Saturday, February 29, 2020

East of Eden



I'm not sure what this says about Stephen's beloved Scottish Tories, because the last time a Tory leader other than Sir Anthony Eden won a general election in Scotland was half-a-century before Stephen was born. In 1935. But hey, that's completely different, yeah?

Wednesday, February 26, 2020

It's "journalism", Jim, but not as we know it : comically amateurish Scotsman article contains multiple wild inaccuracies about the YouGov independence poll

There's still no clarity on whether yesterday's YouGov poll showed a 50/50 split on independence after Don't Knows are excluded, or whether (as some newspapers claim) it was Yes 49%, No 51%.  Neither do we yet know whether 16 and 17 year olds were interviewed for the poll.  I checked the What Scotland Thinks website, which often has access to information that isn't otherwise in the public domain, and it states that only over-18s were polled - which, if true, would cast doubt on the headline numbers and might suggest that Yes have been slightly underestimated.  But a commenter on this blog's previous thread pointed out that there is a discrepancy in the datasets between the total number of respondents and the combined total of respondents from all of the listed age groups.  The most logical explanation is that there are also respondents from an additional age group, which would obviously have to be 16 and 17 year olds.

What I can say for certain, though, is that a Scotsman article about the poll (which has been online for five days because the results on some of the supplementary questions were released early) contains a series of extraordinarily wild inaccuracies.  It's tempting to call them lies, although I suspect they're probably inadvertent blunders caused by either sloppiness or wishful thinking, or possibly by a blend of both.  The writer seems to have allowed himself to be duped by a propaganda press release, or perhaps he just didn't bother to read it carefully enough.  This is one of the offending segments -

"Half of Scots (50 per cent) blame the SNP for the divisions, according to the findings. The prospect of a second referendum is blamed by 41 per cent of respondents, while 26 per cent say everyone bears some responsibility."

Not only is that untrue, it's not even within light-years of the truth.  The YouGov datasets make abundantly clear that only respondents who said that Scotland is divided (57% of the sample) were asked the follow-up question about "blame".  That means only around 28% or 29% of Scots "blame the SNP for divisions" - not 50% as the Scotsman claim.  The prospect of a second independence referendum is actually "blamed" by around 23% of the sample - not 41% as the Scotsman claim.  And only around 15% say everyone is "responsible" - not 26% as the Scotsman claim.

Even the headline of the piece is misleading, to put it charitably.  It states: "Half of Scots believe independence and Brexit division will last generation, finds poll".  Er, nope.  It's not half of Scots, but just under half of the portion of the sample who think the country is divided.  The correct figure for the whole sample is therefore around 26%.

That's misreporting on a truly colossal scale, and there's absolutely no excuse for it.  But on past form, a correction and apology is probably extremely unlikely.

In case you're wondering about the story behind this poll, the datasets imply it was commissioned by Hanbury Strategy, which is described on Powerbase as "a Conservative-led lobbying firm set up by ex-David Cameron adviser Ameet Gill and Brexit campaigner and former British Bankers' Association director, Paul Stephenson in September 2016.  In June 2017, it hired Lizzie Loudon, former press secretary to the Prime Minister, Theresa May".  Strangely, though, when Gordon Brown fronted the release of some of the results a few days ago, it was reported as being a poll for the think-tank Our Scottish Future.  It's pretty clear that it was intended to be an anti-independence propaganda poll of some description, which might explain some of the oddities about it - the non-standard question wording, the highly unorthodox question sequence, and perhaps also the ambiguity over the headline results with Don't Knows excluded.

Tuesday, February 25, 2020

Misery for Boris as another YouGov poll finds a clear majority of the Scottish public demand an independence referendum within five years

I was oblivious to this for most of the day, but the fourth independence poll of the year has been published - and perhaps of most significance is that it's the first post-Brexit poll.  However, it should be stressed that, strictly speaking, it's not directly comparable to the three polls conducted in January.  There are a couple of reasons for that - a) the question asked was non-standard, and b) it appears from the datasets that a supplementary question about whether Scotland is "heading in the right or wrong direction" was asked before the main independence question, which is highly irregular and might conceivably have affected the result by putting respondents into a certain mindset.  For my money, the second problem is far more important than the first, because although the wording of the main question is non-standard, it's not leading in any obvious way.  (That said, it's perfectly reasonable for us to wonder why the client seems to have insisted upon unusual wording.)

YouGov/Hanbury poll:

If another Scottish independence referendum were held today, how would you vote?

Yes to an independent Scotland: 45%
No to an independent Scotland: 46%

Some of the insanely biased newspaper reports of this poll (which are probably lightly rewritten versions of a press release) suggest that with Don't Knows excluded, the figures are Yes 49%, No 51% - but there are no such numbers in the datasets.  If this is simply based on a crude recalculation of the 45 and 46 figures, it may well be inaccurate due to rounding issues, because it comes out very close to Yes 49.5%, No 50.5%.  In other words, until and unless we hear definitively from YouGov, it shouldn't be assumed that No are actually in the lead in this poll on the rounded figures excluding Don't Knows.  It might be a 49/51 split, but it could just as easily be 50/50.

The drop in support for Yes since the last YouGov poll is therefore either one percentage point or two percentage points.  It's not statistically significant either way.  There are three possible explanations for the slight drop.  It could just be random sampling variation (if Yes are on around 51%, you'd expect some polls to put them on 49% or 50% due to the margin of error). The unorthodox question sequence might have distorted the result.  Or there could have been a real but modest slip in support for independence due to the 'Brexit lull' - ie. in some voters' minds, Brexit is 'done' and nothing disastrous seems to have happened, but in reality the cliff-edge is looming at the end of this year when the transitional period finishes.

As things stand, though, an average of all four polls this year continues to show a slight pro-independence majority.

Contrary to the impression you might have got from the press relea....sorry, original newspaper reporting, the poll actually detected considerable enthusiasm for a second independence referendum.  55% of respondents want it to take place within the next five years, and 40% want it by 2022.

UPDATE: It's far from clear whether 16 and 17 year olds were interviewed for this poll.  The datasets for the last YouGov poll specified that over-16s were the base, but this time it just says "2587 Scottish adults".  If by any chance 16 and 17 year olds were excluded, the media narrative about this poll would be completely bogus, because the Yes vote may well be underestimated by 1%.