We almost have the full results for the European elections, so how well did the pollsters do? Here are all the polls taken over the final weekend, plus that final YouGov poll on the eve of polling.

YouGov/Telegraph 03/06/09 26 16 15 18 10 5
ComRes/Green 31/05/09 24 22 14 17 15 2
YouGov/Telegraph 29/05/09 27 17 15 16 9 7
ICM/Sunday Telegraph 28/05/09 29 17 20 10 11 5
Populus/Times 28/05/09 30 16 12 19 10 5
RESULT 27.7 15.7 13.7 16.5 8.6 6.2

Scotland is still to declare, but on the figures so far it looks as if YouGov and then Populus will have the laurels. ICM ended up severely underestimating UKIP support and overestimating Lib Dem support, while ComRes were out on Labour, the Conservatives and the Greens. I’ll update with a proper post, and what lessons we can learn, once the Scottish results have been declared.

UPDATE: I’ve now updated the table to include the Scottish figures so we can look at the final performance of ICM, Populus, YouGov and ComRes (MORI did not carry out any polling for the European election). Looking at the average errors of each company (that is, the average of the difference between each party’s actual share of the vote, and what each pollster had them in their final poll) YouGov performed the best – both their polls in the final week were closer than any competitor, with average errors of 0.83 and 1.23. Second were Populus, who slighly overestimated Conservative and UKIP support, but otherwise performed well with an average error of 1.57

ICM’s average error was 3.16 – this was down to them overestimating Lib Dem support and underestimating UKIP support. As regular readers will know, ICM do tend to produce higher levels of Lib Dem support than other pollsters do, however, in this case I don’t think the error in predicting the Lib Dem vote is part of a wider problem with ICM’s method – rather, I think it’s down to the way ICM and Populus asked the question of how people would vote. ICM prompted with just the main three party names, and then gave supporters of “other” parties a second list of prompts. Populus included minor parties in their main prompt. The results suggest Populus’s approach works better with phone polls (though YouGov’s increased accuracy compared with their 2004 performance suggests it’s the other way round online!)

ComRes’s final poll was furthest out, with an average error of 3.56. Their final poll before the election underestimated Conservative support, and severely overestimated support for Labour and the Green party (for whom the poll was carried out). My guess is that the skew towards Labour away from the Conservatives was due to the lack of any political weighting – exactly what went on with Green support I don’t know, it may well be a prompting issue – but at the moment I don’t know exactly how ComRes worded their question.

81 Responses to “Euro election: How did the pollsters do?”

1 2
  1. I imagine that this election was a particularly difficult one to call. The political situation and, presumably, people’s responses to it have been so volatile for weeks. I suspect that a higher than usual number of voters weren’t all that committed to a particularly party prior to voting and were doing a bit of a “I’ll vote Conservative! No, wait a minute – UKIP! Oh, hang on, NIck Clegg sounds okay …, nah, I’ll vote BNP and stick two fingers up at the lot of them”.

  2. Although most of the opinion polls seem to have come close to the actual UKIP vote, it was interesting to see the editorial comment that accompanied these polls – most seemed to think that UKIP would do rather worse in the actual voting and lose ground to the Greens and the BNP. This clearly didn’t happen.

    It does rather beg the question of what may happen in a general election, if UKIP field 550 candidates as promised. Even if their share of the vote drops by 10% under FPTP, they could still deprive the Tories of success in tight marginals by splitting the Eurosceptic vote.

    There was a useful antidote on the BBC last night to some of the the triumphalist talk on this site about the inevitability of a large Tory majority at the next election. It was pointed out that only twice since the war had any party achieved the swing needed by David Cameron (6.7%) to achieve even a majority of 1. And neither of these were achieved by the Tories (Attlee in 1945 and and Blair in 1997). While it is by no means impossible for that feat to be repeated next time, especially with the current state of the Labour leadership, I think that there still remains plenty of uncertainty about the outcome of the election.

  3. Further to my previous comment, I would accept that calculations based on a uniform swing are probably unrealistic and the Tories could gain a majority with a smaller overall swing.

  4. seems like ICM/sunday telegraph was closest to the cons. but id say the yougov/telegraph poll was probably closest for the mostpart.

    Odd results but about what the polls indicated:/

  5. @ Leslie – “It was pointed out that only twice since the war had any party achieved the swing needed by David Cameron (6.7%) ”

    True but then how many times since the war has an incumbent government registered so many “historic lows” as the current Labour government?

  6. @ Leslie

    Even with your correction, you’re kidding yourself. The BBC also pointed out that there seemed to be a telling mood of “anyone but labour”.

    You know what this means. You’ve seen it before. It means that in all key marginals voters will concentrate on voting for the party most likely to oust labour.

    They are toast in a way that no party in this country has ever been toast before at a single election. Time to start thinking again about that election in Canada in which the Governing C’s almost lost every single seat. If Gordon Brown is still in charge and if the election is delayed until next year, that’s where labour is heading.

  7. Does anyone know when and where (hopefully all in one place!) we can/will be able to access the by-Council area voting data?

  8. Back in Caledonia and the poltical weather is great.

    Bob Worcester of MORI on Scottish radio made the point this morning that the difference between the Scottish SNP Government which is rising and the UK Labour Government which is sinking is that the first has leadership and the second does not.

  9. Do you mean the European election shares in each UK parliament seat?

    I’ve been looking for similar but no luck so far I’m afraid. Too much work in terms of number crunching I suspect and both the Beeb and Sky have been more than a little lazy/shoddy in their coverage this time in opinion!

  10. Looking at that, it’s obvious that certain parties were fairly consistently over-estimated: Labour, LibDems, Green, and Conservative, UKIP and BNP mostly under-estimated. Looks like the shy tory factor now extends to all rightish parties.

  11. Anthony / Others

    Did I read somewhere that we are getting a Populus Westminster Poll tonight ?

  12. David I don’t know if they’re all in the same place. All of Yorkshires are on the Leeds City Council website:

    All the North Easts are on Sunderland city councils website:

    Just pointing out these two as in both Calderdale and Kirklees the Conservatives won and there is a few marginals in these council areas. Also in Newcastle the Lib Dems won, which doesn’t look good for the labour seats in Newcastle!

  13. YouGov and Populous do indeed win laurels, but all the pollsters seem to have slightly under-estimated the BNP and over-estimated the Greens. In the North West and Yorkshire that made a big difference. It would be useful to have more sophisticated, region-by-region analysis next time.

    I’d be interested to hear your take on one thing that YouGov DID get wrong, which was turnout. Last week they were calling a notional turnout of around 43-45% when the final result was 10 points below that.

  14. Ivan, the rersults last night were coming through by local authority area, not by Parliamentary seat. Some councils e.g. LB Richmond, LB Barking (and no doubt others) are putting them up, but it’s a schlep to go round and check ’em out individually.

  15. @Brett

    Your assertions are just that – assertions, and wishful thinking at that. There is no real evidence that the Labour vote has gone for good and not just being lent to fringe parties as a protest. The likelihood under FPTP of a wipe-out of a major party is tiny. To secure a 2-term majority, Cameron is going to have to get people to support the Conservatives, not just dislike the Labour party. That still remains a tall order.

    Given a number of ifs: if Labour replace Brown this summer, if the economy shows some signs of bottoming out, if the housing market starts showing movement then the most likely outcome given our electoral system and the advantage that Labour has through boundaries, is a hung Parliament. Since it is unlikely that all three will occur, I’d put money on a Tory majority of less than 30. However, with a fragmented Commons that may well prove sufficient for two terms. But a Labour wipe-out? I very much doubt it on the evidence so far.

  16. I am pretty pleased with my predictions over on the Regional pages. In 8/11 regions I called every seat correctly – in the other 3 regions I was only 1 seat out in each case.

    In the West Midlands I called the last seat for Conservative – it went to UKIP. In Yorkshire I called the last seat for Labour, it went to BNP. In Wales I called the last seat for Labour, it went to UKIP.

  17. @ Keith – quite right about the shy tory factor. But interesting how the “shy tories” manifest themselves. If you look at the populus poll, you might think that voters intending to vote BNP/UKIP said that they would vote Tory.

    Does this mean that:

    1) We need to downgrade the tory poll figures to account for closet BNP votes?
    2) Cameron has sufficiently purged the Tory brand, people are comfortable about saying they’ll vote Conservative?

    Populus wins this one I think!

  18. The election results seem to have proved Mike Smithson’s “Golden Polling Rule” right: the most accurate polls were the ones showing Labour in the worst position.


  19. P.S. – That is 95.7% accuracy for my predictions – not too bad.

  20. YouGov on 29th May was by far the most accurate. Next most accurate was YouGov on 3rd June; then Populus. Quite some distance out were ICM, and ComRes come last with a prediction that bore little resemblence to the reality.

    The clear winner was YouGov. I didn’t go to the trouble of calculating standard deviation – I just added up how many points out for each party the pollsters were, disregarding whether they over-estimated or under-estimated. The results were:

    YG 29/05/09 – 5 points out
    YG 03/06/09 – 7.4
    Pop 28/05/09 – 9.4
    ICM 28/05/09 – 19
    CR 31/05/09 – 21.4

    I have always thought YouGov and MORI were the two most accurate pollsters. This seems to confirm that YouGov are a cut above.

  21. Congratulations to those who got their predictions right. I don’t know if anyone predicted that Labour’s share would be as low as 15.7%. I predicted 17%.

  22. Looking at the results for North Tyneside (thanks Robert for posting the links), clear evidence of split ticket voting: Labour out-poll Conservative in the Euro poll on the same day that the Conservative candidate is elected as mayor.

  23. Polls plz

  24. Forget the headlines about Labour, the Tories, UKIP or the extremists. The most interesting trend to come out of this weekend was none of them.

    Check the English Democrats results. In time, I see these becoming the protest party.

    Hopefully they’ll help kill off the extremists.

  25. @ Wayne

    I hope so, but i wold not read too much into this poll, still too close to the EU elections, give it a month to allow the parties to recover their natural support which can be lost in the euro-election (con – ukip, LD – green, Lab – BNP, ect) by then we should start seeing patterns.

  26. Does anyone know the national swing from Labour to the Conservaitves on Thursday, from the 2005 county figures to the 2009 County figures?

    I’m struggling to find it!!

  27. @ Sunbeam

    Good luck to the English Democrats, I hope they can replace the BNP as a party of protest.

    @ Leslie

    I agree that Labour won’t be wiped out; their support is far too localised to allow it.

    But that doesn’t mean that they can’t “go Liberal” partially as the Liberal Party did in the inter-war years; if Labour were to plumb the low 20s in a GE, would UNS still be as accurate?

    It would need the LibDems to begin to replace Labour as the party against the Tories for a long-term effect though, be this through defections or splitting or “Michael Footism”. It all depends on how the GE turns out (and it doesn’t look too pretty for the LibDems at the moment, in my opinion).

    I think it’s all too riven with “maybes” and chaos to put any firm guess on the future of Labour in the next 2 years.

  28. My prediction was Cons 27, Lib Dem 17, UKip 16, Labour 15 so three out of four of my predictions were right to within 1%.

    Regarding the Lib Dems I may have let my heart rule my head. In Wales where I am from I did not even get a leaflet through the door from the Lib Dems. Maybe they are saving their money for GE.

    Overall with the 28% Council result the Lib Dems have done quite well. But still not the big break through I, I mean they, were hoping for.

  29. Castle Point constituency is co-terminus with the borough. Breakdown here: http://warelane.wordpress.com/2009/06/07/castle-point-euro-election-percentages/

  30. The polls being snap-snots were pretty accurate.
    Given factors such as having two polls on the same day and probelms estimating liklihood to vote, they were well within the margin of error.

    I suspect the actual ballot paper acted as a final prompt in favour of the minor parties.

  31. Beware party-funded polls is one lesson I guess. Comres would seem to need to be careful to avoid damaging their reputation by (co-incidentally?) overstating their client.

    Is there any data on whether polls funded by political parties are typically less accurate and biased towards that party?

  32. I wish they’d publish the detailed results for the SE and SW. They don’t seem to be on the relevant websites yet. If anyone knows if they’re published elsewhere I’d be grateful.

  33. They should have published all the council areas on the results program last night, in flashes at the bottom of the screen.

    It could have been done like
    North Lincolnshire Con majority over X X%

    then they could have cut to Vine for more detailed results against 2004 of interest (which would have had less waffle if the flashes by all council areas had been provided).

    They need to look at the Decision79 program to learn again how to deliver masses of information quickly and efficiently

  34. I’m talking about the Euro election program – declared by council area.

  35. Although most of the opinion polls seem to have come close to the actual UKIP vote, it was interesting to see the editorial comment that accompanied these polls – most seemed to think that UKIP would do rather worse in the actual voting and lose ground to the Greens and the BNP. This clearly didn’t happen.



    But note that UKIP’s share of the vote rose by only 0.3% of the vote.

    It’s hard to see why sites such as the BBC are calling it a huge success for them.

    UKIP will also find themselves without a group in the European Parliament as other Eurosceptic parties were soundly beaten.

  36. Given we all gave caveats about the size of the Scottish sub samples all the time, the SNP has done very well in the elections.

    It would be interesting to see the Scottish subsample %s for each pollster against the final results and also to see which pollster was more accurate…

  37. @ David in France

    Not Euroskeptics as such, not in the terms the Conservatatives are Euro-skeptc, more Anti-EU vote has dimished.
    UKIP owe most of their gains to Labour drops which may be picked up by Lib Dems, Greens and a small number English democrats and(unfortunatly) the BNP, this will not help them in 5 years time, where they will come next EU election wil depend more on if the Conservatives deliver the referendum that DC has promised.

  38. @David: It’s partly because they beat Labour, which is psychological, but it also has to do with them doing so without Robert Kilroy-Silk (who got a lot of the credit for their good showing last time around).

    In general:
    A slight BNP underestimation is hardly a shock by pollsters…people tend to be a little shy about admitting BNP support, after all.

    As to my prediction, I gave the following:
    Con: 28%
    UKIP: 17% (and edging out Labour)
    Lab: 17%
    Lib: 16%
    Green: 10%
    BNP: 7%
    SNP/PC: 4%

    I got the order right on; my only real error was in overpredicting Labour and LibDem support (by 1% and 2%). Everything else was within 1%.

  39. @David: It’s partly because they beat Labour, which is psychological, but it also has to do with them doing so without Robert Kilroy-Silk (who got a lot of the credit for their good showing last time around).

    In general:
    A slight BNP underestimation is hardly a shock by pollsters…people tend to be a little shy about admitting BNP support, after all.

    As to my prediction, I gave the following:
    Con: 28%
    UKIP: 17% (and edging out Labour)
    Lab: 17%
    Lib: 16%
    Green: 10%
    BNP: 7%
    SNP/PC: 4%

    I got the order right on; my only real error was in overpredicting Labour, LibDem, and Green support (by 1%, 2%, and 1%).

  40. In the 1999 Euro election they did flash all the results along the bottom of the screen. All the results were being counted by parliamentary constituency on that occasion. I remember seeing “Conservatives gain Don Valley” flashing up which was unexpected.

    Jeremy Vine’s virtual touchscreen was a bit difficult to follow because you had to remember what the 2004 results were when he changed it to 2009. It would have been better to have both on the screen at the same time.

  41. Leslie

    “they could still deprive the Tories of success in tight marginals by splitting the Eurosceptic vote.”

    This is one of the great reasons to have STV brought in. Imagine a case where UKIP and Tories both take 30% on a Eurosceptic vote and Labour take 35% on a Eurofanatic vote. The Eurofanatics take the seat even though the sceptic vote was 60%.

    STV would solve this problem and might even allow some minor parties a better chance at gaining seats because people won’t feel compelled to, for instance, vote Tory to keep Labour out. They can happily vote for their preferred candidate, knowing that their secondary preferences can keep out a party they don’t want.

  42. Vine just is a waffler.
    Excellent graphics on the Decision 79 program – all computer generated, and with maps and full percentages.

  43. @Mark |M

    You’re preaching to the converted. While no voting system is perfect, and all can in some circumstances lead to illogical outcomes (as Kenneth Arrow showed), STV in multi-member constituencies is the least bad system, and this is a reason why it is widely used in non-political situations. Whilst not strictly PR, it does eliminate the dreaded list system which in my mind is an abuse of PR.

    Opponents bang on about electoral reform letting in extremist parties – well frankly I’d rather have them where I can see them (and more to the point, where people can see what a waste of space they are once elected). A system that permanently ignores a particular strongly-held strand of opinion is storing up trouble for itself.

  44. Well you’re not preaching to the converted. STV is a disgrace because it re-allocates losing votes second choices (or even 3rd onwards) often to knock out who won first.
    I believe that is fundamentally undemocratic and I won’t change my view on it.
    We should never introduce it in this country – and should repeal it where it is used.
    The least bad form of PR is the additional member system – but it should be drawn simply from the one first past the post vote.

  45. But I would oppose PR anyway.

  46. I think that this election has probably torpedoed Labour’s flirtation (Re Alan Johnson) with the idea of introducing PR. It will make them vulnerable to the Greens and BNP in their heartlands. Moreover, I think that the Lib Dems desire to gain power at whatever cost, and thereby proposing PR is delusional. They could become the victims of any change in electoral reform at Westminster. Instead of bleating on about electoral reform; they should be taking the battle to the Labour Party in their territory and cleaning up the protest vote.

  47. I don’t know if anyone else saw it, but there was an interview last night with Chris Watson who boasted that the ALDE group held the balance of power, and thus made or broke key decisions in the European Parliament.

    This, if it isn’t exaggeration, rather exemplifies an issue with PR: the 3rd-biggest group is (or claims to be) in control of decisions in the Parliament. It seems undemocratic that, almost regardless of the votes of the top 3 groups, the 3rd party will reign over key decisions disproportionately to their vote share.

    No-one ever said FPTP was perfect, but listening to the ALDE leader boasting of disproportionate power in the set-up that he’d like to see in Westminster is, in my view, revealing.

  48. The BNP failed to win 10% in any region; their best was 9.8% in Yorkshire&Humberside. One idea might be to use the d’Hondt system but also have an additional threshold of 10% so that you cannot be elected unless you win at least 10%. Any seats which would have gone to a party in that situation would go to the first placed party.

  49. PR is fine if you do it much like Japan does (where about 2/3 of the legislature is FPTP, and 1/3 is PR…but it’s pointedly -not- a top-up system, it’s separate elections). If you work PR like that, you’ll usually get either a majority or a “strong minority”; multi-party coalitions and deadlocked parliaments will tend to be rare (as they are in the UK).

    Do consider that on the result from the Euro elections, if translated into a German-style PR with a 5% threshold for the 578 England/Wales seats (and allowing for PC to clear in Wales) would give you:
    CON 181
    UKIP 110
    LAB 97
    LD 88
    GRN 55
    BNP 41
    PC 6

    A Con-UKIP coalition would have a slight edge (291-287) in England and Wales; but for the strong UKIP showing you’d likely be facing an unworkable situation due to the BNP interloping…and that’s just on England and Wales.

  50. Jeremy Vine’s graphics were dismal. Whoever thought that pie charts were the best way of communicating the changes in support for several parties needs to go back to Information Graphics 101.

1 2