Poll Analysis: Panelbase 28th – 30th of March 2023

For reasons outlined in this Twitter thread, Ballot Box Scotland has generally had a non-coverage policy for Panelbase polls lately, given the tendency to be commissioned by the Sunday Times who do not then publish the Holyrood vote figures alongside Westminster and Independence. I’ve since flexed this policy such that basically anyone playing silly buggers and not doing full release of core VI figures at one point doesn’t get direct BBS reporting and analysis. I was also aiming for a little bit of a break from BBS after the SNP leadership election was over, since that interrupted a previous attempt at one, but did say I’d cover polls!

Although this one (link to tables) was indeed a Sunday Times (link to original writeup) commission, they’ve actually published the Holyrood figures this time! It therefore meets the criteria for full analysis, hurray! I’m going to take that as a win for Ballot Box Scotland complaining about it, and certainly not suggest it instead may have been the first time the figures were acceptable that paper’s known constitutional preferences.

Even if it had fallen foul of the non-coverage policy, in all honesty this is such a juicy poll it just wouldn’t have done not to cover it. As the first poll (in terms of fieldwork dates) since Humza Yousaf took office as First Minister, there’d have been a much greater interest in this one in any case. Add that to what it says and, well, it’s really worth getting into the meat of it. 

Note that this was published the day after a Savanta/Scotsman poll that had run until the 31st. There are a lot of similarities between those polls and, on the assumption not everyone seeing this analysis piece will have seen the other, a lot of the commentary is going to be a direct copy-and-paste between the two pieces. If you are such an avid BBS follower that you do read both, I’m not repeating myself as such, just covering my bases!

The previous Panelbase covered the 7th – 10th of March. Changes are shown as (vs that poll / vs last election). 

Regional Vote

Starting with that all-important regional vote, this isn’t quite the smallest gap between the SNP and Labour of the term (or indeed since before the 2016 election), but at 4% it’s not far off it. Accounting for fieldwork dates, this would also be the joint-highest Labour figure of those periods too. Although there’s still a lot separating the SNP and Labour from the Conservatives, it’s one of the most three-cornered polls too, as the Conservatives benefit from quite a chunky positive swing that lifts them – just – back into the 20’s.

The Greens and Lib Dems meanwhile are both down two points. For the Greens that’s particularly interesting because 10% has for most of this term been their record with Panelbase, who are one of their lowball pollsters. Instead, they’re now firmly in their rough average at the moment. To be at 12% in their previous entry was therefore remarkably high for this pollster. Although it’s a very slight improvement on 2021, 6% is the lowest the Lib Dems have polled with anyone for a few months.

If we turn to Alba, I’m afraid I’m going to write the first of my little segments that will upset a section of the Twitter-sphere. That 5% is remarkably high, their best of the term, and there will be a temptation in some quarters to view it as a specific backlash arising from the leadership election. I’m afraid for fans of that narrative it doesn’t hold much water. Panelbase didn’t prompt for Alba in their previous poll, and indeed haven’t done so since Alba themselves commissioned them in October, so we can’t meaningfully compare with the immediate pre-contest period.

When they did prompt for Alba in October, they got 4%, so not very far off the figure here. However, most other pollsters who prompt for Alba find them on 1-2% most of the time, including the four other polls since Sturgeon announced her resignation. This has very obvious echoes of the fact that Panelbase were the most consistent firm in massively overestimating Alba before the 2021 election, with three polls finding 6% and their final 4%, against Alba’s actual result of 1.7%. I am inclined to believe Panelbase simply have not been able to fix whatever methodological quirk is leading to this, but that since most of their polls don’t prompt for Alba these days, it’s not generally been a visible issue.

Constituency Vote

The constituency vote displays some pretty similar things, with the SNP down a bit and the Conservatives being the ones making up ground rather than Labour (be aware the Sunday Times initially reported Labour’s share here as 33%, but the tables say 30%). Although it’s mostly within margins of error, the fact the SNP slip slightly further versus the last poll than Labour narrows the gap between the two to the lowest it’s been for this vote, which is the kind of thing that will have consequences we’ll see later. Note too this is the lowest SNP share on this vote of the term so far.

The Lib Dems look slightly healthier on this vote than the list, even though this is the one the matters far less to their seat tally. You’ll also see the Greens are down from what was a record high for them on the constituency vote, but still their second best ever. I’ve written previously that the data we do have from constituencies in 2021 suggests Green constituency shares in polls are always underestimates versus what they’d see if they contested them all, and though that’d apply here, it’s to a much lesser degree. The key to this point is also “if they contested them all” – on the assumption they don’t, they’d obviously be well below this figure.

Seat Projection

Projecting that into seats might give us something like this:

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

Look at all that red! Some recent polls had little Labour forays into projected constituency gains – a wee nibble on Clydebank and Milngavie here, a lick of Hamilton, Larkhall and Stonehouse there. This goes significantly further, projecting to double-digit constituencies for Labour for the first time in a very long time indeed. It says a lot about how dominant the SNP have been lately that it actually felt very peculiar filling in list seat bubbles in Central, Glasgow and West with yellow! Another interesting little titbit is this is the first poll in ages, probably in the history of BBS, to have the SNP project to third in a constituency, placing behind the Greens in Kelvin.

This also becomes only the second ordinary poll of the term, after the recent Redfield & Wilton, to suggest the co-operative government between the SNP and Greens would lose their combined majority. Even with those Alba seats, the first to be projected this term (caveats about this likely being an outlier for them in mind), this still gives an advantage of 66 to 63 for the combined Pro-Union parties over their Pro-Independence counterparts. Bear in mind however that at this point in the previous term, there had been six such polls (in Sept, Oct, Nov and Dec 2017, plus two in Jan 2018) suggesting such an outcome.

This is where I do some further upsetting. Firstly, it’s absolutely the case that Humza Yousaf’s Glasgow Pollok seat projects to a Labour victory here. Whilst that would doubtless be extremely painful for him, he does not “lose his seat”, because the SNP win a Glasgow list seat. Given Sturgeon topped their Glasgow list when she was in charge, you’d expect Yousaf would too.

He would therefore join, assuming everyone contested the same seats as in 2021, Anas Sarwar, Douglas Ross, Patrick Harvie and Lorna Slater in being leader from a list seat. Indeed, of the Holyrood parties, only the Lib Dems’ Alex Cole-Hamilton would hold a constituency, a comedic artefact reflecting the (currently) smallest party’s dominance of their small clutch of seats. Crucially, there is nothing better or worse about which kind of seat you represent – I merely point this out because I know people obsess over constituencies.

Next, I’d like to upset people who are absolutely convinced that if Kate Forbes had won polls would have looked a lot better. I simply don’t think that’s likely at this early stage. What we saw during the leadership contest was a steady slide for the SNP in the polls, as what was a bitter and at times frankly farcical campaign severely damaged the party. Let me emphasise: damaged the party, not simply the candidates.

A Yousaf win has led to the messy stuff around not including Forbes or many of her backers in Cabinet, which likely adds to the feeling of chaos. A Forbes win would undoubtedly have seen a large majority of the Sturgeon cabinet handing over their resignations in the most scathing terms, given her own starkly critical assessment of her government’s record, giving similar feelings of chaos. The SNP are polling (relatively) poorly because the SNP brand has been tarnished, and that’d have been the case regardless of winner. At best any differences would be firmly within margin of error.

In the longer term, of course, a different choice of leader would almost certainly result in different outcomes from what we will see. But we don’t know what kind of different. To go back to the points I made in my piece about what’s next for the SNP after the contest, we really are in the eye of a political storm right now. This is where my final bit of upsetting comes in, because for the folk working on the basis that “line go up” for Labour and “line go down” for the SNP, we need to recognise there has been a sharp and sudden shift in polling over the past three months.

With three years to go until the next Holyrood election, there’s twelve more sets of three month periods to fit similar dramatic swings into. This is undoubtedly a bad poll for the SNP and for Humza Yousaf and it could get… worse or better! We don’t know yet! I know I’m one of very few political commentators not nailing my colours to the mast on whether I think this is a mere blip or a fatal inflection point, but I’m doing that for very good reason. The SNP have spent their time in government slumping in the middle of their term and surging at the election. In recent years, Conservative governments have had cripplingly bad polls then recovered and gone on to victory. In short, it’s not over until the Scottish Electorate sings, and it hasn’t actually done so yet. 

Just like the other votes it’s the Conservatives making the biggest gains here, as Labour slide a little bit, whilst both the SNP and Lib Dems drop a point to both only be 1% above their lowest polls this term. It’s important to emphasise most of what we see here is margin of error movement, but it means that the SNP vs Labour gap here is a touch wider than in the Holyrood constituency figures.

That’s against recent trends, where the gap for Westminster has been narrower (though often not by much at all). Though this is still within that “danger zone” I keep referring to for the SNP, it’s not that far into it, and they’d thus probably maintain a significant lead over Labour in seat terms, and likely even still hold a majority of Scottish seats.

(As an aside, I’m not sure why Panelbase are prompting for the Greens on Holyrood constituency but not at Westminster! They’ll presumably make up most of the Other chunk here, but we can’t say how much of it.)

Over on the constitutional question there’s as good as no change here – a little bit of decimal point based shuffling gives a completely statistically meaningless +1% for Independence with Don’t Knows included. Excluding them is still on the 52-48 that we now know not only as the Brexit numbers but also the SNP Leadership election numbers. We’re increasingly haunted by that split!

The fact that Independence is holding up relatively steady, albeit on the minority side of the divide, whilst the SNP and Greens are on a bit of a slide could suggest one of two things. On the one hand, it could be the start of a trend whereby support for Independence runs ahead of the parties that support it. On the other, it could be that after the political drama cools slightly, this could be the level those parties rebound to. Like so much else in this piece, this is a big question mark we’ll need a few more months to really get a better sense of.


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.

Naturally, since the Pro-Union parties have a majority under the AMS-as-is projection, they’d have a more comfortable one if we made AMS a bit more proportional.

Likewise, a fully proportional model for all parties with at least 3% gives a gap of 69 to 60 for the Pro-Union bloc.

Scandinavian Style Westminster

Finally, we should all know the story about imagining PR at Westminster – it, correctly and fairly, cuts down the SNP’s representation and instead boosts the tallies for Labour and the Conservatives. Given the size of the Other component of the poll and their performance in the Holyrood constituency I guarantee you the Greens would be on at least 3% here and thus projecting to a couple of seats, but since we don’t have an actual figure I’ve just left them out.

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)