Sunday, April 19, 2015

Britain-wide YouGov poll confirms Nicola Sturgeon defeated Miliband in the BBC leaders' debate

Today's Britain-wide YouGov poll is the first to have been entirely conducted since the BBC leaders' debate on Thursday night.  There's a supplementary question about the debate itself, which shows that Nicola Sturgeon was judged to be the winner both by respondents across the whole of Britain, and by the Scottish subsample - although understandably her margin of victory was much, much bigger with the latter.

Which leader do you think came across best in the debate?  (Respondents across Britain) :

Nicola Sturgeon 30%
Ed Miliband 27%
Nigel Farage 17%
Leanne Wood 3%
Natalie Bennett 3%

Which leader do you think came across best in the debate?  (Respondents in Scotland only) :

Nicola Sturgeon 63%
Ed Miliband 22%
Nigel Farage 8%
Natalie Bennett 1%
Leanne Wood 0%

This isn't an instant poll, so respondents will have been influenced by the post-debate reporting, including the widespread reporting of the rather questionable instant poll from Survation that had Miliband slightly ahead (albeit only at Britain-wide level).  So perhaps the most important thing about YouGov's result is the reassurance it provides us that the Survation poll didn't distort people's perception of the outcome - or at least not sufficiently to convince a plurality that Miliband won.

The headline voting intentions from YouGov's Scottish subsample are also of particular interest, as they're the first numbers we have that are entirely post-debate.  They show : SNP 40%, Labour 27%, Conservatives 19%, UKIP 6%, Liberal Democrats 6%, Greens 2%.  If that was the result of a full-scale poll, it would look like the gap had narrowed, but in subsample terms it's very much business as usual, and doesn't give rise to any obvious suspicion that the debate has changed the state of play.

*  *  *

Here's a damning verdict on the leader of Plaid Cymru from one member of the YouGov panel -

"Uniformed spreader of misinformation, stupidly taken in by Nicola Sturgeon."

Hmmm.  Let's be fair to Leanne - she may be a gullible spreader of misinformation, but at least she has the common decency to do it in a uniform.

14 comments:

  1. Ed Balls to Murnaghan: "We're not going to start getting involved in coalitions OR DEALS with a party that wants to break up the United Kingdom."

    ReplyDelete
    Replies
    1. Yeah but that still, tellingly, doesn't rule out vote by vote support, which is the most likely outcome. To really shaft themselves they'd have to say, we won't rely on SNP votes ever under any circumstances...

      Delete
  2. I agree the sub-sample is suspect. I think any Scottish poll that shows combined support for pro-independence parties below 45% is very suspect.

    ReplyDelete
    Replies
    1. The long term average for the sub-sample scores on YouGov is something like SNP 42, Labour 27. Anthony Wells of YouGov has said in the recent past that the typical YouGov panel has too few people born and living in Scotland relative to the actual demographics, which has the effect of decreasing the SNP scores in those polls by 3-4 points. They adjust for this factor in their Scotland-only polls.

      Delete
  3. Speaking to my parents today, and they said that they had McGovern canvassing at their door yesterday. Too ill to stand again in Dundee West, but well enough to canvas for his replacement?

    ReplyDelete
    Replies
    1. It is possible, you know, depending on the nature of his health problems.

      Delete
    2. Jesus dude, cut the guy some slack. 'Recent medical advice' results in him walking away from a 5 year job just weeks before an election, while still outwardly healthy?

      I'll give you pretty short odds he's looking at cancer/chemo or something equally horrible and degenerative.

      Delete
    3. Comments in The Courier suggest he may be increasingly tired and emotional.

      Delete
  4. Millibland cant even win a LabGov Poll, oh dear me! I am sure I saw the latest LabGov poll giving the Red Tories a 3 point lead over the Blue Tories.

    ReplyDelete
  5. That LabGov panellist is really bitter :)

    ReplyDelete
  6. With that accent, Leanne Woods could sell me double glazing, and I already have double glazing!

    ReplyDelete
  7. I have done a little analysis of the Ashcroft polling.

    The average percentage shares across the 32 polls (5 seats polled twice) is:

    SNP 44
    Lab 28
    Con 13
    LD 9
    Others 6

    This actually flatters the Lib Dems and Labour (understates the SNP) a bit. This is because Ashcroft has not bothered to poll the six seats the SNP are defending, obviously because they are unlikely to change hands.

    By then working out the average 2010 scores in the seats polled and comparing it to the national 2010 scores, we can then get a rough and ready idea of how it translates into a national poll share now. Something similar was done with the ComRes "Labour seats only" poll, which came up with a score of SNP 45, Lab 29.

    Using these adjustments, we get the following estimated national shares from all of the Ashcroft polls:

    SNP 48
    Lab 25
    Con 13
    LD 8
    Others 6

    There is some variation between the three batches of Ashcroft polls. The 'best' set for the SNP was the most recent one (50%) and the 'worst' the ones released in early March (45%). Although these differences may simply be due to sampling variation and/or the types of seats Ashcroft polled. e.g. the first batch had a lot of Labour seats, the second batch had seats with some Tory strength, and the third had a lot of Lib Dem seats.

    The main thing I would take from this is that the Ashcroft polls are not out of step with the national polls, as some of the SNPout lunatics on twitter have suggested. The most recent TNS poll and both of the Ipsos Mori polls have shown bigger SNP scores and leads than what I estimate from Ashcroft.

    ReplyDelete
    Replies
    1. Although Ashcroft only gives such "low" (ahem) figures for the SNP because of the 2010 weighting distortion - if he used the Ipsos-Mori approach of no past vote weighting at all, he'd be putting the SNP quite a bit higher.

      Delete
    2. It's impossible to quantify though exactly how much that is worth though, i.e. how much of the adjustment is due to genuinely having too many 2010 SNP voters in the sample and how much is due to voters falsely recalling how they voted (confusion with 2011).

      Delete