I’d have several comments here (and seen countless similar comments on Twitter) asking why, if the Lib Dems got 16% in the local elections this month none of the polls show them with that level of support. They normally proceed to conclude that the polls are wrong.

As I’d said several times here before, people vote differently in local elections than they do in general elections, so polls asking how people would vote in a general election are not a good guide to how they would vote in a local election (and conversely, how people vote in a local election is not necessarily a good guide to how they would vote in a general election).

Anyway to try and finally nail down this nonsense, below is a graph comparing what the Liberal Democrats got in local elections since 1992, and what opinion polls of general election voting intention were showing. Local election figures are the Rallings and Thrasher National Equivalent Vote for each year (you’ll note that 1997 and 2001 are missing, since as far as I can tell R&T did not produce a NEV figure for those years. I have not yet seen the R&T figures for 2012), polling figures are the average Lib Dem score in polls from all companies in the first four months of each year.

As you can see, the contrast between what the Liberal Democrats manage in general election voting intention polls and what the Liberal Democrats achieve in local elections is not new – it has been consistent for the last twenty years. While the gap has gotten bigger and smaller over the years, peaking during the 90s and at its smallest during Cleggmania, on average Liberal Democrat support has been seven points higher in local elections than what they were polling at the time. Over the same period, polls have been largely accurate in predicting Liberal Democrat support at general elections, with the exception of 2010 when they over-estimated Lib Dem support.

In short, the difference between Lib Dem support at local elections and national polls is normal, and Lib Dem performance at the 2012 local elections is wholly consistent with their current ratings in general election polls.

31 Responses to “Lib Dem support at local elections”

  1. So this proves ALL the polls have been wrong for about twenty years ;-)
    (kidding, Anthony)

  2. It will be interesting to look at how the boundary changes would specifically affect the Lib Dems, based on current polling.

    I think the LD’s could lose most of their seats in Scotland and many of their seats elsewhere. Chris Huhne looks pretty safe and I think Tim Farron should be ok, but I think many of the others could be difficult. Could even Simon Hughes be in danger of losing to Labour ? I have automatically presumed that Nick Clegg does not stand much of a chance of being re-elected, if he is a candidate.

  3. Anthony, I think this is an excellent analysis of the difference between VI and local election results for the LibDems.
    Is it also true that differences between VI and local elections results for Labour tend to show VI LOWER than Local election results, with Con about the same?

    When the first YouGov poll came in after the elections I remember feeling that the poll was in line with Locals actuals, if one made these sors of adjustments.

    I think the YouGov polls still broadly fit in with the Local results, if adjusted appropriately.


  4. @Alec


    On DM- mea culpa on that one: for some reason I recalled you as being amongst the batch of people on here who do.

    I still think- for the moment- that EdM will not win Labour an OM: that’s an honest disagreement between us.

    Thread topic

    It would be interesting to see a graphic of the Lib Dems average performance over a parliament in the locals against their number at the subsequent GE.

    Though I do think there is a lack of relevance here: this is a totally different scenario to the period in the graph.

    Clegg and the Lib Dems are now in government (so no home to protest/ NOTA/ ‘principled left’ anymore). They are also supporting a – by most people’s opinion- pretty right wing government. I cannot see where they get more than 10-12% at the next general election and don’t consider valid the view that ‘a concentration of vote in SW and NW means we’ll hold onto 20-30 seats even with 12/13% of the vote’.

    I think current OP numbers (and local elections losses) point towards Lib Dems falling below 20 seats (on current boundaries). They could turn it around of course but I think it is unlikely.

  5. THe increase in LD VI in the early 2000s is interesting.

    On that bar chart they appear to have put on around 10 points then ( Labour fell back to low 30s at the same time).

    In a very kind resonse to an impertinent question from me on another thread, LIZH said that this was when she stopped supporting Lab & switched to LD.

    If this switch , of up to 10% pts was a Labour Further Left rejection of Blairism, the two factors come to mind :-

    The current VI traffic in the opposite direction might well be those self same Labour anti-Blair defectors going home.

    What effect would occur to current Lab VI ( given the above hypothesis) if the trailed EM reshuffle brings back Blairites like Darling, Adonis , DM & Mandy ( currently schmoozing EB ) -does not eject Byrne…..and sees The Permatanned Satan himself return to endorse the young leader.?


  6. aw,

    or should it be ‘local election results for Labour tend to show VI HIGHER than local election results.’

    Sorry, I’m confusing myself — it doesn’t take much.


  7. @AW

    How much predictive power is there in LD GE VI polls versus LE results, especially compared to the other big two parties? Just looking at your graph it looks like there isn’t a particularly strong relationship (e.g., compare 1995-2000 with 2005-2010), but human eyeballs are a notoriously bad statistical test!

  8. @ R Huckle,

    If you use the new boundaries swingometr on the right, you see that on their current scores in YouGov, the LibDems get massacred under the new boundaries.


  9. @Coiln

    “What effect would occur to current Lab VI ( given the above hypothesis) if the trailed EM reshuffle brings back Blairites like Darling, Adonis , DM & Mandy ( currently schmoozing EB ) -does not eject Byrne…..and sees The Permatanned Satan himself return to endorse the young leader.?”

    I wrote yesterday:

    If the policy platform *does* represent the kind of social democratic centrist position that IMO is needed for a chance at an OM, Labour are going to – probably but not inevitably- lose some of the ‘petulant left’ vote that (effectively) voted in the coalition by backing Clegg or the greens at the 2010 election- the sorts of people who were SO annoyed by what Ed Balls said in January. These are people who since ConLib have swung back behind Ed and believe the ‘we have our party back’ rhetoric of some observers.

  10. FPT

    @Colin – “I cannot make my mind up about a Greek default & secession from EZ.”

    You probably don’t have to – it was decided years ago.
    The real tragedy about this is that the blindingly obvious has been ignored by the EU governments in an effort to save their own reputations. As a result, what should have been a planned and supported strategic realignment of the Eurozone, run at a timetable set by policy makers and with sufficient space to allow firewalls and rebooting of failed economies to take place, will now occur at a pace set by panicking markets.
    When it comes it will probably happen in the space of days or even hours, and will cause maximum damage to the poor Greek citizens and likely contagion elsewhere, with an ill prepared ruling elite largely powerless to intervene as they have sat on their hands and refused to countenance such an occurrence as it hinted at their own political mortality.
    I still favour something like a plan for Greece, Italy, Spain, Portugal and Ireland to leave the Euro. They would need to devalue and consequently default, so all creditors would exchange current debts for new 30 year bailout bonds issued as part of a rescue package that would not pay any interest. These would be affordable, but would mean that banks would maintain capital assets that would degrade slowly over time, spreading the losses over a lengthy period.
    In due course, these could be redeemed early or written off in a staged and structured manner as and when economic conditions improved, with the burden being shared between taxpayers and creditors over a long time scale and without a single shock event to paralyse the system.
    I seriously doubt the banks and markets would accept such a deal, but then they are terminally stupid, which brings us back to why we’re here in the first place.

  11. Andyo – yes, generally speaking there tends to be the opposite pattern with Labour. They do worse in local elections than their national standing.

    Colin – what you are looking at there is, at least partially, the Iraq war.

  12. @rob Sheffield – Thanks. For the record, I’m not saying EdM will win a majority either. I rather suspect another hung parliament may be in the offing. I just think that at present, the tides are such that it’s going to be easier for Labour to land on the beach than the Tories.

  13. I guess this is actually pretty bad news for the Lib Dems as the implication is that the current VI polls are an accurate reflection of what they would get in an election tomorrow.

  14. @R Huckle

    Putting yerterday’s dramatic and possibly unrepresentative14 point YouGov Lab lead, with LD on 7% into the swingometers…

    existing boundaries: Con 216, Lab 390, LD 19,

    new boundaries: Con 203, Lab 366, LD 8.

    So Labour losing 24 seats from the review, Con 13, and LD 11.

  15. Anthony

    Thanks for that. I did a much simplified version before the 2011 Locals, trying to work out the ‘local premium’ for the Lib Dems:


    I reckoned that the Lib Dems normally got 6-7 percentage points more in local elections (in non-general election years) than they were currently getting in the polls. I only went back to 2002 and used the Wiki estimates, but the overall picture is similar.

    Interestingly, the ‘premium’ seemed to be independent of the level of Lib Dem support, but looking at your more detailed picture there does seem to be an effect that the higher the Lib Dem polling is, the smaller the premium is.

    You would to some extent expect this, because part of the premium must be because the Lib Dem vote is more likely to come out when they are doing better in the polls. However it is still true that most of the premium is a fixed amount, perhaps due to voters who see the Lib Dems as a local but not national Party.

    I assume R&T missed out 1997 and 2001 because they were General Election years where any effects get swamped by the 30% of the electorate who normally vote in GEs but not locally. This rather than Cleggmania is probably the reason for 2010 being different.

    Having said I’ve pointed out the reasons why polls and LE results differ numerous times on here. No one ever takes the slightest notice and they just keep on announcing that the figures from a Town Council byelection in Somerset in a wet January show that entire history of opinion polling has just been disproved. Honestly it’s worse than global warming denial.

  16. Repost from previous thread; It took a little time, and has mention of the Lib Dems.

    Some graphs on the changes in VI over the past seven weeks. Everytime there’s a new poll, I take a note of the median absolute deviation (MAD) of the parties. This cuts out the outliers and gives (hopefully) a slightly more accurate state of play. The MAD samples used to calc it all use the last 30 polls, so there’s scope of error, but hopefully not too much.

    As time passes the gaps open up, assuming there are gaps. On a daily basis you might see no change or slight change, but over a period of seven weeks (32 polls in this instance) there is significant change.

    First a graph comparing the MAD on 27th March and the MAD from the 14th May:

    h ttp://imageshack.us/photo/my-images/441/thennow.png

    Of particular note are the UKIP gains in the Rest of South and Mid/Wal (with corresponding Con losses; In fact the con VI is down everywhere), and a possible Lib Dem recovery in Scotland.

    Next, the changes graphed:

    h ttp://imageshack.us/photo/my-images/252/changec.png

    Lastly, the sum of the changes by region (hopefully indicating if turnout, is rising or not):

    h ttp://imageshack.us/photo/my-images/33/turnout.png

    Not very scientific, but it does make for interesting speculation.

  17. Interestingly the closer Labour moves towards landslide territory, they still lose the largest number of seats because of the boundary review…. but using the above example, proportionately it is a close run thing with the Conservatives:

    Labour sees a 6.15% reduction in their parliamentary representation, Con 6.01%, LD 57.89%.

  18. @BILLY BOB

    Could it be because now Labour have to eat into Con core votes to make gains beyond what they already have?

  19. You’re all getting over-excited especially those on the red (or pale pink now) side. In areas where there are Lib Dem MPs – including Sheffield Hallam – the Lib Dem vote was much better than in areas where there are not. In some of those areas the Lib Dems not only held their council seats comfortably, but actually gained some. In Colchester for instance they held all nine mostly with increased majorities. The Lib Dems on present showing would lose seats, but not anything near the dire predictions in some of the posts above. This factor should be taken account rather than a uniform swing assumption. Are there any figures on this?

  20. Rallings and Thrasher wrote a short piece on the local elections for The Sunday Times (06/05/12). Their national equivalent share came to –

    Labour 39%
    Conservative 33%
    Lib Dems 15%

    This is of course different from the BBC’s calculations which uses slightly different methods and is probably carried out by Strathclyde/Curtice.

  21. ROB

    Thanks-presumably they will go Green in the event of what you euphemistically describe as “social democratic centrist position “……it would be the only home left for TGB types?


    Thanks -yep I understand that was the big catalyst for the War Crimes Tribunal crowd.


    I replied on the previous thread.
    I had understood that AW wished us not to carry debate across threads.

  22. Just watched Blairite Sally Morgan-now Chair of OFSTED on DP-crossing swords with some teaching Union bloke.

    A Martian would have been unable to decipher that this was not a Conservative politician.

  23. What an excellent, clear and concise article. It won’t stop the twitters and others arguing that they know better, Im afraid.

  24. It is wholly correct not to take local election results as a snapshot of Westminster voting intentions. Since 1990 Lib Dems have always done 5-8 points better at local elections at the expense of Labour.

    I had a look at the local elections share of the vote vs Westminster polls since 1980 below


    Another important point is that governments generally do worse at local elections. Conversely Oppositions do better.

  25. Colin

    “A Martian would have been unable to decipher that this was not a Conservative politician”

    and earlier post on reshuffle (return of the supertan etc): I admire your mischief making today !

  26. I agree that opinion polls underestimate LibDem support at local elections – but at the moment it is by a factor of 2 – that is 100% wrong. To say this is withinthe margin of error is of course willfull misunderstanding.

  27. I see a huge effort has been made in defence of pollsters here, but little effort was made by pollsters to inform of the likelihood of such a discrepancy pre local elections. A case of they protest too much methinks.

  28. There is, however, one crucial difference between the 2011-2012 elections and pre-2010 local elections: the Lib Dems are no longer the beneficiary of the anti-government-and-main-opposition protest vote. So whilst an enhanced Lib Dem vote in local elections over opinion polls is not new, an enhanced Lib Dem vote in local elections WHILST IN GOVERNMENT is.

    My opinion on the accuracy of Lib Dem voting intention is that we simply don’t know. We know from 1992 that the Tory Shy Factor cause the polls to be horribly wrong. We could have a Lib Dem Shy Factor now, and there’s no knowing if the methodology designed to stop the Tory Shy Factor will work for a Lib Dem shy factor. Even testing the accuracy of polls for local elections isn’t foolproof; with eve-of-poll stats many more voters will have made up their minds who to vote for in a local election, and the Lib Dem Shy Factor is those polls may be greater or less than Westminster polls done at the same time.

    This all boils down to a fundamental problem that all polls except the few in a run-up to an election are hypothetical. When an opinion poll says “If there was a general election tomorrow …”, there’s no getting round the fact that there isn’t an election tomorrow. People don’t give that much thought in how to vote when there’s no election coming up. We could settle this once and for all if we created a parallel universe where YouGov takes a poll, a snap election tomorrow is called, and then YouGov comes back with a question along this lines “Actually, there is a general election tomorrow. So, how are you really going to vote?” But as YouGov doesn’t currently have the capability to create parallel universes, there’s not much we can do about that.

  29. A rule of thumb I tend to use is the Lib Dems had terrible ratings in the opinion polls from 1976 to 1979, and although there was a slump in support for them at the General Election, they still returned 14.1%.

    So limping along in high single digits or bordering 10/11 seemed to mean 14% in practice.

    16% in local elections for 2011 and 2012 is pretty much what I’d have expected for them.
    I suppose, for a General Election, whereas I wouldn’t expect a 7% differential out of much lower figures hitting the core
    I guess it could mean that in GE terms the current situation is even more serious than in 1979 – 12 or 13%?

  30. The Thrasher and Rawlings figures of 39-33-15 are interesting,
    because as in 2011, they are higher than the BBC figures,
    this time they have the Lib Dems 1% lower as opposed to 1% higher last year.

  31. I don’t think there is too much of a mystery here. To those who say that the Lib Dems outpolled their VI predictions in the locals, I’d point out that in 2009 the Lib Dems won 28% of the vote in the locals, while Labour won only 24%. So compared to that the Lib Dems are way down in local elections, and that is a direct effect of their being in government.

    Conversely, to those who say the Lib Dems are on 9% I’d say, this is almost fertainly an under estimate. Indeed wasn’t there a poll of Lib Dem seats recently that showed that when people are asked about VI, the Lib Dems do significantly worse than when people are asked to think about their own constituency before giving their VI? Further, those pollsters that redistribute some “don’t knows” to the party they previously voted for have consistently shown higher Lib Dem support.

    Finally I’d point out that a few months before the 1997 election Labour was polling some 20% above the Conservatives, and yet, very shortly later they won by about 10%.

    I think this has a lot to do with how people would like to vote versus how much FPTV constrains their choices.

    This fanmean that parties are often overestimated in their support, or underestimated in their lack of support.

    Suffice to say that Lib Dem support has certainly collapsed, but I’d be surprised if it dropped below 15-17%, which is, after all, more or less where it has been for the past 40 odd years, give or take the odd election.

    22/23%, which was their 2005/2010 support is exceptionally high for them.

    One final point, the Lib Dems have done exceptionally well in parliamentary representation since 1997, when they more than doubled their seats. For the Lib Dems to do well, the Tories must do badly, as the Tories are the main opposition in some 2/3 of Lib Dem seats. This is why the Lib Dems lost seats in 2010, even though their vote share rose, and their total votes also rose. It is also why their seat share increased in 1997, even though their vote share, and total vote decreased.