In the first week back there have been seven polls. The regular weekly Ashcroft poll hasn’t fired up yet, and none of the phone pollsters did fieldwork over the first weekend of the year, but the daily YouGov and twice-weekly Populus polls are off:

Opinium/Observer (2/1/15) – CON 32%, LAB 33%, LDEM 8%, UKIP 17%, GRN 4%
Populus (4/1/15) – CON 34%, LAB 35%, LDEM 9%, UKIP 12%, GRN 5%
YouGov/Sun (5/1/15) – CON 31%, LAB 34%, LDEM 7%, UKIP 14%, GRN 8%
YouGov/Sun (6/1/15) – CON 33%, LAB 33%, LDEM 7%, UKIP 13%, GRN 8%
YouGov/Sun (7/1/15) – CON 32%, LAB 33%, LDEM 7%, UKIP 15%, GRN 7%
YouGov/Sun (8/1/15) – CON 33%, LAB 33%, LDEM 8%, UKIP 13%, GRN 7%
Populus (8/1/15) – CON 33%, LAB 34%, LDEM 8%, UKIP 14%, GRN 6%

All the polls so far are showing a tight race, with the Labour party averaging a very small lead – the current UKPollingReport average has CON 33%, LAB 34%, LDEM 8%, UKIP 14%, GRN 6%. YouGov started the year by changing their methodology to include UKIP in the main prompt, but it doesn’t appear to have had any impact on their level of support (if anything they are marginally down – last month YouGov had them averaging at 15%).

Start of the Campaign

The political parties started the campaign, the Conservatives largely on the economy and spending, Labour on the NHS. In terms of believability at least Labour’s claims went down better – by 48% to 32% people thought the claim that the NHS could not survive five more years of David Cameron was true, and by 42% to 27% that the claim the Tories wanted to cut spending back to 1930s levels was true. For the Conservatives, by 33% to 22% people believed that Labour had made £20 bn of unfunded spending commitments, but their claim that they had reduced the deficit by half was disbelieved by 49% to 24%.

Those, of course, are responses when respondents are prodded and forced to consider some party political claims and have an opinion. Whether anyone actually noticed or cared and whether anything made any difference is a different matter. I doubt we will see much change in the positions at the start of the week when the Conservatives led Labour on the economy by 15 percentage points, and Labour led the Conservatives by 12 points on the NHS, little different from other issue polls over recent months. Where there has been a significant change in the salience of issues. Presumably on the back of headlines about A&E waiting times and crisis in the NHS the proportion of people saying that health is one of the main issues facing the country has risen to 46%, in third place behind the economy and immigration and up 13 points since December. If health remains high on the agenda it will be good for Labour.

OfCom major parties

As I wrote about yesterday, Ofcom released their draft guidance on which parties should be treated as major parties in terms of election coverage. It’s open for consultation so may yet change, but as things stand UKIP will be treated as a major party (meaning broadcasters will have to give due weight to reporting them in editorial coverage), the Green party will not.

Projections

Latest projections from Election Forecast (Chris Henretty et al’s project), Election Etc (Steve Fisher’s project) and the New Statesman’s May2015 site are below. All are predicting a hung Parliament, all with Labour and Conservative within 10 seats of each other. Note that Steve Fisher’s method doesn’t have anyway of factoring in the SNP yet, so will change very soon. I think we should also be getting a regular seat projection from the Polling Observatory team in the next week or two.

Election Forecast – Hung Parliament, CON 284, LAB 281, LD 26, SNP 34, UKIP 3
Elections Etc – Hung Parliament, CON 294, LAB 297, LD 29, OTH 30
May 2015 – Hung Parliament, CON 273, LAB 281, LD 24, SNP 46, UKIP 3


197 Responses to “Seventeen weeks to go”

1 2 3 4
  1. Unless something electorally drastic happens, it looks like being the closest General Election in (my) living memory.

  2. Interesting that the Election Forecast has moved from a regular Labour small seat majority over recent weeks to a Conservative lead for the last week.

  3. A good way to start the final furlong-neck & neck.

  4. First time Greens up to 6% on the “UKPR polling average”. Meanwhile, my own calculation is that a simple rolling mean of the last 20 polls gives Greens on a record high (6.15%), and of the last 5 (7.2%).

  5. Wilson vs Heath I remember I had 10 pounds on labour 7/2 .. 1970s Wilson won by about 6 ,. The bookmakers hadn’t sorted out rules back them and I was worried I wouldn’t get paid out if Heath didn’t resign .Now of course its simple ( in some ways ) most seats and you get paid out . Back then it was “who will win the election “

  6. The elections etc forecast is based on vote share of tory 34.0,labour 31.2,libdem 11.4,ukip 12.6 others 10.8.

    So compared with current polls,tories up one ish,lab down 3 ish,libdem up 3.5ish,ukip down a tad.

    My gut reaction is lab too low and lidem and ukip too high but tory about right(tho poss also too low).

  7. I have them as

    Con 272
    Lab 286
    LD 28
    Oth 47 (excl NI)

    So, not far off those other pollsters….

  8. So it’ll be a LAB/LD/SDLP government with the support of SNP/PC when required.

    Quite a coalition!

  9. Question for AW on UKPR rolling average: are more repeated pollsters weighted lower? As it currently stands, as I understand it, Yougov, as a daily poll, is going to have far more of an impact on the ‘poll of polls’ than other pollsters who only have weekly results. Is this intended/desirable?

  10. Elections etc looks a little odd. 591 seats between Con and Lab, with 29 to Lib? Nah!

  11. @ Anthony

    Thanks for posting your updated polling average and the seat projections for the various models. I hope this will lead to some discussion of the differing assumptions made by each of the models.

    A while ago @ Catmanjeff posted some comments to the effect that he could point to (CUSUM-based) statistical support for the claim that there was a small upturn in the Labour VI at the time. Others accused him of cherry picking, and as a way of avoiding this I suggested using as a snapshot the set of polls contributing to the current polling average.

    I undertook to use this approach to look out for departures from the (linear) trends for each of the parties over 2014. (Bear in mind that if there is no move away from these trends we are headed for a small Labour VI advantage with Labour picking up most seats in the election).

    With the current set of just eleven polls (with 4 repeated from the previous set) we don‘t have a strong basis for making any projections. However, for what they are worth here are the results:

    Conservatives – Still on trend but edging up towards the 90% confidence limits (8 of the current polls above trend and 3 below)

    Labour – Back on trend (6 above; 5 below). This contrasts with the calculations for the December batch of polls, which had them above 2014 trends. So, they briefly edged outside the confidence bands only to return securely within the boundaries.

    LibDems – Still on trend (7 above; 4 below)

    Ukip – reliably below trends for the second batch running (2 polls above trend and 9 below)

    Greens – Still on trend (6 above, 5 below)

    So, it is beginning to look as if there is statistical support for the informal observation that the Ukip VIs are starting to top out. Also there is no sign yet of the accelerated swingback that is projected to occur in the last few months before the election.

  12. On another matter I was interested to see a different kind of GE prediction offered by Tony Travers at:

    http://blogs.lse.ac.uk/generalelection/local-elections-party-prospects/

    Using adjusted vote shares from local elections he argues that “Labour are some way off the level that has been historically necessary for opposition parties to replace the government in UK elections.”

    What his analysis seems to lack (at least in this brief summary) is careful consideration of the *size* of the majority that has to be overturned to bring in a change of government. With substantial pre-election majorities it is easy to see that a big groundswell would be needed to send the removal men to No. 10.

    But it may not require such a heave when the senior coalition partner starts without a majority and the smaller group itself faces big drops in the coming election.

  13. I’m a longstanding New Statesman reader. However, I’m b*ggered if I can understand how they think the SNP will win Orkney & shetland.

  14. The interesting thing with these models is that they have some substantial methodological differences, which basically offset each other. As far as I’m aware:

    – Elections Etc uses swingback (helps Tories) but also uses UNS (helps Labour)

    – May2015 uses a mix of differential swing and strong transition (bad for Labour) but assumes zero swingback – essentially a nowcast (bad for Tories)

    – Election Forecast uses an extremely complicated model based on a bayesian mosaic (which I’m busy unpicking – Chris Hanretty is an extremely nice chap and very helpful) but there’s something I don’t quite get in the marginals…

    I’m also working on my own model, which (to put it mildly) is rather fidly…

  15. @DavidinFrance

    “So it’ll be a LAB/LD/SDLP government with the support of SNP/PC when required.”

    Sounds like a sort of “anybody but the Tories and UKIP” party. Might sign up to that!

  16. @ MOG

    I think that’s a quirk of how their model deals with Scotland… I Agree with you 100%, fwiw

  17. Incredible video of the police entry into the Paris supermarket here http://www.francetvinfo.fr/images/videos/

    The suspect clearly had Butch Cassidy and the Sundance Kid aspirations. Also clearly shows that at least one of the hostages was definitely dead before the police entered. I expect it to be confirmed that all four were dead, as the suspect appears to be focused on fighting the police rather than any last minute exectutions.

    Francophobe that I am, you can’t help respecting and admiring those French police officers.

  18. Miserable Old Git

    I’m a longstanding New Statesman reader. However, I’m b*ggered if I can understand how they think the SNP will win Orkney & shetland.

    I presume it’s to do with then using Election Forecast figures. Their predictions for individual Lib Dem seats are deeply weird because they use the Ashcroft figures where they exist but don’t seem to extend them by implication to similar LD seats. Their general model predicts a certain number of LD seats at a certain amount of risk, but when individual polls show that the seats ‘most likely’ to fall are actually safe (for example the SW London ones), they disappear from the list:

    http://electionforecast.co.uk/tables/LD_seat_losses.html

    and those which haven’t been polled are kicked down the pecking order to compensate, even though the reason they haven’t been polled is that they are mega-safe.

    So according to their current figures O&S is less safe (despite them giving Carmichael a 12 point lead) than Cambridge (with only a 3 point LD lead).

  19. @MOG / Roger

    Presumably the simple approach of taking the VI the LDs had in 2010 (23%) then applying the losses nationally (let’s say two thirds) and seeing if two thirds of the LD losses go to the SNP.

    In Orkney & Shetland, that would be a shift of 8,000 away from LD, with the lion’s share going to SNP, so it’s an SNP win.

    Possible, but not what we expect.

  20. UNICORN
    I wonder if you could weight VI projections by reference to verifiable or “felt” factors, especially in NHS, economy/pound in your pocket and migration. Their salience in the next four months will be affected by daily headline news and commentary, but also by debate and policy commitments affecting VI for the main parties. The crucial movement seems to be that of UKIP swing back to Con and Lab.
    The in out debate and referendum on Europe does not seem to me to have much salience, but as a manageable factor in respective intent and believability of the two parties in controlling the level of immigration and rights of immigrants this could be the game changer.

  21. Related also to the integration of immigrant populations, especially in the light of the events of the last few days in France.

  22. ‘intent and believability of the two parties in controlling the level of immigration and rights of immigrants this could be the game changer’

    Really? I suspect the normal issues will win – 1) hip pocket nerve (wages, rent, mortgage) 2) Health and Education

  23. @Unicorn

    how many defections do you think there’ll be from this putative labour government?

  24. John Pilgrim

    I see that Marine Le Pen has called for the suspension of Schengen .

  25. @ NC

    You make some very interesting comments about the various models. I hope you have time to answer a few questions,

    “Elections Etc uses swingback (helps Tories) but also uses UNS (helps Labour)”

    As I see things, it not the use of UNS per se that makes it too optimistic about Labour prospects. It is mainly the application of UNS without making separate provision for dealing with Scotland and the SNP surge. As Anthony says, Stephen Fisher is promising to fix that very soon.

    “May2015 uses a mix of differential swing and strong transition (bad for Labour) but assumes zero swingback – essentially a nowcast (bad for Tories)”

    Why do you think Strong Trsnsition hinders Labour? My calculations suggest it is broadly neutral in its effect.

    “Election Forecast uses an extremely complicated model based on a bayesian mosaic (which I’m busy unpicking – Chris Hanretty is an extremely nice chap and very helpful) but there’s something I don’t quite get in the marginals…”

    Do you know of any detailed description of the model in the public domain? There are hints and clues on their website but no really specific details about how they calibrate their various adjustments. My impression is that the forecasting part of the model is much cruder than the database of how things stand at the moment. For example, for all parties other than Ukip the VI is assumed to rise or fall by a constant amount between now and the election (constant within one point, that is). I suspect this kind of consistency would not emerge from anything other than a hand-wired tweak.

    “I’m also working on my own model, which (to put it mildly) is rather fidly…”

    What distinctive features does this have?

    Finally, do you know anything about how Electoral Calculus works? I,can find no technical information at all on their website.

  26. ” Ofcom released their draft guidance on which parties should be treated as major parties in terms of election coverage. It’s open for consultation so may yet change, but as things stand UKIP will be treated as a major party”

    “May 2015 – Hung Parliament, CON 273, LAB 281, LD 24, SNP 46, UKIP 3”
    _____

    SNP 46 UKIP 3……..

  27. @Allan

    It’s obvious that the SNP are a major party in Scotland and a zero party elsewhere.

    I’m sure the SNP will take part in some Scottish debates and play a major role. I

  28. @ John Pilgrim

    “I wonder if you could weight VI projections by reference to verifiable or “felt” factors, especially in NHS, economy/pound in your pocket and migration.”

    I think this is the kind of analysis that lies behind the Goldman Sachs report that was being discussed here a couple of days ago (at least with respect to economic influences). The effects are volatile and difficult to quantify. Their influence of VI depends on how effective different campaigners are in setting the agenda, and this is impossible to capture in a set of calculations. For example, with A&E pressures and seasonal flu risks I would expect the NHS to have high salience over the next couple of months (presumably helping the Labour VI). But this could then fade into the background as the winter problems recede. Or perhaps not, if further private contractors ‘hand over the keys’. Events of this kind are not predictable and so in my view there is no point in trying to build them into any model. Better to know what the numbers suggest and tweak the outcomes based on how the campaign unfolds.

    “The crucial movement seems to be that of UKIP swing back to Con and Lab.”

    I don’t think there is good evidence that this is where the ‘crucial’ action will be. Far more important will be the extent to which the Conservatives can feel in tentative Labour supporters and vice versa. The analyses I have done suggest that (plausibly sized) Ukip swings don’t have a profound effect on the Labour/Tory balance (perhaps in part because Ukip support both comes from and goes to both parties to some degree).

    Also more important by a long way will be the shape of the SNP trajectory, and how many seats they take from Labour in Scotland. If the SNP have peaked too soon, ending with May VIs similar to those for Labour then 20-30 seats could shift back to Labour.

    Finally, there is quite a lot of uncertainty about LibDem prospects. Most of the models see them holding on to 20-30 seats. But this is based either upon using the (unsustainable) UNS model or relying upon non-standard Ashcroft constituency polling data – specifically VI support figures based on constituency-specific VI questions rather than the standard polling data. If the models instead used the latter (Q2 data) they would project the LDs getting no seats at all. Well, incumbency effects should prove this wrong. But it seems to be no more than a leap of faith to assume that CVI rather than SVI data provides a more accurate predictor of genuine voting behaviour.

    So, in my view the real action will be almost anywhere *other* than the behaviour of voters moving in and out of the Ukip orbit.

  29. CATMANJEFF

    Pre iny ref I would 100% agree with you but since then the SNP have been polling almost half the Scottish electorate in every opinion poll. They have nearly 100k members and in terms of members they are the 3rd largest party in the UK.

    I know the SNP don’t stand across the whole of the UK but with the closeness and the current dynamics of UK politics I think excluding the SNP from a national debate would be a bad day for democracy.

    UKIP only have representation in Westminster via two defections, they are projected to win just 3 seats. The SNP are projected to win more seats than the Libs and UKIP combined.

    Realistically (on current polling) the SNP are going to have a much much bigger impact on UK politics than UKIP but they are being excluded from a UK national debate.

  30. And just to add.. In the event of a hung parliament, who out of UKIP and the SNP will have the most leverage?

    Ofcom should also take that into account in their so called consultation!!

  31. Allan Christie

    “a bad day for democracy.”

    Now don’t confuse the issue by bringing in irrelevant concepts like “democracy”!

    This is all about party advantage – not about what would help voters to come to a decision.

  32. Excluding the SNP from a national debate only makes sense if they’re guaranteed a place in a Scotland debate with the other parties (well, at /least/ Lab and Con) participating. I’d actually be inclined, if Con and Lab were trying to exclude them from a debate, condition their exclusion on those two agreeing to participate in the Scotland debate.

    Then again, I do have to (at least rhetorically) wonder how the SNP would do if they stood some candidates in seats throughout England (either in their own name or under an allied pro-devolution league of some sort). T.P. O’Connor comes to mind as the model here (the Irish Nationalist elected for a long time from Liverpool). There have been intermittent appearances by them in polls in England (presumably from Scots living south of the Border), but I would wonder at their standings if they did put candidates up…at the very least, they’d wreak havoc on Labour.

    [And of course, the idea of a renegade pro-Tory group standing some “Independent Scottish Nationalists” in a few places comes to mind…]

    Still, the biggest problem is the idea of having a party leader given, say, 20% of the time in a debate where about 90% of the viewers cannot vote for that party’s candidates.

  33. It is very clear – at least as far as the smaller parties are concerned – that different pollsters produce results so far apart that (viewed as measuring true VI) there are systematic errors between them. It is thus invalid to average them – a bit like averaging the numbers in a length measurement made by six people, five of whom measure in cm and one in inches, and two having stretched tape measures instead of steel rulers.

  34. How is it invalid to average them? What else can you do, if you don’t know which are wrong and right?

  35. OLDNAT

    @”This is all about party advantage – not about what would help voters to come to a decision.”

    Is there any other purpose to be found in party political campaigning ?

  36. OLDNAT
    Allan Christie
    “a bad day for democracy.”
    …..
    Now don’t confuse the issue by bringing in irrelevant concepts like “democracy”!
    This is all about party advantage – not about what would help voters to come to a decision
    _____

    Absolutely.

  37. GRAY

    I wrote on the previous thread that the only two people who are realistically going to become PM are DC & EM.

    The whole concept of the leaders debate is (Who is going to become our next PM?) So as I wrote previously the debate should be just between DC and EM pitching against one another and across the English regions, Scotland and in Wales we can have local debates.

    What will UKIP bring to a national tv debate to viewers in Scotland and NI?

  38. Reporting seat-number projections isn’t particularly useful, because of the “I hit a bullseye, on average” problem with trying to identify that specific a result. Actual meaningful predictions would be the percentages of any particular government being able to form.

    When you do so, you do end up finding that while the voting intent is close, this is not a tight race in terms of what the probabilities of the next government forming will be.

  39. To clarify….There is as much chance of Stewart Hosie MP (SNP deputy leader) becoming PM as Nick Clegg and Nigel Farage so why exclude him and not the other two?

    Bringing UKIP onto a national TV debate will only showcase the disastrous immigration policies of Labour and the Tories inept policies to deal with it.

    If that’s what people want to watch then good luck to them.

  40. Incidentally, I intend to update my own model’s page on Monday. Doing it this week would not be an accurate prediction, because of the holiday gap in data.

  41. There is as much chance of Stewart Hosie MP (SNP deputy leader) becoming PM as Nick Clegg and Nigel Farage so why exclude him and not the other two?

    Nobody has mentioned that Stewart Hosie is actually an MP. Nigel Farage isn’t; and it’s not ‘nailed on’ that he will be. Whilst some might argue that incumbency is no guarantee of success in 2015, shouldn’t it be a factor in determining who is included in the leaders’ debates?

  42. “Unless something electorally drastic happens, it looks like being the closest General Election in (my) living memory”

    we all love hyperbole

    but the first election of ’74 was closer in terms of number of seats than is likely this year. Similarly in ’64 only 13 seats separated the total of tory and labour seats….

  43. Amber

    “Whilst some might argue that incumbency is no guarantee of success in 2015, shouldn’t it be a factor in determining who is included in the leaders’ debates?”

    One could make the argument that party representatives on TV debates should NOT be candidates in the particular election, as it might influence voting in their constituency.

    Arguments can be made for or against any position on these debates. Almost all of them will be thinly disguised (or just blatant) partisan positioning.

  44. @Peter Crawford

    I said “my” living memory. As I was born in 1975, it’s not hyperbole, it’s a fact.

  45. @ Jayblanc

    “… this is not a tight race in terms of what the probabilities of the next government forming will be.”

    Too much ellipsis. Not clear how you reach this conclusion.

    There are six plausible ‘types’ of government post-May:

    Conservative majority govt, Labour majority, Tory-led coalition, Labour-led coalition, Tory minority with Confidence and Supply and, finally, Labour with confidence and supply.

    If you think it is ‘not a tight race’ you must see one of the parties being in a comfortable lead. Which one? What type of government do you anticipate emerging? And what is the reasoning that lies behind your verdicts?

  46. yes it will be the closes election in 41 years, regardless of whether anyone can remember the 70s and 60s or not. you did put (my) in brackets, so that covers you. but a politically interested 70 year old would remember the elections in the 60s and 70s quite well, both the 1950 and 1951 elections were also quite close too.

  47. @Unicorn

    My ‘reasoning’ is a monte-carlo model that simulates 5000 national elections seat-by-seat, based on a distribution variated vote-share, taken from a diminishing-revert-to-mean, momentum and average polling. It lets me see how many times out of 5000, in elections near to the predicted vote-share, would any particular government be able to form.

    And just to dispel doubt, yes, that vote-share prediction has been adjusting in the Conservative’s favour (albeit not as much as a niave swing-back), and the Scotland seats account for SNP’s vote share rise.

    And you can wait with everyone else to see the latest figures it produces on Monday – http://ukelectiontrend.blogspot.co.uk

    But I’ll say that it is, like practically every prediction run I’ve made since I started, currently close in vote share but not a tight race for government.

  48. @Peter Crawford

    It’ll be a close election in terms of vote share.

    But to be a close election in terms of seats won, the Conservatives need to be a few points ahead of Labour in terms of vote share.

  49. COLIN
    “Is there any other purpose to be found in party political campaigning ?
    Yes, actually. One purpose is to get their candidate well known. Another is to persuade voters on a cause which is central to the party platform and intended programme.

    I wonder if cynicism about the intentions of parties and their leaders and candidates isn’t a bit overdone on UKPR. The majority make more sacrifices to be in politics than any gain they receive.

    UNICORN
    Thanks for your gentle and useful clarification. In Scotland, I wonder at the sustainability and distribution of the independence campaign effect, which carried much of the fervour of football support; there are very wide constitency variations in which the YES factor could recede fairly rapidly, but whether this will happen between now and May, and whether it will be retained, for example, in boots on the ground in the campaign, Heaven knows.

  50. And yes, things would be hugely different had the Conservatives not backed out of Lords Reform, so triggering the Lib Dems withdrawing support for redistricting. And campaigning against AV was an own goal there too.

    If redistricting had happened, yes this would be a tight race. If AV had happened, yes this would be a tight race. (Also harder for me to model!) Neither did, so we go into the election with built in differentials, that mean a close vote share is actually a small ‘virtual’ lead for one party.

1 2 3 4