New swingometers

A brief note – I’ve updated the two graphical swingometers on the site so they are based on the 2017 election results. The basic version is here, and the fancy version that lets you put in separate Welsh and Scottish figures is here (the old version without the map isn’t yet updated).


828 Responses to “New swingometers”

1 2 3 17
  1. AW

    Have you changed the methodology as well?

  2. Looks a bit complicated.

  3. CambridgeRachel – nope, it’s just the usual uniform swing.

  4. @AW

    Isn’t a conclusion from GE that UNS is so good?

  5. Correction

    @AW

    Isn’t a conclusion from GE that UNS is not so good?

  6. AW
    Should we be expecting many VI polls in the next few weeks?
    I would think some newspapers would commission some because the polls themselves will form a part of the narrative with a minority government

  7. There’s nothing I like more than an updated graphical swingometer.

    Unless it’s two of the rascals.

  8. I’m surprised there hasn’t been a bunch of polls already, I guess all the polling companies need to review their methods

  9. @ CR

    “I guess all the polling companies need to review their methods”

    Just add a few more points to Cons to compensation for shy Tories. ;-)

  10. I would think there will be a few for the Sunday papers?

  11. @Exile

    They also have to make adjustments for feckless youth who it turns out aren’t so feckless after all. And possibly for the GOTV efforts on the day of the not-feckless turning out in their thousands and swamping local organisers.

    If as mooted there’s to be a Conservative Momentum, a Blue Rinsum or whatever, need to allow for armies of boomers on their mobility scooters getting out the vote, those things can shift and cover a lot of ground…

  12. @ Carfrew

    Wouldn’t that be Conservation of Momentum?

  13. As the Yougov MRP was so successful, does that mean we will soon stop seeing polls, with everyone moving to MRP?

    And how does MRP stand with the British Polling Council? I didn’t see any of the tables we all enjoy dissecting so much, I guess that will all disappear with MRP?

    We will just have a nice map to look at and ponder why on earth Canterbury is red and not blue with no data to understand that the reason for that is that the young are voting one way, and the old a different way?

    And how do you do all those interesting supplementary questions with MRP?

  14. @Richard

    MRP needed 7,000 interviews per day.

    Way to expensive for conventional polling.

  15. @ExileinYorks

    Ha ha ha. Yes.

    Are those the ones that are too shy to vote!?

  16. @Exile

    “Wouldn’t that be Conservation of Momentum”

    ——–

    As opposed to energy? Yes, if you like…

  17. Swingometers get me all nostalgic for the days of dear old Bob McKenzie and, occasionally of this parish, the equally dear David Butler. Invented originally by someone else, McKenzie and Butler perfected the swingometer and launched it on the British public in the 1959 election. It then became part of the furniture in the BBC’s election coverage for the next 25 years, as did McKenzie and Butler. McKenzie was one of the first psephologists to become a familiar face and household name in the TV age. He was a Canadian professor of politics and sociology and while obviously an intellectual and serious academic, he made politics and elections digestible and understandable for the laymen who only got interested at election time. The forerunner of people like Peter Snow and Jeremy Vine, I suppose, although, intellectually, probably closer to the Kellners and Curtices of this world, as was David Butler.

    Another little anecdote that reminds me I’m getting old! Bob McKenzie and Cliff Mitchelmore, where art thou??

    :-)

  18. Even I remember Cliff Mitchelmore and I was born in the 1980s!

  19. Michelmore!

  20. I noticed the Swingometer played very little role in the BBC election coverage this time round, compared with previous elections. Is this because the swing was so far from uniform that the swingometer was less applicable?

  21. Inertia would be the Conservative version of Momentum perhaps

  22. Ah well, as I said at the end of the last, brief thread, normality will soon be with us.

    Same government as before: despite all the potential drama, reality bites and we’re back to real world.

  23. Uniform national swing would have approximately correctly predicted the result. It worked fine.

    The fancy models that suggested the Tories would increase their majority even with a lower margin of victory than in 2015 are the ones that failed.

  24. I thought that labour had a boundary advantage but playing about with the swingometer it looks like the conservatives have a big advantage. Is this because of the collapse of the dems? Or is it the rise of the SNP? Or was it a load of tosh to begin with?

  25. I wonder if the political parties will now drop their internal polling and just use the YouGov MRP for elections. It’s free and works.

  26. @CambridgeRach

    “I thought that labour had a boundary advantage but playing about with the swingometer it looks like the conservatives have a big advantage. Is this because of the collapse of the dems? Or is it the rise of the SNP? Or was it a load of tosh to begin with?”

    ——–

    I confess I’m curious about this Rach. Some have said that the loss of Scottish seats in 2015 more or less ended the Labour advantage. Anthony says that in this latest GE, Labour improved their position, and since FPTP favours the victor you have to equalise the vote shares to see who has the advantage.

    You can see how students choosing to vote where their vote counts more might help Labour a bit, but even allowing for victor’s advantage under FPTP, when you look at how just a 2.4% lead gives the Tories 56 seats, it does make you wonder just how much of an advantage Labour really have. Especially given the micro-targeting thing, that saw Tories gain quite a few seats last time despite not increasing their vote much.

    I also wonder how it changes… How does the Labour advantage change as Tories and Labour combined vote share changes? Meanwhile, there’s more than one effect with electoral bias: there’s constituencies of different sizes, and then there’s the efficiency of voting within a constituency. The former is more clearly an issue than the latter, which is dodgy, penalising people for not voting in ultra safe seats. I don’t know how much each effect contributes to the supposed bias.

    It’s hard to find useful info. online. But I did find a couple of things I shall now post…

  27. From the FT…

    “Did the parties spread their votes efficiently?

    Some of the excitement around a possible “Corbyn surge” as Labour ate into the Tory poll lead was tempered by concerns that Labour may indeed gain votes, but only in seats where it was already comfortably in front.

    The early evidence is that these worries were misplaced: many of Labour’s biggest swings against the Tories came in marginal seats, with minimal padding of healthy leads.

    The party either overturned or significantly reduced the Conservative margin even in places like Canterbury, which has voted Conservative since 1918.

    Labour has also widened its lead in areas where its 2015 margin was precarious, such as Ealing Central and Acton. The chart below shows how Labour has performed against the Conservatives in seats where they and the Conservatives were the top two parties in 2015.

    There had been fears that Labour would gain votes in safe seats. Darker red dots and lines show greater changes in the Conservative-Labour margin. Many of the reddest constituencies had margins of around 10 percentage points in 2015.”

  28. Here’s an article from the LSE…

    http://blogs.lse.ac.uk/politicsandpolicy/electoral-bias/

    “Equalising constituency sizes is likely to reduce the electoral bias in favour of Labour but only minimally so. Much of the bias is likely to remain because it is due to other factors such as turnout and vote distribution.”

  29. QUOTE (Cambridge Rachel) “I thought that labour had a boundary advantage but playing about with the swingometer it looks like the conservatives have a big advantage. Is this because of the collapse of the dems? Or is it the rise of the SNP?”

    ————————————————–
    It’s a bit of both in a way. In the late 90’s and early noughties the boundaries favoured Labour as there were quite a few seats that went LibDem in areas that Labour would never take. There was no opposite effect in Labour seats that the tories could not take.

    The collapse of the LibDem vote saw many of these seats reverting back to tory.

    In Scotland, initially, the SNP took almost everything, costing Labour a lot of seats. This wasn’t immediately a disaster as the SNP are essentially social democrats and far more likely to vote with Labour than the tories – BUT – what this has lead to is people voting on the basis of geography rather than class, which has made it easier for tories to take Scottish seats as they have done in this election, bucking the trend of the rest of mainland UK.

  30. FT article has a number of interesting charts besides the voting efficiency. It shows the increasing polarisation according to age, where the UKIP votes went, how the modellers fared (Singh, Fisher etc., though not our Catman!!), how education remains important after adjusting for age, the role of class in elections is getting weaker, and the impact of Brexit.

    One thing I found interesting is that while Cons tend to do better when there’s a higher share in poor health…

    “Our analysis finds that the link between a Tory/Labour swing and health is slightly stronger than the relationship with the DE social grade alone, so it may be that the impact of chronic poor health on attitudes and outlook may also make people more likely to vote Conservative than Labour, all other things being equal.”

    https://www.ft.com/content/dac3a3b2-4ad7-11e7-919a-1e14ce4af89b

  31. “The collapse of the LibDem vote saw many of these seats reverting back to tory.”

    ——–

    Thinking about the role of the other parties makes my brain hurt. Let’s take the case of a party like the Greens in 2015. Say for sake of argument they might secure only a couple thousand formerly Labour votes per seat, but that can cost you a lot of marginals. A smallish split in your vote has quite an effect.

    Of course Tories then suffered their own smallish split with UKIP, which then grew during the Coalition. But then they started taking votes from Labour too. When UKIP collapsed, Tories benefitted most, giving them quite a few marginals but the efficiency will be further complicated by UKIP standing down in some places, summat that may not happen every time. And naturally LDs split across Tories and Labour too…

    This is without considering how behaviour might change in response to boundary changes. People who formerly didn’t bother to vote because no point in a safe seat may now spring to life. Well, not me, obviously, but you see the point…

  32. Mark

    “BUT – what this has lead to is people voting on the basis of geography rather than class”

    I don’t think it’s helpful to classify voting as being on a single axis. People’s votes are determined by a complex set of factors.

    If by “geography” you mean that those voting within the Scottish polity vote on different axes, than those in other polities, that is a relevant consideration – but it has been the case since 1707, so it may not be worth much of a comment!

  33. Can I just take a moment to say thank you for all the work you put into this site. It’s been fascinating and very informative.

  34. Carfew/CR – yes Labour used to get oodles of seats in Scotland for a 40%ish share and the Tories get none for 20ish% in Scotland.

    In 2015 Labour got 1 seat in Scotland for 24% vote share.

    Straight UNS would suggest the Tories now have an advantage even on current boundaries.

    It would gain even more if the review went ahead with current proposals based 600 seats and the old electoral register – using the new one with all the extra registrations might be different, I don’t know.

    Would be surprised if seats reduction went ahead so a new review will be needed and therefore the latest ER can be used, probably not in time for the next GE which would, therefore, be fought on current boundaries.

  35. Given how small many of the majorities are in seats held, these swing calculators are for fun only. They are likely to indicate an overall result that will be proved to be very different with the actual result.

    The parties target seats and it depends on how successful this is, as well as how good the candidates are, what the local issues are etc etc. The Tories enjoyed about 3% additional votes over Labour and only lost out on a majority by a few thousand votes over a dozen seats.

    According to the swingometer, Labour need to be about 5% ahead of the Tories to achieve a very small majority.

  36. R Huckle, yes an a 2ish lead to have the same number of seats but a 2% Con-Lab swing would probably be accompanies by other swings reducing the Tory advantage.

    Tories losing a couple to LD Richmond the obvious one); Labour ‘wasting’ less votes to lose to the SNP in Scotland etc.

    Chances are, therefore, in a hypothetical 2018 GE on current boundaries Labour get an OM for a 3%ish swing and largest party if they ahead by 1%, rather than the 2% UNS suggests. However, in a mirror opposite of what could have happened in 2010, could be Lab most votes but Tories most seats.

  37. @JimJam

    Yes, when you put it that way, the change in Scots voting is pretty demonstrative.

    Are there any figures anywhere that show the supposed bias in the last few elections, including the most recent? ‘Cos I can’t find anyway, but there are surely peeps with much better Google-fu than me?

    I’m still getting my head around it all, but am now beginning to see why I find it disquieting, by thinking through the different things that affect electoral “efficiency”.

  38. @oldnat

    “I don’t think it’s helpful to classify voting as being on a single axis. People’s votes are determined by a complex set of factors.”

    ———–

    He didn’t classify it as being on a single axis, that’s you inventing summat to have a go at. He’s just saying that of all the axes, geography seems to be a dominant factor. You could of course show otherwise if you can…

  39. Regarding Farron’s resignation, I agree Cloudspotter’s position: people are entitled to know the religious views of a candidate, but it does not disbar them from office in any way.

    Farron’s problem was a simple one: the LBGT community were concerned that if elected, and perhaps close to power, his personal views might undermine their legal position. That concern was legitimate, in my view.

    Farron responded unwisely, and that changed perceptions.

    It is interesting that politicians used to make it very clear that they held Christian beliefs, presumably because they thought that to do so would help win votes.

    In today’s much more secular society, that tendency is much diminished.

  40. @ HIRETON – thanks for the Prof Curtice link on Scotland y’day. I agree with his analysis (posted similar findings on page before in y’day thread).

    Scotland was described as 3D politics, in order of impact on election:
    1/ IndyRef (biggest tactical vote factor)
    2/ Left-Right (SNP/SLAB v SCON with SLIB niche player in a few seats but more willing to help SCON in Scotland where as more willing to help LAB in E+W)
    3/ Brexit (as we saw across whole UK, wasn’t a factor and hence least important this time)

    Tactical voting tends to pull down %s (e.g Greens/LDEM in UK who shifted to LAB and SLAB who shifted to SCON). So if IndyRef is put on back burner, Brexit happens (so out for next time) then Scotland reverts to Left-Right. My analysis suggests SLAB gave around 2% swing to SCON and with no tactical voting SCON and SLAB would have been closer with SLAB possibly even beating SCON in %

    SCON and SLIB are pretty well established in defined seats via historic presence (with a large dollop of established tactical voting in 2015 and 2017).

    SLAB versus SNP will be very interesting if we move back to left-right as the single dimension. As many others are saying LAB’s route to a majority in next GE will probably require gaining many more seats in Scotland. SLAB+SNP represent 2/3 of the Scottish vote and if one party can unite the Left vote (as SLAB did prior to 2015) they will not only do better in next GE but will also have a big impact in next Holyrood election.

    P.S. Since Education is devolved the Corbyn mania is Scotland shouldn’t be due to free Uni fees. I’ll look at the high youth vote seats later as I did not add that filter to my Scottish model.

  41. Carfew/Rachel/Jim Jam – “bias” in the electoral system is a very changeable thing. It is a combination of several factors: turnout, vote distribution, different seat sizes and the impact of third parties. The LSE article linked below is from a time when the system was skewed towards Labour, however in 2015 it moved very strongly towards the Conservatives. Turnout and seat size continued to work to Labour’s advantage, but vote distribution and the impact from third parties strongly favoured the Conservatives producing a very large Conservative advantage. Given a further significant change in the pattern of support this time round (SNP no longer taking almost all the seats in Scotland, Labour moving into a winnable position in many more seats and the collapse of UKIP) that advantage has weakened again in 2017, so the Tories now have a significantly smaller advantage.

    From an initial look I don’t think the boundary review as it stands will actually change the balance between Lab and Con seats that much (Tory seats still have larger electorates, so that may be just happenstance and the way the boundary commissions happen to have drawn the seats). That may mean the Tories spend less political capital trying to push them though (the DUP don’t like the proposed boundaries, so it would be a challenge!). Alternatively if the new boundaries don’t help or hurt either side they may go back to being an uncontroversial technical change that goes through on the nod…

  42. Thanks Anthony,
    And I may take the opportunity of you being live on here as it where to voice my appreciation for running this site.

  43. Will there be another GE in 2017? If you think the chance are much higher than 25% and fancy a flutter then..

    https://www.betfair.com/exchange/plus/politics/market/1.132100025

    The odds will probably move a little wider on the announcement of DUP deal and passing of the QS but that looks pretty well priced in at these levels. Several small bumps to get past the QS and then some pretty big hurdles (inside the CON party) once Brexit negotiations get started and we need to start making sufficient progress on the exit bill, etc…

    IMHO the FX markets will stay focussed on UK political risk, hence my ongoing interest… (plus messing around with prediction models has become mildly addictive)

  44. Without opening up an old debate the issue the Government (Cons led) was criticised on the most IIRC was using the ER rather and Census.

    The accusation was this suppressed the number of people the review is based on in safe Labour seats where populations are more transient and many don’t register. Hence in having to equalise within 5% (down from 10%6 Labour lose seats as 6 in Cities become 5 for example.

    The surge in registrations may change this if the new list uses but too early to tell.

    The argument went that the census was fairer as non-registered voters should not in effect have their, and registered votes in their seats, representation diluted. Or put another way Labour MPs would have as many registered voters as Cons ones on average but actually more constituents.

    The Census, of course, tends to be less current than the ER is an argument the other way.

  45. Since 2015 the system/boundaries have been “biased” towards the Conservatives, and ceteris paribus the new boundaries would increase that bias.
    The word bias is misleading, of course – it is important to bear in mind that the advantage could go back the other way depending how the distribution of votes changes.
    But if you enter 42%-42%-8% in AW’s swingometer you can see that Con win 15 more seats.On 41-41-10, Con win 22 more seats.

  46. Sorry, I see now that AW had already answered, and better and more accurately than I could!

  47. @Anthony W

    I was very disappointed that you didn’t allow the last thread to run on a little longer. I was hoping we could have clocked up 3,000 posts, thereby establishing a record for the longest thread in UKPR history!!

    If we’d have got near, I was planning to fire off a quickfire 10 posts so I could have been the 3,000th poster. No doubt some sort of award would have been on offer. A miniature gold swingometer, perhaps?

    :-)

  48. @ CARFREW – FT article.

    They use the word “link” but clearly infer that correlation is the same as causation which it is not. The graph showing the correlation of poor health and shift to CON is almost exactly the same as the previous graph showing higher Leave areas moved more to CON (which itself takes in the graph of areas with high UKIP moving more to CON).
    Correlation is not proof. The truth is we can never be sure but data mining results to look for “proof” is very dangerous if you start looking for correlations on items such as health!!

    Similarly with education level. The article itself puts a big warning on the new assumption that highly educated people prefer LAB. Firstly it was a swing measure, the scatter plot shows the correlation to be weak and as they say themselves adjusting for other factors removes/reverses the finding! It would be interesting to see if University towns represent outliers biasing the LAB-education link. Universities employ a lot of faculty who probably live/vote locally and will be highly educated – lecturers will be educated to PhD (DR) level!!!

    Labour making gains where it matters – all the seats they pick out are student towns!! CON’s targeting failed as the share they hoped to gain from UKIP in their target seats was not enough – in many places they reduced LABs majority but it only gained them a tiny number of seats (Mansfield, Derby NE)

    The big unanswered question is to what extent did complacency have a role. As the modellers analysis shows only YouGov were close (but their final poll in the Times gave reassurance that CON would comfortably win). Turnout was high (much higher in younger LAB leaning buckets but lower in older CON leaning buckets than the EU referendum).

    The only safe conclusion to make is LAB ran a much better campaign – shifting the debate to domestic policies, motivating their core and new voters. CON ran an awful campaign – no sweeties in the manifesto, terrible effort to sell their manifesto, failure to keep the focus on Brexit and maybe not doing enough to motivate their core voters.

    We’ll never know whether “Complacent CONs” reversed “Shy Tories”… well not until we have a do-over!!

  49. Catmanjeff

    Thanks for posting that analysis of green voting. It very neatly answered one of my questions.

    I wonder if the Green party will re-think their emphasis on a ‘progressive alliance’ on the back of their poor performance? It doesn’t seem likely. I don’t really understand their current approach. The Green party badly needs a more proportional voting system and media talk about a ‘return to two-party politics’ hardly increases the pressure for electoral reform, nor does a shrinking Green vote.

  50. The main 2 reasons the Tories have better vote to seat ratio after 2015 would be:

    1) South West England: Lib Dems held 57 sea

1 2 3 17