The weekly YouGov/Sunday Times poll is out here. Topline voting intentions are CON 33%, LAB 40%, LDEM 9%, UKIP 11%.

These would have been perfectly normal a fortnight ago, but contrast with the average Labour leads of two points or so that we’ve had for the last week. All the normal caveats apply – it could be a sign that the post-budget narrowing of the polls is coming to an end and things are headed back to the pre-budget situation, or it could just be random sample error, and next week’s polls will be back to leads of one or two points. Wait and see.

YouGov also asked about European election voting intention, and found figures of CON 24%, LAB 32%, LDEM 11%, UKIP 23%, GRN 5%. Labour remain in the lead (though more convincingly than mid-week), the Conservatives and UKIP remain in a tight race for second place (though this time it’s the Conservatives who are narrowly ahead). Voting intention in a referendum on leaving the EU remains at 42% stay, 36% leave – the same as before the Nick v Nigel debate.

Most of the rest of the poll dealt with comparisons between how Ed Miliband and David Cameron are seen as leaders. The pattern is a familiar one, and one I’ve discussed here many times before – Cameron is seen as stronger, more decisive, clearer about what he stands for and more up to the job of PM; Miliband is seen as more in touch with ordinary people. We can’t easily quantify how much this helps the Tories or damages Labour. Miliband had rubbish ratings last year too and that didn’t stop Labour enjoying 10+ leads in the polls so it is cleary not a complete road block to success… but then, neither is anything else. There is no one, single explanation to voting intention, no one, single thing that leads to failure or success. Parties have won elections with unpopular leaders, they have won elections when behind on the economy – these things do matter, but they are all part of a package and can be outweighted by other things.

484 Responses to “YouGov/Sunday Times – CON 33, LAB 40, LD 9, UKIP 11”

1 2 3 4 10
  1. Ian Bailey
    “All parties want to win and the sad reallity is I trust all of them to sincerely lie to me if they think I might change my VI and vote for them”

    Best post of the day. LOL

  2. Alec (fpt)

    @Roger Mexico – “Spearmint doesn’t identify a cause, but I suspect that this is due to the increase in the income tax threshold in the Budget. ”

    I doubt this very much. There was a poll which I now can’t find, showing what issues people remembered from the budget. The largest score (27%) was ‘nothing’, with the annuities move second at 21%. From memory, the tax threshold thing was way down at 9%. In other words, 91% of voters were not aware they would be less than £2 per week better off after tax in fifteen months time.

    But the Populus poll, which Bramley helpfully found, asked what was “most noticed”, so the people who gave an answer would reply with what they saw most in the news, not that which affected them personally or would influence their voting. Their latest ‘most noticed stories’ poll has Missing Malaysian Plane at 76.3% and Ukraine at 6.3%. Nothing else makes 1% (Budget is third with 0.9%) but neither story will alter UK VI much. And respondents could only give one answer, rather than the two or three we often see. These sort of questions tell us more about media coverage than people’s real motivations.

    The fact that, despite the way the coverage went, 11.5% of Populus’s sample gave “personal Tax Allowance” as what they noticed most (second to Pensions at 21.5%), suggests it does matter to a lot of people. And how Lib Dem retention went up when the policy was trailed before Budget Day suggests that it may be related.

  3. I’d make a guess that Nick Clegg promising to entertain us by debating with Nigel Farage helped LD VI. They didn’t actually watch the debate but they liked that it was happening. :-)

  4. Helped Lib Dem VI up to 9? Am I missing something?

  5. One point that occurred to me the 3% error for a 95% confidence level only applies if the data is normally distributed. Does the assumption of normal distribution hold or is it close enough, at least for the main 2+2/2 parties. The fact that if one party under reports, someone somewhere else (including the will not vote) must gain. The figures for each party are not independent of each other and intuitively get the feeling the sampling errors should be something like sqrt(p*(1-p)/n) instead of sqrt(1/n) for a normal distribution. So the error in Lib Dem support will be roughly half the error in tory/labour support

    There also will be a covariance of -pq/n between party p and party q?
    (p and q being the true proportion of the population from which the sample was taken). If one party reports lower than the true proportion, the other parties will report high relative to the true position.

    I’ll have to think how the act of weighting alters the statistics as in effect what we have is a distribution of distributions reweighted to the first and trying to think about that makes my head hurt. Additionally I’d expect these distributions to be non-symmetric

    I understand that polling is for public consumption and so a +/- 3% is just quoted for that purpose. For creating a more interesting model, starting with the most appropriate statistics seems like a start.

    I wonder if any statistically minded folk out there could point me towards the appropriate statistics to use (I’m a scientist, not a statistician) especially considering the reweighting, as trying to work out how that might affect the distribution is opaque to me.

    I have an idea for a model I’d like to test, but I’d rather get the statistics “right” than make the calculations simple as part of it.

  6. @Shevii: “A lot of argument that the 50% tax rate brings in no extra income…”

    Except it’s not true. Increased taxes always bring in extra income in total although it’s possible that the marginal gain may shrink.

  7. Rogerh

    The Laffer curve disagrees with you.

    I did the test and it seems pretty accurate. My results were
    SD 87% match, G/EFA 73, ALDE 67, EPP 67, EUL/NGL 47, ECR 27, EFD 27.
    As far as national parties are concerned, it showed 93% match with Irish Labour and Czech and Finish Social Democrats. In France I had a 80% match with the Socialist Party and in Greece 80% with PASOK. The parties with which I had 0% match were UKIP, Dutch PVV and North Irish DUP.

  9. The Laffer curve becomes almost useless because it doesn’t specify what the ‘optimum’ tax level is. Some people put it around 75%, some around 40%. I’d be happy to split the different for a maximum tax rate of 57.5%, with an ‘upper band’ threshold raised to perhaps 250k.

  10. @Alan

    We’ve had the “Laffer Curve has never been demonstrated in practice, and even in theory it’s assumptions are based on a flat tax, not a progressive bracketed tax.” conversation here before. To be frank, the use of Laffer Curve to justify all tax cuts no matter what the circumstance has simply become a canard of the online-libertarian.

  11. MRnameless

    I agree it’s a simplistic toy model, I was using it a simple counterexample to Rogerh’s simplistic assumption the revenue is monotonic with tax rate.It’s more illustrative than a serious model.

    It is possible to have an optimal tax rate with regard to revenue is the point, where this point is with depend on your model, which will have to take into account the multiple rates/thresholds.

  12. Jayblanc

    I think it’s appropriate when discussing whether cutting taxes can raise revenue. Perhaps it’s misquoted by those who always think they are on the wrong side of the latter curve.

    Leaving it as “sometimes it’s appropriate to lower taxes and raise revenue and sometimes it is appropriate to raise taxation to raise revenue” isn’t an unreasonable position, where that point is is open to debate.

    I consider the “In all situations reduce taxation” brigade as ridiculous as the “In all situations increase taxation” bunch.

  13. Kate Moss is a model, and we may as well consult her as to the optimal tax rate.

    She would provide about as much insight as the Laffer curve.

  14. @Alec,

    We always like getting back to that Laffer curve don’t we!

    Also, how about what is a fair tax rate? Almost impossible to judge as it’s so subjective, but there is probably a tipping point for those working. Is 45p right?, or maybe 55p over £250k who knows…


  15. “We always like getting back to that Laffer curve don’t we!”

    I don’t. Can we end that conversation now please.

  16. Alan,

    The data from 2013 VI for the 4 main parties is technically non-normal (all have a p value of 0.0000 on the Anderson Darling test).

    However, the curve does look quite normal, and not too badly skewed.

    (Graphs fro Sofastats, Figures from Gnumeric)

    I would be happy to treat it as normal data, all things begin equal. There is enough scope for error in various parts of the polling methodology to make me think worrying about data that look reasonably normal but technically is not worth it.

    You are right about the margin of error for LDs, but if I recall the MOE at say p = 0.10 isn’t much less that at 0.5. Where 0.5 = 3.11 MOE, at 0.1 it’s about 2.4 I think.

  17. Correction

    I would be happy to treat it as normal data, all things begin equal. There is enough scope for error in various parts of the polling methodology to make me think worrying about data that look reasonably normal but technically isn’t, is not worth being concerned about.

  18. @Rich – not wishing to mention the L word, your point on fairness is spot on. This isn’t just about efficacy, in strict economic terms.

    However, in my view, you then make your own error by relating fairness only to a notion percentage rate.

    When we buy a car, we don’t just buy on the price – we look at what we get for that price, in terms of the product. Then we balance the product against the price, so a higher price car could still be better value.

    As with tax. I suspect people won’t mind paying higher taxes, up to a point, if they can be sure that what they get in return is worthwhile. This is why I believe the left has an overwhelming moral duty to ensure government is efficient.

    One of the problems we have is trying to overcome the notion that state provision is always inefficient, while private is always best. In many major areas, this simply isn’t the case. Pensions, for example, are far more efficient through the state system, and present really dreadful value via the private sector.

    Addressing the issue of fairness and efficiency needs to sit alongside economic efficacy, and this is one of the reasons why I find the L word so meaningless.

  19. The other day when discussing energy prices I gave some figures showing that UK has relatively low prices compared with other Euriopean countries. Somebody, I can’t remember who said my comparison was not fair as my figures included energy taxes. This is a fair point as some countries have very much higher energy taxes than we do.
    I have therefore taken the trouble to get to the figures without energy taxes. Somewhat to my surprise it makes very little difference.
    On electricy costs we are 12th out of 22 with taxes included and 11th out of 22 with taxes excluded. On either measure we are below the European average.
    On gas costs we are 15th out of 22 with taxes and 17th out of 22 with taxes excluded. On either measure we are below the European average.

    On this comparison UK energy prices look OK.

  20. TOH
    ” Lovely to feel the sun at last after those months of rain.”

    Here in the bonny Pentlands we had two hours of sunshine last Sunday morning, and I persuaded my son-in-law to bring the two grandchildren for lunch in the garden. It snowed.

  21. @TOH

    “On this comparison UK energy prices look OK.”

    I haven’t looked at the comparisons as closely as you obviously have, so I’ll have to take your word for it, but is it much consolation anyway to those in this country struggling to afford to heat and light their houses and facing annual price hikes by the energy companies? These companies, owners and dispensers of a natural resource, are exploiting a cartel position in a dysfunctional market to profiteer, milking their trapped customers to reward themselves and their shareholders.

    While that’s going on, I’m not sure that I’m overly fussed about what someone in Latvia may be paying for their gas and electricity.

  22. “sometimes it’s appropriate to lower taxes and raise revenue”

    Alice in Wonderland economics.

  23. Some of us wrote that we needed a week or two to see if the hype over the budget would subside. It still needs a few days more IMO and it is noticeable that Labour has not, on YG, gone below 36 and neither has Con gone above 35 IIRC. There was a movement, clearly, and the question is whether it’s a blip or a shift.

    Someone wrote that this movement to Con proved that the press have an effect and wonder whether that could be assumed in the last days of the forthcoming GE, given that supporting Con is the bias of the printed press. I don’t see that as proven, as, the way I saw it, most of the hype over the budget ‘giveaway’ was coming from the supposedly unbiased public broadcasters. We also have to remember how few are reading ‘the press’ nowadays. It’s probably that pure publicity effect again wot dun it.

  24. Catmanjeff

    Looking at sd=sqrt(p*(1-p)/n) gives a MOE (+/-2 sd)of 3% for p=0.35 and 1.9% for p=0.1 (n=1000), seems to concur with your thoughts about where it should be. I made a slight mistake when I said LD error was half of lab/con, in fact it was the variance which was half.

    FWIW the distribution I was looking for was “categorical distribution” which does give the variance/covariance I was thinking of.

  25. @ MrNameless

    I have just had loads of fun with that euvote link – thanks!
    Although its a bit funny here and there, such an excercise would teach folk loads about where they stand on EU issues vis-a-vis their normal voting habit.

    For example – I found, to my astonishment, that I am closest to the Slovakian Social Democrats and Italian Socialists, wheras I had assumed I would be closest to the German Die Linke – I was in fact miles away from them!!
    In UK terms I had a best match with Plaid Cymru (73%) followed by Labour (67%) – since I live in Maidenhead I am at a loss as to how I could vote for PC – so I will vote Labour as the nearest option.

    Interestingly (for me!), my old pre-2010 allegiance of 36 years (Lib/LD) only had a match of 60% – this did surprise me, as I thought I was more with them on EU issues than it appears.

    It all makes switching to Labour quite comfortable after all.

  26. @Howard – ‘Blip or Shift?’

    I can feel a new game show coming on here….

  27. Crossbat11

    I agree it’s not much consolation for those who are finding difficult to find the money for their energy but I do take exception to this…………..

    “These companies, owners and dispensers of a natural resource, are exploiting a cartel position in a dysfunctional market to profiteer, milking their trapped customers to reward themselves and their shareholders.”

    That has yet to be proved and I suspect it won’t be. I look forward to the report of the CMA in due course.

  28. While one can’t take a regional/national sub-sample seriously, the polls, local election results and by-election results all seem to suggest that the Tories will increase on their 2010 vote- in Scotland, where it makes no difference except in a few Tory vs. Lib Dem seats.

  29. Amber Star & John Pilgrim

    Apologies for mentioning the fine weather down south. Sorry it’s cold up there. Last time I went to Scotland on holiday in mid May there was too much snow on the mountains and i could not get to see the special plants I wanted to see. Mind, otherwise the weather was good if a bit on the cold side.

  30. @TOH – interesting work on power prices elsewhere, but a number of issues to consider;

    i) How do these costs relate to relative income levels
    ii) How much are the vendors investing in long term infrastructure development
    iii) What level of taxpayer subsidy, if any, are the energy companies receiving for such investment

    For example, UK taxpayers are picking up a £70B open ended tab for the Sellafield clean up, alongside bills for decommissioning of all other nuclear plants. Ideally, this should be added to energy bills, so unless we know how all costs associated with energy supply across all the countries are allocated, we can’t really make valid comparisons.

  31. BP,

    Do we know what the average improvement in Tory VI in Scotland is? It could be altering national polling numbers a bit.

  32. @Tony Dean
    It’s a bit weird. Apparently I’m closest by far to Parti Radical which seem to be a centre right party in France (I hadn’t come across them) with the Greater Romania Party (far right) close behind, level pegging with a bunch of SOcial Democrats and the Spanish UPyD who say “We begin with a revolutionary assumption: that citizens are not born being either left wing or right wing nor with the card of any party in their nappies” [I’m a bit naive but I didn’t think people were born in nappies at all]
    In UK terms I’m nearest to Green followed by SInn Fein with Labour trailing in third.
    A bit odd!

  33. Labour back to a 7 point lead, funniest joke I have heard today. Yougov polls, a representation of 2000 out of 30 million voters. This equates to a sample representation of 0.007% so this should be recycled.

  34. Unlike all the other polls, eh, C Beaven?

    Are you sure you’ve come to the right site?

  35. C Heaven

    You are right, taking a sample of only 1000 means to a 95% confidence level the sample error is a whopping 3% at most. Completely useless for a rough idea of where things lie.

    It’s quite possible this is one of the 5% that lie outside even this massive margin

    Do you have access to properly weighted poll of 1,000,000 people so we can reduce the sampling MOE to 0.1% (we’ll gloss over the fact that methodological errors will outstrip this) as no one seems willing to fund a daily poll to that accuracy?

  36. @C Beaven

    You do realise you’re on a polling forum?

  37. I have noted a few comments on the weather.
    as we all know there is a north south divide which may well be a passing reason why the south east of England is so popular.
    where I work in brighton and hove there also added attractions hence the first gay marriage taking place down here.
    what I have found over the years is that whenever I brave the northern climate I am pleasantly surprised at the friendliness of the people.
    I worked in Birmingham in the late 1970’s and could not have been made to feel more welcome.
    angela and I went to Edinburgh several years ago and when angela went sight seeing if she stopped to look at her map inevitably someone would ask if they could help her.
    whilst politically many people on this site have very different views on how they would like the country run the fact is that I believe the UK to be one of the most tolerant and well integrated countries around.
    when my daughter was in her teens she and her fellow students were always going on about how much more exciting the rest of the world seemed to be.
    I said that if you took a number of important aspects of life and then applied them to each country in the world and score out of ten for each aspect for each country I would be surprised if we didn’t come near to the top.

  38. Mr Nameless,

    Statgeek has breakdowns for polls on his website.

    You can get a ward-by-ward profile of every seat in the country on Electoral Calculus, and they model the probabilities of various parties’ performances using local election results. I don’t think there’s a seat in Scotland where they don’t expect the Tories to do better than in 2010. They give the Tories good-to-certain chances of winning in 6 different Scottish seats in 2015.

    Wikipedia of course has by-election details.

    C Beaven,

    What matters is not the sample size as a proportion of the population, but the sample size in absolute terms + proper weighting to take account of sampling biases. A 1,000-fold sample of a population of 30 million is better than a 15-fold sample of a population of 100 (because of Bernoulli’s law of large numbers) and a properly weighted sample of 1,000 is better than a badly weighted sample of 1,000,000.

  39. @GuyMonde
    That result IS weird, because I got the same UK results as you but turned out to be best match to the Swedish Social Democratic Workers Party – bit of a stereotypical centre-leftist, and sooooo boring, but at least I wasn’t acccused of being French.

  40. I notice that here and in other sites where averages are used to predict who has most votes that , despite the recent fall in some polls for .Labour the party have retained for some years varying leads. AND after all the stat playing with what we say. The average for the Conservatives seems stuck around 32/33 where ever one looks. Electoral calculus(prediction ) and the bookies are STILL putting Labour odds ON. Tories 3 to one AGAINST. If. like me you had grown up witha father as a professional gambler. you would know that 3 to 1 is for the guessers by horse name , and ” lets have a bet as we are here” They were not in it for the money Unlike , I believe most betting now very much are and care little WHO wins. due to their politics-if any. Of course things will change but surely we have passed the old view of a mid term fall and a recovery as the election becomes close.
    This election is almost certainly going to be uniquely non predictable as we have now Four parties . OK 2 to 1 favoured Farage but his parties effect is surely going to make the result unusually unpredictable. We hear Tories will “go home” . Maybe Labour will too. Could UKIP simply be a place to say (for many ) you are all a load of c… The polls THIS day say different for them(YOU GOV Sunday ) I know what I want but haven’t put any money on yet. Those who have bet vast sums will almost certainly have consequences. -“Money where mouth is”. Read their record in election betting. It appears to me to be more convincing than the best poll results. and for a very long time . They do not adjust the turn out . or how one voted before, etc. JUST where the money is going NOW.

  41. Apparently, I’m either a Green, a Liberal or a Christian Democrat. None of them match up very well, though the Greens do best overall.

  42. Mr N
    I must have done the wrong quiz. I got a score, 220 out of 500, which qualifies me as knowledgeable about EU, I gather. What I got wrong was the ‘who said what’ – that seems to be for anoraks and I would have been ashamed to have got that right (‘get a life’ time). Actually I got a few right just by looking at their faces. Now, there’s a polling topic!

    Was there another one that told you what your politics were?

  43. @ Howard,

    I said if your theory was the budget bump was due to the positive press, then Labour are in trouble once the real campaign starts. I have no strong view on it myself, but other posters have complained that Osborne’s pension changes were insufficiently scrutinised.

    (I do think the broadcasters tend to take their direction from the print media, so a party that gets an easy ride from the papers gets an easy ride on television as well. Whether that happened in this particular instance- who knows? I suspect journalists- many of whom are of pensionable age- just welcome the opportunity to pour their pensions into buy-to-let.)

  44. Spearmint
    Good point on aging journalists. Much financial commentary is probably unconsciously biased. I find our financial journalists, indeed the industry they report on, very insular and also parochial in outlook.

  45. Bookmakers make a book that reflects where the polls are and use standard probability curves to decide the odds. This initial offer is then modified to balance out their liabilities once people start placing bets. They don’t care who wins, they just want to guarantee a profit. At this stage, when betting is fairly light, the odds will usually just reflect the polls. You have to remember that they are not really interested in the likelihood of any particular event taking place just how much that would cost them.

  46. I’ve done the right tst now (EU voting) and old UKPR lags who knew what my colour background was will be astonished at to whose views I adhere.

  47. @ Howard
    I bet you’re not a Green liberal socialist fascist like me.

  48. “Can we end that conversation now please.”

    Thank you! Laffer curve stuff is boring at best and encourage partisan chat at worst.



    “It still needs a few days more IMO and it is noticeable that Labour has not, on YG, gone below 36 and neither has Con gone above 35 IIRC.”

    Con got 36% on the 21st and the 24th.

  49. @Alan “The fact that if one party under reports, someone somewhere else (including the will not vote) must gain.” Don’t forget that the reported poll totals (including ‘others’ – Greens etc) add up to 100%, while in fact some 35-40% of the electorate have not voted this century, so there is plenty of room for the poll figures for any one party to vary without affecting the others. Also the polls are weighted, which might have some odd results in this context.
    ( I too am a scientist, not a statistician. I believe there is potential for some odd systematic errors in polls, as well as rather large random errors. I recall some months ago showing that over a series of 10 YouGov polls the Conservative VI varied in a way consistent with the results being repeated measurements of a constant VI with a sampling error of around +/_ 3 VI points, while over the same 10 polls the Labour VI of about the same size hardly varied at all.)

  50. @Guymonde

    Presenting for your consideration, a real political party:

1 2 3 4 10