Two types of outlier

This morning TNS released a new poll showing figures of CON 28%, LAB 35%, LDEM 6%, UKIP 18%, GRN 7% (tabs). The seven point Labour lead is striking, and out of line from the broader trend. Usual caveats apply, but for once I haven’t seen many people over-react to it. Most have sensibly enough seen it as the just the other side of the coin to the ICM poll earlier this week showing a solid Tory lead – two outliers in opposite directions. However, it is worth looking at the different reasons why these two polls went against the trend.

On average ICM produce figures that are slightly more Con/less Lab than average, but only by a tiny bit, generally they show Labour -v- Tory leads that are much the same as those from other companies. The reason that their poll on Monday showed a four point Tory lead when other companies are showing the race almost neck-and-neck is almost certainly just random sample error. Samples vary from day-to-day, month-to-month and sometimes you get one that’s a bit Laboury or a bit Conservativety in ways that weighting does not correct. I expect ICM’s next poll will be in line with those from other companies.

TNS is a different sort of outlier. The seven point Labour lead may be different from that shown by other pollsters, but its actually in line with TNS’s previous polls. In their previous four polls they showed Labour leads of 6 points, 0 points, 7 points and 7 points. Over on my chart of house effects TNS’s polls on average show a Labour lead three points larger than other companies (and if anything the gap is growing!) This isn’t random sample error, this is a consistent methodological difference between TNS and other companies, and unless they switch methods I expect their future polls will continue to show bigger Labour leads than other polls.

Two outliers, but one is probably just a random blip from a pollster who normally shows the same as other companies, one looks like a typical poll from a pollster who regularly produces polls that show a bigger Labour lead than the pack.

311 Responses to “Two types of outlier”

1 3 4 5 6 7
  1. @Statgeek

    I think all 4 Fife seats are very tricky to call.

    The safest Labour seat Kirkcaldy & Cowdenbeath is complicated by how much of Gordon Brown’s vote Kenny Selbie can rely on.

    Both Glenrothes and Dunfermline had by-elections in the 2005 parliament that probably were still affecting the numbers in 2010.

    Anecdotally the strong Yes areas were Glenrothes and Kirkcaldy which are the seats with higher Lab majorities.

    UNS will suggest Dunfermline being most vulnerable to SNP but it is much more Edinburgh commuter belt and perhaps more No friendly so may stay Labour. But the historically high Lib Dem vote here will collapse and where will it go?

    North East Fife had a huge Lib Dem majority but with their collapse and Menzies retiring it’s a genuine 4 way marginal seat.

    You can plausibly argue for results of:
    4 Lab
    4 SNP
    NE Fife to be Con or Lib Dem with 3 Lab or 3 SNP

    My guess is SNP and Lab will each take 2 out of 4 but I wouldn’t want to bet on which 2!

  2. Mike Smithson’s I must assume is talking about the Survation Scotland poll just out. (I popped over thinking a post on it would be up. Maybe it’s further down the page & I missed it) Anyone who considers it a ‘Labour recovery’ is an extraordinary optimist, to put it mildly. It is:

    SNP 45% (-1)
    Labour 28% (+2)
    Conservatives 15% (+1)
    Liberal Democrats 5% (-2)

  3. On a modelling note it pains me to say it (and hopefully it won’t be necessary after the general election) but Scotland Votes really need to improve the way their Holyrood List prediction models UKIP.

    It’s just not plausible that 7% on the list will deliver 0 MSPs as it would require very even geographical spread and unlucky finishes in each region.

    More likely they’d return 2-3 MSPs on that sort of performance. Depending on which parties lose out it could affect the chances of an SNP majority (though not on this poll’s numbers) so it would be worth someone trying to model it properly.

  4. @ChrisLane1945

    Crosland was an erratic politician, I grant you, but a great intellect and political thinker. This, from his inspirational 1956 book “The Future of Socialism” could apply to today

    “In Britain, equality of opportunity and social mobility… are not enough. They need to be combined with measures… to equalise the distribution of rewards and privileges so as to diminish the degree of class stratification, the injustices of large inequalities and the collective discontents.”

    This is political poetry and much more inspirational than Herbert Morrison’s prose of “Socialism is what a Labour Government does”.

    More Crosland and less Morrison and we’d have had much more successful Labour Governments than those that held office after 1951.

    That’s my view anyway.

  5. @Oldnat

    Ahh, you’re on 2016. I was on 2015.


    I have a feeling that the LD collapse might let the SNP in the back door in NE Fife. It is potentially a 3-4 way marginal if the Lib Dems go to SNP, Con and Lab in the right proportions (assuming they are going at all).

    Pretty much agree with the other three…at one time there was no way they Kirkcaldy or Glenrothes would go anything but Labour, but now there’s any amount of speculation. Dunfermline perhaps Lib Dem, but it has to be Lab or SNP now I feel.

    I’m hoping for an Ashcroft poll or four. :))

  6. Northumbrianscot

    Scotland Votes have said that they will update the Holyrood model to include UKIP.

    My guess is that they are waiting to see the distribution of Westminster votes around the regions, before they construct the model. What else would they have to go on?

  7. @ Spearmint

    I think you made the mistake of evaluating the actor in isolation from the audience.

    The audience for the time being says that they, to put it mildly, don’t like EM.

    However that was true for the later PM you mentioned. It was the audience that quite suddenly (for various reasons) started to listen to the content-free inspirational stuff.

    The question, in this respect, whether there are compelling reasons why the audience would want reflect to their own thought processes. If they find them, they will and EM may fly. I have to say I can’t see any.

    On the other hand, Labour can form the government irrespective to EM. However, then the non-PM quality (whatever that means, it is a mystery to me) would hang just righ above his head.

    EM is good when there is an “enemy” so perhaps they should find one. On the other hand he looks like a peaceful chap. So it’s his fault again.

    Get the people think then is the advice, but it’s the most unrewarding task. so being wooden is perhaps better.

    I’m actually more concerned by his team.

  8. Statgeek

    No. The regions don’t exist for Westminster, but the same geographic crossbreaks for the List are also given for the 2015 VI.

  9. Statgeek – 7 Westminster constituencies to look at on the figures I gave are the 4 Fife ones, Perth, Ochil and Stirling.

  10. OLDNAT

    “Apparently it is the SNP’s smallest lead since the Referendum. It was 24% in December”

    I suppose if you ignore the 10% lead polls it could be.’

    Sorry – I was referring to Survation!

  11. Just before I tuck in for the night.

    Labour currently have outright control of 3 Scottish councils, Glasgow West Dumbartonshire and North Lanarkshire.

    All 3 areas voted Yes and Ashcrofts Scottish constituency polls had the SNP hoovering up Labour seats in the same areas.

    Food for thought………night peeps.

  12. @Oldnat

    Ah ha. Given that Ochil is another Labour seat, doesn’t that translate to less Labour hopes?

    Ach, now I’ll have to work out the Mid & Fife 2010 average…but not tonight, and will be in Aberdeen for a couple of days. Even geeks need chill time. :))

  13. @Oldnat

    I suppose that makes sense re UKIP. I’m increasingly expecting at least 1 UKIP MSP in 2016.

    Given David Coburn seems unafraid of a double mandate this year it might even be him.

    Also he has a number or regions to choose from depending how the GE results look:
    Glaswegian background,
    his possible Edinburgh residency in 2014 (although his registered address for the Euro elections nominations was in Kensington and he didn’t vote in the Independence Referendum owing to not being registered in Scotland).
    New found love of Aberdeenshire as he’s running in Gordon this year.

    My guess is he’ll pick either Lothian or South of Scotland.

  14. @Peter Crawford
    “I think that’s why Foot beat Healey. yes he was on the left, but he was very good in the chamber of the house of commons at a time when the PLP exclusively got to choose the leader”.

    Not how I remember it. being reported at the time. Foot was thought of as a compromise candidate, someone who could bridge the chasm between right and left, chiefly because of his CND credentials.

    His reputation as a great orator was discussed, but wasn’t thought of as a decisive asset: it was usually followed by “but is that what we need in the television age” or some such prevarication (and a perfectly reasonable prevarication it was).

  15. Graham


    4 Survation polls since the referendum

    Nov – SNP 46% : Lab 24% : Con 17% : LD 6%
    Dec – SNP 48% : Lab 24% : Con 16% : LD 5%
    Jan – SNP 46% : Lab 26% : Con 14% : LD 7%
    Feb – SNP 45% : Lab 28% : Con 15% : LD 5%

    That doesn’t seem like a pattern of change to me

    I’ve seen Smithson’s graph – but he is a former LD!

  16. Foot was also helped by the fact that several rightwing MPs who were already intent on leaving Labour to form a rival party – which turned out to be the SDP – deliberately sought to destabilise the party by voting for him rather than Healey. I am thinking of – Neville Sanderson – Tom Ellis – Dick Crawshaw – John Horam. Given that Foot only beat Healey by 10 votes it only needed 5 MPs to vote in such a dishonourable way.

  17. @OLDNAT

    If it is a pattern of change, and I suppose it might be, then it’s a pretty pathetic one. Labour may have gained 4 points in 3 months and since there is no significant change in the SNP figures, that gain can’t have come from the SNP. At that rate, Labour might reach 32% by May but the SNP would still be at about 45-46% (strictly by this pattern).

    It takes a few seconds of plugging that into any swing meter on the internet to see exactly how bad that would be (for Labour. SNP would chuckle and head for Westminster and fun and games)

  18. OLDNAT
    ‘4 Survation polls since the referendum

    Nov – SNP 46% : Lab 24% : Con 17% : LD 6%
    Dec – SNP 48% : Lab 24% : Con 16% : LD 5%
    Jan – SNP 46% : Lab 26% : Con 14% : LD 7%
    Feb – SNP 45% : Lab 28% : Con 15% : LD 5%
    That doesn’t seem like a pattern of change to me’

    Well – a 7% drop in the SNP lead over 2 months is not insignificant.

  19. JR Tomlin

    Let me add the UKIP & Green figures to that list

    Nov – SNP 46% : Lab 24% : Con 17% : LD 6% : UKIP 5% : Grn 2%
    Dec – SNP 48% : Lab 24% : Con 16% : LD 5% : UKIP 4% : Grn 1%
    Jan – SNP 46% : Lab 26% : Con 14% : LD 7% : UKIP 4% : Grn 3%
    Feb – SNP 45% : Lab 28% : Con 15% : LD 5% : UKIP 3% : Grn 3%

    The variations in the rounded figures are too small to suggest much, but it wouldn’t be surprising if some non-Lab Unionists are more motivated by dislike of the SNP than of Lab.

    Murphy’s speech tonight may indicate that he is reaching out to such people –

    Tom Gordon tweet
    @jimmurphymp at RSE tonight: “The Labour Party is not… going to increase the council tax. I’m not willing to increase the taxes on the middle classes, the mainstream middle class”

    or perhaps ensuring he keeps his own seat, at the expense of some colleagues?

  20. Seems to me the only variability in Scotland is between the SNP and Labour. LD’s are stuck on 5%, Conservatives on 16%. The main question is are whether the SNP will get 45% or nearer 50% of the vote, how many being decided by how many transfer from Labour.

    It’s highly probable that the SNP will be involved in running the UK next time to some extent. Or else get in the way of running it, if a major party who they really dislike tries to form a coalition government and the LD’s do as badly as predictions suggest.

  21. Oldnat
    Let me add the UKIP & Green figures to that list
    Nov – SNP 46% : Lab 24% : Con 17% : LD 6% : UKIP 5% : Grn 2%
    Dec – SNP 48% : Lab 24% : Con 16% : LD 5% : UKIP 4% : Grn 1%
    Jan – SNP 46% : Lab 26% : Con 14% : LD 7% : UKIP 4% : Grn 3%
    Feb – SNP 45% : Lab 28% : Con 15% : LD 5% : UKIP 3% : Grn 3%
    The variations in the rounded figures are too small to suggest much, but it wouldn’t be surprising if some non-Lab Unionists are more motivated by dislike of the SNP than of Lab.

    …or perhaps it’s churn.

  22. raham

    It’s only significant if you pick the polls to compare, and ignore things like moe and rounding (Feb lead is actually 17.7%, so 18%, not 17%).

    Indeed the Jan lead was 19.6%, so a 1.9% drop in one month. If that get’s you excited ……. :-)

  23. Funtypippin

    Or indeed, churn.

  24. Looking back at those Survation figures I’m curious about how they fit into Unicorn’s quadratic model.

  25. It’s like the referendum all over again -first it was Keller telling Scots that voting SNP will let Cameron in number 10, now it’s Damian from Survation. I don’t know how the unionists manage to get people with reputations to spin their lines but it is an insult to our intelligence.

    There is no point in a rehash but it is very clear that more SNP MPs will not lead to a Tory government, the only thing that will do that is England voting Tory.

    So why do people that must know the truth persist in this nonsense?

  26. The point is if SNP stay in mid-40s even if Labour gets to 30% Labour will still lose most of their seats.

  27. Really nothing much is happening in Scottish polls or crossbreaks we have McPolldrums

  28. Couper

    Fear of the unknown and potential change will play a part.

    Those in Scotland have had an SNP government since 2007. Consequently, they are liked, disliked or people don’t care much – but they aren’t an unknown.

    Even highly respected people like Lewis Baston found that his preconceptions changed when he left London and talked to a range of people here.

    Scots would be exactly the same if we suddenly faced the prospect of our domestic policies being affected by wild Cornishmen from Mebyn Kernow, that our media had presented as being eccentric pirates and ship wreckers :-)

  29. Well to be fair if UKPR is anything to go by some of you are quite eccentric.

  30. Funtypippin


    That’s what happens when you don’t send in Church of England bishops to keep us under control.


    I can assure you, that ancient nature of the Sheffield Conservative Association is not to be doubted.
    The university CF has recovered to around 10 active members, and the vast vast majority (around 210/~280) of the city-wide party membership is in Hallam, particularly strongholds like Dore – and it certainly isn’t young.

    On Hallam, Ian Walker is a decent candidate, and unlike every other hustings candidate pitching to be PPC, he was the only one from the city – and also the only one who thought he could win.
    From what I understand, his tactical pitch is a “Clegg’s finished, it’s Con vs Lab”, but beyond standard national campaigning points, he’s also pushing a fairly strong “ethics in politics” angle as well.

    If he’d been selected a year ago and campaigning on these points before the first constituency polls started coming through, I would say he might have had a decent chance. But now I think it’s a little late to over-ride both a mindset and a string of Lab-lead Tory-3rd polls pushing a centre-right vote to tactically retain Clegg.

    I somewhat suspect CCHQ doesn’t overly want to lose Clegg (or rather, his nominal ability as leader to make centre & centre left MPs to support centre right government) either, and resources won’t be forthcoming.
    For the opposite reason I feel UKIP have selected their ex-LibDem 20 year old student candidate with a betrayal narrative aimed at the student population (basically messaging to hurt Clegg, but with as little Tory collateral as possible).

    Hallam is certainly the most interesting & non-determined individual constituency result of the election (more than if there’s another green, or any SNP scalp, Farage’s chances etc), and the one to have the most individual impact on the composition / nature / success of any governing arrangement in the next parliament.

  32. @Oldnat / Northumbrianscot

    I had a spare ten minutes: (top four charts)

    So based on Oldnat’s quote: “SNP weakest area – Mid & Fife SNP 41% : Lab 27%”

    We’re seeing:

    SNP 41% (+20%)
    Lab 27% (-14%)
    Lib 11% (-8%)
    Con 9% (-8%)
    UKIP 5% (+4%)
    Green 7% (estimated about +6.5% I think)

  33. @GRAHAM

    It’s only 7 points if you arbitrarily start with the second poll, the December one, rather than the first in November. The December one appears to be an outlier anyway as it is outside the pattern of all the other Survation polls.

    The gap closed five points in three months starting with the November poll. To call that significant is stretching it a bit, especially with only three more months to go, but the movement could be real and Labour gaining votes from somewhere although few if any appear to come from the SNP. Judging by the additional percentages OldNat posted, they probably come from everyone BUT the SNP.


    McPolldrums indeed! *yawns*

  34. There appears to be a very slight recovery in Labour’s position in Scotland, and perhaps it will continue or even accelerate in the course of the campaign proper, but it would need to accelerate a lot to deliver a significant number of seats.

    There is a bit of a precipice, as I recall, when Labour get within about 5 points of the SNP when they might pick up a lot of seats but that seems a long way off and Labour is running out of time.

    I think Ed should be calculating that he needs at least 270 seats in England and Wales to be the largest party. That would probably guarantee him Number 10. But that’s a net gain of about 50.

    Does anyone know what E&W swing is required to deliver that?

  35. The Indy is running another Tax Avoidance/Tory Donation story. I doubt that this issue will impact the GE much but it is putting the Conservatives on the back foot for a few more days and delaying the opportunity to start the recovery they need.

    Conventional wisdom and history suggests that there is little if any Con/Lab swingback in the campaign itself so DC has about six weeks to get the Con VI moving. Unless UKIP/Con swing back is of the late variety. No-one knows.

    Without 3 or 4% of Blue kippers losing their nerve and returning home GO will need to deliver a very good budget. Or the debates will have to give DC a boost.

  36. Today’s YG Scottish crossbreak

    SNP 43% : Lab 26% : Con 17% : LD 3% : UKIP 4% : Grn 7%

    Mean of last 20 YG Scottish crossbreaks

    SNP 42% : Lab 26% : Con 19% : LD 5% : UKIP 4% : Grn 4%

  37. @Statgeek.

    Very interesting thanks.

    If we apply a UFS (Uniform Fife Swing) on those numbers which seats change hands?

  38. @ OLDNAT Scot X breaks

    I will be interested to se what Chris Lane thinks of the LD number.

    Good Morning; windy here on the beach.

    I think the LD figure looks high in Scotland, England and Wales; inexplicably

    In terms of England and Wales it looks like Labour are on track at the moment to pick up forty Tory seats and ten LD seats, which takes Labour to about 298 seats.

  40. All this talk of Fife made me think of this:

  41. Chris Lane

    Good morning, little breeze here in Delhi today.

    Does your 298 suppose that Labour will lose only 10 seats in Scotland? I would expect more.

    So that would be 258 minus (say) 28 Scots losses = 230 plus 50 (your figure) = 280

    SNP on about 40

    So other parties (PC. GP, SDLP) plus the SNP would be necessary to achieve a bare majority or Lib Dems to achieve a working majority.

    Yes, I think there is little chance that either Tories or Labour will have an overall majority. The Tories will pick up, I think, about twenty Lib Dem seats., losing about forty seats to Labour, taking them into the 287 region.
    Labour and the SDLP and Green could well ‘draw’ with the Tories.

    77 days!

  43. I still think that Labour will win a majority, because I expect them to get about 34%, and the Tories to lose votes to UKIP during the campaign. I don’t think that the Greens will cost Labour many seats, if any, because I suspect that Green voters tend to be quite tactical and will vote Labour in Labour-Tory and Labour-Lib Dem marginals.

  44. Well if we are all predicting what happens in May, I go for:

    Tories: 37.8
    Rest do not matter, much

  45. I’m waiting for the next salvo from @Pressman’s arsenal.

    At the turn of the year he told is that certain sections of the Press would hit Labour in two ways: firstly, by getting senior Labour figures to lay into him and, secondly, by using prominent celebrity figures to do the same.

    We’ve already had the former in the form of Alan Miburn on the NHS, and the latter with Messrs Ant and Dec. Predictably these events have had no discernable impact on the polls.

  46. This makes me think of Fife…

    Great place, lovely people….wired accents!


  47. @Bill Patrick

    I think what’s going to make the May election one of the most fascinating, intriguing and unpredictable of recent times is the extent to which the ground war will supersede the air war. By that I mean that while the posters, hoardings, press conferences, TV debates etc will gain inordinate amounts of media coverage, the key to the outcome will reside with the many local battles that will be waged, constituency by constituency, region by region. The air war will make great noise, and could swing some votes, but how Labour fare against the SNP in an entirely Scottish affair, and how the Tories get on in battling UKIP in the marginals, and what the Greens do when confronted with tactical choices, will determine who gets to Downing Street. Local campaigning strategies, efficacy in distributing postal votes, success in boosting voter registration, key local issues, the quality of candidates, the strength of local party organisations; this is where it’s going to be won and lost in what promises to be the closest race since February 1974.

    I’m less convinced that the national media and social networking wizardry is going to swing it. Rather, perversely some might argue, I think the old fashioned electioneering techniques and virtues will hold sway in May.

  48. @ blubob –

    so you are predicting that the tories will increase their vote share on 2010 (never been achieved by a full term government) and that UKIP will have no impact on them? Meanwhile lab will not benefit from the lib dem collapse?

  49. @ChrisLane1945
    @little Red Rock

    Current polling loses the Tories a good deal more than 40 seats to Labour. Latest Yougov with its Scottish crossbreak gives Lab 311, Con 253, LD 18, SNP 43, Others 25

    On those figures SNP wouldn’t matter to anyone – Lab – Lib deal would be enough.

    But we’re all waiting for swingback!

  50. Bluebob

    I can agree with you on Tory 7.5% lead but I’m not bold enough to predict the actual numbers at the election, at least not yet.

    Another lovely morning here, too good to waste. Have a good day all.

1 3 4 5 6 7