Recent Posts

Chichester is a good example of a seat which seems to be hard to predict under current polling and with tactical voting.

In 2019, it was Con 58%, LD 21%, Lab 16%.

The conventional wisdom is simply to look at the previous election, or at notional result for changed seats. And the LDs were second, albeit a very distant second. However, it is worth remembering that there have been big movements in Voting Intention since 2019, and polls typically show 30-40% of 2019 LDs switching to Labour and only around 3-4% of 2019 Lab switching to the LDs. So Labour may well be better placed of the two challengers. 

This is reflected in two of the three Models or MRPs that I have looked at . YouGov's is the most promising for the LDs as it shows:
Con 32% (-26), LD 28% (+7), Lab 21% (0).

However, Survation and EC both have the LDs in a distant third. Survation shows:
Con 36% (-22), Lab 31% (+15), LD 18% (-3)

And EC's nowcast is:
Con 27% (-31), Lab 38% (+17), LD 15% (-6).

Of these three, I think Survation is likely to be the most accurate, though their Con VI is a bit higher than I would guess, even in the context of the Tories reviving to around 28-29% nationally. However, as a rider to that, I think Survation's March MRP badly understates the LDs in those seats where they are in close contention. 

One of the reasons for this is the EC finding for the LDs overall performance in their 'Top 63' and their 'next 44' seats. We've discussed this before, but the simple point is that the LDs have an overall polling lead in their 'Top 63' seats, but trail in the 'next 44' where EC found overall shares of Lab 30%, Con 28%, LD 21%.

www.electoralcalculus.co.uk/blogs/ec_lib2seats_20240226.html

But to me, the bigger reason to expect that Labour will fare the better of the two challenging parties is the precedent of GE1997. The LDs fared well in seat gains in 1997, but the seats where they improved their overall share of the vote were those where they were in 2nd place from 1992, AND were within 20% of the Conservatives from the previous election, so as to seem the clear challenger. In seats (such as this) where the LDs were in second place, but more than 20% behind, their vote actually fell a little, while Labour's rose by 9 points, much in line with their overall GB performance. 

I've looked into how seats such as this behaved in 1997 - that is seats with a large (20%+) Con majority, LDs in second and Lab 3rd. It looks to me like there is a clear pattern:

Where the LDs were 14 or more points ahead of Labour from the previous election, they generally stayed ahead of Labour (so would have been the better tactical choice). The only major exception to this I have seen was Hastings where Labour moved up from 19 points behind the LDs to 7 ahead, winning the seat in the process. The other exception was Cambs NE, but that seat was seeing a large unwinding of the LDs previous tactical vote as they had held the seat up to 1987. 

However, wherever Labour started out less than 14 points behind the LDs from 1992, they finished ahead of them in every single example I have seen. This is why I think YouGov's modelling of such seats may prove to be wrong: the tactical benefit for the LDs came into play where they were seen as really being in contention, but not in seats where they started in a distant second place. And I think that the '14-point LD/Lab Gap' may provide a useful rule-of-thumb now, given that the overall polling movements seem set to be very similar to 1997 : the Conservatives falling back sharply, Labour up by about 10 points, and the LDs down fractionally. Incidentally, while I have not checked the 1997 results exhaustively, in those which I have analysed of this type ( Con more than 20% ahead, LD 2nd, Lab 3rd), the average 'net improvement' for Lab in relation to the LDs was 15 points, with most being in a range from 10 to 18. 

If similar dynamics can be applied now, it means that the LDs are the better tactical choice in places such as Wimbledon (18% gap to Lab), South Cambridgeshire (26%), Didcot & Wantage (16%), or Woking (just over 14%, but in any case the LDs are less than 20% behind the Tories there, so it isn't 'safe').

On the other hand, Chichester with a 5% 'LD to Lab gap' looks to me likely to now have Labour ahead. And the same is true of seats such as Reigate (3%), St Neots (8%), Chelmsford (8%). 

www.electoralcalculus.co.uk/blogs/ec_lib2seats_20240226.html
www.dannydorling.org/wp-content/files/dannydorling_publication_id1318.pdf

pjw1961
I've gone for a similar figure: 28.5%. But my seats prediction for the Tories of 131 is lower than yours by 27.

To answer LL's point, I think we are likely to see some movement in the next week or so, if it is really going to happen. 2017 is not the only time when the polls have moved during the campaign. One rather forgotten example was 1997, when the polling average moved from Lab 52.5%, Con 28.5% in late Feb/early March to a final position of Lab 48%, Con 30% in the final eve-of-election polls. So a 6-point net movement. I think we are quite likely to see a similar movement in the next 5 weeks, but more Con gain than Lab losses.

The 2019 election saw little change in the Tories' polling lead, because both Con & Lab rose by similar amounts. The first polls of the 2019 campaign (and the subsequent movements were):
Con 38% ( went up by 7)
Lab 28% (went up by 5)
LD 15.5% (went down by 4)
BXP 10% (went down by 8)

Many thanks for that. 

Looking at these figures and the notional figures for the new boundaries, there may be more close contests than I had expected. However, I think the result is 95%+ certain in 15 of the 18 seats in Northern Ireland. The three that I think may change hands are:

East Belfast, Alliance challenging DUP
Foyle, SF challenging SDLP
South Antrim, UUP challenging DUP (but Alliance may also have a slim chance)

If all three of these changes happen, the seats total would be:
SF 8 (+1)
DUP 6 (-2)
All 2 (+1)
SDLP 1 (-1)
UUP 1 (+1)

Predicting seat totals under First Past The Post is difficult in any UK election, but especially so in current circumstances. Vote shares are somewhat easier, and on these I think that the likely outcome is at the lower end of the current polling range; so my best guess is that Labour are going to win by about 15 points. On the other hand, it seems clear to me that with the Conservatives' proportional losses, and corresponding Lab & LD gains in their heartlands, we are likely to see a record-breaking number of seats changing hands. I have mentioned before how the UNS-based target for a majority for Labour (or for the Conservatives) shifts significantly at most elections. I expect a record-breaking shift, such that the apparent target for a Labour majority will fall from 12% per the 2019 results to around 3%. As I mentioned yesterday on the main thread, I would expect Labour to gain around 17 seats for each 1% of swing from the Tories they achieve in the popular vote - which is a lot more than the 10-11 shown by UNS, and also larger than the 14 shown by YouGov's most recent MRPs.

The main reason for this is the proportional swing, but tactical voting against the Tories is an additional factor. The Conservatives win few constituencies on much less than 40%, and their current losses suggest that they will fall below that 40% threshold in many formerly safe seats. The LDs performance in seats varies in relation to the gap between them and the Tories, and they are likely to prosper with the Conservatives' share being at a record low.

In vote shares, I am guessing around Lab 43%, Con 28%, LD 11%, Ref 7% (?), Green 6%. My attempt to put this into seats is:

Lab 421
Con 131
LD 51
SNP 22
PC 4
Grn 2
NI 18 ( SF 8, DUP 6, Alliance 2, SDLP 1, Alliance 2, UUP 1)
Speaker 1

The larger parties' figures must be taken as very approximate. A single point swing would increase or reduce the size of the majority by about 35 seats. Given the range of current polling, a more cautious approach might be to say that Labour are likely to get between 375 and 450 seats. My figures are around the middle of that range.

I have taken a similar approach to Scotland. The difference in seats between an SNP lead of around 7 points and a similar lead for Labour is around 35 of Scotland's 57 seats. I now think that the most likely outcome is a narrow lead for Labour in both votes and seats, but where SNP and Lab are close, a single point swing can shift 4 or 5 seats between the two. Wales is not too hard to predict, as all polling put the Tories close to wipe-out, even though Labour's vote share there is not hugely better than in 2019 - PC will do well, as will the Greens in their English targets. In Northern Ireland, I anticipate just a few seats changing hands, that being East Belfast from DUP to Alliance; SF may also take Foyle from SDLP, and UUP could gain Antrim South from the DUP. And turnout is likely to be significantly lower than in 2019 - probably around 60%.

It's a useful tool, but it does produce what may seem surprisingly large numbers of Lab and LD seats.  

If you use UNS rather than Electoral Calculus then this produces an implausibly low figure of just 10 or 11 Lab gains for each 1% swing. This isn't going to happen, but I thought I should mention that as a starting point. 

www.electionpolling.co.uk/battleground/targets/labour

However, UNS does not work for large swings, and  Labour in fact  achieved significantly better at 14 seats per 1% swing in 1997 (147 gains divided by 10.5% GB swing), when UNS would also have produced just 10. I am very confident that figure will be exceeded this time.

I've done some analysis of the six 'proper' MRPs we have had in the past 12 months, two each by YouGov, Survation, and FON*. In each case, I have compared the number of Labour gains shown, divided by the Con to Lab swing. So this is a measurement of the efficiency of the way labour are gaining and the Tories losing voters.

The 'Lab gains per 1% swing' are :
YouGov  14.7, and 14.1 (but the latter figure is reduced because of their 'unwinding' adjustment**)
Survation 15.7 and 17.6
FON  16.5 and 15.9

All of these show a higher movement in seats than GE1997, although in some cases not by much. The average across all of these is 15.75 seats per point of swing. So with, say, a 14-point swing (Lab wins by 16.3%) they would get close to 420 seats.

The figures produced by Electoral Calculus's seats calculator are just a little bit higher than this average, but well within the overall range of these MRPs. From various figures I have entered based around a 15% labour lead, their 'Lab gains per 1%' are around 16.8. So it would generate an extra 13 Labour gains in the example I have quoted above, with a 14% swing. 

I have not mentioned Con or LD seat losses/gains, but to be brief - Survation's MRPs produce what I think is too low a figure for the LDs, but I think both YG and FON may be a bit generous to them. I think that YG and FON are wise to use samples and modelling specific to LD seats and targets, whereas it looks to me like Survation do not. However, it is a tricky issue as to how to model for the substantial LD to Lab movement shown by polling in seats where both may potentially be competitive. As for the Tories - their losses per 1% swing against them range from 16-20, growing larger the lower their vote share goes (down to around 20-23%, where would be few left).

What I haven't mentioned in this is the additional factor of tactical voting. While the benefit of this in the Lab/Con marginals is, in my view, exaggerated, it will add a few more seats and it is clear from both by-elections and polling such as that by Ashcroft that a significant proportion (50%?) of LD and Lab voters are willing to cast their vote tactically for each other. Because of this, and above all the continued proportionate pattern of the Conservatives' vote-losses, Electoral Calculus could well be about right in showing close to 17 Labour gains for each point of swing. 


* See:
en.wikipedia.org/wiki/Opinion_polling_for_the_2024_United_Kingdom_general_election#Seat_predictions
I have ignored the 'Stonehaven' MRP with a total sample only 2,000, and would also ignore Focaldata, as their previous MRPs have contained basic errors, IMO.


* see YouGov's explanation of their 'unwinding' adjustment here:
yougov.co.uk/politics/articles/49061-yougov-mrp-labour-now-projected-to-win-over-400-seats

It's worth having a read of joeboy 's comments on Northern Ireland. I don't think there are any DUP/SF contests, or at least those seats that are close are already held by SF, and will probably move to becoming safer for them.

The two seats that might change hands are Foyle, where SF is challenging the SDLP, and East Belfast, where Alliance are challenging the DUP. Other than those, I'd not expect any changes. The DUP may well be able to claw back some of the support they have lost to TUV, just as the smaller parties are often squeezed under FPTP in England, Scotland and Wales. 

So, in seats, it is likely to be

SF 7 , SDLP 2
or SF 8, SDLP 1 

...and

DUP 8, Alliance 1
or DUP 7, Alliance 2

To beat the 10.5% swing which Blair got in 1997, Labour need to be 9.3% or more ahead in the popular vote.

As every single poll we have had for the past 20 months shows a Labour lead of 10% or more, it seems reasonable to expect that "unprecedented record-breaking swing".

(The last sub-10% lead was a 4% in Sept 2022 from Kantar, who now appear to be defunct/relegated)

I had a go at working out what kind of percentage votes could produce those seat figures. 

Using a largely proportional swing model, and factoring in tactical voting, I get figures which look pretty much like current polling -

Lab 45%
Con 23%
LD 11%
Ref 9%
Green 6%
Others, inc SNP & PC 6%

The crucial figure for the sort of figures Steve has guessed is the Tory VI. If it really is barely half of the 45% they received in 2019, they would be likely to go well below 100 seats. But at around 28%, they could expect in the region of 125. 

"Increasing turnout is also about people earning enough money and having affordable essentials so that they aren’t burnt out juggling three jobs and have time to take in and reflect on the news (including having the time to read more than just one right wing newspaper like the Guardian. Maybe even some left wing stuff!). "

What's your favourite  left-wing news source,  c-a-r-f-r-e-w ?

A bit of further comment on that Observer article, linked again below.

www.theguardian.com/politics/article/2024/may/11/labour-gains-leave-brexit-swing-majority-tories-starmer-victory-election

To be fair to the authors, there is at least some acknowledgment from Robert Ford that the pattern of movement has been evident in polls since 2021. However, there is still a vast divergence between the Voting Intentions of 2016 Leavers and Remainers. The current YouGov 6-poll averages are:

Remain
Lab 59% (+11)
Con 11% (-8)

Leave
Lab 25% (+11)
Con 32% (-42)

So the difference between Remainers' and Leavers' voting intentions is still an enormous 55 points, albeit that it has reduced significantly from 90 points at the 2019 General Election.  Labour's gains the same with Remainers and Leavers, at +11%. So it is the Tories' losses with Leaver voters which are fuelling the very different swings. Note too that these figures are for respondents who intend to vote and state a voting intention (as opposed to be undecided). As I mentioned a couple of days ago, Remainers are far more likely than Leavers to tell YouGov that they are 'certain' or '9 out of 10' to vote at the next GE, by 82% to 65%. 

But it is worth looking back on the structure of the Conservatives' 2019 vote as per YouGov's Dec 2019 analysis. YouGov found that 74% of 2016 Leavers who voted in 2019 backed the Tories, but only 19% of 2016 Remainers. Given that there were approximately equal numbers of Leavers and Remainers in the electorate, these percentages can also serve as good rough approximations for the proportions of Remainers are Leavers in the Conservatives' 2019 vote: that is, around three-quarters of them were Leavers, less than 20% were Remainers, and around 5% were non referendum voters.

Current polling shows the Tories share of the vote being roughly halved since 2019 from 45% to 23% - a loss of 22 points. And it is a loss of 26 points per YouGov, whose figures I have given above. A 'Uniform National Swing' on the Remain/Leave axis produces an impossible outcome on these figures, as it would imply a Conservative vote share of less than zero with 2016 Remainers. But it seems to come to a great surprise for many in the media that this isn't happening.

The Conservatives' 14% share in Scotland is actually quite good for them in comparison with how they are doing elsewhere.

R&W's GB figures give the Tories just 21% - their last three polls have all been around that figure. That's 24 points down on GE 2019, so those who think UNS is still a good predictor might expect them to be on 1% in Scotland as they got 25% there in 2019, or 8% in London (from 32% at GE2019). Similarly, they 'should' also still be above 30% in the South and Midlands of England.

The reality is that per most polls they are on around 15% in Scotland, 16-20% in London, but also only around 25% in the South and Midlands. 

Those are stunningly bad figures for the Conservatives.

Survation's 'no prompt by party names' methodology normally produces some of the highest Con VI figures, and also the lowest for RefUK. Before this, the lowest Con VI they have shown since Oct 2022 was 26%. This is also their highest Labour lead for 18 months. 

Taken with the Techne and YouGov figures we have had in the past 24 hours, we now need just one more for '4 in a row'. With fieldwork of 9-10 May, these figures from Survation are the most recent that we have. 

c-a-r-f-r-e-w

"Read the other day that Starmer almost resigned after Hartlepool but was talked out of it, and then Truss happened."

Rather a lot happened in the 16 months between the Hartlepool by-election in May 2021 and Truss's brief stint as PM in Sept-Oct 2022. And as a matter of fact, Labour under Starmer moved ahead in the polls just 7 months after Hartlepool at the start of Partygate. This lead was more than 10 points at times in Summer 2022.

I can't help feeling that either you or your source is trying to airbrush away the advances that Labour made when Johnson was PM. If we take that 10-point Con lead in early May 2021, then two-thirds of Labour's subsequent advance to the 20-point lead (as seen for the past 18 months) happened pre-Truss; the polls in the last month or so of Johnson's time as PM showed Labour on average 11 points ahead.


On the more general point about Labour's ability to win under Corbyn - this poll from August 2023 seems to me the best relevant evidence..

"Westminster Voting Intention (Corbyn as Labour Leader):

LAB: 36% (-10)
CON: 35% (+7)
LDM: 15% (+4)
RFM: 6% (=)
GRN: 5% (=)
SNP: 3% (=)

Via
@moreincommon_
Aug 2023."
Changes w/ Regular VI.
12:42 pm · 18 Aug 2023

The big difference is the adjustment for 'Don't Knows', as applied by Opinium, MiC, JL Partners and (I think) BMG. Pollsters who do this typically show Labour around 16 points ahead, but bear in mind that this has been adjusted downwards by 4-7 points. YouGov have also used this for their two large MRPs this year , and these have shown Labour leads of 14 and 17 points .

There is another group of pollsters who do not prompt for Reform UK and this results in a higher Conservative VI by several points (but often a relatively high Lab VI, too). The pollsters who do this are Survation, Savanta and (I think) Deltapoll.  These pollsters average around an 18-point Labour lead.

If we look at the remaining pollsters, who don't adjust for DKs, and do prompt, that leaves YouGov, We Think, R&W, Techne and Ipsos. This group is currently averaging around a 23-24 point Labour lead. 

You Gov's "18-48" poll may be either an outlier or a halo effect from the various elections last week; we don't know yet, as it is easily the most recent poll. The JL figures above are almost a week old. The gap between the 'adjusters' and 'non-adjusters' has been fairly consistent at about 6 points. My best guess is that when we eventually get an election, the outcome will be closer to that of the 'adjusters' and/or the likes of Survation.