United Kingdom General Election: July 4, 2024

[Citizen (The) Doctor]

https://twitter.com/survation/status/1808888765970096600?s=46

LOTO Ed Davey is possible!!! (Had a brain fart)

[Torrain]

Tom Watson has spent election day campaigning against Corbyn in Islington North.

Which is one of the most THIGMOO things I’ve seen in quite some time.

[DL]

Survation has Reform winning 15 seats....no one else has them getting anywhere near that many - let's see where they land

[IceAgeComing]

I think a big part of that is presentational - Survation use probabilistic estimates for their top line number so if reform is just second in a bunch of seats that will add extra to them. I think it’s the better method although might fall apart in a scenario like that.

In the UK the exit poll is very accurate - I think the biggest miss for the top party since the new methodology was adopted in 2005 was by 15 seats in 2015 and last year had everyone within 5 seats. So really we will know the result at 10 unless the dynamics of this election throw it off and I suspect they’ll know and communicate if that’s a risk.

[DL]

Watching election night in the UK always makes me marvel at two things:

1. After the exit poll is announced at 10pm - then there is like a 2 hour wait for there to be any actual results to analyse, meaning the anchors have to talk and talk and talk for over two hours and its all just idle speculation.

2. At 6am its still the SAME people anchoring the election coverage. how do they do it? Do the on-air personalities all do lines of cocaine together or take methamphetamines to stay awake?

[YL]

I think a big part of that is presentational - Survation use probabilistic estimates for their top line number so if reform is just second in a bunch of seats that will add extra to them. I think it’s the better method although might fall apart in a scenario like that.

In the UK the exit poll is very accurate - I think the biggest miss for the top party since the new methodology was adopted in 2005 was by 15 seats in 2015 and last year had everyone within 5 seats. So really we will know the result at 10 unless the dynamics of this election throw it off and I suspect they’ll know and communicate if that’s a risk.

I think there's a risk that if the result is very extreme it might be less accurate than usual, because that's how statistical modelling usually works: the method is tuned to typical situations not to outlying ones. But even if that's the case we will still know that it is extreme, and they may well tell us.

Although the headline figures are usually very accurate, they don't have enough data in individual constituencies to forecast what's going on in unusual races. In particular it probably won't tell us much about Independents' chances, and not much more about the Greens.

[MillennialModerate]

Watching election night in the UK always makes me marvel at two things:

1. After the exit poll is announced at 10pm - then there is like a 2 hour wait for there to be any actual results to analyse, meaning the anchors have to talk and talk and talk for over two hours and its all just idle speculation.

2. At 6am its still the SAME people anchoring the election coverage. how do they do it? Do the on-air personalities all do lines of cocaine together or take methamphetamines to stay awake?

I always wondered this too. I saw some old coverage where David D was anchoring 9:50-7:00 and then again from 9:30-1:00. The sleeping schedule must be wild.

And why do the declarations take so long? And why are some around 1am and others 4am and then others not until 9am. It makes no sense.

And politicans. The ones with 4am declarations - do they stay up or go to bed and wake up very early And I’d guess ones with cabinet jobs have to alter that plan if they have to be in London for cabinet reshuffles etc

[MillennialModerate]

I think a big part of that is presentational - Survation use probabilistic estimates for their top line number so if reform is just second in a bunch of seats that will add extra to them. I think it’s the better method although might fall apart in a scenario like that.

In the UK the exit poll is very accurate - I think the biggest miss for the top party since the new methodology was adopted in 2005 was by 15 seats in 2015 and last year had everyone within 5 seats. So really we will know the result at 10 unless the dynamics of this election throw it off and I suspect they’ll know and communicate if that’s a risk.

I think there's a risk that if the result is very extreme it might be less accurate than usual, because that's how statistical modelling usually works: the method is tuned to typical situations not to outlying ones. But even if that's the case we will still know that it is extreme, and they may well tell us.

Although the headline figures are usually very accurate, they don't have enough data in individual constituencies to forecast what's going on in unusual races. In particular it probably won't tell us much about Independents' chances, and not much more about the Greens.

I get the feeling if the raw result they get is off from the polling in the direction of “insane” they’re likely to alter it downward out of fear of egg on there face

[beesley]

And why do the declarations take so long? And why are some around 1am and others 4am and then others not until 9am. It makes no sense.

A number of reasons why in previous elections:

1. The travel time from polling station to count, and whether such travel involves land, sea or sky.
2. Whether there were local elections being held concurrently.
3. To a degree, the number of counts being held in one place.
4. Previously, the size of electorate. The turnout in any case.
5. Whether they began operations at 10pm at all.

[DL]

And why do the declarations take so long? And why are some around 1am and others 4am and then others not until 9am. It makes no sense.

Some of that makes sense...rural seats are often the last to report since they need to take all the ballot boxes to a central location from across a large area. In theory seats with more voters would take longer than seats with fewer votes to count - but sometimes the first seats are ones with lots of votes - presumably because the powers that be in that constituency wanted to be part of that pointless competition to see who can complete their count the fastest and so they employ more people to count. Some declarations also take longer because its super close and one candidate or another has asked for a recount.

I prefer the Canadian system where the ballot boxes are opened on the spot and counted on site and you get reports on 5/200 polls and then 10/200 polls etc...

BTW: Is it true that because all the ballots are counted together at once, there is no tally of how individual polls voted within each constituency - all you get is a grand total - making it that much more difficult to gerrymander or for parties to know where they did best?

[beesley]

BTW: Is it true that because all the ballots are counted together at once, there is no tally of how individual polls voted within each constituency - all you get is a grand total - making it that much more difficult to gerrymander or for parties to know where they did best?

Generally. It's possible to do ward by ward tallies, but very few ever get released or even leaked.

[DL]

BTW: Is it true that because all the ballots are counted together at once, there is no tally of how individual polls voted within each constituency - all you get is a grand total - making it that much more difficult to gerrymander or for parties to know where they did best?

Generally. It's possible to do ward by ward tallies, but very few ever get released or even leaked.

If that is the case, how do they create "notional" results of the previous election in redrawn constituencies after a redistribution?

[JimJamUK]

If that is the case, how do they create "notional" results of the previous election in redrawn constituencies after a redistribution?

If varies by source, but generally it’s based on using local elections to estimate the distribution of the vote within a constituency and then using these wards as building blocks for the new ones. This generally works better when the preceding local elections looked like the general election result.

[beesley]

BTW: Is it true that because all the ballots are counted together at once, there is no tally of how individual polls voted within each constituency - all you get is a grand total - making it that much more difficult to gerrymander or for parties to know where they did best?

Generally. It's possible to do ward by ward tallies, but very few ever get released or even leaked.

If that is the case, how do they create "notional" results of the previous election in redrawn constituencies after a redistribution?

Educated guesses sounds harsh but it's not inaccurate - Professors Railings and Thrasher who provide the 'official' notional used by broadcasters make detailed guesses about each locality based on other data to help them make an estimate.

[LAKISYLVANIA]

I think a big part of that is presentational - Survation use probabilistic estimates for their top line number so if reform is just second in a bunch of seats that will add extra to them. I think it’s the better method although might fall apart in a scenario like that.

In the UK the exit poll is very accurate - I think the biggest miss for the top party since the new methodology was adopted in 2005 was by 15 seats in 2015 and last year had everyone within 5 seats. So really we will know the result at 10 unless the dynamics of this election throw it off and I suspect they’ll know and communicate if that’s a risk.

They got Brexit notably also wrong.

[afleitch]

I think a big part of that is presentational - Survation use probabilistic estimates for their top line number so if reform is just second in a bunch of seats that will add extra to them. I think it’s the better method although might fall apart in a scenario like that.

In the UK the exit poll is very accurate - I think the biggest miss for the top party since the new methodology was adopted in 2005 was by 15 seats in 2015 and last year had everyone within 5 seats. So really we will know the result at 10 unless the dynamics of this election throw it off and I suspect they’ll know and communicate if that’s a risk.

They got Brexit notably also wrong.

There was no official exit poll for Brexit.

[afleitch]

As much as uniform national swing is lamented, a quick reminder and perhaps some caution that in 2019 a UNS GB wide would have given a majority of 74; a lot better than most other models.

[President Johnson]

When will have a clear picture how many seats Labour is going to clinch? Or at least whether it gets over 400?

[soundchaser]

When will have a clear picture how many seats Labour is going to clinch? Or at least whether it gets over 400?

Presuming Ipsos does an exit poll again, shortly past 10 PM U.K. time.

[CumbrianLefty]

Also, any news on turnout? If it as expected, lower or higher?

Election day turnout rumours in the UK are usually nonsense.

Though it has been claimed by a Times journalist, quoting a senior electoral official, that turnout is on track to be "broadly similar" to the last GE - contrary to some predictions of an extremely low figure.

(and, less reliably, Tory HQ has apparently sent out an email saying it is "higher than expected")

[Oryxslayer]

I think a big part of that is presentational - Survation use probabilistic estimates for their top line number so if reform is just second in a bunch of seats that will add extra to them. I think it’s the better method although might fall apart in a scenario like that.

In the UK the exit poll is very accurate - I think the biggest miss for the top party since the new methodology was adopted in 2005 was by 15 seats in 2015 and last year had everyone within 5 seats. So really we will know the result at 10 unless the dynamics of this election throw it off and I suspect they’ll know and communicate if that’s a risk.

I think there's a risk that if the result is very extreme it might be less accurate than usual, because that's how statistical modelling usually works: the method is tuned to typical situations not to outlying ones. But even if that's the case we will still know that it is extreme, and they may well tell us.

Although the headline figures are usually very accurate, they don't have enough data in individual constituencies to forecast what's going on in unusual races. In particular it probably won't tell us much about Independents' chances, and not much more about the Greens.

I commenting cause I can't stress this enough. While for it is obviously noteworthy for the smaller parties, this matters for the big ones. In general yes, the exit poll has had a small MOE topline, maybe more now. But by region or individual seats...it is often very off in the places it matters, things just average out. If you want, go back and watch BBC coverage from 2019/17/15/etc like I have done at times. Every now and again they'll pull up a graphic/visual/data tool and highlight seats they expect to fall. And every time, with hindsight, we see there are a bunch of holds that flip and flips that hold. (My reading is that at some point they can't take things lower than a certain level of breakdown and are forced to do a uniform swing. This misses certain behaviors like Labour doing relatively better in the south and smaller cities/towns in 2017, and then holding up better than the national picture in these same areas in 2019 while losing seats with even larger majorities in the north.)

Also, In a similar vein, when the first seats are announced, there will be reflexive comparisons to the exit for the same reason. Usually this leads to 'doubts' because the Northeast in the past decade has always been an extreme example of only one trend taking place on election night. This time there will probably be a large reform presence in the NE, probably passing the Tories in a number of seats. But the exit poll doesn't handle such regional behavior too well for the reasons I described above, even though the overall numbers will likely average out.

[Antonio the Sixth]

Northern Ireland is, I think, the one part of the UK in which turnout figures can be seen throughout the day.

https://twitter.com/Saraita101/status/1808836161017724974

Does Britain not even do nationwide turnout reports? I guess its election administration is stuck in the early 19th century just like its constitutional structure.

The British seem to subscribe to the philosophy of less information being better, which has its own merits. (What do on-the-day turnout reports or partial results actually add other than drama, anyway?)

There is a legitimate public interest in timely updates on the state of democratic participation. The lack of partial results I'll give you is defensible, but the real travesty here is the absence of results by precinct which seriously hampers electoral analysis. I get that there's a quaint charms to the archaism of it all, but let's be real, this is largely another example of "we're doing it this way because we've always done it this way".

[oldtimer]

Also, any news on turnout? If it as expected, lower or higher?

Election day turnout rumours in the UK are usually nonsense.

Though it has been claimed by a Times journalist, quoting a senior electoral official, that turnout is on track to be "broadly similar" to the last GE - contrary to some predictions of an extremely low figure.

(and, less reliably, Tory HQ has apparently sent out an email saying it is "higher than expected")

Some are talking about a "working class" turnout surge.

[MABA 2020]

Just got back from the polling station, dead quiet but it is pretty late in the day

[Punxsutawney Phil]

https://twitter.com/oldenoughtosay/status/1808816391077683517

My favorite part of polling day (before the exit poll).

Still more evidence British democracy has gone to the dogs!