Sunday, July 6, 2014

Extraordinary : the opinion poll the No campaign were gloating about last night actually shows a DECREASE in the No lead

There have been a lot of contradictory suggestions about what "the main lesson" of the new TNS-BMRB referendum poll is, but to my mind the most important lesson is that when the No campaign crow about a poll in advance, it's no longer safe to automatically conclude that the No lead must have increased, or even remained static.  With the TNS datasets having been released on Tom Hunter's website, we now know that the No lead has in fact fallen by 0.6% after undecideds are stripped out.

Should Scotland be an independent country?

Yes 41.4% (+0.3)
No 58.6% (-0.3)

41.4% is also (albeit only fractionally) a new high watermark for the Yes vote with TNS-BMRB in the campaign so far - the previous high was 41.3%.

Even on the headline numbers that take account of Don't Knows, it turns out that No have been heavily flattered in the published figures by the effect of rounding.  Here is the position when rounding is only to one decimal place -

Yes 32.2% (+2.6)
No 45.6% (+3.2)

So the "widening of the gap" that the No campaign are wittering about (spin which one or two media outlets have stupidly - if predictably - fallen for hook, line and sinker) amounts to a statistically insignificant 0.6% - exactly the same amount by which the gap has narrowed on the more important measure that excludes undecideds.

The one silver lining I can see for No in this poll is that, in contrast to the last couple of TNS polls, respondents who say they are certain to vote are only very fractionally more likely to say they will vote Yes.  However, the TNS figures after turnout filters are applied have tended to be much more volatile than the headline numbers, so not too much should be read into that.  And the news is actually pretty good among respondents who say they are either certain or very likely to vote - the Yes vote among that group (with undecideds excluded) stands at 42.6%, which is very similar to last month's figure of 43.4%.

*  *  *

In other polling news, the Tories once again seem to be standing on the brink of crossover in the GB-wide polls for next year's general election.  The last couple of YouGov polls have shown Labour's lead shrink to just 1 and 2 points respectively.  What will be the implications for the referendum if it becomes unambiguously clear over the next few weeks that David Cameron is heading for another five years in Downing Street?  I think we can all hazard a guess...


  1. How does 2.6 round down to 2 and 3.2 round up to 4?

  2. It doesn't - 32.2 rounds down to 32 (a 2 point increase on last month's rounded figure of 30), and 45.6 rounds up to 46 (a 4 point increase on last month's rounded figure of 42).

  3. On the UK polls, last couple of months have had the Tories and Labour crossing over within the range of values provided by each pollster, i.e. the Tory % found by one pollster is not uncommonly higher than the Labour found by another.

    That means we've see a few 'Tory narrow leads'.

    If the average gap falls below 2%, then we will see increasing 'Tories overtake Labour as the race for No 10 is on' in the headlines. The last thing BT needs right now but something the right-wing UK media are itching for.

  4. Would a Tory lead really have much effect on the referendum? You'd have to be a very short-term thinker to vote Yes simply to escape a Tory victory in 2015. After all, a No vote guarantees that, even if Labour do win next time, the Tories will rule Scotland again many, many times in the future.

  5. Would a Tory lead really have much effect on the referendum?

    No, probably not much at all. What it should do is affect Y/N polls by boosting the number of people admitting they plan to vote Yes, but currently saying DK or No to pollsters. People like my mate's mum who told porkies to MORI.

    As we have seen, various polls give Yes a lead if the Tories are back again in 2015.

    This is not giving people a reason to vote Yes, more an excuse to justify what they plan to do. It's the same for the SSAS bribe of a few hundred quid a year better off; suddenly it's a big majority for Yes. Really, for the price of a cheap egg majo sandwich a day?

    We saw a similar section of the population lie to pollsters over the course of a year ahead of 2011. In 2009, the SNP were already on 40%; all they needed was a wee boost from defecting libs due to the coalition.

    However, as the anti-SNP propaganda was piled on Scotland ahead of 2010 and 11, up to 15% of the population then started lying to pollsters about their most probable intention, telling them Labour when they planned SNP. However, ask them if they were happy with Labour/Gray and they said no, yet ask them if they were happy with the SNP and they said 'Yes, decent enough'. On the doorsteps they told SNP canvassers what they actually planned, hence the disparity between what the SNP found and what the polls were getting. Likewise the utter collapse of the Labour vote in the space of a few weeks. It didn't collapse because it was never really there. A mirage created by the electorate to take the heat off themselves and Scotland. Which it did quite successfully.

    So, the prospect of the Tories allows someone who is shy Yes to say 'Well I'm voting Yes because we're getting another Tory government'. It gives them something the feel they can justify their intention in 'public' with. Of course that's not actually the reason they're voting Yes, well only part of it.

    In 2011, people said 'It was Labour's campaign that lost them it'. This is bollocks. The terrible campaign just allowed people to justify saying they were not going to vote Labour but SNP. Hence over the campaign period the polls started to show what the real result would be. Even then, all pollsters underestimated the huge scale of the win. Not because their methodologies were wrong, but likely because a good few people remained coy right up to voting day.

    Remember, if you are voting Yes then you are about to do something huge. Something you are publicly attacked for by senior figures in politics and the media. You are a 'blood and soil nationalist who hates the English and is breaking up Britain' etc. That makes people less than honest when asked. Trying to make people feel bad about voting Yes - a no campaign ploy - won't stop them doing so, but will make them much less likely to admit it.

    Hence the much higher level of Yes in online polls where anonymity is much greater.

  6. The America pollster who called Obama state by state wins said,The Scots wouldn't risk the Union because of a Tory government ?

    any comment on that


    sorry I can't remember his name

    1. The Scots and Americans are as different as black and white and the flop of Milibands campaign with Mr Yes We Can is the glaring example.

  7. Water : Do you mean Nate Silver? If he did use those exact words, "risk the union" illustrates perfectly his astonishing lack of perspective on this issue.

    We discussed this last year when he first made his comments - a lot of the things he said just didn't make any logical sense whatever. For example, he cited Quebec 1995 of an example of how Don't Knows in a referendum like this can supposedly be expected to break for No, apparently unaware of the fact that the No campaign in Quebec threw away a handsome lead and ended up scraping over the line by a tiny margin. And his suggestions of events that might be game-changers were faintly ludicrous as well. I just don't think he has much of a feel for non-US politics. Doubtless if No wins he'll still claim it as another accurate prediction, but that one would be by complete luck rather that sound judgement.

  8. The America pollster who called Obama state by state wins said,The Scots wouldn't risk the Union because of a Tory government ?



    was silver--thx

  10. Looking at the data tables, the figures for Voters under 55 (Yes: 36% No: 41%) and C2DE are very encouraging (Yes: 36% No: 41%)

    There's a possibility that the decrease in don't knows is down to the length of this survey - the sample which agreed to take part may have been more politically aware than usual for TNS polls.