Thursday, July 3, 2014

More on Kellner

Just to let you know I've written an article for Newsnet Scotland about this week's ongoing YouGov saga - you can read it HERE.  That's what I was doing last night instead of writing the detailed commentary I'd promised on Kellner's blogpost!  But I'm delighted to say there's no shortage of such detailed commentary from others - I can recommend Survation's very forceful response, Scottish Skier's comments, and of course the piece by Alasdair Stirling that was published here earlier this afternoon.

20 comments:

  1. Kellner appears to have lost the plot. Why aggressively risk so much credibility on a referendum that is inherently unpredictable?

    Good on you for keeping the pressure on.

    In general I've long thought that "he who pays the piper" might apply a little too much to some polling.

    And agree with others who say polling should be banned during campaigns / vote run-ins. Far too often it sets, rather than reflects, the agenda.

    All that said, it does seem to me that the gradual narrowing of the yes / no gap has stalled somewhat recently.

    Hugh.

    ReplyDelete
  2. A very comprehensive and informative roundup James.

    I shall repost this particular piece of information as it really does need emphasising as this extraordinary saga plays out.

    John Curtice is the current President of the British Polling Council.

    Yet what has his response been to these completely unprecedented events in his industry?

    Silence.

    To be fair, that's not quite the whole story.

    Curtice was more than happy to be quoted in Murdoch's Times article featuring YouGov's polling using it's controversial methodology.

    It may or may not come as a surprise to readers of ScotGoesPop that Curtice used the YouGov polling to proclaim almost certain doom for the Yes campaign. The fact that the Times comprehensively misrepresented that polling seem not to have bothered Curtice a jot. (somewhat unsurprisingly it didn't seem to bother Murdoch's clients Kellner and YouGov either)

    Something else for ScotGoesPop readers to ponder is that the phrase Curtice used, that Yes "needs a gamechanger", has been pounced upon by the No campaign and splashed all over the unionist friendly papers massive coverage of this one MOE YouGov poll using YouGov's eccentric and questionable methodology.

    Surely one BPC pollster forced to defend itself against another (who commissioned their own research solely for the purpose of criticizing another polling company’s methodology) is a small price to pay if the President of the BritishPollingCouncil can get widely quoted in all the papers?


    Not to mention even more time on the BBC as Curtice profile and status as 'celebrity psephologist' rises.

    ReplyDelete
  3. High : Well, we don't know. Survation and Panelbase both showed the gap narrowing to the lowest point ever in their most recent polls, so there are various possibilities - YouGov could be wrong and the gap has continued to narrow, they could be right and the previous trend has gone into reverse, or the truth could be somewhere in between the two extremes, in which case nothing much has changed.

    My instinct is to agree with you about banning polls during the campaign, although of course it wouldn't be wholly effective in the internet age. But it might help a bit.

    ReplyDelete
  4. Are we talking about the future Lord Keller here?

    ReplyDelete
  5. RE Survation's response...

    I think if they were entered into the 'Tear someone a new one as politely and eloquently as possible' competition, they'd be in with a good chance.

    ReplyDelete
  6. Juteman, you may well think that but I could not possibly comment. Actually I could and I will.

    It was reported (by Michael Crick) that Cameron was actively considering making a certain Patrick Rock a Lord before things went, well, pear shaped doesn't quite cover it, does it?

    So I will never be surprised by who is or is not made a Lord from now on.

    Let's face it, with some 800 Lords and counting I'm only surprised Cameron and Clegg aren't handing out Peerages on a "BOGOF" basis. ;-) That was a joke of course since we know there is not now and has never been any 'funny business' whatsoever involving Lordships. Apart from the Lavender List unpleasantness obviously. And some minor "Cash for Honours" kerfuffle involving the Saintly Tony Blair that was of course unproven and given a clean bill of health by the CPS. (If by 'clean bill of health' we mean the CPS stating "that while peerages may have been given in exchange for loans, it could not find direct evidence that that had been agreed in advance".)

    Where was I? Oh yes, Peerages are in no way an anachronistic laughing stock and the very notion that they are given out as sweeteners, for favours done or even cash is a vile falsehood. You should be ashamed of yourself Juteman!

    ReplyDelete
  7. So let's see, unionists have attacked panelbase (back in 2013 over the 'new sign-ups' issue etc).

    Then they attack survation and by association (given they use similar 2011 weighting) ICM and panelbase again.

    All this done in 'public' no less.

    Do I detect panic?

    ReplyDelete
  8. Poll of Polls without the inclusion of YouGov would be interesting.

    ReplyDelete
  9. Mick Porks last Comment about the lords, haha!!

    Narrowing of the polls: The thing that has struck me is that in all the debates we see in which before and after polls are taken Yes gains and gains a lot.

    This is why BT have been pulling out of debates and why they with the help of the MSM are trying to ensure people don't hear both sides of the debate, by filling the airwaves with stories of nastiness and cyber attacks.

    However the grassroots campaign trundles on like a big happy steamroller and I see people on twitter every single day announcing that they have went on the journey from No to Yes and have now joined local groups etc.

    The problem isn't that the gap (if there is one, which I'm not sure about) is narrowing, the problem is that we have a MSM that is rarely owned by Scots and if it is is owned by very wealthy Scots, and which is unashamedly part of the BT campaign in all but name, so we aren't getting any positive information about how the campaigns are going as this would trigger the one thing that the Politicians know is unstoppable and that's the Yes campaign creating a positive momentum.

    One thing the referendum has taught me, is that we are not living in a genuine democracy as I ever understood it.

    ReplyDelete
  10. LOL if this is true, which it would seem it is.

    Survation (feb 2014) unweighted recalled 2010:
    18% Con
    32% Lab
    14% Lib
    33% SNP
    3% Other

    2007 Holyrood election result (adjusting for the higher 'other' in Holyrood):
    16% Con
    32% Lab
    14% Lib
    33% SNP

    It seems people are recalling who they 'solidly supported' in 2010, not who they 'voted for'.

    Two different things of course for a tactical voter.

    This agrees well with Holyrood vs Westminster polls in 2009 too.

    That's funny as hell. Scotland and FPTP.

    The ironic icing on the cake is this:

    which party they would support, or had supported, in the 2010 General Election. Data are weighted to 2010 party identity wherever this information is available. The weights used for party identity are consistent with the outcome of the 2010 General Election.

    From Yougov.

    ReplyDelete
  11. Great find scottish_skier. I must admit the idea that there was little to no tactical voting for a westmisnter GE is one of the more preposterous notions I've heard. It's blatantly obvious it happens. The fact that there are still more pandas than scottish tory MPs is wee bit of a giveaway as well. YouGov and Kellner are making a complete laughing stock of themselves so they had better get used to their referendum polling being laughed at from now on as well.

    Speaking of funny as hell, there seems to have been some incompetent fop of a PM trying to extoll the virtues of westmisnter governance in scotland on the day his chum and spindoctor Coulson was sent to prison for 18 months.

    'Better together' Cammie? So when will you be visiting your chum Coulson in jail? Hmmm?

    So much for that politicalbetting non-story, eh James?

    *chortle*

    Thurlbeck and Miskiw were also sentenced to 6 months in prison. Thurlbeck is a world class buffoon who was actually extolled on PB by the likes of Plato and others while I pointed out just how bizarre and foolish his actions and outbursts were as the hacking trials got ever closer.

    Speaking of trials there are 11 or so still to come for hacking so anyone deluding themselves that this is all over had better think again.

    I saw this on twitter which is well worth repeating.

    "In the ultimate irony, prison officers will be monitoring Andy Coulson's phone calls..."

    Indeed.

    ReplyDelete
  12. "The existence of three polling firms that currently show Yes within a hair's breadth of victory is plainly a huge threat to the Kellner narrative, and his blogpost appears to be an attempt to bully (or perhaps 'spook' is a better word) those firms into adopting a more No-friendly methodology. At least that way there would be safety in numbers for Kellner and YouGov, with all of the pollsters standing or falling together."


    I have to say, there's an awful lot of speculation going on there. Kellner has always been quite bullish about the accuracy of his polling. If he thought he'd put his foot in it then he'd change YouGov's methodology, not go down guns blazing by pointlessly trying to encourage other polling companies to change theirs.

    YouGov might be hopelessly wrong (nobody will know for sure until the vote) but I don't doubt for a minute that Kellner thinks he's getting it right. He'd be setting himself up for an awful big fall if he didn't - and for what? It's not as if artificially inflating the No vote would actually help the No camp to win. As you rightly stated in another post here, giving Better Together the impression they're further ahead than they are is far more likely to just breed complacency and hurt the mobilisation of No voters on Election Day.

    ReplyDelete
  13. "It's not as if artificially inflating the No vote would actually help the No camp to win."

    That's fundamentally wrong. The whole reason for the propaganda war over the polls is that a perception of a decisive No lead would squeeze the life out of the whole campaign in the media, and lead to undecideds and soft Nos thinking that they don't need to take the vote seriously, because the whole thing is just a formality and that we already know the result (the latter is something that - jaw-droppingly - Kellner has actually SAID).

    So, yes, Kellner could well be hoping (whether consciously or subconsciously) to make this a self-fulfilling prophecy.

    As for your earlier point about how Kellner would alter his methodology if he thought he was getting it wrong rather than bluffing his way through, how do you square that with his wayward claim on the BBC results show that Labour were winning the popular vote in Scotland in the Euro elections? I certainly haven't seen him admit to that error, let alone apologise for it - like YouGov's inaccuracy in the 2011 election, it's something we're just supposed to set aside when considering the credibility of their other claims.

    OK, it was a simple arithmetical blunder and we've all made those at times, but I find it hard not to suspect that he made the error partly because he wanted to believe that Labour were winning, and because it fitted a narrative he was all too keen to push.

    ReplyDelete
  14. "As you rightly stated in another post here, giving Better Together the impression they're further ahead than they are is far more likely to just breed complacency and hurt the mobilisation of No voters on Election Day."

    Just to clarify, it wasn't me who said that - it was a guest post by Alasdair Stirling.

    ReplyDelete
  15. I think the final point I'd make is that I'm not actually suggesting Kellner has privately concluded that he's getting it wrong. I think he wants to believe that he's getting it right, and the strength of his previous assertions have left him with nowhere else to go. It probably was genuinely a shock to him that the SNP didn't make a Devo Max question a deal-breaker, and that's mucked up his narrative.

    The end result is him making declarations of absolute certainty that are utterly irrational and unrealistic in this most uncertain of campaigns.

    ReplyDelete
  16. From what I can see, the Kellner - 'I understand the minds of Scottish voters when nobody else does and they still love Labour - correction doesn't make a huge difference to the final result.

    What it would do if you removed it is push No down below the 50% mark and Yes up closer to 40, although this is very hard to calculate due to Yougov hiding so much information in their tables against standard industry practice. I can see why keeping No above 50% before DK are excluded is attractive due to the symbolism of it.

    As we saw in Yougov's own poll, the Kellner correction essentially means voter recall for the final weighted panel is actually wrong for both 2010 and 2011, falling somewhere in between, with SNP voters suppressed and Labour + Labour/SNP No voters up-weighted in both cases.

    The massively skewed (to England) country of birth problem in panel respondents should be adding to this and combined, in a binary polarising question, creates a much bigger gap than other pollsters are getting.

    What I said in my earlier post does seem to neatly explain what is happening with 2010. If you ask people about previous elections you'd think they are far more likely to tell you who they supported / ID'd with rather that what they did in terms of a tactical FPTP vote, particularly if that failed to achieve what was hoped for. We must remember that the average person will not be thinking 'Ah, they're asking this so they can correctly re-weight respondent vote recall to actual election patters so I should...'. Instead, people see 'who did I support / want to win in 2010'.

    Again Yougov's own data support this. The Kellner correction balances recall for Labour / SNP (making both incorrect as noted), but doesn't for (tactical) Libs where the recalled 2010 is still 6% too low, just like Survation etc.

    ReplyDelete
  17. @James "Just to clarify, it wasn't me who said that - it was a guest post by Alasdair Stirling."

    In that case I'll give Alasdair the credit for it then as it's still a sensible comment. You've presented one reason here why it might be hypothetically in the No camp's interest to skew polls intentionally. You've said it's "fundamentally wrong" to state that it's likely to do more harm than good for the No camp.

    I think at best that case depends on the circumstances. In certain cases skewing the polls intentionally might help (e.g. if Yes was polling incredibly low numbers then the bottom might fall out of the campaign) but it can just as easily hurt the No campaign in other situations (e.g. it's actually neck and neck, but the polls are giving No a relatively comfortable lead). We're far closer to the latter situation in this case - I think YouGov probably is unintentionally inflating the No vote and that it is closer than what's being suggested.

    Your point about Kellner believing he's getting it right through wishful thinking might well be true, but it's somewhat different to the quote I posted above where you're suggesting he's trying to bully other companies into adopting his own methodology (thereby meaning there's "safety in numbers" if he gets it wrong). That's fundamentally inconsistent because if he genuinely thought he was getting it right then the last thing he'd want to do is encourage his competitors to steal his more accurate (in his eyes) methodology. He's running a business ultimately and having a more accurate polling record than his rivals is the only real guarantee of future contracts.

    I think it's perfectly possible that YouGov are getting it wrong (and Kellner's bluster in that regard probably is going to come back and bite him) but I have a natural aversion to any suggestion of intentional polling bias. You might not be making that argument explicitly, but I can see in the comment section at Newsnet Scotland that it's absolutely what some people believe.

    ReplyDelete
  18. "but it can just as easily hurt the No campaign in other situations (e.g. it's actually neck and neck, but the polls are giving No a relatively comfortable lead). We're far closer to the latter situation in this case..."

    I'm really struggling to see how you think that scenario harms No. It's actually precisely the scenario I was talking about earlier that could help No - although fortunately it's not so much of a problem as long as other pollsters are showing a tight race. But obviously that would be threatened by Kellner getting his way and other pollsters falling into line with YouGov's No-friendly methodology.

    "Your point about Kellner believing he's getting it right through wishful thinking might well be true, but it's somewhat different to the quote I posted above where you're suggesting he's trying to bully other companies into adopting his own methodology (thereby meaning there's "safety in numbers" if he gets it wrong). That's fundamentally inconsistent..."

    No, it's not. I am suggesting Kellner is being dishonest, but only about the level of certainty he feels. There's no inconsistency in suggesting that he's hoping that he's right, but would feel a lot more relaxed if he had a consensus backing him up. In any case, you don't accept the premise of my argument about how a polling consensus could suck the life out of the campaign and thereby affect the final result, which in my view is another likely motivation for Kellner - he's trying to turn his prediction into a self-fulfilling prophecy.

    "I have a natural aversion to any suggestion of intentional polling bias. You might not be making that argument explicitly, but I can see in the comment section at Newsnet Scotland that it's absolutely what some people believe."

    I may not agree with every last dot and comma of those remarks, but people are perfectly entitled to make them. By stating explicitly that the result of the referendum is a foregone conclusion, Kellner is now acting in the interests of the No campaign, and YouGov can no longer be regarded as taking a neutral approach. Nobody has forced him to play the cowboy, but it does have consequences.

    ReplyDelete
  19. If I had a few k I'd get yougov to do a Y/N poll using 2011 recall as remembered right now and weighted to CoB to ensure correct demographic representation.

    Hell, Mr Kellner could do this himself easy enough. Something I would do as a scientist to test my theory anyway. It would give us the results of standard industry practice vs the kellner correction + 'nationality isn't important in a nationality referendum' approach for comparison.


    ReplyDelete
  20. Note that Mr Kellner has probably already done the above / has data recent enough to do it.

    In fact having met many of the type of scientist in the course of my career who try endlessly to prove their own theories rather than, more correctly, attempting to disprove them, I suspect this Kellner correction may be a result of him having done this. His article did have a 'I must be right' feel to it.

    That is in the end how good science works. You have a theory, and if you've tried everything to disprove it with no success, then it's probably correct. Finding evidence to support a theory in contrast is much easier as you can readily grab bits of data from here and there that do ostensibly agree with what you propose whilst overlooking some anomalies; correlation not always implying causation of course.


    ReplyDelete