Poll Analysis: Survation 15th – 18th of August 2023

Following on from last week’s YouGov, we got a second Wednesday poll release in a week from Survation (link to tables) on behalf of True North (link to site; no writeup as yet?). That continues the unique trend of a lobbying firm commissioning and publishing voting intention figures in addition to, ahem, carefully selected and worded questions on areas of relevance to their policy aims. Most publicly-facing polling is commissioned by newspapers, nerds, or at the pollster’s own expense… who sometimes also ask carefully selected and worded questions relevant to their aims, admittedly.

It also continues the trend of cruelty to this poor polling aggregator, as Wednesdays are the one day a week where I am both in the office for my day job (so I do not have access to my reporting toolkit at lunchtime) and have a standing social event after work (so I can’t report until much later in the evening.) Obviously folk aren’t doing this deliberately to spite me, but if there was to be another Wednesday release next week, I might start to get conspiracy-minded!

The previous Survation covered the 23rd – 28th of June 2023. Changes are shown as (vs that poll / vs last election).

Regional Vote

I have to admit there’s a really very pleasant, purely aesthetic, neatness to the figures in this poll, which with two exceptions lie firmly within the realms of margin of error. Effectively, the two pairings you might expect to be happiest to work together are both completely equally matched.

The SNP and Labour would be the leading party in their respective blocs, both on 30%, which is a slight uptick for Labour which gives them a tie for their best figure on this vote of the term and indeed since the referendum in any poll BBS tracks. Meanwhile, their junior counterparts in the Greens and Lib Dems are level on 9%, which is a slight downward motion for the Greens but consistent with Survation’s still-recent flip from a high to low Green pollster.

The shift that lies outside the margin of error looks on the surface to be a 3% direct swing from the Conservatives to Reform UK, giving the latter their best figure of the term. That’s basically nonsense though. Reform UK do not meaningfully exist as a serious party; they did not contest a single council ward in last year’s local elections, for example, and even with a sitting MSP at their head in 2021 managed a paltry 0.2% of the vote. They do pop up for UK parliament by-elections but that doesn’t make them any more relevant than the Monster Raving Loonys. What they are, and this applies equally to UK-level polling where they often poll comparatively well, is an easy option for expressing dissatisfaction in a completely inconsequential way in the middle of the term.

In that sense, they are no different to their predecessors in UKIP or their old badging as the Brexit Party. Ahead of 2016, some polls suggested UKIP would win seats, but on the day they got 2% and David Coburn basically fled the Highlands and Islands count in embarrassment. Similarly, at the peak of Brexit-Mania in 2019, the Brexit Party polled as if they’d win a few seats, and we’ve already touched on how that went come 2021. In short, it’s not happening at an election, even if a rogue poll like this briefly suggests it might.

Constituency Vote

A margin of error bonanza on the constituency side of things too, with single point gains for the SNP and Labour counterbalancing a drop of two points for the Lib Dems. This is the outright best figure Labour have had on this vote in any poll I’ve tracked, either this term or since the referendum. Survation don’t prompt for anyone else on this question.

Seat Projection

Projecting that into seats might give us something like this:

Please see this page for how projections work and important caveats.

Although on the regional vote the two prospective government pairings are equal, the SNP’s continued lead on the constituency side of the equation keeps them ahead of Labour. A large part of that is that with very weak Conservatives, the SNP are projected to remain largely dominant in the North East constituencies despite their slide overall. Their junior partners are tied however, and this is nonetheless the joint-best seat projection I’ve ever had for Labour. I know everyone’s favourite game at the moment is “does the First Minister lose his seat?”, and this time the answer is no – he loses his constituency, but the SNP do get a Glasgow List seat which unless they were complete dafties he’d top the list for.

Note too the presence of a Reform UK MSP for the South Scotland region. As a point of interest, I’ve long since effectively amalgamated what I would term the hardline anti-Devolution vote from 2021 (including All for Unity, Abolish the Scottish Parliament, Reform UK and UKIP) into a single pile in my model, which is why it’s South they get that seat, albeit that SNP dominance in North East constituencies knocks them out of second seat there. As discussed above, this is 99.9% likely to be an ephemeral mid-term polling artefact, and not something anyone should expect to see at a real election.

There’s also a lot of discussion as to what happens in a scenario like this, where there is no majority for either obvious governing duo, and where the SNP-Green grouping is larger than Labour-Lib Dem, but with a Pro-Union majority overall. I remain of the view that the Conservatives would have little option but to vote for a Labour First Minister, and would be in far too weak a position to make policy demands for that vote. However, budgets would be torturously difficult, as their support would be much harder to buy on the issue of actual government expenditure and programme, and there wouldn’t be the numbers for Labour and the Lib Dems to try and peel the Greens away from the SNP.

There’s not much movement where it matters on the Westminster side of things, in that the SNP are static and Labour only gain a margin of error 1%. That does however once again give Labour their best showing in a long time, making this poll overall, on every measure, the happiest one for them since the referendum turned everything on its head. The Lib Dems will be rather less so, dropping a whole third of their share versus the last poll, but of the four Westminster parties they are the one whose national share matters least. Like the Holyrood constituency, they don’t prompt for anyone else here, so mystery abounds.

Although the SNP’s 2% lead isn’t the narrowest they’ve had – June’s Panelbase tied them with Labour – it is at a point where when you consider how each party’s vote is distributed, I’d expect Labour to win more seats. The Central Belt would return to its old red tint, albeit still with rather more streaks of yellow than it ever had prior to the SNP’s period of dominance.

Look, absolutely hee haw has happened here since the last poll – this is a masterclass in the effects of rounding, which is what leads to a +1 for Independence here. It’s totally meaningless in statistical terms, but it does put things on the dreaded 52:48. What remains notable with Independence polling overall however is the degree to which support for Independence continues to outstrip support for the Pro-Independence parties at the moment. Plenty of commentators by this point have noted how the SNP’s misfortunes haven’t carried over to the wider Independence camp.

This poses potential challenges for both sides, depending on what vote is “softer” and whether constitutional politics continues to drive voting patterns at actual elections. If the party vote is softer, that might mean there’s space yet for the Pro-Independence bloc to bounce back, bearing in mind that mid-term slumps haven’t been uncommon for the SNP, albeit this is clearly and unarguably the worst period in their recent history, and feels a lot like late-government chaos. If the constitutional vote turns out to be soft, it might just need the jolt of a Pro-Union government to come tumbling down. 


As ever, the last little bit of analysis concerns those hypothetical and more proportional voting systems that BBS likes to play about with. The use of pure FPTP at Westminster is an affront to democracy, and though Holyrood fares far better, AMS is still deeply imperfect. The examples here simply transpose the poll findings onto more proportional voting systems – the reality is that different systems would of course result in different voter behaviour.

Note that on both this model and the below Scandinavian model, the exact tie in votes is matched by a tie in seats for the two prospective governing pairs, again hinging everything on the Conservatives.

Scandinavian Style Westminster

Note that this is the first Survation poll since the (almost certainly) final new boundaries for Westminster were published. Everything here has therefore been re-baselined to 57 rather than 59 seats, including the 2019 and previous poll results for the model.

If you find this or other Ballot Box Scotland output useful and/or interesting, and you can afford to do so, please consider donating to support my work. I love doing this, but it’s a one-man project and takes a lot of time and effort. All donations, no matter how small, are greatly appreciated and extremely helpful.
(About Donations)