Categories
politicalbetting.com » Blog Archive » Combination therapy. An occasional reminder that using seat pr
politicalbetting.com » Blog Archive » Combination therapy. An occasional reminder that using seat predictors on current polling is stupid
You see it happen regularly, on here and on twitter. A new opinion poll comes out, showing dramatic news. Immediately, we decamp to Electoral Calculus and Flavible, to discover that such a poll, if replicated at a general election, would produce a hung Parliament with Plaid Cymru the largest party, able to form a coalition with the Greens and Lady Sylvia Hermon. The oracles have spoken. “Cor blimey”, we expostulate.
0
This discussion has been closed.
Comments
Have we ever had a large scale nationwide protest vote in a general election ?
The attraction of these models, of course, is that we have an absurd voting system where knowing the levels of support for the parties tells you remarkably little about how they are likely to be represented.
69.4% of the vote
88% of seats
The reason why these models are likely to be less reliable isn’t solely because we are expecting dramatic falls in the level of support for both major parties - it is also because the demographic and hence geographical basis of support for both of them is changing.
If the level of party support shifted dramatically, but still along the same lines as the previous GE, the models would be more reliable. Yet we can see that UNS has already broken down - because it’s a model that *ought* to work backwards as well as forwards (on the same boundaries), yet you can key the 2010 or 2015 election results into a UNS model and get nowhere near the actual result.
In particular the models are struggling with identifying where the possible 10-15% extra vote share for the BXP and LibDems and the possible extra 5% for the Greens might fall. The one thing that is unlikely is that it arrives as a straight swing (after all, in the LibDem’s case it didn’t depart as one).
The Liberals did exceptionally well in 1974 (esp February - indeed Feb 74 remains the third party high water mark in many southern seats), yet, not having a core constituency beyond the then-called Celtic fringe, their seat tally was barely into double figures.
An undercommented observation on recent polling is that, just as the Conservatives lose their base in the middle class of all ages and move toward a more evenly spread vote across the census categories, the LibDems are acquiring a base of educated working age people which has some of the same geographical characteristics. At vote shares into the 20%s, this ought to offer a sounder basis for winning seats across a broad swathe of the south.
A further feature of the more even distribution of Conservative support is that, at lower vote shares, they start to run into similar problems as the LibDems. This was beginning to show in some of the polls prior to the ascent of Bozo, which gave significant Labour seat leads for relatively small vote leads. There is a tipping point opening up beneath the Conservatives in the high 20%s-30% that is broadly the same as the one that always eluded the LibDems coming from below.
If Project Fear is even partly true, Bozo may get to explore that for himself.
As Rob Ford pointed out, anyone doing seat projections off current polling is either a fool or a knave.......
.....and as for those doing seat projections off hypothetical polling questions ("Imagine Britain has left the EU with/without a deal and Jeremy Corbyn is still leader of the opposition, and there's an 'r' in the month... - extremely dodgy at the best of times) a cad, a scoundrel and a poltroon to boot (or a tramp, a tart and an unfit mother, as the case may be.)
https://twitter.com/AllieHBNews/status/1160288482524356608?s=20
The only 'Sadiq' article I can find is the Khan one:
https://www.theguardian.com/politics/2019/aug/10/sadiq-khan-boris-johnson-parliamentary-poker
Flavibile is a spotty student in a bedsit and regularly predicts the Lib Dems sweeping screeds of Scottish seats from 3rd place. I’ll pay attention when the site owner needs to buy a razor.
....Smelling salts for HYUFD please
And Lord (George) Foulkes is getting medieval on Shadow Scottish Secretary Lesley Laird. The right-wingers smell Corbyn blood.
Income 2018: 259,425
Income 2016: 1,073,108
Donations I Leonard’s first year fell from 153 thousand to 36 thousand.
PPCs in full panic mode, fearing snap elections.
Move by HQ to prevent SLab branches selecting their own candidates: HQ wants to do selecting.
https://twitter.com/simongerman600/status/1114514352709820416
https://twitter.com/ElectionMapsUK/status/1160275495847628800?s=19
Maybe a statistics/politics student is what is needed to break away from broken UNS.
Don't we owe much of the stuff on the internet to such people?
Or are the MEPs in Brussels right now?
What Swinson needs to do is convince people the Liberal Democrats are the only meaningful and sane Scotland-wide alternative to the SNP and best placed to both protect the divers unions the people of Scotland voted for and deliver the promises made in 2014 and since ignored. Helpfully the Tories and Labour are doing a very good job of making this a reality without her having to lift a finger.
Flavible's predictions of some LibDem Scottish gains from third place could easily be realistic in such circumstances.
For the UK, for the EU.
PC is a slightly different kettle of fish, as I think even PC think Welsh independence is some way off, and LDs have long been pro devolution and geographic economic redistribution.
All alliances fall apart on local egos. It seems that no one hates a politician from X more than his neighbour with slightly different politics.
I think the Liberal Democrats need to show right now they are strong enough to be a serious force on their own. If they cuddle up to the SNP in an anti-Tory pact it could easily be counterproductive. There must be a number of seats where a Unionist vote decisive in the last elections for the Liberal Democrats swinging behind the Tories would topple an incumbent.
But tbh, I think this is like yesterday's ridiculous claims. It's kite-flying by people who very much want it to happen, and believe it worked in B&R because hey are pig ignorant of what happened there, but have no understanding that it wouldn't work where the dynamics are not solely Leave/Remain - which is basically Scotland, Wales, the semi-rural Norh of England and the West Country (or 'most places' for short).
As for a niche position, I'm not sure I agree. There are studies suggesting as many as a third of SNP supporters voted for Brexit including a substantial number of MSPs. Which implies here must have been a fairly solid vote for the EU among Unionists. Probably many of them Labour - but Labour is rapidly abandoning both unions. The Liberal Democrats could easily hoover up support from there, as they are already doing over what appear to be a number of issues.
And what about leading UKIP? That surely counts.
https://en.wikipedia.org/wiki/Richard_Braine_(politician)
A thin public CV, very thin, but went to King's, which is not too shabby. Perhaps he will surprise and do great things in the populist far right space.
Sorry but I have had to delete some of Ian's comments to get this in, which I really didn't want to do.
Agree good article, but also very much agree with Ian's points. Exactly my thoughts.
UNS assumes the parties have the same make up of voters as before with the edges flaking off to other parties as percentages go up and down. But we don't know that is the case this time and there is evidence that it isn't. Both the Labour and Tory party may well have changed their core vote with Momentum and Brexit and who is to say the LD vote of 20% is made up of the same 20% they had before they plunged to 5% for nearly a decade, particularly as although they have always been pro EU they are very, very much more clearly defined as the Remain party now. just look at their support in London now. Clearly very different to the past when they were on the same percentage
As evidence of this look at Scotland. When I was young Scotland was Tory. It became solid Labour, now it is solid SNP. Apply UNS in the past did not produce those changes. Something else happened.
UNS works when the voters are basically the same and you apply the percentage changes to those voters. If the voters change it doesn't work.
Now Justin might be right in JRM seat as might HYFUD in all the analysis he does. After all they at least have the starting position right so the probability is on their side. However I don't think the probable results are anywhere near as high as they think they are. We would be in a different world.
Meanwhile the LDs have won every single local by-election in which they've stood for three weeks' running now.
Has Meeks not noticed that Electoral Calculus (after all run by a mathematician) predict the PROBABILITIES of seats being won by each party ?
E.g., Belfast North has a probability of 0.6 DUP and 0.4 SF ?
Not DUP 1 and SF 0.
Brecon and Radnorshire has LibDem 0.4, Tories 0.36, TBP 0.18, Lab 0.05, PC 0.01
Not LD 1, Others 0.
When you construct the seat totals, it is better to weight by the probabilities in the adding.
That is another guide to the number of seats likely to be won by each party.
I would guess the "truth" lies between the total number of seats weighted by the probabilities and the total number of seats taking 1 for a win and 0 for a loss.
Of course, Baxter probably has the full pdfs, so he could do a still better job, but there is enough data on Electoral Calculus to do better job than implied in the wordy header.
Why, only today we have the Clown Prince promising more money for this and that.
What? No alternative proposal for leaving the EU beyond "You come up with something"? Look, another spending pledge!
[There is a case to be made for leaving with no deal, namely that the backstop prevents us leaving it without the EU's permission. The way the Clown is proceeding does not fill one with confidence in his abilities].
Given this, I think we need to break the prediction down into four groups:
1. Labour marginals (Con 2nd): How much bigger is the 2017 Lab voter switch to LD switch than the Con to BXP one? This determines the number of Cons gains.
2. Cons marginals (Labour 2nd): The opposite. This determines the number of Labour gains.
3. Cons (LDs second) By-election style tactical voting squeeze leads to LD gains. Similar possibilities for the Brexit Party.
4. Any Cons seat in Scotland: How many 2017 Cons voters will switch to other parties leading to a loss of that seat, typically to the SNP?
Add up the group gains and losses to get to the net loss or gain
On current projections, I suggest 1 will happen with Cons gains at the expense of Labour; 2 won't happen; 3 will happen but perhaps not huge numbers and very little for the BXP; 4 will happen with the loss of most of most Con seats in Scotland.
The question is whether (1) > (3) + (4)
Overemphasing tactical voting is also not wise as most voters are not that politically astute, they will vote for the party they support or have always supported, only a minority of generally highly educated voters will tactically vote.
Plus Electoral Calculus reflects the rise of third parties as much as the pollsters if they get a big enough swing
I think most people believe both front benches are rubbish.
The Borders seats also had amongst the highest Leave votes in Scotland as did seats like Moray the Tories would also hold
https://yougov.co.uk/topics/politics/articles-reports/2019/08/08/voting-intention-con-31-lab-22-lib-dem-21-brex-14-
http://www.electionpolling.co.uk/battleground/targets/snp
The London effect for the LibDems could be a weakness - it is possible that they might stack up a shedload of votes with good second places across Camden, Islington, Haringey, Lambeth, Southwark and the like with no seats to show for it.
We just need to remember the incredulity that the YouGov model predictions for Canterbury and Kensington produced (from me as well) before ruling out a larger batch of similar surprises next time.
Will Remainers vote 'smarter' (i.e. more tactically) than Leavers?
Will Remainer turnout % be higher than that of Leavers?
I think Corbyn's chances of replacing Johnson as PM depend on there being a clear YES to both questions.