Wednesday, March 5, 2014

Ipsos-Mori poll : SNP open up big Scottish Parliament lead

So now we turn to the other side of the coin when it comes to the Ipsos-Mori enigma - they may be the most extreme No-friendly outlier in respect of referendum voting intentions, but they've nevertheless reverted to being very much part of the mainstream in showing a solid SNP lead over Labour.  These figures will ease any concerns that had been building up (fuelled in particular by the Cowdenbeath result) that the tide was gradually starting to turn in Labour's favour.

Scottish Parliament constituency vote :

SNP 38% (+2)
Labour 29% (-5)
Conservatives 17% (+2)
Liberal Democrats 9% (+2)

(Note : Ipsos-Mori don't ask for voting intention on the regional list ballot.)

The only slight fly in the ointment is that Johann Lamont continues to have a modestly better net personal satisfaction rating than Alex Salmond, and it was certainly the case in 2011 that leadership ratings were a better early predictor of the election result than headline voting intention figures.  However, Lamont's lead is not all that it appears - in absolute terms more people are satisfied with Salmond, with Lamont benefiting from the fact that fewer people have an opinion about her one way or the other.  I know that a number of us have a reasonable amount of confidence about which way things will go once voters become better acquainted with the charms of the Scottish Labour "leader".

At this stage, of course, we're most interested in what the Holyrood figures say about the referendum race.  There is more than one possible interpretation - you could argue that if a pollster is showing a good position for the SNP, there's no reason to doubt its No-friendly referendum findings.  But the counterargument is that Ipsos-Mori have essentially just corroborated the Holyrood figures produced by Panelbase and Survation, meaning that there is also no longer any particular reason to fear that those pollsters have overstated the strength of Yes due to having a "Nat-heavy" sample.  So you pays your money and you takes your choice.

The other intriguing question is this : how can pollsters that show such similar Holyrood figures be coming up with wildly diverging pictures of the state of play in the referendum?  It seems to me that the explanation must be one or both of the following -

1) As a telephone pollster, Ipsos-Mori are falling victim to 'shy Yes' syndrome, something which the internet pollsters Panelbase and Survation (and indeed ICM) are immune to.

2) Although Ipsos-Mori are broadly getting the balance between the parties right, there may still be a fundamental difference in the types of SNP and Labour voter that they are interviewing, which would explain why Scottish national identity is suspiciously weak in their figures.  In particular, other pollsters may be better at reaching the type of patriotic Labour voter who is moving to Yes.

*  *  *

I have a confession to make - I actually quite like the UK's Eurovision entry this year.  Yes, it's a do-it-by-numbers effort - although it would be unfair to call it derivative of last year's Danish winner, it's not hard to see how someone could use Only Teardrops as a very rough template and end up with Children of the Universe.  But it's a strong song all the same, with a good live performer, and if there's any justice it should easily secure the UK's best showing since Jade Ewen finished fifth with a Lloyd-Webber-penned ballad in 2009.  It might even have an outside chance of winning if the staging is phenomenally good (on past form I doubt that it will be).

I still don't feel that the song represents me, though.  The old public vote system was a kind of ritual that gave us all a stake in the entry, whereas nowadays the whole thing just feels like a private enterprise by the BBC.  This year I've voted in the French and Irish national selections, and for a second year in a row I voted for the winner in Ireland (Heartbeat by Can-Linn).  I haven't had a chance to vote for or against the UK song, and for obvious reasons I don't particularly identify with the UK as a country anyway, so guess which entry I'll feel that I have a stake in?

Incidentally, as I traditionally point out at this time of year, Molly Smitten-Downes' selection means that there has still been no Scottish involvement in any UK Eurovision entry (either as a performer or as a writer) since Scott Fitzgerald performed in Dublin way back in 1988.  Bizarrely, France and Cyprus have both been represented by Scots more recently than the UK have.  That's the typically copper-bottomed type of 'union dividend' that is being brought to the table in referendum year.

In case you've forgotten it, here's a clip of Capercaillie's Karen Matheson singing for France (yes, that's France) at Eurovision in 1996 alongside Welsh singer Elaine Morgan, with the Breton language song Diwanit Bugale.


  1. James I've heard it mentioned that the stats for YouGuv, show that the same or similar percentages of people say they will vote Yes, from both the poorest and wealthiest respondents.

    All other polling statistics show a far higher percentage of the poorer people of Scotland will vote Yes, so there's something wrong with this polling companies results.

    I think the best way to analyse the referendum polls, is to filter them through 'circumstantial evidence' ie circumstantial evidence suggests that the poorer will vote Yes more than the wealthier. circumstantial evidence suggests canvasing results have a surge in support for Yes throughout Scotland etc etc.
    The further away from the evidence on the ground that these pollsters are, the less credible they must be.

    Circumstantial Evidence is excepted in a court of law, so i don't understand why we give credibility to a Tory owned company, who's often tiny samples, contradict the huge canvas returns...especially since we have saw how rotten and dishonest Tory owned companies are recently.

  2. I was disappointed too to read that Salmond was now in negative figures, and completely amazed at the notion that Lamont was doing better...

    But I'm comforted that this probably has more to do with the fact that a fair number of the population have no real idea who she is.

    She appears weekly at FMQ, which is not, I imagine, widely watched. Apart from that she is more a position than a person.

    I suspect that many people don't even know what she looks like.

    I've yet to meet anyone who seriously thinks that she does a good job. Even those who agree with her arguments seem to think that she has a communication problem!

  3. Patrick : Actually, it was YouGov's last-but-one poll that was so suspicious in that respect - when Don't Knows were excluded, the No lead among lower-income people was actually slightly higher than among higher-income people, which was plainly garbage. The Saturday YouGov poll showed that the No lead was seven points lower among lower-income people, which is more realistic (albeit the gap is maybe still a touch on the low side).

  4. What I've noticed is Yes and No shares among supporters of each political party are similar in TNS polls to what the online companies show, suggesting the online polls and offline polls both find representative samples.

    (Although I personally feel TNS has by far the most representative sample overall because they use methodology that weights by past voting behaviour - 50% of the sample didn't vote in 2011 or can't remember, 23% voted SNP and 16% voted Labour ect. The online posters underestimate people who didn't vote in 2011 very heavily, which may be why TNS shows so many undecided voters.)

  5. The problem with what TNS do is that it waters down the benefit of weighting by past vote recall. And they're also giving a lot of weight to people who are never going to vote in a million years.

  6. I agree with you partly, but the online companies overestimation of those who voted probably overestimates no support too - both in the full figures and likely voters figures.

    I believe TNS's method can give a better picture on where things stand because it takes into account C2DE voters who may not have voted in 2011 but may vote yes in the referendum for example.

    TNS's very likely to vote figures are very interesting.

    Whole Electorate:
    Yes: 29% No: 42% DK: 29% -13

    Certain/Very Likely To Vote only:
    Yes: 27% No: 35% DK: 18% -8

  7. That is indeed interesting Calum.

    We need to factor in the drive to get people who have never voted in the past to sign up, though, because there does seem to be some success with this recruitment drive.

    I think the Yes voters will get themselves down to that voting booth come what may, however the negative campaign of BT will leave a bitter taste in a lot of peoples mouth and this will almost certainly result in a lot of natural 'No' people, deciding that they don't feel comfortable voting against Scotland.

    I think it all comes down to the trends at the end of the day, and every one of the polls have shown the trend is for Yes.