The Lib Dems have been out and about briefing journalists about what their polling shows. This morning the Guardian, May 2015 and The Spectator have all written about it.

Private polling always has a certain allure when spoken of in the media, there is that whiff of forbidden, insider knowledge. It really shouldn’t – beyond asking whether such private polling actually exists, the first major caveat to any claims that private polling shows something different to publicly available stuff is “why should it?”. Political polling doesn’t really make much money, it’s normally done as a shop window to get a polling company’s name known and to create a reputation for accurate research. It is in polling companies’ interest for their voting intention polls to be as accurate as possible, so the publicly available stuff really is the best we can do, there is no more accurate version of the data held back for private clients. Therefore, most of the time when political parties claim their private polling shows them doing better than the public polls it suggests they are making it up, or they have been commissioning stuff that asks things in a skewed way for propaganda purposes (what Lord Ashcroft has called in the past “comfort polling”).

Even if the full details of private polls are released, with all the methodology and tables available for scrutiny, you should still view them sceptically. Parties decide which polls to release or brief journalists on, if any. One thing you never see is a political party releasing a poll that is not helpful to them, so even if the polling itself is above board, there is a strong publication bias; only the stuff that helps the party is published.

To illustrate that, cast your mind back to the polling of Lib Dem seats conducted by Lord Ashcroft. One of the most obvious findings was how much the Lib Dem performance varied – in some seats like Eastborne, Birmingham Yardley or Sutton & Cheam the party is doing very well indeed. In other seats like Somerton and Frome, Chippenham and Brent Central they are doing atrociously. Imagine the very different narratives that could be created by selectively releasing polls from those first three seats, as opposed to selectively releasing polls from the latter three.

So what should we make of the Liberal Democrat claims? Well, the polling does genuinely exist – Survation are a proper company and while the newspaper reports don’t include specifics, several journalists have assured me they were shown the actual figures. For once, there is also a legitimate reason why the Liberal Democrat polling might show them in a better position than the published polling – the articles suggest they prompted using candidate names. We know that a lot of Lib Dem support is reliant upon tactical voting and personal votes, so it seems reasonable that polls that include the candidate names in Lib Dem held seats might show the Lib Dems doing better. Until we see tables we can’t tell what other methodological factors may have been at play.

The articles claim that the Lib Dems have done about 100 constituency polls, while it seems journalists were shown results from about a dozen or so, so it is impossible to know how representative this group were, or whether they were cherry-picked to create a good impression of the Lib Dem performance.

Even if you take the claims that the Liberal Democrats make at face value, they don’t actually show much that contrasts with existing publically available data that much. According to the Guardian the party “is on course to remain “competitive” in seats that would fall if there was landslide against the Lib Dems, such as Cheltenham, St Ives, Cardiff Central, Eastbourne, Solihull, Cheadle, Leeds North West, Cambridge and Bermondsey.” From talking to several other journalists who were there, I’m told they were also shown figures from St Austell and some other Labour facing seats, so perhaps a dozen in total.

The word “competitive is vague” – it could mean both a little ahead… or a little behind. As it happens, Lord Ashcroft has conducted polls in nine of those seats, and in most of them the Lib Dems are indeed “competitive”. In three of them (Cheltenham, Cheadle, Eastborne) Ashcroft found clear Lib Dem leads, in two others (Bermondsey and St Ives) he found the Lib Dems just ahead (though I’m told the Lib Dem polling shows them doing better than that). In Cambridge Lord Ashcroft he found them a point behind… but that counts as “competitive” in my book. The other three are St Austell & Newquay, Solihull and Cardiff Central, where Ashcroft found the Lib Dems trailing. If the Lib Dem data was kosher, then it may well show the Lib Dems doing better in those seats… but polls vary, and perhaps there are other, unmentioned, polls that show the Lib Dems doing worse than in Ashcroft’s polling.

87 Responses to “Lib Dem private polling”

1 2
  1. “I understand Survation will be releasing some figures later on today, so we shall be able to see for ourselves…”

    Sounds like the Lib Dems have shot themselves in the foot. If they hadn’t named Survation they wouldn’t have needed to publish the polls.

  2. They should’ve (the pollster is supposed to identify themselves under those circumstances). In practice it makes it harder to smoke people out, but in this case, it was known who was doing constituency polling for the Lib Dems anyway. It’s a small world.

  3. Lib Dems have announced record donations for 2014.

  4. Hi,

    A bit of non sequitur, but I was wondering if there has been an article on the uns projections. I continually see them on the website, and I would love to know what they are predicting would happen in terms of seats should the votes be equal, what each party would need for an overall majority, would the conservatives have more seats should they poll 1% above Labour etc.

    Just a thought, but it would make the polls even more interesting should I be able to convert them (as a forecast) into seats.

  5. Yeah, but all the records are The Chipmunks and Perry Como.

  6. I would think the Lib Dems are probably quite happy to release the tables. It will only be the tables for the seats mentioned I guess – the good ones – which should help counter the narrative that the Lib Dems are finished that seems to be going around.

  7. “the Lib Dems have done about 100 constituency polls, while it seems journalists were shown results from about a dozen or so,”

    They must be holding back the 88 showing them doing even better until closer to E day.

  8. #WinningSomewhereAtLeast #StrongerEconomyForPollsters

  9. @MrNameless

    Please don’t mention the dreaded Chipmunks.

    My 2 year old has an obession with that film and the songs within it and I would happily send those records to Nick Clegg for him to listen to for all eternity.

  10. Interesting stuff, and if true, obviously good news for the Lib Dems and, possibly, Labour too. If the Lib Dems are going to beat off Tory challenges in the LibDem/Tory marginals, then that will reduce the number of seats the Tories can win to offset likely losses to Labour in Lab/Con marginals. To counter that, Labour may not win as many Lab/Lib Dem marginal seats as once seemed likely, but there aren’t as many of those. Accordingly, the arithmetic of a stronger than expected Lib Dem showing in May looks, prima facie, to favour Labour over the Tories.

    Of course, there is also the question of what the Lib Dem vote does in Lab/Con marginals where Labour are chasing the Tories.

    Interesting times indeed.

  11. On the article subject it says they have conducted around 100 Constituency polls in the last 12 months.

    Lets assume they’ve decided not to bother polling really safe English seats where Con are the main oppposition like Westmorland.

    This suggests they’ve polled all the potentially vulnerable seats about twice with maybe some done 3 times.

    The fact they have polled Leeds North West and Old Southwark & Bermondsey suggests even the safest Labour vs Lib Dem seats have been polled. So we need to assume all the vs Labour seats have been polled including Sheffield Hallam where they specifically talk about squeezing Con rather than Clegg being ahead at present.

    The fact that they don’t mention polls from other large majority vs Labour seats like might suggest that even on their own numbers they are in trouble in seats like:
    Bristol West
    Hazel Grove
    Hornsey & Wood Green
    and all the Scottish seats

    In fact claiming you are “Competitive” in your 10th safest seat (Leeds NW) is rather a worrying sign in itself.

  12. @Ian Watson

    I think Anthony has mentioned UNS has flaws, better to use one of the more complex models – Anthony sums them up in his weekly roundup – see last week’s here – end of post

  13. Typical good sense from Anthony in his campaign to throw buckets of cold water over the media’s more excitable coverage of polling (ie 99% of it). Though given the past form most of these people have, he might be better giving up on the buckets and aiming one of Boris’s water cannon down the Street of Shame.

    The only thing I’d disagree with is Private polling always has a certain allure when spoken of in the media, there is that whiff of forbidden, insider knowledge. It doesn’t have any allure to people outside the media at all – most of us just go “Well they would say that, wouldn’t they?”. If anyone is taken in, it’s the media in love with their own insideryness.

    Actually I genuinely don’t understand what the Lib Dems are playing at here. They can’t fake stuff because they have Ashcroft’s figures to keep them straight. As NorthumbrianScot points out, by mentioning some seats where you’re doing well, you just draw attention to those you don’t mention where everyone will assume things are going badly. Especially if you brag about having done 100 polls.

    And because they have Ashcroft’s figures, they should need to spend less money on polls not more. I can see this sort of thing causing great resentment among Lib Dems all over the country (there will be thousands fighting council seats as well in May).

    The whole Lib Dem set-up rather reminds me of the Yes campaign in the AV referendum – which was also dominated by the Lib Dem HQ types.

  14. That guardian piece has this to say about Sheffield Hallam:
    Party officials are confident that Clegg will retain his seat in Sheffield Hallam despite three polls showing he is behind. They admit the seat has switched from being a Lib Dem/Tory fight to a contest against Labour, but are confident that the large Conservative vote in the seat can be squeezed by warning of the dangers of a Labour government with an overall majority.

    I note that Sheffield Hallam doesn’t appear in the list of private polls journalists were shown.

    Also, Anthony, just FYI your Spectator link is pointing to the May2015 site.

  15. I go away for one night, and we get a Scottish thread (opened and closed again), and QT and bias BBC gets mentioned. No trusting some folk!

    All the polls seem in agreement that the SNP are mid 40s this week.

    All the UNS calcs have the SNP on 45-55 seats.

    Room for improvement. :))

  16. @Roger Mexico – “I genuinely don’t understand what the Lib Dems are playing at here.”

    Perhaps they want some up-to-date data to use in bar charts for their leaflets?

  17. Almost all of the information in the public domain suggests that things are dire for the LibDems. In Anthony’s Polling Averages and in in the various graphs, their current VIs are not much above those of the Greens. The only hope for them is the last minute appearance of the swingback cavalry (perhaps taking the form of unusually strong incumbency effects).

    Unfortunately for LD supporters, there is already quite a lot of swingback already built into some of the – already sobering – current seat projections and this raises the question of whether it is plausible to suppose that there is more to come. With this in mind, I thought it would be interesting to look at the role that anticipated swingback plays in tilting the balance of seat projections in the model that is currently most optimistic about their election prospects (Electionforecast).

    The Electionforecast (EF) projections today are given below with an indication of the size of the swingback [1] adjustment built into each party’s seat prediction:

    Labour 282 seats (based on a -0.5% swingback adjustment)
    Con 281 (+2.5% swingback)
    SNP 39 (-0.3%)
    LD 25 (+3.8%)
    UKIP 2 (-4.4%)
    Greens 1 (-2.2%)
    Others (inc. NI, PC etc.) 20

    So, the model’s projections are based on the premise that the LDs (and the Tories, of course) will enjoy something of a surge in support over the remaining 10-11 weeks, and that the remaining parties will suffer a loss of VI.

    If these are the kinds of swings you expect to see, then no doubt you will also see a lot of plausibility in the overall seat projections, and perhaps see little need for secret polling to boost morale. However, in the absence of evidence that this is what is happening, it might be worth looking at the effect of making different assumptions about the shape of future changes. I have done this using two scenarios: (a) continuation of 2014 trends until May 7 and (b) continuation but with tweaks to take account of systematic departures from trend over the last few weeks. This gives the following projections:

    Labour 288 (286 after tweaks)
    Con 266 (268)
    SNP 50 (50) {Ignoring 2014 trends: Largely based on @OldNat’s stable SNP averages}
    LD 19 (21)
    UKIP 5 (2)
    Greens 2 (3)
    Others 20 20

    Clearly these projections are somewhat less favourable to parties that are expected to benefit from swingback. Where you place your bets will depend on the confidence with which you expect swingback to make a late but dramatic appearance.

    [1] Swingback percentages are calculated by subtracting the current from the projected VI for each seat, and then calculating the average discrepancy.

  18. There is practical need for Lib dem Central to keep it’s activists as positive as possible, even if the pudding is being over-egged.

    If they don’t get out, put up Council candidates and get working, they stand to face utter oblieration in May.

    I really do admire the Lib Dems cheerful disposition in adversity. To most folk, it looks like the run up to May is their ‘the Green Mile’ – with death by ballot box at the other end.

    They are facing it with some aplomb.

  19. @Allan Christie, previous thread – That’s somewhat fair, though it is a weaker comparison.

  20. @CatManJeff – Regarding the supposed aplomb of the Lib Dems, remember that by definition these are the devoted rump after vast swathes of their less “aplombatic” activists quit over the last five years.

  21. @ AW

    “The articles claim that the Lib Dems have done about 100 constituency polls”

    I’m sure I remember you saying once that it costs about £10k for a proper constituency poll. If my memory (and my maths!) is right then it seems unlikely to me they would spend £1m on doing 100 of them???

    I’m very sceptical they would want to spend any money on anything outside the 57 they hold or that they are doing any serious work at all in more than about 60.

    As far as any leading questions go it’d be interesting to see what a properly commissioned poll came out with if they went down the naming the candidate route, calling it a two horse race, showing respondents not to scale graphs and so on. I actually think those will be the conditions on the ground so I’m not sure whatever leading questions the Lib Dems put into the polls to get voting intention will be that unreasonable.

  22. @Chris Grren

    Indeed yes.

    The hardcore left are the ones who will cheerfully march off the end of a cliff. for the cause.

  23. Shevii
    The articles AW posted above claim they’ve spent £350,000 – I don’t know what period/how many polls this covers. If it does cover all 100 I’d be interested in seeing the tables to see if they’ve tried to scrimp by doing smaller samples than they should.

  24. Unicorn,

    excellent post, mate. I was wondering what swingback assumptions were built in, because 281 seats for the tories looked ambitious on current polling.

    When you write 2.5% swinback, is that the effect of the swingback on total vote share or is that the swing itself….If the former, the swing is 1.25%, no?

  25. Maybe Survation do them cheaper, maybe they got a bulk deal! Maybe it was £350k a year, or £350k for the campaign. Who knows.

    The 100 polls question though makes more sense. The Guardian article talked about swings in groups, changes since 6 months ago, etc, etc. That suggests to me they’ve polled some seats more than once.

  26. Forgive my ignorance, but is their an established date for the dissolution of Parliament? If not, when is the likeliest date for the end of the phoney war?

  27. @Unicorn

    Lib Dem v Green updated (bottom of page – F5 to refresh if you have been here recently)

  28. Apologies that this is completely off topic, but something struck me.

    I’m on the Populous panel and often get emails inviting me to participate in surveys. I don’t normally get a chance to do so until the evening, and after putting in my demographic details I invariable get politely turned down as they have enough people of my demographic. Today I was working from home so had time to complete the survey before starting, and after the demographic questions went straight to the VI questions before moving onto the body of the poll.

    All of which got me wondering – if all their VI polls only accept the first batch of people fitting each demographic, does this skew the polls in favour of those who, for whatever reasons, can regularly participate in polls shortly after being invited? At least for those demographics which typically generate too many respondents.

  29. Good Afternoon All.
    CROSSBAT 11.
    I think that in the Feb 1974 GE that the Liberal revival under JT was a vital element in Harold Wilson’s surprising ‘victory’.

    Maybe, as you are suggesting, the Lib Dem revival will help Ed M, by stopping Tories gaining many seats to off set the gains Labour expect in their top 40 target seats.

    Simon Hughes would be a great asset for the Lib Dems, I think; just my view.

  30. “In other seats like Somerton and Frome, Chippenham and Brent Central they are doing atrociously”

    Nice shout out for my ole Granny’s seat of Somerton and Frome, She once told a Lib/Dem to get off her path or she would boot him in the Mendips.

    Anyway back to Lib/Dem polling. My feeling is that the Lib/Dems vote share will plummet in May but their losses might not be as bad as the national polls have us believe.

    However they do look to be in for a hammering in Scotland and I would guess in England they will lose most if not all of their urban seats and that includes the jewel in the crown, Nick Clegg’s seat.

  31. This has been a good news week on the economy, with today’s borrowing figures out and showing some healthy numbers. If we are going to experience swingback, these are the kinds of weeks that one might expect to make a difference.

    Having said that, I posted last night an analysis of the latest Markit Cips Household Finance Index findings, although having my posts tied up in auto mod probably means most people miss them.

    In a nutshell, the February figures are showing a sharp decline, with the headline index falling for the first time in five months, and the perceptions of the next twelve months taking a sharp downward turn.

    Personally I found this unexpected, as the macro economic data was pointing to a more sustained period of low inflation, wage increases and increasing employment activity. Quite why the HFI has dipped is unclear, but may help to explain why the apparent good economic news isn’t yet showing through in movements in VI.

  32. Paul – 30th March I believe (though it may be prorogued before that)

  33. thank you

  34. Well the Tories do seem worried that their campaign is turning off women

    So that does chime with the Lib Dems increasing their support amongst women.

    And as a reminder, look at those historical graphs on the right and what happens to the Lib Dem vote share in the last weeks of every historical campaign since the 80’s.

    Yes, it may be different this time with the Lib Dems being part of government but I suspect the mud throwing does turn people off and sends them to the non mud slinging parties.

    And most of the undecided voters are 2010 Lib Dem voters, they have to go somewhere….

    Many thanks for your detailed post.

  36. I believe they actual story here is not the seats that they mentioned but the seats that they failed to mention. i.e. seats which the Lib Dems have polled but the results are so bad they don’t want to mention them for example a lot of the South Western marginals and a whole bunch of Labour facing seats.

  37. The Scottish Spring Party Conference season has kicked off with the Tories in Edinburgh.

    I predict zero change in the polls by the end of the conference season.

  38. Unicorn

    Interesting that EF are expecting what appears to be a swing back from Freen to LD, since the direction of travel of most of these voters has been 2010LD–>2011-13Lab–>2014 Green. Is it realistic that these voters will complete the circle?

  39. @ Statgeek

    Thanks. Always interesting. Perhaps crossover has already happened up North…

  40. ANTHONY.
    It is on March 30, I think.
    March 25 is when the Civil Service goes into ‘purdah’ and cannot give Ministers advice on policy proposals.

    At least that is what my Politics students have been told; I heard a very good Radio 4 programme about this recently.

  41. @ James Peel

    When you write 2.5% swinback, is that the effect of the swingback on total vote share or is that the swing itself….If the former, the swing is 1.25%, no?

    No – I am not measuring swingback here in the same way as the classical 2-party swing. The figures I gave in brackets are the average increase (or decrease) to a party’s VI between now and the election.

    So, if the model has Party A on VI = 20% now, and the same party is projected to have a VI 22.1% on May 7, then I would enter this as a positive swingback of 2.1%.

    (Conventional swing calculations are all about the VI change for one party relative to that for another. This is not very illuminating when each and every party isor reding along its own individual swingback journey).

  42. chrislane1945

    It is on March 30, I think.”

    Yes, 30 March:

  43. @ LeftyLampton

    Is it realistic that these voters will complete the circle?

    The model obviously doesn’t assume any particular voter itinerary. So, if it posits a LibDem increase and a Green VI drop, it does this without making any assumptions about how this is effected. There could be any number of underlying shifts to produce the overall effect.

    Also – as you no doubt know – swingback/regression models are based on patterns of VI change that occurred in previous elections. The working assumption is that the patterns will repeat themselves. If you believe instead (say) that left-leaning LibDems have departed forever, then it is important to disregard the handle-turning aspects of models based on the premise that patterns of the past must always be reinstated.

  44. @Chrislane

    Feb 74 was the first election I voted in and so it sticks in my memory. I think you are right to say that the Liberal “revival” took votes from the Tories, but we have to remember that they took votes from Labour too. Both parties lost vote share; Heath came off worse.

    Thorpe was clearly a personable fellow and a creative campaigner, but a bigger promoter of the Liberals was the “plague on both you houses” school of thought. Neither Heath nor Wilson were trusted. Thorpe was the only alternative.

  45. @Roger Mex
    ” The whole Lib Dem set-up rather reminds me of the Yes campaign in the AV referendum –which was also dominated by the Lib Dem HQ types.”

    Very much agree with you, and a crying shame it was.

    Many thanks also for your comments last night on why some at the top of the Labour party are comfortable in opposition. “What oft was thought but neer so well expressed”.

  46. “Maybe Survation do them cheaper, maybe they got a bulk deal!”

    Tradesman: “We are doing a job in the next street. We can do your double glazing at half price.”

  47. Interesting, albeit anecdotal story from a constituency that unlikely to change.

    For my shock a Green Party leaflet arrived. Unfortunately my Kuvasz ate it. So I phoned the local Green Party office to ask if I could pick one up (I was interested as their last leaflet was really bad). Got the answering machine that they are open from 9-5 which is not a very good sign for a challenging campaign.

    So I tried the others. The Conservatives don’t even have an answering machine. UKIP although came second in the EUP elections don’t have a local office. LibDems immediately tried to make me their activist (perhaps only such people ring them), in spite of my quite strong accent. Labour asked if I had already been down as a voter for them and would I put out a poster in my window (I assume they meana vote Labour poster).

    On this basis I predict a 6% swing to Labour in this constituency with about 62% share of the vote.

  48. As the GE approaches I expect that all the Parties will issue statements saying that they are doing better than it is generally thought. The LibDem articles of this thread are just the first. The Parties will hope that the undecided voter will choose to back the winner; or at least the undecided voter is unlikely to back a looser.

    Nigel Farage’s recent statement that DC is likely to be PM after the GE is an interesting variant. Presumably he is afraid of “vote Ukip get Miliband” gaining more traction.

  49. UKIP have got a campaign day tomorrow in the Endcliffe Student Village, in Hallam. I estimate they might get about five votes. It’s actually probably helpful for Labour that they do so, since it might scare Labour people into thinking UKIP might win.

  50. @Laszlo

    Nice story.

    My guess is that your strong accent doesn’t stop you from being a LibDem activist, because nobody is listening anyway. You might as well just do the whole thing in Hungarian.

    They could deploy your talents in Wales to good effect. I’m sure your rendition of A Walesi Bardok would get a rapturous welcome.

1 2