Let's start with a detail from yesterday's new full-scale Scottish YouGov poll that may not be the most important, but is certainly the most amusing. As we all know, pretty much the sole selling-point of Owen Smith's campaign to become Labour leader is that he is supposedly much more "electable" than the incumbent Jeremy Corbyn. Well, this poll begs to differ -
If you had a vote in the Labour leadership election, who would you vote for?
Jeremy Corbyn 27%
Owen Smith 25%
Owen Smith 25%
Those figures are for the whole electorate. Ironically, among the rump Labour vote, Smith does a little better (he's tied with Corbyn at 35% apiece), which gives the lie to the idea that he's better-placed than Corbyn to grow Labour's support. At least in Scotland, Smith is very much the 'comfort zone' candidate for Labour, whereas Corbyn is the more attractive candidate for people currently minded to vote SNP. 42% of SNP voters (who of course make up roughly half the electorate) would back Corbyn if they had a vote in the leadership election, and just 20% would vote for Smith. It's true that Corbyn has so far failed to make good on his suggestion last year that he could actually coax those SNP voters back to the Labour fold, but based on these numbers, the idea that Smith would have a better chance of doing so is risible in the extreme.
Turning now to the meat of the poll...
Journalists always (and rightly) get a good roasting when they make misleading claims about opinion polls for the sake of a good headline, or to push some political agenda. It's therefore only fair that YouGov themselves should get a similar roasting when they do exactly the same thing. Matthew Smith's article on the YouGov website about this poll makes two extremely dodgy claims -
1. "Scots don't support a second independence referendum...With the SNP set to relaunch their campaign for independence, 50% of Scots oppose a second referendum."
In fact, the poll didn't ask about the general idea of a second referendum. Respondents were instead asked whether they support a referendum before the UK leaves the EU, which could of course be in the very near future. There will be people who answered 'no' to that question because they don't favour a referendum until 2021 or whenever, but who have been wrongly categorised by YouGov as opposing a second referendum, full stop. If YouGov want to make such sweeping claims without being criticised, they shouldn't ask such narrow questions.
2. "Should [the SNP government] be successful in forcing another vote, the results would be almost identical to last time, with 54% of Scots voting against independence and 46% in favour."
The question on independence specifically asks how people would vote if there was a referendum "tomorrow". The chances of the SNP getting the requisite legislation through by teatime today do seem rather remote. It's disappointing to see YouGov fuelling the lazy journalistic myth that polls are predictions, rather than snapshots of public opinion at a given moment in time (and indeed snapshots which may not be entirely accurate).
On their Twitter account, YouGov also make a third claim -
"Ruth Davidson is now more popular among Scottish voters than Nicola Sturgeon"
To be fair, that's a fractionally less outrageous statement, because there is a convention of headlining the 'net satisfaction' figures for political leaders, ie. subtracting the percentage of voters who have a negative view of the leader from the percentage who have a positive view. If you do that, Ruth Davidson's satisfaction rating is indeed 1% better than Sturgeon's (a trivial difference which is of course well within the margin of error). But the fact remains that Sturgeon is significantly more popular than Davidson in absolute terms - 53% think Sturgeon is doing a good job, compared to 46% for Davidson. It's the negative ratings that swing the balance, and given that Sturgeon is actually in government making decisions, it's hardly surprising that she arouses stronger feelings than Davidson does among those who don't like her. But bearing in mind that the target figure for an absolute majority under our electoral system is the mid-to-high 40s, the question has to be asked : if 53% of voters like you, does it really matter that much if 33% of voters don't?
On all measures, Sturgeon is considerably more popular than Theresa May, in spite of the fact that May is still enjoying her honeymoon as Prime Minister. Just 35% of the electorate think that May is doing a good job - 18% lower than the figure for Sturgeon. And May's net satisfaction rating is 7% lower than the First Minister's.
Alas, I'm not done with dodgy claims about the poll yet. Step forward the maestro : Mr George Eaton of the New Statesman -
"Scottish independence support falls again: No side ahead by 54-46"
What? I mean, what?! There have been five credible polls on independence since the EU referendum : an online Panelbase poll showing a significant increase in support for independence, telephone and online polls from Survation which both showed a significant increase in support for independence, an online YouGov poll showing a modest 1% increase in support for independence, and then yesterday's online YouGov poll showing that the 1% increase had been reversed, returning us to the position in the poll before last. On what planet does that sequence of results justify the statement "Scottish independence support falls again"? Answers on a postcard, folks. Eaton is either cynically intending to mislead, or just doesn't have a clue what he's talking about - I'm struggling to see a third option.
Can YouGov's figures of Yes 46%, No 54% be regarded as reliable? As I alluded to in my TalkRadio article yesterday, every pollster has its own 'house effect', and there's no guarantee that other firms will report the same basic trend, let alone the same headline numbers. One point about YouGov's methodology that has become increasingly controversial is their failure to include 16 and 17 year olds in the polling sample. You could call this institutional inertia, but really it boils down to Anglocentricity. If the Westminster parliament had reduced the voting age to 16, it's unthinkable that YouGov would still be conducting Westminster voting intention polls without interviewing 16 and 17 year olds. It's unlikely in most cases that this deficiency will lead to the Yes vote being underestimated by more than 1%, but when the contest is so evenly balanced, even that can make a big psychological difference. By definition, the Yes vote being 1% too low means that the No vote is 1% too high, so it's perfectly possible that the No lead is being overestimated by a full 2% in the new poll - even assuming the methodology is otherwise correct. YouGov might shrug their shoulders and say that this is a relatively minor flaw, but if they're not polling the correct electorate and some of their competitors are, we're entitled to point out that their results should be regarded as comparatively unreliable.