Tuesday, September 16, 2014

Yes campaign just 2% from victory in new ICM poll

ICM poll :

Should Scotland be an independent country?

Yes 48% (-6)
No 52% (+6)

The percentage changes are of course probably highly misleading, because the previous ICM poll does now look like a very obvious outlier.  This is the joint second-highest ever Yes vote in an ICM online poll.

New Survation and Opinium polls are also out tonight - full analysis of all three polls will appear HERE.

31 comments:

  1. Is this a phone poll or back to using a panel?

    ReplyDelete
  2. Almost something fishy about that previous outlier - as if they might be laying ground so that if dark Yes springs a surprise, they can grasp the straw of that previous poll and say "we came the closest amongst our competitors." Or is that pure conspiracy theory?

    ReplyDelete
  3. That's two with four point lead...if it's like this on the day were screwed :/

    ReplyDelete
  4. Alastair, I don't think you're far off the mark- I think if the pollsters settle around 50% they will say to future clients, "we were within 5%" when all's said and done.

    Getting this wildly wrong would be really bad for the polling industry, maybe its the kick up the arse they need.

    ReplyDelete
  5. I may be repeating myself a lot, but this looks pretty much as a statistical tie. All polls in the last two weeks are basically within margin of error of 50/50, perhaps with a slight advantage for No.

    The most surprising thing is the level of agreement among all pollsters.

    Xabi

    ReplyDelete
  6. It's as if they can't call it so are going down the middle

    ReplyDelete
  7. James, why are you comparing it with the online ICM? Even the Scotsman is calling it a swing to Yes, in comparison with the last telephone poll.

    ReplyDelete
  8. "An ICM poll compiled in August put support for Yes at 45 per cent when undecideds were taken out – indicating that the Yes campaign has gained three percentage points to 48 per cent.

    This has been accompanied by a slip of three percentage points for a No vote which has fallen from 55 per cent to 52 per cent over the same period."
    http://www.scotsman.com/news/politics/top-stories/scottish-independence-poll-icm-puts-yes-up-to-48-1-3543614

    It would be
    48% (+3)
    52% (-3)

    ReplyDelete
  9. Good poll for Yes, but I'm starting to get concerned that since the YouGov poll (Yes at 51%) took place, Yes has lost its small lead and fallen behind.

    ReplyDelete
  10. They might be converging around a certain value, so are becoming more precise, but they could be converging around a bad number- their methods are being stress tested to destruction here so the errors will be much larger than the size of the difference between the sides.

    ReplyDelete
  11. What a truly weird post at 9.11pm, "If you don't have enough votes you lose"?

    ReplyDelete
  12. If the Survation poll is confirmed then we have more movement to YES.

    It is to close to call. I big turnout will swing it for Yes.

    ReplyDelete
  13. Yes FitzyMan, I think it was inevitable that there would be a hit to Yes after the financial wobbles at the end of last week. In this final push the question we must press is whether fear and money are to triumph over hope and justice.

    ReplyDelete
  14. Let's face it: to win this there'll have to be a logistical effort on the day. This is a total tie.

    Xabi

    ReplyDelete
  15. @ Scott

    Care to elaborate in more detail?

    ReplyDelete
  16. If there's anything consistent and we are calling the last ICM an outlier...it's a small lead for No again and again...can't believe this :/

    ReplyDelete
  17. The swings back and forth of a percent here and there are totally meaningless- when the error in the poll is probably of the order of 6 or even 10%, obsessing about a swing of 1 % is like worrying about the colour of the hammer someone is hitting you with.

    ReplyDelete
  18. Correct me if I am wrong but this ICM Poll and the last ICM Poll are different polls. One was an internet panel and the other was a telephone poll.

    I would like to see the internals and what percentage of the undecideds are swinging to YES.

    Regardless the referendum has become a statistical TIE.

    We know that even though 90% of eligible have registered to vote, actual percentage will be less.

    Whoever physically gets their voters to the polls will win. This is classic GOTV which has been used by the Democrats in the USA for decades.

    ReplyDelete
  19. @ Scott

    Do you think it's possible that Yes could be ahead by that sort of percentage?

    ReplyDelete
  20. @Fitzyman

    The only certainty here is that every poll is wrong. By how much, well we don't know, but 3% error is a laughable claim from the pollsters given that refers only to "sampling error", they simply have no idea how much error is in their published data which in reality arises at every single stage of the polling procedure (resulting in systematic error)- because get this, they have no idea what the "true" value is.

    My line of work involves modelling of certain environmental realities using computer simulations, sometimes our results are "right" but we have no idea why- i.e. what part of the process made our result "right". The pollsters are in a similar position- they're trying to "model" an outcome which is too complex for the tools they have at their disposal- and none of them are telling you that.

    ReplyDelete
  21. @Fitzyman

    yes for sure, if Yes is at say, 48 plus or minus 5% (it could be much more), and the same uncertainty for the No vote, there are plenty of scenarios within that range where Yes could be ahead.

    ReplyDelete
  22. @ Scott

    Thanks for getting back to me on this, I appreciate it.

    ReplyDelete
  23. "Is this a phone poll or back to using a panel?"

    Ignore my original comment (above.

    They were both online polls, it's just that the last ICM, the one with a sample of 705, was part of a UK-wide poll.

    It was Survation that did the out-of-the-blue telephone poll.

    This is all getting very confusing.

    ReplyDelete
  24. @Fitzyman

    no problem, I guess what I would say is treat the polls with huge caution, but also that what I would look for is signs of large shifts- I'd argue there has definitely been evidence of that to Yes in the past few weeks, shifts of 1% are meaningless in the context of 5, 6, even 10% errors.

    ReplyDelete
  25. Final Yes Surge needed.
    We can do it.

    Chest dip over the finish line!

    ReplyDelete
  26. @ Scott

    So are you saying that there is the possibility that polls that show significant shifts away from average estimates made by other polls could potentially be more accurate?

    ReplyDelete
  27. @Fitzyman

    No I don't think so- there's simply no way to know. The only way the pollsters could tell you whether they were accurate at any given point was if they could tell you the actual value..... there's no way for them to do that.

    What I meant was, taking the group of polling companies as a whole, if they, as a group, have shown a shift to Yes over the past few week then that might mean something. But be aware if there's anything inherent in ALL of their methodologies that would favour/harm one side or the other, well the entire group could be wrong. Say not being able to contact people without either a landline or a PC for instance.

    ReplyDelete
  28. If you lose - and I accept that it is still in the balance - it is due to the fact that you have zero credibility on the currency.

    Your leader is a man who claimed in 1999 that sterling was a 'millstone around Scotland's neck,' and who backed the disastrous Euro, which has caused misery around Europe, but who now says with a straight face that Scotland has to adopt the pound.

    He and YES have ZERO xredibility on the currency.

    ReplyDelete
  29. This poll has been fixed to favour the NO campaign. Wouldn't believe ANY of the BPC polling. Would like to see data tables. A little tweak to the weightings in one or two areas an Le voila!

    ReplyDelete
  30. This poll has been fixed to favour the NO campaign. Wouldn't believe ANY of the BPC polling. Would like to see data tables. A little tweak to the weightings in one or two areas an Le voila!

    (Pressed wrong button before)

    ReplyDelete