New in the
@SavantaGroup
/
@SavantaComRes
#covid19
daily tracker:
- Boris Johnson’s approval rating is now at -1%. It was at +19% four days ago.
- Overall government approval is now at -2%, dropping 16 points in a day.
More:
We all mistakes. This was a silly but significant one. It will almost certainly not be the last public error by a pollster this year.
If Savanta finds out we've made a mistake, we will own it and fix it - being open is vital to building trust in our industry.
Upon further inspection, there is an error in our West Midlands Mayoralty poll from last week, where we have incorrectly applied the 2019 past vote weight, using past vote targets for whole of the West Midlands region as opposed to the West Midlands Combined Authority area. (1/4)
New poll from
@SavantaComRes
finds just 11% of the public and 16% of Conservative voters can correctly name Penny Mordaunt when shown a photo of her.
Sunak 66%
Truss 33%
Mordaunt 11%
I make this:
- Largest lead for any party in any poll since May 2021
- Lowest Conservative vote share in any poll since September 2019 (Lib Dems were on 22%, Brexit Party in 16%)
- Largest Labour lead in any poll since February 2013
Went back to
@ColU_Official
for the first time since I lost Dad today. I’m sure he’d have been smiling after a scrappy 1-0 win, wearing his old shirt and scarf. This one’s for you.
🚨NEW Scottish Westminster VI for
@TheScotsman
📈First Savanta poll since GE announcement shows Labour lead over the SNP remaining static.
🌹LAB 37% (=)
🎗️SNP 33% (=)
🌳CON 17% (=)
🔶LD 7% (=)
⬜️Other 5% (-1)
1,067 Scottish adults, 24-28 May
(change from 3-8 May)
Thread below on our methods, record & plans for the future:
@Savanta_UK
is an independent and well-established polling company that purely seeks to accurately measure public opinion, not promote or suppress one party over another, and any implication otherwise is entirely false.
My two cents on the national service policy is that it’s an obvious pitch to Reform UK voters, in that it’s incredibly similar to Reform UK policy i.e. it looks like it’s come from a minor party that can say anything because it knows it will never have to implement it.
And specifically on Penny Mordaunt:
- two people thought it was Adele (their other open end responses were high-quality, so it didn't *look* like a joke)
- one respondent said "Penelope Mordaunt (gorgeous)"
A 14pt lead from us and a 27pt lead from YouGov on the same day. Confused? Don't be.
There could be a number of unknown factors leading to differences (and there probably are), but there are two we do know about: how we treat Reform UK, and how we treat undecided voters. 🧵
It's the day before polling day, so time to stick my neck on the line & make some polling and perhaps non-polling predictions.
There's going to be lots of noise and hot takes that age badly, but there are a few we can already see ahead of time that are worth challenging now.
🧵
This is going to do the rounds now. It's an interesting hypothetical aiming to understand what voters think they might do in a number of different Brexit scenarios. As we know, voters are terrible predictors of their own behaviour, and there's a long way to go in any GE campaign.
Westminster voting intention...
if "the deadline for the UK to leave the EU has been extended beyond the 31st of October 2019":
LAB: 27%
CON: 26%
BREX: 20%
LDEM: 18%
GRN: 4%
via
@ComRes
, 16 - 17 Oct
The change poll-on-poll is minimal, but this is the lowest Conservative vote share we've recorded since mid-September.
For context, it hasn't been lower than 26% since the final poll of the Truss tenure.
Very pleased to have accurately called the London Mayoral race for
@centreforlondon
. Still, plenty to learn from voter behaviour and lots of hard work from the
@Savanta_UK
team between now and the general election
Realised that I didn't tweet last week's
@SavantaComRes
VI for the
@Independent
, so here it is for completeness. Health warning given the fw dates!
🌳Con 29 (-4)
🌹Lab 42 (-2)
🔶LD 12 (+3)
🎗️SNP 3 (=)
🌍Gre 4 (+1)
⬜️Other 9 (+1)
2,272 UK adults, 22-24 July
(chg from 21 Jul)
If Labour’s majority is massive, but they lose some seats to Greens, lose Islington North, and don’t beat SNP in terms of Scottish seats, I could see some spin saying Lab have underperformed.
If this happens, it’d be the worst takeaway I’ve ever seen at a general election.
This is quite comfortably our largest ever Labour vote share and largest ever Labour lead - even back in the ComRes days, don't think
@PigsAndPolling
saw anything like this.
Sunak keeps harping on about Jeremy Corbyn at
#PMQs
- does anyone actually care?
Latest
@SavantaComRes
data found that 46% of the public say attacking Starmer for Labour's association with Corbyn is unfair.
All
Fair 27%
Unfair 46%
2019 Con
Fair 40%
Unfair 38%
The strongest prediction I’ve got is that “Portillo Moment” will become the most incorrectly-and-over-used phrase of the night, mostly because there is no assumed heir to Sunak in the way Portillo was to Major.
Not sure I have a huge hot-take on this
#ITVDebate
, but it does seem like Sunak's answering audience/moderator questions, and then Starmer is being made to answer Sunak's own questions which feels...unfair?
A 21pt lead for the SNP last June in
@Savanta_UK
’s polling for
@TheScotsman
has now completely eroded.
The decline started well before Sturgeon’s resignation/Yousaf’s election, but has seemingly accelerated since.
🚨NEW Scottish Westminster VI for
@TheScotsman
📈Labour tied with SNP for first time in a Savanta poll.
🎗️SNP 35% (-3)
🌹LAB 35% (+1)
🌳CON 19% (+2)
🔶LD 6% (-1)
⬜️Other 4% (=)
1,002 Scottish adults, 6-11 October
(change from 9-14 June)
Of course, the beauty of Savanta (and others) being
@BritPollingCncl
members is that our methods are transparent and our tables are available on our website for people to make their own minds up.
Lib Dems may well be overlooked, and depending on Reform’s performance, the threshold which they need to win seats in Con-Lib races reduces. The fact Ed Davey’s chance of being LOTO isn’t zero is mad, and if they can beat 2005’s seat total (62) you’d think they’d be very pleased.
Obviously what this election campaign needs is more polls and
@Savanta_UK
are only too happy to oblige: our latest for the Sunday
@Telegraph
will be released later this evening and well, let's just say there's some movement.
Savanta do not include Reform UK in our initial VI prompt, we include them (and the Greens) in a secondary sub-prompt once a respondent has clicked 'another party'.
That may not account for all of the Savanta 8pt / YouGov 13pt disparity, but it could account for a lot of it.
This weekend I saw Kalush Orchestra, Verka Serduchka, Loreen, Sam Ryder, Roger Taylor, Netta, DaðI Freyr, Duncan Laurence, Hannah Waddingham and Graham Norton. Liverpool was an amazing host and
#Eurovision
is simply epic.
EXC: Labour MPs are being urged to launch a leadership challenge to
@Keir_Starmer
amid anger over the two-child benefit cap.
A group of former advisers wanted
@Ed_Miliband
to help topple him, but the ex Labour leader has made clear the plot is 'nuts'.
The
@SavantaComRes
MRP poll released yesterday at a
@LabourList
fringe showed that *all 632 GB constituencies* support public ownership of the railways. Average support is 73.1%, high of 77.4% (Dunbartonshire East) and a low of 69.7% (Harrow East)
NEW: Labour commits to renationalisation of the railways. Shadow Teansport Sec
@LouHaigh
: “Labour in power will bring our railways back into public ownership as contracts expire.”
We are confident in our figures as they stand and won’t be bullied into making methodological changes by politicians, simply because they don’t like our results.
Half also say the party has changed for the better since Corbyn stepped down as leader.
All
Better 47%
Worse 16%
Same 25%
2019 Con
Better 50%
Worse 11%
Same 31%
2019 Lab
Better 57%
Worse 23%
Same 14%
Most interesting part of this for me is Starmer wins comfortably on many of the metrics we tested, yet only narrowly 'won' the debate.
Sunak definitely had upper hand at times last night, and although our figures suggest he 'lost', he probably outperformed expectations.
🚨🚨🚨
NEW: Starmer beats Sunak in televised debate overnight poll
Who won the debate:
Starmer (44%)
Sunak (39%)
Don't Know (17%)
1,153 UK adults, 4-5 June
Today we held our first
@SavantaComRes
PMQs focus group, where we asked a group of 2019 Conservative voters, but who wouldn’t vote Conservative now, what they thought of the Prime Minister, Leader of the Opposition and Partygate through the prism of
#PMQs
. Here’s what we found…
The SNP remaining on 46% is notable. I would not have been surprised to see Labour perhaps eat into that SNP vote share, but the fact that they did not all-but confirms that they're fishing in the unionist vote pool rather than a specifically centre-left one in Scotland.
Rishi Sunak's actions this week may well go down as the greatest act of electoral self-harm in modern UK political history.
So much of the Conservatives' electoral strategy was pinned on older, Reform UK-curious supporters.
Sunak may have upset this group of voters.
🚨SNAP POLL for
@Telegraph
Two thirds of UK adults, including three in five Conservative voters, say it is unacceptable that Rishi Sunak left D-Day commemorations to do a TV interview.
All
Unacceptable 68%
Acceptable 21%
DK 11%
Con 2019
Unacceptable 61%
Acceptable 31%
DK 9%
It was conducted via online panel, in English, which will likely have some impact on how truly representative it is. Older Muslims, those for whom English isn't their first language, and those less likely to be online, aren't likely to be represented in this poll.
Just 57% of 2019 Conservative voters say that they'd vote Conservative tomorrow.
Switch to Labour - 16% choosing to vote Labour Switch to Reform UK - 8%
Switch to Lib Dems - 6%
Undecided - 11%
Even if all undecideds voted Con, their deficit to Lab would remain around 14 points.
If Nigel Farage and Richard Tice would like to commission Savanta to get an accurate assessment of their party’s current performance, they are welcome to.
Important to throw a little cold water on the change displayed here. Irrespective of it being well within MoE,
@Savanta_UK
's last three Labour leads:
15-17 Sept: 20pts
22-24 Sept: 14pts
29 Sept - 1 Oct: 19pts
If anything, it seems last week was an outlier.
No idea what fuelled it in May, but we saw lot of baseless rumours about Susan Hall winning London Mayor, and I wonder if we’ll see similar again.
I’d also expect the rumour mill to be in overdrive in Bristol Central, Chingford and Woodford Green, Islington North and Clacton.
Great to speak to
@lewis_goodall
yesterday morning on why we're all a bit guilty of assuming the Conservatives will avoid complete wipeout in 3 weeks time just because, well, they're the Conservatives, even though the data tells us it could be bad. Very bad.
🔊"We're in a position, pollsters & commentators & journalists, where we just can’t quite comprehend the scale of defeat that could well be coming to the Conservatives"
@ChrisHopkins92
spoke to
@lewis_goodall
about the political & mathematical problems facing Sunak this election
Most said that the threat of Corbyn pushed them towards voting Conservative in 2019, as opposed to the lure of Boris Johnson. This group didn’t necessarily believe in the magic winning touch some Conservatives think the Prime Minister possesses.
Reminder on the PM's current favourability rating. His net score in December was -27 - the lowest it's ever been - and we'll be launching the January wave this weekend.
The polling picture – despite what you may have heard – has been remarkably consistent in this regard. There are margins of error, and some disparity, but the industry knows what’s at stake.
I’ll be sharing further analysis throughout election night – happy to respond to Qs.
Latest
@Savanta_UK
Westminster VI shows Labour with a 17pt lead over the Conservatives.
But the real story is in our best PM ratings, where Starmer leads Sunak by 7pts - his largest ever lead over Sunak, and the fourth consecutive time he's beating the incumbent PM.
🧵(1/5)
There is an increasing trend, and a positive one at that, of some pollsters urging caution around their own findings, particularly those that show change beyond the margin of error. Be interesting to see if this one follows that pattern...
We were the most accurate polling company at the last London & West Midlands mayoral elections, with an average error across both polls of zero percentage points when it came to Reform UK’s performance.
In May, just a third (31%) of Scots said they wanted
#indyref2
"in the next two years" - driven primarily by those intending to vote Yes. We'll find out with
@TheScotsman
this week how things have changed and if more favour Sturgeon's intention to hold one by the end of 2023.
When asked who ‘won’ PMQs, all participants decided that the Prime Minister, who came out fighting and looked less dejected than last week, beat Keir Starmer – surprising, given the week the PM has had.
Ordering beer at 9 in the morning feels *a little* wrong, but when it's to support businesses like
@TheWansteadtap
stay afloat as their landlords won't give them a rent holiday during the lockdown, who gives a shit.
Will be some interesting seats to look out for, namely Cabinet big-hitters that would expect to be frontrunners for leader (assuming Sunak loses and resigns).
Imagine it’ll be these seats where the most rumours from the count come from. Portsmouth North feels ripe for this.
I think it'd be a bit lazy to just assume that all of YG's 'extra' Reform UK voters would switch to Con if YG didn't prompt Reform UK - in fact, YG's own research indicates they won't.
But it may at least make our Con vote share disparity (Savanta 29%, YG 20%) a little closer.
Discussing tax rises is fine, but against the backdrop of a pandemic, swing voters have sympathy for the government needing to take difficult decisions. Yet Partygate rumbles on, and some swing voters just will not entertain the idea of voting for Johnson again.
Labour’s vote has always been good in these early declarations, but we’re looking for efficiency this election – piling up votes where they’re needed, and not in safe seats like these (and like they did in 2017).
However, the increase in no. of MRPs makes me a bit uncomfortable, because they’re so open to misinterpretation (as is basically any form of modelling).
Differences in vote share by fractions of % point could make a big diff to seat outcomes, as many pollsters have already said.
Does everyone remember that hilarious Censuswide ‘poll’ that claimed Labour would win by a huge margin, except it was unweighted and included those too young to vote? Good times.
We've made a methodology tweak in our latest London polling which will, at least in part, account for the significant difference between this latest London poll for
@MileEndInst
and our previous for
@centreforlondon
.
There is electoral evidence from previous general, local and by-elections of the Liberal Democrats justifying their inclusion in our main prompt. This evidence does not currently exist for Reform UK.
Other British Polling Council members will have their own legitimate methods, and many share in great detail how they've come to their results.
All will have data tables online, including at , for anyone to make their own mind up on our methods.
This isn't great reporting of poll findings and change versus an actual post-election poll where being undecided is, of course, impossible. Compares apples with oranges.
In summary, there isn’t anything in
@Savanta_UK
's research (including fundamentals, long-term trends or game changers) or anyone else’s, that suggests anything other than the Conservatives getting a proper pasting on Thursday.
Starmer has led Sunak in this metric - which seems to divide opinion over its value (fwiw I think it could be more valuable than VI in some sense) since May '23, with this lead now his largest ever. We'll be tracking this weekly from now until the GE.
🚨NEW Best PM rating
📈Starmer's biggest lead over Sunak ever in Savanta polling
🌹Starmer 44% (+4)
🌳Sunak 30% (-1)
◻️Don't know 27% (-2)
2,239 UK adults, 24-28 May
(Changes from 17-19 May)
Tables are on our website but the 'Other' breakdown is...
Omilana 5%
Rose 2%
London 2%
Fox 1%
Gammons 1%
Hewison 1%
Kurten 1%
Binface 1%
Obunge 1%
Fosh 1%
Reid 1%
All others <1% each.
Also worth saying that if the response to this isn't too horrendous then we'd certainly look to repeat, but if academics, peers and the like have any *constructive* feedback, then of course I'm all ears.
As per our BPC obligations, tables with a methodology statement are online for those interested to make their own minds up on its value, but internally we felt it was important enough to attempt and were satisfied enough with its quality to release.
We did use in-field quotas, and the poll is weighted to correct for sample imbalance, but it's unlikely this will entirely make up for the above.
Still, it's far less self-selecting, and is far more representative, than some of the surveys of Muslims we've seen recently.
We currently include smaller, UK-wide parties under a sub-prompt, based on their recent electoral performances.
We do this for accuracy, because we believe this prevents the artificial over-inflation of smaller parties in national polling.
Worth being very clear that not all predictive models are this bullish on Labour's chances - pretty sure
@BNHWalker
's SOTN for New Statesman shows Lab and Con basically neck and neck at a GE fought on current boundaries
Our recent MRP model implied that Peter Bone's seat of Wellingborough would switch to Labour at a General Election.
🌳CON 35% (-27)
🌹LAB 47% (+21)
🔶LD 7% (-1)
➡️Reform UK 7% (new)
⬜️Other 5% (-1)
7,409 GB adults, 22-25 September 2023
When exit poll drops, so may a few jaws. There are 2 broad scenarios that could play out:
1) Exit poll points to historically large Lab maj that, although many polls and seat models have pointed to, still feels difficult to comprehend, as I've said:
)...
Great to speak to
@lewis_goodall
yesterday morning on why we're all a bit guilty of assuming the Conservatives will avoid complete wipeout in 3 weeks time just because, well, they're the Conservatives, even though the data tells us it could be bad. Very bad.
Clearly our MRP published yesterday jointly with
@ElectCalculus
for the
@Telegraph
is startling, but perhaps it shouldn’t be.
Thread time!
TLDR: I’d argue it's a natural extension of current national polling and political pictures.
🧵
🚨NEW Savanta MRP for
@Telegraph
📈Lab projected for 382 majority
📉Cons fighting with LDs to be Official Opposition
Seat forecast
🌹Lab 516 (+316)
🌳Con 53 (-319)
🔶LD 50 (+42)
🎗️SNP 8 (-40)
🌼PC 4 (+2)
⬜️Other 0 (-1)
17,812 UK adults, changes to last GE notional results
Now candidates are finalised, many pollsters move to a ballot prompt method, where all candidates standing in a given constituency are prompted to survey respondents.
Savanta has done this in the past, and intends to do so again, as we have long planned.
Important context around this one:
1) we know it’s Tory members and not just Cllrs that matter. Doesn’t make this not worthwhile and interesting
2) our Cllrs poll in 2019 was closer to the final result that some members polling