Survation, 20th-22nd January 2020

I feel like every other time I do a post about a poll, I’m complaining about the bus-like nature of Scottish polling. We just got two different poll releases on the same day, because apparently it would kill people to try and stagger these things. Survation was conducted and released first, so we’ll start there. We haven’t actually had a full Survation poll since March last year, so this is a welcome bit of diversity in terms of polling agencies. Perhaps given the UK General Election wasn’t very long ago, there’s no Westminster VI in this.

Display format;

Holyrood Voting Intention and Projection (Tracker)


  • SNP – 38% (+6 / -4)
  • Conservative – 21% (-1 / -2)
  • Labour – 19% (nc / nc)
  • Green – 9% (-2 / +2)
  • Liberal Democrat – 9% (-2 / +4)
  • Brexit – 2% (+2 / +2)
  • UKIP – 1% (-2 / -1)

It’s been so long since Survation did a poll that they completely missed the rise and fall (in projection terms) of the Brexit Party, who come out with their now-usual “no seats” share. Compared to the last poll, the SNP are up at basically everyone else’s expense, but that still puts them below their 2016 figure whilst the Greens and Lib Dems are higher. When you compare this to the Panelbase poll, the performance of the SNP and Conservatives is notably lower versus stronger results for Labour, Greens and Lib Dems.


  • SNP – 51% (+8 / +4)
  • Conservative – 23% (-1 / +1)
  • Labour – 17% (-5 / -6)
  • Liberal Democrat – 7% (-2 / -1)

Survation are generally quite bad for even prompting for the parties outside the four with Westminster seats, so bear that in mind here. In any case, that is a remarkable share for the SNP, especially in the context of being down on the list vote compared to 2016. That’s largely at Labour’s expense, who are down substantially no matter which way you look at it. The difference with Panelbase is much less stark here too.

Translating that into seat terms it might look like;

  • SNP – 64 (+5 / +1)
  • Conservative – 24 (-2 / -7)
  • Labour – 24 (+2 / nc)
  • Green – 9 (-2 / +3)
  • Liberal Democrat – 8 (-3 / +3)

The inherently awkward nature of AMS is evidenced in the Conservatives and Labour projecting out to a tie in seats despite the latter being behind on both votes. Notably, the SNP fall one seat short of a majority. Even with a shade over half of the constituency vote, their much lower list vote in this poll would be a barrier. That gap naturally means there is a stonking overhang here, which is how the Greens drop one of their Lothian seats, as a consequence of Labour and the Lib Dems losing the constituencies they hold in Lothian. Correcting for the overhang in New Zealand style would add 9 seats to Holyrood – 5 more for the Conservatives, 2 for Labour, and one apiece for Greens and Lib Dems.

As ever, to demonstrate how AMS is imperfectly proportional, we can run the list vote (the proportional element) through a more proportional Scandinavian style system (vs AMS projection);

  • SNP – 51 (-13)
  • Conservative – 28 (+4)
  • Labour – 26 (+2)
  • Green – 12 (+3)
  • Liberal Democrat – 12 (+4)

Whereas there’d be a very comfortable pro-Independence majority under AMS, here they’d come a couple of seats short. Given the truly enormous gap between the SNP’s polled support between votes, it’s worth emphasising more than usual that in actual implementation of such a system, voter behaviour would be quite different.

Constitutional (Independence Tracker)

Changes for this one are versus the 10th-11th of December, as Survation did poll Independence just before the snap GE.

Usual Independence question here;

  • Yes – 45% (-1)
  • No – 45% (-2)
  • Don’t Know – 10% (+3)

Excluding Don’t Knows;

  • Yes – 50% (+1 / +5)
  • No – 50% (-1 / -5)

So a dead heat on the big constitutional issue in this one. That just for fun universal swing projection from 2014 comes out at 17 councils with a No lead and 15 with a Yes.

As it didn’t have attached parliamentary VI there wasn’t a full post about it, but a YouGov conducted around the same time came out 51-49 after Don’t Knows were excluded. And the Panelbase that’ll be covered in the next post came out at 52-48. That’s more movement than has typically been seen since the 2014 referendum, and it’s not since the EU referendum in 2016 that two polls by different agencies in quick succession suggested Yes leads. It’s early days yet, so we’ll need to see over the course of the year whether this is a passing trend or whether it beds in.