Calm down, Mr Sarwar - don't get too enthusiastic over Hamilton result

The Herald Scotland

13-06-2025

While the Labour win was indeed a surprise, it did not indicate a fundamentally different political environment than the one we had previously thought we were in, based on Scottish polling. As other analysts have pointed out since – including Ballot Box Scotland's Allan Faulds, who in these pages provided a sobering corrective to the wild narrative swing that took place last weekend – this was a by-election victory full of caveats.
Read more
I won't recount all of those caveats, but it suffices to say that going into the by-election, national polling suggested that, based on a proportional swing model, the SNP would win 33.7% of the vote, Labour 28%, and Reform UK 19.2%. The actual results were SNP on 29.4% (a 4.3-point gap), Labour on 31.6% (a 3.6-point gap), and Reform UK on 26.1% (a 6.9-point gap). As far as the SNP and Labour go, the polls had their results within the margin of error for a model of this kind. The only party that significantly outperformed their polling was Reform UK.
As Dr Eoghan Kelly, a postdoctoral researcher on the Scottish Election Study team, found by extrapolating the result in Hamilton to the whole of Scotland, if the result were replicated in next year's Holyrood elections then Labour would come third, with 23 seats to Reform UK's 26 seats and the SNP's 49. There's a lot more work for Scottish Labour to do to get back into power in Scotland, and this one by-election win isn't sufficient to suggest otherwise. As we are all reminded before by-elections, and as most of us seem to forget in their aftermath, we cannot draw grand conclusions from them about national politics.
But Hamilton was, nevertheless, a shot across the SNP's bow, and not least because it pointed to significant fragility in the supposed SNP leads in what will be closely contested constituencies next year – putting at risk the projected, disproportionate plurality of seats the SNP are expected to win by dint of dominating the constituency results.
Let's take a few widely cited models to sketch out where we expect these three parties to end up if the polls do not change dramatically by next May (which they might). Ballot Box Scotland currently expects the SNP to win 59 seats, Labour to win 21, and Reform UK to win 15; Professor Sir John Curtice's latest projection for the Sunday Times had the SNP on 54, Labour on 20 and Reform on 18, and Dr Kelly's model (using my poll averages as inputs) suggests the SNP on 62 seats, Labour on 16, and Reform on 17. My model would have the SNP on 60 seats, Labour on 17, and Reform on 17.
The Hamilton, Larkhall and Stonehouse by-election was called after the death of SNP MSP Christina McKelvie (Image: PA)
Read more by Mark McGeoghegan:
Historically, a party winning the SNP's current level of constituency and regional list support would only win around 47 seats, even short of Professor Curtice's relatively low projection of 54. The SNP's projected overperformance is entirely a result of the split unionist vote, meaning that the SNP can win as little as a third of the constituency vote nationally while sweeping the vast bulk of Scotland's 73 constituency seats. However, if they failed to win those excess constituency seats, they would not be compensated for them with regional list seats, which are allocated proportionally.
So, how fragile is this constituency boost? And what does it mean if more close contests, like Hamilton, Larkhall and Stonehouse, slip through the SNP's grasp? My model suggests that a swing of under 5% from the SNP to Labour, compared to current polling, is needed for the SNP to lose a further eight constituency seats. A further two are vulnerable to the Liberal Democrats in the same way, and six to the Conservatives.
The swing from the SNP to Labour in Hamilton was, compared to the modelled result, around four points – enough to flip eleven of those 16 close seats. In other words, if the unmodelled swing away from the SNP in Hamilton is replicated elsewhere next May, the SNP would win eleven fewer seats than expected. Only a couple more would be needed to bring them down to the 47-seat mark.
Why does this matter? Even with 54 seats, the SNP would likely be able to govern as a minority requiring the support of only one other party, likely the Greens or Liberal Democrats, to pass budgets and legislation. At 47 seats, they would need at least two partners unless Labour or Reform backed them, making the 2026-31 Scottish Parliament the first without a realistic two-party majority since 2011.
That isn't necessarily a bad thing. Voters constantly tell us that they want politicians to work together, and establishing a more consensual style of politics has long been a lofty and distant goal of Scotland's political class. And it's an outcome our politicians should be preparing for in the background ahead of next year's elections. Of course, the respective leaders won't admit it in public, least of all John Swinney, as he pushes to secure a big enough minority to rely on the Greens and Liberal Democrats. Still, such an outcome looks much more likely today than it did two weeks ago.
The Hamilton by-election result neither indicated that John Swinney's days as First Minister are numbered nor that Anas Sarwar is on track for Bute House. But it did send an obvious signal: the SNP's position is far more fragile than we thought, and little swings in local areas could dramatically change the complexion of the Scottish Parliament next year.
Mark McGeoghegan is a Glasgow University researcher of nationalism and contentious politics and an Associate Member of the Centre on Constitutional Change. He can be found on BlueSky @markmcgeoghegan.bsky.social