Back now, and ready to go through the rest of the Sunday polls apart from the voting intentions and pick out some of the more interesting findings:

  • YouGov’s poll for the Sun on Sunday asked about tactical voting (it didn’t actually use the term, as I think many people use it to refer to different things. It asked if people were voting for their first choice, or a different party that would beat a party they disliked from winning). 77% of people said they were voting for their first choice, 11% tactically. A word of warning about interpreting this – 8% of Tory voters said they were voting tactically, 11% of Labour voters, 12% of Lib Dem voters, 11% of UKIP voters… but we don’t know if that means 8% of Tory voters are voting Tory for tactical reasons, or that 8% of would-be Tory voters are actually going to vote for someone else for tactical reasons (or a mixture).
  • The YouGov Sunday Times poll had a number of questions on British Muslims and on terrorism. People were split over how well integrated British Muslims are into British society and the extent to which they share British values. 46% of people think the majority or almost all British Muslims share British values, 46% of people think that only a minority or hardly any British Muslims share British values. 42% of people think that most or the majority of British Muslims are well integrated, 50% think a minority or hardly any are. UKIP voter’s attitudes towards British Muslims are far more negative than supporters of other parties – 73% of UKIP supporters say most Muslims don’t share British values, 79% say most British Muslims aren’t well integrated. Nigel Farage’s comments about areas of Britain being like ghettos with sharia law were rejected by most respondents – 33% though they were broadly true, 41% thought they were false. 75% of UKIP’s own supporters believed them.
  • There was a particularly interesting immigration question in the Sun on Sunday poll, essentially asking people to choose between a multicultural approach and an integrationist approach. Slightly to my surprise a multicultural approach was the more popular – 36% thought it better that immigrants leave their own cultures and traditions behind and integrate fully into British culture, 48% thought it better than immigrants retain and celebrate some of their own cultures and traditions.
  • Going back to the terrorism and surveillance questions in the Sunday Times poll, just over half of respondents (52%) thought that the security services do need more access to the public’s communications in order to fight terrorism, 31% thought they already have all the access they need. A similar proportion (53%) would support requiring internet and phone companies to retain users data for 12 months and provide it to the security services on request, though by 51% to 35% people think accessing someone’s personal communications should require the consent of the Home Secretary. While people think accessing personal communications data should require the consent of the Home Secretary, when asked whether they trust the authorities to use the information they obtain responsibly they actually trust Ministers & the Home Office less than the police and the intelligence services. 50% trust the police to use the information responsibly, 63% the intelligence services, 45% the Home Office, Ministers and civil servants.
  • In the ComRes poll they asked (via my old favourite the agree/disagree statement, grr!) whether people agreed with the statement that “Ed Miliband is using the issue of the NHS for his political advantage, not because he cares about it”. 49% of people agree, 26% disagree. That looks bad, but I have my doubts about questions about politicians’ motives. I suspect they largely just reflect a general cynicism towards the motives of all politicians, rather than opinions about particular politicians or decisions. YouGov asked a very similar question in their Sun on Sunday poll, but asked it about Ed Miliband AND David Cameron, and got answers that were almost the same. 46% thought Ed Miliband was using the NHS for political gain, 19% doing what he thought was best for it, 20% both equally. 48% thought David Cameron was using the NHS for political gain, 15% doing what he thought best for it, 19% both equally. As you’d expect, in both cases supports of the Conservative and Labour party both thought their own leader was doing what was best, but the opposing leader was just using it for political reasons.
  • The debate over the debates rolls on, and so do poll questions about it. Opinium asked about whether particular leaders should be invited – 61% think the leader of UKIP should, 46% the leader of the Greens, 30% the leader of the SNP, 23% the leader of Plaid Cymru. The current proposals for debates between Con, Lab, Lib Dem and UKIP leaders was supported by 37% of people, opposed by 31%. YouGov in the Sun on Sunday asked people to pick from some possible combinations. The most popular individual option was the widest, the Lib, Lab, Con, UKIP, Green and the SNP and Plaid. This was picked by 35% though, so while it was the most popular single option, 49% favoured a narrower option – 20% favoured the proposed Con, Lab, LD & UKIP, 17% Con, Lab, LD, UKIP and the Greens (but not the two nationalists). 12% supported an even narrower option, excluding UKIP. In their Sunday Times poll YouGov found people still think David Cameron should take part even without Natalie Bennett – if she is excluded 31% think Cameron should refuse to take part, 52% think he should take part anyway. However, asking about the other side of the deadlock, if Cameron refuses to take part without Bennett 52% think the broadcasters should call his bluff and invite her, 28% think the debates should go ahead without him, 8% think the debates should be cancelled.

193 Responses to “Gleanings from the Sunday polls”

1 2 3 4

    Great link to that SKY report.

    I find it absolutely frightening -and a total turnoff.

    They are supposed to be selling ideas to improve our lives-not washing powder or mobile phone packages.

  2. Barney Crockett


  3. MRN

    @”the ideology behind a party are surely other vital factors in motivating people to vote a certain way.”

    Perhaps for some who have a family background deeply attached to a particular party.

    But I would have though “ideology” was the last thing which appeals in todays open , global society-particularly to the very young.

    Ideas -yes-Ideology-no.

  4. In fairness to VfP they now have a note on their site pointing out that it’s based on old material (although I think this has only gone up in the last couple of weeks, it isn’t very big, and it hasn’t stopped them ramping up their twitter promotion in the last month or so).

    I think the major issue with it is that it doesn’t allow for preference voting – when I took it I was 80% Green and 20% Tory (because I’m generally left wing but support nuclear power and a form of national service) but I’m fairly certain that all of my second preference votes would have been Labour policies. So while I might prefer individual Green or Tory policies to Labour ones, in reality I have major objections to their platforms as a whole and find Labour’s offering much more acceptable.

  5. JIM JAM

    What is an “implied manifesto” ?

  6. Colin,

    I’m not sure what you think I mean by ideology, but I mean the guiding principles which act as the basis for a party’s policies.

    For instance, Labour’s ideology could be summed up as “More equality is a good thing, and the needs of the worst off in society should in general be placed before those of the wealthiest”.

    The Conservatives’ is along the lines of “Private enterprise builds strong economies, and people should wherever possible take responsibility for looking after themselves and their families”.

    I think you are truly mistaken if you think parties ought not to be guided by basic principles any more. Rigid ideology of the kind I think you mean – that written down in dusty books with names like The Road to Serfdom or Das Kapital – is unknown and irrelevant to most people, I agree.

    But people still have principles, and those principles (and the perception of the principles each party holds) are a major factor in deciding votes.

  7. MR N

    I’m happy with your revised emphasis on “guiding principles” -and pleased you didn’t use the word “values” :-)

  8. Colin,

    I first heard the term used by Roy Hattersley but I suspect it’s older. It refers to the idea that however well-meaning a manifesto, a manifesto cannot predict events. The implied manifesto means the public perception of what each party will tend to do in any case presented.

    So in the case of an economic collapse, you could expect the Labour Party to cut VAT and invest in Keynesian infrastructure building. You could expect the Conservative Party to cut personal taxes and public expenditure. That’s the implied manifesto.

  9. MR N

    Ah-I see.

    A sort of guessing game about what they will do-as opposed to what they say they will do :-)

  10. Colin

    I think it’s less ‘what they will do as opposed to what they say they will do’ than ‘what they will do where they haven’t said what they’ll do’

  11. (clear as mud)

  12. FUNTY

    Thanks for clearing that up. :-)

  13. @stat

    I was not trying to be offensive or rude and I don’t think Barney was either; he carefully chose to use the collective and you went off on him! Would you have preferred him to have made it personal? You know full well that many here think that this site gets hijacked from time to time by the “Scottish Question” but you don’t care about the views of these people and you post away regardless, and that’s fine, but you really ought to be a bit more thick-skinned when others gently, and dare I say, quite affectionately, call you out on it from time to time.

  14. Populus

    Lab 36 Con 35 – LD 8 – UKIP 15 Grn 4

    Scotland Crossbreak SNP 32 Lab 28 Con 25 LD 6 Ukip 5 Grn 3

  15. Correction – UKIP 13%!

  16. David Colby

    When we are discussing voting for the Westminster Parliament in May, the “Scottish Question” is “How many seats will each party win in Scotland?”

  17. Colin – re data mining

    Surely part of improving people lives, is finding out what factors will actually improve them, from the perspective of said person.

    I suspect it’ll be increasingly used in many aspects. Politics being no exception. I’d trust proper data analysis of what people are saying over politicians anecdotes from isolated chats with a voter.

    Perhaps the future of political strategists are going to be replaced with data scientists.

  18. Graham

    As always with Populus online, you need to look at their idiosyncratic weighting of current party ID to 2010 estimates.

    In this case (as usual) little/no change for Con/Lab : LD upweighted 163 -> 204.

    UKIP downweighted 215 -> 81
    Green downweighted 89 -> 41
    SNP downweighted 75 -> 38

    That is taking reversion to the status quo ante-bellum to rather ridiculous extremes!

  19. Alan

    “Perhaps the future of political strategists are going to be replaced with data scientists.”

    1. Wrong tense! It’s already happened.
    2. “Scientists”? That’s giving the number crunchers (as opposed to Number Cruncher, of course) a very inflated status.


    Have to disagree with you on the Vote for Policies site. You have a point re identifying parties, although I suspect that the large majority of the electorate would fail to do so.

    It might make life for Lab harder, though. I worked right through it as honestly as I could and gave the postcode of my elder son’s constituency [Bristol E] whereupon I was presented the statement that I am 55.56% Green, 22.22% LD, 11.11% Lab & 11.11% Con, with “the results from 1283 completed surveys in your constituency“:

    Green 33.80%
    Lab 21.75%
    LD 18.24%
    Con 11.62%
    UKIP 8.30%
    BNP 6.29%

    Hope the Greens are canvassing hard there!

  21. “Perhaps the future of political strategists are going to be replaced with data scientists.”

    Perhaps. And if this is how policy is determined in future we may find our politics developing in the way Henry Ford surmised: politicians will be busy developing a faster horse.

  22. Colin –
    Mr N has summed the implied manifesto idea quite well but it is not just about the hyperthetical it is about broad themes.

    One of my concerns for exampleis that Labour may allow themselves to be seen as not being sufficiently pro-business or even anti.

    There is a bit in Bridget Jones Dialry where the eponymous heroine is at a dinner party full of Tories (remeber this is early 1990’s think).

    There are taliking about Economics etc and she says to paraphrase ‘who cares being Labour is about Gays, Nelson Mandela and stuff’.

    DC sanatised this of course for the Tories.

  23. @JimJam

    Don’t underestimate what you can do with (good) data science. Identifying opportunities and innovation is part of that…

    However neither Sky, nor the political parties, are at the forefront of that. There were some interesting bits of work done as part of the Obama 2012 campaign though these were largely tactical rather than strategic.

  24. Statgeek

    Re party VI concentration in the Scottish regional crossbreaks [1]

    While wee unweighted samples can never tell us very much, they can sometimes suggest possible effects.

    SNP above average in Central (52%): below average in H&I and South (32/33%)
    Lab above average in Glasgow and West (27%): below average in H&I and South (17/18%)
    Con above average in South and Lothian (16-19%): below average in H&I and Central (7/8%)
    LD above average in H&I (18%) and Mid/Fife (11%) : below average in Central and NE (1/2%)
    Grn above average in H&I (13%), West (7%) and South (10%)
    Most undecideds in South and NE (19/20%) : fewest in West (7%)

    They don’t look too outlandish – especially for the minor parties – Con, LD and Green

    If Ashcroft shows anything like that pattern in Lanarkshire, and/or we get a good LD/Green contest in the Highlands, then May could be even more interesting.

    [1] SNP at 52% in Central (Lanarkshire & Falkirk for those unsure of Scottish political geography) could lead to a new UKPR mantra – “Don’t look at the Central Region crossbreaks!”

  25. Oldnat

    Data science (done properly) is about applying the same sort of rigour to data analysis that one might apply to “real” scientific data.

    Done badly and you can pretty much show any trend you like and is pretty dangerous as it leads you to have a great deal of confidence is what in reality is no better than gut feeling of which trend to show.

    I suspect we are in a state where it’s being used a bit and with more rigour than in the past and the trend will be and increased level of use (to the point where causation and prediction will be heavily used)

    As for whether they are being held in high esteem or not is kind of irrelevant, I’ve never felt as if I was held in high esteem as a scientist, if a new field is opening up where I can transfer my skills and pays a hell of a lot more, well it seems as if I’d be more valued by doing that.

  26. Alan

    I have the same problem with the term “data science” as I have with “social science”.

    Adding “science” to the name of an intellectual activity doesn’t qualify it as a “science” – in thee sense that the word has been used for the last couple of hundred years.

    Still, if it makes people feel good, while paying them buttons, it has no doubt served its purpose. :-)

  27. Following up on an earlier comment, the difficulty arising from people being ignorant about the quantum of some amounts is that they can be fooled into believing something is possible that simply isn’t, for instance I have heard each of the below things being listed as a way to substantially address either the deficit or the long-term NHS shortfall (depending on the speakers’ priorities):
    – abolition of Trident
    – leaving the EU and stopping paying EU subscriptions
    – putting the bankers’ bonus tax back in place
    – halving overseas aid
    – stopping benefit tourism

    The gulf between the income generated / cost saved from each of these ideas and the amount needed is at least ten fold, and as much as 1,000 fold – but if people don’t understand the size o the numbers they have no chance of evaluating the effectiveness of the proposal.

  28. Is Ashcroft due today?

  29. BFR

    That was the context of my point.

    Simon Wren-Lewis has been screaming about the lack of any attempt to properly educate the public on what Government spending goes on. If we don’t have an informed public, we’ll get badly uniformed opinions.

  30. And the polls show that the public across the world have seriously uniformed ideas about key issues.

  31. Bigfatron,

    It’s not about people being fooled at all, it’s politicians succeeding by telling them what they want to hear.

    All the examples you quote relate to the fact that we spend far more on the things people like and value than things they don’t, a natural comsequence of democracy; To get elected you need to offer what people want.

    The next consequence is that when there is pressure on something they like, usually a large budget, they target something they don’t’t like, usually a small budget that won’t fill the gap.

    It’s not ignorance, that they don’t know, it’s that they don’t want to know.
    When I was a Councillor we had about a decade of having to find close to £10m in savings every year and every time we consulted the public said “Well it’s easily solved by cutting back on Councillors allowances and expenses”

    No matter how often it was pointed out that if every member worked for nothing and didn’t claim a penny it would save less than £5m, not even 1 years savings target, they simple got angry and accused you of not caring or feathering your own nest.

    It’s confirmation bias, the public filter the facts to get the result they want.

    Why do so many people believe that we can solve the problems of the NHS by slashing Foreign Aid……..because they want to!


  32. Jack R

    At 4pm.

  33. OldNat,

    Ta. I couldn’t remember if it were 2 or 4.

  34. Oldnat

    As a real scientist, I can see genuine parallels between the techniques used in data science and the physical sciences.

    Yes, there aren’t any underlying “laws” to be discovered (or the underlying patterns are too complex to be modelled), but establishing a rough approximation has worked for physics for hundreds of years. We haven’t thrown away F=Ma even though it’s predictions are hopelessly inaccurate in certain situations. We establish where it falls apart, find a model that improves it’s predictions and carry on. It helps if this model fits our understanding, but quite often that requirement has been completely ignored and a new field of science has been established.

    I’m not saying that the same levels of prediction will ever be possible in data science, some systems are simply too complex for that to happen. It can however still answer some questions, even if the answer is a bit fuzzy.

    It seems that highly complex, and chaotic physical systems can be analysed from a data point of view, and it would at least give insight into which are the important mechanisms involved.

    Largely it’s about answering questions in the best possible manner, it’s about replacing hunches and assumption with data driven evidence. Yes, scientific measurements are precise enough to allow you to create a system of laws that allow us to fool ourselves we understand what is going on. The methodology of the number crunching is largely the same, with less robust data you have less robust predictions, you are still trying to answer a question, just one with a lot less certainty.

    If asked “What can you say about the polls?” There will be various things you can say with varying levels of certainty. Just because you can’t uncover a fundamental law doesn’t make the answers necessarily worthless. It should also improve the data collected to be the data that actually helps you answer the various questions that might be asked.

  35. Alan

    The smiley should have alerted you that I wasn’t entirely serious!

    The improvements in data analysis are excellent. When you get to the Hari Seldon level, I’ll promote you from proto-science to science. :-)

  36. ALAN

    Yes-I can see that. And I have no objection to accumulation of data which informs policy.

    Its just that I hate the idea of being “targeted” by my local candidate as part of this or that group-rather than an individual.

    Guess I’m old fashioned :-)

  37. JIM JAM

    Thanks- yes the impression that a party gives about itself.

    Call it what you will-but it is certainly important.

    Not at all sure that DC has “sanitised” impressions of Cons as “for the rich”.

    I wish he could-but don’t think he is the man who can do it. Given another term & a fair wind on public finances, I do believe that he can & will demonstrate it in Social Policy .

    But I’m a great advocate of “seeing is believing” , and if he fails in May-it will need a David Davis type background ( not the man though) to change that impression.

  38. Colin

    I agree although going around everyone isn’t feasible (or possible to get everyone’s views even if it were).

    With improvements, you will at least feel part of a smaller, more accurately represented group.

  39. I appreciate that it is a minority interest here, but there are a number of public models posting election projections and I think it makes sense to be aware of the strengths and weaknesses in each case.

    Recently I have been looking is some detail at Electoral Calculus (henceforth EC) and I am not that impressed with the projections it makes.

    Full model details can be found at:

    This site offers a wide range of resources including projections down to the ward level and facilities for users to run their own simulations. It claims to be “Uk’s top General Election predictor”. It is currently offering the following national seat projections:

    Labour – 321; Tories – 242; LibDems – 19; Ukip – 0; SNP & PC – 49 and others – 19.

    Labour supporters may be heartened by this and others may be surprised by prognosis for Ukip and the complete failure to mention the Greens. But how much faith can we place in their predictions? Not a lot – is my assessment, for reasons given below.

    EC makes its projections (for an election taking place today) by using national polling data together with a rather sophisticated algorithm for reallocating VIs between parties (the Strong Transition Model). This differentiates between ‘strong’ and ‘weak’ supporters with churn figures being calculated differently in the two cases. Some of the analyses refer to the use of betting spreads but this may relate to an earlier version of the model. As far as I can tell, the model makes no use at all of constituency level polls such as the Ashcroft series.

    The model has been going for a long time and the authors report that it did a good job in predicting the results of the 2005 and 2010 elections. However, my personal view is that the exclusion of Ashcroft constituency polls makes it ‘outdated technology’ (more on this below).

    Prior to May 7 2015, how can we evaluate the inner workings of the model? Recently I carried out an exercise assessing how accurately three other models predicted the outcomes of Ashcroft’s December batch of constituency polls. I have now done the same for EC and the result is that their predictions were further off-target than any of the other three. The mean Euclidean Distance measure (amount off target) was 13.34 for SVI and 13.1 for CVI, compared with figures of around 10 for each of the other models. In this batch the model consistently overestimated Labour VI and undershot with its Ukip predictions. On this evidence, it doesn’t seem that they have a particularly good method for reallocating voter support.

    These shortcomings are all the more stark given that the EC constituency projections were based on figures downloaded today (Jan 19) at an interval of about a month after Ashcroft published his polls. This is not surprising if the model take no account of Ashcroft constituency polling data. (Analyses of Electionforecast data showed a sharp drop in the off-target measure after the polls were published, the new information rapidly being incorporated into the model.)

    A dramatic illustration of the effects of this ‘ignore constituency polls’ stance comes from the model’s current predictions for Clacton (not included in the December batch). Despite mentioning the by election, the model projects a Conservative win with a 35.98% margin over Ukip (itself predicted to get just 10.46% of the vote). Ukip’s chance of winning the seat is estimated to be just 0.6%. This obviously flies in the face of the by election result itself as well as two constituency polls (Ashcroft and Survation).

    My own assessment is that by ignoring constituency polling data – an increasingly valuable resource – this model hasn’t kept up with the curve. For all the impressive graphs on the site and facilities for running your own simulations, there seems to be good evidence that their intrinsic figures are out of line with what is really going on at the moment.

    When the next batch of Ashcroft constituency polls turn up, I plan to repeat this benchmarking exercise and I’ll be interested to see whether I have to eat my words by then.

  40. ALAN

    Perhaps-but the idea that the clip-board holder is pre-armed with a list of “buttons to press at this address” will not facilitate a very lengthy conversation :-)

  41. That populus weighting is absurd.

    SNP downweighted 75 -> 38???
    UKIP 215 -> 81 ???

    Who do they think they are kidding?


    As an active political inhabitant of the Clacton Constituency, I can certainly confirm that projection is complete and utter nonsense. UKIP will win by a country mile.

    P.S. Don’t tell Tim Young I said so!

  43. Skippy

    Those kind of figures have been common since Populus introduced their new online methodology.

  44. Colin

    If they did their work properly, they would already know this about you and not knock on your door!

  45. Ascroft:

    Con: 29% (-5)
    Lab: 28% (=)
    LD: 9% (+1)
    Ukip: 15% (-1)
    Green: 11% (+3)
    Others: 9% (+3)

  46. Er, Ashcroft.

    Lib Dems look high, and either he’s captured an unprecedented Tory -> Green swing or his numbers are, as usual, basically random.

  47. The polls are all wrong – well Rob Hayward thinks they are – they are understating the Tories and overstating Labour.

  48. Ashcroft Scottish crossbreak –

    SNP at 58% – from a sample of 43.

    Don’t think ‘ll crack open the bubbly. :-)

  49. Yeah… the last two ashcroft polls have been…. .. err.. at best… “controversial”?

1 2 3 4