Survation

[CatholicLeft]

Jun 15, 2024 20:17:19 GMT andrewp said:

Jun 15, 2024 20:12:09 GMT doktorb🏳️‍🌈🏳️‍⚧️ said:

That's both Survation and YouGov with Fylde as a Labour gain. I just can't see it.

If Lab are going to win 450+ seats, maybe nearer 500, they are going to win some seats that we can’t see happening. In the SW, they would win some or all of Central Devon, Bridgwater, SW Devon potentially, SE Cornwall- all of which would be in my I can’t see it category, but the polls think we are in that territory.

I could just about see Labour gain SE Cornwall, where the swing to them has been pretty substantial in recent times, and Bridgwater has always been a possible if they could get the Labour turnout up and regain former strength in Highbridge.

[ricmk]

Jun 15, 2024 20:12:09 GMT doktorb🏳️‍🌈🏳️‍⚧️ said:

That's both Survation and YouGov with Fylde as a Labour gain. I just can't see it.

That's about the only thing here I can believe - with the Mark Menzies scandal affecting the local party, and both Blackpool N and (ex) Blackpool S Tory MPs beset by scandal, I'd expect an epic swing against the blue team in Fylde.

[CatholicLeft]

Jun 15, 2024 20:51:32 GMT ricmk said:

Jun 15, 2024 20:12:09 GMT doktorb🏳️‍🌈🏳️‍⚧️ said:

That's both Survation and YouGov with Fylde as a Labour gain. I just can't see it.

That's about the only thing here I can believe - with the Mark Menzies scandal affecting the local party, and both Blackpool N and (ex) Blackpool S Tory MPs beset by scandal, I'd expect an epic swing against the blue team in Fylde.

Well, yes, possibly. That isn't why they are predicting it though, it they were, it would mean they were taking local factors into account, which Survation obviously haven't in Exmouth and Exeter East.

[batman]

I still find a Labour gain in Fylde hard to believe. But then I spent a lot of time there when I was growing up. Grandma usually voted Labour if she voted but other Labour voters were very few and far between. All her relatives and friends there were Tories. It was one of the very safest Tory seats in the land but somehow it has rather slipped down the list of those.

[carolus]

Jun 15, 2024 20:22:22 GMT cogload said:

Someone on twatter gas offered Survation 10 grand if the LD's finish 3rd in North Shropshire.

I've said it before and will say it again - it is absurd not to take into account byelections in these sort of predictions, and I note that other providers (e.g. Yougov) now clearly do.

[The Bishop]

Jun 15, 2024 20:22:36 GMT finsobruce said:

Jun 15, 2024 20:13:18 GMT graham said:

Did Labour not win Castle Point in 1997?

Yup, on a 16% swing or thereabouts. The swing required now is just under double that if my maths is correct.

Of course a lot of Tory majorities in 2019 are artifically inflated but even so, this would be proper nose bleed territory.

In 1997 there were not one but two anti-EU candidates, that polled significantly more between them than the Labour majority.

Partly for that reason, it was one of the handful of seats that Labour lost to the Tories in 2001.

[finsobruce]

Jun 15, 2024 21:53:49 GMT The Bishop said:

Jun 15, 2024 20:22:36 GMT finsobruce said:

Yup, on a 16% swing or thereabouts. The swing required now is just under double that if my maths is correct.

Of course a lot of Tory majorities in 2019 are artifically inflated but even so, this would be proper nose bleed territory.

In 1997 there were not one but two anti-EU candidates, that polled significantly more between them than the Labour majority.

Partly for that reason, it was one of the handful of seats that Labour lost to the Tories in 2001.

One of the two, Linda Kendall, also stood in Rayleigh and Wickford in 2015 opposing local housebuilding.

She wrote to the Daily Mail last October saying that she "preferred Rishi Sunak's sincerity over Sir Keir's advocacy of change for changes sake".

[iain]

Jun 15, 2024 21:12:21 GMT carolus said:

Jun 15, 2024 20:22:22 GMT cogload said:

Someone on twatter gas offered Survation 10 grand if the LD's finish 3rd in North Shropshire.

I've said it before and will say it again - it is absurd not to take into account byelections in these sort of predictions, and I note that other providers (e.g. Yougov) now clearly do.

It clearly is absurd, but, in defence of the polling company, it is also inconsequential. There are a small number of seats which no regression analysis can cope with, and these are best ignored - by-elections, seats with strong indies, Green targets, etc. There is little point putting a load of effort into attempting to perfect the results in these areas because they are, by definition, beyond the scope of the model.

More broadly, it is perfectly clear that YouGov are the only pollster who have managed to create an MRP model worth a damn. Other MRPs may ‘beat’ them in being closer to seat numbers due to more accurate polling, but in terms of the patterns of support only YouGov has a track record which allows one to take them seriously.

[carolus]

Jun 15, 2024 22:47:12 GMT iain said:

Jun 15, 2024 21:12:21 GMT carolus said:

I've said it before and will say it again - it is absurd not to take into account byelections in these sort of predictions, and I note that other providers (e.g. Yougov) now clearly do.

It clearly is absurd, but, in defence of the polling company, it is also inconsequential. There are a small number of seats which no regression analysis can cope with, and these are best ignored - by-elections, seats with strong indies, Green targets, etc. There is little point putting a load of effort into attempting to perfect the results in these areas because they are, by definition, beyond the scope of the model.

More broadly, it is perfectly clear that YouGov are the only pollster who have managed to create an MRP model worth a damn. Other MRPs may ‘beat’ them in being closer to seat numbers due to more accurate polling, but in terms of the patterns of support only YouGov has a track record which allows one to take them seriously.

Not true. It is certainly very possible to build a model which algorithmically deals with byelections. Survation have simply chosen not to do so - which is revealing about their model and their objectives.

[Forfarshire Conservative]

Jun 15, 2024 19:06:07 GMT hullenedge said:

Lots of swing seats.

I find it hard to believe that Berwick would go before DC&T.

[The Bishop]

Their general Scottish findings are so bizarre (SNP sweep every seat in Glasgow, but lose Falkirk??) that they are best simply binned tbh.

[Tony Otim]

So their model works poorly in Scotland and Wales, works poorly where there's been a by-election, works poorly where the Greens are a factor, the Reform results also look very odd in some cases and TBH, some of the other results are a bit iffy...

[hullenedge]

[batman]

2 weeks and a day to go. And the Tories are actually losing ground in the campaign so far. They are going to lose some extraordinary seats unless there's some sort of a miracle, despite Labour's share not being particularly high (though of course much higher than last time).

[jamesdoyle]

Jun 19, 2024 9:03:06 GMT batman said:

2 weeks and a day to go. And the Tories are actually losing ground in the campaign so far. They are going to lose some extraordinary seats unless there's some sort of a miracle, despite Labour's share not being particularly high (though of course much higher than last time).

We are, I think, definitely into the polling zone where the operative factor is not the Lab-Con gap (you've flooded out all the targets that are achievable without a significant change) but the Con-LDm gap, where there are still quite a lot of seats vulnerable to a small shift. I still think that 100-150 is the likely landing zone for the Cons, but we're 1 or 2% away from 50-100 becoming more likely.

[Adam in Stroud]

Jun 19, 2024 9:24:30 GMT jamesdoyle said:

Jun 19, 2024 9:03:06 GMT batman said:

2 weeks and a day to go. And the Tories are actually losing ground in the campaign so far. They are going to lose some extraordinary seats unless there's some sort of a miracle, despite Labour's share not being particularly high (though of course much higher than last time).

We are, I think, definitely into the polling zone where the operative factor is not the Lab-Con gap (you've flooded out all the targets that are achievable without a significant change) but the Con-LDm gap, where there are still quite a lot of seats vulnerable to a small shift. I still think that 100-150 is the likely landing zone for the Cons, but we're 1 or 2% away from 50-100 becoming more likely.

Tricky gap to measure, for the pollsters.

The big picture of Lab leading Con by c.20 points is clear, as are the rough sizes of the RefUK, LD and GPEW vote shares.

But whereas up to 3 point moves in the Lab or Con shares don't effect that Lab-Con lead much, especially if they shadow each other (ie both down a bit/both up a bit), a few points on the Con&LD lead could flip a lot of seats.

Just on this poll the changes shift the Con-LD lead from 13 points to 8. I think that would put a lot of Tory seats at risk even on UNS, much more if it is the result of the national figure reflecting tactical voting shifts.

But it wouldn't be shocking if those numbers were reversed in the next poll.

[batman]

Jun 19, 2024 9:24:30 GMT jamesdoyle said:

Jun 19, 2024 9:03:06 GMT batman said:

2 weeks and a day to go. And the Tories are actually losing ground in the campaign so far. They are going to lose some extraordinary seats unless there's some sort of a miracle, despite Labour's share not being particularly high (though of course much higher than last time).

We are, I think, definitely into the polling zone where the operative factor is not the Lab-Con gap (you've flooded out all the targets that are achievable without a significant change) but the Con-LDm gap, where there are still quite a lot of seats vulnerable to a small shift. I still think that 100-150 is the likely landing zone for the Cons, but we're 1 or 2% away from 50-100 becoming more likely.

I would tend to agree. I totted up the likely Conservative holds in my head last night, without looking at notionals and seat profiles in detail, and I came up with a figure of about 137. Before I post detailed predictions I will look carefully at notionals and profiles.

[The Bishop]

A few pollsters showing "Others" at about 5% now - even given there are some high profile and credible Independents standing this time, that still looks a bit high.

[rc18]

Bear in mind that this election has broken the record for number of candidates standing, by some margin, with virtually all of the increase coming from 'Others'. There are no constituencies where there are fewer than 5 candidates. 2019 and 2017 on the other hand were modern low points, i.e.;

democracyclub.org.uk/blog/2024/06/08/2024-uk-general-election-candidate-summary/

The increase is made up of a creditable number of Worker's Party (and SDP) candidates, and a huge increase in Independents. A fair few of these could be described as Gaza single-issue candidates. It wouldn't be surprising to me if they siphon off a couple of percentage points of leftwing main party votes.

[The Bishop]

I acknowledged that, but significant "Gaza Independents" and the like are only standing in a relative handful of seats out of 650.

Most "fringe" candidates will prove to be exactly this, and barely trouble the scorers.