Paul's Election Blog

Well then that was fun wasn't it!

If you were a Green supporter that is. Not so much of you were a Labour or hopeful Reform supporter though.

The result
G 14,980 40.7% (Pred - 29.0) Error 11.7
R 10,578 28.7% (Pred - 28.1) Error 0.6
L 9,364 25.4% (Pred - 26.8) Error 1.4
O 1,892 5.1% (Pred - 16.2) Error 11.1%

T/O 47.4%

Was I close?

Well i got the winner, and the right order, but actually I wasnt that close as my overall error was high.

Although I pretty much nailed the Reform vote, and was within 1.5 of Labour, I was nearly 12% under on the Greens and similarly over on the lesser parties. Why?

Well there was a complete collapse of the Lib dem vote which most likely went Green. That went against the polling which seemed to indicate that the lib dems would hold onto their 4%. Pred the late polling squeezed that vote as the lib dems are a fluid lot and readily lend their votes out tactically.

There was also an even bigger collapse in the tory vote. They dropped 6% which was much higher than expected. One might have thought the tory vote just goes to Reform, but I'm not actually so sure. The tories have much more to fear from Reform than Labour, I suspect tactically voting Green for the one nation conservatives to keep Reform out at the last minute was a thing as the polls showed it was close. The tories got nearly 3,000 in 2024, they got 706 here. 50% of them absolutely were going to Reform but the other 800, I think they went to the Greens.

That accounts for about 1,600 of my miss. The other 1,700 i think is what I expected to happen after my analysis of the 1st Omnisis poll, a late shift from Lab to Grn as the 'anti -reform' party became more clear. It wasn't a big one, but it appears to have been significant and accounts for the rest.

The turnout didn't quite increase as I thought it might. It was a touch higher than 2024 but i really expected a surge from the Denton area which appears not to have materialised as although my % for Reform was about right, the actual number of votes missed by 700 or so, which was partly what I added on from voters who didn't vote in 2024.

The result, a clear win for the Greens and almost a sort of vindication for FPTP as the voters did manage to sort themselves out in the end and gave one party a clear mandate.

Also, the 'anti-reform' group needn't have worried here in the end. The total Lab + Grn vote was more than double the Reform, meaning that there was no combination of numbers which would have allowed Reform to come through the middle. They simply had no path to victory. The vitriol between them was just unnecessary.

A final note, well done to Sir Oink. He finished best of the rest with 159 votes, beating out 5 others including Advance UK for a creditable 6th place.

So, final update and prediction for Gorton & Denton.

So I've amalgamated the two polls from Omnisis and Opinium in my own special way, giving me a sample subset of 695 opinions.

The top lines from these are as follows:
- Labour are retaining about 50 - 55% of their vote from 2024
- It's going roughly, 18% to the Greens, 11% to Reform, 3% LD and 12% to others/won't vote
- The new poll gave a good indication that Reform, Greens and the Tories are pulling in new voters that didn't vote in 2024, at a rate of 5:3:2 in that order - this is a mixture I think of never voted and Con voters who stayed home in 2024,
- From the new poll, the workers party voters in 2024 almost en masse are going to turn out for the Greens - 89% of their reported vote is headed that way.
- 20% of the total sample were still undecided.

So where does that leave us and how have I modelled this.

So I start off with the base from 2014, and redistribute the Labour vote. I then have redistributed the WP vote. From that, I have shed off the disillusioned Tories and LDs. I then start adding in the Green and Reform support using their 2024 numbers, adding in the estimated amounts of new voters.

The final point is to reallocate the don't knows. And his is the hardest part. Most of the reported undecideds were likely to vote.
The Opinium poll did ask the undecideds which way they were leaning, and to be honest, it was all over the shot with no real pattern. A fair few of the undecideds were Labour voters from 2024, and this makes sense, they are wondering whether to vote Green, Labour or vote for a third party or Reform. And the numbers show from the 41 reported don't knows, they roughly split 30/30/20/20 between those 4 categories.

So adding those on what do I get as my final call:

G: 29.03% - 11,650
R: 28.15% - 11,299
L: 26.82% - 10,762
C: 6.54% - 2,624
LD: 3.81% - 1,531
Adv: 2.47% - 994
O: 3.35% - 1349
Maj: 351
T/O: 51.47%

Oh and a final thought, Sir Oink might not be last, 3 of the 550 indicated they were going to vote looney plus 5 of the 41 undecideds for a total of 8, that's 1 more than Rejoin EU and 2 more than the Libertarians and the SDP. The Communists will end up last.

So we have a new poll, commissioned by Forward Democracy, who's #1 goal is to advise people on how to vote to keep out right wing parties, formerly focusing on how to beat the tories, and now how to stop Reform.

I'm not sure they got the answer they wanted:
https://www.bylinesupplement.com/p/exclusive-poll-anti-reform-tactical

They've dressed it up as the Greens are ahead 30-28 over Labour and Reform also on 28 amongst the 339 who gave an indication that they would vote, but of those who are a bit less sure on the full 401 Respondents, it drops to G 28, L 28, R 27.

Now the tabs haven't been released yet, presumably to allow the Byline Times who published the top line results to get some airtime, but I expect them out tomorrow.

Opinium are a good pollster so the numbers will be accurate and the methodology good. But I want to dig in to see if there's anything more in there and how they have dealt with undecideds and also to look at the past recall for the sample to see if it's representative or they've got too many from certain areas that they've had to try and correct for.

Then I'm going to try and put this and the omnisid polls together so that I can collate 600 responses and generate a vote prediction for thursday from it.

No idea if it will work but I want to try and model this thing and see how close I get!

It's really hard to see how this one will break out now. That was an indication from the omnisis poll that the greens were best placed to see off the Reform 'threat', but those casts doubt on that.

There's also an interesting undercurrent. Whilst the omnisis poll did not pick up any real support for Advance UK, the Wikipedia entry for this poll has him on 6%, consistent with recent national polls for the newly branded Restore Party, year to be properly registered, led by Rupert Lowe - who was kicked out if Reform for an alleged assault and is currently the subject of a parliamentary investigation he is trying to stop via judicial review.

He seems to be leeching about 7% of the reform vote on that latest national polling, and the policies published appear to be back to the lines of the BNP who in their heyday were getting about 8-10% in European elections and even got themselves 2 MEPs at one point.

The point here is if the Advance UK candidate is are pulling 6% here, that's going to cost Reform any chance of winning here, and they might even end up 3rd.

I'll try for one more update before the polls open on thursday morning!

Exclusive Poll: Anti-Reform Tactical Voters Poised to Defeat Nigel Farage’s Party in Gorton and Denton By-election Greens and Labour are in an effective dead heat against Reform UK, with tactical voters leaning towards Zack Polanski’s party

We have a poll. And a fairly good one with a reputable source. Tables are here: https://www.omnisis.co.uk/poll-result/constituency-polling-for-by-election-in-gorton-and-denton/

It's headline -
Grn 33
Ref 29
Lab 26.
Con 5
Oths 5

Now annoyingly, when you dive in, there's no recall to who voted for the Workers party in 2024 to see where those votes have gone. Intuitively, you would expect them to go mainly to the Greens, but it would have been nice to see if for example, 60% of them actually were planning to stay home.

But what it does tell you is that this isn't 1 election, there's 3 going on in different areas.

In the Dentons, Hannah is a clear 3rd on 13 behind Angela on 18 and Matt way out in front on 36.

It's a different story in Levenshulme and Longsight though, where that's flipped, Hannah 34, Angela 21, Matt only on 8.

Then there's the middle ground of Gorton and Burnage. Here, it's Hannah 20, Angela 15, Matt 18.

On the whole though, there's is a pattern. The labour vote is down 40% apart from in 2 of the 3 dentons, where it's pretty much the same.

Longsight sees the biggest shift in green vote, from only 10% in 2024 to 42% this time. Levenshulme, from 25% to 58%. If you wanted empirical evidence of where the WP votes are going, that's your evidence.

The other areas are more modest, with a significant improvement in all wards but not of that magnitude.

Matt's vote is up the other way, but from a much higher base level. Reform were already 2nd in the Dentons, which has now gone to 1st. He matches Hannah's increase across Gorton and Burnage, but is flatlining in Levenshulme and only a small increase in Longsight.

Looking at the raw numbers and only looking at people who say they will vote (historically, that's where you look - exit polls have repeatedly shown that anyone saying less than they are very likely more often than not don't show), and showing changes from 2024:

Lab 109 to 68
Ref 30 to 77
Grn 28 to 87
Con 17 to 12
LD 8 to 6 (seriously, why are they standing)

That gives a statistically significant lead (over 5%) for Hannah over Matt and for Matt over Angela.

What you also need to account for is turnout. In the Dentons, the poll shows turnout is likely to increase by 40% with 72 saying they voted in 2024 and 100 saying they will vote this time. That is exceedingly rare. Bye elections have historically low turnouts compared to GEs but only 46.8 bothered showing up in 2024, presumably cos it was a safe seat so why bother. Now it's not, you might well see it go up.

Gorton shows a similar increase, but Longsight, Levenshulme and Burnage all show a decrease. Perhaps some WP are staying at home, or perhaps the Lab voters are doing so.

But it also could mean that the poll might be overcooking the Denton numbers as too many people are saying they will vote when actually, they won't.

There's no intention that anyone is gong to vote for Advance UK, the proxy here for Restore. Whether than changes or not, will be interesting, but if he does start to pick up, it will be curtains for Matt.

What does this mean for my predictor?

Well on these, I give you:

Green 35
Reform 31
Labour 27
Con 5
Ld 2

But actually, I am not predicting that cos the poll will actually change the numbers by being released.

Where's the first poll in January showed labour just behind reform on 33 to 36, that's changed significantly. With the Greens now leading and labour not that near Reform or the Greens, the labour vote I predict will collapse further to sure up the Green vote. And the polling when asks about a straight fight between Hannah and Matt, Matt only leads in 2 of the 3 Dentons, and Hannah's vote in the Longsights and Levenshulmes jumps to way over 50.

So there's not a lot of change from my original prediction with this:

Green 35-40
Reform 30-35
Labour 15-20

Regrettably, the poll didn't look good for Sir oink-a-lot.

Not much new to report apart from a locally organised hustings last week which Reform declined to stand citing that he quotient get a fair crack at the whip due to the rules of debate. Not sure that's going to do down too well with anyone who was unsure about which way they were going.

He is going to turn up to the BBC one next week though, and that will be interesting.

No new polls yet, social media keep spreading around the electoral calculus MRP, I do wish they would turn it off or update it. Most other political models are also not updated to deal with the bye election.

I'll try to figure something out, but it would be nice to have a proper poll with some cross tabs to work with other than that 153 vote sample which although want useless, isn't great either.

https://www.manchester.gov.uk/directory_record/538138/statement_of_persons_nominated_and_notice_of_poll_for_the_february_2026_gorton_and_denton_by_election

So we have the final candidates. No Workers Party and no Your Party candidates as the fringe left have decided that taking votes away from the Greens is more likely to let Reform in.

Now it starts to get interesting as we start to see if the labour voters stick with them or they start to mass migrate to the Greens.

Legal notices for the Gorton and Denton By Election 2026 - Statement of persons nominated and notice of poll for the February 2026 Gorton and Denton by election | Manchester City Council Legal notices for the Gorton and Denton By Election 2026

Well no sooner than I talk about the threat of the splinter from the Worker's party, George has decided to not enter the race. The Greens are supportive of the plight of the Gazans and a such, there would be an inclination for the 5-10% who would have supported the worker's party to vote green instead.

It certainly tips the greens back into a slight favourite in my mind to move from a ceiling of 30% to 40% against Reform's probable 35% assuming the tory voters try to tactically win the seat.

Once all the candidate's are finalised, I expect yougov to commission a poll and see where the starting gate is at. Then it will be a bitter battle and not for the faint hearted. It will be full of smear and hate and will generally be the worst kind of politics about winning rather than genuine policy discussions, unfortunately.

Well it's reasonably fun having a by election in your own backyard and alas although I was Gorton before 2023, I now fall into Rusholme so I can't cast my vote for Sir Oink a Lot (although I do encourage others to do so!)

As the candidate deadline falls on tuesday, a clearer picture will fall then about who is standing but it's quite clear that officially no-one is doing any sort of pact. Candidates from the current main Westminster make up of Lab, Con, LD as well as the up and coming Reform and Greens are confirmed.

It is likely that The Workers Party will nominate someone, it was thought to be the local Longsight Councillor but George the Cat himself might throw his hat in, although I dont think he'll find his way into a 6th different constituency to represent.

Added to that are some minor candidates from parties like Rejoin EU, Advance UK and of course, the Monster Raving Loonies so quite a selection.

A word on Polls and 'Projections':

There's have been an early poll with a very small sample size - 146, but it seems to be representative and shows Reform 30, Labour 27, Green 17, Don't Know 18.

A caveat, it did not prompt the workers party, but the don't knows I think will include them and genuine ones which haven't made up their minds.

Polls with small sample sizes are not inaccurate solely based on that. If the sample is representative of the make up of the electorate, it will be in the ball park. What it will have though is a whacking great margin of error for each top line. 146 people is about 5%. So Reform are 25-35, labour 22-32, green 12-22.

The published figure says Ref 36, Labour 33. But ignore that. They have reallocated the dont knows. This far out, that is unwise. Only pay attention to the tabs which include don't knows.

Big sample sizes which are not representative would be far worse. In 1948, a poll of 2m confidently predicted Dewey would thrash Trueman in the Chicago Tribune, but as the poll was not representative at all, it was way way off the mark.

Wheat you can correctly discern from this is that it's not quite Reform v Greens yet, but a genuine 3 party race at the start.

There's also been points to electoral calculus and UKPR's MRP projections. Now I use these alot when I do my mass predictions but there are not good at by elections.

MRP is multiple regression prediction. It's quite clever in that it looks at the national polls, discerns trends from how people voted last time to what they are going to do next time, then applies that to the make up of each constituency, I.e., making sure it had the right number of each gender, education level, demographic, income levels etc.

It's very good at looking at if all 650 voted, what would the overall numbers for each party likely be. What it's awful at is predicting each individual constituency, even at a general. It's very awful at local by elections.

Do not take any heed of them, they wont tell you anything about what this result will be.

My original prediction of Green 35-40, Ref 25-30, Labour 15 i think will be a little off. I think the workers party will be a solid 4th with 5-10 and I didn't factor them in. That will come from the left and will hit that top line of the greens.

So I've revising it down to:

Green 25-30
Ref 25-30
Lab 15-20
Work 5-10.

No-one else keeps their deposit (although come on the Loonies, get into 5th!)

I'll update once a week or when something significant happens.

How to throw away a 14,000 majority at a by election - don't allow the guy who would win it comfortably to stand.

Burnham would have romped home and they had nothing to fear from a Mayoral election, Labour are well liked in Manchester.

Now, the Labour voters in Gorton will stick it to Keir as he isn't popular in the Manchester region because (whisper this bit) he's not really Labour.

The Greens will get 35-40% at least. Reform 25-30%. If Labour get 15% ill eat my hat. If i was a lib dem, I'd vote green.

Well I wasn't quite right, Reform outperformed everyone's expectations. The Runcorn result was close, but the council results were not. They took from labour and the Tories and are now a real challenge with 30%.

However, they now have to govern which is not quite as easy as the Greens found out when they took over Brighton for the first time and it all went a bit south a bit quickly.

What they probably will do is cut staff & cut services. It might work, if it does and the locals like what they see, then watch out the big parties as everyone is gong to be watching very closely.

Labour probably need to take a hard look at themselves. Was cutting winter fuel payments worth it? The voters appear to have had that at the front of their minds when they cast their ballot. 20% was a historic low.

The tories were always gong to do badly, the electorate hasn't finished punishing them yet and I don't think this will be the last bad result. They only got 15%, but I think they could go down to 12 or so if Reform show they aren't useless at governing. But they are in opposition, they can spend a few years out in the wilderness and rebuild.

The lib dems had a reasonable night. They picked up the dissasafected left, and again, maximised their vote in the right places. They got 17%, 3 councils (very nearly 5), but their old bread and butter - by election swings, was not apparent here.

The greens on 10% had I think a dissapointing night. They didn't capture the west of England mayor, and they didn't really breakthrough in terms of councillors. I think they hoped for better.

Interesting times ahead. Nigel says 2 party politics is dead, I'm. Not quite sure I'm there yet. Locals has always been the chance for a third party to do well, the liberals have often hit mid 20s in these, and if you look at the EU elections, they were not dominated by the lab/cons for a while before they creased to be.

This is a year 1 local in a 5 year government. It's a bloody nose to Labour, but if they listen, the voters might come back in 2 or 3 years time. If they don't however, well, 2029 might be very very interesting indeed.

I've not done anything on tonight elections as there's not much to write home about. The tories are starting from a relatively high base, as these are last contested in 2021 when Boris was still going, they went do well, they may even lose all their councils.

Labour won't make alot of gains but might make a few, they should hold onto the Runcorn seat though despite a strong shoeing from Reform.

Reform will do well but not amazingly so and both the lib dems and greens might make small gains as they often do in these types of elections.

They don't mean very much though, and are very local. The vote share will be interesting, expect Reform to be 2nd and the Tories way back. But no party will get more than 30% when John Curtis extrapolates the equivalent national share

Turnout will be awful. Perhaps low 30s.

2024 Debrief.

We are pretty much there with the results, the final numbers will take a few more weeks to be verified, but we can se where the trends are.

The current Popular vote is Harris 73M, Trump 76M. There are (about) 4M to come in. Most are Californian, Harris could well win these 3M to 1M, giving the final vote share somewhere in the region of Harris 76M, Trump 77M.

2020 for reference: Biden 82M, Trump 75M.

So Trump has gained 2, but Harris lost 6.

Trump did not win many votes from Harris, they just stayed home (quite similar to the Tories over here in May). Trump didn't win this election, Harris lost it.

Senate, all but one in, 47D, 52R, the PA one is close, the R lead is shrinking by about 5K per day, but I'm not sure there are 5 days left where that will happen.

But look at where the D Senate candidate won, and Harris lost:

Michigan: Harris 2.73M Trump 2.81M
Wisconsin: Harris 1.67M, Trump 1.7M
Nevada: Harris 704K, Trump 750K
Arizona; Harris 1.56M, Trump 1.75M

Michigan: D 2.71M; R 2.69M
Wisconsin: D 1.67M, R 1.64M
Nevada: D 700K, R: 676K
Arizona: D 1.65M, R 1.57M

Harris and the Senate Democrats batted even, generally.

On the Republican side however, this is an entirely different story. In all states, Trump got far more votes than his Republican colleague. 120k in Michigan, 60K in Wisconsin, 74K in Nevada, 180K in Arizona. In PA, its 150K (interestingly Harris also did about 25K better than Bob Casey).

Trump voters voted for Trump. They did not care down ballot.

That goes even closer when you look at the house. The current score stands at D 212, R 219. 4 to call, 3 in California, 1 in Alaska, that national vote will likely end up with the D's winning it, but the R's having control on a knife edge, perhaps a change of 1 seat if it finishes 220-215.

Its a strange system america have where they have so many races and ballot initiatives all being voted on at the same time. Unlike the UK where if this happens (rarely) you get a different ballot slip for each election gong on, you just get the one slip in the US to mark for all the races. That means people are getting their pencils and consciously only filling in the top race.

What they are not doing, is voting either a straight ticket (some states allow you to check a box at the top which will actually count for all the races if you want to do that), or a split ticket where the marker votes for one party in one race, but the opposite one on a different one. (The only case where that happened was the NC Governor race, although not in huge amounts, most R's just didn't vote for him, but the governor got enough votes which if Harris had got, she'd have won).

It just seems really strange that if you go to the effort to either go down to the polling booth, or you open up your absentee ballot, that the vote doesn't vote for everything.

What it does tell you though is that Trump appears to have a personal following, over and above the rest of his party. Harris didn't.

That makes 2028 interesting, because if those say, 500K people who just voted for trump don't turn up for whoever is on the ballot with an R next to their name, they ain't gonna win.

I leave you with one message to get you through the next 4 years, or the next two until midterms:

Covfefe