Survation sensation as new poll puts SNP on course for OVERALL MAJORITY in the Scottish Parliament

My apologies - I'm really playing catch-up with this latest Survation poll.  I've been away on a long weekend in the Highlands which has just been extended for another night because of bad weather, but I'll get home and post some proper analysis eventually!  In the meantime, here are the voting intention numbers for Westminster and Holyrood.  It's intriguing that this poll was commissioned by the SNP and that they've elected to publish the results in full, because as you can see they're a bit of a curate's egg affair.  The last Survation poll back in the summer was conducted by online panel and this one was conducted by telephone, so a meaningful comparison isn't possible, but I've added in the percentage changes from one to the other in italics for the sheer hell of it, as it's the only comparison available.  On that basis the new results are perhaps a bit disappointing in Westminster terms (although even on these numbers the SNP would be gaining seats), static on the Holyrood constituency vote (albeit static from an outstandingly good position) and mind-bogglingly wonderful on the Holyrood list vote.  But if you think the direction of travel on the list is a bit too good to be true, you're probably correct - if it's not an illusion caused by the switch to telephone fieldwork, it looks like it may have been an illusion caused by other methodological differences.

Scottish voting intentions for next UK general election:

SNP 37% (-5)
Conservatives 28% (+4)
Labour 26% (+3)
Liberal Democrats 6% (-2)

Scottish Parliament constituency ballot voting intentions:

SNP 44% (n/c)
Conservatives 24% (n/c)
Labour 23% (+2)
Liberal Democrats 8% (-1)

Scottish Parliament regional list ballot voting intentions:

SNP 40% (+7)
Conservatives 25% (+6)
Labour 22% (+1)
Liberal Democrats 8% (-2)
Greens 4% (-7)

Other than polling by telephone, Survation have done two things differently that could have contributed to the huge changes on the list vote - they haven't offered UKIP as a specific option this time (thus presumably boosting the Tories) and they haven't referred to the list vote as a "second" vote, which may have helped to bring about an astonishing 7% 'swing' from Green to SNP.   Of course it may just be that telephone respondents are far less into the Greens than online respondents are, but nevertheless it's startling to ponder the possibility that the deletion of a single word might have had a rather big effect.  It obviously raises the question of whether the cause of the SNP's suspiciously poor showings on the list in recent online Survation polls was that many SNP voters were misled into thinking they were being asked for a second preference vote, and were thus plumping for the Greens.

As noted above, the Westminster figures are underwhelming but OK to be getting on with.  The SNP have been as low as 37% in a post-election Survation poll before, and their 42% in the last Survation poll was unusually high, so there's no reason to jump to the conclusion that telephone polls will always show a lower SNP vote for Westminster.  When rounded to one decimal place, the figure is 37.4%, which is still a touch higher than the 36.9% achieved in June 2017.  The SNP's lead over both the Tories and Labour is also slightly higher than it was last year, which is why they remain (just about) on course to gain seats at Westminster rather than to lose them.

As far as Holyrood is concerned, it's been rare in recent times for any opinion poll to point to a pro-independence parliamentary majority, let alone an outright SNP majority.  But arguably, on these figures, the SNP would just about make it over the line on their own.  And that might just be (at least partly) because of the omission of that rogue word 'second'.  Never let it be said that misinformation about how the voting system works doesn't matter, or shouldn't be called out.

UPDATE: Please see Calum's comment below for important information that I'd overlooked (which contradicts some of the above).