Marginal polls

Back in May when ComRes first launched their marginal seat Omnibus I wrote about some of my reticence towards marginal polling, why it isn’t usually quite as useful as it should be, and why I hoped that might change. Marginal seat polls matter because they are the seats that might change hands, and therefore the seats that will decide the election. If they behave differently to the national polls, and if different groups of marginal seats behave differently to one another, it’s obviously a very big deal.

What has limited their usefulness in the past is their infrequency, and the lack of comparability and empirical testing. Marginal polls used to only come along occasionally, varied a lot, polled different groups of seats, and didn’t often happen right before elections so weren’t tested against reality, meaning methods weren’t finessed and improved over time in the same way national polls are.

In practice their rarity and inconsistency rendered them a very blunt tool when we’re looking to spot quite subtle differences – the reality is that marginal seats aren’t that different from the country as a whole:

  • In English & Welsh seats at the last election (the swing in Scottish seats is consistently different) the average swing from Lab to Con was 5.8%. In the 50 most marginal seats the swing was 5.6% – no real difference at all. In the real core battleground (Lab-v-Con seats), there was a slightly more noticeable difference, but it was still small. Amongst all Lab-v-Con seats the swing was 6.7%, amongst those with a majority of less than 10% the swing was 8% – so 1.3 percentage points bigger.
  • In 2005 the average swing in all English seats was 3.2%. In the Lab-v-Con battleground seats it was 3.5%, in Lab-v-Con marginal seats the swing was also 3.5%. No difference.
  • In 2001 the average swing in all English seats was 1.6%, the average swing in Lab-v-Con seats was also 1.6%, the average swing in marginal Lab-v-Con seats was -0.5% (that is, overall there was a small swing to the Conservatives, but on average there was a tiny swing to Labour in the Lab-v-Con marginals).

You can see that marginals do behave a little differently sometimes – the Conservatives managed a better swing in their target Labour marginals in 2010, Labour did better in those seats where they had fresh incumbency in 2001 – but the differences aren’t huge. We’re talking 1 or 2 percent difference. That’s enough to make a genuine difference in seat numbers, but is very difficult to determine from a single opinion poll. The difference between the national picture and the marginal picture will normally be so subtle that it could easily be lost under or mistaken for normal sample variation, or the methodological differences in doing marginal polls (or vice-versa, normal volatility or methodological impacts could be mistaken for a different pattern in the marginals when there is none).

More recently though things have been looking up. We’ve seen an increase in marginal polls and, more importantly, we’ve seen an increase in regular marginal polls – Lord Ashcroft and ComRes are both doing regular polls of the same groups or group of marginal seats. Different pollsters are also doing marginal polls of roughly the same marginal seats – Ashcroft, ComRes and this week Survation have all done polls that include ultra-marginal Conservative -v- Labour seats. However, despite covering the same ground, the results are very different.

The table below is an attempt to make the results roughly comparable. There are much more obvious differences between different battlegrounds (that is, between seats that are Con-v-Lab battles and seats that are Con-v-LD battles), so I’ve looked at only the Con-v-Lab battleground – those marginal seats with the Conservatives in first place ahead of Labour. Sample size for each poll is just for the Con-v-Lab marginals, the swing just those seats, and I’ve compared it to the average of national polls at the time of each marginals poll’s fieldwork.


As you can see – three companies, three completely different stories. ComRes show the Conservatives doing much better than nationally in these key marginals. Lord Ashcroft shows very little difference between the national picture and equivalent marginals. The Survation poll today showed Labour doing much better in similar marginals.

Some of the differences will be methodological. For example, Ashcroft uses the two stage question wording to try and coax out local considerations though frankly it makes very little difference in Con-v-Lab marginals. I don’t think ComRes prompt for UKIP in their marginal polls, but Survation and Ashcroft both do. The weighting regimes are very different – I think Ashcroft weights by age, gender, social class and recalled vote; ComRes weight by the same plus housing tenure; Survation appear to weight only by age and gender, with no political or socio-economic weights. Lord Ashcrofts poll are also, it’s worth noting, of a substantially larger size – they are an aggregate of full size single constituency polls, rather than a poll of a group of marginals.

You pays your money, you takes your choice. My own expectation is that, if there is a relatively small Con to Lab swing the Conservatives will do very slightly better in the marginals thanks to the double-incumbency effect – the historical evidence for such an effect is extremely strong and I see no obvious reason for it not to happen this time round. If, on the other hand, there is a hefty swing towards Labour then it might well be cancelled out due to a stronger performance in Labour target seats, like we saw for the Conservatives in 2010 or Labour in 1997. Time will tell. Either way, I wouldn’t expect Con-Lab margins to perform radically differently to the national picture – if there’s a systemic difference between marginals and the country as a whole, I’d expect it to be a small one. In a close election that could still be the difference between a majority and a hung Parliament, so don’t underestimate its potential importance, but it would be a remarkable election if the swing in marginal seats really was 4 or 5 points bigger or smaller than the national picture.

102 Responses to “Marginal polls”

1 2 3
  1. Smithson is a big troll who makes posts he must know are total psephological nonsense as clickbait.

    But he does make nice graphs. And he digs into the figures and notices trends, which allows him to pick up on stuff like this.

  2. I doubt whether this will be welcomed by many folk. It might just have a teenzy effect on VI, depending perhaps on whether or not you bought shares in Royal Mail.

  3. @Spearmint

    It’s not polite to make personal remarks. Not all of us can be young and beautiful like you.

  4. “‘Output associated with private sector housing remained some 8.2% below its pre-downturn level in Q2 2014, while public sector housing output was 76.1% above its pre-downturn level in the same period: the highest quarterly level since 1997.’ – It is public sector housing driving the growth – surprising IMO”


    Yes, that is a bit of a surprise. We’ve been of these new housing starts… we didn’t hear anything about it being public sector!! ‘Course, we could do with knowing the absolute proportion of public vs private housing starts…

  5. DAVE
    “(I don’t mean that all the constituency work a good MP does is aimed solely at getting re-elected. Just that his future opponent has no chance to do it.)”

    That’s not entirely true: Lab candidate intervention with which I am familiar led to the reinstatement of the Sixth Form College place and bus rights of a constitiuent; review if a waste water system reconstruction charge of a group of families; and a highly public fight for the saving of a Roman town. Incumbency of the Lab candidacy could also be seen to carry support from popular association with public sector workers and unions, for example in the context of local hospital and factory closures.

  6. …been told of these new housing starts…

  7. @Spearmint

    Though I thought Mr Wells had covered this Populus thing a bit ago? He didn’t do a graph, just some numbers. Have you been away and missed it? (Other possibility is that I’m bonkers, which as Trollson would tell you, is a fair bet). X

  8. AW
    “John Pilgrim Your comment is awaiting moderation.”

    True though, and not intended to be partisan. Opposition candidacy of any stripe, especially if continuous over several elections, can lead to the candidate being “well regarded” and does carry incumbency support.

  9. AW
    Back to the Marginals, which of them does Lord A use/ define as ultra?
    Sorry if I have missed this info elsewhere.

  10. @Roger M

    Loved the post about Blair, the Stuarts et al. I believe you are wrong about James I/VI though. The convention, I believe, is that future monarchs should be numbered by the highest number, so there’s no need to rewrite all the textbooks (which always brings to mind the Priests of Amun chipping the names of Akhenaten and Hatshepsut of the walls of temples to pretend it never happened).

    It’s worthwhile pointing out that unless you’re a fairly recent immigrant to the UK and have no known ancestors in the UK, you’re bound to be a descendant of William the [email protected] Conqueror and for that matter, by way of the Old English royal line, from Woden too!

  11. @Ewen

    I’ve not seen him give a definition beyond “constituencies that will be among the most closely contested”.

    I think it’s worth remembering though that if Ashcroft didn’t do this polling – just for his own amusement you understand – then the Conservative Party (which he is a supporter of) might have to pay to do it themselves.

    Of course the polls in themselves are perfectly respectable.

  12. Give us the math on the William the B descent. My ex-wife claims she is descended from his uncle, Rollo the Deeply Unpleasant !

  13. @ Postageincluded

    Did you see the 10th Anniversary of “Who do you think you are” the other night. Compulsive viewing if you didn’t. They showed Boris looking totally confused, he was standing in the large hall of a German Royal Palace while an expert showed him one picture at a time of his early ancestors until the last which was William II.

    They also showed Xander Armstrong speechless as he found his family line went right back to our Norman friend William and the best was Matthew Pincent as he watched an expert unroll a vellum family line which went back to God via Adam & Eve. Priceless!

  14. @ Postageincluded,

    It’s not polite to make personal remarks

    I say it with affection.

    (And yes, Anthony did cover the Monday/Friday thing earlier and showed that statistically there was no difference in the polling from earlier this year. But in the light of Smithson’s graph I think it might be worth revisiting. The lead distributions barely even overlap- a T-Test is giving me p = 0.00047. That’s probably the wrong thing to test because the lead isn’t actually what the polls are measuring, they measure VI, but you have to admit it looks odd.)

  15. @Ewen

    Well I suppose it’s better than being descended from Rollo the Chocolate Covered Caramel!

    Can’t give you the maths I’m afraid, though I don’t think there’d be much problem finding it on line, It’s something I first saw calculated 30 years ago – in the Observer I think – though for a much more recent figure (Commonwealth period?), not William the Genocidal. You’ve also got a good chance of being descended from Genghis Khan. What is it with your family?

  16. @Spearmint

    I’d remembered that completely wrong then. Apparently vitamin D is the cure…


    Did you see last night episode.

    A real eyeopener on the British Aristocracy in Ireland. Truly appalling.

  18. Ah, yes, Ewen Lightfoot, my wife can trace back to Rollo (and Charles the Simple) too, via some Saxon maid who married into Norman posh. I had only a quick glance at what that means, but once you hit the big names genealogy becomes a lot more ‘recorded’; and so far as I could see that Rollo/French royalty connection leads back to – among many notables – Boadicea and Old King Cole, if Colus was old King Cole, the royal families of the Goths and the Burgundians, Roman emperors such as Claudius and Nero, Livia, Mark Anthony, not Julius Caesar, but via the Livii back to the old aristocratic families of Alba Longa. Which means what then? Do as you’re told when your wife tells you, I guess.

  19. @Carfrew

    ‘we could do with knowing the absolute proportion of public vs private housing starts…’

    Much smaller than the private sector, I think, but it is interesting what you can do with statistics – even the ONS does it

    ‘Annual housing starts totaled 133,650 in the 12 months to
    March 2014, up by 31 per cent compared with the year before, which sounds good until you find that the year before was only 102,023, so a total of 31’627 more houses started in a year ‘

    ‘Starts are still 26 per cent below their March quarter 2007
    peak,but are now 113 per cent above the trough in the
    March quarter of 2009’ – 62,747

    and i read somewhere that we need 200,000 to 250,000 new properties to keep up with demand


  20. Postage
    Back to these ‘ultra – marginals ‘ the use of this phrase surely implies prescision does it not , he’s going after the most knife edge situations. And yet he also has the bludgeon of his huge sample size of over 12, 000. Presumably to silence doubters of his polls’ validity . It would be nice to know which individual constituencies Lord A considers ‘ultra’. ( Spearmint you’re usually pretty good at explaining the doings of various Noble Lords, any thoughts ? )

    Bluey , welcome !

  21. @ Colin

    Certainly did, our family has Irish roots and had big troubles with Palmer the *******!

    To be fair, there were landowners who didn’t react in the same way when times became tough, the Marquis of Rockingham being the most notable. A lot of Irish land was rented to the Irish by the Irish, it’s a bit of a myth that Ireland was wholly owned by the Aristocracy!

  22. Ewen – actually it’s my term, not Lord Ashcrofts!

    His poll covers the 12 *most* marginal Con-v-Lab seats, hence “ultra marginals”. He also polled Great Yarmouth and South Thanet because he thought they were particularly likely to have interesting UKIP results – I exclude them from my calculations because they were chosen specifically because they were atypical.

  23. @Ewen

    He’s not a member of the Polling Council, despite being invited, so presumably we are not going to find out, ever. This does mean that Mr Wells weights his results down by 50% in the UKPR Average – which is just as well as it’d be all over the place otherwise.

  24. @ Postageincluded,

    No, no, you remembered exactly right. I just think someone ought to look at it again. (And I’m trying to tempt Roger Mexico or Phil or possibly Anthony into doing it so I don’t have to.)

    @ Ewen,

    You’re never going to let that Oakeshott thing go, are you? ;)

    But in this case I can speak definitively, because Lord Ashcroft always tells us which marginals he’s polling when he publishes a poll. If you’re curious you can just go to his website and look at the summary pdfs.

  25. @Mr Wells

    You’re right.I was reading Ashcrofts introductory articles, which are vague on the point. The full results explain clearly (possibly becasue they were written by someone with less of an axe to grind?)

  26. @Spearmint

    Haha! and there was me trying to tempt you into doing it.

  27. @RM and JP
    Yes, “no chance” was a bit strong. I didn’t include the truth that any MP’s future opponent can do “good works” and enhance his/her reputation locally. However, the original comment noted that such opponents are often not selected by their party until quite near to the election and so have less chance to have their work related to their status as the potential MP.
    The serving MP’s profile is higher and he has a very considerable advantage in that he can raise issues in parliament, and if a member of the governing party, has contacts and sometimes a close relationship with ministers.

  28. Ewen Lightfoot

    Back to these ‘ultra – marginals ‘ the use of this phrase surely implies prescision does it not , [Ashcroft’s] going after the most knife edge situations. And yet he also has the bludgeon of his huge sample size of over 12, 000. Presumably to silence doubters of his polls’ validity . It would be nice to know which individual constituencies Lord A considers ‘ultra’.

    Well the ‘ultra’ bit is what we’re calling then. The core 12 that Ashcroft has polled twice and Anthony analysed above are:

    *Amber Valley (#9)

    *Broxtowe (#7)

    Cardiff North (#4)

    Hendon (#3)

    *Lancaster & Fleetwood (#8)

    *Morecambe & Lunesdale (#12)

    *North Warwickshire (#1)

    *Thurrock (#2)

    *Sherwood (#5)

    Stockton South (#6)

    Waveney (#10)

    Wolverhampton South West (#11)

    Ashcroft also asks about Great Yarmouth and Thanet South as these are UKIP targets. I presume his larger sample is so the individual seats have good sample sizes and that he can also examine the intentions of sub-groups of voters validly.

    The #no is the seat’s target position on Anthony’s list of Con-Lab seats with the smallest percentage majority.

    The * seats were also asked by Survation, who also polled in Brighton Kemptown, Lincoln and Cannock Chase.

    ComRes (who may have actually stopped doing these polls – certainly they didn’t in July) didn’t give a list but it look like their list is by smallest absolute majority rather than percentage, so the list may vary a little.

    The problem is that, if everyone concentrates on the same dozen or so seats, those seats may not be typical of marginal seats as a whole but we will never know.

  29. Green membership is likely to reach 18000 in the next few days and is expected to reach 20000 by year’s end by some.

  30. iPopulus give us the likelihood to vote to look at first – which I find an interesting pass-time. Scotland seems regularly to have the highest ‘certain to vote’ figures, which is probably a result of the highly charged atmosphere of the past few years.

    According to Populus, in comparison with 2010, SNP are up about 8%, whilst Labour are down about 5%. Tories up 4% and LDs down 9%.

    But in seat terms, (Westminster Scotland Votes site) this makes almost no difference to Labour, whereas the SNP gain only 2 and the Tories 3, with LDs losing 4.

    I would caution against people getting too worried/enamoured of the idea concerning Labour losing a lot of Scottish seats to the SNP. Although the SNP have a strong heartland in the North East, they are fairly evenly scattered around the rest of the country and it isn’t until Labour drop to 32% and the SNP reach 36% that things really start to get difficult for Labour. SNP have to be 7 points ahead of Labour in Scotland to go ahead in terms of Westminster seats held. That doesn’t seem very likely in the foreseeable future, IMO.

  31. It’s difficult at this stage to forsee what the general poll prediction of a reasonably narrow No would do to the SNP going forward.

    I don’t think I’m sticking my neck out in saying that they would collapse if there was an emphatic No, and presumably benefit in Scotland’s final Westminster election (but not necessarily the next Scottish Parliament election) in the event of a Yes of any magnitude, but for me it’s too early to talk about Westminster north of the border given that the most likely referendum outcome (based on the polls) is the one with the least predictable political effect.

  32. @CH

    ’emphatic’ might need to be defined a bit. Less than 40%? Less than 35%?

    And discussion on this issue needs to assess also the relative strength of the other parties’ pitches (remember, this is Westminster GE, not Holyrood).

    The LDs are not showing much sign of returning to their 2010 levels; the Tories may win one or two, perhaps three more seats – but at the expense of the LDs, I think, not at the expense of the SNP. IMO. Difficult to see where Labour can do any better than it already is doing.

  33. Something for Scottish colleagues to consider is this. In Belgium, we are almost at the stage where a centre-right (very right actually) government will be formed there at federal level.

    The expression given to this formation is the ‘Swedish’ solution. This has nothing to do with Sweden, other than its flag which has yellow (the Flemish nationalist party colour, the blue of right wing liberal (orange booker) party colour and the cross which they associate with Christian Democrats. The last is only the Flemish CDs, because the Wallonian French CD party (which is a a bit leftier) had made a deal with the Wallonian socialists for their region’s government.

    This goes against what was a ‘principle’ in Belgium, that at national level one should have coalition parties, but compulsorily both wings from the two regions . Still with me?

    It’s the overwhelming performance of the Flemish Dutch speaking nationalists that has meant that the previous idea has had to be abandoned. There is no Walloon nationalist party, to speak of, except some who want to join France.

    Now to UK. If Labour gets a small majority in 2015, it could govern with LD and perhaps add SNP, especially if a deal was offered for Scottish independence or devo max. This would then parallel the Belgian situation, because you would have possibly an SNP UK minister or SoS, guiding legislation applying to England and Wales.

    I can’t think of a nation’s flag to deal with that one, oh yes I can -er, Belgium?

  34. What is one to make of people like Andrew Cooper – just made a Tory peer? In 1979 he campaigned for Labour – joined the SDP in 1981 – stayed with the Owenite rump following the Liberal/SDP merger before joining the Tories shortly before the 1992 election. Drifting ever rightwards when will he join UKIP I ask myself. I am not sure that people so uncertain as to their political leanings can ever inspire much confidence whether the shift is left to right or vice versa.

  35. Or the ‘German’ coalition too?

  36. Graham
    Sounds like he is a fervent supporter of the Andrew Cooper Party?.

  37. So what can we make of these results

    Epping Forest District – Broadley Common, Epping Upland and Nazeing:
    C 155, Ukip 122, Green 23, Lib Dem 7. (May 2011 – C 585, Green 69, Lib Dem 43).
    C hold. Swing 15.5 per cent C to Green.

    Huntingdonshire District – Warboys and Bury:
    C 619 ,Ukip 560, Lib Dem 78, Lab 72. (May 2014 – C 864, Ukip 576, Lab 172, Lib Dem 138).
    C hold. Swing 6 per cent C to Ukip.

    Malvern Hills District – Wells:
    C 317, Lib Dem 227, Ukip 158, Ind 76, Lab 71. (May 2011 – Two seats C 822, 784, Green 491, 386).
    C hold.

    Stroud District – Valley:
    Green 291, Lab 230, Ukip 76, C 67, Trade Unionists and Socialists Against Cuts 16. (May 2011 – Green 417, Lab 279,C 190).
    Green hold. Swing 3.3 per cent Green to Lab.

    Worthing Borough – Castle:
    Ukip 568, C485, Lib Dem 242, Lab 197, Green 49. (May 2014 – C 702, Ukip 653, Lib Dem 414, Lab 324, Green 88).
    Ukip gain from Lib Dem. Swing 5.1 per cent Lib Dem to Ukip.

  38. AW, Spearmint, Postage, Roger M.
    Thanks for answering my questions on Lord A’s polling.
    Seems pukka -ish, but… is at variance with other such marginal polling.
    Still,his next set of marginal polls will be something to look forward to.

  39. @Graham

    The political equivalent of El Cid.

  40. No one quoting Populus – Con 36, Lab 35, UKIP 11, LD 9

  41. “No one quoting Populus – Con 36, Lab 35, UKIP 11, LD 9 ”

    There you go.

    I think most posters here are more interested in trends than one-off polls – especially when they are out of step.

    The idea that one is scared to acknowledge polls of differing stories is rather fanciful in my view.

    As Phil O’Soffical once said:

    “Things is what they is, so they are, so they are.”

  42. I haven’t been commenting on it because I’ve been busy trying to configure a GameCube and Game Boy Advance emulator to play nice with each other so I can stream some gameplay with commentary tomorrow night.

    I’d like to think the entire reason we’re here, my Latin counterpart, is that we want to discuss polls in a relatively spin-free environment. What good would it do those of any persuasion not to acknowledge when something bad happens in the hope it will go away?

    Fact is, as Paul said, it’s just not that far outside the norm, especially with Populus. If it continues as a trend, I’ll offer a view. But it doesn’t do anyone any good to be throwing around accusations of bias, and I’m sure our fellow users on the right would agree.

  43. @ R&D – lol yes of course, never thought of putting it like that !

  44. Mr Nomine

    That’s ‘cos yer not a qualified Flossifer, like wot my mate is.

  45. Populus: I joined the panel and they pre-select (if you are the wrong demographic they don’t include you) and you need to respond to the email within an hour. Also, the questions can be interminably long, with politics mixed in with which kind of coffee dispenser you prefer (I once gave up after 50mins). This might be the reason for the Friday/ Monday pattern: the different people available to put aside an hour, instantly, in the morning during working hours or on a weekend. Oh, and unlike yougov you cannot do the poll on a mobile phone so you have to be near a computer.

  46. For all the talk of a 36.9% GB ceiling for the Tories, they’re not far off that in the latest poll, and with some time to go until the election.

    (My prediction of a Labour win hasn’t changed, I just like pointing the unreliability of generalisations based on things like “elections since 1945”.)

  47. Chatterglass

    I am amazed. What motivates you to join?

    You could no more get me to join an online pollster site (presumably in hope of influencing the result) than I could join a site that wanted to investigate if there was such a person as JC of Nazareth or whether Suarez really bites or is somehow propelled by a supernatural force to sink his teeth into people’s arms.


    “…Did you see the 10th Anniversary of “Who do you think you are” the other night. Compulsive viewing if you didn’t. They showed Boris looking totally confused, he was standing in the large hall of a German Royal Palace while an expert showed him one picture at a time of his early ancestors until the last which was William II. They also showed Xander Armstrong speechless as he found his family line went right back to our Norman friend William and the best was Matthew Pincent as he watched an expert unroll a vellum family line which went back to God via Adam & Eve. Priceless!…”

    My favorites were the Kim Cattrall (ancestor was a bigamist) and Alan Cumming (ancestor died during Russian Roulette) episodes. The most harrowing was Lisa Kudrow who is descended from Eastern European Jews, none of whom made it thru the Holocaust.

  49. DAVE
    ” opponents are often not selected by their party until quite near to the election and so have less chance to have their work related to their status as the potential MP.”

    My comment related more to “incumbent” opposition candidates, in the sense of being selected for a substantial period or for successive campaigns. However, I would also draw attention to the relevance which candidate activity has for its realisation of a policy position or ideological approach of the party in areas relevant to the constituency, whether or not specifically for the manifesto. This can bring national policy into the local setting, for example, support for the retention of public or local authority responsibility and funding for veterinary meat inspections, or support for the provision of care for the elderly in a campaign for sheltered housing, or for maintaining local authority residential care. On the other wing, support for specific enterprise and training projects under EU structural funding, have all been part of campaigning, and pre campaign substantive political activity of opposition candidacy and branch and party action in West Country over recent decades which enter into VI.

  50. RAF
    “No one quoting Populus – Con 36, Lab 35, UKIP 11, LD 9”

    Actually, Richard did at 12.20pm – it’s on page 1.

    There was then some debate about the Populus anomaly that shows increased numbers for Cons on Fridays when compared with Mondays.

1 2 3