We have three new polls so far today. TNS have put out a new GB poll, which has topline figures of CON 33%(nc), LAB 32%(nc), LDEM 8%(+1), UKIP 16%(-1), GRN 5%(+1) – clearly no significant change since their previous poll (tabs are here).

ComRes have a new poll of the 40 Labour held constituencies in Scotland (that is excluding Falkirk, where Eric Joyce sat out his term as an independent). In 2010 the share of vote in these seats was CON 14%, LAB 51%, SNP 19%, LDEM 14%. The new ComRes poll found support standing at CON 13%(-1), LAB 37%(-14), SNP 43%(+24), LDEM 2%(-12). The seven point SNP lead represents a swing of 19 points from Lab to SNP, the equivalent of a sixteen point SNP lead in a national Scottish poll (tabs are here).

Finally YouGov have a new London poll for the Evening Standard, which has topline figures of CON 34%(+2), LAB 45%(+1), LDEM 8%(+1), UKIP 8%(-2), GRN 4%(-1) – changes are from YouGov’s previous London poll a month ago. The eleven point Labour lead represents a swing of 4.5 points from Con to Lab since the general election, the equivalent of a two point Labour lead in a national GB poll (tabs are here).


444 Responses to “ComRes in Scotland, YouGov in London and latest TNS poll”

1 2 3 9
  1. First?

  2. But I was intending a post more relevant to the last thread.
    I wll pay more attention to an election prediction which says something like:
    Labour L seats, Conservatives C seats, turnout T%.

  3. For those who haven’t seen the previous thread, the ITV/ComRes poll gives crossbreaks by the size of the 2010 Lab lead.

    I’ve done some analysis of these, which you can see on that thread.

  4. Thanks to Oldnat for the analysis, which required some work on his/her part.

  5. oldnat, others,

    what would the comres poll suggest about the overall number of seats for the different parties in scoltland?
    Sorry if youve already done this on previous thread.
    best andyo

  6. @Peter Crawford

    See last thread for disagreement with your views on the accuracy of projections coming from the Electoral Calculus model.

  7. i.e., if the comres figures were repeated in the GE.

  8. @andyo

    Prof. Rallings suggested 12 Labour holds out of the 40.

    Using nothing stronger than guesswork for the other 19 seats, that would give:

    SNP 43 (per above + Falkirk + 6 holds + 8 LD)
    Lab 12 (per above)
    Con 2 (1 hold + Berwickshire)
    LD 2 (Orkney/Shetland + Charles Kennedy)

  9. @James
    Thankyou very much sir.
    Could be better for Labour (could be worse!)
    best, andyo

  10. JAMES.
    Thanks for this.
    Is this a glimmer of light for Labour? (Like the sunset I have just seen here in Bournemouth on the beach).
    If Labour can knock their losses to 25 in Scotland and pick up 45 seats from Con and LD in E and W they are on 277.

  11. @Unicorn

    The BES predictors offer up very similar seat totals to EF. In fact they hedge their bets quite a bit but ultimately range from a Con/Lab dead heat to Con ahead by 10-15 seats.

  12. @ chrislane1945,

    That’s sort of what I thought — hence my ‘could be worse’ comment!

  13. Good evening all from a windy and cold Giffnock.

    ” The seven point SNP lead represents a swing of 19 points from Lab to SNP, the equivalent of a sixteen point SNP lead in a national Scottish poll”
    _______

    I would had thought this sort of poll would underestimate the SNP’s VI (across Scotland) because Labour are going to poll better in the seats they currently hold than in other seats across Scotland.

    So I reckon the true SNP lead will be around 20% when factoring in the VI in seats they currently hold and in the Lib/Dem seats where the SNP VI will be much higher than in the “current” Labour held seats.

    And the Lib/Dem shift to Labour in Scotland is somewhat hiding the true loss of Labour supporters to the SNP.

  14. Tactical ? Doubtless many would wish to vote SNP-IF it was going to assist independance so could these 40 seats be saying , whatever Labour might / might t not do : Damn the Tories. The older people’s VI is interesting. maybe the young are less pro Labour but will have been largely from generations of Labour families.

    Interesting ??

  15. ProfHoward

    ITV Scottish poll: has anyone run this swing through the Labour seats to see how many are lost?

    But that’s the interesting thing – the swing isn’t uniform. If ComRes split the 40 seats[1] into three groups depending on how far ahead Labour were in front of the SNP in 2010. If you look at the tables:

    http://www.comres.co.uk/wp-content/uploads/2015/03/ITV-News_Scottish-Labour-Constituencies-Poll_31st-March-2015.pdf#page=13

    the headline VI results[2] for each set are revealing:

    2010 Lead under 30pts (19 seats):

    SNP 43%

    Lab 35%

    Con 14%

    2010 Lead 30-40pts (11 seats):

    SNP 39%

    Lab 39%

    Con 13%

    2010 Lead over 40pts (10 seats):

    SNP 45%

    Lab 39%

    Con 10%

    Obviously if the swing was uniform the Labour VI would be higher and the SNP lower in each successive group.

    If you applied the swing separately for each group, the result may be that the SNP win everything. Certainly it does hint that Labour may have a better chance of retaining the seats in the middle group. But if you look at these seats:

    Edinburgh South West
    Glasgow South
    Glasgow North
    Cumbernauld, Kilsyth and Kirki
    Dumfries and Galloway
    Glasgow Central
    Airdrie and Shotts
    Paisley and Renfrewshire North
    Dunfermline and Fife West
    Glasgow East
    Inverclyde
    Glasgow North West

    Ashcroft polled many of them and found healthy SNP leads. So it could be that the fact that SNP and Labour are ‘only’ tied in this group is just sampling.

    [1] With a total sample of a thousand that means that the number polled in any one seat was only about 25. Can we please have no more calls for a breakdown for each constituency.

    [2] Part of Comres’s standard squeeze questioning is to ask what Party people “Generally speaking think of themselves as” and Labour is still ahead in the whole sample here (33%) ahead of SNP (31%), despite the SNP’s big lead on voting. As this answer is used to allocate undecided voters this would help Labour a little. It might also give them some hope for the future, although a lot of people could be answering out of habit and may not return.

    Interestingly only one voter who thought of themselves as Green said they would vote SNP.

  16. To put the gap between SNP and Lab into context –

    On average, in Labour’s 10 most heartland seats (where the 2010 Lab lead over SNP was greater than 40%), the SNP lead over Lab is now 6%.

    In their 18 “weakest” seats (where the 2010 Lab lead over SNP was between 3 and 29%), the SNP lead over Lab is now 8%.

    In the middle group of 12 seats (where the 2010 Lab lead over SNP was between 30 and 40%), SNP and Lab are both on 39% – but the Greens are showing a 7% VI, which is likely to be concentrated in only some of them.

  17. @Andy Shadrack

    For some time you have been expressing concern about distortions that may arise from allocating “don’t know” and “won’t say” responses on the basis of 2010 voting shares.

    Given this, you might be interested to look at the recent set of projections by the British Election Study. These include different projections assuming (a) that D/Ks are handled in the normal way and (b) when they are just left out of the calculations.

    From a quick look, it seems as if this doesn’t make a great deal of difference and the gainers and losers are not the parties you have been highlighting in most of your comments. However, you may take a different perspective on this report.

  18. I still reckon we are heading for Con having the most seats but well short of a majority and unable to form a coalition.
    Lab will do better in Scotland than current predictions but still pick up less seats than Con but able to form some kind of patchwork coalition. I cannot see anything other than turbulent times ahead.

  19. @chrislane1945

    It’s at the “better” end for Labour of the bulk of the Scottish polls, which have had SNP leads between 16 and 20 points. There have been a few outside that range (e.g. Ipsos Mori), but about 2/3 of them have been in that group.

    It’s hard to tell if it’s a genuine “improvement” for Labour or not, as ComRes haven’t done a Scottish poll before. Labour are claiming it is an improvement. I think they are comparing it to the Ashcroft polls, which showed 20+% swings, but then he has concentrated mostly on the “safer” (based on 2010) Labour seats.

    The one health warning I would attach (picked up by someone else on the last thread) is that the recall of 2010 vote lumped SNP under “some other party”, which might mean the weighting is wrong.

  20. @oldnat

    Which seats do you think the Greens are doing well in? I would have thought Edinburgh East, but it’s in the “less than 30%” group.

    Of the “between” group (the one where Greens are scoring well), I don’t see too many where they would obviously do well. Glasgow North, maybe?

  21. Roger Mexico

    “Interestingly only one voter who thought of themselves as Green said they would vote SNP.”

    Not surprising though. I suspect that most of the growing Green vote in Scotland has already decided to vote SNP now, and Green in 2016.

    Those who have a Green VI for May are unlikely to change now.

    However, that they are concentrated in the middle group of seats, suggests that their vote is concentrated in all or some of –

    Dumfries and Galloway
    Dunfermline and West Fife
    Edinburgh South West
    Glasgow South
    Glasgow North
    Glasgow Central
    Glasgow East
    Glasgow North West
    Paisley and Renfrewshire North

    where they have candidates.

    Pat Harvie’s popularity in his own territory of Glasgow may be playing a role.

  22. Election Forecast prediction updated today. Changes are from yesterday. 1 week ago and 15th March (when I started recording these).

    Con 285 +3 +1 -1
    Lab 279 -2 +2 +6
    SNP 37 +1 -3 -5
    LibDem 26 -1 = =
    Plaid 2 -1 = =

    Tories have restored the lead they had yesterday. The EF model seems to be quite sensitive to individual polls which don’t seem to be very exciting. Also, over the last couple of weeks, Labour seem to be creeping back in Scotland.

  23. @RAF


    The BES predictors offer up very similar seat totals to EF. In fact they hedge their bets quite a bit but ultimately range from a Con/Lab dead heat to Con ahead by 10-15 seats.

    Yes – if the matter were to be decided by a democracy of modellers it would all be settled by now. (Consensus: Tories a little ahead, but not enough to form the next government).

    But if we want to try to be a bit ahead of the curve in predicting the actual outcome, we have to take account of the strengths and weaknesses of the various models (though admittedly this is not always all that evident with the recent proliferation of somewhat obscure models).

    For example, two of the BES models are somewhat more Tory favouring than the other two. Which is more plausible?

  24. James

    See my response to Roger.

    We don’t have enough data to know where the Green vote is strongest, but anecdotally (and purely as an example) my daughter’s street in Glasgow South was fairly solid Yes at the referendum, judging by the spread of window posters – but most of them were Green Yes ones.

  25. PETE B and JAMES.
    Thank you for these posts, and I agree with both.

    MIKEY.
    Yes, it looks like Cons may have small seat-lead over Lab, unless there is a swing back to Lib Dems in seats they hold, where the Tories are challenging in second place.

  26. @Pete B

    Election Forecast prediction updated today.

    Yes – but how plausible is it , for example, that the LibDem VI will rise to 13.3% in a little over five weeks, especially given that it has been stuck at close to 8% for around five months now.

    I have always thought that there Nowcast was more useful than their forecast.

  27. Unicorn
    Yes, they have LibDem at 13.3% and UKIP at 9.7%. I think the reverse is more likely.

    I agree with you about the Nowcast, but I can’t seem to find it any more. They seem to have changed the site layout a bit.

  28. @unicorn

    What surprises me is they’re expecting like 5% swingback to the Lib Dems, like you said, while saying the Conservatives are going to stay put on ~34%. It doesn’t make much sense.

  29. As EF are predicting 26 LibDem seats on 13.3% of the vote, how many do people think they will get if their vote is (say) 9%?

    I’ve been saying for ages that I don’t think they’ll get more than 12. Any other ideas?

  30. @Pete B

    I agree with you about the Nowcast, but I can’t seem to find it any more. They seem to have changed the site layout a bit.

    For some reason some weeks ago they stopped publishing the seat-by-seat probabilities of winning (for current polling figures) and their seat projections for what would happen in an election held today. Fortunately they still publish their VI profiles for each of the seats (based on up to date polling data).

    Anyway,mthe teo should converge quite rapidly over the next few weeks.

  31. Re use of 2010 vote as a weighting factor in Scotland

    Roger (I think) suggested that you can’t use 2011 vote as a weighting basis because the boundaries are different.

    Technically, you could because the postcode is known for each respondent to ensure they are in the correct constituency [1].

    However, I’m not sure that it wouldn’t just introduce another level of error in constituency polling, since people may have changed address since 2011.

    The only past vote/party ID weighting I’m really happy to trust is where an online pollster gathers the information near the time of an election – and update it.

    [1] I discovered when cleaning up the pupil database for a local authority, the wrong postcodes are often given – especially in areas of high multiple deprivation.

  32. Omni

    For some reason (time lag?) Con seem significantly behind Lab on the current polling. I think their factor to stabilise the VI is doing something odd. A few weeks ago we saw a crossover which then went away as if it never existed.

    Getting up to 34% is from a supposed current position of a bit below that. It worries me their prediction on polling day will suffer this artifact which could wreck their final prediction.

  33. ON
    Can you explain what ‘high multiple deprivation’ means? Is it something like poverty, lack of government services and no food banks?

  34. Mikey
    Agree, it’s been my opinion for a while that as counterintuitive as it sounds the specific seat numbers don’t really matter. Labour just need to make sure they have a similar number of seats to the Tories then simply due to the number of anti Tory MP’s and a lack of Tory allies Labour will be in the best posistion to form a government even if they are behind.

  35. Although I am not a UKIP voter I do believe that the extensive coverage given to Nigel Farage over the next 5 weeks will result in an increase in the UKIP vote at the expense of the Tories. Farage has not had a look in on TV since his two by- election victories last year. He is a very good TV performer, better than all the other leaders, and his direct attack on the Tories immigration policy (as seen in UKIPs poster unveiled today) will resonate with many undecided Tories.

  36. @alan

    Yes I agree, something seems wrong. They must have noticed though. The seats tab is updated daily, but maybe they just haven’t bothered/forgot to update the VI tab? The VI numbers look like they would have made sense a few weeks ago, but not now.

  37. Pete B

    There are lots of kinds of “deprivation” – most, but not all economic.

    The Scottish Index of Multiple Deprivation is probably not much different from that used elsewhere, but the factors taken into consideration, and the weighting they are given, can be seen here –

    http://www.gov.scot/Topics/Statistics/SIMD/SIMDindicatorshandout

  38. @Omnishambles

    What surprises me is they’re expecting like 5% swingback to the Lib Dems, like you said, while saying the Conservatives are going to stay put on ~34%. It doesn’t make much sense.

    The difference between the LibDem and Tory adjustments is exactly what you would expect from the reversion-to-mean hypothesis. In a crude form this states that a party’s vote share will go back to about half-way between the current VI and its vote share in the previous GE.

    For the Tories the current VI is about 34% and the 2010 vote share was 36%-37% (depending upon whether you use UK or GB figures). So the pure reversion projection is for a share of a little more than 35%. (The slightly lower projection can be accounted for by the fact that the weight assigned to the reversion adjustment is now being steadily lowered.)

    For the LDs the two figures are 8% (current VI) and 23% (2010 vote share). So, here the pure reversion projection is 8 + (23 minus 8)/2 = 15.5%. Again, the actual projection is now being lowered a bit.

    But as I say the difference between the scales of adjustment makes complete sense. (If you believe the hypothesis)

  39. Tom

    Yes, I have noticed a sudden increase in Nigel on our TV News, and I am sure that will benefit UKIP.

  40. All the Scottish polls are pointing towards an SNP landslide if taken at face value. The big question is will this seismic shift hold.

    IMO the key to how many seats they actually capture is going to be the relative turnout of the Lab diehards vs the new (ex Lab) Nats. The conventional view is that swingback and incumbency effects should soften the blow for LiS and instead of wipeout they might get away with a just a very bad night.

    I’d suggest that a counter view is that post referendum the SNP converts will be very keen to given Lab a bloody nose and turnout in droves, while the diehards who haven’t a great turnout record will at best live up to their historic norms.

    My guess is that there is a strong probability that SNP will outperform the opinion polls when it comes to real votes.

  41. Thanks ON. Interesting.

    It would be very interesting to calculate some deprivation index by constituency across the UK, and correlate to which party held the seat. There might be some surprises. For instance in England, Tories tend to hold countryside seats which are often assumed to be wealthier than inner cities. However, rural unemployment is often quite high, and I notice that one of the measures of deprivation is drive time to various services. This would be likely to be higher in the countrryside.

  42. @unicorn

    Thanks – that makes more sense. If you’re checking the numbers regularly, how often are those VI numbers changed? Because I could have sworn it was 13.3% for the LDs about a week ago.

    You’d think they would update it daily with the seat updates, wouldn’t you

  43. Can anyone explain how weighting can work. When you weight a ‘Some other party’ group to the 2010 %. When ‘some other party’ includes SNP, UKIP, Greens. How do they assign the weighted voters between the various parties.

  44. Omni

    I think their prediction is based on a current position of Con 32% Lab 34% with a 2% transfer predicted. If we bump it up to 34 34 we’d likely see a stronger con position predicted (but not the full 2% due to the way autoregression works) a few weeks ago it seemed sensible with a slight Tory crossover after a few good polls, then it disappeared from the history and has been quite resistant to closing despite polls which have been very level.

  45. Labour’s task in Scotland is a salvage job with damage limitation the order of the day. Some polls have suggested an almost total wipeout, but this election is going to be so tight in the rest of the UK then if Labour can hang on to a dozen or so seats in Scotland, it could make all the difference between them being the largest party or not.

    I don’t entirely buy this argument that the largest party factor is unimportant. My view is that Labour need to nudge ahead of the Tories in terms of seats to form a sustainable administration after May 7th. All bets off for me if the Tories emerge as the largest party.

  46. EXILE In YORKS

    In the referendum we seemed to get a high turnout from both sides, so one possibility is that the turnout will be up for both sides but might cancel out.

    However, the high level of SNP membership suggests that there may be a lot of activists willing to knock on doors and GTVO.

  47. @ExileinYorks

    The most likely scenario at this stage is that the SNP will poll around 20% higher than Labour in Scotland. Various Scotland polls have said this now so it’s very unlikely to change in any material way. It’s also likely for the same reasons that Labour will lose around 30 seats to the SNP.

    I think we’ll see quite soon Labour shifting their resources to areas they need to win, such as the Con-Lab English marginals (and the 10-15 Scottish seats they can win). As @Couper said months ago, Labour should probably forget most of Scotland. It’s gone (for this election).

  48. Peter Kellner is reported tonight as calling a Minority Tory Govt as the most likely outcome.

  49. @James

    Glasgow North and its Holyrood equivalent Kelvin has been the Green’s strongest area in the west for some time.

    The other 2 strongest seats for Green are probably Edinburgh South and Edinburgh East but neither is in that middle group.

    It could be a general Patrick Harvie boost in Glasgow seats.

  50. Pete B

    All that data is available here by datazones (aggregations of postcodes), council areas, Holyrood constituencies, Westminster constituencies and various other planning areas.

    It’s an essential tool for strategic planning. I would be amazed if other parts of the UK didn’t have something similar.

    In Scotland the data is accessible via Scottish Neighbourhood Statistics.

    http://www.sns.gov.uk/Reports/AreaProfile.aspx

1 2 3 9