Lord Ashcroft this morning announced he was going to start publishing weekly voting intention polls in the run up the election. The polls are conducted by telephone so will be by far the most frequent phone polls (the ICM/Guardian, Ipsos MORI and ComRes/Indy series are all monthly efforts).

Topline voting intentions for the first on, conducted over the weekend, are CON 34%, LAB 32%, LDEM 9%, UKIP 15%.

I misread that to start with, and began to write a post about it being a useful addition to the polling world, but showing figures similar to everyone else… but then realised it wasn’t showing a 2 point Labour lead, it was showing a two point Tory lead. The first for a couple of years. As ever, unusual results demand closer scrutiny and currently we don’t know anything about it other than it was done by telephone – we don’t have a track record to compare, or any clue about the methods used. I’ll update later one I’ve some details to rake over…

UPDATE: Methodological details of the poll are as follows – the poll is past vote weighted (accounting for false recall, so the Tories are actually weighted slightly lower than than in 2010, Labour slightly higher), the voting intention question is prompted for the Conservatives, Labour and Lib Dems like most other companies. Results are weighted by likelihood to vote and a proportion of people who say don’t know are re-allocated to the party they voted for in 2010. Fieldwork was 50% landline, 50% mobile (weighted to be 15% mobile only households).

There is nothing here that should produce unduly pro-Conservative figures, in fact it’s broadly the same methods as Populus used to use in their telephone polls before they switched to online polling, and their polls were normally inline with other companies. What to make of the Tory lead then? Well, the poll seems methodologically sound, but it’s subject to the same margin or error as any poll, so treat it with the same caution you would if ICM or YouGov or MORI had popped up with a slim Tory lead. It might be a sign that the Tories have overtaken Labour, or might just be a an outlier, wait and see if it’s repeated it in other polls.


100 Responses to “Ashcroft – CON 34, LAB 32, LDEM 9, UKIP 15”

1 2
  1. Gasp! Oxygen please!!

  2. Well the much mooted crossover has occurred! A complete outlier, within MOE or has something changed out there?

    A very unusual one indeed when you consider the weekend Survation, Opinium and You Gov polls.

  3. @ Anthony

    Do you know which firm is doing these polls?

  4. @ Crossbat11

    We don’t even know yet when this poll’s fieldwork was conducted!

  5. I think that’s an outlier. It certainly is unless & until any other polls show something similar.

  6. It does seem strange when other 3 polls show increase in Labour lead.

  7. If this isn’t an outlier, then the YouGov one is. The 2 polls are too different from each other to be within the MOE. It would seem to be contrary to the current political narrative for this one to be accurate though.

  8. So without a prompt UKIP still score 15%. Might that be a reason for a slightly higher Tory share? That there was no UKIP prompt? Though, as you’ve said, some of the other firms don’t prompt for UKIP either.

  9. Woah Crossover!

  10. Paul D – nope. It’s only Survation who prompt for UKIP, so it certainly can’t be an explanation for a difference between this and YG, ICM, ComRes, MORI, Populus, etc.

  11. I know that Lord Ashcott has been doing lots of marginal constituency polling but anyone know if he has published many national polls? That would us an idea of potential ‘house effect’.

    However, to preempt some of the conspiracy theorists, I believe Lord Ashcott has previously been quite ‘bearish’ on Tory chances based on his constituency polling.

    There’s no evidence I’m aware of of a deliberate bias in his poll findings.

  12. The raw data of this poll doesn’t show crossover – a statistical dead heat ( which I agree still isn’t good for Labour ) somehow ends up being spun as Tory 34% Lab 32% – can someone explain?

  13. I think this is the beginning of the poll shift and we will see more and more of these narrow leads over the coming weeks then as the pressure cranks further on miliband, can we get into the 6 to 7 point majority zone ?

  14. Pigeons…cat amongst.

    What’s more, my comment last week of,

    “I’ll be amazed if Lab get less than 33%” (in the GE 2015)

    is looking a little premature. I just didn’t think Lab could get less than 5% above their 2010 level (34%). Ok, so it’s only one poll, but if SW’s assessment of the poll is accurate, it’s a bad poll for Labour.

    We’ll have to see other Ashcroft polls really. If all other pollsters are giving Lab 35%-38% and Ashcroft is lower, we’ll know.

    @R&D (there’s that sub-35% poll :-p)

  15. But how much of an outlier?

    The poll is only 1,000 as opposed to 2,000 for YouGov, so a margin of error of 4% could mean Labour 2% ahead which isn’t that far out from some of the recent YouGov polls.

    I am inclined to wait till The autumn after the spotlight is off the Euro’s and UKIP and the holidays. We are closing in on the silly season followed by the Referendum and the Party Conferences (the really silly season).

    In a packed year with UKIP getting the limelight then Scotland we really won’t get a clear picture till nearer November, which isn’t a long time for anybody to change tac if their strategy isn’t ‘t working.

    Polls six months out tend to be good indicators of the result but right now the crystal ball is cloudy!

    Peter.

  16. That’s AW’s assessment.

    Oh, and note that Con + Lab are at 66% of VI.

    Con + UKIP = 49%
    Lab + Lib = 41%

    (Shift to the right?)

  17. Thanks for the link Anthony. There’s lots of interesting stuff in there. I’m wondering whether you can guess which firm did it from the method &/or the style of tables. I hope you’ll do us an ATL or update once you’ve had more time to cogitate. :-)

  18. Amber – updated the original post already.

    Ashcroft don’t confirm who does the fieldwork, but my understanding remains that it is Populus given the tables and the methodological approach.

  19. Actually, there’s a part which says Con 33%, Lab 32%, Lib 8%, UKIP 16%. Strategical rounding?

    Regardless, at that level of the poll, we’re talking less than 500 people nationally. I would write that one up as bunkum.

  20. Statgeek,

    I’m amazed too. The LDs going into coalition with the Tories was a golden opportunity for Labour, but they don’t seem to have capitalised very well on it.

    Might this be a case of a Green bounce? The Tories, UKIP and the LDs are about where they are on other polls (barring a batch of shocks for the Tories recently) so the surprising detail of the poll is really the low Labour score, and the Greens doing well due to the Euro elections could explain that.

  21. Btw I know that Lord Ashcott is Ashcroft. Damned automatic spellchecker…

  22. @ Anthony

    Thank you for the update – which also answers Paul A’s question. Do you think the supplementary questions are useful? Are they worth further consideration, do you think?

  23. Statgeek – those are the figures prior to the re-allocations of don’t knows.

    DJ – nope, this is the first of the series. He has occassionally done much bigger national polls, but this is the first of these smaller weekly ones.

  24. Amber – haven’t even looked at them. I wish companies wouldn’t put polls out when I’m supposed to be working :)

  25. Somewhat beguiling thread over at PB, wherein there’s a Twitter exchange between Lord Ashcroft and board favourite, Glenda Jackson’s son…

    http://www1.politicalbetting.com/index.php/archives/2014/05/07/lord-ashcroft-teases-us-about-his-latest-poll-of-the-marginals-can-anyone-decipher-this/

    “DH: One other thing. After that PPB let’s not have any more denials from Labour insiders about there not being a 35% strategy.

    LA: …it’s all in the marginals Dan! National polls mood music only…

    DH: Maybe, but Ed’s not getting 40% is he.

    LA: …wait until you see the marginal polls released at ConHome’s spring conference”

  26. Not worth getting too excited over. With other polls showing 1% Labour leads we were bound to get a small Tory lead eventually.

    Still a poor result for Conservatives. Uncapable of scoring above 34% in any poll. Can’t rely on Labour being even more unpopular to win.

    If Cameron can’t comfortable break the 35% mark then he can’t win regardless of how embarrassing Labour are.

  27. @ Anthony

    I wish companies wouldn’t put polls out when I’m supposed to be working :)
    ————
    You & me both! :-)

  28. May 24th is when the ConHome conference is.

  29. Anthony – Just curious how much weight you would put on these Ashcroft polls in the… Since it looks like you give populus online polls a weighting of around 0.96 in the UKPR average (before factoring in time decay, multiple polls in the period, etc)… Is it reasonable to treat these the same way?

  30. I actually feel slightly sorry for Lord Ashcroft because the immediate reaction will be “Oh what do you expect!”. It really would have been better for the credibility of his polls if the first of them had been more in line with other polls around at the time. Instead the partisan will either rejoice in such a way that discredit the result or will go all Mandy Rice Davies.

    I also need to do the usual reminder for phone polls that because far more people in them say Don’t Know or Won’t Vote or whatever, the effective sample size from which the VI percentages come are smaller. In this case margin of error is about 4.5%.

    From the tables, Labour actually have a one point lead among all voters, but after likelihood to vote is taken into consideration it switches to a one point Conservative which goes up to two when there is some reallocation of past vote.

    The make-up of the UKIP vote looks odd. It actually has only 25% coming from 2010 Tories but 23% from Labour (after LTV), though sample sizes are low of course. UKIP voters are also lower in LTV than we normally see.

  31. Also on PB, a look at the polling on Labour’s policy proposals…

    “Key LAB policy moves are polling well but including EdM’s name in questions appears to be a negative”
    May 11th, 2014

    http://www1.politicalbetting.com/index.php/archives/2014/05/11/labour-makes-heavy-weather-when-it-comes-to-leader-ratings-but-key-policy-moves-are-polling-well/

    “Above are four key policy areas tested in the latest YouGov poll and as can be seen the responses have been quite positive.

    For three of the four areas even CON voters give them their backing. For the other, renationalising the railways, Tory voters split 42-42.

    The rest of the poll is not so good for the red team. To all questions that have Ed Miliband’s name in the wording the response is negative.”

  32. It’ll be good to have another semi-tracker, even (especially?) if it’s out of synch with the other polls. Ashcroft is probably doing this in part to enable an academic comparison between the phone and online polling, so it should further our understanding of psephology, if nothing else. I actually feel a little sorry for him- I get the impression he would have preferred a less controversial result for his first ever poll!

    Does anyone understand why Populus is weighting this poll normally when their weighing for their online panel is so peculiar?

  33. Just for fun and giggles I put these VI figures into the Advanced Swingometre and got the following seat count:

    CON – 292
    LAB – 305
    LD – 23
    OTHERS – 30

    So on seats (UNS implied) technically crossover has still not occurred.

    It would be fun to see a Lab/LD govt on just 41% of those voting…but then of course we all know Lab would never ever ever go into coalition with Nick so I would probably be denied that amusement.

  34. Number cruncher –

    Dunno, options are to treat as a Populus poll (since essentially the only difference between this and a Populus phone poll is the badging) or as a newcomer with no record. I need to revisit all those weights sometime before the end of the year anyway.

  35. Non Ukip others at 10% here, presumably. This seems remarkably high, and may hint at Green’s suppressing Lab vote share as part of the Euro poll effect?

    Having said that, we’ve had a couple of Euro polls which show Lab trailing UKIP badly, and only just holding second place, so there is some other polling data to suggest Lab are suffering.

  36. GRHinPorts,

    I still expect that we’ll see a Labour government on about 33-36% of those polling, probably with a 15-30 seat majority. I don’t expect Tory voting % to hold up under sustained UKIP coverage over the next 12 months.

  37. @Roger

    “I actually feel slightly sorry for Lord Ashcroft because the immediate reaction will be “Oh what do you expect!”

    Indeed. A tweet I received on the subject:

    “Who did he poll? His tory mates? Utter rubbish.”

    sigh!

  38. Thanks Anthony!

  39. ICM poll due this afternoon as well, apparently.

  40. Depends on what you mean by ‘shift to the right’, statgeek. Many of UKIP’s members are quite left-leaning on economic issues. UKIP supporters tend to support re-nationalization of rail and energy in considerably larger numbers than the Conservatives. (The fact that UKIP politicians are radical free marketers seems to have eluded them).

    I think we can safely say that it’s a shift to the right on cultural issues though. (immigration, political correctness, etc.)

  41. To ruin Anthony’s work rate even more Opinium have finally got a Euro poll out after much nagging:

    http://news.opinium.co.uk/sites/news.opinium.co.uk/files/eu_060514_2.pdf

    Conservative 23 +1

    Labour 27 n/c

    Liberal Democrats 8 +1

    UKIP 28 -3

    Plaid Cymru 1 +1

    Scottish National Party 3 -2

    Green 5 +1

    BNP 3 +1

    Opinium Research carried out an online survey of 1,972 GB adults aged 18+ from 6th to 8th May 2014. Results have been weighted to nationally representative criteria and filtered to those likely or certain to vote, taking into account past voting behaviour in European Parliament elections. Changes refer to our previous unpublished poll (data tables available shortly) conducted between the 29th April and the 1st May.

    They seem to be the only people to have asked how people voted in 2009, which is interesting. They also have adopted a ‘ballot’ approach listing the Parties available for each region depending on where the panelist is[1]. Though I suspect this has led to a programming error giving the BNP 16% in London.

    [1] Except they seem unaware that Mebyon Kernow and the Christians aren’t standing the South West and there are other additions/omissions. Anthony will be annoyed that the Roman Ave Party aren’t listed separately.

  42. SNP down, Tories up- Oldnat, what have you done?!

  43. @Anthony Wells – Your list of methodological issues does not cover sampling methodology, which I don’t think should be taken as read.

  44. I agree with bill. though i think labour will fall just short.

    this reminds me of the obama/romney fight in 2012. there’ll be the odd polls showing tory leads which will be seized on as evidence of dave’s victory. there’ll be hype etc.

    just as we were told ahead of nov 12, that no sitting president had been re-elected in such dire economic circumstances, we will be told that no leader of the opposition has won with such bad personal ratings etc…

    there’ll be much sound and fury, many false dawns, over optimistic analysis and then the result will be ED M in no. 10…depressing thought, but that’s what i think will happen.

  45. Anthony, did anyone complain about Lord Ashcroft’s polls when he was showing clear Labour leads?

  46. @GRHINPORTS: “…but then of course we all know Lab would never ever ever go into coalition with Nick”

    On those figures it might not be dealing with Nick.

  47. The best odds still on offer for GE 2015:

    Con Maj 7/2 (implied probability 22%)
    Lab Maj 13/8 (implied probability 38%)

    NOM 11/8 (implied probability 42%)

    So Labour or NOM still approx 75-80% probability.

    The Tories have probability against them but with a few more polls like this, it will all move in their favour.

  48. “Anthony, did anyone complain about Lord Ashcroft’s polls when he was showing clear Labour leads?”

    Many lols . :-)

  49. @GRHINPORTS: “…but then of course we all know Lab would never ever ever go into coalition with Nick”
    On those figures it might not be dealing with Nick.

    True enough. But Others will be a fair hotch potch of parties which might be difficult (or not) to harmonise into one coalition.

    The thing about those figures is that they leave Lab short of a majority by 21 and the LDs could easily bridge the gap.

    Im sure there will be much posturing from all parties about how they would never work with such and such a party in coalition ever.

    I seem to remember endless Tory declarations before May 2010 how they would never work with the LDs.

    However if the parliamentary maths dictate a certain result like above then IMO all bets are off.

1 2