Christmas fun

Naturally I wouldn’t expect any polls in the next few days, so in the meantime here’s a Christmas toy to play with – a nice graphical version of my swingometer. Before you click on any links, it uses Javascript and SVG, which Internet Explorer does not yet implement, so if you are an IE user you’ll need to download a plugin here to get it to work.

The basic version is here – Swingometer map, which is basically just the swingometer we already have on the site, but with a pretty map. There is also an advanced version here – advanced swingometer map, which does something a lot of people have asked for, allowing Scottish (and Welsh) figures to be entered seperately – the swingometer then calculates what the swing must be in England and Wales if the GB figures are X and the Scotland figures are Y.

This is very much a beta version, so let me know of any obvious problems you find – in particular I don’t know whether this is going to work nicely for Mac users (Safari does support SVG so it should). Note the figures don’t match those in the current swingometer exactly because they take account of Buckingham being the Speaker’s seat and the pact between the Conservatives and UUP.


89 Responses to “Christmas fun”

1 2
  1. Ordinary Swingometer very neat. I got this error message on advanced version:
    XML Parsing Error: not well-formed
    Location: http://ukpollingreport.co.uk/images/conmap_advancedbk.svg
    Line Number 3, Column 5:HTTP/1.1 200 OK

  2. Labour MP David Taylor has sadly died following a heart attack.

    As a key English marginal, the subsequent by-election may be an interesting barometer of voting intention in marginal seats ahead of the general election. It may also provide an insight into whether the uniform swing is representative of sentiment in marginals, and subsequently whether we are heading for a hung parliament or a conservative majority.

  3. I love it. It is better than watching television

  4. Anthony

    Thanks for the present!

    A couple of minor points.

    Because the figures for other parties are automatically adjusted when numbers are changed, to automatically produce 100%, one can’t replicate the rounded numbers that most polls give if they total 99 or 101.

    Is it possible for the borders between English and Scottish/Welsh constituencies to have a bolder line? That would help those like me whose knowledge of the geography of the Welsh Marches is a bit hazy!

    Great fun though!

  5. Agree totally with Glenn Otto. It’s brilliant fun! I’ve been acting out a fantasy to see just how many votes Labour would have to lose to be annihilated in Wales. It seems that there’s a hard core of Labour seats in the Valleys that will stay Labour to the bitter end but that the Tories and Plaid divide the rest of the country quite quickly with a sharply falling Labour vote.

    Any chance of a Super-Advanced version where we can play with vote shares in the English regions?

  6. Some of the constituencies show up grey.

    Is this a bug? Does it imply they’ve been ‘won’ by OTH?

  7. Not that people can’t elect from other parties, obv. but does it count all the Other votes as one party is what I was wondering…

  8. Further (very) minor point – the Scottish 2005 results are actual, not notional. Doubt if it’s worth doing anything about it though.

  9. Pete B (and anyone else who gets an error message), can you let me know what browser you are viewing it in. It’s all browser based code, and the support different browsers give to SVG differs a lot.(That includes what version of it you are using, for example – on Firefox 2 I can’t get the link to each constituency on the UKPR guide to work, on Firefox version 3 it works like clockwork).

    Oldnat – SVG support for typing numbers in is apparently really dodgy in some browsers, so rather reluctantly I decided to only use the up/down arrows. Just give the “others” score an extra or one less point – it doesn’t make any difference.

    Sindel – grey means other, it doesn’t count others as one party, it takes the highest individual other score in each constituency. If you see wierd constituencies turning up grey, tell me, I’ve rounded the percentage support of each party in each constituency to cut down on size, but occassionally that produces a dead heat in some seats which it reports as an errorneous other gain. As I spot them, I’m putting in more detailled figures for the constituencies affected to solve the problem, but there’s lots of combinations so you need to tell me!

  10. Oldnat –

    So are some other seats with no boundary changes, but no, I’m not putting that in!

  11. Thanks Anthony – good fun. Works just fine on my Mac.

    Fiddling around with the numbers makes it clear just how hard a task the Tories have to obtain an overall majority, if Labour scores reach into the low 30’s (pretty likely by polling day in my opinion).

  12. Anthony

    Thanks

    Odd result – East Dunbartonshire. When I enter figures as Con 18%, Lab 37%, LD 12%, SNP 27%, it shows as an SNP gain. Adding one to SNP from Labour (Lab 36%, SNP 28%) turns it back to an LD hold. Deducting one from SNP to Lab (Lab 38%, SNP 26%) makes it a Lab gain.

  13. Oldnat – try again, it should be behaving more rationally now!

  14. Oooh, a prediction I made (not too disimilar from the current polls) showed a gain for Others in Buckingham!!!

  15. Anthony

    It’s working now – pity, I quite liked a scenario in which the SNP gained East Dunbartonshire! :-)

  16. Great stuff-

    reinforces the fact of how thoroughly ineffective are all those piles and piles of Conservative votes heaping up on top of each other in SW/ SE and East England !!

    Even more reason for UKIP votes in their heartlands to vote on principle rather than tactically.

  17. Very Cool Anthony. Thanks.

    And a Very Merry Christmas to you. God Bless.

  18. I’ve found a humerous bug, though I presume it’s just a problem with UNS.

    Wondering what percentage of the vote it would take for SNP/Plaid to cleanup in Scotland/Wales, I ramp up the % for Others and find that with shares of the vote of

    Con 0%
    Lab 0%
    Lib 0%
    Others 100%

    The predicted seat tallies are:

    Con 205
    Lab 268
    Lib 95
    Oth 81.

    Hmmmm. There goes our reputation for democracy I fear.

  19. Seems like there won’t be a by election for David Taylor’s seat as the guardian and times are saying Labour is going to avoid the by election.

    Labour is using the excuse that by tradition you don’t have to hold a by election if it is so close to the election. The tradition holds you can leave a seat empty for 6 months but I don’t know when this tradition has been held.

    I looked at wiki at uk recent by elections. And in 1983 an MP died as late as february 12th and a by-election was held on march 24th before the 1983 election. In 1986 an MP died on december 22nd and a by election was held the following march 12th before the election in 1987.

    In 1992 with the tory govt about to lose its majority an mp died on november 4th and a by election was held on march 12th. Major according to labour’s thinking could have skipped the by election that year because the election wasn’t until may 3rd which was less than 6 months.

    It seems by elections have been held in the past. The 6 months might be a tradition that isn’t used and labour is using it to their advantage. It might technically be a tradition you can use but it seems to work against the spirit of the tradition since so many other by elections have happened and the tradition wasn’t meant to game the system. Major and Thatcher worked within the spirit of the tradition and allowed the by elections to go ahead. They also worked the system in delaying the glasgow by election from the summer.

    When was a time when the tradition was used and a seat was left empty for up to 6 months. Does anyone know of another time of a seat that will have been left open for so long until may 6th before an upcoming election. I’m sure other seats have been left open but they were much closer to the election.

    This seat would have been interesting so close to the election and the smaller parties were going to play here and it would have given a good read right before the election.

    Obviously this is a sad day I just found the politics of it interesting.

    RIP Mr Taylor and condolences to the family

  20. Joshua – all scenarios give you an “other” gain in Buckingham, since there are no Conservative, Labour or Lib Dem candidates standing there!

    Stephen W – for the SNP and PC you need to use the advanced version and increase the SNP and PC vote. Because you can’t differentiate the other vote for minor parties in the basic version it just keeps the SNP and PC vote static.

  21. James

    Thanks for that. I’d intended to look up those details, and you’ve saved me the bother.

  22. Anthony,

    One major issue, I can’t seem to get teh SNP to get a majority at Westminster….

    Peter.

  23. James, in 1983 and 1987 the elections were held at the four year point. Thatcher could have gone on for another year, so we weren’t “within 6 months” of the election, it was more “within 18 months”.

    The 1992 election was April 9th, not May 3rd. However it could have been held as late as the beginning of July, so November 4th is not within 6 months either.

    But as it happens we’re not “within 6 months” this time either as it could be held as late as 3 June. (3 June 2013 in fact, but I am sure Brown will not try to rely on the Triennial Act).

  24. Anthony,

    More seriously, there is a strange tipping point where when the SNP go from about 33% Lab- 32% SNP to 31% Lab- 34% SNP.

    The seats total goes from Lab 35 seats to the SNP’s 11, to 31 Labour to 16 SNP.
    Depending what happens elsewhere this could really have an impact on the election outcome.

    Peter.

  25. Peter, Anthony –

    Surely the model should be accurate at the extreme? If you put in 100% of the votes going to Others, surely they should get 100% of the seats. For the model to deliver any other result looks wrong.

  26. @James

    The clearest analogy here is 1979:

    Liverpool Edge Hill- sitting Labour MP died on 15th December 1978. David Alton picked up the seat from Labour for the Liberals.

    That By-Election took place on 29th March so almost *4 months* after the death of the sitting MP and only 1 month before the General election.

    Alton held onto it at General Election on 3 May. I cannot find any indication of what you say:

    “Seems like there won’t be a by election for David Taylor’s seat as the guardian and times are saying Labour is going to avoid the by election/ Labour is using the excuse that by tradition you don’t have to hold a by election if it is so close to the election”

    Other than in the Murdochian Times (what a surprise)

    http://www.timesonline.co.uk/tol/news/politics/article6968864.ece#cid=OTC-RSS&attr=2015164

    Two scenarios hold here:

    (a) General Election in March = no By-Election;
    (b) Brown hangs on = By-Election in March as a tester of the ground i.e. 2005 = Lab 45.5 Con 36.0 LD 12.1 Others 6.5

  27. Anthony

    Blaenau Gwent – the 2005 results show PC at 58.2% of the vote!

  28. John

    I meant 1997 election when major’s majority turned into a minority. The by election was in march that year after a conservative mp passed away in november 1996. The election was on may 1st not may 3rd.

  29. According to the list of polls Labour have staged a hell of a recovery in 2 days. What’s happened?.

  30. The calculate button seems to be missing in both Chrome and Firefox on Linux.

  31. Peter – from memory I don’t think that’s a mistake, there’s just a big group of SNP vs LAB marginals that need a very similar swing.

    John Lilburne – but there isn’t a party called “Other”, hence the problem. Realistically, swing calculators can’t predict where a fring party would do well and how many seats they’d get if they suddenly made a major breakthrough because it almost certainly wouldn’t be uniform.

    Oldnat – Ta. I thought I’d corrected that.

    David P – Huh? Where. All looks as normal to me.

    James D – that’s wierd. I can’t see any reason why that in particular would disappear when all the other text doesn’t (just to make sure you are looking in the right place, it isn’t actually a button – it’s just the word CALCULATE in the bottom of the last box).

  32. Anthony – it seems to be working now, thanks. And as others have said – great fun.

  33. Anthony,

    I didn’t think it was a mistake either.

    Its just one of those things that FPTP does. Two ot three percent of a difference could be the difference between not noticing there had been an election and the biggest turn around in Scottish Westminster politics in almost forty years.

    Peter.

  34. I try to avoid posting an opinion without some evidential basis – but it’s the holidays!

    The 2010 election may well be the most critical election ever for Scotland, but Scottish votes are largely irrelevant!

    If England selects a Tory Government (as it is clearly entitled to do), then the irrelevance of Scotland to the governance of the UK will be clear. If Labour manage to hang on through Scottish votes, then many Scots will continue to vote Labour for Westminster. On the other hand, the English may rebel at the subsequent GE!

  35. Re the SNP

    it can be shown that generally under FPTP, evenly spread parties don’t start converting votes inio seats untill they exceed 33.33% of the vote – then they start wining them rapidly.

    Rule of thumb for SNP:
    for every 3 percent gain below 33% they gain 1 seat.
    for every 1 percent gain above 33% they gain 3 seats.

  36. The Tories’ “short by” calculation seems to be wrong – it implies they need 328 for a majority.

    I know you intend going with 326, but in practice the winning post will be 323 if SF continue to abstain (and hold their existing 5 seats).

    One could even argue 322, if a “majority” includes the Speaker’s casting vote.

  37. Thank Anthony. Will try that again.

  38. @ OldNat

    “Scottish votes are largely irrelevant!”

    Why? Are they more or less relevant than any other collection of 59 seats in the country? They played a key part in passing many of Labour’s Acts, notably tuition fees in England&Wales. Perhaps the Liberals and the SNP will support a minority government of Labour, like in 1979?

    “…irrelevance of Scotland to the governance of the UK will be clear.”

    Oh, come on. Of the 19 PMs of the last 100 years, 5 had Scottish constituencies, including the current one. At one point in this Parliament, all 3 of the major leaders were either Scots or half-Scots, and of the 7 parties with more than one seat in the 2005 election, only one (the Conservatives) had their leader’s seat in England, 3 were from Scotland (Lab, LD, SNP), 1 from Wales (PC) and 3 from NI (SDLP, DC, SF).

  39. Well, my December average of polls gives scores of 41,27,19 for this map. It gives a very blue picture and a Conservative majority of 52.

    As ever, I sign off with a shameless plug for my blog :)

  40. @Ian S

    41;27;19 ?

    Are you including Angus Tory, er, Reid in that calculation perchance…

    40;29;19 and a Conservative majority using map of 18 for December excluding AR.

  41. Anthony-for me the GB/Scotland/Wales boxes are all unclickable.
    Any thoughts please?

    Looks like great fun.
    Thanks.

  42. It’s just Labour rose 5 points between the 18th and 20th.

  43. @Rob Sheffield

    Yes, I do include Angus Reid. Is there a good reason not to, except not liking the look of their results? They are open, they publish their tables in full, they publish their methodology. I see no reason not to include them in an average of polls.

  44. Rod – I was doing it in a stupidly complicated manner, done far more straightforwardly now and should be working correctly. It is indeed going for a formal 326 seats, rather than the practical reality of a few less than that.

    Colin – what browser (and version) are you using. It’s not Safari is it?

    David P – Ah! You need to compare polls to previous polls carried out by the same company, otherwise changes can be due to differences in methodology rather than actual changes in support. In this case the second to last poll was conducted by Angus Reid who for methodological reasons connected to their weighting and possibly the question they ask always show Labour at a much lower level than other companies.

  45. @Ian S,

    It’s precisely because of the look of their results – AR’s mean for Labour scores is way outside the margin of error of the other polls, so you’d be taking a mean from two substantially different sample distributions.

    Of course the other polls all have their own different sample distributions (with their own respective means and variances that are different from each other), so taking a mean of them is statistically dubious anyway.

    However, the differences between means are far less than their margins of error, so the statistical faux pas is less noticeable (and we all do it anyway :-).

    Secondly, ICM, Populus and YouGov have a good track record (as recent as the last Euro elections) so what reason does anyone have to trust AR (as a masure of voting intention), which is outside margin of error of these poll?

    If we presume AR is right about voting intention, we have to conclude that three pollsters with a long track record have suddenly started to over-estimate Labour by 5-6%, and that this has happened suddenly since the last Euro elections when these pollsters were within 0.3%, 0.3% and 2% respectively of Labour’s percentage. Implausible, so AR is wrong.

    One effect of including AR in UKPR’s weighted average is to bring Labour down by about 0.5-1%, which is probably not unreasonable given Labour’s past history of being over-estimated in polls (e.g. see last Euro election polls and results). But I would prefer to exclude AR and make that adjustment myself.

  46. Anthony,

    The WMA currently stands at 40:28:19, a Conservative Majority of 36.

    How does this compare with the WMA exactly 12 months ago and what conclusions, if any, can we draw from this?

    Thank you.

  47. Anthony- Internet Explorer ( Windows XP)-not sure of the version- and Firefox 3.5 which I have just downloaded.

    Same effect on both-cannot click/change party %s for GB/Scotland/Wales.

    Thanks

  48. Colin – well it definitely works happily in Firefox 3.5 (ditto with the plugin in IE), so it shouldn’t be a browser incompatibility (unless you have funny settings, put presumably Firefox has the standard ones if you’ve just downloaded it).

    Anyone else having any problems using it with XP?
    Is it working at all – are the constituencies highlighted in party colours?, can you click on individual constituencies?

  49. Anthony
    Yes the map shows constituencies in party colours.These are all clickable & the left side box updates on doing so.

    On loading , the top right hand (GB) box shows 33/36/23/8….and yet Overall Projections shows Con Majority 60. On clicking Calculate, this changes to Lab Maj 38.

    That’s about it.
    Thanks again.

  50. @Statto

    Fair enough reasoning. I’m going to keep them in my numbers for now because generally it’s the change in the vote that’s of interest as much as the actual number, and the change from November to December is quite similar whether you include Angus Reid or not (including AR the December average CON lead was 12.7%, down 0.2% from 12.9% in November – excluding AR the lead was 11.4%, down 0.2% from 11.6%).

    Also the general trend for pollsters for over-estimate the Labour vote in elections means that I think AR’s methodology may be on to something, and although I do think it’s probably going too far the other way it balances out the other polls reasonably well.

    I can’t wait for the election so we can get everyone’s predictions in. That will answer many questions and raise many more.

1 2