at one pollster is a true like-for like comparison. As internet pollsters, YG and Opinium did not make the methodological change which phone pollsters needed to apply during May 2016. This is very helpful in establishing whether there was really any polling movement (as opposed to methodological changes producing different outputs). 3. Re Polling error: yes, there was a significant overall error, although the pollsters I have quoted (YouGov and Opinium) were fairly accurate. But your point was about polling movement during the campaign - which is an entirely different thing. See also the paper linked below by John Curtice in May 2016 re Internet and Phone polls. [On page 3 of this there is some very interesting data which divides the period from sept 2015 to May 2016 into 3 periods. In each of these, Internet polls show on average a dead heat, at 50/50. It was only the phone polls which moved towards leave, and these underwent heavy methodological change in the campaign, when it was found that educational level was a strong determining factor in Voting, and phone polls included too many graduates.] 4. While you quote events such as Partygate and Truss's short leadership, these don't appear to have had any noticable effect on EU polling (as opposed to General Election VI). There is slow,steady movement towards Rejoin, and in fact, Truss's time in Downing Street (and the Con recovery under Sunak) have had no impact on this. And there is no need to keep quoting my name (as you do 4 times). It should be clear from the quotes who your comments were addrressed to. www.whatukthinks.org/eu/wp-content/uploads/2016/05/Analysis-paper-6-The-Divergence-Between-Phone-and-Internet-Polls.pdf
|
|
|
|
Post by James E on Aug 14, 2023 12:34:21 GMT
That 62.5 / 37.5 lead for 'Rejoin' is a new record for YouGov; their past three polls on rejoining were 2 on 58/42, and a 60/40 - all within the past 3 months. docs.cdn.yougov.com/7myo59zisk/Internal_Brexit_230809_W.pdfThis poll will also tip the average of recent Rejoin/Stay Out polls over 60/40 (link below). As per my long post this morning, there has not been a consistent (or average) 60/40 lead in EU membership polling since 2012, and then it was the 'Leave' side which had a 20+ point advantage. www.whatukthinks.org/eu/opinion-polls/poll-of-polls-uk-eu/As ever, the age divide on EU Membership is stark. It's 34/66 for 'Stay Out' among the over 65s, and 72/28 for 'Rejoin' for those aged 18 to 64. Over the course of the next 10 years, these age demographics alone could be expected to add a further 6 points to the 'Rejoin' side. |
|
|
|
Post by James E on Aug 14, 2023 10:58:26 GMT
Well that may be the situation, but Remain was some way ahead at some point before the ref. After a round of campaigning and buses etc., things could change. (Though as the margin increases it maybe becomes less likely they could overhaul remain. I do wonder what would happen if the US joined CPTPP, and maybe the EU did a deal with CPTPP, but these are future hypotheticals)  A narrative seems to have been developed among Leavers that the campaign swung significant numbers of voters towards 'Leave'. I'm not convinced by this, although the graph of all polls above does show some movement from the start of 2016 to the final polls. They go from a 4% remain lead to a tie. However, if you look at YouGov's polling from Sept 2015 to 22 June 2016 on 'Should the UK Remain a Member of the European Union or Leave the European Union', there appears to be no movement at all. The first 5 polls and the final 5 polls both average as effectively a tie. www.whatukthinks.org/eu/questions/should-the-united-kingdom-remain-a-member-of-the-eu-or-leave-the-eu/?removed=removed&pollster%5B%5D=yougovSo perhaps it would be more acurate to say that the Leave campaign succeeded in holding support for leaving the EU rather than increasing it. And it is rather to selective to focus on the 'peak Remain' polls in mid 2015. These averaged at around a 54/46 Remain lead, but even that is partly caused by one huge outlier ( 66% Rem, 22% Leave from Ipsos 14-16/6/15). If you look further back - or just at the start of 2013 per the graph above, it was Leave which had a lead. It was around 55/45 during 2013, and averaged 61/39 back in 2012. The same polling from YouGov only shows a clear 'Remain' lead for a short period from Feb to July 2015. www.whatukthinks.org/eu/questions/if-there-was-a-referendum-on-britains-membership-of-the-eu-how-would-you-vote-2/?removedLooking at these figures, and the known inaccuracy of most pre-referendum polling, it seems to me that as a best appproximation the two sides in the referendum started the campaign on level-pegging.
And of course not all pollsters got it wrong. Opinium were almost spot-on in their final poll, so it worth also checking how their figures moved (or failed to move) during the campaign. www.whatukthinks.org/eu/questions/should-the-united-kingdom-remain-a-member-of-the-eu-or-leave-the-eu/?pollster%5B0%5D=opinium&removedSo YouGov and Opinium, both Internet pollsters with panels, showed effectively zero movement. So why did the overall polling average shift in the campaign? Probably the main cause was a methodological change for those pollsters doing telephone polls - of which there were many in 2015-16. Some might recall the large difference between phone and internet polls observed early in the campaign. This resulted in methodological changes, which made the polls much less favourable to 'Remain' quite late in the campaign. Ipsos, who I have mentioned above, stated that this resulted in a 3 point Remain to Leave movemennt in their headline figures. |
|
|
|
Post by James E on Aug 13, 2023 10:40:10 GMT
Omnisis's figures are a little bit on the high side for 'Rejoin' compared to other pollsters, but the average of recent polls is 59% Rejoin, 41% Stay out. What always strikes me about polling on EU Membership is how consistent and slow-moving the figures are (after allowing for house-effects and random variation). It's nothing like as changeable as Voting Intention, for example. This has not always been the case, but clearly views have become very entrenched post -2016. YouGov's figures show a move of about 5 points in each of the past two years, from 49/51 for Stay Out in July 2021, to 55/45 for Rejoin in July 2022, and most recently 60/40 for Rejoin. 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=yougovIncidentally, it's worth noing the consistent difference between those who think that Brexit was the wrong choice and those who want to Rejoin. It's not the same question, and difference has been a consistent 3-4% over the 30+ times that YG have asked about 'Rejoin' rather than 'Right or Wrong in Hindsight'. |
|
|
|
Post by James E on Aug 11, 2023 21:07:24 GMT
Westminster Voting Intention: LAB: 48% (+1) CON: 24% (-1) LDM: 10% (-1) GRN: 6% (+1) RFM: 6% (-1) SNP: 3% (=) Via @omnisis , 10-11 Aug. Changes w/ 3-4 Aug. That's a smashing poll, it really is . It's pretty good, but for lovers of two-to-one ratios it would be even better with LDs on 12%. |
|
|
|
Post by James E on Aug 11, 2023 19:02:13 GMT
Westminster Voting Intention: LAB: 48% (+1) CON: 24% (-1) LDM: 10% (-1) GRN: 6% (+1) RFM: 6% (-1) SNP: 3% (=) Via @omnisis , 10-11 Aug. Changes w/ 3-4 Aug. Been waiting for the poll that showed a 2 to 1 advantage for Labour. This is the first in August, Crofty, but there were 3 such polls in July -from Deltapoll, Omnesis and YouGov. In fact we've had at least one "Lab = at least 2 x Con" poll for in every month of the past twelve (the exception being April - due to the 'Windsor framework bouce'). |
|
|
|
Post by James E on Aug 10, 2023 20:56:40 GMT
Anent the earlier discussion about the Greens winning Bristol West. A fortnight today there is a council by-election in Bishopston and Ashley Down ward, historically a Labour seat within Bristol West, it might give us all a few clues. I will keep my ear to the ground. Under the new boundaries, most wards in Bristol West are becoming part of the Bristol Central constituency, which Electoral Calculus currently predicts as a Green gain, by around a 7% majority. Bishopston and Ashley Down ward is predicted by EC to vote Green, but will in fact move to the redrawn Bristol North West constituency (which stays Labour despite this). www.electoralcalculus.co.uk/fcgi-bin/calcwork23.py?seat=Bristol+North+West |
|
|
|
Post by James E on Aug 9, 2023 21:44:57 GMT
Electoral Calculus giving Greens Bristol West. That's according to Stats for lefties (full disclosure- now a Green member) who is saying it is this one although I can't find individual seat projections, but EC does give a second Green seat as their central prediction and and this is the Greens top target. www.electoralcalculus.co.uk/prediction_main.htmlNot sure what this is based on (perhaps local results in the mix) and personally I'd only be expecting a much reduced Lab majority with Greens looking to take it at the following General Election. Greens were a quite close 2nd in 2015 but then went backwards because of the Corbyn effect and a massive Lab majority in 2017 & 2019 but it is quite a volatile seat (bit like Sheffield Hallam in some respects) and the huge Lab majority there at present is misleading. Those are the figures in yesterday's 'FindOutNow/Electoral Calculus' MRP seat projection. EC's central seat estimate is a different thing. If you enter their current polling figures as your own prediction, it shows Bristol West as a Green gain from Lab. It looks to me like EC have made a few tweaks to their seat predictor, as it is producing higher LibDem seats, as well as a 2nd Green seat. This may be based on Local Election results, which would be consistent with the Greens doing well in Bristol . Incidentally, if you take current VIs as Lab 45, Con 25, LD 12, Grn 5, Ref 3 and then give 50% tactical voting between Lab, LDs and Greens, EC shows a tie for 2nd place between the Tories and LDs with each taking 51 seats. Lab have 505 and Green 2. So maybe Ed Davey and Suella Braverman taking turns as Leader of the Opposition? |
|
|
|
Post by James E on Aug 9, 2023 18:58:04 GMT
The tories will be lucky to keep any seats in Scotland
www.electoralcalculus.co.uk/blogs/ec_mrppoll_20230809.html
Whether they are accurate or not in individual seat predictions on current VI or not, they are probably reasonably correct that lots (22) of the seats will be marginal (<5% between winner and runner up), while around the same number (23) will be relatively safe (>10% winning margin).
Of the current Con seats, 3 are predicted to go SNP -
Moray by 5.9%
Aberdeenshire West and Kincardine by 5.2%
Dumfries and Galloway by 3.4%
while Con would cling on in 3 -
Berwickshire, Roxburgh and Selkirk by 2.4%
Banff and Buchan by 2.3%
Dumfriesshire, Clydesdale and Tweeddale by 0.6%
(though with a Lab VI in these three of 19%, 17% and 22% respectively and SGP VI of 4% in each, one might reasonably expect sufficient extra votes to go SNP to defenestrate Con - again).
In the SNP/Lab contests, there are 7 of the 12 Lab gains predicted to be by less than 5% (Glasgow Central by only 0.1%), while 9 of the SNP retentions also have a predicted margin of <5%.
I wouldn't want to risk any cash on predicting the eventual seat numbers per party!Many thanks for that, ON. I would strongly reccommend the EC blog which you have linked, and the full tables are also available. One detail of these which I think it well worth sharing is the overall VI for Scotland, with a sample of 900 (so close to a full scottish Westminster poll). SNP 40% (-5) SLab 30% (+11) SCon 12% (-13) SLD 8% (-1) S Green 7% (+6) That's a fair bit better for the SNP than most other recent polling, and to my surprise, the SNP to Lab seat movements are more nor less than the 8% swing would suggest. I do think that such a small sample is still not sufficient for good seat projections though, and I guess the same should probably apply for Wales (with a sample of 570). |
|
|
|
Post by James E on Aug 8, 2023 20:44:32 GMT
A few comments on those 'Find Out Now' MRP resuts. etc .... Thanks for the additional analysis, James E . One thing that does occur to me is that, whilst the headline CON and LAB figures are broadly in line with yesterday's YouGov and R&W, the sample in the C4 poll is apparently c11,000. I couldn't readily see from the polls yesterday their sample size, but I suspect it will have been substantially lower (1-2000?). If that is indeed the case, it suggests a lower MOE to me in its findings, purely on the basis of a much larger sample than is typical? Grateful for your thoughts. NB Few people, I would suspect, expect a seats result anything like this at the GE. However, the mood music certainly suggests to me that LAB is on course to be the next UK Govt, and CON look like they're needing snookers. Those sample sizes for recent polls are 2,000 for R&W and 2,313 for YouGov, but then you really need a very large sample for worthwhile MRP results. A sample of 11,000 would only include 1,000 in Scotland, so this would make it difficult to get really reliable seat-by-seat projections. YouGov's pre-GE MRPs have used samples of around 50,000. MoE must be lower with a larger sample, but actually the FON headline figures with 11,000 are not far at all from the first 6 other polls we have in August, for which the combined samples are around 9,400. The Labour average in these is exactly 46% (if you take the unadjusted Opinium figure as Lab 44% for a like-for-like comparison) while the Tories have averaged 25.7%. |
|
|
|
Post by James E on Aug 8, 2023 19:37:58 GMT
A few comments on those 'Find Out Now' MRP resuts. The headline VI figures were Lab 46%, Con 24%, so at the top end of recent Labour polling leads. This would be an overall swing of 17%, with Con down 21 and Lab up 13. Their seats projection leaves the Tories with only 90, whereas UNS would give them around 150-170, even on these dire figures. That pattern of the Tories faring considerable worse than UNS - and Lab + LDs a lot better - should be apparent from all detailed polling findings. However, FON's figures show large numbers of 'Don't Knows', and these are treated as non-voters rather than re-allocated per Opinium. Such as adjustment could reduce the Labour headline lead by 4-6 points. Previous 'Find Out Now' MRPs have shown even worse outcomes for the Tories: in January, they had one showing the Conservatives on only 45 seats, so they have doubled this in the past 6 months: en.wikipedia.org/wiki/Opinion_polling_for_the_next_United_Kingdom_general_election#Seat_predictionsFor parties other than Con and Lab, I would trust FON's seat figures rather more than those of Focaldata, whose methodology recently gave the LDs, PC and Green Party a combined total of just 7 seats (9 May 2023). It will be useful to see the overall figures for Scotland, but I for one am not too surprised by the SNP retaining 38 seats - presumably with Labour taking 11-15 in Scotland. This would be consistent with the SNP being around 5% ahead of Labour in Scotland. This sounds credible enough to me in the light of other recent Scottish polling showing the SNP 0-3% ahead, but with a likely squeeze of the SGP, who have taken a sinificant share of the voters the SNP has lost since 2019. The other factor to bear in mind with Westminster seats in Scotland is that there will almost certainly be a lot of very close contests again - like in 2017, but probably even more so in the SNP/Lab contests. |
|
|
|
Post by James E on Aug 8, 2023 17:22:35 GMT
Latest YG 6 poll Scots average, with changes from 2019 GE (done as a kay9 experiment) SNP: 34% (-11)SGP: 7% (+6)SLab: 34% (+16)SCon: 13% (-12)REFUK: 3% (+2)SLD: 8% (-2)EDIT : Didn't work  No experimental formatting but just to note that the last 6 Deltapolls' averaged Scottish sub-samples are fairly similar: Lab 36% SNP 35% Con 15% The most recent full Scottish Westminster polls still show a narrow SNP lead of 0-3 points. |
|
|
|
Post by James E on Aug 7, 2023 15:23:48 GMT
I occasionally look into the detail of cross-breaks, 2016 Remain Lab 56% (+7) Con 13% (-6) 2016 Leave Lab 26% (+12) Con 42% (-32) Is the number of people who either didn't vote in 2016, or were unable to vote then, now significant enough that they could get their own cross-break? Conveniently, those who were too young to vote in June 2016 are now exactly those in the youngest age cohort of 18-24 year olds. Over the past 5 YouGovs, they average Lab 63%, Con 11%. It would be harder to get an accurate figure for those who could have voted in 2016, but didn't. I've seen figures for non-voters at GE2019, and these break heavily for Labour (a little more than the headline voting figures). It can be noted that the combined swing of those who voted in the referendum is around 14%, as opposed to 16% in YG's overall averaged figures for the past 2 months. |
|
|
|
Post by James E on Aug 7, 2023 11:31:33 GMT
sheviiAgree - polls have really been much the same for many months. The last significant movement was the Conservatives' partial recovery in the early months of Sunak's time as PM, last Autumn. I occasionally look into the detail of cross-breaks, but even there we have little movement. Here are the averages of the past 9 YouGovs, covering all their GB polls in June and July. The overall average has been a 20% Labour lead, so a 16% swing. Overall, we're only a few points different to where we were 6 months ago. London Lab 53% (+5) Con 18% (-14) Swing Con to Lab 9.5% (was 11% in Jan 2023) South England Lab 38% (+16) Con 30% (-23) Swing Con to Lab 19.5% (was 20% in Jan 23) Midlands Lab 45% (+12) Con 30% (-24) Swing Con to Lab 18% (was 19% in Jan 23) North England Lab 52% (+9) Con 21% (-18) Swing Con to Lab 13.5% (was 15% in Jan 23) 2016 Remain Lab 56% (+7) Con 13% (-6) Swing Con to Lab 6.5% 2016 Leave Lab 26% (+12) Con 42% (-32) Swing Con to Lab 22% ABC1 Lab 45% (+12) Con 23% (-20) Swing 16.5% C2DE Lab 43% (+10) Con 26% (-22) Swing Con to Lab 16% Comparisons are to YouGov's own analysis of GE2019. YG tend to show the Tories faring better among C2DE voters, whereas most other pollsters who provide cross-breaks by class show the reverse.
|
|
|
|
Post by James E on Aug 6, 2023 18:14:00 GMT
Talking about MP's second income. Nadine Dorries hasn't actually voted or spoken in parliament for over a year. She has been noticeable by her absence at her [mid Bedfordshire] constituency for even longer to the point where the largest town there have written about their concerns about not having parliamentary representation. It could be argued that pocketing at least £86000 + expenses while doing nothing at all is fraud, however there isn't actually a requirement for MP's to do anything and many are notable for doing bugger all for decades. However there is a 1801 law that requires MP's to attend parliament at least sometimes in the course of a parliamentary session and Nads hasn't. If she continues to take her salary while sulking about not being allowed to do nothing in the lords instead it's possible for the parliamentary standards committee to recommend a sanction. They used to lock wayward mps up in the clock tower but a ten day suspension from parliament triggering a recall petition would do the job and it's envisaged that if she doesn't depart under her own power soon that's precisely what will happen. I think locking her up under big ben would be more fun. On top of this, Dorries also still employs her daughters as 'senior secretaties' at a combined annual cost to the public of around 80K, to assist her in doing nothing as an MP. Neither of them lives near her constituency, nor Westminster www.indy100.com/politics/nadine-dorries-bbc-nepotism-daughters-b1932514 |
|
