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. I think a hefty swing to Labour could have a double-effect, from those who might have voted UKIP were they not in a winnable marginal.

  2. AW – ” 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. ”

    There is one good reason, namely that the expenses scandal had already eliminated the incumbency bonus of many sitting MPs in 2010, especially those which had been badly damaged personally by the scandal. That won’t eliminate a positive incumbency effect for the new MP, but there will be no further bonus where the party of the former MP loses in 2015 their previous incumbency bonus. So a single incumbency effect in contrast to previous elections where there was a double one?

  3. @AW

    You rightly concentrate on English (and Welsh) seats because these are the Lab-Con variables. But what about Scotland? If there were to be a tight race (not that that is being indicated by the UNS projection at present) is anyone doing any similar work to understand what Scottish marginals are likely to do? Or is that all on hold until after September?

    As regards ‘marginals’ polling in general, is there any indication that people who know they live in very marginal seats are more aware of the influence they may have through polling results than those whose seats are unlikely to change hands?

  4. Phil –

    Well, the high retirement rate last time will have (incumbency had already gone in 2010, so can’t go again), but those seats where Labour MPs did stand and fight they still did better. I think people had half expected the expenses scandal to erode the incumbency benefit (or even reverse it) but in the event Labour still did better in seats with incumbents than those without.

  5. John B – nope, there isn’t.

    I think it’s largely worthless until after the referendum anyway, though I’m not certain it will happen after then either :(

    Perhaps Ashcroft will do some specific Scottish marginals stuff after the referendum has died down.

  6. Sorry – I’m getting tired. But the my question really regards the potential bluff and double bluff nature of some polling results. If I live in a very marginal seat and say I’m going to vote one way but in reality I’m going to vote a different way, that gives a false sense of security to my opponent.

    I have a friend in the States who, although 100% committed to the Democrat cause, is a registered Republican. Her view is that she can thereby push the local Republican party to vote for extreme right wingers and thereby make the Democrat much more electable.

    People can be like that, you know!

  7. @AW are those ComRes swings really zero? The first one, I thought, was 1 point CON to LAB IIRC and the second was a bit more…

  8. @AW

    Thanks for your reply (10.20). Perhaps the Scottish situation is now so different from that in England (and Wales?) that the London based organisations find it difficult to work out how to approach the task of interpreting the polling results here.

    Do the Scottish organisation have the resources to do the job? And even if they do, how do results from Scottish polling feed into the overall UK picture unless there is some agreement on how to interpret them?

  9. I agree, the weighting in the Survation marginals poll was very lightweight especially since age and gender by themselves predict relatively little of the variation in VI.

    The effect of this can be seen in the LD VI, which Survation report as falling from 21.2% in these seats to 4.4%, an astonishing implosion even for the LDs. But within their actual sample, after reweighting, those who say that they voted LD constitute only 13.0% of the sample of all who say they voted in 2010. It’s that 13.0% that’s become 4.4%.

    What’s more likely? Should we discount 2010 VI because of false recall and accept that the fall in LD VI was far further off the scale than anything we’ve seen before, losing four-fifths of their 2010 share? Or should we conclude that Survation got their weightings wrong, and that in reality their poll shows the LDs only(!) losing two-thirds of their 2010 share? Given how superficial the Survation weighting scheme is, I think the balance of probability has to be with the latter.

  10. Number Cruncher – figures up there are only for the Con-Lab seats in the sample, so excluding the Lab-Con seats which did have a swing.

  11. Is there no Yougov tonight?

  12. @ AW: Thanks

    @ Sine: Yes
    CON 33%
    LAB 38%
    LD 7%
    UKIP 12%

  13. Good Evening All.

    Thank you, Anthony for your analysis of the Marginals.

    I am surprised that Ed M is holding the lead, and his nerve, at this stage.

  14. @AW
    I referred to the elimination of an incumbency bonus for many, not all, sitting MPs. I wasn’t implying that there wouldn’t still be an incumbency bonus for some MPs in 2010, just that the average effect would be smaller. (And yes, there was an unusually large number of sitting Labour MPs retiring in 2010, which also didn’t help Labour.) I’ve certainly seen reference to this in I think the British Election Study as a possible reason why the Conservatives achieved a slightly higher than average swing against Labour in marginal seats in 2010.

  15. now I don’t know if its just me or not but common sense dictates as regards marginal con v lab seats and as I keep pointing out the libdem 2010 vote as mainly gone straight back to labour especially in the con v lab marginal and I can tell any pollster in my neck of the woods amber valley Sherwood Broxtowe and erewash are a certainty to go back to labour so that’s 4 cons down already so how the hell can they form a majority government and the libdems are going to be slaughtered at the GE gods sake look at actual libdem election results since 2010

  16. @ Anthony,

    I hate to say it after you’ve made such a pretty chart, but I think something is off about the July 14 Ashcroft poll. Either you have the national and marginal swings transposed, or the difference should be + 0.5 Lab and the row should be faintly pink.

  17. @John B
    In fairness to AW, the absence of reference to Scottish marginals is I think because there has been almost no recent polling specifically in them. Apart from Matthew Oakeshott’s polling for Inverness, I think you have to go right back to Ashcroft’s original March 2013 polling to find some:

  18. @John B

    The selection of those seats can be found on Page 16 here:

  19. nottsdave

    I agree about Con seat losses to Lab being probable and also the high improbability of a Con majority.

    Then again Lab gains from Con may be ofset to some degree by Con gains from LB and, perhaps, some SNP gains from Lab. So while Lab may do well in the Con marginals, that doesn’t automatically mean Lab will be in as strong an overal possition as it may at first seem.

    Good Evening to you; and welcome here; some UKPR men have suggested that LDs are heading for trouble in May.

  21. The most relevant number for me is the 2001 GE when the double incumbency effect was last seen in a significant number of seats.

    The difference was around 2% of swing in marginals and I would not be surprised if he cons had a similar lift again.

    Assuming approx the same gains from LDs for Lab and Con ( reckon less than 5 in one of their favours) Labour need just under a 2% swing to emerge as the largest party and it would be surprising if they did not achieve this.

    Factoring in 2% less swing in Con-Lab marginals means a UNS of around 4% needed which may be beyond them.

    Not mentioned above is that the cons threw the proverbially Kitchen sink at many of these seats in 2010 and Lord A himself says this can not be repeated so the incumbency bonus maybe less.

    Of course, the LD vote share in these seats is crucial as Lab are on course to gain 20% or more net votes per constituency.
    Also double incumbency will be tempered in some seats by the 2010 Tory victor standing down or the defeated former Lab MP re-standing (presumably selected as the local party reckon popular).

    For both these reasons, inter alia, as we know some high targets will hold and lower ones fall

    My non-scientific hunch is that a 4% UNS will not be necessary for Labour to take 25+ Tory seats but that it will need to be more than 2%.

    Psephologically, it will be interesting if there is a 3% swing.

  22. Anthony

    When you compare the marginal swings with national swings are you doing it like for like (ComRes wit ComRes, Ashcroft with Ashcroft and yes I know Survation is a problem) or using your own rolling average?

    I think it’s worth pointing out that the three surveys have different methods. ComRes is online, Survation landline (though they keep on using that phrase about ‘targeted lifestyle data’), Ashcroft half landline, half mobile.

  23. “The most relevant number for me is the 2001 GE when the double incumbency effect was last seen in a significant number of seats”

    what few acknowledge in all this “double incumbency” chat is the exceptional nature of the 97-01 parliament.

    Labour secured a 179 maj. in 1997. The effect of this was that they only ever needed about 250 MPs at any one time to get legislation through the house….their most marginal MPs spent the whole of that parliament in their constituencies…the effect of all this was seen in 2001 where, even though the labour vote share fell by abt. 3%, most of those marginals were held.

    This phenomenon gave rise to all this double incumbency effect chat, even though the exceptional circumstances which gave rise to the effect had been forgotten.

    Tory marginal MPs have been spending their time this parliament in the house trying to get the coalition’s business through the house.

  24. Candy (fpt)

    Most people would consider Blair to be upper-middle class with a slight classless veneer (but then members of the British upper middle class are the ones most likely to proclaim themselves classless). I was unaware of his father’s working class upbringing till now[1], so I doubt if it has that wide an impression, especially outside Scotland.

    Technically we should probably refer the James VI[2] because Churchill passed a law so Anglo-Scottish clashes of regnal numbers should always go by the higher one. Which has at least saved us going round for the last 60+ years referring to Elizabeth II and I.

    Anyway Cameron is not only descended from Jimmy but (illegitimately) from William IV. Whereas Boris is only descended (illegitimately) from George II, three generations further back. This means that DAVE WINS.

    Technically Margaret Beaufort, like all the descendants of Katherine Swynford was barred from the throne by an Act of Parliament. But then technically all the descendants of Margaret Tudor were also barred from the throne by Henry VIII’s Will. Which would have meant that there would have been no union of the Crowns and and you would have ended up with Queen Anne I[3]. Though as a consolation it would have meant union with the Isle of Man. And you would have had some awesome sex scandals.

    [1] Bizarrely his (biological) parents, like Major’s were circus performers. You may feel this tells you everything you need to know about British politics.

    [2] That TV series on the Stuarts is quite good by the way, especially by the usual BBC standards of “We must fit this topic into 3 x 1hr episodes, no matter how long or short the period”.

    [3] From Mary Tudor to Eleanor Brandon to Lady Margaret Clifford to Ferdinando Stanley (great name), her father.

  25. Spearmint – got them the wrong way round. Better change that!

  26. Incumbency: nothing magical in it. An MP who has been working the seat since the previous election, with a paid for constituency office and staff has a huge advantage against a candidate, probably only selected a year or two before an election and reliant on donations and volunteers.

  27. @Roger Mexico

    I think most people had instinctively pegged Blair as “aspirational class” rather than upper middle class or toff.

    And that’s to do with his peculiar accent. For example he’s the only person I’ve heard to pronounce both t’s in “rotten” – which indicates some effort a while previously to change his natural accent, no? Though he sometimes slipped effortlessly into estuary, and in his late years into hip-hop argot (he kept saying “diss” etc and I half expected him to go full-on in and say things like “Yo, don’t diss ma man Milburn”).

    His parents weren’t circus performers. His biological grandparents on his father’s side were music hall actors. They were unmarried, so gave the baby (Blair’s father) to the Blairs who lived in Glasgow tenements. And then the Blairs refused to give the baby back.

    Blair’s dad left school at 16, was one of the unemployed desperados of the 1930’s. And then had his world view completely changed by WW2. Got an education in night school, started practicing law. scraped to send Blair to public school, and then when he had his stroke and couldn’t pay any more, Blair’s uncle on his mother’s side paid the fees. Classic aspirational stuff.

    They didn’t make much of his back story in the lead up to the 1997 election (not the way they would now for instance), but ours is such a class based society, that most people can place you just by listening to how you talk and what you say (though you do get exceptions who are quite tone deaf about these things). He pulled in the aspirational voters because of that sixth sense in a public that probably wasn’t paying much attention to policies but had grasped the essence of who he was.

    He had and has a very different vibe from Cameron and Boris.

  28. Chatterclass

    And your evidence is…

    Do you seriously believe that the current instinct of non/aligned voters is to reelect their current MPs ?

  29. @ Welsh Borderer

    there is a large academic literature on incumbency effects – a search on Google scholar ( will find you vast numbers of sources.

    In academic circles I think incumbency affects are very widely accepted as a useful explanation of electoral outcomes.

    “Do you seriously believe that the current instinct of non/aligned voters is to reelect their current MPs ?”

    You only need a small proportion of voters to change their vote according to incumbancy. Indeed, they don’t have to consider themselves non-aligned – in constituencies like Bermondsey many years of hard work will mean voters always (often literally) having voted for a party that they would never have voted for if that party didn’t have an MP in the constituency.

    I think you can get a first-time incumbency bonus in a constituency that changed hands at the last election among voters who no longer consider themselves non-aligned. They are now firm supporters of the MP they have had for the last 4 years, having been impressed, even though they didn’t vote for them in the first place.

  30. This is an interesting article on the G spot: “Ukip has not gone away – don’t be fooled by the lull” as it discusses the potential impact of UKIP in marginal etc.
    I’d put the link but anything to do with G spot enters automod.

  31. Attending school fetes, sorting our a constituents problem (even if it is probably the MPs staff), being involved with a popular local campaign and just been visible etc.

    Intuitively, it makes sense that their will be an incumbency bonus.

    James may well be right that 2001 exaggerated the effect which kind of supports my hunch (no facts) that Labour will require an extra 1% swing or so to compensate.

  32. Mike N: could you post the link please, so that when AW approves the link (hopefully) we can see it if we come back here?

  33. @Welsh Borderer
    The advantage Chatterclass speaks of is not that the sitting MP has a greater chance of persuading waverers to vote for him/her. It is in persuading definite supporters to commit to actually voting.
    Evidence? Just look at the figures. Changes in VI from one party to the others are (usually) a few % of the electorate (pace UKIP, perhaps).
    35 to 40 % of the electorate is not voting, while the incumbent has the highest percentage of the 60-odd % who do. The key to winning elections is to get your own supporters to vote. Ask Norman Tebbit.
    Don’t you hear comments like this? “I’m not keen on some of his party’s policies, but he did a good job supporting keeping the local hospital open, and helped my brother sort his tax out.” That goes whichever party the incumbent belongs to. (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.)

  34. 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.

    I agree with the rest of what you said, but I’d quibble a little here. An MP’s opponent(s) don’t have no chance, though it’s true they have many fewer resources and opportunities. But they can use casework and constituency campaigning to build up a profile and a network of helpers, and they may have some ability to do things, particularly if they are are a local councillor. The Lib Dems won many of their seats on this basis, especially where the current MP wasn’t particularly assiduous and had a moribund local party organisation.

  35. Well done Obama-the right decision in all conscience.

  36. @PH

    Thanks for those Ashcroft links. Problem is that a year on they are not much use, I think. Looks like we’ll have to wait for the dust to settle after September 18.

    My guess is that a lot will depend on the line taken by Labour. If they insist on further cuts and if there is no acceptable offering on further devolution then Labour are not safe in several seats, and look unlikely to gain any more up here. On the other hand, incumbency matters, especially in Scotland, where outside Glasgow and Edinburgh all MPs represent recognisable communities in which an MP’s work (or lack of it) is talked about on occasion.

    The real question then follows as to what will happen in the Holyrood elections in 2016. If the Yes vote gets around 45% then SNP will be in good spirits and be returned to power (perhaps with the Greens). Less than 40% and big questions will be asked. IMO!

  37. Poor trade figures for June

    UK’s deficit on trade in goods and services was estimated to have been £2.5 billion in June 2014, compared with £2.4 billion in May 2014. ONS

    Imports and exports both decreased. Oil and aircraft exports were a big reason for the poor figures. However, cars exports were up.

    I can’t see this gap narrowing this year and the visible trade gap will probably be between £105 to £110 billion in the year – same as last year.


    Better news from the construction industry

    construction output for June 2014 increased by 1.2% when compared with May 2014, or -0.03% decline from Q1 to Q2, which the ONS rounds to 0% – just about all the growth in GDP from Q1 to Q2 came from the service sector

    The ONS notes

    ‘housing construction output has grown relatively quickly in recent months,(but) those sub-components of construction that are not consumer facing have performed less strongly. i.e commercial and industrial construction and infrastructure

    I found this interesting though

    ‘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

  38. AW

    I suppose it will be important, if marginal polling becomes more refined and regularly held, that the polled don’t get to find out that that is why they are polled. ‘Ooh, I live in a marginal’. It could alter answers possibly?

    We have previously identified that people are rather ignorant about whether they live in a marginal. It does not stop some people voting tactically (‘XXX can’t win here’ histogram leaflets?). Otherwise it just becomes a conscience vote.

    In short, it seems doubtful that such a bias could appear in polling in marginals, but I suppose the pollster could ask a supplementary about what the voter knows about his constituency history. Come to think of it, that would be a good question to ask everyone, every time, to build up a picture of voter awareness..

  39. Just seen that an old friend of mine Chris Fox, has been made a life peer by Clegg.He was heavily involved in the coalition negotiations then left his job as LD chief executive. Is Clegg preparing for a post coalition new reality.

  40. ?

  41. There is likely to be some incumbency effect for the Tories in 2015 but I would expect it to be diluted by three factors – some Tory MPs standing down – a fair number of defeated Labour MPs seeking to regain their seats – and five years of demographic change which will make the existing constituency boundaries more helpful to Labour next year than in 2010.

    I note that Andrew Cooper – one of the Populus founders – has been made a Tory peer. He started his political journey in the Labour party and campaigned hard for them in East Surrey in 1979.

  42. Dunno

  43. Tories take back the lead.

    Populus [email protected] · 42 mins
    Latest Populus VI: Lab 35 (-2), Con 36 (+1), LD 9 (=), UKIP 11 (-1), Oth 8 (-1). Tables here:

    Yes, one poll, wait for the trend, etc, etc. But it gives the newspapers something to talk about.

  44. fascinating…there’ll be the usual gnashing of teeth on the left and frantic panic. these guys don’t really do sang-froid…the underlying picture isn’t that different.
    labour lead of abt. 3 points…

  45. Interesting ‘poll of polls’ analysis which shows Greens (the hidden party not included in headline figures or graphs) steadily edging up to an average of 6% now.

    Looks like Lid Dem defectors are now headed to Greens rather than Labour.

  46. I know Anthony has dismissed Mike Smithson’s Monday/Friday Populus theory, but I have to say, based on the summer polls it does look like there might be something in it:

    Has anyone tried comparing the Monday/Friday weighings? Populus topline figures are so contingent on their weighing scheme, it seems like that could produce the pattern even if the underlying samples were identical. (I know, I know, Populus are responsible and transparent pollsters who would announce any methodology variation. Still, it’s worth a look, isn’t it?)

  47. smithson, i’m afraid, is the daddy of commentators. he’s right, of course…the bar charts are pretty compelling.

    populus is pretty dodgy, if you ask me.

  48. Congrats to Lord Populus ! Nifty coincidence there.

1 2 3