I’m delighted to have joined
@TheEconomist
as a Political Data Scientist this week. I’ll be working on polls, forecasts and everything political data. Especially pleased to have joined in the year that 4billion people worldwide will go to vote - lots of exciting things coming up!
YouGov asked people to rate Sunak and former PMs
(% good minus % bad)
Blair -4
Major -4
Brown -8
Sunak -25
Cameron -25
Johnson -36
May -37
Truss -76
Surprising how high Blair is and how low Cameron and May are
What's going on in Birmingham, Bethnal Green, Bradford and Leicester?
Labour is directing activists to some of its safest seats in the country, constituencies where Indian, Pakistani and Bangladeshi people make up >30% of the population:
The "poor people don't drive" narrative is such a London bubble thing.
The pink areas on this map are where people use public transport more than driving.
With today's polls included, the Conservatives are at their lowest average voting intention since Liz Truss was Prime Minister
Lab - 44%
Con - 24%
Lib Dem - 11%
Greens - 7%
Reform - 6%
Very pleased to see our swing model published today.
If an election were held tomorrow, our median estimate of the result would be:
Lab - 372
Con - 198
SNP - 32
Lib Dem - 21
Plaid - 2
Green - 1
Historically, we're moving into the part of the Parliament when the polls begin to move in the Government's favour.
Labour still outperforming the "winning opposition" average:
We’ve spent two years talking about how uniform swing doesn’t work when one party is 20pts ahead… not enough time worrying about how it might break the MRPs
🚨 NEW: Tory minister Steve Baker has been accused of "giving up" after jetting off to Greece instead of campaigning in his constituency
[
@daveburke12
]
New
@IpsosUK
polling explores public attitudes to Lee Anderson's defection to Reform UK.
1. First things first. He's not that well known. Just three in ten familiar. More than his new boss mind - and it does rise to 40% amongst 2019 Cons.
Had a little look at differences in British and American public opinion. By far the biggest difference is belief in God (81% in US, 48% in GB)
Followed by:
- Private/public ownership
- Abortion
- Death penalty
Tory MPs returning to Westminster today wistfully talking about Sunak fighting a ‘92 election: green shoots, risky Labour. Not so much “change” as “keep a-hold of Nurse”, as one minister put it. Analysis featuring Hezza and
@nickmacpherson2
We're practically at the inflection point where the 1987-1992 Parliament went from big Labour leads to being very close. This time the Conservatives don't have an unpopular Prime Minister they can depose:
Data increasing looking like a pincer of Reform, Labour and Lib Dems taking Con19 voters with the geographic distribution that does maximum damage. Will have more on this next week hopefully! 🦕
One reason the Conservatives might do worse than uniform swing implies: tactical voting. In BES data, I found that the number of voters considering more than one progressive party is at its highest in recent history
The tone of many responses to this speech feel really off. Fine to poke fun at council candidates speaking about foreign policy, but the insinuations that this is extremism are islamophobic
Very pleased to see our US Presidential MRP in
@politico
today - one year out:
Trump 45% (298 EC votes)
Biden 46% (240)
RFK 8%
West 1%
Libertarian 1%
Green 1%
Head to head:
Trump 48% (292)
Biden 49% (246)
@StackStrat
12th Oct - 3rd Nov, N=15,205
Funny looking back a month to "the Conservatives need to make gains every week" and it having been so much worse than that:
Lab 41% (-4)
Con 20% (-4)
Ref 16% (+5)
Lib Dem 11% (+2)
Green 6% (-)
Interesting dynamic this election that every day the government doesn't make gains increases the likelihood of Labour victory.
Average of 5 polls since the election was called, with change from pre-announcement:
Lab: 44 (-)
Con: 23 (-)
Ref: 12 (+1)
Lib Dem: 10 (-)
Green: 6 (-1)
niche but the Greens are actually nightmarish from a modelling POV because they love winning loads of votes in places with none of their core demographics
These numbers are so small that I find it striking in the opposite direction. The U.K. has a higher % Chinese residents than Japan despite being 000s km further apart
One of many stupid reasons Westerners think Japan is super homogeneous is that they go there and see a bunch of Asian people, not realizing that a lot of the people they're seeing aren't actually Japanese.
Government ends the week on an average voting intention of 24%.
The UK govt has had a VI of 24% or lower three times in the history of polling:
- October 2022
- May/June 2019
- May/June 2009
Even if you account for Leicester and Slough having unique local factors, it looks as though wards with higher British Indian populations swung away from Labour and towards the Conservatives in the local elections
Quite the line from Starmer on the Tories:
They're building a sort of Jeremy Corbyn-style manifesto where anything you want can go in it, and none of it is costed... load everything into the wheelbarrow, don't provide the funding, and hope nobody notices the money isn't there.
Potential for ecological fallacy here. In polling data I’ve seen, non-Muslim voters in more Muslim areas have swung just as far (sometimes further) than Muslims from Lab->Green
Some more
#gaydata
. Overall, LGB+ women outnumber LGB+ men by around 12:10.
In West Lancashire, this ratio is 17:10 while in Westminster and City of London it is 4:10.
Local election vote share in Johnny Mercer's Plymouth Moor View constituency:
Lab 46% (+4)
Con 29% (-12)
Other 10% (+1)
Green 9% (+5)
Lib Dem 6% (+2)
Changes with 2022
What do they have in common? All of them are over 30% Indian, Pakistani or Bangladeshi. As you go down the list, Labour are sending activists to many other areas with significant South Asian populations: Bolton, Ealing, Huddersfield, Luton, Oldham
@patricksturg
There's no evidence trans rights advocates are disproportionately men or their opponents are disproportionately women... Male MSPs were more likely than female MSPs to vote against the GRR bill in Scotland, for example. Polling consistently shows men hold more transphobic views.
Very interesting paper published this week, showing how the adoption of PR in New Zealand has meant fewer centre-right governments and greater income redistribution since 1996 (from
@Maciej76718052
and
@pierzgal
)
@Labour4PR
@robfordmancs
not a fun take but I’m not sure election year local elections tell you anything about the fate of the government
2019: Con -1,330 (Con+48 MPs)
2017: Con +563 (-13)
2014: Con -236 (+24)
2009: Lab -291 (-91)
2004: Lab -464 (-48)
2000: Lab -574 (-6)
1996: Con -607 (-178)
The geography of subjective social class. People in Scotland, south Wales, parts of the north of England and London are significantly more likely to identify as working class, once you control for income and education. (inspired by
@TomLath36874686
)
Average Labour lead of 19%. There have only been four periods where the opposition has had such a large lead:
🔴 Oct 2022 - now
🔵 Jun - Sep 2008
🔴 Oct 1993 - Apr 1997
🔴 Mar - Apr 1990
🔵 Mar 1968 - Jun 1969
Labour currently leads by ~20%. Here's how that compares to opposition parties at this point before historic elections:
* 1997: 25%
* 1970: 18%
* 2010: 17%
* 1964: 11%
* 1974: 6%
1987: 6%
1959: 5%
2015: 3%
* 1979: 1%
2019: 0%
* Went on to form a govt
But there are some seats suggesting serious weakness. Here are those with the largest notional Labour majorities:
1. Birmingham Ladywood (73%)
2. Bethnal Green & Stepney (63%)
3. Bradford West (61%)
4. Birmingham Perry Barr (48%)
5. Leicester South (46%)
6. Blackburn (3%)
The seats selected suggest more than one factor at play. Some have substantial Muslim populations, where Labour struggled in the 2024 local elections and where there are a number of Independent or Workers Party candidates standing on pro-Palestine platforms
Biggest Conservative majorities that Labour is targeting:
1. Tamworth (43%)
2. Cannock Chase (43%)
3. Great Yarmouth (41%)
4. Warwickshire North & Bedworth (39%)
5. Redditch (39%)
6. Amber Valley (37%)
7. Leicestershire North West (37%)
8. Harlow (35%)
Brilliant new poll from
@RedfieldWilton
shows yet more growing support for PR 📈
54% back changing the voting system while only 12% are opposed 💪
Join us to make sure we
#MakeSeatsMatchVotes
!
On this, there isn't a single constituency in Britain where more than a third of people think equality for gay people has gone too far (estimated using BES data+MRP)
In May 2022, 18% of 2019 SNP voters said that equality for gay people had gone too far, compared with 25% nationwide. Meanwhile, 37% said equality had not gone far enough:
@democlub
Wow, since I posted this someone clearly updated the Green Party candidates:
Labour - 573 (+15)
Green (GPEW + Scottish Greens) - 551 (+147)
Lib Dem - 483 (+12)
Reform UK - 469 (+1)
Conservative - 368 (+14)
Workers Party - 162 (+6)
SDP - 98 (-)
SNP - 56 (-)
Independent - 54 (-)
I had a look at where Labour is sending activists from its “how to help” webpage. I found 256 constituencies where activists are being directed - seats Labour thinks are in play. They include all but one seat in Scotland so from here I am referring to England and Wales only.
Could Conservative "Don't Knows" be causing the next big polling error?
Looking at
@StackStrat
data, the decision to weight a survey before vs after excluding "Don't Knows" is the difference between a 16% and 21% Labour lead:
Not sure we've properly adjusted to the possibility of a result like this. Extremely close to a total Tory collapse (talking <40 seats) or a Reform breakthrough in former Tory safe seats. Kinda surprised none of the MRPs so far have shown it...
Labour leads Reform (!!) by 23%.
Highest Reform %.
Tied-lowest Conservative %.
🇬🇧 Westminster VI (19-20 June):
Labour 42% (-1)
Reform 19% (+1)
Conservative 18% (–)
Liberal Democrat 11% (-1)
Green 5% (–)
SNP 3% (–)
Other 1% (–)
Changes +/- 14-17 June
For balance, most likely Lib Dem gains:
1. Cheltenham
2. Guildford
3. Esher and Walton
4. St Ives
5. Winchester
6. Wimbledon
7. South Cambridgeshire
8. Lewes
9. Carshalton and Wallington
10. Cheadle
Most likely Reform seats in our model:
1. Doncaster North
2. Barnsley North
3. Rawmarsh and Conisbrough
4. Rotherham
5. Pontefract, Castleford and Knottingley
(but bare in mind this is a swing from the Brexit Party - Reform's coalition appears quite different)
Focus groups are important for understanding voters' feelings + reasoning around issues, but presenting them like this (i.e. presenting qual as if it is quant) is actively unhelpful imo. We have representative surveys to ask whether voters prefer Sunak or Starmer
RED BOX PODCAST
🏴 Focus Group
2019🟡SNP voters, now undecided give their verdict on Sturgeon, Yousaf, Sarwar & Ross.
@jamesjohnson252
is in the chair.
🎧LISTEN
And if forced to pick, who do they prefer Sunak or Starmer?👇
On the whole, the seats chosen are very bullish. They include 174 Conservative seats from 2019 (notional). That is every Conservative seat with a majority under 28% and over half of seats with a majority up to 32%:
Brighton and Hove's vote for PCC was:
Paul Richards (Labour) - 22,053 votes
Katy Bourne (Conservative) - 10,308 votes
Jonathon Kent (Green) - 9,042 votes
Jamie Bennet (Liberal Democrats) - 3,198 votes
This is not the full results for Sussex PCC
But they also include seats with significant Hindu and Sikh British Indian populations. These groups have been trending away from Labour and were a particular weak point in the 2023 and 2022 locals, when Cons made big gains in Leicester, Slough and Harrow
Even if you account for Leicester and Slough having unique local factors, it looks as though wards with higher British Indian populations swung away from Labour and towards the Conservatives in the local elections
Big problem for Muslims dissatisfied with Labour that the electoral system+geography give them strikingly little leverage. Few Muslims will go Con, Greens don't pose a challenge in any highly Muslim seats (Bristol C. is 6%). Only 1 of Lab's top target seats in E+W is >20% Muslim
We'll find out soon whether that YouGov poll is an outlier, but with it included in the average you get:
* Lowest Con VI since October 2022
* Biggest Labour lead since February 2023
* Highest Reform share ever
* Lowest Lib Dem since March 2023
* Reform in 3rd place for 1st time
I think this is evidence that Reform has *some* real support but given they stood in so few wards, you'd expect the ones they did stand have much higher support than nationwide (i.e. closer to 20% than 10%). Not so true for the other four, which stood in most wards
Aggregate Result in the 197 Wards where all 5 Major Parties Stood:
🌹 LAB: 35.7%
🌳 CON: 22.9%
🔶 LDM: 13.8%
🌍 GRN: 10.8%
➡️ RFM: 9.6%
Looks like the Reform polling numbers are real, even if they can't convert it into seats.
First results of the day are through and they're for the West Midlands PCC 🚨
In Solihull, there were 33,159 votes for the Conservative's Tom Byrne and 21,184 for Labour's Simon Foster
Meanwhile in Wolverhampton, Foster got 31,972 votes while Byrne received 20,863
Every week since Sunak became PM it’s felt like The Narrowing is finally about to happen, then you check a polling average and they’re a point lower than the week before
Since around March, Labour's lead has been bigger than any opposition which has gone on to lose. So a Conservative win would be a historic turnaround. But... stranger things have happened:
Explore our interactive tool to discover how many people with advanced and higher education qualifications stayed in or returned to the towns they grew up in 📘
Striking polling for
@MakeVotesMatter
:
How well do you feel the political system is working for the people of the UK?
Very badly - 24%
Somewhat badly - 25%
Neither well nor badly - 25%
Somewhat well - 16%
Very well - 4%