|
|
Post by James E on Apr 11, 2024 20:12:17 GMT
Hmmm... To quote part of the article in question: "If the polls are level or even if Labour is five points ahead, it could be extremely damaging for Labour and could mean they fail to win a whole host of English bellwether marginal seats.” "Polls have shown Labour about 20 points ahead and heading for a majority of over 150..." So they appear to be talking about a scenario where there is a substantial swing back to the Conservatives. And using current YouGov 6-poll averages, Labour are still making gains relative to 2019 with the youngest cohort of voters, enjoying a polling lead of 53 points. Those age-based cross breaks are: Age 18-24 Con 8% (-13) Lab 61% (+5) Swing Con to Lab 9% Age 25-49 Con 12% (-19) Lab 55% (+12) Swing Con to Lab 15.5% Age 50-64 Con 21% (-29) Lab 41% (+14) Swing Con to Lab 21.5% Age 65+ Con 35% (-29) Lab 25% (+9) Swing Con to Lab 19% What the article gets right is that there is good reason to expect Labour's performance to be very uneven. However, the seats where the swing is low or negative are likely to be city seats which are mostly ultra-safe Labour seats. There are likely to be just a few seats such as Bristol Central and perhaps Sheffield Hallam where they are facing challenges from the Greens or LDs, but the article acknowledges that these are very much the exceptions. In fact, it is notable that despite the title 'Labour may fail to gain target seats' that they do not mention a single Labour target seat (as opposed to a couple of labour-held seats looking vulnerable). |
|
|
|
Post by James E on Apr 11, 2024 19:28:22 GMT
A significant number of voters who supported Ukip at the 2015 GE and the various by elections and local elections leading up to it would likely have voted LD prior to the party entering the 2010 Coalition. For many Ukip - and later the Brexit party - replaced the LDs as the party of protest or the NOTA option. Were such candidates not to stand, it might - rather ironically - benefit the LDs and the Greens. I don't think this is right, I don't think a significant number transfer direct from Ukip/Reform to Lib Dem or vice versa - their values are so different. Much more likely its a combination of right wing Tory to UKIP/Reform and centrist Tory to Lib Dem movement. I did some analysis of the final YouGov polls prior to the 2015 General Election some time ago, as there does not seem to be any full churn analysis, as they have provided in 2017 and 2019. UKIP were recorded by YG as taking about 15% of Con2010, 10% of LD2010 and 5% of Lab 2010. The remainder of their 12.5% vote share would have come from those who voted for UKIP or other small parties in 2010. So perhaps 1-in-5 or 1-in-6 of those who voted for UKIP in 2015 was a transfer from the LDs. I think at the time many assumed it was far greater than this because the LDs vote share losses and UKIP's gains were similar shares of the overall vote. |
|
|
|
Post by James E on Apr 3, 2024 21:10:31 GMT
shevii The samples used for the MRP-result for each constituency are drawn widely from respondents with the same indicators or demographics, rather than from those in the constituency itself. So on balance, I think it may simply be that the Green Party has chosen its targets well. Also, a part of YouGov's modelling depends on the dynamics of the seat per previous results. This is certainly the case for seats where the LDs are in contention, and they may have done the same in seats where the Greens have performed well previously - although if they did this they would be using quite a small group within the total of 18,000 respondents.
|
|
|
|
Post by James E on Apr 3, 2024 16:59:42 GMT
wb61 I agree with your comments, and they are supported by the regional pattern I have noted with YouGov and other pollsters with the Tories' support holding up better in London or Scotland. The quote above ("...the drop in Conservative party support estimated by this model is still proportional..") is YouGov's words, not mine. Their 'unwinding' adjustment moves their seat estimates somewhat away from proportionality and more towards UNS. This increases the number of seats which the Conservatives hold. I really don't know how appropriate this is, although I suspect that it may turn out to be valid. To add - there is a comparison to be made between different MRPs and models as to how many Labour gains they show per point of Con to Lab swing. It's notable that YouGov show somewhat fewer seats going from Con to Lab (in relation to their headline figures swing) than other companies, but still far more than UNS. UNS typically shows about 10 Lab gains per 1% swing. This YouGov MRP shows 14.3 Lab gains per 1% swing. FON/EC's MRP in Feb showed 15.8 Lab gains per 1% swing and Survation's recent MRP showed 17.5 Lab gains per 1% swing. (my own seat estimates are based on around 15 Lab gains per point swing)
|
|
|
|
Post by James E on Apr 3, 2024 16:29:50 GMT
It's well worth looking at the detail behind the YouGov MRP - in particular note: "The headline results based from this MRP model would be Labour on 41% of the vote, the Conservatives on 24%, the Liberal Democrats on 12%, the Greens on 7%, Reform UK on 12%, and others on 1%. Vote intention derived from our MRP will look different to regular headline vote intention figures published by YouGov because the MRP model probabilistically matches 'missing voters' (i.e., those indicating they do not currently have a vote intention) to similar respondents who do express a vote preference, rather than excluding them from the estimations." yougov.co.uk/politics/articles/49061-yougov-mrp-labour-now-projected-to-win-over-400-seatsThat means that the VI figures are adjusted in the same way as by Opinium or More In Common. A 17-point Labour lead is quite high under this methodology - and it might be a 23-24 point lead if YouGov were not doing this. the 41/24 lead can be compared to the 13.5% lead in a similar YouGov MRP over the Christmas period, but not to their normal polls which have recently averaged around a 24-point Labour lead. On top of this, YouGov are using a new technique which they call 'unwinding' to partially counteract the proportionate effect of MRP. They explain this as follows: " To address this, we have developed a new technique called ‘unwinding’. The unwinding algorithm looks at historical results and learns from them what the typical distribution of party vote shares tends to look like (for each party), and re-fits constituency-level shares in the posterior distribution to better reflect this variance. This has the effect of ‘unwinding’ the posterior distributions to better reflect the spread of constituency-level results at British general elections, and in turn reduces the proportionality of the swing.
That said, the drop in Conservative party support estimated by this model is still proportional. They are for instance dropping much more in places like South Devon (-24 points) than they are in Liverpool Riverside (-3 points). This is however a mathematical necessity – if a party is dropping by a large amount nationally (as the Conservatives currently are), there are constituencies (such as Liverpool Riverside) where their total vote share is smaller than their national decline. It would be impossible for the Conservatives to be on minus vote share in Liverpool Riverside. To make up for this, it must be the case that Conservative party shares are dropping faster where its shares are larger. What the unwinding algorithm does is create a more realistic picture of where, and the extent to which, this is happening."So their seat figures are the product of adjustments for both 're-weighting' (per Opinium etc) and 'unwinding'. These factors probably explain why they have the Tories winning as many as 155 seats, compared to Survation's recent MRP with them on just 98. |
|
|
|
Post by James E on Apr 1, 2024 19:59:34 GMT
graham I've posted on this subject dozens of times before (see top post on link, below) but to be brief : UNS predicts that the Tories' vote losses and Labour's gains are the the same number of voters in both their strong and weak regions (or seats). So if the Tories are down by about 20 points, they fall from 55% to 35% in the South of England or the Midlands, and from 32% to 12% in London. Detailed polling findings show a totally different pattern to this. In fact, the Tories are losing a larger proportion of their support in their strongest regions (such as the South and Midlands of England), but their small vote shares in Scotland and in London are holding up better. This is the opposite to the effect which UNS would predict. So the number of seats changing hands is likely to be far higher than that predicted by UNS. Labour are going to get small, or even negative, swings in their safe seats, and very large swings where they are attacking - a more exaggerated pattern of what happened in 1997. (If you had applied UNS to the 1997 results compared to 1992, Labour would only have made around 105 gains on a swing of 10.5%, and the LDs 15 or so. In reality, Labour made 147 gains and the LDs 28. ) ukpollingreport2.proboards.com/user/48/recent?page=3
|
|
|
|
Post by James E on Apr 1, 2024 17:32:56 GMT
Redfield Wilton Labour leads by 24%. Westminster Voting Intention (31 March): Labour 46% (+4) Conservative 22% (–) Reform UK 14% (–) Liberal Democrat 10% (-2) Green 5% (-1) Scottish National Party 3% (+1) Other 2% (–) Changes +/- 24 March ; "I've put those numbers through electoral calculus. Labour should be safe with its majority of 412 the Liberal democrats are the official opposition in 51 the Tories manage third place 9 ahead of the SNP on 26 and the poor little refukers win refuk all." Yes, these figures with a 24-point Lab lead are about the point at which I reckon the Conservatives would cease to be second in HoC seats - maybe even a bit less in the event that the LibDems rise by a few points. But the current polling average of a 20-point lead would still leave the Tories with about 100-120 seats. The Tories' vote is now rather evenly spread, and there is a point in the low-20s, below which their seat losses would really start to accelerate. |
|
|
|
Post by James E on Mar 31, 2024 14:28:59 GMT
MRPs are variable in quality, and although I consider Survation's polling figures to be among the most reliable, they way this is translated into seats looks rather simplistic. In particular, they do not allow for the different dynamics of seats where the LDs are better placed. YouGov's MRP in January included a specific reminder to voters of the position of their seat as of 2019, which gave a significant boost to the LDs at Labour's expense. I would consider that to be a better guide to the likely outcome when we get an actual election, although there is still going to be a high level of LD to Lab switching.
One part of the Survation MRP which I would want to highlight is the question of the possible effect of Reform UK standing aside. Reform get 8.5% of the vote in the headline figures, while without them, the outcome is:
Con 29.3% (+3.1)
Lab 46.5% (+1.5)
LD 10.9% (+0.5)
SNP+PC 4.1% (+0.2)
Other 4.5% (+2.6)
So Labour still lead by 17 points, and the overall effect is a net boost of just 1.6 points to the Tories, who get about 40% of the previous Reform UK vote. But the effect in seats is far greater, with the Conservatives winning 52 more and Labour 46 fewer. This is because the benefit to the Tories is concentrated into the seats where they were strongest in 2019 - places where they might have been 40 points ahead, but which are now marginals.
For an example of this, I've noted the figures for South Holland and The Deepings. Here, the headline figures are Con 35%, Lab 31.3%, Ref 22.2%. But without Reform standing, this becomes Con 43.3%, Lab 33% - so a swing of 3.3% in a 'new marginal' even though the GB-wide effect is a swing of only 0.8%.
And if we look at a bellweather seat (I've chosen Worcester), Reform UK standing down delivers 3.2% to the Tories and 0.6% to Labour, so just a 1.3% swing from a Reform UK vote estimated as 8.3%.
|
|
|
|
Post by James E on Mar 31, 2024 11:33:45 GMT
Both the two MRPs done recently by different pollsters are actually pretty similar apart from LD and small parties. Both suggest 40 seats for the SNP and ballpark figures for Lab and Con that I would suggest are more or less accounted for by the different headline voting intention percentages. Find Out now: Lab 452 (42%), Con 80 (22%), LD 53 (11%), SNP 40 (4%),PC 4 (1%), Green 2 (7%) Survation: Lab 468 (45%), Con 98 (26%), LD 22 (10.4%), SNP 41 (3.2%), PC 2 (0.7%), Green 0 (4.2%) The Green (and PC) could also be put down to headline percentages rather than anything in the MRP, but it's the LD that clearly shows some sort of different methodology going on given how close their headline percentages are to each other. To predict the LDs' seats, the figure you need to look at is not the LD Voting Intention, but the gap between them and the Conservatives. Looking at the previous 3 MRPs we have had from FON, YouGov and Survation, this seems to me to work: Survation (Sept 2023) Con 29%, LD 11%. 18-point LD deficit : 25 LD seats YouGov (Jan 2024) Con 26%, LD 12.5%, 13.5 point LD deficit : 48 LD seats FON/EC (Feb 2014) Con 22%, LD 11%, 11 point LD deficit : 53 LD seats These figures are all in line with what has happened at previous General Elections. The 18-point LD/Con deficit is close to the 17 and 19 points in the 1983 and 1987 results. The LDs then won 23 and 22 seats respectively. The 13.5% LD/Con deficit is close to the 1997, 2001 and 2010 results. The LDs then won 46,52 and 57 seats respectively. The 11% LD/Con deficit is close to the 2005 result (Con 32, LD 22). The LDs then won 62 seats. The difference between now and those elections from 1997-2010 is that the LDs are no longer in the running for the 5-10 Lab/LD contests which they then won; and the result is 5-10 fewer seats. |
|
|
|
Post by James E on Mar 31, 2024 10:49:12 GMT
The Scots VI figures (and projected seats) in this Survation poll are -
SNP 37% (41) : Lab 32% (14) : Con 16% (0) : LD 7% (2) : REFUK 4% (0) : SGP 3% (0) Thank for this, ON. It's worth adding the other details you posted : "On Survation's mean figures, 22 of the 57 seats in Scotland would have a winning margin of less than 5% under the silly FPTP system. 1 of these is given as an SNP win over Con 9 are given as Lab wins over SNP 12 are given as SNP wins over Lab." Scottish Westminster Polling since the Rutherglen by-election has varied between around a 4% Lab lead and a 10% SNP lead. So this poll is a bit better for the SNP than the average, but not by much. And I think this may be mainly due to Survation's practice of not prompting by party names: this boosts their Con VI (at the expense of Ref), their Lab VI (at the expense of Greens) and their SNP VI (at the expense of SGP). If that 5% winning margin for the SNP were to turn into a tie between them and Lab, that would deliver those 12 seats from SNP to Lab, and possibly one from SNP to Con. So we would have near-parity in seats at SNP 28, Lab 26. Labour won just 7 seats in Scotland when 9.8% behind the SNP at GE2017. Comparing this with the MRP, and the possibility of parity in votes and seats, it looks to me like each 1-point swing between SNP and Lab can flip around 4 seats, or even more near the point where the two are close to being tied. With this in mind, I would tentatively suggest the following as a very rough guide to translating votes into seats for the GE in Scotland: SNP/Lab seats SNP lead by 10% 45/7 SNP lead by 8% 42/10 SNP lead by 6% 39/13 SNP lead by 4% 35/17 SNP lead by 2% 30/22 Tie 26/26 Lab lead by 2% 23/29 Lab lead by 4% 19/32 |
|
|
|
Post by James E on Mar 29, 2024 10:12:21 GMT
Savanta 21pt Labour lead - largest since Jan '23 .... We're now in the position where polls are coming in one of three kinds... Favourable to the Conservatives, so they still lose, and lose badly, but still come second place in terms of seats. Those unfavourable to the Conservatives, where they plunge to third place in terms of seats. And third the polls where they're treading water just above relegation. I think what will happen next is that we'll have the local elections, and the panel based polls will make their adjustments, and others might change their weightings a bit, and polls will group up together a bit more on one side or the other of that relegation line. Finding out how many said they voted Green/Reform, and how many of those say they intend to do so again in the GE will be important. But not in terms of knowing if the Conservatives could win, they clearly can't if the polling is anywhere near true. It's now not even a question of if it's a Canadian Conservative style collapse, it's going to be a collapse, but is that collapse closer to 1988's collapse or 1993's wipeout. Only one or two pollsters attempt Local Election Polls, so there is no basis for adjusting in line with the actual LE results. I don't think I have seen any of YouGov Opinium or Survation producing a Local Elections poll, and nobody weights by LE vote. In 2019 we had the European Elections 6 months before the General Election, for which most pollsters did produce polls, and this was probably a factor in producing largely accurate polls at the 2019 General Election. Based on past experience, we might expect Net Equivalent Vote shares around Lab 35%, Con 25%, LD 21%, in line with the normal pattern of the LDs getting about 8-10 points more than their Westminster polling figure. This will be mostly at the expense of Labour, who have underperformed Westminster polls by anything up to 10% in the past (including last year), but this is a variable figure. The general pattern is that both Lab and Con underperform when doing well, but get a NE|V more in line with their polling figures when they are polling badly. |
|
|
|
Post by James E on Mar 28, 2024 11:27:17 GMT
pjw1961 Re the Polling on rejoining the EU, above. The figures on this have become fairly steady by individual pollster, but show some significant 'house effects'. Deltapoll are the most frequent pollster to ask this and their figures are clustered around a net 55/45 for Rejoin. However, recent figures from YouGov, R&W and Omnisis are all around 60/40. I think these are likely to be more accurate, due to YouGov's panel, and the fact that R&W do not weight by 2016 referendum vote. As I mentioned last week, Deltapoll's tables consistently show them finding more 2016 Remainers than Leavers, and some odd figures for their 'Remain' cross-break with the Conservatives' support holding up at almost its 2019 level. Meanwhile, YouGov have provided tables for their latest poll (Lab 40, Con21) which include their Brexit 'Hindsight' tracker. The figures for this are much as they have been for about the last 6-9 months, working out as net 62/38 for 'Wrong to Leave'. (The figure for 'Wrong' is normally about 3 points higher than YG's figures for those who want to 'Rejoin'). Breaking this down into age groups, the sample for over 65s is 36/64 for 'Right to Leave' so barely changed from the position in 2016 (35/65 for Leave per YG's analysis) The under 65s divide as 71/29 for 'Wrong', a 19% swing on 2016, when those under 65 voted overall 52/48 for Remain. This pattern should add 0.6% per year to the lead for 'Wrong', or Rejoin. d3nkl3psvxxpe9.cloudfront.net/documents/TheTimes_VI_240327_W.pdf
|
|
|
|
Post by James E on Mar 27, 2024 23:23:47 GMT
Redfield Wilton Wales 20mph limit really damaging Welsh Labour... Labour lead by 33% in Wales. Lowest Conservative vote % EVER in our Welsh polling. etc..... "Ever'' ?!!- Redfield and Wilton have been polling Wales since April 2023, and since then have done a total of 12 Welsh polls. en.wikipedia.org/wiki/Opinion_polling_for_the_next_United_Kingdom_general_election#WalesAnd their GB polling only started in 2020, so they have yet to test their findings in an actual election. |
|
|
|
Post by James E on Mar 27, 2024 20:06:46 GMT
YouGov have been asking their Brexit 'Hindsight' tracker less frequently recently, but when last asked in February, the current Labour VI divided 82% for the decision to Leave being 'Wrong', 10% for 'Right' and 8% Don't Know. This is barely changed from 2019, despite Labour having won over considerably more Leave voters, though Leavers continue to favour the Tories overall.
On the question of Re-joining: the most recent YouGov figures I have for Labour supporters are for their 2019 voters, as of a few months ago. These were 81% Re-join, 8% Stay Out, with 11% Don't Knows. The 'Stay Out' figure will be a bit higher for current Lab VI, but probably only by about 3% (in line with the '2019 voters v current VI' movement in the Hindsight tracker). As has been noted before, Leave voters switching to Labour is strongly correlated with their giving up with supporting Brexit.
The other relevant piece of poling from YouGov is the question of whether Brexit is 'done' on 'not done'. Labour voters divided by 60% to 20% for 'not done', with 20% Don't Knows.
|
|
|
|
Post by James E on Mar 26, 2024 19:39:47 GMT
Re steve 's "Could the refukers win in Blackpool? Old , white, badly educated and brexity they'll have few better opportunities." youtu.be/b4g4-Ga5EFw?si=vKdocI2KW0a6TFcv Reform UK has not yet come near to matching the vote share achieved by UKIP in GE2015 in any by-election. The most recent examples are: Mid Beds 3.7% (v 15.4% UKIP2015) Tamworth 5.4% (v 18.5% UKIP2015) Kingswood 10.4% (v 14.8% UKIP2015) Wellingborough 13% (v19.6% UKIP2015) Rochdale 6.3% (v18.8% UKIP2015) So it would be a significant milestone for them if they can match or beat UKIP's 2015 vote share of 17.3% in Blackpool South. That might be a reasonable expectation for their share of the vote, given that their current polling average of 12.3% is right in line with what UKIP achieved at GE2015. The Kingswood, Wellingborough and Rochdale by-elections all happened when their polling average was 10%, while for Mid Beds and Tamworth they were around 6%. |
|
|
|
Post by James E on Mar 25, 2024 15:57:13 GMT
graham"I well understand the point, but suspect the Tories will have a fair degree of success in seeing off the LD challenge in BlueWall seats." While the LibDems appear to be slightly down in 2019, their performance in terms of seats in England rests on their vote share relative to the Tories rather than its absolute value. And looking at YouGov's detailed figures (5-poll average) for Southern England and ABC1 voters, the Tories do not currently seem to be having much success: South England Con 25% (-30) Lab 39% (+17) LD 13% (-4) ABC1 Voters Con 19% (-24) Lab 48% (+15) LD 10%
|
|
|
|
Post by James E on Mar 25, 2024 14:18:43 GMT
I will be surprised - albeit pleasantly so - if the Tories end up with much fewer than 200 MPs. I suspect that this is due to your trust in UNS. If you use this as a model, it would take around a 16% swing, so a Labour lead of 20%, for the Tories to go under 200 seats. However, all MRPs and other detailed and proportionate models show seats changing hands in far greater numbers than UNS does. The obvious recent example was the YouGov MRP which put the Tories on 169 seats, in the context of an overall swing of 12.6%, and a Labour lead of 13.5%. Using this as a guide, I would suggest that the Tories would need to be within 10% of Labour in the overall vote in order to retain 200 seats or more. yougov.co.uk/politics/articles/48371-yougov-mrp-shows-labour-would-win-1997-style-landslide-if-election-were-held-today |
|
|
|
Post by James E on Mar 25, 2024 13:58:03 GMT
Re That Times report :'Reform now ahead among male voters'
This might be true, but it may well be just normal variation. The previous two YouGovs showed the Conservatives 2 & 4 points ahead of Reform in their male sub-sample. The Tories are actually faring somewhat "less badly" with C2DE voters, males voters and in the North of England than they are with ABC1s, women and those in the South & Midlands of England.
As for '12% (only one in 8) switching directly to Labour':
This is true of this one poll, but the range of 2019Con to Lab switching is from 12-19% in the past 6 YouGov polls, averaging 15%. 1.5% of 2019Lab voters are shown as switching the other way, which by my calculations would produce 6.2% of 'Direct Swing'.
This can be compared to YouGov's own analysis of GE2019, when their research found that 11% of 2017Lab had switched to the Conservatives, and 4% of 2019Con had moved the other way. This would have produced 2.7% 'Direct Swing' from Lab to Con at the last election.
|
|
|
|
Post by James E on Mar 24, 2024 11:28:31 GMT
Further to my last, EC is predicting 27.3% Con vote in P-mouth North (more than halved, but still more than what Lab managed last time). Con have never polled less than 36.7% in this constituency (that was 2001). Not sure if boundaries have changed but I'd take their prediction with a large pinch of salt. It's very white working class and rough AF. They're also predicting Fareham and Waterlooville (Braverman's hovel) to go red - 48% to Con's 47%. As if. Those 48%/47% figures are probabilities rather than vote shares. EC's current prediction for Braverman's seat is Lab 31.1%, Con 30.8% - this is a swing of 22.2% with the Tory vote being halved. www.electoralcalculus.co.uk/fcgi-bin/calcwork23.py?seat=Fareham+and+WaterloovilleThis is very much in line with what current polling shows. I have repeatedly posted 'regional' figures here showing the Tories faring very badly indeed in the South of England, and more generally shedding votes in a proportionate or 'greater-than-proportionate' way. YouGov's current average is a 22% swing for the South of England, as a whole. And there are very good reasons for expecting the Tories to do considerably worse in the South than 1997 or 2001, which was the closest that Lab and Con have ever been. In 1997, the overall vote shares for the 'combined South' (SE+SW+E) were: Con 40% Lab 29% LibDem 24% But these are the current 6-poll averages from YouGov or Opinium - with comparisons to GE1997: YouGov Lab 38% (+9) Con 27% (-13) LibDem 12% (-12) Opinium Lab 37% (+8) Con 30% (-10) LibDem 13% (-11) So an 11% Con to Lab swing compared to 1997 per YouGov, or 9% with Opinium's adjusted figures, in the context of average GB Labour leads of 24% with YG and 16% with Opinium. Electoral Calculus's figures are (I think) based closely on the MRP they did with FON a couple of months ago, but look to me consistent with YG, Opinium, and other southern polling such as Survation's constituency polls in Clacton or Godalming & Ash. As you have mentioned Portsmouth North - a little observation on EC's figures for the Portsmouth seats. They show a 23% Con to Lab swing in Mordant's Portsmouth North, but just an 11% swing in Labour-held Portsmouth South. In both cases, they show the Con2019 vote being roughly halved, but being reduced by more (even on a multiplied basis) in North. This is consistent with the detailed pattern of other polls and MRPs. Dave Re Opinium. It's not really correct to describe Opinium's figures as a 'prediction'. They adjust for is the high levels of 'Don't Know' responses from previous supporters of unpopular parties. In most polls, these are treated the same as those who answer that they will not vote. So when 25% of Con2019 voters say they Don't Know, but only 10% of Lab2019, you get headline figures with a far higher turnout of Lab2019 than Con2019 - which is what YouGov (for example) do. Opinium's new methodology (also used by MiC) is to upweight the positive responses to reflect the actual 2019 turnout. This reduces the Labour lead by about 6 points. They stated when they introduced this that they expected the difference to narrow towards a General Election, but this is dependent on the' Don't Knows' coming off the fence, which has yet to happen. I would now expect their figures to stay lower than others all the way to the election. Opinium do not re-allocate all DK's back to the party they voted for last time. Their adjustment reflects the positive responses they they get from those Con2019 who do state an actual voting intention, which of course means that the Tories' VI gets a boost. Because of this, their adjustment is, in effect, quite similar to that applied by ICM in the 1990s and early 2000s, and which was the main reason for ICM to be the only pollster to call the 1997 result accurately. (ICM really did 'predict' respondents' VI, adding half of each party's DKs from the previous election to those who stated an actual VI) |
|
|
|
Post by James E on Mar 22, 2024 19:25:23 GMT
With today's 47/24 poll from WeThink, the 7-poll average currently shows the largest average Labour lead (22%) since Feb 2023. www.politico.eu/europe-poll-of-polls/united-kingdom/Current Politico averages, compared to this time a year ago are: Lab 45% (-1) Con 23% (-5) ref 12% (+6) LD 9% ( 0 ) Green 5% ( 0 ) SNP 3% (-1) The average Lab VI (of 45%) is down by 1 point from where it stood exactly 12 months ago, but also up 1 point from the previous 2024 average of the 7 most recent pollsters to release figures (WeThink, Techne, MiC, YouGov, Delta, R&W, Savanta) |
|
|
|
Post by James E on Mar 22, 2024 18:41:31 GMT
Brexit bonus alert ( no really). Of c course Brexit is an economic and social disaster for the UK, stealing our freedoms and democratic mandates and diminishing our economy and credibility . But there's a slim glimmer of hope. As European union member educational fees were exempt from VAT. Now we've taken back control it's entirely possible for Labour to impose vat on fee charging elitist educational establishments such as Eton, Harrow or Westminster. Looks like the parents of Tarquin's and Phoebe's will have to pay a bit more to provide their offspring an unfair bonus. youtu.be/Dlthhp270zM?si=Ee-zpQVdeqE1sFVtThere is nothing to prevent an EU Member State from imposing another separate tax on something exempt from VAT per EU Law. The UK did this in 1994 when Insurance Premium Tax was introduced. A 15% Private Education Tax would probably raise more revenue than 20% VAT, due to the way that VAT can be recovered on costs. So it would be an option for any government minded to keep the option of Rejoining open, and/or a way of annoying anti-EU obsessives. ( steve - thanks for confirming my point below: it is perfectly possible for the UK to tax Private Education AND rejoin the EU, if we choose to do so. And in fact it would (IMO ) be preferable to create a separate tax rather than applying VAT to Private School fees, so as to prevent Private Schools from re-structuring to take advantage of the VAT exemption for private tuition, or the reduced VAT rate for long stay accommodation). |
|
|
|
Post by James E on Mar 22, 2024 15:56:15 GMT
leftieliberalThanks for the MiC analysis of Reform UK voters' second preferences. I find these surprising - particularly when compared to the YouGov figures for where they are drawing support from by past vote. Average Ref UK VI 13.5% (per last 6 YouGovs - those giving a Voting Intention ) Con2019 28% Lab 2019 1.7% LD 2019 1.5% Remain2016 2% Leave 2016 29% (and close to 0% of those who did not or could not vote in the referendum)
|
|
|
|
Post by James E on Mar 21, 2024 12:54:56 GMT
On the subject of 'squeezed LibDem voters' there were some good examples from the effects of this from the 2001 election. There were many seats where Labour had been in 3rd place going into 1997, but where they either won, or came very close in 1997. This produced unusual results in GE2001 when the LDs then went into the election in third place. The effect was a boost for Labour, but also a noticeable boost for the Conservatives, well above the overall 1% increase which they managed for GB as a whole (while Lab was down by 2.5%) So in Shrewsbury, where Labour had won from their place in 1997, the movements in 2001 were Lab up 7.6%, and Con up 3.4% In Dorset South, where Labour came an unexpectedly close second in 1997 after starting in third, the figures were Lab up 6.1%, Con up 5.5%. But, by happy coincidence, the best example of all for this unnoticed phenomenon came in Hastings. This had been Labour's most long-range gain in 1997 where they had taken just 16% of the vote in 1992. In 2001 Lab were up by 12.7%, but the Tories also substantially increased their vote share, by 7.4%. The LDs' vote fell considerably in all these constituencies. And after factoring out the national trend, with Lab down, and the Tories slightly up, it looks to me like the overall split of the votes they lost went roughly three-quarters to Labour and one-quarter to the Conservatives. en.wikipedia.org/wiki/Hastings_and_Rye_(UK_Parliament_constituency)en.wikipedia.org/wiki/South_Dorset_(UK_Parliament_constituency)en.wikipedia.org/wiki/Shrewsbury_and_Atcham_(UK_Parliament_constituency) |
|
|
|
Post by James E on Mar 21, 2024 12:04:48 GMT
jimjam Where's your evidence that 'fewer LD voters will vote tactically Labour (indeed some may vote tactically Tory agin Lab) than Lab voters for the LDs'? Both YouGov (data from Dec 2019) and Electoral Calculus (data from Sept 2023) suggest otherwise. I think JimJam is broadly right on this. There was an interesting illustration of how LD voters divide local to me in the 2021 Cambridgeshire mayoral election. This was, I believe, unique, in that only Con, Lab and LDs fielded candidates, and the LDs came third, resulting in their second preference being taken into account. So it was only the LDs' voters being re-allocated. The split of 61,885 LDs was: Lab 61% Con 23% No second choice 16% en.wikipedia.org/wiki/2021_Cambridgeshire_and_Peterborough_mayoral_electionThis works out as 73% to 27% for those to made a choice, and I think this is a reasonable guide as to how LDs might divide where squeezed. So Lab might benefit by net 50% (75/25?) of a squeezed LD vote, whereas in the scenario of a Labour vote being squeezed the LDs might get more like 90% (95/5?). However, this only comes into play where voters' first choice is not the party which they believe to be in contention, and the very large net movement in VI from LD2019 to Lab must be considered first. |
|
|
|
Post by James E on Mar 21, 2024 11:47:49 GMT
Yougov Labour lead at 25 points in latest YouGov poll for The Times CON 19 (-1) LAB 44 (=) LIB DEM 9 (=) REF UK 15 (+1) GRN 8 (+1) Fieldwork 19 - 20 March Tables are here: ygo-assets-websites-editorial-emea.yougov.net/documents/TheTimes_VI_240320_W.pdfSome interesting details: Reform UK now lead in the Leave cross-break. This one shows: Ref 23% Con 22% Don't Know 19% Lab 15% Won't Vote 11% And the Conservatives VI by region is particularly interesting: whereas they ranged from 25% in Scotland to 55% in the South of Midlands of England in 2019 , their support has now effectively been flattened. Wales 24% (-12) South of England 22% (-33) Midlands 19% (-36) North of England 17% (-22) London 16% (-16) Scotland 16% (-9) To be fair, this Wales figure is on the high side, as their 6 poll average for Wales is 19%, compared to 20.5% for the whole of GB (and typically around 26% in the Midlands and South). |
|
|
|
Post by James E on Mar 21, 2024 10:17:31 GMT
Re Con/Lab/LD 3 ways. I think the number of councillors is a good indicator of the number of likely committed activists on the ground. In some of these seats the LDs extra resource will have been in use for many months and in some cases enough to cement them as clear challengers. Labour may well come from 3rd place in a few to be the main challenger to the incumbent Tory MP but imo the LDs will maintain that position in most. As polls narrow (and they will we just don't know by how much) the Mid Beds type scenarios will become fewer and fewer. In 2019 there were many seats where the LDs overtook Labour to take second place. I fully expect that to be reversed this year. LD support is not really advancing at all - if anything it has declined - and it may well be that the 2017 and 2015 results offer a better guide as to the underlying strength of parties in a given constituency. On this, I tend to agree with Graham rather than JimJam. Large movements between Lab and LDs have been a feature of just about every General Election for the past 30 years, resulting in a very high proportion of seats in the South of England where the tactical position has changed, and in most cases, changed repeatedly. The pattern of GE2019 is no more set in stone than that of 1997 or 2001. Looking at movements since 2019, YouGov's averages currently show 40% of 2019LD voters switching to Lab, and 3-4% of Lab2019 switching to LD. This would be sufficient on its own to reverse a result of LD30%, Lab 20% in 2019 to LD19%, Lab 31% now. As a rule of thumb, Labour are now likely to finish ahead of the LDs wherever they were within 15% of them in 2019. So to take examples from the discussion above, I would expect this to happen in Chelmsford (where the gap is only 7%); Didcot and Wantage with a 16% gap is a borderline case. But my own constituency of Cambridgeshire South with a 28% gap is one where the LDs are clearly better placed, even though Labour did come second here in 2017 and 2015. The 'Blue Wall' polls we have had from R&W and very recently from MiC are relevant here, as they contain a high proportion of LD targets, both short-range ones and those which might now appear to be 3-way contests. For the latter, there is some useful comment on this recently on Electoral Calculus (apologies for a link repeated from a few days ago). www.electoralcalculus.co.uk/blogs/ec_lib2seats_20240226.htmlTo Add: I've linked detailed analysis on the 1997 Election before so won't do so again, but the pattern then was for the largest Con to Lab swings happened in the safest Tory-held seats, and not in the close or medium-range ones which Labour had targeted. |
|
|
|
Post by James E on Mar 20, 2024 21:40:17 GMT
domjg You have quite a dilemma voting in Didcot and Wantage. Electoral Calculus (link below) currently predicts it as: Lab 30.4% LD 28.7% Con 26.8% YouGov's MRP at the turn of the year was: LD 32% Con 31% Lab 24% MiC have shown it as a Lab gain, but applying their overall voting movements to the EC notional figures shows: Con 30% Lab 29% LD 27% www.electoralcalculus.co.uk/fcgi-bin/calcwork23.py?seat=Didcot+and+Wantage(In your position, I would wait for the pre-election YouGov MRP, which is on a far larger scale than the one done by them 3 months ago, and the recent FON/EC one)
|
|
|
|
Post by James E on Mar 20, 2024 19:48:55 GMT
...."More in Common have published a poll of the thirty-nine seats which the Conservatives won in 2019 and saw the largest total swing towards Labour and the Liberal Democrats in the 2017 and 2019 elections." Fieldwork 20 Feb -2 Mar 2024, sample of 1,005. Lab 33% (+13) Con 32% (-19) LD 20% (-5) Ref 10% (new) Green 5% (+3) Swing of 16%, compared to 13% as an average of the last 3 MiC polls, which show Con -17, Lab +9. In proportionate terms, the Conservatives' vote losses are identical to those in MiC's 3 most recent GB polls: as 32/51% =28/44.7%. ...... www.moreincommon.org.uk/our-work/research/blue-wall-collapse/I was particularly taken with their observation that, if a 25% tactical voting element was added to the swing noted, Theresa May's seat of Maidenhead would fall to LAB. Pretty unlikely in reality, I feel, but a pretty damn sobering thought for CON, nonetheless. What makes these figures all the more remarkable is that More in Common show the smallest average Labour lead of any pollster so far this year. They adjust in the same way as Opinium, and have recently shown around Lab 42%, Con 28%. The 32% Tory VI in this poll is really no worse than their 28% for GB as a whole - and a great deal better than almost any other polling company would suggest. But that does not stop them finding some very unlikely sounding Labour gains. Even without tactical voting, their figures suggest Labour taking Didcot and Wantage, Mid Sussex, Runnymede and Weybridge, South Devon, St Neots and Mid Cambs, and Woking - all from 3rd place in 2019. |
|
|
|
Post by James E on Mar 20, 2024 12:35:35 GMT
I've found some figures for the MIC 'blue wall' poll. "More in Common have published a poll of the thirty-nine seats which the Conservatives won in 2019 and saw the largest total swing towards Labour and the Liberal Democrats in the 2017 and 2019 elections." Fieldwork 20 Feb -2 Mar 2024, sample of 1,005. Lab 33% (+13) Con 32% (-19) LD 20% (-5) Ref 10% (new) Green 5% (+3) Swing of 16%, compared to 13% as an average of the last 3 MiC polls, which show Con -17, Lab +9. In proportionate terms, the Conservatives' vote losses are identical to those in MiC's 3 most recent GB polls: as 32/51% =28/44.7%. It's an interesting choice of seats, in that it is not strictly-geographical (though very South-East based per their maps, and with 8 in London) and based purely on the swing away from the Conservatives - and note that overall the LDs were in second place with 25% in 2019 to Labour 20%. Given that the largest swings against the Tories in 2019 were in strongly Remain-supporting seats, this also provides further reason to doubt those strange Deltapoll figures for Remainers that I highlighted yesterday. www.moreincommon.org.uk/our-work/research/blue-wall-collapse/ |
|
|
|
Post by James E on Mar 19, 2024 21:17:21 GMT
I have looked further back into this, and the same pattern is apparent. While YouGov show ConVI as around 12% with 2016 Remain voters, Delta have the Tories' support with this group as averaging 18.7% across their past 12 published polls. This is almost at the level it was at the 2019 Election - measured as 19% per YG's analysis or 20% per Ipsos. But these two companies show similar levels of support for the Tories in their 'Leave' samples, 35% per YouGov, 38% per Delta. Assuming good faith all round, then it strikes me the two companies somehow are selecting their samples differently, hence they are polling a different sort of remain tory voter. Yougov make extensive use of permanent panels of respondents, I dont know if Delta do. Problem with a permanent panel might be that while it was initially constructed to meet certain criteria of being representative, Brexit threw a coach and horses through that because it was an issue cutting across previous selection groupings. Not at all guaranteed that a pre brexit panel would now be representative of the diversity of the leave/remain groupings. YouGov have a panel, and Delta do not. The huge advantage that YouGov have with their long-established panel is that all members who were part of it in June 2016 will have been asked at that time how they voted in the referendum (and the same for the 2019 election). But they do also add new members, and they are then asked about past votes on joining. This should be far more accurate than asking many years later. It is the sample of each poll (normally 2,000 respondents) which needs to be representative by past votes and other demographics. This is a small subset of the entire UK Panel which is of course many times larger than this - YG say that globally they have a panel of 24 million. yougov.co.uk/about/panel-methodologyAnd an Anthony Wells article from summer 2019 about false recall of the 2017 election - but note that the linked tables also show anomalies by 2016 vote. yougov.co.uk/politics/articles/24349-false-recall-and-how-it-affects-pollingAs an aside - those who were around on UKPR1 about 3-4 years ago will recall that at one time we had a regular contributor who somehow 'forget' that he actually campaigned for Leave rather than Remain in 2016. |
|
