|
|
Post by James E on Feb 24, 2023 17:38:19 GMT
|
|
|
|
Post by James E on Feb 24, 2023 10:19:57 GMT
"Techne UK
NEW POLL: Labour extend lead to 22 points:
Lab 49% (+1)
Con 27% (nc)"
Some polling movement at last from Techne !
Their eight polls so far this year show leads of 21,21,19,21,21,21,21 and now a 22%.
I don't know why this is, but their figures are always the most static of all pollsters. Surely proper random sampling "should" produce the variances that other pollsters show? And their tables are among the most impenetrable, making any analysis very difficult.
|
|
|
|
Post by James E on Feb 23, 2023 16:06:52 GMT
Kantar's 17-point Lab lead is still the equal-largest they have ever shown.
Their figures have for some time been the most positive-for-the-tories of any pollster. And their 36/36 tie in March of last year was the last poll not to show Labour ahead.
|
|
|
|
Post by James E on Feb 22, 2023 16:14:57 GMT
There can't have been many polls in the run up to '97 that were better for Labour than that can there? I thought I would have a look. If you assume that the next GE will be October 2024 (which is my guess), then we are roughly 19 months away frin the past om it. The equivalent period compared to the May 1997 election would therefore be about October/November 1995. The average Labour lead of the 5 polls in October 1995 is a staggering 30.3% (range 17-39.5%), the average for the 3 polls in November is 25.3% (range 17-30%). So Starmer is not quite as far ahead as Blair was, although the leads are still huge of course. The 1997 result (UK figures) was a 12.5% Labour lead, so about half that November 1995 (GB) position. en.wikipedia.org/wiki/Opinion_polling_for_the_1997_United_Kingdom_general_electionWhile checking the old polls from 1995-97 a few days ago, what I found was that the Labour lead came dowm by 10 points from around 20-22 months before GE1997 (from an average of 28% to 18% in the final polls). Those final polls then overstated the Labour lead by about 5 points. I think it is best to look at the polling movement and polling error separately, because although it has been common for polls to understate the Conservatives, they did not do so in 2017 or 2019. And polling has moved on in the past 25 years, with many pollsters now re-weighting so as to remove the 'Dont Knows' effect. We know that Opinium do this, and it seems likely that Delta, Savanta and Kantar do too. So for a comparison to 1995-97, we should compare these pollsters to the only company which then adjusted their figures in a similar way. This was ICM, whose Lab lead was typically around 20% in 1995. |
|
|
|
Post by James E on Feb 21, 2023 20:55:37 GMT
That's an 18.5% swing in R&W's 'Red Wall' poll compared to 19.5% in their recent GB poll (the one with a 27-point Lab lead).
And that's despite the Tories having two advantages in these seats: the 'first-time incumbancy bonus' , and a very low LibDem vote for Labour to win-over/squeeze.
|
|
|
|
Post by James E on Feb 21, 2023 18:14:32 GMT
Increased Labour lead again with Deltapoll - this is their highest lead for months.
🚨🚨New Voting Intention🚨🚨
Labour lead is twenty-two points in latest results from Deltapoll.
Con 28% (-)
Lab 50% (+2)
Lib Dem 9% (+1)
Other 12% (-4)
Fieldwork: 17th - 20th February
Sample: 1,079 GB adults
(Changes from 10th - 13th February 2023
|
|
|
|
Post by James E on Feb 19, 2023 18:15:35 GMT
Quota sampling is a non-probability method where interviewers try to interview particular types of people until they meet demographic target quotas. As with other non-probability methods, quota sampling has no grounding in statistical theory and is likely to suffer from non-random error. A particular problem with quota samples is availability bias – if a relevant attribute such as vote choice is correlated with being available to take the survey within demographic groupings (i.e. being at home when the interviewer calls), then quota samples are likely to give incorrect answers. Inaccurate quotas and availability bias were suggested as major factors in the 1992 polling miss (Market Research Society Working Party 1994; Moon Reference Moon1999)." My sister once did a couple of months working for an opinion poll company as an interviewer. Their quotas were based on age, gender and socio-economic group (there may have been others). Once all the quotas had been reached then that was job done for the day. If one was unlucky, then it could be that, to meet the quotas, they would end up looking for an 18-25 company director or other unlikely groupings. Of course, those who had been doing it for more than just a summer job had become canny at picking just enough people to meet all the quotas, without having to interview several "additional" people just to get, say, the final "AB" group. Which means that, not only would there be availability bias, but there could also be selection bias - ignoring some people simply to fit as many of the quotas in as few as possible. No doubt the polling companies are aware of these factors, and hence each will have a different methodology to compensate. I hope. There is a little quirk of YouGov's sampling which illustrates this problem. YouGov weight by political attention, to guard against the problem that polls are notoriously prone to disproportionately reflect the views of those with a high level of political engagement. They aim for 19% of their sample to have a 'low' level of political interest, and of course have samples for 'medium' and 'high'. However, there is an obvious practical difficulty in getting people with a low interest in politics to give their political views. YouGov typically only get about half of the 'weighted' figure of about 400, so have to upweight the 200 or so responses that they get from those with little interest in politics. This may be part of the explanation for the volatility of their results, though it may help a bit that YouGov take an overall sample of 2,000 for their GB polls compared to about 1,000 or so for Deltapoll, Ipsos and Omnesis. |
|
|
|
Post by James E on Feb 18, 2023 21:30:12 GMT
James E 'So if we get a the same Tory recovery to 2015 (10 points in the polls) AND the same level of polling error to 2015 ( 6 points), we'd end up with Labour 6-8% ahead.' Thanks for the analysis, my antennae has a Lab win over the Tories of 5-10% at the GE which may just sneak an OM. However, I think at lower end and with incumbency bonuses (often double) I don't think a 5-6% margin would be enough for an OM. I guess this is why the Scottish Seats discussion is so relevant as an extra 5-10 there could make all the difference. That 10-point recovery in the polls is also in line with what John Major achieved from mid-1995 to May 1997; the Labour polling lead was reduced from 28%in May-July 1995 to an average of 18% in the pre-election polls. And those pre-election polls overstated the actual Lab lead by 5 points. But of course, my figures also allow for a polling error in line with the near-record 6 points of GE2015 (only GE1992 was worse). With a 'normal' polling error (understating the Tories by a net 3 points) Labour would win by 10%. And if the polls are broadly right (as they were in 2019) that becomes 13%. |
|
|
|
Post by James E on Feb 18, 2023 20:47:56 GMT
Looking back to mid-2013 - so 2 years before the May 2015 General Election- the bulk of pollsters put Labour 10% ahead on average. However, ICM, who adjusted their figures for 'Don't Knows' averaged more like 4%, ranging from 6% Lab leads to a 36/36 tie. Today, the 'unadjusted' polls e.g. YouGov put Labour about 22-25% ahead, while for those such as Opinium who re-weight the Don't Knows in line with previous votes it's about a 16% lead. en.wikipedia.org/wiki/Opinion_polling_for_the_2015_United_Kingdom_general_election#2013So if we get a the same Tory recovery as in 2015 (10 points in the polls from 2 years before) AND the same level of polling error to 2015 ( 6 points), we'd end up with Labour 6-8% ahead. |
|
|
|
Post by James E on Feb 18, 2023 20:06:11 GMT
There has been a second Scottish Westminster poll post YouGov's "SNP 38%, Lab 35%", this time from Survation with a 13 point SNP lead. (SNP43%, Lab 30%) www.survation.com/scotlands-political-landscape-after-nicola-sturgeon/Looking at the last 10 Scottish Westminster polls, starting from late Nov, we have, in chronological order, SNP leads of 26,14,13,13,14,13,13, 3, 10 and 13. So this 13-point lead per Survation is right in line with the mean, mode and median of recent Scottish polling. YouGov was an outlier, as was Ipsos Mori's 26% lead in early December. The cross breaks for switching between SNP and Scottich Labour per Survation are back to normal as well. Their tables show 11% of 2019SNP>SLab, and 10% of 2019 SLab>SNP. In a marginal with SNP less than 15 points ahead, this produces less than 1% of direct SNP>Lab swing. |
|
|
|
Post by James E on Feb 17, 2023 19:42:50 GMT
That's a bit of a contraction of the Labour lead in Wales to 29%, but even so, a 12.5% swing would be enough to unseat all eleven current Tory MPs in Wales. Monmouth is the hardest, needing a 10% swing, but 8 of the Conservatives' 11 Welsh seats are tight marginals, needing swings of 3.25% or lower. The swing of 12.5% in this poll, or 14.5% in YouGov's previous one 2-3 months ago, is still somewhat lower than their general GB swing of around 17%. However, it is much in line with cross-breaks for the (combined ) North of England where Labour led by about 4% in 2019 , just as they did in Wales. |
|
|
|
Post by James E on Feb 17, 2023 16:40:03 GMT
pjw1961 "I think that EC's MRP, and probably other similar analyses, are overestimating the number of non-urban seats Labour will take, which in turn makes a small working majority Labour government a better bet than a landslide. " There is some evidence for a lower swing in rural areas in a poll YouGov did for the Fabian society in December. This showed Con and Lab as on level-pegging, compared to a 33% Tory lead in GE2019. This is reported by the Mirror as an amazing turnaround, but a comparison with YouGov's GB polling in December with Lab leads of 24% and 25% shows that the rural swing appears to be a couple of points lower: 16.5% compared to 18-19% generally. www.mirror.co.uk/news/politics/rural-voters-just-likely-back-28803967EDIT - But there is also very good evidence that Labour are achieving lower swings wherever they are already ahead - especially in London. Given Labour's dominance in the cities, evidence of lower-than-average swings both there and rural areas might be taken to imply some very large swings in the intermediate territory: the large or small towns. Either that, or the polls are just wrong. Incidentally, I'd see the Huntingdon constituency as an almost equal mix of 'small town' and rural voters, as it covers both Huntingdon itself (pop 26K) and St Ives (17K pop). It has shown a significant overall trend to labour since 1997 when it was the Tories' safest seat in the country.
|
|
|
|
Post by James E on Feb 17, 2023 12:13:50 GMT
A couple of polls have come out this morning.
The Labour lead is up 2 points to 23% with Omnesis, and down 2 points to 27% with People Polling. Both of these pollsters are in the group with the highest Lab-leads, along with YouGov and Ipsos.
|
|
|
|
Post by James E on Feb 17, 2023 10:15:50 GMT
My analysis of Scottish Westminster cross-breaks shows why a simple x% swing from SNP to Lab is unlikely to reflect reality. oldnat may well have better figures, but (for what it's worth) here are mine: 2019SNP>Lab switching averages 11% per the recent cross-breaks I have found. However, there is also 2019Lab>SNP switching of about 10%. This produces around a 3% SNP>Lab swing in the context of the overall 2019 shares of SNP 45%, Lab 19%. But that swing would be only 1% (or lower) where the difference between SNP2019 and Lab2019 vote shares is 15% or below. As a polling-based formula for the SNP-held Labour targets, I'd suggest applying this: 11% Of SNP2019>Lab 10% of Lab2019>SNP 15% of Con2019>Lab 25% of LD2019>Lab This works at the level of the overall movements since 2019, turning SNP 45%, Lab 19% into SNP42%, Lab 29%. And it is consistent with current polling, particularly YouGov. However, it is not an attempt to predict the Con and LD shares (there are further losses esp for the Tories to Ref), and it's rather simplistic for the SNP for whom there are both gains and losses. graham . This is an attempt to work out what the detailed trends of current polling would be if they were reflected in Labour's SNP-held target seats. Obviously, if the result of the next GE is SNP37%, Lab 33% as you predict, things would be very different. |
|
|
|
Post by James E on Feb 16, 2023 20:53:59 GMT
To see the likely outcome of the SNP/Lab contests, it's important to note the actual movements between the parties since 2019. I've done some rough calculations based on the most recent YouGov Westminster poll ( SNP 42, Lab 29 etc on 23-26 Jan, tables below) . This shows 12% of the SNP2019 voters moving to Lab and 10% of SLab2019 moving to SNP. The net effect of this would be a 1.75% direct SNP>Lab swing. There is also some Con>Lab swing (@ 2%) and LD>Lab swing (@1.6%). The problem for Labour is that the actual movements from Con or LD will be lower wherever the Tory vote was below the 25% they achieved across Scotland in 2019 and/or the LD votes is less than 9.5% they achieved last time, or where the Lab vote was higher than the 19% they achieved then. This is the case in most, maybe all, SNP/Lab contests. Tactical voting could provide some assistance, but not enough to overcome the basic problem for Labour : in Scotland they look likely to get lower swings in the marginals. docs.cdn.yougov.com/q06xam6c7j/TheTimes_VI_ScotTabs_Jan23_W.pdfThere is no obvious reason why the big Labour to SNP swings seen in many seats over the period 2017 - 2019 cannot be reversed over the much longer period 2019 - 2024. Had there not been an election in 2019 so that we were viewing 2017 as the starting point, Labour's prospects would look rather different.I've tried doing exactly as you suggest: use GE2017 as the starting point, and then see how current Scottish Westminster polling compares. Here are the current average Scottish VI figures, with changes from 2017: SNP 43% (+6) SLab 29% (+2) SCon 17% (-12) LibDem 6% (-1) Even though Labour are up two points, the SNP has made a larger gain of 6 points, so a 2% Lab>SNP swing. And if you look at the seven Scottish seats that Labour won in 2017, five of them had majorities of less than 4%. These five would all now revert to the SNP on UNS from GE2017, leaving Labour with two seats in Scotland, Edinburgh South and East Lothian. For what it's worth, my comparison from the GE2019 results looks just a little bit better for Labour than this . I think Kirkcaldy and Cowdenbeath currently looks a likely Lab gain, and Midlothian plus a few others may be very close (but with SNP narrowly ahead). |
|
|
|
Post by James E on Feb 16, 2023 15:07:54 GMT
That Polling Report projection for Ashfield prompted me to check what kind of result they predicted for Cambridge. pollingreport.uk/seats/E14000617Unsurprisingly, it's a near 40% Lab majority, with the LDs in second. And while the Conservative vote will presumably slump even lower here, it seems a little OTT to predict Con 0% (-16). But I think the more important point is that this illustrates why a simple UNS model is not a good predictor when there is a large swing. |
|
|
|
Post by James E on Feb 16, 2023 9:47:55 GMT
To see the likely outcome of the SNP/Lab contests, it's important to note the actual movements between the parties since 2019. I've done some rough calculations based on the most recent YouGov Westminster poll ( SNP 42, Lab 29 etc on 23-26 Jan, tables below) . This shows 12% of the SNP2019 voters moving to Lab and 10% of SLab2019 moving to SNP. The net effect of this would be a 1.75% direct SNP>Lab swing. There is also some Con>Lab swing (@ 2%) and LD>Lab swing (@1.6%). The problem for Labour is that the actual movements from Con or LD will be lower wherever the Tory vote was below the 25% they achieved across Scotland in 2019 and/or the LD votes is less than 9.5% they achieved last time, or where the Lab vote was higher than the 19% they achieved then. This is the case in most, maybe all, SNP/Lab contests. Tactical voting could provide some assistance, but not enough to overcome the basic problem for Labour : in Scotland they look likely to get lower swings in the marginals. docs.cdn.yougov.com/q06xam6c7j/TheTimes_VI_ScotTabs_Jan23_W.pdf
The pattern in the YG poll is, however, fairly standard across most of the Scots polls I've looked at (lots!) -
A large swing from SCon to SLab : A small swing from SCon to REFUK : A small swing from SNP to SGP : A small swing from SNP to SLab : A big swing (though of a small number) from SLD to SLab : A higher level of DKs among SCon & SLD (though that's normal for SLD).Could I ask if you have more data on these figures? Particularly the movements between SNP and SLab. I am very aware that I am drawing conclusions from the swings from a single YouGov poll, while others are using approximations and 'what-if' figures. One of the thing which struck me is that the proportions of 2019SNP>Lab and 2019Lab>SNP are actually very similar with YouGov (12% v 10%). It's the much higher number of SNP2019 voters which creates the 2%-ish direct swing. But of course this would be much lower in places where the SNP and Lab votes were more evenly matched. Hence the scenario where SLab could increase its vote share by 10 points but gain ouly 2 or 3 seats. [EDIT 8pm - I think I am some way to answering my own question. After checking two YouGovs and a Survation Scotland Westminster poll, the SNP>Lab movement averages 11%, and Lab>SNP 10%. If this pattern holds in the SNP-held marginals, then this a further factor severely limiting the SNP>Lab swing. ] |
|
|
|
Post by James E on Feb 15, 2023 23:47:39 GMT
Just a gentle reminder to some on here, that there is no UK Parliamentary seat called "Western Isles".
If you don't even know the name of a constituency, perhaps better not to try commenting on it? Donald Stewart is recorded as having won the Western Isles from Labour at the 1970 election -a seat he retained until 1987. The constituency name became Na H-Eileanan an Iar about 20 years ago (before the 2005 Election). But I can imagine you expecting it to change back to how it was in 1970. |
|
|
|
Post by James E on Feb 15, 2023 20:26:40 GMT
oldnat Thanks, I realised that the YouGov Scottish poll I was using was a rather favourable one for Labour. But the movements are, as you say, fairly consistent. For @jimjam, graham and others, I'd point out that even the YouGov figures would produce relatively few Lab gains in Scotland, if their movements were replicated at a GE. To take an example of why this is so, look at Coatbridge, Chryston & Bellshill, which is Number 5 on Labour Scottish target list. It needs a 6% swing, and in 2019 was: SNP 47% Lab 35% Con 13% LD 3% Firstly, note that YG show Lab taking 12% of the 2019SNP vote, with 10% SLab2019 going the other way. This produces a lower net SNP>Lab swing wherever Labour is doing better than their overall 19% 2019 vote share . Then there are the transfers from Con and LD to consider. These are worth about 7 points combined in the overall Scottish figures, but with the lower Con and LD votes in Coatbridge, this falls to little more than 1 point. As a result, while the YouGov figures do show Lab gaining 10 points overall in Scotland compared to 2019, the detailed cross-breaks suggest to me an SNP hold here - probably with the majority reduced by about 5 points. And it would be a similar story in all of the Glasgow contests, with their ultra-low Con and LD vote shares.
|
|
|
|
Post by James E on Feb 15, 2023 19:22:45 GMT
To see the likely outcome of the SNP/Lab contests, it's important to note the actual movements between the parties since 2019. I've done some rough calculations based on the most recent YouGov Westminster poll ( SNP 42, Lab 29 etc on 23-26 Jan, tables below) . This shows 12% of the SNP2019 voters moving to Lab and 10% of SLab2019 moving to SNP. The net effect of this would be a 1.75% direct SNP>Lab swing. There is also some Con>Lab swing (@ 2%) and LD>Lab swing (@1.6%). This is mainly due to Lab taking 19% of Con2019 and 42% of 2019LD - so similar figures there to England. The problem for Labour is that the actual movements from Con or LD will be lower in seats where the Tory vote was below the 25% they achieved across Scotland in 2019 and/or the LD votes is less than 9.5% they achieved last time, or where the Lab vote was higher than the 19% they achieved then. This is the case in most, maybe all, SNP/Lab contests. Tactical voting could provide some assistance, but not enough to overcome the basic problem for Labour : in Scotland they look likely to get lower swings in the marginals. docs.cdn.yougov.com/q06xam6c7j/TheTimes_VI_ScotTabs_Jan23_W.pdf |
|
|
|
Post by James E on Feb 14, 2023 23:02:37 GMT
domjg "Newbury, a wealthy, educated, majority remain constituency was Libdem from 93 to 2005. I don't understand what's stopping it from also making that list? Lab have never been close there, it's always been a Tory/LD race. The constituency I live in, Wantage, is seemingly as likely to go Lab as LD in 24 running the risk of it therefore going to neither." Newbury is the LibDems target no 43, but is not included by R&W as the Tories' majority in 2019 was over 15,000. I agree that Wantage is a difficult one to call. Labour were 17 points behind the second-placed LDs in 2019. On the other hand, this poll shows a net movement from LD to Lab of 30 points since 2019, so even if you allow for some Con or LD recovery at the expense of Labour, the latter may be the better bet. A simple application of the changes in this (and previous) 'Blue Wall' polls puts Labour narrowly ahead in Wantage; using today's poll it would be Lab 35, Con 34, LD 22.
|
|
|
|
Post by James E on Feb 14, 2023 21:26:42 GMT
pjw1961R&W did some sort of 'MRP' with their previous figures 2 weeks ago. This showed a total of 24 Labour gains from their 42 'Blue Wall' seats. 9 of these must have come from the LD targets listed here (with Lab in 3rd place in 2019).
|
|
|
|
Post by James E on Feb 14, 2023 21:04:45 GMT
Re the Redfield and Wilton 'Blue Wall'. Compared to GE2019, this latest poll is: Lab 41% (+20) Con 34% (-16) LibDem 17% (-10) So an 18 point swing compared to 17 points per this week's G&W poll of GB - all fairly typical. The slight movements since they last did this poll are probably just 'noise' as other polls don't show a rise in the Con share and a loss for the LibDems. But as for the LibDems : even taking their 'normal' share in these polls of around 19% (so -8 points compared to GE2019), this really is awful polling for them. I've cross-checked the LibDems' target seats with the constituencies polled here by R&W, and the two coincide almost exactly, with a smaller number of Southern Labour targets thrown in. To List the 27 LibDem targets polled here (with their target numbers): 2 Wimbledon 5 Chetenham 6 Winchester 8 Cambridgeshire South 9 Esher and Walton 10 Lewes 11 Guildford 13 St Ives 14 Cities of London and Westminster 16 Hitchin and Harpenden 17 Finchley and Golders Green 18 Wokingham 19 Surrey SW 20 Wells 21 Sutton & Cheam 24 Cambridgeshire SE 26 Woking 27 Taunton Deane 28 Wantage 29 Chippenham 32 Romsey and Soton North 33 Dorset West 34 Thornbury and Yate 36 Henley 37 Chelsea & Fulham 38 Totnes 44 Tunbridge Wells Plus 15 more where Labour were 2nd in 2019. www.electionpolling.co.uk/battleground/targets/liberal-democrat |
|
|
|
Post by James E on Feb 14, 2023 20:32:58 GMT
Re 'Labour back 20 points ahead with Delta': That's a high Labour lead for Deltapoll. They've been averaging 16-17 points for the past 2 months (or 9 polls). As I've mentioned before, their figures make me think that they altered their methodology in the Autumn (perhaps an Opinium-style adjustment of DKs?) but I have not seen any details or confirmation of this. It looked like a slow drift from the Tories towards Labour for me. Taking just this year's polls (6) Con 31%, 29%, 30%, 29%, 29%, 28% Lab 45%, 45%, 44%, 46%, 47%, 48% Other regular pollsters like R&W seem to be pretty level over the same time period. I'm looking at the longer-term trend in Deltapoll's figures. They averaged a 12 point lead last summer in their final 4 polls when Johnson was PM. Now over the past 2 months they are averaging around a 17-point lead. As most pollsters have shown the Labour lead growing by 10-12 points over the same period, it looks to me like Delta must have made changes at some point of the Autumn. |
|
|
|
Post by James E on Feb 14, 2023 12:59:31 GMT
Re 'Labour back 20 points ahead with Delta':
That's a high Labour lead for Deltapoll.
They've been averaging 16-17 points for the past 2 months (or 9 polls). As I've mentioned before, their figures make me think that they altered their methodology in the Autumn (perhaps an Opinium-style adjustment of DKs?) but I have not seen any details or confirmation of this.
|
|
|
|
Post by James E on Feb 12, 2023 20:56:04 GMT
On rechecking YouGov's figures, they found that (compared to GE2015) Conservative voters had split 39/61 for Leave, Labour voters 65/35 for Remain, LDs 68/32 for Remain, and Greens 80/20 for Remain. Only 2015 UKIP voters were monolithically on one side, being (unsurprisingly) 5/95 for Leave. d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/oxmidrr5wh/EUFinalCall_Reweighted.pdfBut things have moved on so much since then: YouGov's most recent use of their tracker, on 24-25 Jan (Con26, Lab 45) showed the current Tory VI dividing 78/22 in believing Brexit to be Right, and Labour's current VI dividing 14/86 for Wrong (excl a small number of DKs). Current LDs divide 16/84 for Wrong, too. Such data may simply represent views on the Brexit deal finally agreed - rather than the principle itself. Many of those now hostile to Johnson's deal could well still be content with a much softer Brexit - eg a Norway type deal. The question. "In hindsight do you believe Britain was right or wrong to vote to leave the EU" seems to me to about the principle (or at least the initial decision) rather than the withdrawal agreement and subsequent Trade and Cooperation agreement. Besides, the figures are very close to those for recent Rejoin/Stay Out polls. en.wikipedia.org/wiki/Opinion_polling_on_the_United_Kingdom_rejoining_the_European_Union_(2020%E2%80%93present) |
|
|
|
Post by James E on Feb 12, 2023 20:18:48 GMT
You are right, of course, in pointing out that no party is a monolithic block of individuals with identical views.I seem to remember that when, I think it was YouGov, looked at party affiliations after Brexit they found 34% of Labour voters, 33% of SNP voters, 30% of Lib Dem voters and 25% of Green members voted Leave (along with about 68% of Tory voters), so on the Left the EU wasn't universally popular even though all the leaderships of the parties were asking them to vote for Remain. It may well be that there is both a majority for independence in Scotland and a majority for re-joining the EU, but they are composed of different people. On rechecking YouGov's figures, they found that (compared to GE2015) Conservative voters had split 39/61 for Leave, Labour voters 65/35 for Remain, LDs 68/32 for Remain, and Greens 80/20 for Remain. Only 2015 UKIP voters were monolithically on one side, being (unsurprisingly) 5/95 for Leave. d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/oxmidrr5wh/EUFinalCall_Reweighted.pdfBut things have moved on so much since then: YouGov's most recent use of their tracker, on 24-25 Jan (Con26, Lab 45) showed the current Tory VI dividing 78/22 in believing Brexit to be Right, and Labour's current VI dividing 14/86 for Wrong (excl a small number of DKs). Current LDs divide 16/84 for Wrong, too. |
|
|
|
Post by James E on Feb 10, 2023 17:17:07 GMT
I agree that it is an interesting analysis worthy of consideration, especially to political elements of a hung parliament. However Kellner only considers two options for the voting (a) a uniform swing and (b) a uniform swing plus tactical anti-Tory voting. However, that is not what the polls are pointing to. Rather they suggest an non-uniform swing with Labour picking up more votes where it does them good rather than areas where they already dominate. That - especially with tactical voting as well, which seems likely - could make Labour's task at the next GE easier that an uniform swing would suggest. I was rather surprised that Kellner has based his article on Uniform swing when there is so much evidence of a non-uniform (or 'ratio') swing. There are two main factors that are driving this: firstly, there is a clear pattern across all pollsters of the Conservatives losing more support where they are strongest; and secondly, all pollsters show a significant LD>Lab movement compared to 2019, with between 30% and 45% of 2019LDs (dependant on which polls you believe) switching to Labour. The first point is illustrated by Redfield and Wilton's polling in Conservative-held seats in the North and South of England for the past 2-3 months. In their Northern "Red Wall" polls, they show a 16-17 point Con>Lab swing and a slight rise (1%) in the LD share where they were a poor third place in 2019 (4.5%). However, their Southern "Blue Wall" polls show a 19-20 point Con>Lab swing, which is boosted by a significant fall in the LD vote. They are down 7 points from the 27.4% they took in these seats in 2019, which is all the more significant as the seats sampled include most of their prime targets. The same polling showed Labour taking more seats from 3rd place (8) than the LDs were gaining from 2nd (6). This has huge implications for any tactical voting. Then there are the regional cross-breaks: all 4 companies whose recent figures I have analysed ( Opinium, Delta, YouGov and R&W) show a swing at least 3 points above their prevailing average in the South of England ( SE+SW+E), a swing at least 2 points above their average in the Midlands, and a swing at least 6 points below their GB average in London. What would be really interesting would be some polling, or even a by election, from a relatively safe-looking Tory seat. |
|
|
|
Post by James E on Feb 10, 2023 15:54:06 GMT
"If you had to choose, which Prime Minister has been the best since the 2019 general election?
Boris Johnson: 26%
Rishi Sunak: 15%
Liz Truss: 2%
Via
@peoplepolling
, 8-9 Feb.
2:45 PM · Feb 10, 2023"
57% of respondents either don't know or prefer not to say. And there's a little quirk that Liz Truss scores significantly higher in Scotland (@5%) than England (1 or 2%) - presumably for similar reasons that applied to the late Graham Taylor.
|
|
|
|
Post by James E on Feb 10, 2023 15:24:57 GMT
Almost no change with YouGov; figures are right in line with the overall polling average for February 2023 (so far):
Westminster Voting Intention:
LAB: 47% (-1)
CON: 24% (=)
LDM: 10% (+1)
GRN: 6% (=)
RFM: 6% (=)
SNP: 4% (=)
Via
@yougov
, 7-8 Feb.
Changes w/ 31 Jan - 1 Feb.
|
|
