Tuesday, August 12, 2014

Mystery of "new TNS poll" showing another boost for Yes

A number of media outlets were seen to briefly report a new poll a few hours ago, seemingly from TNS-BMRB.  The figures given were -

Yes 38%
No 46%

- which if true would be the narrowest gap of the campaign so far from TNS, regardless of whether those are the headline figures or the turnout-filtered figures.

Mysteriously, the reports disappeared soon afterwards, which may indicate that it is an embargoed poll that was accidentally released too early.  Or it could just be an old poll from another firm - time will tell.

51 comments:

  1. I clocked this earlier on the Herald and the DR websites just after lunch... It had a sample size of 730(ish) people who said they were certain to vote and gave the numbers you quote in your blog post. I'm pretty sure the articles said the fieldwork was carried out before the Commy games but i may be wrong

    But as you say now vanished... odd.

    ReplyDelete
  2. I can certainly believe that the fieldwork took place before the debate (because TNS are so slow), but I find it hard to believe it took place entirely before July 23rd. Maybe it just started before then.

    ReplyDelete
  3. The Daily Record article (before it was edited) seemed to suggest it was a "certain to vote" figure, although this wasn't obvious until you looked down further and it mentioned a different overall figure, and then looked back up to see the phrase "who will vote in September", or words to that effect. Think it mentioned 773 people as well.

    The overall figure seemed to have Yes stuck at 32% and No up 4% at 45%. So even that's interesting - the gap increasing in the overall figure, but decreasing amongst those who'll actually vote? Hmmmm.

    ReplyDelete
  4. Those where indeed the figures The Record and the Herald quoted.

    On my rough calcs that's a 45/55 Yes/No split without the DK's

    As usual long lead time, covered both end of the Games and debate

    Does tend to provide further evidence of the lack of the #DarlingWin bounce our Tuba playing friend was trying to fan over the last 7 days.

    And that Survation at the weekend may well be a 'rogue'

    Who normally commissions TNS polling?

    And isn't the monthly Survation/5 Million Q's/Record poll due tomorrow or Thursday?

    ReplyDelete
  5. "Who normally commissions TNS polling?"

    I'm not sure anyone commissions it as such, but they seem to have some kind of relationship with the Herald.

    ReplyDelete
  6. Apologies i meant during the games

    ReplyDelete
  7. "And that Survation at the weekend may well be a 'rogue'"

    I've just assumed that the swing in the Survation poll is a result of the margin of error favouring Yes last time, and it favouring No this time.

    ReplyDelete
  8. Doug : The three previous Survation polls all had Yes at 47% (with DKs excluded). Admittedly, all three polls had potential problems with them because of severe upweighting of certain groups.

    ReplyDelete
  9. Sunshine on CrieffAugust 12, 2014 at 3:56 PM

    TNS shows positive figures for Yes, but the story immediately disappears from view?

    Spooks?

    There is no such organisation as TNS. There never was an organisation called TNS. Move along now.

    ReplyDelete
  10. If Doug is right about the figures for the overall sample, rest assured we'll still be getting another of Blair McDougall's beloved "No Thanks Extending Lead" Twitter graphics.

    ReplyDelete
  11. Here is the original google cache of the Daily Record story that was removed. Can someone please archive.org it quickly thanks.

    http://webcache.googleusercontent.com/search?q=cache:5TFPkH9jzlUJ:www.dailyrecord.co.uk/news/scottish-news/support-no-vote-stalling-says-4039764+&cd=11&hl=en&ct=clnk

    ReplyDelete
  12. Here's an archived version

    https://archive.today/AOo3p

    ReplyDelete
  13. Well, straight away I can see an inaccuracy in the Record's reporting of the poll - it clearly implies that Yes were on 32% among the whole TNS sample as far back as November, which is completely untrue. Yes have made significant gains since then.

    ReplyDelete
  14. It's an odd (and long)period of time to be polling over but, excluding undecideds, that's Yes 42% No 58% over the whole sample and Yes 45% No 55% among those certain to vote. Those numbers give a 73% turnout.

    ReplyDelete
  15. Somebody posted a picture on twitter of the front page of the Herald showing this story. I'm sure that I saw it this morning.

    ReplyDelete
  16. The fieldwork was between May and July(!).

    How does this gel with the TNS poll from 25Jun-9Jul showing Y 32%, N 41%, UK 27%? For some reason people polled during those 2 weeks were significantly more undecided.

    I have no idea why polling companies do this, similar with the YouGov poll whose dates spanned the debate, FFS!

    ReplyDelete
  17. o/T Straw poll on the Scottish Banter Facebook page Yes 70%, No 27%, 10,000 votes in 2 hours (only one vote per respondee). Yougov,Ipsos-Mori polls debunked.

    ReplyDelete
  18. Well I'm certainly growing more and more convinced by the day that the pollsters are operating in an open and honest manner while the unionist media and No campaign are in no way looking like a bunch of desperate and incompetent amateurs.

    LOL

    ReplyDelete
  19. Anonymous said...

    "The fieldwork was between May and July".

    Those dates refer to the Social Attitudes Survey. The TNS poll must have been in the field between some date after 23 July to some date after 5 August - which isn't as long a time as I'd thought!

    ReplyDelete
  20. So the changes between the one published 14 July and this (now unpublished) one is (excluding undecideds) -

    Certain to vote : 0.7% swing from No to Yes

    Others : 2% swing from Yes to No

    Proportion of definite voters down by 1%.

    ReplyDelete
  21. ""Who normally commissions TNS polling?"

    I'm not sure anyone commissions it as such, but they seem to have some kind of relationship with the Herald. "

    Why do I suddenly feel the need to have a shower?
    relationship with the Herald?
    whahahaahahahaha

    ReplyDelete
  22. oldnat - thanks! Hopefully it was conducted over a few day period, rather than weeks.

    ReplyDelete
  23. Anonymous

    It was probably a couple of weeks. Since its seldom that anyone pays TNS to do political polling, they tack these questions on to their normal "Scottish Opinion Monitor" where they ask questions that their paying clients want.

    Their predecessors, System 3, gave their political polls free to the Herald, basically as a cheap way of advertising themselves. My understanding is that TNS still give their political polling data (which costs them little to produce) to the Herald (and perhaps others) for the same reason.

    ReplyDelete
  24. oldnat - Fair enough for them wanting to advertise, and I suppose we should be grateful for any polling whatsoever!

    ReplyDelete
  25. That Daily Record article is so extensive (even including a direct interview with the Head of TNS Scotland) that this is obviously just a case of someone messing up and putting the article out before they were supposed to - whether that's because it was intended to appear at a later date or the figures need to have a final check/have some error, or otherwise.

    ReplyDelete
  26. Kinda odd that there'd be an embargo on a standard (no client for political Qs) TNS poll.

    And if it wasn't standard, but for a client what was it doing in the hands of the herald + Record who so readily ran with it then withdrew?

    And if it was a paid for poll, why was it in the hands of non-clients before the client reported it?

    It's out there already on social media, so why hasn't it been released as quickly as possible to quell rumours etc?

    ReplyDelete
  27. Gordon Robertson

    That was my assumption when I initially saw it on the Herald site. However, since the same thing happened on the Record, it clearly wasn't the papers who made the mistake.

    Either, they were both taking it from a news agency (quite likely since papers don't have many reporters these days) where the error has made, or TNS themselves have asked for their press release to be withdrawn for some reason. I can't think of other reasonable explanations.

    ReplyDelete
  28. Scottish_Skier

    I emailed a contact at TNS to ask about this. If I get a reply, I'll share it - though I'm not holding my breath!

    ReplyDelete
  29. I think the following tweet might explain what is happening

    Mike Smithson ‏@MSmithsonPB 4m

    Another #IndyRef poll due to be published at midnight. Most of fieldwork, though, was before the Salmond-Darling debate.

    ReplyDelete
  30. Laukat

    Thanks. Sounds like a TNS error.

    ReplyDelete
  31. "Well I'm certainly growing more and more convinced by the day that the pollsters are operating in an open and honest manner while the unionist media and No campaign are in no way looking like a bunch of desperate and incompetent amateurs."

    I don't really see much benefit in doing this - why would you put an article out and then immediately pull it on purpose? I think oldnat is probably on the money here and TNS or a news agency have just released it prematurely or made some error that needs rectified.

    ReplyDelete
  32. "TNS released it prematurely or made some error that needs rectified"

    I've been wondering about the latter given the circumstances. A bog standard / non-client poll released too early while becoming further out of date by the day needs such a fuss?

    ReplyDelete
  33. Indeed Scottish_Skier. I somehow doubt the Record enjoyed the appearance of ineptness that having to hurriedly withdraw their copy quite clearly gave. They had all their spin lined up as well.

    We'll know more eventually, but the very fact that these mere polls are being regarded with almost the same level of press hysteria and 'sensitivity' as an article that had got a D Notice slapped on it by westminster, is somewhat telling.

    It's just a poll. If you have to release it early then do so for god's sake.

    ReplyDelete
  34. Scottish_Skier.

    But midnight tonight places its publication after tonight's Inverness debate - though I can't see the changes being so great as to affect things.

    ReplyDelete
  35. "But midnight tonight places its publication after tonight's Inverness debate"

    New poll shows movement towards Yes after Inverness debate!

    ;-)

    ReplyDelete
  36. Bit of a minter for the Herald and the Record (following suit) in releasing this earlier today.

    ReplyDelete
  37. I saw the story of the poll on the Herald website this morning and thought it was rather strange. What the article said conflicted with the actual figures released which, as James said, would have resulted in an all time high Yes vote for TNS. Blair McDougall tweeted an alert earlier that a TNS poll will be out at midnight (With the fieldwork done before and after the Salmond-Darling debate), and as we all know he tends to alert everyone to polls that are favourable to No. So all will be revealed shortly.

    ReplyDelete
  38. Very shortly, in fact.

    Yes: 32 (N/C)
    No: 45 (+4)

    Seems the figures released earlier were some sort of error.

    ReplyDelete
  39. good news for no

    up 4%

    3rd poll since the debate

    all good for NO

    ReplyDelete
  40. Stoat

    The figures haven't changed but the reporting of them has.

    Who told the Herald to rectract and put up a more pro no statement?

    See the fornt page of the herald here https://pbs.twimg.com/media/Bu3v2GXIcAAEHzW.jpg:large

    ReplyDelete
  41. The problem is that the article I saw was framed as being favourable to know, yet the figures were as James said at Y38, N46. This leads me to think that it was an erroneous article they got from a news agency. That would also explain the fact that the Daily Record (According to others here) said the same thing.

    ReplyDelete
  42. Have a look at the daily record page form earlier today here https://archive.today/AOo3p

    The figures for all 1003 people polled are 32Y/45N

    However the figures for those defintely going to vote (only 733 people) becomes 38Y/46N

    Looks to me like the same poll but the reporting has changed. Who made the papers change the way in which it was reported?

    ReplyDelete
  43. Numbers are the same as in what the Record article posted earlier - though the numbers can be (and will be) spun for both campaigns.

    Undecideds down by 4% among those not certain to vote, but only 2% among those certain to.

    Underlying position remains the same. During the couple of weeks of fieldwork, there was a shift from Undecided to Yes among certain voters. Among those less certain to, the No vote rose.

    ReplyDelete
  44. OK, thanks all for clarifying that.

    ReplyDelete
  45. So mad english public school ranter and failed chancellor Darling has shot his load and achieved nothing.

    He's now totally screwed by millipede's admission that he would deny us a currency union out of spite no matter the damage to his english electorate. Well done. Ready, aim, fire, Bang! No feet left!

    ReplyDelete
  46. Oldnat

    I agree with everything your saying on the figures

    However there is a far bigger story with this poll than the numbers.

    The Herald earlier today ran a very similar story on these figures to the Daily Record one. Have a look at the way the Herald reports them today and ask why did it change? Why was the pro yes reporting replaced with pro no reporting?

    ReplyDelete
  47. The "certain to vote" vs overall numbers aspect is relatively interesting from a political science perspective. What it implies is that overall people are supporting No (and this poll shows that gap widening) but that the No side is likely to have more trouble mobilising its vote.

    It's interesting because that brings other issues into play: e.g. the usual talk you get about the weather affecting turnout (so if it rains on 18 September that might be a good thing for the Yes side and a bad thing for No).

    ReplyDelete
  48. Anonymous: 12:03 AM:

    According to John Curtice, the fieldwork was done between 23 July (Commonwealth Games opening) and 7 August (Two days after the Salmond-Darling debate). So I don't think this poll could give us any insight to result of the debate.

    ReplyDelete
  49. Within the campaign period certain to vote is usually used by pollsters so the 38/46 should really be the headline figure

    Two points the Herald and Record headlines were pulled because they were pro-Yes so I now believe the polls are being manipulated and being used to destroy the confidence of the Yes campaign.

    2. Better Together always get details first so are able to spin any poll as favourable to them.

    We should assume the polls are no more reliable than any other BT propaganda but how to stop confidence draining away?

    Could Yes commission a poll? Or perhaps we could produce a poll from RIC canvassing?

    ReplyDelete
  50. I'm still positively bouncing with confidence Denise.

    You'd be better employing someone to kill a goat and examine it's entrails. Cheaper too.

    ReplyDelete