Saturday, September 14, 2013

Boost for pro-independence campaign as ICM show a No lead only half as big as suggested by YouGov

I was slightly worried earlier this evening when Blair McDougall and Euan McColm eagerly retweeted the news of an impending (and extremely rare) ICM poll on the independence referendum. ICM are of course one of the most credible and evergreen of pollsters, so if they had shown a whopping No lead in line with YouGov, it would have been a cause for concern. But thankfully that hasn't happened, and on balance I'd say that Yes should be the campaign to take most heart from this poll - the figures are roughly equidistant from Panelbase on the one extreme with its one-point Yes lead, and YouGov on the other extreme with its thirty-point No lead.

Should Scotland be an independent country?

Yes 32%
No 49%


The No lead is some five points lower than in the recent TNS-BMRB poll, and a full thirteen points lower than in the recent YouGov poll with its dodgy preamble. ICM also join a clear majority of pollsters in putting active support for Scotland remaining in the UK at less than 50%.

We now essentially have three camps of pollsters - Panelbase with an outright Yes lead, Angus Reid, ICM and TNS-BMRB with middling No leads, and Ipsos-Mori and YouGov with much larger No leads.

Panelbase - Yes lead of 1%
Angus Reid - No lead of 13%
ICM - No lead of 17%
TNS-BMRB - No lead of 22%
Ipsos-Mori - No lead of 28%
YouGov - No lead of 30%


That really ought to lay to rest the suggestions from commentators such as Alex Massie that it's "Panelbase against the field" - if Panelbase are outliers on one end of a very broad spectrum, then YouGov are plainly outliers at the other end.

So what can explain the fact that ICM's No lead is markedly lower than YouGov's? Well, basically, they've shown a degree of professionalism that puts YouGov to shame. They do use a preamble to the referendum question, but it's neutrally-worded, and doesn't cast independence in pejorative terms. They also ask how people think they will vote on the actual referendum date, rather than "tomorrow" (eliminating the risk that voters might assume that they should respond more cautiously to a hypothetical question about "tomorrow", because they don't yet have enough information to vote Yes with confidence). Perhaps most importantly, ICM weight by recalled Holyrood vote, not by recalled Westminster vote as YouGov do. The problem with the latter approach is that it's overwhelmingly likely that some people who voted for the SNP in the 2011 Holyrood landslide will falsely recall that they also voted for the SNP in 2010 - and those disproportionately pro-Yes voters will be wrongly scaled down in the weighted figures as a result.

In a nutshell, it's now very much game on.

* * *

UPDATE : There are also some whispers of another Panelbase poll tonight, although so far I can't find any details.

7 comments:

  1. False recall - dead on. I don't think any pollster has addressed the issue that only England has had a single general election in the last 4 years. Everyone else has had two, and Westminster wasn't the latest one.

    Weighting by recalled Holyrood vote may well be wrong, but less wrong than weighting by the 2010 vote.

    ReplyDelete
  2. From Wings

    Iain has posted on WoS:

    ST Panelbase poll
    Support for independence at 37% (nc) while backing for the union is 47%(+1) and 16% are undecided (-1).
    Most interestingly, when the 16% of Scots still undecided are asked how they would vote if the referendum was today the No lead falls to just four points with the gap narrowing to 52:48.


    Just as a lot of people have been picking up lately the NO vote is much softer than expected a year out before the actual vote.

    ReplyDelete
  3. James,

    This is about the Panelbase poll.

    http://www.thesundaytimes.co.uk/sto/news/uk_news/scotland/article1314221.ece?CMP=OTH-gnws-standard-2013_09_15

    ReplyDelete
  4. The Scottish Lib Dems have now fallen to 5th place, behind the Greens. First time I can ever recall seeing this.

    Panelbase/Sunday Times
    Scottish Parliament voting intention
    Sample size: 1002
    Fieldwork: 30 August - 5 September 2013
    (+/- change from Scottish GE 2011)

    Constituency vote

    SNP 45% (n/c)
    Lab 32% (n/c)
    Con 12% (-2)
    LD 5% (-3)

    List vote

    SNP 46% (+2)
    Lab 28% (+2)
    Con 12% (n/c)
    Grn 6% (+2)
    LD 4% (-1)

    ReplyDelete
  5. Thanks for the updates about the Panelbase poll, guys - I haven't been able to write a post about it because I've just left to go on holiday.

    Stuart, I think there have been at least one or two polls in the past showing the Lib Dems in fifth? I remember the Greens being quite excited about it the first time it happened, although I can't recall exactly when it was.

    ReplyDelete
  6. Have a nice break. You have earned it. My advice is just to forget about the internet for a while. You'll be more effective when you get back. Cin cin!

    ReplyDelete