United Kingdom General Elections: December 12th, 2019

[cp]

FYI, Dominic Raab is having another wretched night at the hustings in Esher & Walton. Follow Lewis Goodall for a play by play.

[cp]

I'd argue the focus on Labour anti-semitism is perfectly proportionate - quite aside from the various horrorshows with e.g. holocaust deniers in the party, there are still way too many people who clearly have antisemitic attitudes to some degree (thinking in particular of the people who can't discuss it for two sentences without mentioning Israel). It's bad and we deserve the kicking we've been taking for it.

There is an issue in that racist attitudes in other parties don't receive sufficient attention, but I don't think we should be getting an easier ride to compensate.

Proportionate focus maybe, but the level of vitriol directed at Labour/Corbyn is totally out or proportion to the substance of the allegations being made. Granted, the tenor of the dialogue isn't much worse than, say, any random flame war about Israel or Trump or the EU/Brexit. But that's a pretty low bar to set.

[cp]

So Corbyn had a horrible interview with Neil, though it wasn't as well roasted as Sturgeons. I can't wait for the Boris interview, since it appears Neil will bat three for three.

Yeah, I watched it. Corbyn came off peevish and was evasive at times that didn't make sense. That said, at other times Corbyn made good points and came off principled, especially on the waspi issue, while Neil seemed a bit obsessed with bean counting and bizarre specifics and hypotheticals (he did this with Sturgeon, too).

Truth be told, the more entertaining - and who knows, maybe informative - viewing was the online commentary reaction. Corbyn's backers were out in force saying he did well. Corbyn's opponents were as shrill and hyperbolic as ever. I struggle to see how either side's efforts will change anyone's opinion.

[cp]

Turns out that the BBC have yet to schedule a feature-length Neil interview with Johnson. It appears that everyone else signed up and were allotted slots on the assumption that everyone was one board. I don't like Corbyn and I don't like feeling bad for him,* but this is really serious, a potential breach of the BBC's commitment to balance and political impartiality. Negotiations are supposedly ongoing, but that is just not good enough: they now have to set a date and empty-chair Johnson if he doesn't show up. This is very disturbing.

_{*I also think Neil's style of interview does not do our democracy any favours.}

This feels like more of one of those macho, strutting 'power moves' that campaigns like to pull; distract and frazzle your opponents for half a day by doing something that hints at an action you'd never actually take, but plausible enough that people worry.

If Johnson pulled out of the interview at this stage he would likely incur a far greater level of censure than anything he might trip over saying during a grilling. It would fuel the narrative of Johnson being evasive and untrustworthy - things that he's already weak on and doesn't want to reinforce if he can help it.

I could be wrong of course. Could be that the Tories are playing hardball with Neil behind the scenes, trying to get him to promise not to ask about a certain topic (Johnson's family, mistresses, etc.)

[cp]

Con 359 (+42)
Lab 211 (-51)
Lib Dem 13 (+1)
SNP 43 (+8)
Plaid 4 (-)
Green 1 (-)
Speaker 1 (-)

Game over I think. It might not be accurate but the two main parties will use the data to squeeze mercilessly. The Hard Brexit duopoly wins.

Well that is concerning. The Tories would have to seriously underperform for there to be a chance of stopping BoJo's hard Brexit (and like it or not, that can only happen is Labour overperforms. I'm not going to count on polls underestimating Labour again (if anything, I'm afraid they overcorrected from 2017 and might now be overestimating it), but I guess there is still hope of more LibDem voters coming home between now and December 12 (the very thing you fear, but which is the only hope or stopping BoJo). We shall see.

Just so we're all clear, this is a forecast for an election two weeks away based on a poll that's nearly a week old using a model that specifically doesn't account for local campaigns. I'm not saying it's worthless, but it is hardly conclusive.

[cp]

https://twitter.com/britainelects/status/1200824832042164225

Labour closing the gap with accelerating speed. Still outside the margin of error for a tie, though.

[cp]

Lib Dems are 13 in every recent poll. Labour is 31-34. But the Conservatives are 39-46, a much wider range.

I wouldn’t be shocked to see the Lib Dems underperform popular vote polling while exceeding the number of predicted seat wins at this point. Something like 11% with 26 seats would not surprise me at this point.

I suspect this will be the case. To what extent it occurs may decide the outcome of the election (i.e. Tory majority or not). Mounting evidence suggests substantial numbers of voters in the SE are either not voting for the Tories this time and/or tactically supporting Lab/Lib as the best chance to defeat the local Tory candidate.

On the subject of polling that Oryx brought up, it occurred to me last night that we have to control for at least two different types of error when assessing how much a polling firm, or the industry more generally, might be indicating something inaccurate. The first error has to do with how polling firms measure a person's propensity to vote. Whether a person is 1/10 likely to vote on e-day, 5/10, or 10/10 affects how polling firms weight their samples and, hence, how their top line numbers turn out. This is what was behind the 2017 GE and 2016 referendum polling error. Most firms assumed lower propensity, Labour/Leave leaning voters wouldn't turn out, but they did.

The second error has to do with a person's choice of vote, i.e. the likelihood that a person, of any though usually high propensity to vote, will choose Party X instead of Party Y or Z. This type of polling error is what was behind the missed calls in 2015 and 1992. Lots of people who said they could vote Tory but also Labour or Lib Dem decided, largely at the last minute, to back Tories in large enough numbers to deliver them a majority that the polls indicated wasn't likely.

So, which error is more likely to be at work in this election? Because 2017 saw a failure to accurately measure propensity to vote, it's reasonable to assume pollsters have adjusted for this. That doesn't mean they've *fixed* it, mind you, just that they've made a new set of assumptions to determine voter propensity. High rates of voter registration, greater interest/investment in the election, and changing demographics are all confounding factors, too. I'll also note that in virtually every survey taken in the past two months, the pre-weighting results for Lab/Tory has been a tie.

On the other hand, how likely is it that pollsters will be embarrassed because a crucial bloc of voters change their minds at the last minute? Obviously, we won't know until the votes start getting counted, but there's evidence to suggest there may be potential for this. If one is to believe the constituency polls (yes, yes, I know) for Portsmouth South, Esher & Walton, Beaconsfield, Warwick, and slew of Northern Labour held seats , there are substantial numbers of voters willing to shift their support. This indicates the potential for further sudden changes that pollsters might not be equipped to anticipate or detect. There's also the matter of the Tories draining  their well of Brexit/UKIP voters but Labour still having lots of potential Lib Dem voters to squeeze in the next 10 days.

All told, I don't know which of these errors is more likely to manifest. But then, neither do the pollsters.

[cp]

Maybe the YouGov poll is actually the most interesting one out of that lot, given their tendency going back almost two years now as Tory-friendly - Labour equalling their highest rating since the launch of Change UK (remember them?) and the second lowest YouGov lead since Johnson became PM.

As for Opinium - lol. Literally nobody - including in Tory HQ - actually believes they are ahead 46-31.

You are probably mostly correct.
But is it possible the other pollsters are over herding to 2017 results?  Is it possible they might over compensentate for tactical voting?

Because a pollster notifies the polled person with the tactical choice, does not mean all the voters really get the notice.  It also does not mean all of the possible shifters will actually shift.

And finally there may be no reliable way to gage shifting Leavers, who may be shifting in greater numbers than in 2017.

Let’s wait till 12/13 to judge the pollsters.

Yes, the big worry is that pollsters have overcorrected since 2017 (as pollsters tend to do when their previous house effect was off) and BoJo is on track for an even bigger majority than it seems thanks to previous nonvoters. Hell, their previous overcorrectiong from when the pollsters missed 2015 was part of the reason why Corbyn's surge was totally off the radar. The issue with this line of thought though is that the MRP poll seems to suggest that overcorrection is going on to a slight degree, but it's main effect is the solidification of already expected Tory flips, rather than padding an already large margin. But yes, the best play as I always say, is to watch the tracking average models and see what 10 days bring.

As I recall the MRP weighted awfully heavily for Brexit views, thus the extreme result in the North and Midlands. But the assumption that Brexit matters to voters more than it actually does may be why Labour is being underestimated and why they seem to be surging. Voters in Labour-Leave constituencies aren't going to suddenly flock to the party of Jacob "you should have just left the building" Rees-Mogg because of Brexit, even though the prevailing narrative is that surely they'll do exactly that. It's a really dumb narrative, honestly. Red Labour Leave voters don't become Tories, they become simply Red Labour Leave voters. That's why the decision of the Brexit Party to leave those constituencies uncontestes is by no means an automatic gamechanger.

Should that prove to be the case, the likely result will be a Labour government. It also ought to (but probably won't) prompt a great deal of reflection among commentators and psephologists. The ur-narrative of the past 3 1/2 years has focused on this nearly-mythic group.

[cp]

The Lib Dems have actually been fairly stable for the last two weeks or so at 15% or a few points south of that - I think the Labour increase might be more due to undecideds and possible a few soft Tories swinging their way. Whether that will be enough for 2017-redux who's to say. Mostly this time I think the Lib Dems do have a higher floor than last time though because of Brexit, and it looks like they might have reached it. 'Might' definitely being the operative word there.

A quick thought on the polling atm (from the general discussion thread)

I'm not sure the top line numbers are the best way to read the crucial dynamics at work. Lib Dem fade over the past few weeks might have been remainy types switching to Labour OR it could be dissatisfied red Tories getting scared of Corbyn and boosting the Tories. Similarly, Labour's slower rise compared with 2017 could be because they're picking up remainy votes in the south but losing leavey votes in the north - some of the local polling would seem to corroborate that.

Something worth keeping in mind whatever the case: there are a *lot* of wavering Lib Dem/Labour voters, and a relatively large number of true undecideds. Depending on how they shift by next Thursday, the Tory lead could be double digits or completely gone.

[cp]

Bit of random fun/observation: In a fit of masochism I decided to look up the forum thread from the 2017 election, focusing on the last 10 days before the vote, to see what useful trends I might be able to discern in the debate/commentary people had. Here's what I found ...

1. Everyone agreed May had run the worst campaign in living memory, but a good few people were still bleating about her being a 'strong' leader. A true testament to the power of cognitive dissonance.

2. Corbyn hating was, if anything, more intense than it has been this year. Seriously, I was shocked at how vitriolic the rhetoric was; I sort of assumed it had gained volume and intensity over the past 2 years.

3. The YouGov poll that projected a Tory minority was very much an outlier and not taken very seriously by anyone posting. Even when a couple of other polls published subsequently (YouGov came out about a week before polling day) showed the Tory lead down to single digits - and, in one case, a -1 to Labour - most people dismissed them.

4. Relatedly, if there was an overall tone to the final days of the 2017 race's commentary on here it could be summed up as "The Tories really stepped in it, but they're still going to win a majority." I couldn't find a single prediction that didn't have the Tories winning 330+ seats.

5. There was a much greater number of different posters contributing to the thread at the time. This was not to the thread's benefit.

[cp]

2. Corbyn hating was, if anything, more intense than it has been this year. Seriously, I was shocked at how vitriolic the rhetoric was; I sort of assumed it had gained volume and intensity over the past 2 years.

Even though I'm here now? 😁

Lol. Your Corbyn Derangement Syndrome exhibits relatively mild symptoms compared to the terminal cases from 2017.

In all seriousness, the 2017 thread was *way* heavier on the 'Corbyn is a terrorist' angle. This year the collective wisdom centres on something more like 'Corbyn is incompetent/a Brexiter/shifty'.

With regards to the polls published tonight, they obviously show Labour closing the gap. Maybe more importantly, they're showing the Tories stagnant and hitting their ceiling. Taking the margin of error into account, it's conceivable the BMG and Survation polls are actually showing the same thing (herding?).

At the risk of belabouring this gimmick: compared with 2017, these polls have Labour *and* the Tories about 3-5 points shy of where they were 10 days out. The only difference, really, is how well the Lib Dems and Brexit Party are doing. Depending on which votes these two smaller parties siphon from the main parties - and where they do it - the seat totals could vary enormously.

[cp]

With Labor gaining, it looks like this could end up as some sort of 2017 redux, where conservative PM calls for new elections in light of strong polling numbers, but then ends up with a bare victory and another hung parliament as leftwing voters come home to Labor.

From your keyboard to God's ears.

[cp]

The Tories' polling numbers weren't that great when Johnson called the election, however, and while there has been a tightening of the polls, the Tory lead is still around 2-3 points larger than at the same point in 2017.

Well, they were hovering in the mid to high 30s had a 10-15 point lead by most measures. It's not the heights they had during May's pre-2017 honeymoon, but pretty good overall.

The Tory lead is definitely a few points shy of this point in 2017, but I think that has more to do with stickier Lib Dem votes than a more successful Tory (or less successful Labour) campaign thus far.

[cp]

It seems like the WWII-present UK system is Labour wins a landslide once every generation, Conservative pluralities/narrow majorities otherwise.

If we're looking for decades-long cycles and patterns in Westminster elections, the standard account portrays Labour as the dominant party from 1945-1979 (when they won 6/10 elections and were in government for 18/34 years) and the Tories ever since (7/10 wins, 27/40 years). It's also worth noting that during these periods, when the 'non-dominant' party took office they did not substantially challenge the overarching consensus set by the other party.

If one were inclined to put faith in these observations, it would be quite easy to craft a narrative of a generational change in 2019 that would see the Tories on the losing end. I think that approach is far too simplistic, alas.

[cp]

I don't think that does show them going different ways. That would be a 5pt swing towards the Tories nationwide and a 2.5pt swing in London.

And, in any case, they are by different polling firms...

And in any case, they cannot possibly capture the full scope of Jezz-mentum which will be realized only in 10 days.

That's a nice thought for a Lab supporter, but even I'm not willing to go that far

On the poll, something to keep in mind is that YouGov publishes two polls per week with slightly different methods for collection/weighting. The relevant poll to compare this one to is the one conducted Nov 26-28, which showed the Tories at 43 and Labour at 32.

Side question, but was wondering why in Liverpool Labour tends to have some of the highest margins in the country and Tories struggle to crack double digits.  I can see in some parts of Birmingham or London which are predominately non-white why this might happen or in university towns, but Liverpool is not especially young and it is fairly white or at least close to 90% white.  Any particular reason?

Mostly historical reasons relating to the miner's strike and the local government. Liverpool was run by one of the most ardent left councils and was the centre of power for the Labour Party's leadership through most of the mid 20th century. Think of it as to Labour what the Cotswolds are to the Tories.

[cp]

I also can't help but feel the shortness of this campaign has compressed time in a weird way; the Prince Andrew news cycle alone clearly blocked off a significant chunk of the campaign, and the parties/campaigns have been so pushed with time- post votes were dropping last week!

I'm really thinking out loud; but I can't work out if the short campaign means we're going to see the result everyone predicted (Tory Maj between 30-60) because there was no time to change it, or are we going to see a weird bizarre result (Lab win & or Tories up to 400) because it was a short campaign where no-one knew what was going on, and one side failed to turn out or inspire their vote.

I've been completely wrong in both 2015 (predicted difference of 10 seats between Labour and Tories) and in 2017 (predicted healthy Tory majority) but this election I'm still completely baffled

Totally agree. The time of year has had a palpable effect on the campaign, too. Local canvassers can't really operate for more than a couple of hours before needing a break from the cold (certain Canadian transplants excepted, of course).

I think part of the seeming ambiguity about Labour's rise this year is a function of the 2017 campaign being so long. Labour gained in the polls pretty steadily from day 1 through to the end; it wasn't a sudden 'surge'. However, because those gains took place over 7-9 weeks it was possible to build a narrative of acceleration or momentum. It was a matter of perception and context, not the raw numbers themselves.

This year, Labour has been gaining at pretty much exactly the same rate, but it doesn't look like a 'surge' because it hasn't been going on for very long and the Tories have also had a boost by poaching from the Brexit Party (which collapsed later in the campaign than UKIP did in 2017). Put another way, the Tories hit their peak later in 2019 than in 2017, while Labour has time yet to hit theirs.

[cp]

Getting a bit ahead of ourselves here aren't we?

And this may be an occasion when the polls "average lead" turns out to be not much use.

Indeed. There's no doubting the Tories are in the lead. That said, a case can be made that every major election in the UK (and the Scottish referendum) since 2005 had a crucial final week. Personally, I'm more bearish about a Labour victory than I was a few weeks ago (more bullish about a Lib Dem upset in E&W, tho), but if I was working for the Tory campaign and said we could put our feet up and coast right now I would expect to be fired.

[cp]

Since we're doing stories from the front lines, here's mine: I handed out flyers at the train station this morning. Mostly encouraging responses, with one glaring exception. A lady in her 50s started yelling at me, saying I was trying to overturn a democratic vote, the EU was corrupt, and various other Daily Maily agitprop. I stayed calm until she walked away, at which point she turned back and hollered at me: "You're not even English!"

For the record, I was born in Canada and still speak with that accent. I'm also a UK citizen and have lived here 12 years.

I'm really never going to forgive the Tories for what they unleashed these past three years. They deserve oblivion, not a majority.

[cp]

Apparently two comRes polls were conducted, the one shown above, was commissioned Remain United. The Other, for the Telegraph, shows an 8 point Tory lead.

British Prime Minister Boris Johnson’s Conservative Party’s lead over the opposition Labour Party has narrowed to eight points from 10 with days to go before Britons vote in a Dec. 12 election, according to a Savanta ComRes poll for the Sunday Telegraph.

The Conservatives were on 41%, down one point from a survey published on Wednesday, while Labour were up one point to 33%. Savanta ComRes surveyed 2,034 British people between Dec. 4 and Dec. 5.

A separate Savanta ComRes poll for Remain United, an anti-Brexit group, released earlier on Saturday showed a six-point gap, with the Conservatives on 42% and Labour on 36%.

https://uk.reuters.com/article/uk-britain-election-telegraph/johnsons-lead-over-labour-narrows-savanta-comres-poll-idUKKBN1YB0IH?il=0

Picking up on Oryx's point, that would seem to align with the idea of the Tories being 7ish points ahead and falling. For comparison's sake, at this point in the 2017 election polls showed a Tory lead between 1 and 12 points, and polls over the following week showed a similar distribution. This was also the point in 2017 when the second terrorist attack of the campaign (the one in London) happened, which I will point to as proof that there very much is still time for changes to happen one way or another.

With that said, I think we're close enough to the close of the campaign to be confident about one observation: this election isn't *precisely* like 2017, but it's unfolded more like 2017 than any other recent election.

[cp]

Three more constituency polls, from three varying seats.

https://twitter.com/britainelects/status/1203396078604865536

I currently consider this a three-way tossup, and similarly does YouGov. When presented with two way battles between Lib-Dems and Labour, the race remains marginal. Essentially, it's an unknown if non-Tory votes can or even want to consolidate here.

https://twitter.com/britainelects/status/1203394680697872387

This once was a three way marginal, but it has slid off the triple  battlefield, prompting the question of where the Lib-Dem base would go. Turns out that answer was nowhere. Conservatives with in both a Blue-Red and a Blue-Orange hypothetical if only one had a realistic opportunity here. Numbers align with YouGov's projection.

https://twitter.com/britainelects/status/1203394822133944320

I knew I was right to give this Surrey Lib-Dem target to them in my model. Conservatives lose in every prompted realistic 2-way matchup. This seat voted fairly hard for remain, and had a Lib-dem base in 2017. It's also an open seat. YouGov has the seat as lean Conservative, but there is a lot of Lib-Dem and Tory crossover, and I don't need to repeat myself on how every swing model, even YouGov's, unshoots the parties that focus on specific targets rather than the entire board.

Worth noting these are all nearly a week old and date from when Labour was closer to 10 points down on average compared with 6-7 now. Putney could be a Labour lead now and Southport a tie.

Completely agree with you about Guildford and the Lib Dems, though. The rumours on the ground there are, if anything, even more encouraging than they've been in Esher & Walton.

On that subject, the Tories seem to have actually shown up to campaign for the first time in weeks. I spotted two or three new Tory lawn signs this evening, still outnumbered by Lib Dems, though - and this is in the *suuuuper* Tory part of the ward (Cobham). My partner went to an even in Walton today with Gina Miller and the actor Hugh Grant. Said there was great turnout.

[cp]

Just putting this here for later reference: a rough timeline of election night declarations and their relative significance.

https://www.theguardian.com/politics/2019/dec/08/election-night-timetable-polls-tory-labour-lib-dem?utm_source=dlvr.it&utm_medium=twitter

[cp]

snip

Oh yeah I saw some labourite mentioning this guy earlier. You know your side isn't doing so hot when the Romney style "unskew'ers" come out of the woodwork.

I think that's a little unfair. The unskewed polls guy from 2012 assumed incorrectly that because Republicans in 2012 were more 'engaged' than Democrats, any polling that didn't have GOP/Dem party identification at least equal were using unrepresentative samples. (Note the logical leap: 'engagement' =/= 'party ID').

The Dr Moderate thread/logic is, as far as I can gather, all about weighting, not sampling. They note few polling companies (read: Kantar) are underweighting youth turnout and overweighting 65+ turnout, leading to abnormally high Tory numbers in individual polls and an inflation of the aggregate polling average. This is quite readily observable in the relevant pollster's data, and has been commented upon on here more than once. They also argue Leavers are being overweighted and new (likely to be Remainer/young/non-Tory) voters are being underweighted. The former point I'm not too sure about, but the latter point seems valid, as polling companies do appear to double weight young, previous non-voters in their modeling.

Obviously there's a degree of motivated reasoning going on, at least for me, but I think there's more credibility here than with the unskewed polls guy.

[cp]

snip

Oh yeah I saw some labourite mentioning this guy earlier. You know your side isn't doing so hot when the Romney style "unskew'ers" come out of the woodwork.

I think that's a little unfair. The unskewed polls guy from 2012 assumed incorrectly that because Republicans in 2012 were more 'engaged' than Democrats, any polling that didn't have GOP/Dem party identification at least equal were using unrepresentative samples. (Note the logical leap: 'engagement' =/= 'party ID').

The Dr Moderate thread/logic is, as far as I can gather, all about weighting, not sampling. They note few polling companies (read: Kantar) are underweighting youth turnout and overweighting 65+ turnout, leading to abnormally high Tory numbers in individual polls and an inflation of the aggregate polling average. This is quite readily observable in the relevant pollster's data, and has been commented upon on here more than once. They also argue Leavers are being overweighted and new (likely to be Remainer/young/non-Tory) voters are being underweighted. The former point I'm not too sure about, but the latter point seems valid, as polling companies do appear to double weight young, previous non-voters in their modeling.

Obviously there's a degree of motivated reasoning going on, at least for me, but I think there's more credibility here than with the unskewed polls guy.

Polling companies by their nature are supposed  to try and get accurate numbers. They do not want to release bad data because the people paying for these polls won't give the  polling company money in the future if the results are just going to be off. I remember how hungover it was around the YouGov office after Brexit because we had put out the 'exit poll' calling a Remain victory, and I also remember the proverbial champagne coming out after 2017 for a similar reason.

If you start messing around with naturally low crosstabs then you are asking for trouble. How many times on this forum have people noticed how high the GOP vote in the crosstabs is with African Americans? Or what about the Urban/Suburban/Rural breakdown? Or how about in this very thread, where we have to be cautious about low-response constituency polls.

Now, why might groups be weighted in some fashion? I dunno, perhaps because the electorate is going to be different than 2017? Is that really hard to believe, especially considering how different the circumstances are between the two elections? All but the worst pollsters (McLaughlin) have no reason to lie with their data, so this is their best estimate. If they are off, they will be off, and we will know either with YouGov on Tuesday or the exits on Thursday.

If the polls are going to be off, it won't be because we went diving into the weights and found errors. Rather it is what I alluded to earlier: there is a high number of undecided voters this late into the campaign. British polls love to remove these guys from the topline, but if they move as a block (who knows...) than the polls can be both right and off.

Well, yes, but then the issue at hand is *how* one believes the 2019 electorate will be different and what premises led one to that conclusion. I'd be fascinated to read what YouGov (or any other pollster's) logic is for adjusting the weighting as they do. If they make a convincing case based on reasonable assumptions and/or polling, then they lend credence to their results. If not, or in the absence of such explanations, we're left with little more than reputation and venerability on which to base a conclusion. The Twitter thread referenced above at least gives something of a justification.

[cp]

So this happened today.

It gets worse (for the Tories). Matt Hancock, the Health Secretary, made an unscheduled trip to the hospital. He was heckled as he left, but the BBC falsely reported it (based on 'Tory sources', of course) as a Labour activist punching a Tory aide. Considering this is the event in question, I'd say it's *just possible* that was a misleading source.

[cp]

https://twitter.com/britainelects/status/1204306504276414464

Unclear which ComRes method is being used here, i.e. whether it's the same they used for Gina Miller's poll last week or their regular methodology. Either way, 7 point gap pulls the overall averages of the race closer to a draw.