In the Sun this morning there is a new YouGov poll of Lib Dem seats and marginals. The poll is based on the provisional new boundaries, covering the 46 seats that the Lib Dems will notionally hold on the new boundaries, plus the 30 most winnable seats for them beyond that (this obviously includes many areas that currently have Lib Dem MPs on the old boundaries, but wouldn’t on new boundaries – such as Redcar, Sutton and Cheam, Yate and Manchester Withington)

Lib Dem MPs depend more on other parties on personal votes and tactical voting – for this reason I’m always wary about voting intention polls in Lib Dem seats and marginals, as they tend to pick up the national picture rather than local factors. For this reason the poll repeated something I’ve done before in the marginal polls for PoliticsHome and which Populus have done in some of their marginal polling for Lord Ashcroft. First YouGov asking the standard voting intention question, then asked people to think about their own constituency and the candidates likely to stand there, and asked them how they are likely to vote there at the next election.

As we’ve seen before, this makes a large difference in Lib Dem seats, boosting their support by nine points. This suggests the are still benefitting to a substantial degree from personal and tactical voting. They are just in such a bad starting position it doesn’t really help much!

Vote share at last election: CON 32%, LAB 19%, LDEM 41%
Vote share (normal question): CON 29%, LAB 34%, LDEM 15%
Vote share (constituency prompt): CON 28%, LAB 31%, LDEM 24%

This would suggest the Conservatives are down 4 points, Labour up 12 and the Liberal Democrats down 17. If these shares of the vote were repeated at the next general election and there was a uniform swing across this group of seats the Liberal Democrats would hold only 7 seats, with the Conservaties taking 21 of them and Labour 18. There was little significant difference between the position in Con -v- LD and Lab -v- LD battlegrounds.

One thing to note is that YouGov’s national polls show the Lib Dems down by about 15 points, while this poll, in their strongest seats where they will benefit most from MPs personal votes and tactical voting has them down by 17 points. This is not necessarily as strange as it might seem, it could just be a sign of a floor effect. At the last election there were 132 seats where the Lib Dems got less than 15% of the vote, by definition they cannot lose 15 percentage points in those seats, so to be down 15 points on average they must be losing more in some of their stronger seats. We saw a similar pattern in the Scottish Parliament elections in 2011 – the Lib Dems had a bigger than average drop in their support in all but one of the seats they held due to a floor effect of seats where they started off low.

Needless to say, polls measure opinion now, they don’t predict what will happen in three years time. The Lib Dems may recover, all those former Lib Dem voters saying “don’t know” may come home, as we get closer to an election people may become more aware of which parties can win in their own area or more willing to consider voting tactically for the Liberal Democrats. Right now, however, things really don’t look good.


40 Responses to “YouGov poll of Lib Dem seats”

  1. Is there a list of which 77 constituecnies were polled?

  2. Not sure some people are being honest about tactical voting.

    If you are a Labour voter in a constituency that only the Tories have a real chance of defeating the sitting LD candidate, I suspect that many will still vote LD. Why ? Well they would not want the Tory to win and atleast they have a chance of a coalition with the LD’s, if Labour failed to get a majority.

    I therefore suspect that the LD’s won’t lose as many seats as predicted and that the Tories won’t gain the most of those seats lost.

  3. LD wipeout at the next GE seems increasingly highly plausible.

    And it seems to me that if the boundaries changes go through many voters will not know how TV might affect the outcome in their constituency which I imagine will encourage Con and LD voters who in the past may have voted tactically for the LD candidate to instead vote for their own party.

    Annihilation beckons for the LDs.

  4. @ R Huckle

    That might have been true before 2010 but I am afraid many left inclined voters who might have voted as you suggest now see the liberals as little different than the tories so will not vote for them again.

  5. @Robert C (9.42 – previous thread)

    “I’m not sure how the supposed total meltdown in Lib Dem support across the country squares with the fact that so far this year we have got 27% of the vote in local government by-elections. I know AW will not like this, but it all hinges on what question you ask and I believe the difference between asking people how they would vote locally versus asking them generally how they would vote in a general election is significant.”

    Robert,
    It seems to me that you are contradicting yourself.

    Your second statement re the difference between local and general elections I would agree with. In fact, in 2010 canvassing as a LD member I was amazed at the number of people in Newcastle who confirmed they would be voting LD in the local election (we held the city council at the time) but would be voting Lab in the GE.

    As someone who has now resigned from the party due to tuition fees and support of the NHS bill, I am now in the same position, still giving practical support to the local party but will not give support and may not even vote LD in the next GE.

    Thus your apparent glee at 27% support in local by-elections is imo misguided re the Sun poll indicating wipeout in a GE. As the poll shows, the majority of respondents believe the national party has “sold its soul” and while this will have some knock-on effect locallly (we lost Newcastle Council in 2011) the impact will be significantly greater in GEs.

    I forsee the LDs having a reasonable (although reduced) presence locally but a wipeout for the next decade or so nationally. I do believe there will be improved fortune for the party longer term (assuming a return to centre left politics) as dissillusionment with the two main parties increases. I note that even with the demise of the LD VI(10% – the best result for a week or two) , Lab+Con is only 82% with a high number of “don’t knows” as well as “others”.

  6. Colin –

    Abingdon and Oxford North CC
    Argyll, Bute and Lochaber
    Ashfield CC
    Bath BC
    Bermondsey and Waterloo BC
    Berwick and Morpeth CC
    Berwickshire, Roxburgh and Selkirk
    Bideford and Bude CC
    Birmingham Yardley BC
    Bodmin and Newquay CC
    Bournemouth West BC
    Bristol North West BC
    Bristol West BC
    Caithness, Sutherland, Ross and Cromarty
    Cambridge BC
    Cardiff Central and Penarth
    Cardiff East
    Ceredigion and North Pembrokeshire
    Cheadle BC
    Cheltenham BC
    Chesterfield BC
    Colchester BC
    Croydon Central and St Helier BC
    Cupar and St Andrews
    Deeside and Gordon
    Derby West BC
    Eastbourne BC
    Eastleigh BC
    Edinburgh West
    Falmouth and Camborne CC
    Glastonbury and Wincanton CC
    Gower and Swansea West
    Guiseley and Yeadon CC
    Harrogate and Knaresborough CC
    Hazel Grove and Poynton CC
    Hereford CC
    Hexham CC
    Hornsey and Wood Green
    Inverness and Skye
    Kendal and Penrith CC
    Kingston and Surbiton
    Kingston upon Hull North BC
    Leeds North BC
    Manchester, Withington BC
    Newcastle upon Tyne Central BC
    Newton Abbot CC
    North Devon CC
    North East Somerset CC
    North Norfolk CC
    Northampton North BC
    Norwich South BC
    Oldham and Saddleworth CC
    Orkney and Shetland
    Redcar BC
    Richmond and Twickenham BC
    Rochdale North and Rawtenstall CC
    Rochdale South BC
    Sheffield South West BC
    Sheffield West and Penistone CC
    Solihull BC
    South Powys
    Southport BC
    St Albans CC
    St Ives CC
    Sutton and Cheam
    Taunton CC
    Teddington and Hanworth BC
    Thornbury and Filton CC
    Torbay BC
    Trowbridge CC
    Truro and St Austell CC
    Warrington South BC
    Weston-Super-Mare CC
    Willesden BC
    Yate CC
    Yeovil CC

  7. Robert C –

    “I know AW will not like this, but it all hinges on what question you ask and I believe the difference between asking people how they would vote locally versus asking them generally how they would vote in a general election is significant.”

    Anthony does like it, and agrees. For polling in marginal seats asking the normal voting intention question underestimates Lib Dem support, you need to ask a localised question. These figures though, are based on just such a question.

  8. All this is based on the false assumption that most voters think about party politics between elections, they dont. Most people actively dislike party politics & will avoid thinking about it till they have to, a few months before they intend to vote. For a quarter of the electorate, thats never; for another third its not for 3 years.
    The only voters actually thinking about party politics now are those intending to vote in 2 weeks, maybe 1 in 10 of the total electorate.

    Thats why the results in 2 weeks will bear very little relation to the VI polls.

  9. Thanks Anthony!

  10. I am not sure why such a fuss is being made here. The national polls show the LD’s down 15 points, this poll shows them down by 17 but the margin of error is + or – 3 points or so so this poll is within the margin of error for what the national polls show. Is it not?

  11. Paula, the fuss would be that the UNS projection doesn’t forecast LibDems retaining only 7 seats.
    This more specific poll does. That should certainly stir up a fuss, at least among the 46 MPs who stand to loose their seat.

  12. @ Anthony

    This is a really great poll for informing Tory & Labour about the current situation. If the Dems vote to approve the boundary changes, they’re probably toast.

    LDs who are placing their hopes on tactical votes may well be disappointed because the new boundaries will make tactical voting very difficult. And you’ve dealt with that to a certain extent by asking based on constituency.

    So thanks again, for a fascinating poll, great analysis & also for giving us the seat list.
    8-)

  13. What would be interesting would be to have a similar poll on the current boundaries. The boundary changes may be attenuating constituency effects by bringing in voters without a current LD MP, which would reduce personal voting, and by introducing uncertainty into tactical calculations.

  14. Anthony – it would be really helpful to know (separately) the vote share at the last election in the Con-Lib battlegrounds, and the vote share at the last election in the Lab-Lib battelgrounds.

    That would help in interpreting the cross-breaks. Thanks for anything you can provide!

  15. The [not so] Magnificent 7 :twisted:

  16. Tim –

    The 2010 shares in the Con v LD seats was CON 37%, LAB 13%, LD 42%

    In the Lab v LD seats it was CON 20%, LAB 33%, LDEM 37%

    Hence the change in the shares of the vote was:

    In CvLD C down 4, L up 11, LD down 17
    In LvLD C down 2, L up 12, LD down 16

    So the Lib Dems are performing much the same in both types of battleground

  17. I think tactical voting will probably reduce the impact of that, though, right? And not just on CvLD, in LvLD. Oldham & Saddleworth showed that Conservatives are willing to make tactical votes for the Lib Dems against Labour. I don’t think that’d show up because lots of people aren’t aware of the situation in their constituency until they get the traditional “it’s a two-horse race, and we’re one of the horses for once!” Lib Dem leaflets I used to deliver. So I’d imagine it’s not quite as bad as 7 seats of 600.t

  18. Don’t suppose each constutuency figures are available? Or even by county?

  19. Darren – no, and they’d be meaningless anyway due to small sample size.

  20. Thanks Anthony – that’s very helpful.

    Interesting that, despite the similar movements in both types of seat, the situation is different in each for third-party tactical voters: the Labour vote outweighs the Con-Lib vote in Lab-Lib battlegrounds, whereas the Conservative vote does not outweigh the Lab-Lib vote in Con-Lib battlegrounds…

  21. The net gains are very different, in the different types of seat, though.

    In CvLD: C -4 LD-17= C net gain wrt LD 13

    In LvLD: L+12 LD-16= L net gain wrt LD 28.

    In other news, google has started telling me how many times I’ve visited this site, which is slightly disconcerting.

  22. Does anyone have any information on the change in SNP vote in Scotland?

  23. Paula, you should not assume that just because the polls are within the margin of error, that means a statistical tie. The margin of error is usually a 95% confidence interval, which is what I assume it is here. That means that there is a 5% chance that the true figure is outside this range (either above or below) is 1 in 20, or 5%. But this range is not linear, the true figure is not equally likely to be in the middle to at the bottom end of the range, it is still far more likely to be in the, mddle, close to the mean, than it is to the bottom end of that range. If we fitted 90% confidence intervals, the margin of error would be smaller, and would likely not overlap, and in that case there would be a 5% chance that the true population average was above the margin of error, and a 5% chance it would be bellow the margin of error. (The 95% margin of error means a 5% chance of the true statistic being outside the range, but a 2.5% chance of it being above, and a 2.5% chance of it being bellow).

  24. Anthony – hope you don’t mind a local question, but I don’t see Lewes in the list (which will become 2 seats under the proposed boundary changes: Lewes and Brighton East, and Uckfield). Given Norman Baker had a pretty hefty majority in 2010 (52% to the Conservative’s 36% and Labour’s 5%), does this mean you feel he’ll not only lose his seat in the boundary changes, but also not be in a winnable position in either of the new constituencies? If it’s no longer going to be a close LD/Tory marginal in Lewes (where I live) that would certainly turn me into a swing voter as I weigh up my options.

  25. Alex F – I was surprised as well actually, when I noticed it was missing I went back and double checked in case I’d made an oversight.

    The Lewes & Brighton East has a notional Tory majority of about 7%, Uckfield of about 10%. I went up to Lib Dem targets that needed a swing of 3% or so.

  26. Sample size across all seats is around 20 then?

    I would not deny that a trend is there. But asking 1500 people across 77 seats? Now, if it were 1500 in each seat – that would be a poll.

  27. Anthony – Cheers for that. Given that Lewes itself is solidly left of centre (as seen in Town Council and District Council election results), but only 5% voted for Labour in 2010, I think it’s reasonably safe to say there’s a lot of tactical voting going on here. Do you know if under the new boundaries Lewes and Brighton East would still be a Con-LD marginal, or will it be a Labour target as well? Interesting times ahead. Assuming the boundary changes go through as planned. Given there’s no obvious social/cultural reason to lump Lewes, the county town of East Sussex, in with Brighton suburbs, surely the LDs will fight hard to keep one of their safer seats intact.

  28. Cogload, I think that’s a pretty decent sample size. Look at it like this, usually a poll might be 1,000 people or so from across the entire country of what? 40 million or so registered voters? The variability there is pretty high. This is a sample of 1,500 from a kuch smaller and almost certainly more homogeneous population, what is the number of registered voters in 76 seats? Roughly 5 million? So a much smaller and more homogeneous sample than a typical poll. In that sense we might expect there to be a far smaller error than for a typical national poll. These seats all share something in common, a high Lib Dem support base, that most seats don’t share.

    You are right that making predictilns about any given seat is dubious based on the data collected from each seperate constituency, but as a measure of Lib dem support in Lib Dem seats, it should be taken seriously.

    The only way to make a comparison was if there was a similar poll of Lib Dem seats taken, say, in 2007, that we could compare to. As it is, all we can do is compare national polls now, with national polls in 2007, or 2003, to estimate how the Lib Dems are doing now, relative to how they are usually doing two years after a general election.

  29. Fair enough – but Ashcroft style polling shows slightly different results.

    I haven’t looked at the tables in great detail – did they weight by newspaper readership?

  30. A very interesting poll. I wish You Gov would do more of these instead of basic voting intention 5 times a week!

    In my humble opinion there are only two issues deciding the next election. As ever the economy is massively the number 1 issue. The other is where the 2010 Lib Dem vote ends up.

    I can’t see any way that in Lib Dem/Lab areas the vote will go anywhere other than Labour in the percentages we have already seen in polling- almost regardless of the economy. The vote in Lib Dem/Tory areas is much less clear.

    I’d have been happier with a more leading question- however wrong it might be- of ‘given you have a Lib Dem MP and Tories are in second place how would you vote’- it might be too leading but that will be what the Lib Dems will be presenting at the next election.

  31. This should silence those that tell is that “the centre-left vote has reunited and will now beat the Cons”.

    Actually the collapse in the LD vote (even though most of it has gone to Lab) would actually result in the Cons netting three more seats.

    Thanks for providing this proof!

  32. Thanks for all this Anthony. Out of interest, what are the 7 held seats?

  33. Playing around with Electoral Calculus, my guess would be: Orkney and Shetland (Alistair Carmichael), Ceredigion and Pembrokeshire North (Mark Williams), Colchester (Bob Russell), Bath (Don Foster), Norfolk North (Norman Lamb), Yeovil (David Laws) and Kingston and Surbiton (Ed Davey).

    The same figures, on the current boundaries, give the above, or their predecessors, along with: Ross, Skye and Lochaber (Charles Kennedy), Sheffield Hallam (Nick Clegg), Twickenham (Vince Cable), Southport (John Pugh), Thornbury and Yate (Steve Webb), Lewes (Norman Baker), Westmorland and Lonsdale (Tim Farron) and Portsmouth South (Mike Hancock).

  34. Paul Barker

    Would you post some of the assertions you make on LDV regarding your view that LD will be at around 30% in the LE

    Perhaps you can also share your wisdom on the use of local by-elections to challenge the national polls. As this is a polling-focused site I am sure you will find some responses

    You only have to look at the demographics of the seats fought to see why the LD clutching at straws is a nonsense

    As I have said befor the Tories are the real threat to LD but there seems to be a headless chicken response with consistent policies and rantings that will alienate tactical voters in the south and strategic voters in the north who come from the left

  35. Adrian B

    “This should silence those that tell is that “the centre-left vote has reunited and will now beat the Cons”.Actually the collapse in the LD vote (even though most of it has gone to Lab) would actually result in the Cons netting three more seats.”

    Well no, whereas it is true that in these seats, where the Lib Dems have a very high and unrepresentative turnout, the Tories do marginally better than Labour, this is because of the current 57 Lib Dem seats, some 66% havt the Tories in second place, so for Labour to win these, they must overtake both the Tories and the Lib Dems, I haven’t checked the situation in the 30 other seats, but would not be surprised if more were Tory held than Labour held. In fact, of the current Tory seats, more than half have Lib Dems in second place, so the Tories main opposition is the Lib Dems, and vice versa.

    But with so many Lib Dem votes goint to Labour, this will also have a big effect on Tory/Labour marginals, which is not modelled here. So the Tories may well pick up 3 more Lib Dem seats than Labour do, but Labour will take a hefty slice directly off the Tories if current polling is to be believed.

    The Tories and Labour are, after all, not only fighting for Lib Dem seats.

  36. “In my humble opinion there are only two issues deciding the next election. As ever the economy is massively the number 1 issue. The other is where the 2010 Lib Dem vote ends up [in LD/Con areas]”

    Agreed that in LD/Lab areas, the LD vote is in meltdown, and may not come back in time for the general election.

    In LD/Con areas, the question is whether the local LDs can persuade enough of the 2010 tactical vote to stay LD or whether they will switch to Labour and hand the seats to the Tories.

    But you miss the critical third question: how many seats will the SNP take in Scotland? If there are 40+ SNP MPs and Labour lose thirty-odd seats, then there could well be a hung parliament even if Labour do very well in England.

  37. Richard Gadsden

    ” In LD/Con areas, the question is whether the local LDs can persuade enough of the 2010 tactical vote to stay LD or whether they will switch to Labour and hand the seats to the Tories.But you miss the critical third question: how many seats will the SNP take in Scotland? If there are 40+ SNP MPs and Labour lose thirty-odd seats, then there could well be a hung parliament even if Labour do very well in England.”

    I disagree, the crucial question is how many Tory seats Labour will gain from Lib Dem voters switching to Labour in Tory/Labour marginals.

    Labour won a majority of English seats in 1997, though I’m not sure about 2001.

    A significant Lib Dem movement will hurt the Tories, especially if the Tories poll less than 34%.

  38. On tactical voting – the Aberdeen South Scottish parliament constituency has been Lib Dem since it started, the local MSP was 9.2% ahead of the SNP in 2007. In 2011 (some boundary changes) I knew people who voted Lib Dem tactically to keep the SNP out, but in the election the Lib Dems came 3rd behind Labour. People may think its a 2 horse race when it isn’t.

    The (almost) equivalent Westminster seat has been comfortably Labour since 1997 – its quite common for different parties to be elected to different parliaments, so one cannot assume the SNP will win a lot of seats from Labour in a GE.

  39. @Jimmy

    I was just going to make the same point about the SNP in Scotland. In the 2007 Scottish election Labour won 32% of constituency votes, in 2011 they lost only 0.45% putting them on 31.5%. Labour did not lose vote share (though more in list votes). But the Lib Dems went to the SNP in Scotland in 2011, Lib Dems down 8%, SNP up 12% (clearly they picked up vote share elsewhere). So there was no collapse in Labour support between 2007 and 2011.

    More importantly, in 2010 in Scotland Labour won 42% and the SNP 20%, so as the evidence stands Scottish voters are happy to endorse the SNP in the national parliament, and Labour in Westminster.

    The question remains, will those disgruntles Lib Dems stick with the SNP in 2015, or switch to Labour. My hunch is that they will switch to Labour as a more viable opposition to the coalition. But we shall see.

  40. Alun Parsons “So there was no collapse in Labour support between 2007 and 2011″.

    This was the story put out by Labour immediately after their defeat. The truth was different as per later academic studies. Yes, many Lib Dem former voters switched to SNP but a significvant percentage switched to Labour offestting (by chance) almost identical direct switching of former Labour voters to the SNP.

    The Labour vote in Scotland contains a large element of anti-Tory rather than pro-Labour votes. They cannot be relied upon to stick with Labour if the SNP remain rampant-although the perversion of FPTP for Westminster elections means that even if the SNP substantially outpoll Labour, the Labour party may still have more Scottish seats.