United Kingdom General Elections: December 12th, 2019

[DistingFlyer]

Here's a three-day average of polls, with my own MP estimates (based on nationwide swing figures):

Cons - 43.3% (-0.3%), 345 MPs (+27)
Lab - 33.0% (-8.0%), 219 MPs (-43)
Lib Dem - 13.5% (+5.9%), 17 MPs (+5)
Nat - 3.5% (-0.1%), 50 MPs (+11)
GP - 2.5% (+0.9%), 1 MP

Overall majority: 40
Overall swing: 3.9% to Cons


Will update figures regularly, then make an actual prediction a day or two before the vote.

[DistingFlyer]

Three-day poll aggregate update:

Cons - 43.0% (-0.5%), 344 MPs (+26)
Lab - 33.0% (-8.0%), 220 MPs (-42)
Lib Dem - 13.2% (+5.6%), 17 MPs (+5)
Nat - 3.8% (+0.2%), 50 MPs (+11)
GP - 2.8% (+1.2%), 1 MP

Overall majority: 38
Overall swing: 3.7% to Cons

[DistingFlyer]

An overall majority like that would be a good result for Johnson, who hasn't exactly run a stellar campaign.

And the overall majority is likely larger since their should be at least 7 non-voting members: 6 is the average Sinn result right now and 1 Lab MP is the speaker.

It might also be a rare occasion where a party's lead was about the same - or maybe even larger - on election day than when the campaign began, though it's close.

[DistingFlyer]

Three-day poll aggregate update (6 Dec):

Cons - 42.8% (-0.8%), 344 MPs (+26)
Lab - 32.8% (-8.3%), 219 MPs (-43)
Lib Dem - 12.5% (+4.9%), 17 MPs (+5)
Nat - 4.2% (+0.6%), 51 MPs (+12)
GP - 3.0% (+1.4%), 1 MP

Overall majority: 38
Overall swing: 3.7% to Cons

[DistingFlyer]

Three-day poll aggregate update (7 Dec):

Cons - 43.4% (-0.1%), 348 MPs (+30)
Lab - 32.8% (-8.3%), 215 MPs (-47)
Lib Dem - 12.4% (+4.8%), 17 MPs (+5)
Nat - 5.3% (+1.7%), 51 MPs (+12)
GP - 2.3% (+0.6%), 1 MP

Overall majority: 46
Overall swing: 4.0% to Cons

[DistingFlyer]

The polls were pretty widespread, but the average suggested a clear Tory majority. Kind of like now...

There had been polls showing only very small Tory leads  (1% and 2%) so the writing on the wall was there it's just most people (including me) chose to ignore it. The YouGov forecast also fairly accurately predicted what was coming though most people thought it was a joke. This time the Tory lead hasn't dipped that low in any poll (well at least yet) and has remained much more steady over the course of the campaign. Whilst I am ruling nothing due to there being a chance that all polling is massively out, the fundamentals do look more rosy for the Tories than at this point in 2017.

It was rather interesting last time - Labour steadily rose throughout the campaign in almost every poll until the final week, when some showed them continuing to rise while others showed them flatlining. Obviously the former turned out to be right, while the latter were not; I remember being hopeful that the reverse was true, but neither was I really shocked when the exit poll was released.

Will put up a chart showing poll numbers over the course of the campaign as we get closer to the end.

[DistingFlyer]

Labour haven't been steadily rising this time, have they?

Not really; they've gone up over the campaign (high 20s to low-mid 30s), but so have the Tories (high 30s to low-mid 40s), leaving the Tory lead either unchanged or maybe a little larger.

Most of the Tory rise came early (in the first ten days or so), while most of the Labour increase came at the midpoint.

The Liberals have steadily slid, however.

[DistingFlyer]

Three-day poll aggregate update (9 Dec):

Cons - 43.4% (-0.1%), 352 MPs (+34)
Lab - 32.9% (-8.2%), 216 MPs (-46)
Lib Dem - 12.4% (+4.9%), 17 MPs (+5)
Nat - 4.2% (+0.6%), 46 MPs (+7)
GP - 2.2% (+0.6%), 1 MP

Overall majority: 54
Overall swing: 4.0% to Cons

The polls indicating 6% and 14-15% leads to have come out in the last day or so appear to have created a net change of almost nil; the three-day averages have remained almost identical for the last week. Sometimes such steadiness is a precursor to a surprise last-minute swing, but just as often it's not.

[DistingFlyer]

Obviously these predictions may not come about - a number of long-held Labour constituencies didn't go down last time when it was thought they might - but if they do, we may be witnessing the political patterns in Britain get a little closer to what we've seen in Canada for a few decades now, with well-off urban & suburban seats seldom going Tory but many blue-collar ones being strong for them. That's slowly been coming to pass over the last few decades anyway, with 1997 being a big step in that direction, but it still hasn't shifted to the same degree that we've seen in Canada since the 1960s (and even more so since the 1990s).

(For instance, if Canadian voters still went the same way as British ones we'd probably have seen the Tories win Rosedale but lose Fort Mac.)

[DistingFlyer]

When was the last time any of that solid red block of seats in northeast England along the coast ever went for the tories?

1931, but it took a nationwide vote of 61%-31% (and a seat count of 521-52) to do it. The old 'North' region (Cumberland, Durham & Northumberland) went Tory 54%-39% and 28 seats to 3.

[DistingFlyer]

Here's a graph showing the parties' poll numbers over the campaign (will update if/when more come in before tomorrow night):




As for my own prediction:

Cons - 342 MPs (43%)
Lab - 226 MPs (34%)
Lib Dem - 17 MPs (12%)
Nat - 46 MPs (3%)
GP - 1 MP (3%)

Overall majority: 34
Overall swing: 3.0% to Cons

Margin of error: +/- 15 MPs (equivalent to roughly +/- 2% of swing)

[DistingFlyer]

I'm hoping for some anti-Conservative tactical voting like there just was in Canada.

There's a fair degree of that already, and has been for some time - one reason why a Tory lead of 10% will probably not get them a majority greater than 50 - and why an actual lead of 8% in 1992 got them a majority of just 21 - while a Labour lead of 9% in 2001 produced a majority of 167.

[DistingFlyer]

Here's how the polls moved in some past elections:

1970


Feb 1974


Oct 1974


1979


1983


1987


1992


1997



2017


The darker line disregards the polls indicating the Labour rise stopping in the final week, while the brighter line takes all polls into account. Maybe the former were overcorrecting for 2015.


In general, a party's lead shrinks over the course of a campaign - sometimes just a little, sometimes quite a bit (1997 & 2017), but rarely does the party trailing actually finish up ahead (1970 & 1992). Unless there's a big surprise tomorrow, this election looks to follow the general model: the party leading at the start of the campaign still leads at the end, but by a somewhat smaller margin.

I was struck by Prof. King's remark in 1992, commenting on the surprise Tory win, that parties trailing usually fell further behind during a campaign; not sure on what he based that, but it doesn't seem accurate (even in pre-1970 elections: the Labour lead in 1964 & 1966 ended up being a little less than at the campaign's start, while the Tory lead in 1959 also shrank a tiny bit).

[DistingFlyer]

One more chart - this one looks at swings vs. seats that changed hands. (Will update it once tomorrow's results are complete.)

A note of explanation: the x-axis denotes the swing at each election, while the y-axis denotes the number of seats that changed hands between the two big parties as a percentage of marginals - for instance, 1979 saw a net change of 52 seats between Labour & Conservative out of 58 Labour marginals, while 1997 saw a net change of 144 seats between Conservative & Labour out of 68 Conservative marginals. Additionally, the color of each dot denotes the direction of the swing at the election.



One can observe the ability of Labour to defend vulnerable seats when there's a swing against them (1979 & 2001), and their ability to make extra gains when there's a swing toward them (1964, 1974, 1992, 1997 & 2017). Occasionally the Tories will make more gains than the swing would suggest (2010) or lose fewer (1974). Tactical voting obviously has something to do with this, as does what was called a 'tactical vote unwind' in 2005 & 2010.

Given that there's a certain degree of anti-Tory tactical voting already happening (look at 2017), I'm not sure how far from the line tomorrow's result will stray - perhaps on the order of 1987 or 2005, but I'd be very surprised if it was like 1979 or 1992.

[DistingFlyer]

Nobody can complain about the polls this time around - they led pretty much exactly to the right spots:



It's also one of those very rare occasions when a party extended its lead over the campaign instead of shrinking it.

(There are two constituencies still left to report - will update this chart if necessary when they come in.)


Here's an updated swing-vs.-gains chart; somewhat to my surprise, given all the talk about anti-Tory tactical voting, the Conservatives overperformed relative to the swing to a greater degree than they ever have - it's not 1992-level, but it's close:

[DistingFlyer]

https://twitter.com/danbloom1/status/1206146416793526273?s=19

I love you John. You're a fking hero.

Honestly, as someone who never liked the guy (even back when I liked Corbyn), massive respect. It takes real guts to apologize so thoroughly.

Yeah, it seems pretty clear that McDonnell & Corbyn are trying to very publicly soak up as much of the blame as possible, so that when they leave front-bench politics, the party can start up again with a clean slate. And that's a very f**king admirable & mature strategy.

Seconded (or thirded) - nothing in their front-bench careers became them like the leaving of them, to bastardize a phrase.

[DistingFlyer]

McDonnell, yes. Corbyn... I can’t see much contrition coming from him.

True; given the rather shameless message issued by his kids (and his other die-hard supporters), I doubt his graciousness will be much more than he showed on the night itself.

[DistingFlyer]

Here's a link to my (finally completed!) spreadsheet of UK elections from 1885 to two weeks ago; obviously when the full Electoral Commission report comes out the latest figures may change a bit, but this should do for now. Have also thrown in the 2014 & 2016 referenda for good measure. Data come from F. W. S. Craig's books (1885 to 1992), Electoral Commission reports (2001 to the present) and Walker's Ireland books (1885 to 1918).

A couple of items:

Firstly, the italicized constituencies (1885 to 1970) indicate constituencies that roughly fall within modern-day Greater London;

Secondly, the Irish figures from 1885 to 1918 are taken from Walker's Parliamentary Election Results in Ireland, 1801-1922, which lack tables totalling the vote by party at each election; therefore, there may be some errors in my figures as I wasn't able to cross-check my totals with anything definite (though I've gone over them several times, I can't promise there are no mistakes!).

Thirdly, there are some hidden columns indicating swing figures for elections from 1959 to the present (doing it pre-1945 didn't make as much sense due to parties not always contesting a seat two elections in a row - Labour, then Liberal, then both, then just one again, and so forth, for instance).

Fourthly, the vote percentages from 1885 to 1945 have been adjusted for dual-member constituencies rather than simply being taken from the raw totals (this is why the 1945 percentages, for instance, are 48.0%-39.6% as opposed to 47.7%-39.7%).


https://drive.google.com/open?id=1dYxOhd1afsae8DNN-tSN77Uuk6ZUKi0O

[DistingFlyer]

Since 1951 only two Labour leaders have won general elections; Wilson, by appealing to moderates and Tories then governing a little left of that and Blair by appealing to moderates and Tories and governing not so much to the left.

The last transformative government, in terms of leaving a legacy and an imprint on society was Blairs. Only four PM's can claim that legacy since the war; Attlee, Wilson, Thatcher and Blair. Three of them Labour.

The solution is simple and obvious. But it's now two defeats away now.

(FWIW I don't include Macmillan as it was continuity 'war coalition/Butskellist.' Cameron is possible, constitutionally, but too early to tell.)

One also might throw in Ted Heath, given that he took Britain into the EEC. Not much he did was of great note, but that alone was pretty significant.

[DistingFlyer]

Three times in the twentieth century the Tories, after having dominated politics for about two decades, were turfed from office in spectacular fashion - facing double-digit swings and losing more than half their MPs.

First, but least-known today, is 1906: the Conservatives had been in power for most of the previous twenty years, but went down to a big swing to the combined Liberal & Labour forces.
From 402 MPs in 1900, the Tories picked up five but lost 251, giving them a net loss of 246 and bringing them down to 156 MPs - their fewest since the 1832 Reform Act.
The Liberals, starting at 183, gained 223 and lost nine, giving them a net gain of 214 and bringing them up to 397. Labour went from two to 29 (gaining 28 but losing one).
A total of 267 seats changed hands - almost 40% of the total.
Swing is difficult to calculate, given the large number (163) of Tory acclamations in 1900, but if one looks only at seats that were contested in both elections it amounts to about 11% from Conservative to Lib-Lab.

Second, and probably still the best-known, is 1945: Tories had run the country for most of the inter-war period, but this finally came to a close with the end of the Second World War. Opinion polls predicted the outcome (in fact, they exaggerated the Labour lead) but most people didn't pay much attention to them.
Gains & losses are a little harder to gauge here, given the semi-redistribution, but a reasonable estimate is that twenty-four of the new seats should be in the Conservative column, and one (Thurrock) in the Labour one.
This gives the Tories 453 MPs going into the election, but they picked up only five while losing 248, bringing them down to 210.
Labour had 155 MPs before the vote, gaining 241 and losing only three, leaving them with 393.
Almost exactly the same number of seats - 266 - changed hands as in 1906, but with a smaller house this amounts to 42% of the total.
There were still some acclamations in 1935 (26 Conservative & 13 Labour), but not nearly as many as in 1900, so the swing - 12% - is much easier to calculate.

Finally, we have 1997, which ended eighteen consecutive years of Tory government (the longest streak in modern times). Just as in 1945, pre-election polls overstated Labour's lead, but an efficient vote and strong anti-Tory tactical voting delivered them a majority comparable to expectations anyway.
There was a redistribution here as well, but notional results were calculated to give the Tories 343 MPs and Labour 273 - not too much change from the actual 1992 figures.
Unlike 1906 & 1945, there were no constituencies that moved in the 'wrong' direction this time: the Conservatives lost 178 seats and gained none, leaving them with 165. Labour picked up 146 and lost none, leaving them with an all-time best of 419.
Fewer seats changed hands here than in 1906 or 1945: 184, or 28% of the total. Still the most since 1945, though.
In keeping with fewer seats changing hands here than in the other two blowout defeats, the swing - 10% - was a little less as well.
Perhaps the biggest difference between this election and the others is that, while in 1910 and 1950 the Conservatives won back a lot of seats and nearly regained power after just one term in opposition, the Tories stayed down - with fewer than 200 MPs and a very low vote share - for more than a decade after the 1997 defeat.

Will public opinion finally boil over in a similar fashion in five or ten years' time? As everyone here keeps saying, a lot will depend on who the new Labour leader is and how the departure from the EU is handled (and with a recession forecast in the near future, the Tories' popularity will probably suffer in the next couple years anyway). But, as 1997 showed, even if the economy is very good, this doesn't guarantee that a government will survive.

[DistingFlyer]

Here's a graph showing turnouts from 1945 to 2019; the color-coding should be pretty self-explanatory, and the vertical line indicates the lowering of the voting age.

[DistingFlyer]

Here's a map indicating both constituencies that changed hands and marginals that didn't (or, to put it another way, all marginals as well as non-marginals that changed hands).

The picture on the left shows the pre-election situation, and the one on the right shows the 2019 results. Constituencies aren't shaded according to my usual system, but simply colored according to marginal (<10%), moderate (10-25%) or safe (>25%).

If nothing else, it can provide a useful quick-glance guide as to how different parts of the country shifted.




To compare/contrast, here's one for the last election:


Here's one for 1997:


And here's one for 1979:

[DistingFlyer]

A similar chart to what I put up earlier; this one, instead of comparing the swing to gains made as a % of marginals, compares the swing to the so-called 'effective swing' - that is, the swing that would notionally provide the number of net gains that actually were made (for instance, the 144th most vulnerable Tory-Labour seat in 1997 needed a 12.4% swing for it to fall, while the 53rd most vulnerable Labour-Tory seat in 2019 needed a 5.6% swing).

The graph produces very similar - though not exactly the same - results as the earlier one.

[DistingFlyer]

What the two almost identical outcomes in the bottom corner?

1951 & 1959.

[DistingFlyer]

What the two almost identical outcomes in the bottom corner?

1951 & 1959.

So where's 1955?

Have never seen notional 1951 results on the new boundaries, so didn't include it. Originally didn't put in February 1974 until I finally found some redistributed 1970 figures.