Saturday, December 7, 2019

YouGov independence polling

You may have seen that there are polling numbers from YouGov that purport to show a sudden and statistically significant drop in support for independence - down from 49% to 44%. Now, the first thing to say (and any serious analyst of any political complexion will tell you the same) is that we need to exercise a touch of scepticism about a single poll that shows a big and unexpected change. All polls have a margin of error, but statistically it's inevitable that a small proportion of polls will fall outside the margin of error - ie. they'll be rogue polls. It's possible that's what's happened here. Another possibility is that there may have been a genuine drop in support but that YouGov are overstating it due to the normal margin of error - a reported Yes vote of 44% could mean that the true figure is as high as 46% or 47%.

But of course a third possibility is that the poll is accurate. If so, it's surprising, because two previous Scottish polls during this campaign showed support for independence holding steady at a historically high level. The Ipsos-Mori poll conducted between the 19th and 25th of November showed a 50-50 Yes-No split, so if something has changed it's happened very rapidly. But that's not totally inconceivable, and there are precedents for this sort of thing happening in previous UK-wide campaigns. Remember that the media coverage of general elections compels voters to view political issues through a Britain-wide lens - they're bombarded with political parties making pie in the sky promises to transform Britain for the better. If and when Boris Johnson is returned with a working majority and gets on with delivering a Hard Brexit, that effect is likely to wear off pretty rapidly.

The other factor is that, during general election campaigns, media discussion of independence is heavily slanted in favour of unionist spokespeople. Scottish coverage generally sees the SNP outnumbered 3-1, and at UK level the ratio is probably more like 8-1 or higher - indeed it's not unusual for independence to be discussed without any pro-independence voice being present at all, as happened in last night's rigged BBC leaders' debate. It's the polar opposite of what happens in a referendum campaign, where there has to be parity between both sides. So if the (relative) balance of an indyref campaign causes Yes support to rise, as it did in 2014, it's not unreasonable to suppose that the hopeless imbalance of a general election campaign might cause Yes support to fall, and that any such fall may be only temporary.

Bear in mind that even if this is a genuine drop, it's not as if the floor has caved in - it just takes us back to where we were a year or so ago, before the big Yes surge. Until recently YouGov's normal range for Yes was 43-45%, so 44% is bang in the middle of that. I'm not too concerned about it unless there's a corresponding drop in SNP support.  There's contradictory information about what the fieldwork dates for this poll were - one article seems to suggest it's the same poll that showed the SNP with a huge 16-point lead over the Tories, and if that's true there'd be nothing much to worry about.  However, Britain Elects are saying the fieldwork dates were the 3rd to the 6th of this month, which would mean that it's a different and newer poll, and there might still be a risk of an SNP decline.  We may find out more tonight, because there's usually a flurry of polls on the final Saturday night of a campaign (although whether any of them will be Scottish polls remains to be seen).

*  *  *

I have three new constituency previews in today's edition of The National - this time it's Edinburgh North & Leith, Dunfermline & West Fife and West Dunbartonshire

32 comments:

  1. Cannae see it being accurate in the current political climate.

    ReplyDelete
  2. Yougov Tory bias!

    I take any poll with a pinch of salt these days

    50 seats for the SNP come Friday!!

    ReplyDelete
  3. As we move into the last week, who is left to decide? We'll, tactical voters for one
    ....

    ReplyDelete
  4. A key argument must be to persuade Labour and Lib Dems that the bigger risk is an immediate irreversible Brexit and 5 years of Johnson, than letting SNP in, where there is plenty of time to stave off an IndyRef another day.

    ReplyDelete
  5. I completed a panelbase poll a few days ago. Their polls are usually commissioned by the Sunday Times so, tomorrow? On the YouGov Poll, doesn't it seem odd that they are publicising the Independence question before the GE voting intention figures?

    ReplyDelete
    Replies
    1. The Times did that with the poll this week; teasing the approval ratings before publishing the actual voting intentions.

      Delete
  6. worst case---the poll is accurate. Just as the Lib Dumbs get squeezed in England by the Brit Nat media (take a bow, BEEB), so the SNP get a peripheral role in Scotland by this very British slanted process. As long as they win a few seats they will be doing fine.
    Then agitate for Indyref2, insist on it, demonstrate for it, make a "nuisance" as the French do. Block traffic in London--get onto any stage where Boris is--burst into BEEB studios--superglue ourselves to any passing Tory bigwig etc etc etc

    ReplyDelete
    Replies
    1. I think a #AUOB in London would be worth considering? It might get some attention, more attention that the marches ignored buy broadcasters in Scotland.

      Delete
  7. Poll shows 44% Yes and 44% SNP, with the Green's on 1%. SNP has been trailing yes lately by quite a bit and purportedly, 25%, at least, Labour voters are planning to vote Yes, so a little strange.

    For me, either vote has broken on yes/no lines, which given Brexit and everything else, unlikely, or something just isn't right about these figures.

    It could also mean SNP is understated, if there are not enough Yes supporters in the sample and the other polls showing a 50/50 indy split are right.

    Let's hope it's an anomaly.

    ReplyDelete
    Replies
    1. I clearly didn't read last part of the post, but if it is the same poll showing SNP on 44% then the above may apply.

      Delete
  8. As I said in my posts on the previous thread, unless SNP share has suddenly dropped like a stone, all the evidence says its a statistical anomaly, and one which is possibly bad news for unionists.

    I'm kinda surprised if it was new fieldwork. Why do an extensive poll in Scotland, ask VI for Holyrood and Westminster, but not do the indy question? Then, do another poll just days later all over again, even though your previous one agreed with all the others in that not a lot was changing?

    If this is the same yougov poll and/or SNP is still on 44% or so, then unionists should be more nervous than indy folks probably. If the SNP were getting 44% Westminster on a 44% (ex DK) Yes base, then they could be riding much higher for Dec 12th if this poll is low Yes outlier.

    We'll need to see the tables.

    All that aside, it's only Westminster VI that actually matters right now given there isn't an indyref in the next few weeks.

    ReplyDelete
  9. I don't think it helps that Stu Campbell is stating that Nicola Sturgeon should resign before the GE and that newspapers like the Courier have picked up on it. That man is an absolute liability to the Independence cause. I've even got to the point that I'm wondering if he's been locked up in the Tower of London and that Gove has replaced him in Bath.

    ReplyDelete
    Replies
    1. There is some nobility in being a lone wolf or loose canon but when you start damaging your own cause, you've gone too far. By own cause I mean not just independence, but last time I looked people were exiting the site pronto.


      Delete
    2. Whilst I could not be called a fan, I have read what he had to say. Recently I have seen a lot of anti-independence stuff. So, I am leaving pronto, for that is not what I think nor what I think most supporters of independence think. It appears different, somehow.

      I cannot guess what has happened to him, but it is not good.

      Delete
  10. James, first of all we need to remember that irrespective of the statistical issues, this is the kind of news the Unionist media are wanting/ waiting for and obviously will go large on it. In short the political issues exceed any statistical/ technical issues.
    Secondly, if we assume waking up on Friday morning to discover that Johnson has won his majority and that horrors like Brexit are about to become reality and not just a talking point, what price another similar poll showing something like the reverse?

    ReplyDelete
  11. Labour's first 36% I believe.

    Tories need 43%+ to be sure of a majority. As soon as they drop to 42% and lower, they enter hung parliament territory.

    Westminster voting intention:

    CON: 42% (-)
    LAB: 36% (+4)
    LDEM: 11% (-1)
    BREX: 4% (+1)
    GRN: 2% (-1)

    via @SavantaComRes, 02 - 05 Dec
    Chgs. w/ 03 Dec

    ReplyDelete
    Replies
    1. Someone said the other day that the Tories have eaten up all the Brexity votes they're going to get, and nowhere else to go.

      You'd think they already have the NO TO INDYREF 2 vote on board?

      Delete
    2. Probably down to extent of tactical voting now?

      Delete
    3. Everything depends on tactical. I think Labour in England are gaining here and maybe a bit faster than polls suggest. If they are, we'll see some other straws in the wind like a 38% in the next few days.

      Running averages are just no use if things are changing rapidly. People are often plotting polls based on date of release not date of earliest fieldwork (bulk will be done in the first day or two), making them out of date by a couple of days. This lag is made worse by the fact that older polls are still being included.

      The lag is really obvious in e.g. the BE tracker, and that's before the fact it's plotted by end of fieldwork date, which means there's further lag.

      Then there is the problem with averaging. There is no guarantee that makes things more accurate. It can be skewed badly. Better look at the range and point clustering for patterns. You are looking at a dart board and sometimes a lot of the darts hit too low, which a couple hit too high. The average is then actually far away from the actual number, which is the middle of the range.

      Anyway, it is quite possible Labour have hit 35-36% a few days ago, and that's only now starting to filter through.

      Delete
  12. Scottish Skier
    Note to clear up the confusion re: today's @SavantaComRes polling, one showing Con +8, and one Con +6:

    Con +8 published by a newspaper (FW: 04 - 05 Dec); Con +6 published by 'Remain United' (FW: 02 - 05 Dec).

    The one published by Remain United did not prompt candidates.

    Westminster voting intention:

    CON: 41% (-1)
    LAB: 33% (+1)
    LDEM: 12% (-)
    BREX: 3% (-)
    GRN: 2% (-1)

    via @SavantaComRes, 04 - 05 Dec
    Chgs. w/ 03 Dec

    ReplyDelete
    Replies
    1. Aye, but still a 36% means that could be where labour have reached. It's not possible to know, but a 35% ICM then a 36% extends their range now to 32-36.

      Looking at data point clustering, and 34% as of as 2/3 December, i.e. 4-5 days ago seems possible, so 36% as wee speak not impossible, with this one a straw in the wind.

      Again, only time will tell.

      Tory bums should be squeaking as things are tightening, and they have not made progress on 2017 in terms of share. If they drop just a couple of %, the can probably kiss a majority goodbye.

      Things changed a lot with brexit, and although the Tories piled up the votes in blue shires, that didn't help in 2017. 42% of the vote yet no majority under FPTP is a serious problem.

      Delete
    2. I haven't a clue how it will go down south but my gut feeling is on the limited amount of canvassing I've done the SNP are going to do well on Thursday .Nothing scientific just years of campaigning.
      I wouldn't be surprised if the SNP poll in the high 40s

      Delete
    3. The new Comres poll for the Sunday Telegraph Scotland subsample:

      SNP 43%
      Con 27
      Lab 16
      LD 8
      BXT 2
      Oth 3

      Delete
    4. The gap between the top 2 is 16%. The gap in 2017 at the same stage was only 8%, 40 to 32, the Tories have fallen back despite their UK numbers furth of Scotland. The SNP has the party with the highest degree of intention to vote, that should help.

      Delete
    5. Nothing fundamentally changed in Scotland (considering last week polls). Good news about something, finally.

      Delete
  13. Tonight's round of polls certainly ain't looking so good for continuation of the labour surge. Looks more like a stall / running out of steam.

    ReplyDelete
    Replies
    1. Scottish Skier any numbers for the Scottish sub samples from these polls?

      Delete
    2. Within the usual ups and downs of such things, I don't see anything of note. My average still very much like it's been the whole campaign. At least for the SNP.

      The Tories have obviously eaten up the brexit vote over the past month or so to get numbers more akin to 2017 like a little England in our neck of the woods.

      Delete
  14. I have looked at the Opinium Scottish sub sample. Over sample of Tories but interesting check with the 2017 vote and the 2019 voting intention.

    The 2017 vote was;
    Con 32%
    SNP 30
    Lab 26
    LD 5
    UKIP 3

    now they intend to vote

    SNP 44
    Con 29
    Lab 15
    LD 11
    Oth 1

    ReplyDelete
  15. A Scottish poll by Panelbase - margin or error changes.

    SNP: 39% (-1)
    CON: 29% (+1)
    LAB: 21% (+1)
    LDEM: 10% (-1)

    via
    @Panelbase
    03 - 06 Dec

    ReplyDelete
  16. That's not a good poll .I think SNP will poll over 40%

    ReplyDelete