Dylan Difford @Dylan_Difford Twitter profile

Pinned Tweet

Dylan Difford

3 months

Flow of the vote, 2019-24, provisional version (will wait for the BES data to be released to make a final version, plus some deeper cuts).

28

120

402

Last Seen Profiles

@Pauucdz

@PMBENAMIEL

@tdgarden

@iuefsosyoloji

@1MikeNewton

@antonella77771

@bamswrld

@MikeMadera2

@missmhelly

@i_ke_san

@iraq_country_Ar

@College010

@ros_abdelrahm

@Nikobad_boy

@bokeplokalmalam

@ObisikeChi24962

@T99437654

@72me_2nd

@T99437654

@60balconies

@UnNhibitedLibra

@44photography

@Evydoresumao

@stw_pdg

@satoshivision75

@stw_pdg

@BeniKabula59119

@Abdulaz78871036

@OhKayArt

@MBorn1974

@kalala_honore

@blackbyrdevents

@Bluecheeseso

@bottom51515

@R_Buchanan_77

Dylan Difford

@Dylan_Difford

7 months

@janhopi Also clearly set in Times New Roman, which was released in the 1930s.

5

4

2K

Dylan Difford

@Dylan_Difford

9 months

Going through the YouGov MRP data to add some tactical squeezing. If just one third of Lab-LD-Grn voters in England and Wales vote tactically for the strongest party, the result changes to: Con 69 (-100 on MRP) Lab 463 (+78) LD 70 (+22) Nat 28 Grn 1 More to come...

58

352

1K

Dylan Difford

@Dylan_Difford

9 months

Where are 2019 voters today? With each square 100k voters in the colour of their 2019 vote (white = too young), this is how polls currently suggest they intend to vote this year. Am sceptical Labour have gained that many non-voters, but key story is scattering of Tory vote.

94

187

777

Dylan Difford

@Dylan_Difford

4 months

Another entry in the increasingly common series "I've just remembered that I'm concerned about the level of executive dominance in the British political system for reasons totally unconnected to my side definitely losing power".

Jenrick: I share voters’ frustrations but backing Reform means a Labour dictatorship

Ex-immigration minister says the success of Nigel Farage’s party threatens to kill off conservatism

www.telegraph.co.uk

16

188

770

Dylan Difford

@Dylan_Difford

5 months

Wow! Can't believe that opinion polling turned out to be more reliable than the reckons of some bored journalists and party people.

6

67

685

Dylan Difford

@Dylan_Difford

4 months

You can write a lot about this election, but it really does all come back to the public's assessment of the government being a succinct 'we hate you and nothing works'.

14

121

686

Dylan Difford

@Dylan_Difford

5 months

Over the past four-and-a-half-years, the 14m people who voted Tory in 2019 have splintered in various directions, with just 4 in 10 now saying they'll vote for the party - less than under Truss. Here's how they've said and currently say they'll vote, each block 100k voters.

19

100

492

Dylan Difford

@Dylan_Difford

4 months

Still seeing it going around, but a reminder that "x *needs* a swing of y to get a majority of one" is actually bollocks and people parroting it are likely to give you sub-par election analysis. For reference, here's a non-exhaustive list of why it's wrong:

11

108

463

Dylan Difford

@Dylan_Difford

2 months

Feels like the scale of what happened in the South isn't always fully appreciated - nothing like 1945 or 1997, 1906 only comparable heartland defeat for the Tories. Also notable how much more southern the current Lib Dems are than in their early-2000s peak.

23

150

431

Dylan Difford

@Dylan_Difford

5 months

Blackpool South by-election - 26.3pt swing is 7pts greater than the seat in MRP averages, 10pts higher than national polls and, most crucially, larger than my graphic was designed for.

18

78

416

Dylan Difford

@Dylan_Difford

3 months

A lot of people mocking, but she is right in some respects - the scale of the Conservative defeat would have been different had Sunak bravely decided to run a pro-Truss campaign.

Will Hazell

@whazell

3 months

EXCLUSIVE @Telegraph : Liz Truss has blamed the scale of the Conservatives’ general election defeat on a decision by Rishi Sunak to “trash my record”

263

35

150

8

37

331

Dylan Difford

@Dylan_Difford

4 months

If we didn't have polls, the last few weeks would have included commentary about how the national service policy was winning back older voters and no doubt wishcasting around a narrowing race.

Matt Chorley

@MattChorley

4 months

Should we ban polls during an election campaign?

35

3

12

7

39

311

Dylan Difford

@Dylan_Difford

3 months

If only there had been some kind of proposals, maybe around 2011, for a single-member preferential voting system that would have a) minimised the effects of split votes and b) likely have produced larger Tory majorities in 2015 and 2019?

Neil Henderson

@hendopolis

3 months

MAIL: Boris: Britain CAN still sweeve Starmergeddon #TomorrowsPapersToday

496

79

222

14

52

300

Dylan Difford

@Dylan_Difford

3 months

Frankly think there's too much focus on individual seat results from the Exit Poll, which isn't really what it's designed for.

10

31

270

Dylan Difford

@Dylan_Difford

2 months

Bit of a big one, but it's a weekend treat, the most popular party in each county at each GE since 1885. And by county I'm using 1974 boundaries for consistency, because I know it's something everybody is totally normal about.

16

48

272

Dylan Difford

@Dylan_Difford

4 months

The PR accelerationist's dream result - a 300-seat majority on 38% of the vote and an absolute clusterfuck of confusion for 2nd, 3rd and 4th place.

Sam Coates Sky

@SamCoatesSky

4 months

Sky News / YouGov exclusive poll ** Tories joint lowest VI share in Parliament ** Reform just ONE point behind ** Lab down 3; LD up 4 LAB 38% (-3), CON 18% (-1), RefUK 17% (+1), LDEM 15% (+4), GRN 8% (+1) Fieldwork Mon/Tue (around Lib Dem manfiesto coverage)

263

769

3K

7

33

252

Dylan Difford

@Dylan_Difford

7 months

Polls currently imply a record 15½-point swing to Labour. But where is this swing coming from and can it be undone? The largest component is Tory defections to Labour (around a third of the swing), but losses to Reform and Don't Know/Not Voting are also big parts of the story.

15

67

249

Dylan Difford

@Dylan_Difford

4 months

It's interesting to compare the tone of coverage of Starmer in this election with Johnson in the last. They have virtually identical approval ratings, yet one is cast as a popular hero, the other struggling. The reality, both relatively unpopular by historic standards.

14

60

245

Dylan Difford

@Dylan_Difford

8 months

I'd say there's a good chance of the Tories going sub-20 in at least one poll in the next fortnight, which has the potential to trigger some really bad ideas.

Redfield & Wilton Strategies

@RedfieldWilton

8 months

Lowest Conservative % since Sunak became PM. Just two points above lowest under Truss. Westminster VI (11 Feb): Labour 46% (+1) Conservative 21% (-3) Reform UK 12% (–) Liberal Democrat 11% (+2) Green 5% (+1) SNP 3% (–) Other 2% (–) Changes +/- 4 Feb

82

412

1K

11

27

244

Dylan Difford

@Dylan_Difford

5 months

Ostensibly ‘bit of fun’ territory, but using the most recent council election in each ward, what it would actually look like if you apply local election results to Westminster boundaries. By no means a prediction, but does it tell us nothing…

28

47

236

Dylan Difford

@Dylan_Difford

8 months

This doesn't strike me as a party that is politically ready to take govt during an era of a volatile electorate, impatient at failures to deliver. If they can't even commit to a fairly popular policy while in opposition, how are they going to cope the minute the honeymoon's over?

John McTernan

@johnmcternan

8 months

There are days when you have to wonder, what is the point. Labour to ditch £28bn annual green investment pledge, party sources say “The change, after a spate of recent government attacks” No 10 must love that at least someone takes them seriously

178

226

937

10

35

234

Dylan Difford

@Dylan_Difford

4 months

Labour's gap over Tories here is 19pts. If you look at the 'weighted sample' numbers in the tables, which presumably is where the old methodology would have settled, there's an implied gap of 25pts. Quite a difference.

Britain Elects

@BritainElects

4 months

📊 Labour lead at 18pts Westminster voting intention LAB: 37% (-1) REF: 19% (+2) CON: 18% (-) LDEM: 14% (-1) GRN: 7% (-1) via @YouGov , 12 Jun

360

617

3K

7

60

227

Dylan Difford

@Dylan_Difford

3 months

Using the BBC's tallies, the final GB-wide swing to Labour is 11-points. A little shy of Attlee's swing from 1945, but larger than Blair's in 1997.

Dylan Difford

@Dylan_Difford

3 months

An average of final polls indicates a swing to Labour of 14.6 points, comfortably larger than the next three largest post-war swings. Even the narrowest poll predicts a swing larger than 1945, with even a 1992-grade average polling error pointing to one as large as 1997.

2

11

38

3

60

223

Dylan Difford

@Dylan_Difford

5 months

Based on last year's BES data, roughly a third of the local LD and Grn votes are Labour for the general, adding about another 6-8pts to the Labour vote. If you are in No10 and advising electoral strategy based on not understanding this, you should in fact quit.

Kate Ferguson

@kateferguson4

5 months

This being seized on by Tories They think they can bridge a 9% gap in the polls That thing you are hearing is a big sigh of relief in no10

253

16

114

16

63

220

Dylan Difford

@Dylan_Difford

3 months

To put the scale of last Thursday's election into perspective, the number of seats changing hands at each post-war election. In 2024, 303 seats (47%) switched party relative to 2019, 65% more than in 1997.

5

105

225

Dylan Difford

@Dylan_Difford

5 months

This is key - actually looking at council results within Westminster boundaries paints a different picture to UNS projections - LDs winning clearly in their 59th target, Labour winning in some of their 200th-odd targets, in both cases with clear squeezes of each other possible.

Lewis Baston

@lewis_baston

5 months

Here's the Lib Dem version of the target seat spreadsheet for the local elections. They're hitting their targets from the Conservatives quite a long way down the list.

12

57

173

6

64

213

Dylan Difford

@Dylan_Difford

5 months

Because it's inevitable for somebody to compare raw council losses to last year (2023 had three times more seats up), here's some benchmarks relative to total number of contests: -266 = losses are 10% of all seats -350 = as bad as last year -436 = worst for a govt since 1996

9

73

208

Dylan Difford

@Dylan_Difford

4 months

Getting the impression the real winner from this debate is me, for not watching it.

9

14

209

Dylan Difford

@Dylan_Difford

4 months

Averaging the five mid-campaign MRPs so far gives the following seat totals: Con 103 Lab 447 LD 57 SNP 19 PC 3 Ref 2 Grn 1

13

58

208

Dylan Difford

@Dylan_Difford

9 months

If you increase the tactical voting rate to 50% of progressive voters - probably a bit high - then we really are talking Canada 93-level stuff. Con 24 (-145 on MRP) Lab 502 (+117) LD 76 (+28) Nat 28 Grn 1

7

46

203

Dylan Difford

@Dylan_Difford

4 months

Roughly one third of all opinion polls to ever put the Tories below 20% have been conducted since the election was called.

3

46

202

Dylan Difford

@Dylan_Difford

4 months

Merging the three MRPs together: Con: 139 Lab: 429 LD: 38 SNP: 21 PC: 3 Grn: 2 I.e., basically the YouGov results.

8

31

199

Dylan Difford

@Dylan_Difford

7 months

Amazing how much the second referendum policy damaged Labour myth persists. In Jun/Jul 2019, Labour were polling at 13% of Leavers and 32% of Remainers, behind the LDs on 33%. In GE2019 - after they adopted a second ref policy - they won 14% of Leavers and 49% of Remainers.

Richard Johnson

@richardmarcj

7 months

A party that was polling over 50% in this same Parliament, now at 18%. Few will say it, but I genuinely believe (as I said at the time) that removing Boris Johnson was an act of electoral self-sabotage by the Tories on par with Labour’s embrace of a 2nd referendum in 2019.

286

133

771

24

40

201

Dylan Difford

@Dylan_Difford

3 months

Alternative uses for the MRP data - a dress rehearsal...

21

60

200

Dylan Difford

@Dylan_Difford

4 months

Surprising how many people are suddenly coming round to the idea that political power should not be disproportionately distributed and requires strong policing and accountability mechanisms.

Harry Cole

@MrHarryCole

4 months

EXC: Starmer will be "Blair on steroids" if he gets landslide, leading Tory right-winger warns. @miriam_cates says unchecked Labour will run amok on mass immigration, wokery and "constitutional vandalism"

535

78

239

6

27

188

Dylan Difford

@Dylan_Difford

4 months

For all the talk of the (un)popularity of the leaders, there are only two elections in the last 60 years with a larger gap in net approval ratings of the two main leaders - 1983 and 1997. Starmer's ratings might be weak, but the relative gap is large by historic standards.

6

38

186

Dylan Difford

@Dylan_Difford

9 months

The estimates for where the 14m 2019 Conservatives are today: 6.4m (46%) Still Conservative 1.7m (12%) Don't Know 1.7m (12%) Labour 1.7m (12%) Reform 1.2m (8%) Deceased 0.5m (4%) Lib Dem 0.5m (4%) Would Not Vote 0.2m (1%) Green 0.1m (1%) Other

14

52

173

Dylan Difford

@Dylan_Difford

9 months

There's an assumption that the Conservatives' polling position is largely down to the shocks of Partygate and Liz Truss, but I think you could argue they were both distractions from a slower but consistent decline that began in about mid-2021. Cost of governing + cost of living?

24

29

172

Dylan Difford

@Dylan_Difford

4 months

The problem with the 'don't let us die, we need a strong opposition' message is that a) it reeks of weakness and desperation and b) there's every chance voters - who crucially view the govt as incompetent and incapable - go 'we'll have a strong opposition, but not you'.

Sam Freedman

@Samfr

4 months

In full flailing around, panic mode now.

44

107

686

8

31

169

Dylan Difford

@Dylan_Difford

3 months

Survation and Electoral Calc MRPs closest so far in terms of constituency results.

12

31

167

Dylan Difford

@Dylan_Difford

3 months

Ultimately, this election is a total mess results wise. You've got multiple dynamics happening as part of a fragmenting party system. Hard to tell what the final tallies will be.

4

23

167

Dylan Difford

@Dylan_Difford

3 months

Also - as a bit of fun - a projection of the results onto three alternative voting systems - Scottish-style PR, a roughly county-level List PR system and STV.

26

58

168

Dylan Difford

@Dylan_Difford

2 months

Labour might not have managed the feat of being the most popular party in each GB region, something never achieved in a modern election, but they do now hold the most seats in all 11, a feat last achieved by the Liberals in 1906.

7

43

154

Dylan Difford

@Dylan_Difford

4 months

Based on the MRPs earlier this year, without a significant poll change during the campaign, this election is set to see seats change hands on a scale not seen since 1945. Even assuming a worst-plausible-case for Labour, a shift as seismic as 1997 should be expected.

10

32

153

Dylan Difford

@Dylan_Difford

7 months

Problem: Only around a 4-pt swing is needed for the Conservatives to lose their majority. Labour have likely attained this purely through natural turnover (deaths/new voters) and gains from other parties/non voters. Tories could gain back every voter they’ve lost and still lose.

3

28

152

Dylan Difford

@Dylan_Difford

4 months

To be honest, Westminster getting high on data is preferable to the stuff they're usually getting high on.

Rafael Behr

@rafaelbehr

4 months

Too much polling. Seriously. Moratorium on polling for a few days at least. A week or two. Westminster Twitter has become a drugs bore, getting high on data and only able to talk about how good the shit is, or whose stash is better or where the next hit is coming from.

36

63

506

3

11

150

Dylan Difford

@Dylan_Difford

6 months

Relative to public opinion, Liz Truss's small-state quasi-libertarianism is one of the most overrepresented views within the Westminster bubble, including being one of the dominant viewpoints in the governing party. The idea it is newfound or silenced is laughable.

Kate McCann

@KateEMcCann

6 months

Liz Truss may be an imperfect messenger, but the response to her book highlights one of Westminster's biggest flaws - it has stopped listening to ideas it doesn't agree with:

632

115

638

1

24

148

Dylan Difford

@Dylan_Difford

5 months

One of the most likely ways of the final vote lead being narrower than the polls is a degree of Labour supporters thinking the election is a done deal and not bothering. Baselessly talking up the prospect of a hung parliament seems to be a great way of getting more to turn out.

The Telegraph

@Telegraph

5 months

🔴 Labour would have to enter a “coalition of chaos” with other parties to win office, No 10 sources have suggested, after an analysis said the general election could produce a hung parliament

675

41

87

10

22

144

Dylan Difford

@Dylan_Difford

7 months

@Diogenes1 Basically, Tories lose about 25k votes a month through deaths, but only gain about 2k from new voters, while Labour lose about 6.5k and gain 8-10k.

9

46

145

Dylan Difford

@Dylan_Difford

3 months

The swing in post-war elections, including what current polls point to for this election.

8

35

142

Dylan Difford

@Dylan_Difford

5 months

Conversely, here's how Labour's coalition has evolved over the last few years. Gains from the Tories are the biggest group (and count double!), but sizeable chunks of ex-LD, ex-SNP and first time voters are also key, with them also winning over many who didn't vote last time.

Dylan Difford

@Dylan_Difford

5 months

Over the past four-and-a-half-years, the 14m people who voted Tory in 2019 have splintered in various directions, with just 4 in 10 now saying they'll vote for the party - less than under Truss. Here's how they've said and currently say they'll vote, each block 100k voters.

19

100

492

6

18

145

Dylan Difford

@Dylan_Difford

3 months

A summary of the 14 final projections listed on @inglesp 's wonderful website. 382 (60%) unanimous predictions, 118 where 12 or 13 agree, 84 where most (9 to 11) agree, 48 with too little agreement to call.

5

38

144

Dylan Difford

@Dylan_Difford

3 months

What a UNS (uniform national swing) model would project on these vote shares: Con 243 Lab 297 LD 32 SNP 53 Oth 25

16

24

142

Dylan Difford

@Dylan_Difford

4 months

Looking at the preliminary data for this week, things getting way more shocking. Across all polls, Tories now within 3pts of being *5th* among under 35s, nearly 3rd among both Leavers and Remainers and bordering on <25% in all regions. Appalling distribution for a FPTP election.

4

43

140

Dylan Difford

@Dylan_Difford

6 months

Across the trio of MRPs released this year, only 460 (70%) seats have held the same projected winner on all three, with nearly a third showing a different winner on at least one (shown in paler tones) and seven even producing different results each time (light grey).

10

30

140

Dylan Difford

@Dylan_Difford

4 months

Across the MRPs, prominent prediction models and the median of regional poll projections over the last week, the 403 seats with the same implied result across the board. 342 are for Labour, just 37 unanimously Conservative. Not a prediction, just what the models agree upon.

13

36

141

Dylan Difford

@Dylan_Difford

3 months

While no Canada 93, it is very much a Canada-esque result - volatile multi-partyism under FPTP, parties winning big majorities from under 40%, etc.

1

24

140

Dylan Difford

@Dylan_Difford

4 months

A polling error is not impossible and it's always wise to be a little cautious, but I don't find the key claim here, which is not substantiated, enormously convincing as the potential issue around Don't Knows is already (possibly over-)corrected by some pollsters.

Guardian politics

@GdnPolitics

4 months

Labour lead over Conservatives may be overstated, says Tory election expert

17

5

21

8

26

137

Dylan Difford

@Dylan_Difford

4 months

How the polls really moved (week to: 12/06). Dips for major parties, but Labour's leads steady. Reform surge largely where you'd expect it, with Lib Dem bump concentrated in the South of England (suggesting tactical voting).

Dylan Difford

@Dylan_Difford

4 months

Last week's edition of how the polls really moved - this was largely pre-Reform bump, so still not major movements, though the tightening in the Scottish breakdowns disappeared. (Ignore Wales, one very off sub-sample skewed things a lot there).

1

4

28

8

43

133

Dylan Difford

@Dylan_Difford

1 year

A lot of Conservatives seem to forget that, not just is there a party between them and Labour, but that party poses a growing electoral threat to them in one of their core seat types and could easily move to the centre-right, posing a greater long-term threat.

2

13

132

Dylan Difford

@Dylan_Difford

3 months

Across the five mid-campaign MRPs, the min (agreed by all five) / median (agreed by three) / max (on at least one): Con: 21 / 93 / 190 Lab: 368 / 454 / 534 LD: 23 / 56 / 73 SNP: 6 / 15 / 44 PC: 2 / 4 / 4 Ref: 0 / 2 / 10 Grn: 0 / 1 / 4

James Austin

@JamesDAustin

3 months

Excellent overview of what the MRP's do and don't tell us from @Dylan_Difford here The big takeaway is that there are only 21 seats that all MRP's show the Tories as retaining. Shows just how wide their battleground is - and how quickly we could go from landslide -> wipeout

8

33

86

7

32

131

Dylan Difford

@Dylan_Difford

3 months

Was going to do a good night/bad night thing, but the fact that the worst plausible result for Labour is objectively a good result considering their starting position and the best plausible result for the Conservatives is objectively quite bad means it'd be taking the piss a bit.

5

23

132

Dylan Difford

@Dylan_Difford

4 months

Increasingly, the question is simply: why wouldn't it be worse than 1997? Comparing the polling data, particularly beyond the headlines VIs, there's a bit that's as bad for the Tories now as then, quite a lot that's worse and very little that's better.

8

14

128

Dylan Difford

@Dylan_Difford

6 months

An Easter treat: Breakdown of demographics, including some I haven’t reported before, from all opinion polls this year. Excluding Leavers, pretty much a clean sweep for Labour, with big swings (the little arrows) among all target groups.

13

55

125

Dylan Difford

@Dylan_Difford

4 months

As always, fundamentals getting a bit neglected relative to headline VIs, but it's notable that Labour's fundamentals are improving, Tories' getting worse. It's why I'm sceptical at some of the readings around Labour's headline 'dip'.

YouGov

@YouGov

4 months

The number of people with a favourable view of Keir Starmer is at its highest level since Nov 2022 Favourable: 39% (+3 from 4-5 Jun) Unfavourable: 51% (-2)

3

28

80

0

31

129

Dylan Difford

@Dylan_Difford

6 months

Yet again, another MRP that is far worse for the Tories when you look at the details. Tactical voting consideration appears limited (hence the LDs being comically low), 98 seats a best-case scenario on those figures.

The Times and The Sunday Times

@thetimes

6 months

🔺 EXCLUSIVE: As Rishi Sunak hurtles towards a general election, a new mega poll offers little hope he can resurrect his party’s electoral fortunes. The forecast even suggests the prime minister is at risk of losing his own constituency ⬇️

104

186

465

7

27

125

Dylan Difford

@Dylan_Difford

3 months

Pre-campaign, there was a view that Labour's narrower lead on this question than on the headline VIs suggested that their vote lead was somewhat soft. Today, the leads basically match.

Survation.

@Survation

3 months

Best PM: Rishi Sunak: 24% (-3) Keir Starmer: 45% (+1) Don’t know: 31% (+2)

4

46

115

5

18

125

Dylan Difford

@Dylan_Difford

2 months

The real swingometer - i.e., what built the seismic 11pt swing to Labour? Con losses, Lab gains and natural turnover created a potential 19pt swing against the Tories, but this was dampened by 8pts of swing worth of Lab losses and Con gains.

4

29

123

Dylan Difford

@Dylan_Difford

4 months

Despite Labour's slightly weakening voting intention figures, their fundamentals are strengthening. Hard to square with a decline in support, with two most likely culprits being increasing vote efficiency (e.g. tactical and reverse) and/or VI numbers going rogue.

Deltapoll

@DeltapollUK

4 months

Net approval for the Prime Minister @RishiSunak falls by twelve points, while net approval for @Keir_Starmer is up by two points since our last poll.

12

136

383

7

32

122

Dylan Difford

@Dylan_Difford

5 months

Crucial to note: Projected share is roughly where we expected based on the polls, but seat gains/losses seem to be a good 30% worse than expected - suggests an additional efficiency to the anti-Con vote.

4

31

124

Dylan Difford

@Dylan_Difford

9 months

Paging Tory MPs - this chart might help you identify the 'parallels' with 1992. Going into that campaign, the Tories had the more popular leader, led on the economy and some key issues, as well as having a decent party image. That isn't true right now.

Adam Payne

@adampayne26

9 months

Tory MPs are talking about 1992 again — are there parallels? Are they onto something? There are vague similarities But whether it's Tory party unity, the economy, or the state of the polls, Sunak faces a much bigger mountain than Major did 3 decades ago

23

13

37

10

39

122

Dylan Difford

@Dylan_Difford

2 months

Did 2024 represent the final death of uniform swing? The narrower + taller the frequency distribution below, the more uniform the swing. In Oct 1974, 92% of seats had a swing that fell within a 5pt band. In 2024, no such band could be drawn that included more than 28% of seats.

7

28

119

Dylan Difford

@Dylan_Difford

5 months

Fun fact: The highest Green 'most recent locals vote' at the constituency level (Bristol Central, 57.4%) is higher than the highest share for the Conservatives (East Wiltshire, 57.3%).

Dylan Difford

@Dylan_Difford

5 months

Suffolk, Sussex and Severn: The underlying Green vote building up in local elections.

10

18

78

7

42

116

Dylan Difford

@Dylan_Difford

7 months

A useful reminder that, by the time The Sun actually 'backed Blair', Labour had been leading in the polls for four-and-a-half years, including among Sun readers (36% of which had already voted Labour in 1992) for nearly all that period.

James Heale

@JAHeale

7 months

Interesting to read in the i claims that The Sun is “leaning towards” backing Labour at the election. Here’s how Newsnight covered the paper endorsing the party, 26 years ago next month:

9

31

144

5

21

117

Dylan Difford

@Dylan_Difford

6 months

Here’s an updated version of the median winner map, substituting YouGov’s latest MRP for their January one. 105 seats that at least two MRPs agree are Conservative, though they account for only 49 of the 467 unanimously agreed seats. But what else can we learn from the MRPs?🧵

Dylan Difford

@Dylan_Difford

6 months

Across the trio of MRPs released this year, only 460 (70%) seats have held the same projected winner on all three, with nearly a third showing a different winner on at least one (shown in paler tones) and seven even producing different results each time (light grey).

10

30

140

3

19

114

Dylan Difford

@Dylan_Difford

6 months

The fact is, we would *expect* polls to narrow, but there's no evidence that they definitely *will* narrow.

John Rentoul

@JohnRentoul

6 months

Mark my words, the polls will narrow & Labour will panic before the election. My weekend article for @Independent

380

79

329

14

9

116

Dylan Difford

@Dylan_Difford

4 months

Regardless of discussions about the merits of differing methodologies, it should be noted that the Labour dip in numbers does primarily seem to be driven by a swing from Lab to LD in the south - presumably tactical voting that makes the Anti-Tory vote more efficient.

Dylan Difford

@Dylan_Difford

4 months

Labour's gap over Tories here is 19pts. If you look at the 'weighted sample' numbers in the tables, which presumably is where the old methodology would have settled, there's an implied gap of 25pts. Quite a difference.

7

60

227

5

21

112

Dylan Difford

@Dylan_Difford

4 months

Ultimately, we won't know which methodology was right until the election, but I am getting curious about the growing divergence between Labour's weakening VI figures and strengthening fundamentals - something that has often presaged a polling miss.

10

15

110

Dylan Difford

@Dylan_Difford

3 months

You can tell there is a lack of engaging in the actual detail of polls by basic things like Sunak and Starmer being treated as equally unpopular (there's the third largest approval gap in the last 60 years) and ignoring the Lib Dems (who polls say 1/5 chance of being opposition).

4

25

109

Dylan Difford

@Dylan_Difford

5 months

For those wondering the other day, here's the full current voter movements chart, in classic format. As always, each block represents 100k voters.

Dylan Difford

@Dylan_Difford

5 months

Over the past four-and-a-half-years, the 14m people who voted Tory in 2019 have splintered in various directions, with just 4 in 10 now saying they'll vote for the party - less than under Truss. Here's how they've said and currently say they'll vote, each block 100k voters.

19

100

492

10

22

110

Dylan Difford

@Dylan_Difford

8 months

I know paying attention to Euro politics is verboten in the UK, but Labour should really spend more time learning from the successes and failures of w Europe's other social democrats in the last decade, rather than constantly relitigating factional debates from 30 years ago.

Dylan Difford

@Dylan_Difford

8 months

I understand the attachment to 97, but just because it's *The* recent Labour victory doesn't mean it's the only way for a centre-left party to take power. Have some imagination. The world of 97 is about as relevant to today as Wilson's victories were to the mid-1990s.

6

12

92

7

16

108

Dylan Difford

@Dylan_Difford

2 months

There should be a cap for how much Brits are allowed to talk about the US elections, set at 3x the highest amount you have talked about an election in any other G7 country. Excession = deportation (not necessarily to the US, just somewhere I don't have to hear about it).

10

11

106

Dylan Difford

@Dylan_Difford

7 months

Fundamentally, the only sensible strategy for the Tories is one of damage limitation. There will be a swing to Labour, it's just a matter of big or very big. Targeting Reform voters and Don't Knows can help, but losses to Labour count double (as would winning them back).

7

6

104

Dylan Difford

@Dylan_Difford

3 months

@inglesp The range across the prediction models (unanimous - on at least one): Con: 6-207 Lab: 340-550 LD: 26-89 SNP: 5-39 PC: 2-5 Ref: 0-26 Grn: 0-5 Oth: 0-3 Of course, the predicted outcomes generally fall somewhere in the middle, but there's quite a bit of leeway at the seat-levels.

7

20

105

Dylan Difford

@Dylan_Difford

9 months

After some reasonable criticisms of the original chart, here's a slightly modified version that should make certain things clearer.

12

19

102

Dylan Difford

@Dylan_Difford

5 months

Conservative losses just ticked below 436 on the BBC, meaning that, relative to seats up for election, yesterday was the worst set of local elections for a govt since 1996 and second worst in last 40 years. Still 9% to declare, but unlikely to top 1996 now.

Dylan Difford

@Dylan_Difford

5 months

Because it's inevitable for somebody to compare raw council losses to last year (2023 had three times more seats up), here's some benchmarks relative to total number of contests: -266 = losses are 10% of all seats -350 = as bad as last year -436 = worst for a govt since 1996

9

73

208

2

33

101

Dylan Difford

@Dylan_Difford

9 months

Even if you add @LukeTryl 's Reform absence numbers on top of one-third of progressives tactically voting, it's not getting much better for the Tories: Con 114 Lab 426 LD 63 Nat 28 Grn 1 Probably a more plausible result.

3

20

99

Dylan Difford

@Dylan_Difford

5 months

Labour gaining Rushmoor is significant. Contains Aldershot which is a 400+ result indicator for the general election.

3

21

98

Dylan Difford

@Dylan_Difford

4 months

In big swing elections, however, this becomes an issue - parties can't grow or fall forever in every seat - they have a ceiling and a floor. It's worth stressing that if you apply uniform swing to current polls, the Tories would be on negative votes in ~80 seats. Not possible.

3

99

Dylan Difford

@Dylan_Difford

4 months

Going through the MRPs, collectively they project Labour will win the most votes in these counties (73 borders) for the first time ever: Berkshire, Bucks, Cambridgeshire, Cornwall, Devon, East Sussex, Essex, Hertfordshire, Isle of Wight, Kent, North Yorkshire and Oxfordshire.

5

19

97

Dylan Difford

@Dylan_Difford

7 months

Natural voter turnover - i.e., the churn of deaths and new voters - is one of the hidden trends of the next election. Alone, it has made Labour an estimated 1.36m votes better off relative to the Tories between the 2019 election and today, but this will continue to rise.

7

31

99

Dylan Difford

@Dylan_Difford

3 months

Sure there are some who'll still wang on about the 'great festival of democracy', but I've found the conduct of the campaign, particularly (though not exclusively) by the government, to be hugely dispiriting and a reminder of why I view British politics as fundamentally rotten.

7

13

96

Dylan Difford

@Dylan_Difford

1 year

@Samfr Not just in MRPs, you can also see this in standard poll sub-breaks - swing much higher where they need it (swing among Remainers, Londoners, under 34s is very low). Almost scarily efficient, with real potential to be devastating.

4

19

100

Dylan Difford

@Dylan_Difford

5 months

Defections to Reform have obviously been the main story of the past year, but arguably more crucial has been the consistent and unbudging 1.7m losses to Labour, who of course count double in swing terms. Reform voters only affect scale of defeat, Lab losses cause the defeat.

Dylan Difford

@Dylan_Difford

5 months

Over the past four-and-a-half-years, the 14m people who voted Tory in 2019 have splintered in various directions, with just 4 in 10 now saying they'll vote for the party - less than under Truss. Here's how they've said and currently say they'll vote, each block 100k voters.

19

100

492

8

30

100

Dylan Difford

@Dylan_Difford

3 months

One underdiscussed element of the MRPs is that the average MRP being right implies the average polls are wrong, as the MRPs have the Tories about 3pts higher and Reform 3pts lower than polls, thus also having a smaller Labour lead.

6

16

100

Dylan Difford

@Dylan_Difford

6 months

The same unprecedented nature of one party leading across the board on the fundamentals is also true if you switch from 'change' elections to the 'big defeats'. Even in 83, 97 and 19, the losers led on at least some issues and didn't have party images as universally tarnished.

Dylan Difford

@Dylan_Difford

6 months

Important to remember the polling numbers below the headline voting figures. Today's are more one-sided and anti-incumbent than those ahead of 1979, 1997 or 2010 elections. Govt behind on leadership, key issues and party image, with worse economic optimism and general approval.

7

32

90

7

25

96

Dylan Difford

@Dylan_Difford

7 months

The thing that people forget about Theresa May is that she presided over - by some margin - the largest increase in an incumbent government's vote in the modern era. It's only down to the absurdities of FPTP that May is regarded an electoral failure.

9

14

98

Dylan Difford

@Dylan_Difford

5 months

Been saying it for a while, but hard not to see how Greens don't become an electoral problem for a Labour govt. One of my few solid longshot predictions is they'll have double-digit MPs by middle of next decade - they will be the party of university towns.

18

12

95

Dylan Difford

@Dylan_Difford

1 month

Labour's extreme performances in the GE. Lots of things to note - Scottish surge, revolt of the suburbs in England, huge swing against Labour in urban areas, particularly those with high Muslim populations. But also the extremeness and variation of the changes.

5

22

96