As usual, the second of our annual reviews looks back over the polling for both parliaments throughout the year, and it has once again been a year. A UK General Election was always more rather than less likely, as the latest it could have been was 2025. It was also looking likely the SNP were going to lose significant ground to Labour, but the scale of Labour’s victory proved a little larger than most of us expected, pushing the SNP into single-digit MPs for the first time in a decade. Important as it was, the election wasn’t the only source of shocks to Scottish politics this year.
Who could have foreseen Humza Yousaf blowing his political career to smithereens on the basis of some shockingly bad advice and a complete misunderstanding of his governing partners’ voter base? Did anyone have Douglas Ross to all intents and purposes similarly ending his own career, by deselecting an unwell colleague and parachuting himself in as candidate, on their bingo card?
Could we have imagined Anas Sarwar would go from perceiving Prime Minister Keir Starmer as a prerequisite to his own First Ministership to (one can only assume) the primary cause of some classic void-screaming? And would anyone have anticipated the heart-warmingly rare consensus around catching Holyrood up to the Senedd and Stormont by banning dual MP/MSP mandates, after Stephen Flynn cleverly decided to announce his intention to do so (for up to three years!) at the exact moment electoral reform legislation was in progress?
Finally, did anyone really expect to end this year on a low for Labour and with the feeling the SNP had stabilised? In last year’s review I reflected on how my initial uncertainty following Sturgeon’s resignation, were the SNP experiencing a blip or the beginning of the end, had turned more towards the latter. Yet the SNP end this year with a clear advantage over Labour, and possible parliamentary arithmetic that’d make it next to impossible for Anas Sarwar to form a stable government.
Polling Caveats
Although these pieces always work on the basis of a Q4 polling average, in order to have some kind of consistency from year to year, which pollsters are in that average is highly variable. This year we only had three polls in this quarter, two of which came from Norstat and one from Survation. Indeed apart from those two pollsters, the only other full Scottish polling since July’s General Election has come from Opinium. Savanta, YouGov and Ipsos are missing in action.
Especially given the Norstat double dip, that means we’re not really averaging out the different house effects that exist between pollsters. This year that means I reckon the Labour, Green (both being lowballed) and Alba (highballed, Norstat being the only firm that ever finds them above 3%) shares in particular might be a bit iffy. I’d ask you to bear that in mind throughout the piece, though don’t worry, I’ll drum it into your skull at each relevant point too.
There is actually, I am told, a poll due for publication after I publish this. I’m not covering it, I’m not coming back to add it into this, which was mostly done when I heard about it. I’m not going to lie, that’s partly because it feels like a complete and utter piss take to publish polling in the inter-festival lull and I’m quite angry at the circumstances. But genuinely having seen some of the figures I simply do not view it as a credible poll in any case, and I will be adding the firm (not one of the big ones) to my blacklist moving forward, having already felt like they were quite suspect previously anyway.
Regional Polling Average Through 2024
With the General Election having fallen at the start of July, 2024 neatly divides into pre-and-post-GE halves, which show some stark differences. If we start though with one of the relatively stable parties, the SNP haven’t really varied much throughout the year, with their peak only about 3% above their trough. Whilst that represents a significant decrease compared to 2021, it backs up the sense that the SNP have arrested their decline.
Labour however found themselves varying much more widely, with around 8% separating their top and bottom shares in the average. After a few largely unchanging months, the triggering of the General Election saw their support ramp up enough to pull them ahead of the SNP for the first time in a decade. However since the election, their share has taken a severe hit, leaving them much worse off than this time last year, and about on a par with where they were at the end of 2022. As touched upon in the caveats I do think the exact figures are exaggerated by Norstat’s particularly pessimistic view of Labour’s chances, but the direction of travel is clear and matches UK-level polling.
The Conservatives have also had a pretty bumpy time of it. No matter how you slice it, their great post-IndyRef revival is well and truly over. Not a single poll this year saw them enter the 20’s, which feeds a real sense that seat losses are baked-in for them come 2026. After cratering a bit following the election, they have at least seemingly recovered a couple of points compared to their rock-bottom.
Most dramatic of all though is that Reform UK figure, and indeed the fact I’ve come to them before the Greens and Lib Dems is itself a sign of how much they’ve surged. They are now polling comfortably in double-digits, a level never achieved by UKIP or even themselves under their old Brexit Party title. It feels very much like the genie is finally out of the bottle for the further-right in Scotland, having never caught on previously when the SNP provided a (very different) outlet for those seeking an alternative to the major Westminster parties. With the SNP’s appeal weakened after nearly two decades in power and Labour’s shaky start in Downing Street, Reform finally appear to have an opening.
After starting the year with a clear Green over Lib Dem lead, it ends with a slender advantage for the Lib Dems. Some of that is in my view genuine – the Lib Dems have generally been ticking up across regional polling – and some of that might be down to what pollsters were in the mix last and this year. Last year had a lot of traditionally high-Green pollsters; the post-GE era has been exclusively low-Green pollsters. Whether that means the Greens are “really” at higher support now, or were at lower support last year, it’s hard to say, and your instinct may vary depending on preference.
Last but not least, Alba’s uptick can similarly be attributed to the fact Norstat are a dominant factor in averaging. Previously Panelbase, they have consistently had Alba at higher shares than other pollsters. Nobody else thinks they’ll get any higher than 3%, but Norstat has had a few 4-5%s in their mix. Recalling that they most over-egged Alba in 2021, that needs viewed with caution.
Regional Poll Average Q4 2024
How that looks when we look just at the Q4 polls is a relatively substantial SNP lead over Labour, at 7.6%. That compares to just 1.2% last year, and is entirely down to Labour losses rather than SNP gains, as their share is basically unchanged. The Conservatives are also weaker than they were last winter, and it’s Reform UK that appear to be benefitting most from the woes of both parties.
Although right now Reform aren’t that far ahead of the Lib Dems and Greens, there remains a possibility that they will go on to challenge not for fourth or fifth place, but third. However, one thing I’m unsure about – and this applies across all of the polling – is how much Reform’s gains represent real support, versus easily expressed dissatisfaction in the absence of an actual election.
Remember, the next Holyrood election is about 18 months away, and we’ve just had a Westminster one. Polling is low stakes at these times, and Reform UK are an obvious protest vote option for people scunnered with both of the two primary Pro-Union parties. It’s not guaranteed this level of support will hold up all the way to 2026. I’d emphasise though I see this as the difference between Reform living up to polling or instead getting 6-7%. It’s not a question of whether or not they get MSPs, but instead how many they get.
Constituency Polling Average Through 2024
As is so often the case, in all honesty you can just transpose most of what I said about the regional vote onto the constituency and it’s basically the same. This shows a little bit more variance for the SNP than regional polling did, with about 5.6% between their highest and lowest shares. It’s also notable the Greens have been on an upward trend over the latter half of the year with this one: that’s because all of the pollsters in the average actually asked for their constituency support. Previously that hasn’t been the case, which is where lower points in the average come from.
Note that I’ve only included Reform UK figures from the point after the General Election, as they weren’t consistently polled for prior to that. When they were, as you can imagine it wasn’t much to write home about, whereas now it’s pretty remarkable.
Constituency Poll Average Q4 2024
The SNP’s wider variance across the year translates to a more meaningful loss of support on this vote than the list. However, losses are again biggest for Labour, with the Conservatives about as hard-hit as the SNP are. That gives the SNP a 10% lead over Labour, up from 6.7% this time last year. AMS aficionados will know what happens to proportionality when one party widens its lead over its closest opponent…
Seat Projection
Projecting that into seats might give us something like this:
Please see this page for how projections work and important caveats.
… And that is it gains seats, regardless of the fact it has lost vote share overall. This is one of the big problems with AMS, that even though it’s broadly proportional, the fact it has a huge FPTP element can create significant distortion. In a truly proportional system the SNP would remain roughly steady if their proportional share was about the same, yet here Labour’s falling share allows the SNP to notionally regain some of the seats projected as lost last year.
The Conservatives and Greens are also weakened relative to last year, leaving the Lib Dems as the only other one of the existing Holyrood parties to make any projected gains. Instead, the remaining difference flows towards Reform UK, their substantial list vote share naturally translating to a big pile of MSPs.
The sum and total of this is to leave the SNP and Greens one seat short of a majority, though that wouldn’t be a joint government anymore regardless. Accordingly the Pro-Union parties together have a majority, but that by no means automatically hands Anas Sarwar the keys to Bute House – or at least, doesn’t guarantee he can pass a budget. Whilst Reform have outright stated they’d back Sarwar over Swinney for First Minister, would Labour actually want to work with them to pass a budget, and what legislation would Reform UK demand be progressed to keep him in office longer term?
As we approach 2026, if as expected Reform remain a significant component of the next parliament, that question is only going to get more pressing. If it proves impossible either to re-form an SNP minority with the Greens adding up to a Pro-Independence majority, or to form a majority purely from the three “mainstream” Pro-Union parties, or even to imaginatively pass votes on a “traffic light” Labour-Lib Dem-Green basis, what’s the alternative? Do established parties really want to bring the thoroughly anti-establishment Reform UK into the governing fold?
One option that has been on my mind lately following the recent Irish General Election is whether the SNP and Labour might eventually have to bury the hatchet. That may seem utterly impossible given the deep-rooted distaste each has for the other and the yawning constitutional chasm. Yet it once seemed that Fianna Fáil and Fine Gael, both parties of the right but which had been on the opposing side of a literal civil war, were eternal enemies… until they weren’t. 2016 saw Fianna Fáil offer confidence and supply to Fine Gael, before formally entering coalition together in 2020, and this year it was just taken for granted that coalition would continue.
As much as neither might like to admit it, as self-professed social democratic parties, there isn’t actually that much of an ideological gulf between today’s SNP and Labour. The constitution may seem like a big one (and it is), but it’s hardly a civil war either. If Fianna Fáil and Fine Gael can get over their animosity to run an otherwise ideologically coherent coalition when all other options are unworkable or unpalatable, surely the SNP and Labour can come to some kind of working arrangement?
One small caution here might be that my model doesn’t actually account for Reform UK presence in constituencies. If it did, I reckon that’d tip the Banffshire and Buchan Coast seat back into the SNP’s column, which would turn this into the barest possible Pro-Independence majority. Again, that’s the kind of unpredictable chaos you introduce to a proportional system when you insist on maintaining a non-proportional element.
Possible Majorities
Note: these majorities relate simply to passing a vote in the Scottish Parliament. They do not imply the formation of a full coalition government.
- Traffic Light: Labour, Lib Dem and Green
- Independence Bloc: SNP, Green and Alba
- Grand Coalition: SNP and Labour
- Union Bloc: Labour, Conservative, Lib Dem and Reform UK
Turning, as I so enjoy doing, to the hypotheticals, we first look at a reformatted version of AMS. This maintains the same constituency seats as the real system does, but adds a further 16 list seats, divvied up between redrawn regions and with one per region allocated based on the national vote. This gives a more proportional outcome without massively changing the workings of the system. Although these maps are too fidgety to do for every post, you do get them as a Festive treat to see how they shake out.
Similar to AMS-as-is, this has the Pro-Independence parties just short of a majority, on 71 seats to the 74 for the Pro-Union parties. That includes an Alba contingent, benefitting from the 3% threshold and degree of national allocation. Otherwise this would present much the same challenges as real AMS does.
If we get rid of the FPTP element entirely, replacing it with a more fully proportional model, this is where that Fianna Fáil-Fine Gael solution starts to look like less of an option and more like a necessity. Unlike the two FPTP-inclusive systems, this correctly does not hand the SNP more seats despite standing still in vote share, and in fact they lose out relative to last year due to more of the vote overall being for parties above the 3% threshold. Although this gives the widest gap between Pro-Union and Pro-Independence parties at 74 to 55, it does so with an even larger Reform UK contingent.
No wonk-y analysis for this chart, it’s just a visualisation of a range of different changes we could make to the voting system, and how those might pan out on current figures. I’d draw particular attention to the “Pure List” line, as that’s also effectively the “what if AMS was working flawlessly” one that shows, within the system’s own rules, how much of an advantage FPTP has handed the SNP. Here, it’s a full 10 seats more than they’d have if every region was, as it is meant to be, proportional to the votes cast within it.
Westminster Polling Average since GE24
Given that we had a General Election in July, it obviously doesn’t make sense to look at Westminster polling for the whole year. An actual election is an effective hard reset for polling, both because it gives a new baseline for comparison, and because pollsters need to adjust their methodology to account for the outcome and any lessons they’ve learned.
Otherwise this looks pretty much the same as Holyrood’s constituency polling: the SNP basically stable, Labour haemorrhaging support, and Reform surging. It’s also got little upticks for the Conservatives and the Greens, for similar reasons: coming back a bit from rock-bottom for the former, and being “nationally” present for the latter (remember at the GE they only stood in about three-quarters of seats, albeit that was a record for them).
Westminster Poll Average Q4 2024
What that looks like just for Q4 polling should put the fear into both Labour and the Conservatives. The SNP have a solid 7% lead as Labour slump back into second place, losing a whole third of their July share. Although the Conservatives have increased their support a bit since the General Election, Reform UK are truly nipping at their heels here.
There’s already been one poll with Reform in third, and I expect we may see more of those next year. Whilst I still think the Labour figures from Norstat are exaggeratedly low, that also raises another possibility: we see a poll where Labour are in third, because Reform are in second. If that happens, the takes are going to be absolutely nuclear.
Seat Projection
I generally avoid doing detailed projections when we’re dealing purely with FPTP. Annual reviews are my one break with that policy, when I refer to Electoral Calculus for an indication. FPTP projections are extremely difficult to get right, as everyone established ahead of this year’s election, Electoral Calculus included. Yet, it’s what we have to go on, and it’ll do for these purposes.
All of the changes here are between Labour and the SNP, and despite the latter’s substantial lead in votes, Labour maintain a narrow estimated lead in seats. That’s down to the fact Labour’s vote is so heavily concentrated in the Central Belt which means that although the SNP are picking up some of the most vulnerable seats, they aren’t able to sweep Labour out of them the way they were themselves.
Scandinavian Style Westminster
Just like Holyrood though, I do always like to look at Westminster if it wasn’t mortifyingly, miserably, indefensibly undemocratic. For the 2015, 2017 and 2019 elections, the democratic distortion played to the SNP’s advantage, before flipping back to being Labour favourable in this year’s election – regardless of who it benefits, it simply isn’t reflective of the electorate, a point hammered home by the FPTP projection earlier.
This correctly puts the SNP ahead of Labour in seats, but it also ensures everyone else gets a fairer distribution of seats. That means the Lib Dems going down one whilst the Conservatives add to their tally, and both Reform and the Greens are given representation rather than their voters being ignored.
(Note I don’t have a fun map for this one as I’ve discovered I’ve got a mismatch in seat allocation to constituencies between what exists in my main calculator and my post-GE24 PR model. This doesn’t affect the total number of seats per party, as that’s nationally proportional, but it does impact the constituency level distribution and it’d be too much of a faff to sort right now!)
Looking Ahead to 2025
In theory, we should be looking at a nice quiet non-election year in 2025. There is a possibility that a failure to pass a budget, now that the Scottish Government doesn’t have a majority, could trigger an early Holyrood election. I don’t however think that’s particularly likely at the moment, both because the Lib Dems especially seemed very open to a deal when the draft budget was passed, and because the political calculus behind an early election has changed dramatically.
A few months ago, both the SNP and their opponents could have seen some advantage in an early election. Labour and the Lib Dems may have seen it as a way to speed Anas Sarwar into Bute House, and hoped the scale of SNP defeat would be enough that everyone would be happy to pass emergency legislation to cancel the still-required 2026 election. Now it looks like the SNP will do a lot better and Reform would be necessary partner for any non-SNP government, a snap election may be less appealing.
Likewise the SNP might be cooler on collapsing their own government, because they might have previously hoped to either benefit from painting their opponents as reckless to pull of a shock win, or at least if they lost hoped a short sharp shock of non-SNP government might galvanise their support. If they and possibly the Greens were able to withhold the votes necessary for cancelling 2026 (it’d need to be two-thirds of the Parliament), they could have tried for a quick turnaround return. For the moment it looks like the worst is over and they don’t need to take such a gamble.
What we will see next year however is the final version of new Holyrood boundaries. That will make my job a little harder for a short while as I’ll need to estimate 2021 shares on the new boundaries in order to do projections from polls, so you may need to bear with me at that point. There may also be some truly terrible changes to the regions, but I live in hope that Boundaries Scotland find some way to avoid that.
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)
