Wednesday, January 17, 2018

Body blow for hapless Leonard as Labour slip back in first Scottish poll of 2018

Many thanks to Stuart Dickson for alerting me to the first full-scale Scottish poll of the New Year, conducted by YouGov for the Scottish edition of The Times.

Scottish voting intentions for the next Westminster general election (YouGov, 12th-16th January):

SNP 36% (-4)
Labour 28% (-2)
Conservatives 23% (n/c)
Liberal Democrats 6% (+1)
Greens 3% (+2)
UKIP 3% (+2)

Scottish Parliament voting intentions (constituency ballot):

SNP 38% (-4)
Conservatives 26% (+1)
Labour 23% (-2)
Liberal Democrats 7% (+2)
Greens 3% (+1)
UKIP 2% (+1)

Scottish Parliament voting intentions (regional list ballot):

SNP 32% (-3)
Conservatives 25% (+2)
Labour 22% (-2)
Greens 10% (+4)
Liberal Democrats 7% (+1)
UKIP 3% (+2)
SSP 2% (-1)

It may seem obvious that a drop of four points for the SNP is significant, but it's impossible to know that for certain.  If, for example, support for the party in Westminster terms has remained steady at around 38%, the margin of error could have flattered them by two points in the last YouGov poll in October, and understated them by two points in this poll, thus producing an entirely illusory four-point shift.  It's also conceivable that there has been a genuine drop, but that margin of error effects are exaggerating it.  Certainly there was no sign at all of the SNP going backwards in the Survation poll conducted in early December, so I'd be more inclined to the view that nothing much has changed - at least until we see another poll confirming the trend reported by YouGov.

Should Scotland be an independent country?  

Yes 43% (-1)
No 57% (+1)

The Times' interpretation of the above finding is ludicrous to the point of being almost embarrassing - they claim that support for independence has "dwindled", but in fact a 1% drop is of no statistical significance whatever in a poll with a margin of error of 3 points.  The 43% share for Yes is firmly within the 'normal range' produced by recent YouGov polls - indeed the last-but-one YouGov poll had Yes on exactly 43%.  So this is essentially a no change result, and categorically not a "setback for Yes".

We'll have to see the datasets to be sure, but the likelihood is that YouGov have persevered with their reprehensible practice of excluding 16 and 17 year olds from their independence polling, which leaves open the theoretical possibility that the reported Yes vote is 1% lower than it should be (after rounding).

Much is being made of the finding that 36% of respondents want an independence referendum within the next five years, and 54% don't - but that just appears to be a 'house effect' of YouGov's polling.  They've been asking that question for quite a while and have always produced a negative result, in complete contrast to the 50/50 splits that have often been reported in Panelbase's polling on whether there should be an independence referendum within as little as a couple of years.  We can only speculate as to whether YouGov's panel is for some reason more hostile to a referendum than Panelbase's, or whether there's something about the way YouGov pose the question that produces such markedly different results.

In fairness to The Times, it's not just the SNP and the independence movement they're spinning against - they're also reading far too much into a small drop in Labour support that may or may not prove to be genuine.  However, one detail from the poll can't even conceivably be explained away by the margin of error - Jeremy Corbyn's net personal rating has dropped catastrophically from +20 in October to -3 now.  I would imagine that has been caused quite simply by the fact that we're three months further away from the hoo-ha of the general election campaign, and that people are gradually reverting to the view they held of Corbyn before the Labour surge during May and June.  The million dollar question is whether they would once again swing to a more favourable opinion in the heat of a general election campaign - and on the answer to that question may hang the fate of several SNP-Labour marginal seats.


  1. Is the Corbyn rating for Scotland or the UK? If Scotland then I think it eases the SNP's position considerably. Move the troops to the Eastern Front!

  2. - ”one detail from the poll can't even conceivably be explained away by the margin of error - Jeremy Corbyn's net personal rating has dropped catastrophically from +20 in October to -3 now”. Yes, that is the one and only significant finding (so far: we haven’t seen the full tables). And it is counterintuitive. I must admit I’m a bit perplexed. Could it be his fence-sitting on Brexit that is damaging his reputation?

    1. That's probably part of the explanation. He clearly got some of the anti brexit vote but has now gone back on it.

  3. He is proving to be ineffective in opposition. His apparent "wait and see" approach on Brexit alienates both sides.
    So many open goals and he has failed to score. Starmer and Gardner look pissed off.

  4. The only significant change in the leader ratings is Jeremy Corbyn’s plummet. Theresa May is at Gordon Brown levels of unpopularity. Ruth Davidson is still popular, but only mildly so: several other Scottish leaders have had much higher ratings in recent decades. Nicola Sturgeon is doing fine, eleven years in to government, which always takes its toll. Any other 11 year old government would be delighted with these levels of support, and we ought to be too.
    Richard Leonard is still, unsurprisingly, largely unknown by the general public: 60% of respondents felt that they did not know enough about him to rate him (compared to only 13% who did not respond to the Sturgeon rating question).

    Leader ratings (net):
    Ruth Davidson +15 (-2)
    Nicola Sturgeon +/- 0 (n/c)
    Jeremy Corbyn -3 (-23)
    Richard Leonard -24 (n/a)
    Theresa May -47 (+2)

  5. James, can you provide an average Poll of Polls for the primary questions asked? It would help put these kinds of polls in perspective if there was a running average of support for independence / support for a referendum in the next 5 years, etc.

  6. Why, after discovering that Westminster election weighting is inaccurate in Scottish polling, are pollsters back doing it again?

    Holyrood VI should be weighted to the last Holyrood election, not to 'UK union elections'. We don't weight English general election VI polls to e.g. EU elections for good reason.

    Anyway, renders Holyrood data rather useless as we can't actually judge if anything has changed; current polls are not referenced against the last election. By the looks of the data in the wings panelbase where we can see Holyrood 2016 recall, if you correctly weight to May 2016, not much has changed at all it appears. Quite the contrast to what 2017 weighting suggests.

    Way too much tactical voting happens in UK union elections; that's why recall is still bad for them. That was the problem ahead of 2014 and hasn't gone away.

  7. One of the news papers reporting on the Westminster poll over projected for the Unionist parties with Labour winning 17 seats and the Tories 14. If you put the figures into electoral Calculas you get 14 Labour and 11 Tories. Instead of gaining Perth the Conservatives loose Stirling and Gordon to the SNP. The SNP still win a majority of 30 seats and not a minority of 27.