We’ve almost arrived at the Christmas break. Today we have new polls from Opinium (their last of the year) and YouGov (their penultimate of the year – there is one more to come on Monday night). I’m not sure when Populus put out their final poll of the year, and Survation have a Scottish poll being published next week, but that should be it for the year.

Topline figures for today’s two polls are:

Opinium/Observer – CON 29%(nc), LAB 36%(+2), LDEM 6%(nc), UKIP 16%(-3), GRN 5% (tabs)
YouGov/Sunday Times – CON 32%, LAB 34%, LDEM 6%, UKIP 15%, GRN 8% (tabs)

The Observer write up of the poll, incidentally, is particularly poor, or at least, contains one particularly poor sentence. Toby Helm writes “It is the second poll in a week showing that the Tories have lost ground since chancellor George Osborne’s autumn statement earlier this month”. Now, I have long whined about newspapers treating only their own poll as being meaningful and pretending others don’t exist, so well done for putting a poll in context… but it’s a rather extreme case of cherry-picking context to create a narrative that doesn’t exist.

The Opinium poll is the second one this week to show Labour’s lead growing, in fact it’s the third as there was also TNS. But there were also rather a lot of other polls that didn’t… there were another ten polls who the Observer has chosen not to mention. There was an Ipsos MORI poll this week (no change in lead), a ComRes phone poll this week (no change in lead), a ComRes online poll last weekend (shrinking Labour lead), two Populus polls (who have shown smaller Labour leads in their four post-Autumn Statement polls than their four before the statement) and five YouGov polls (whose post-Autumn statement polls have shown essentially the same Labour lead as those before). Lord Ashcroft hasn’t polled this week, he’s already finished for the year, but his post Autumn Statement poll had Labour’s lead down one point. As you can see, there as as many polls showing Labour’s lead falling post Autumn Statement as rising, and overall I expect what we’re seeing is a simple case of normal random sample variation. Taking a crude average of the Labour leads in November would give you an average lead of 1.6 points, take a crude average of the polls in December so far gives you an average Labour lead of 1.6 points.

There’s always a temptation to see narratives in polls, to ignore those showing no movement, latch onto those showing exciting looking changes and build an explanation and a story around them. It’s normally wrong to do so.


159 Responses to “Latest YouGov and Opinium polls”

1 2 3 4
  1. Barney
    Its unbelievable isn’t it. Texans not knowing about their states independence from Mexico. However, on one occasion in California, I mentioned to a young waitress “the gold rush”. She asked me if I meant the one in my own country, Australia !

  2. Regarding the question are Labour moving ahead.

    Based on You Gov’s VI data for December, the answer is likely to be yes.

    Please see the CUSUM chart below:

    https://drive.google.com/file/d/0BzTTW1ecy-NDSGxoWE1maFhzUjA/view?usp=sharing

    In December, Labour mean VI was 32.9% If Labour’s VI had just been bobbling around their mean by the MOE randomly, their data points would not show a steady direction, but have a few points up, a few down etc (just like the other parties on the graph).

    The fact that Labour have enjoyed five polls in a row of increase looks like a trend. If the mean is 32.9% and their VI is reported to an integer, then each poll must be above or below. The chance of each occurring randomly is 50%. As there are five points above the mean in a row, the chance of all this being down to just random MOE (ie Labour VI is the same) is 1/32 (3.125%).

    So given that, YouGov at least are very likely showing an improvement in Labour’s VI, starting about the 15th December. This increase is small, but likely.

    The other parties are not showing any change other than random MOE, so perhaps a few voters who were WNV/DK have crept back? I haven’t checked, but I guess @Spearmint’s churn might help here.

    It can all change the other very fast too!

  3. @CMJ

    I agree, if you are looking shorter term than the whole-month averages that AW quotes, it looks like a small move towards Labour, just about MoE (remember, even an aggregation of 10,000 still has a ±1 point MoE), but nevertheless an improvement for Labour

  4. @CMJ

    I think there is a flaw in the logic. If you notice a pattern in data and then test it for significance the probabilities are wrong. Imagine a run of random data, From time to time there will be some very odd-looking patterns, but to observe them and then test for non-randomness is obviously wrong. It is a known fallacy, the name of which I forget for the moment.

  5. @CMJ

    Hmmm…dubious statistical support for your Labout upturn. Granted a binomial distribution would give you a one in 32 chance of getting a run of five VIs in excess of the prior December average. But you make several arbitrary choices to get there. Why look at just the last FIVE polls. What p-value would you get if you went for the last 6 or 7 or 8 polls? As a baseline average, why use the the start of December as a beginning and five polls back as an end? Also, why base the judgement on Labour VI alone? Arguably, the Labour/Tory gap is just as relevant and this would tell a different story.

    I’m all for agreeing ways of detecting VI shifts, but the first thing would be to avoid potential accusations of cherry-picking and set ground rules for making future judgements. What should be the agreed period for the baseline average? How many recent polls should be compared against the baseline?

    Well spotted on the run-of-five, but not entirely convincing as it stands…

  6. Look, I’m a Labour supporter and I can tell you the polls are defiinitely showing a trend towards Labour.

    No, no, I’m a Consevative supporter and I can prove the polls say no such thing.

    This is what’s so good about the polls, they are impartial and tell the truth so we all know where we stand and have no option but to agree with one another…er….

    Oh, I forgot, I’m a LibDem supporter…..oh dear…..

  7. Month on month (Nov-Dec) AWs arithmatic must be right that the 1.6% lead for Labour was the same.
    I do recall that the UKPR Average lead for labour dropped to 1% 2-3 weeks ago and held for a some days (maybe a fortnight).
    The 1.6% lead being the same for November and December would therefore appear to disguise a narrowing in the Lab v Con lead followed by a more recent widening.
    What we don’t know is if the narrowing then widening was actual VI changes or just statistical movement due to MOE.

    My guess is that there was probably a modest narrowing then widening based on Labour headline average VI rising whilst the Cons stayed more or less flat.
    Truth is we dont know and it matters little as there is almost 5 months to go; except Labour supporters will go in to the festive season with a tad more cheer than Cons ones …..probably.

  8. @Ernie

    Snap! Albeit in different words.

    This is why a postiori hypothesis testing is normally carried out using adjusted (i.e., more conservative) probability values. Best for the community to agree ‘upturn criteris’ in advance and so avoid quibbles of this kind.

  9. Roland
    Agreed. Of course a lot of people in Texas support the Mexicans in that battle! Replicas of the battle flags at the Alamo decorate Aberdeenshire’s HQ. Apart from our contemporary oil links, a lot of those at the Alamo had ancestry linked to north east Scotland. Dallas is a village in Moray.

  10. “Maybe the Autumn statement combined with Xmas has made CON look a little like Scrooge after all!
    (Though I expect things will return to normal in the new year).”

    Looking exactly like ole Scrooge you mean?

  11. I just did a crossbat anagram and find it goes into automod.

    Must be because it uses a forbidden word.

    Lets just say, for those who do crosswords, that it refers to the decomposition of blacklegs (5,3).

  12. I stand by my analysis.

    CUSUM analysis is a form of control chart. Applying SPC principles to Labour’s December VI puts the last five data points above the mean. If the next two are also above the mean, then you get to seven points in a row and special variation is present.

    It’s close to that. That’s my point.

  13. @CMJ

    I used to know a catmanjeff. Geoff, actually, and he looked after people’s cats for them when they were away.

    Not related are you?

  14. MOG

    Got one that I daren’t post-but most enjoyable to find :-)

  15. @MOG

    As a staunch trade unionist, I’ll have you know that I’ve never been a blackleg in my life!!

    :-)

  16. Dthough I’m sure you’d agree with the sentiments in the answer to MOG’s anagram!

  17. @Norbold

    Ah yes, scab5 rot….

    ..in hell.

    :-)

    P.S. Automod avoidance use of 5!!

  18. By the way, the phone lines have no closed for entries into the Crossbat anagram competition and, after much deliberation by a judging panel that included luminaries such as Anthony Wells and former Archbishop of Canterbury, Rowan Williams, the winner is
    .
    .
    .
    .
    .
    .
    .
    .
    .
    .

    Jim Jam for “R Boss Cat”, with the runner up, RosieandDaisie for “sorts cab”. This e.mail is laid out in this way as a tribute to the inspiration for R&D’s anagram, the lamented and much missed Ally C.

    JimJam will win a deflated Size 5 football, signed by the entire first team squad of Aston Villa FC.

    A big thank you to all the other contributors.

  19. @MOG

    I’m not related, but our house has eight cats (a number of them rescue cats) plus another couple of strays who live outside, but we feed in our garden.

    There are two signs of a civilised house – cats and books.

  20. @CMJ

    I think I qualify as civilised then; 3 cats and 7 full bookcases. However, for total civilisation I feel one also needs at least 80Gb of ripped/downloaded music on one’s PC, as well as shortcut links to Radios 4 and 5.

    And, of course, an electric reclining armchair ;)

  21. Am so excited!!

    First time I have won anyhing so meaningful since a game of Beetle as a 10 year old.

    I guess the second prize is 2 deflated Size 5 football, signed by the entire first team squad of Aston Villa FC.

  22. @MOG

    That sounds most agreeable.

  23. Barney Crockett

    I went to Texas earlier this year and wondered if they would appreciate me in a coon-skin hat but I am afraid that like here younger people don’t know the story very well.

    Barney must be so grateful that he’s a member of Labour rather than UKIP. Otherwise he’d probably have been expelled for that by now. :P

  24. @CMJ

    “CUSUM analysis is a form of control chart. Applying SPC principles to Labour’s December VI puts the last five data points above the mean. If the next two are also above the mean, then you get to seven points in a row and special variation is present.”

    You’re not presenting the whole story, are you? If you define ‘the mean’ as covering just the single Populus poll conducted on 12-14 Dec (with its Labour VI of 36%), then given that the following ten polls were lower, then on the basis of your description you should conclude that there is ‘special variation’ (namely a reliable drop in Labour’s support).

    If you shift on a couple of days to the Ipsos-Mori poll of 13-15th December (with its 29% Labour VI), you then find that this is followed by at least eight polls with higher Labour VIs, and so – for essentially the same period – you infer that there is ‘special variation’ in exactly the opposite direction.

    So, at the very least you need a more specific definition of ‘the mean’ before you can develop a persuasive argument along these lines.

  25. @Unicorn

    So, at the very least you need a more specific definition of ‘the mean’ before you can develop a persuasive argument along these lines.

    I did quote this in the original post:

    Based on You Gov’s VI data for December, the answer is likely to be yes.

    Isn’t that specific enough? (ie the polling company and time period?)

    I inferred nothing beyond that data set.

  26. Survation/Record poll –

    “THE Record survey shows that the SNP continue to dominate the voting intentions of ordinary Scots for Holyrood AND Westminster.”

  27. Survation

    Westminster – SNP 48% (+2) : Lab 24% (nc): Con 16% (+1) : LD 5% (-1) : UKIP 4% Grn 1%

  28. At the risk of looking like I’m trying to reopen a contest already won by the very talented and wonderful Mr (or Ms) Jim Jam…
    do you think that the recent absence of he who Sorts Cab could be connected to the desire, oft implied on this blog (including I think our anagrammatic hero) to bar Scots?

  29. BY our anagrammatic hero

  30. More from Survation

    Scotland Westminster 2015:
    CON 16% LAB 24% LD 5% SNP 48% UKIP 4% Others 2%

    Holyrood Constituency:
    LAB 25% SNP 51% LD 5% UKIP 2% GRE 2% AP 1%

    Holyrood List:
    LAB 24% CON 14% SNP 40% LD 6% GRE 9% UKIP 7% AP 1% => The Scottish Green Party and UKIP would secure Holyrood representation

    Jim is going to be a very busy man…..

  31. So much for Survation absolutely not being released before Monday…

  32. Guymonde

    “To bar [1] Scots”

    I think you are referring to Murphy’s latest failure of an initiative.

    http://www.heraldscotland.com/news/home-news/murphy-why-should-football-fans-face-booze-ban-when-rugbys-middle-class-are-allowed-t.1419156833

    [1] To bar : to alienate the female vote, by encouraging drunken abuse.

  33. @GuyMonde

    If your superb Bar Scots anagram hadn’t come in just after the competition closed, then I may just have deemed you the overall winner (sorry JimJam).

    I was thinking of launching a second anagram competition based on the name Colin. The c and the n were coming in handy but I then realised I was short of the other requisite letters.

    :-)

  34. The Greens on 1% for WM is the Yes tactic of uniting behind the SNP for WM working. The Holyrood vote had Greens on 9% on the list.

  35. ON,

    In English football it is permitted to drink anywhere not in sight of the pitch – seems he is proposing nothing more than equalisation of the law.

  36. Mr N

    In that case, we can look forward to Murphy recommending the raising of the permissible blood alcohol for drivers to rUK levels.

  37. If you had a situation where Con was static and Lab very slowly deflating and then it switched to Lab static and Con slightly deflating then it might look like Lab is moving forward.

    I don’t know if that is what is happening – or just starting to happen – but if it was then it would neatly fit the timing of those telly tales from some months back of an improving economy creating one tide and that tide starting to roll back when no visible sign of that improvement showed up (for most people).

  38. Mr N

    Scotland has had historical problems with health in particular alcohol, smoking and unhealthy diets; Leading to low life exoectancy. Many measures have been taken to combat these problems and at last they are having some success. Murphy’s plan is a retrograde step – the idea that you can enjoy yourself without alcohol is an important point to reinforce. The police, the Scottish NHS, domestic abuse charities are against the plan and it is going back to the problems of the ’80s. Murphy does not drink – so it won’t be him spending his cash with the resultant health and social consequences.

    His priorities since taking office – overturn the anti-sectarian laws and the alcohol ban at football matches.

  39. “Every day lately seems to have another story of how the NHS is having difficulties”

    The PC (politically correct or political class, pick whichever you prefer) argument has always been nonsense on its face.

    Even if it was true that rapid mass low skilled immigration could possibly be good for an economy (unless it was filling empty land) then any benefit couldn’t possibly be instantaneous.

    So importing millions of people *before* building the necessary infrastructure for the extra numbers could only ever have one outcome.

    It’s been obvious for fourteen years but hidden in the sort of areas Lab could lose half their vote and still win. Now it’s too late and gradually spreading all over.

  40. Have you ever thought what state the NHS would be in without immigrant labour, Mr. J.?

  41. Perhaps it would have encouraged Britain’s universities to offer more medical and nursing places to UK students? Just a thought. Apparently we have a few young people who are struggling to find work..

  42. Murphy’s problem I feel will be the laws he wants to ditch the other parties want to reform, but maybe that’s his plan for minority SNP or Lab. Leave negotiation room?

    Polls won’t be useful in Scotland till he gets more time but the amount of work needed in Scotland is large and I wonder if he’s taking the lead on that. Whose to blame if current polls are the vote at Westminster, Murphy or Miliband?

    Either way Murphy should be announcing the big policies, he needs to utilise his team and MSPs, makes it look like drinking at football is manifesto material.

  43. @CMJ

    Great data. One tiny thing. Do you have the Holyrood Constituency Conservative VI?

  44. @Fraser

    Not just big policies but even policies (of any size) that can and will be delivered. That’s part of the problem.

    The other part is that the Conservatives want to go back to the 80s, Labour the 90s, UKIP the 50s and the Lib Dems…I’ll guess at April 2010. :))

    There’s a lack of forward-looking, and with that a lack of forward thinking.

  45. Tables are here:

    http://survation.com/wp-content/uploads/2014/12/December-Scotland-Daily-Record-Scottish-Voting-Intention-ONE.pdf

    Holyrood

    Lab 25% / 24%

    Con 15% / 14%

    SNP 51% / 40%

    Lib Dem 5% / 6%

    UKIP 2% / 7%

    Green 2% / 9%

    Other 1% / 1%

    First figure constituency, second regional. SSP, BNP and Christian not asked on main list, but on second.

  46. Fraser

    “Polls won’t be useful in Scotland till he [Murphy?] gets more time”

    I suspect the underlying problems for LiS are more serious than can be fixed by some presentational/policy changes by a new local “leader”.

    Have a look at Prof Curtice’s comments after the YG Scottish poll.

    http://blog.whatscotlandthinks.org/2014/12/new-leader-old-troubles-labour-scotland/

  47. I suspect the underlying problems for LiS are more serious than can be fixed by some presentational/policy changes by a new local “leader”.

    I’d agree with that whole-heartedly.

    Though assuming that the SNP comfortably top the popular vote in 2015, the real question is what level of improvement LiS would require before they are no longer regarded as having “underlying problems” by the neutral?

    And I say “the neutral” not to devalue the opinions of my many nationalist friends on this site, but because of course no Labour supporter is ever going to say “the SNP have no problems now and will be the dominant party going forward”, nor would an SNP supporter ever make a reciprocal comment about Labour if electoral performances were to revert to pre-2007 levels in Holyrood/pre-2015 levels at Westminster.

    Would an end to their underlying problems be signalled by re-establishing themselves as the dominant party in Scotland?

    Or would it be reflected by Labour being competitive with the SNP and winning roughly as many elections as they don’t?

    Or somewhere in between? Normally comfortably top (though not necessarily as far ahead as 2010) at Westminster, normally second at Holyrood but usually strong enough to prevent an SNP majority, for instance?

    Obviously the root cause of political problems run far deeper than individual elections or even electoral trends, but speaking as a pluralist the fact that the SNP’s baseline has risen is not, in and of itself, an underlying problem for Labour (though it’s certainly a massive thorn in their side). I say that because the more competitive parties there are, the easier it is for a party to position itself to consistently appeal to a core of voters and then try to reach out from there.

    Labour’s underlying issue (at least, if the past four months are anything to go by), is their seeming inability to draw voters who – other than being highly unlikely to ever vote Tory and nearly as unlikely to ever vote UKIP – are open to the idea of voting for several parties under the right circumstances.

  48. Mr Jones
    @Norbold: Have you ever thought what state the NHS would be in without immigrant labour, Mr. J.?

    Not to mention the empty lands of E.Anglia, where labour is the main constraint on the mass production of green vegetables for UK supermarkets.

  49. There’s always a temptation to see narratives in polls, to ignore those showing no movement, latch onto those showing exciting looking changes and build an explanation and a story around them. It’s normally wrong to do so.

    The one upside to doing so is that it can at times provoke (admittedly mostly junior and mid-ranking) politicians into making statistical claims that they later regret.

  50. NORBOLD
    “There’s a lack of forward-looking, and with that a lack of forward thinking.”
    Given that, in, say, the US, Con and Lab policies are indistinguishably centre left, might you, I wonder, rephrase your post as one related to a relative lack of research based policy and its evidence in planned investment? Will this be apparent in policy statements and manifestos, and have salience in the VI of the thinking swinger?

1 2 3 4