Wednesday, September 17, 2014

Extraordinary : ICM still aren't weighting by country of birth, even though they KNOW they have 8% too few Scottish-born people in their sample

Well, the title says it all, really.  I just don't know what credibility ICM think their published voting intention figures can have when they ask for people's country of birth, and then fail to weight by that measure when it turns out that their sample is completely skewed.  Just 73.6% of their weighted sample was born in Scotland, whereas according to the census results the figure should be roughly 81.5%.  English-born people account for 16.6% of the sample, whereas it should be 9.6%.  You'd be forgiven for thinking that ICM believe this is a trivial matter, but in fact country of birth is still one of the strongest predictors of referendum vote - in this poll Yes lead by 52.2% to 47.8% among Scottish-born people, while No lead by 75.6% to 24.4% among English-born people.

My rough calculation suggests that if country of birth weighting had been applied, the No lead among the whole sample would have been just 51% to 49% - that's before the turnout weighting (which helps Yes slightly) kicks in, and that in turn would have been likely to take us to roughly Yes 50%, No 50%.

*  *  *

YouGov have published a poll for Sky News asking about issues like the currency and more devolved powers.  I'm guessing these may well have been bolt-on questions from the voting intention poll which will be published tonight, in which case it's clear from the cross-breaks that No are likely to have some kind of lead in the headline figures (but probably not a huge one).

UPDATE : It seems from what I've seen on Twitter that YouGov are still polling, so perhaps the 2000+ respondents from the Sky News poll will only make up about two-thirds of the 3000 sample that is apparently expected.  In that case obviously the calculation is thrown up into the air again - it could be better or worse.

37 comments:

  1. Could it be that those born outwith Scotland are more likely to vote?

    ReplyDelete
  2. Anon : No, because the whole sample is weighted to target figures which don't take into account likelihood to vote. However, what is abundantly clear is that English-born people are far more likely to join volunteer online polling panels. Why that should be the case is anyone's guess.

    ReplyDelete
  3. Anyone might think YouGov are trying to manipulate the result of their poll, but I doubt the future Lord Kellner would do such a thing.

    ReplyDelete
  4. @James

    Why do you understand that YouGov published numbers suggest a No lead?

    ReplyDelete
  5. It's something that's common not just for online polling panels, but across a wide range of organisations. Scots are less likely to be 'joiners' than English. Don't know if it is a cultural difference or if people don't feel that UK-wide organisations are 'theirs'... it's a topic for further study though.

    ReplyDelete
  6. Xabi : You can tell from the cross-breaks - in most cases the figures for Yes voters are a bit further away from the average than the figures for No voters. It's impossible to calculate exactly what the No lead will be, though, because there isn't a cross-break for undecided voters.

    ReplyDelete
  7. Robert, West LothianSeptember 17, 2014 at 2:26 PM

    James is that the same poll which showed that 61% of the electorate don't know what all these "extra" powers actually amount to?

    ReplyDelete
  8. James - you say about the weighting of English-born people on polling panels, "why that should be is anyone's guess."

    In the 1970s and 80s I spent 4 years as a VSO volunteer in Papua New Guinea. Most of the "clubs" and such like were run by expatriates. Also, working as VSO volunteers we all hoped to be training nationals to take over our jobs, but it was a painfully slow process. Few of our national counterparts were able to match the energy of expats, whether at work or recreation. The bottom line in those days was that expats ran most things, often with a national figurehead to keep up appearances.

    I never felt close to the expat world and lived amongst the nationals and at their pay grades. But I puzzled at the expat-national differential in levels of public engagement. Over time I came to see what I called "the Expatriate Effect." The reason why expats had all the energy was partly that they were materially more secure, but mostly because, having left their original homes and families far behind, they had few social ties to absorb their energy and give a sense of belonging.

    The nationals were forever going off work for funerals, weddings, or to care for sick relatives. At weekends they'd have their "wantoks" (one-talks, kinsfolk) all come to stay and feast together. We expats had much less of that. We had nothing else to do but run things and make up "pastimes". We did it very well but in so doing, sometimes inadvertently colonised the soil from which more indigenously-run activities might have grown.

    My mother is English and I was born and raised in England until I was 4. On moving to Lewis in 1960 she used to say that public activities for fellow incomers was frequently a matter of "therapy for the middle classes". I know that the polling panels are socially stratified, but can they stratify for propensity to take part in the panels? I suspect not, and such is why they attract a disproportionate proportion of incomers to Scotland: people seeking a new sense of meaning, purpose and recognition.

    Normally it wouldn't matter, but with the Referendum it does. Similarly, I often raise an eyebrow at some of what social attitude surveys supposedly tell us about ourselves as Scots. Are these showing us what Scotland thinks, or only what those who offer to tell what they think Scotland thinks, thinks?

    ReplyDelete
  9. 10pm thurs

    Do all the polling companies publish their predictions

    thx

    ReplyDelete
  10. @James

    If No voters' answers are only a little bit closer to the average, could this be caused by undecideds? If this were so, perhaps we could still conceivably have a Yes lead, or a negligible No one, even though in these secondary questions undecideds are closer to No voters.

    ReplyDelete
  11. James, the writer for the Scotsman has been told of private polling that says "YES at 53-57% and NO at 55%."

    Someone is blowing smoke but who? Is it possible to test YES at 53% and reverse engineer what they may be thinking at Team YES?

    There are now 4,285,323 registered voters in Scotland. That represents 300,000 brand new registered voters as part of GOTV.

    We know turnout was 63.8% in 2010 and 50.4% in 2011 with a voting roll of 3,985,323.

    We are not going to have 100% turnout but we are not going to get 50% turnout. Lets assume we have 80% turnout as the talking heads are saying that and its midway.

    80% turnout is 3,428,258 which is an increase from 2,542,636 in 2010. Guess what, there is the so called missing million!! We have found thee...........

    I am going to assume that everyone who voted in 2010 votes in 2014. That means we have 885,622 fresh voters who have never voted before or were chronic non voters.

    The pollsters are doing something right and telling us something, I have no doubt that a pollster is very accurate based on existing modelling. Thats why I think Kellner from Youguv has the heebeejeebies.

    For the benefit of the doubt, lets assume that the pollsters are accurate in that the split is in fact going to be 48/52 on the first 2,542,636 which I think is the pool they are correctly forcasting. If the vote goes that way, the question becomes what percentage of the missing million (885,622) does YES need to get in order to be at 53% which is the lower number of the YES internal forcast.

    The answer is that YES would need to get 67% of the missing million (885,622) in order to have 53% of the total vote.

    But what percentage of the missing million (885,633) do they need to get just to get 50% plus one vote? They need 60% of the missing million.

    Is that doable? At a glance it appears so but do I have any way of checking.

    RIC did a canvas in August amongst the working class non-prior voters. I believe this was a followup to a voter registration drive. They are indicating the 18,000 plus canvas was at 63.4%.

    So I think I may have reverse engineered the YES Campaign logic and it appears very doable.

    As I said, I cut my teeth on US Campaigns and there is great similiarity to Democrat efforts amongst the formerly disenfranchised.

    IF on Thursday there is an active effort amongst YES to check off who voted and physically get your voters to the polls then YES will win.

    Again, someone is blowing smoke and me thinks its the tory toffs from London that are blowing. If locals are running the groundgame for YES, then get the vote out to YES.

    In 2012 in the USA, Romney (Republican) in the end was blowing smoke as his ground game was built by fancy high paid consultants not local street smart people that Obama used.

    Has there been any attempts to model the so called missing million?

    ReplyDelete
  12. "If No voters' answers are only a little bit closer to the average, could this be caused by undecideds?"

    Don't read too much into my use of the words "a little" - the exact number varies between questions, but it's a fairly consistent pattern overall. It's theoretically possible that it could have been caused by the undecideds, but that would assume their responses broke in a very weird and illogical way. I'm fairly convinced it's a No lead - it's just a question of the size of that lead.

    ReplyDelete
  13. @James, " I'm fairly convinced it's a No lead - it's just a question of the size of that lead."

    A NO lead in what set of voters?

    Amongst all voters as in the 97% registered or the 60% that voted in 2010?

    The "missing million" or 800,000 plus could be a rather unique voting pool.

    ReplyDelete
  14. The Guardian has a good item on GOTV, if YES is modeled after Democrat (US) efforts, thats worth two to three percent.

    I also liked how NO was importing help from south of the border;

    "The no campaign is anticipating help from activists travelling up from other parts of the UK , including Labour MPs and Welsh assembly members, and there have been some suggestions that relying on diminished Scottish Labour party networks could leave them underpowered."

    Local is always better!!

    http://www.theguardian.com/politics/2014/sep/17/scottish-independence-activists-voters-referendum

    ReplyDelete
  15. George : I'm not talking about a No lead on the ground, but in this one particular YouGov poll. Obviously pollsters use target figures that cover the whole population, so in theory they do take account of the missing million. (But of course "in theory" may not be enough.)

    ReplyDelete
  16. I've seen elsewhere that all three polls released today are 52% NO to 48% YES. Absolutely the same. Seems odd, noticeably out of place.

    ReplyDelete
  17. James, Scots born has been as high as 78% in the weighted base and up to 79% in unweighted base in ICM.

    Has been dropping. The value in the recent one is the lowest it has been.

    That might imply that its not Scots are less likely to sign up to poll panels, but rather they have recently been less likely to respond to iref polls.

    Even MORI have had trouble here. Are Scots born / natIDers less likely to have landlines, or just less likely to agree to be polled?

    The fact that Scots respondent numbers are dropping yet Yes has been rising (even though Scots are more likely Yes) is a good sign.

    ReplyDelete
  18. @James

    Fair enough, James. Let's see if the numbers are again with margin of error.

    ReplyDelete
  19. You mean the ones that came out last night? Yes, that's true - ICM, Survation and Opinium all had 48/52.

    By the way, my assumptions about YouGov may have been wrong - it seems that they're still polling and that their sample size for tonight is 3000, which is bigger than the poll I was looking at. Perhaps the 2000-odd for Sky News poll are only part of the sample for tonight, in which case the calculation is thrown up into the air again.

    ReplyDelete
  20. Note the drop in Yes the last yougov was closely associated with a large drop in response rates amongst Yes supporting groups.

    ReplyDelete
  21. Panelbase on Twitter

    Final indyref poll Y48, N52 (excl DK)

    ReplyDelete
  22. 4 pollsters now showing identical result, very strange or very accurate?

    ReplyDelete
  23. and Panelbase last 5 pols have been 48, 48, 48, 49, 48

    is it me or it that odd?

    ReplyDelete
  24. Chris Knowles

    Very tired?

    ReplyDelete
  25. It seems like a amazing case of extreme convergence!

    ReplyDelete
  26. @James, thanks for the clarification.

    The comments from Kellner of Yougov make me think he does not have a grasp as to what the missing million will do.

    @Scottishskier, thats an interesting observation that Scottish born has dropped but YES has firmed up.

    @All, a couple months ago it looked like ground game would not matter because YES was too far behind.

    YES claims 35,000 activists working on Thursday. If they each get two-three brand new voters to the polls on Thursday, this is done and dusted.

    Again, as I read about BT and their GOTV it sounds more like Romney the republican in 2012 which was a spectacular GOTV failure.

    ReplyDelete
  27. So the entire might of the Britnat propaganda machine has had no quantifiable effect at all?
    Now there's a shock!

    Still getting the same message out though. Don't bother voting because your side will lose. Truly vile.
    Encouraging people not to use their franchise.

    Anti-democracy scum, then, now and forever.

    ReplyDelete
  28. They have been all round the table and one of them says "cards on the table I do not know the f***k whats happening ,What about we all put in the same figure and then we can all be equally right or wrong and then no-one loses their reputation"
    Too conspiracy theory ?

    I'm going for a 58 yes ...on gut feel and looking at the canvass results ....

    ReplyDelete
  29. Anon,

    Very tired? Yes, but will work my arse of tomorrow regardless to GOTV

    ReplyDelete
  30. Perhaps the disproportionate number of English-born people in the polling panels can be explained thus: English people who move up here to study or work are more likely to be middle-class. A working-class English person is more likely to stay closer to home.

    ReplyDelete
  31. Chris Knowles

    Well said, Chris.

    ReplyDelete
  32. Perhaps the low Scots response could be explained by this.

    Media to Scots / Yes people:
    - You are a racist
    - You are vile
    - You are intimidating to innocent people
    - You are violent
    - You are like the Scottish BNP
    - You are a separatist wishing to destroy our wonderful Britain
    - You are supporting the liar Salmond
    - This is all about Braveheart
    - Your country could not survive alone
    - You are an economic fool
    -etc

    Pollster:

    Now, do you wish to be asked what you intend to vote in the iref after I get all your personal details?

    If we get a Yes tomorrow despite being behind / equal in polls, it will be a lot to do with this.

    ReplyDelete
  33. Social media will be going mental for YES tonight, canvass returns showing 40% yes in posh areas, 70% in poor areas ..,. It's all going to be about turnout on the day - if we get 80% we should be the winners!

    ReplyDelete
  34. @Bigalan, yes agreed.

    1. The elderly vote is already maxed out in numbers for NO. They can not "mine" additional votes out of that source.

    2. YES has to get the large part of the 800,000 disenfranchised but new voters to the polls, period.

    Thinking about this, I had been afraid of Gordon Brown drumming up the NO vote with C2DE but no longer fear that. Thse new voters are not paying attention to new/old labour.

    ReplyDelete
  35. Kelner was interviewed on Radio 2 today and stated that he didn't know what the missing million were going to do.

    Rob

    ReplyDelete
  36. Lady came to our street stall today and said she'd voted No by post, but after hearing the Dimbleby programme last night she would have voted Yes on the day.

    I hope that effect caught a bunch of people without postal votes!

    ReplyDelete