SP Poll – Survation, 8th-12th of September 2017

For the last few months ahead of the 2016 Scottish Parliament election, I kept a little polling and projection blog. Since then polling hasn’t been as regular or as pressing to analyse, for obvious reasons. But I’ve been champing at the bit for a poll ever since the shock Westminster results here in June. After a false start, constituency only poll at the weekend, Survation and the Daily Mail have obliged with an actually useful poll. Here’s how my calculator reckons it might look (for fun, as I’m sure everyone is actually aware accurately projecting national results to a local level is next to impossible);

Straight off the bat it’s worth noting there’s a huge SNP decline there. Although they hold up well with 42% (-4.5%) at the constituency level and therefore would potentially hold all but 7 of their constituencies, that regional vote is quite the knock, down a whole 10.7%. As a nerd aside, that demonstrates a problem with the split voting that AMS offers. The SNP are doing so much better in constituencies that it warps the list result – doing Labour out of 4 seats, and the Tories and Lib Dems out of 3 each.

For Labour and the Tories there’s a reverse 2016 going on here – a Tory lead in the constituency vote but lag in the list. That would return Labour to second party status without winning any additional constituencies, whilst the Tories could lose a handful of seats but perhaps double their constituency haul. Ah, the joy of mixed electoral systems.

The big surprise is the Lib Dems coming out at 10% on the list. Past voting patterns make me rather dubious of that finding, especially given a constituency vote of 7%. The Lib Dems have typically polled much better at constituency than regional level, with their list vote being two-thirds of their constituency vote for the past two elections. A relatively static showing in May’s council elections and a national decline in votes in June’s Westminster and zero Lib Dem visibility or impact since add to my sense that is just a bit off. It’s not implausible however, given the gap between Tory constituency and list, that there’s some minor tactical voting going on to keep the SNP out of constituencies.

Finally, it’s a positive poll for us in the Greens, offering the potential of entering double figure MSPs for the first time and an MSP in every region. It’s worth noting that we do typically poll higher than we manage on the day so always treat with caution. By the same token though it does show the huffy “why won’t you do everything the SNP want/why are you doing anything the SNP want, no one is ever voting Green again” brigades are just vocal cyber-zoomers.

Oh, smaller parties, right, I used to talk about them too. Fair’s fair, here we go. No reporting whatsoever of any RISE/SSP/Other Trot Grouping, meaning they still aren’t a thing worth anyone’s time. 3% for UKIP is surprising given they have died on their arse since the Brexit vote and only got 2% in 2016 anyway – definitely just a margin of error thing.

Methodology Note

As with the predictions I did ahead of 2016, I’m using a completely unacademic “regionally modified swing” model rather than the traditional “universal swing”. In theory I’m taking the 2011-2016 swing in each region and working out what it was relative to national swing, then assuming similar regional distributions in swing between 2016-poll. However, a lot of Scotland was very, very swingy last time – the North East, for example, had the SNP decrease 3.5x as much as they did nationally. I’m averaging that raw figure with a universal swing as a moderating force as a result. It’s never going to be perfect, but it hopefully captures regional variations slightly better.

Additionally, the Daily Mail journalist who originally tweeted this poll has further different seat projections – notably, with the Lib Dems on 13. That’s down to the use of figures to a decimal place. As the convention in Scotland is to publish polling figures rounded to the nearest whole number, I follow that convention for predicting seats. That decimal place can make all the difference though.

For example, a universal swing on whole numbers would give the Lib Dems 6.63% of the Central vote and no seat. The Greens would get 7.12% and the second last seat, and the Tories 14.17% thus a figure of 7.085% to secure the last seat as their second in the region. It’s easy to see how even a fraction of a percentage difference by using figures to a decimal place could give the Lib Dems a seat. Lothian is similarly close between Labour and the Lib Dems.

For Fun – How Bout Dat Scandinavian Result?

Since I just finished rebuilding my alternative electoral system, it’d be a waste not to apply it to a poll. Using the list vote from the poll, it’d look a bit like this;

A wee curiosity worth noting here is that although the Lib Dems take a lead on the Greens, we’d win more of the “direct” seats. The Lib Dems would have to go to the bonus seats allocated nationally to pick up 5 of their 13. The national distribution of the Green vote is still healthier than the Lib Dems, who have deep wells of support in some places and barely register in others, even with a 10% national share.

Leave a Reply

Your email address will not be published. Required fields are marked *