SP21 – Final Ballot Box Scotland Projection

Keep tabs on all the latest polling, articles and information ahead of the 2021 Scottish Parliament election in the Ballot Box Scotland Holyrood Hub!

We’re finally here – polling day for the 2021 Scottish Parliament Election! I don’t know about you all, but I’m absolutely knackered. Since the start of the year, I’ve covered a grand total of 34 Holyrood polls, written profiles for the five major parties, analysed the 20 key battleground constituencies, and collated data for over a thousand candidacies. It has been a lot of fun, and people seem to have really appreciated and enjoyed my coverage, which has definitely made it all worth doing.

We’re not done yet however! Those who haven’t already voted by post will be casting their ballots today, and then the pandemic is giving us an abnormal breather, so make sure you get your sleep. The counting will then begin at 9am tomorrow (Friday), and continue again on Saturday, by the end of which we should know who the 129 new MSPs are. I’ll be covering results live on Twitter and here on the website on both days.

The last thing to do before all that is the final Ballot Box Scotland projection for this election. This is a jumbo post, so I’ve tried to provide quick links for analysis of each party and region, so you can jump straight to the area you are most interested in if you don’t want to read the whole thing.

Final Polls

Over the course of the Holyrood campaign, a total of eight different pollsters contributed their reading of the campaign. Seven of those conducted polls with fieldwork concluding at least a week before the election, and we can average them out to take the final temperature of the campaign. This will hopefully balance out some of the extremes at either end of each party’s polling trend.

Regional List Vote

Starting, as is my wont, with the crucial proportional element of the system, and the SNP remain out in front though substantially down versus the start of the year and their 2016 result. There are smaller declines for the two other large parties, with the Conservatives closest to their 2016 performance. If this is close to reality, Labour have a huge gap to close if they want to regain second place.

Gains instead would be for the smaller parties, most notably the Greens who are polling half as well again as they managed at the last election. The prospect of even marginal gains for the Lib Dems are likely to be a relief after a difficult decade. New party on the block, Alba, also manage to find themselves bang on that 3% that I’ve previously described as the point of maximum uncertainty.

Constituency Vote

In contrast to their poor list performance, the SNP’s constituency polling has improved on 2016. Add in slightly weaker Conservative and Labour results and this is the vote that is likely to be make-or-break for delivering an SNP majority. Crucially though, expect some of the non-SNP vote to be gathering behind candidates in key seats. It wouldn’t take many such holds for the three pro-Union parties to shut off the SNP’s path to a majority.

Labour are just a hair’s breadth from the Conservatives on this vote, but unfortunately for them that’s the wrong way round. On these averages, it’s extremely easy to envision Labour repeating their 2016 performance of placing second in the constituency vote but placing third in seats due to lagging in the more important (for them) list vote.

Polling Variance

As ever, there has been quite a lot of variance between pollsters. This was most starkly illustrated the day before the election, with one poll suggesting a comfortable majority of 69 seats for the SNP, whilst another suggested a contextually dreadful 59. Trying to determine at this point which pollster is “most accurate” is a fool’s errand, and even comparisons with previous elections may not be that illuminating, as the most accurate pollster last time may prove worse this time.

As there’s a lot of data here, I won’t be popping it into nice fancy interactive charts, but you can see how each pollster placed each party in the images below – or via the table showing every poll this term here.

Seat Projection

Before we get into the projection, let’s lay out four big caveats that come with this. I am;

  • ASSUMING that polling is roughly accurate
  • NOT Predicting a specific number of seats
  • NOT Predicting which party will win key battlegrounds
  • NOT Predicting all other seats are safe and definitely won’t switch

This is very different to a standard poll projection, where I present specific winners for every seat. Those projections are offered on an “if polling and my model were both completely accurate, then…” basis. Obviously, we all know that perfection is unlikely, so for the actual election, it pays to be more cautious. With that said, let’s get into it…

SNP ~ 58 to 71 seats (model: 65)
Conservative ~ 24 to 32 seats (model: 29)
Labour ~ 18 to 25 seats (model: 19)
Green ~ 5 to 11 seats (model: 11)
Liberal Democrat ~ 4 to 8 seats (model: 5)
Alba ~ 0 or 1 seat (model: 0)

The broad range of most of these projections is a reminder of how volatile polling has been, and how even small changes in vote share can have big impacts on seats. That’s especially true if it’s the SNP gaining in constituencies and causing overhangs that have (negative) knock-on effects on proportionality. You can see that my model also puts some parties quite close to the edges of their possible distribution.

Note though that if you sum up the minimum case for each party, it amounts to 109 seats – that’s around 85% of the total at Holyrood! That means by my estimation only 20 seats (or 15% of the total) are really in contention. Those will make a huge difference to the final shape of parliament, but things aren’t as up in the air as your initial instinct may have been on reading the above numbers.

In general terms we can say the SNP will certainly emerge as the clear leading party, the Conservatives are the most likely largest opposition with only a small prospect of Labour taking that role, the Greens will most probably grow their number of seats, the Lib Dems may prove static, and Alba are likely only in the running for one seat at best.

Remember those caveats though! There could be plenty of surprises in store, especially in the constituencies. As broad as these ranges are, there are things that could happen outside them on a truly wild day. To really dive down into what may happen, let’s take a quick look at each party and each region.

SNP - 58 to 71 Seats

This year’s big question is whether the SNP will regain the outright majority they had in 2011, or continue in minority as they have since 2016. Though the party has been pushing strenuously for both votes, polling for the Constituency vote massively outstrips their Regional vote. Though it may comfort the party to think that’s mostly tactical, as ever I’d caution people against assuming what we think and discuss in the political bubble is what ordinary voters are doing!

I do think they are unlikely to have the whopping 11% gap between ballots we’ve seen here, but how it narrows is harder to guess. I think there may be a narrowing at both ends, with a slightly lower constituency but higher list vote than that average. Regardless, they are unlikely to pick up many list seats, certainly outside the South and Highlands & Islands, though on a relatively bad day in the constituencies they would pick up a few. That’s precisely why (alongside the basic reality of politics that no party just gives up their votes to rivals) they are pushing so hard to pull their list vote upwards.

To emphasise just how tight things are, on this final polling average my model has them bang on 65 seats. However, that includes Edinburgh Southern, which I reckon is likely to prove more resistant than the model thinks and stay in Labour hands. That alone would drop the party below the majority line.

Conservatives - 24 to 32 Seats

Likely to remain in second place, the Conservatives have recovered from some poor polling at the start of the year. Their modelled total of 29 seats is towards the top of their projected range, reflecting the most consistent regional polling trend of any party.

Part of what makes this band so broad is that the Conservatives are quite likely to be heavily impacted by any SNP overhang – for example, they’d have list seats for the chop in regions such as the North East and West based on current polling. On the other hand, they also have some of the best prospects for gaining SNP-held constituencies.

Labour - 18 to 25 Seats

Though Anas Sarwar has consistently polled as the second most popular major political figure in Scotland through this campaign, voting intention polls suggest that hasn’t converted into votes – at least, not the correct votes. Since the start of the year, Labour have substantially closed the gap with the Conservatives on the constituency vote, but have been effectively flat on the regional vote.

That’s why their most likely scenario is to lose seats, with the model reckoning 19 based on polling averages. However, a solid constituency performance could see off further overhang-related seat losses, with Edinburgh Southern in particular likely to be much more secure than simple models can account for.

Greens - 5 to 11 Seats

Across the final seven polls in the average, the Greens were the only party to consistently increase their number of seats in every projection. That’s why their maximum projected figure is the same as what the model was saying at 11 seats. The reason the range dips so far below that is that we know Greens do tend to underperform polling somewhat.

However, they are polling around 2% better than their equivalent 2016 multi-pollster average, so even an error of the same magnitude as last time would almost certainly see the party on a record result. That’s why the seats in West and Mid & Fife, which were narrow wins in 2016, are down as pretty secure this time. At most risk in their current seat total is the second Lothian seat, given that region’s rather extreme potential for overhang. 

Liberal Democrats - 4 to 8 Seats

Though the Lib Dems have some potential for growth, and are the only current Holyrood party besides the Greens to be polling better than their 2016 result, they’ve had a pretty wide range over those polls, including a record low of 4% and 3 projected seats. The model case suggests holding steady on their 5 seats, reminding us that they do need more than a little bit of growth to be sure of seat gains – though they are the masters at consolidating their vote in key areas, mind.

I think holding on to their current crop is more likely than losing any, as I expect them to be reasonably well entrenched in Edinburgh Western. In addition, the North East list seat should be pretty secure, and though it’s one to watch, I’d be astonished if they lost North East Fife.

Alba - 0 or 1 Seat

Launched with bold claims of contributing to a (still undefined) “Independence Supermajority”, Alba’s polling has by and large been lacklustre. By the end of the campaign, even the one pollster that initially had them in clear seat winning territory had them on much shakier ground,. Averaging out at 3%, that’s what I’ve dubbed the point of maximum uncertainty, where it’s probably evens as to whether they get Salmond in. It should go without saying that one MSP does not a supermajority make.

Folk looking to past examples of “small shares, get one MSP across the line” should be cautious. For example, it’s true that the SSP got an MSP in Glasgow in 1999 on 2% nationally, but Tommy Sheridan was very popular in the city then. It’s possible that Salmond is as popular as Sheirdan, but that’ll be as popular as Sheridan is now, which is to say “not very”. Only around 10% of voters say they have a favourable view of Salmond according to polling, and in his own region he’d need more than half of that number to back him to be sure of a seat.

I have to say that I reckon, much like the Greens, Alba are more likely to be over-estimated in polling than under-estimated. If that is the case, it’ll be even harder for them to get across the line anywhere. Still, all that said, as per the caveats above that doesn’t mean I’m telling you they’ve got no chance, I’m just going on what seems within the realms of the possible. There could yet be a shock Alba surge that shakes Holyrood’s foundations.

Central

Projected Seat Ranges

SNP ~ 9 seats
Labour ~ 4 seats
Conservative ~ 2 or 3 seats
Green ~ 0 or 1 seat

Summary

It’s unlikely the SNP will fail to repeat their 2016 clean-sweep of constituencies in this region, and a rare point of security for Labour in polling has been consistently projecting to four seats here. Changes are likely to be very modest, and effectively boil down to whether the Greens will gain one of the three Conservative list seats.

Glasgow

Projected Seat Ranges

SNP ~ 8 or 9 seats
Labour ~ 3 or 4 seats
Conservative ~ 1 or 2 seats
Green ~ 1 or 2 seats

Summary

The SNP should be pretty secure in 8 of the region’s 9 seats – the one exception being Glasgow Kelvin, where they’ll be facing a challenge from the Greens. Though the latter have been polling well in this election, the pandemic may have dampened their campaign, and I’d put the SNP as the likely winners.

Over on the list seats, we can expect three to be securely Labour and one apiece for the Conservatives and Greens. Those parties will then be in the running for the last two seats – Labour need to win twice as many votes as both parties to be sure of holding their four current seats.

Highlands and Islands

Projected Seat Ranges

SNP ~ 6 or 7 seats
Conservative ~ 3 or 4 seats
Liberal Democrat ~ 2 or 3 seats
Labour ~ 1 or 2 seats
Green ~ 1 or 2 seats

Summary

Both of the Northern Isles seats should be safely in Lib Dem hands, whilst the Western Isles and three mainland seats should be securely SNP. That leaves Moray as an SNP vs Conservative contest which looks like its leaning the SNP’s way (note: on reflection, I’d say “leans” rather than “likely” but it’s very late and changing the graphic would be a pain). Meanwhile, Caithness, Sutherland and Ross is one the Lib Dems have in their sights, hoping to replicate their success in the area at Westminster.

Whatever happens constituency wise, the Conservatives are looking at three or four seats overall, and both Labour and the Greens can expect to return a list MSP. Labour’s second seat is at risk and the Greens could take it on a very good day, as could the Lib Dems if they haven’t taken a third constituency. The prospects of a very good Green day may be limited by the presence of former Green MSP Andy Wightman as an Independent candidate, but I think he’s unlikely to pick up a seat himself.

Lothian

Projected Seat Ranges

SNP ~ 6 to 9 seats
Conservative ~ 3 or 4 seats
Labour ~ 2 or 3 seats
Green ~ 1 or 2 seats
Liberal Democrat ~ 0 or 1 seat

Summary

Lothian has the most potential to really scupper a number of parties. Proportionally speaking the SNP were only entitled to 6 seats here in 2016, which matched their number of constituencies. It’s vanishingly unlikely they’d be entitled to more this time. That means if they do pick up Edinburgh Central from the Conservatives, Southern from Labour, and/or Western from the Lib Dems, they’ll distort proportionality and knock otherwise expected list MSPs off.

The two parties most at risk of that are the Greens and Lib Dems. Though the Greens got two MSPs here in 2016, I’d say only one of them is secure, as those possible overhangs could still hurt them even with vote share growth. Similarly, the Lib Dems could be locked out of the region entirely if the SNP manage a clean sweep. That said, I’d expect a degree of tactical voting may help keep Southern in Labour hands and Western with the Lib Dems.

Mid and Fife

Projected Seat Ranges

SNP ~ 7 to 9 seats
Conservative ~ 3 or 4 seats
Labour ~ 2 or 3 seats
Liberal Democrat ~ 1 seat
Green ~ 1 seat

Summary

Back in 2016, Mid and Fife was the only region to have an overhang, which may be more common this time round. That came at Labour’s expense, which is also why this region is the only one where they are projected with the possibility of gaining seats versus 2016. If the Conservatives gained either of the Perthshire seats, that might undo or at least loosen the overhang enough for Labour to pick up a third MSP.

Of those seats, Perthshire South and Kinross-shire would be more likely to flip, as long-serving MSP Roseanna Cunningham has stood down, whereas in Perthshire North, John Swinney is still on the go. I’d say South leans SNP and North is likely. North East Fife is also one that could theoretically flip from the Lib Dems, but I’d be pretty shocked if it did. The Green list seat was shaky in 2016 but there’d have to be a big polling error for them to lose it this time.

North East

Projected Seat Ranges

SNP ~ 8 to 10 seats
Conservative ~ 4 or 5 seats
Labour ~ 1 or 2 seats
Liberal Democrat ~ 1 seat
Green ~ 0 or 1 seat
Alba ~ 0 or 1 seat

Summary

A historic stronghold of the SNP, they suffered some of their worst vote share losses here in 2016, which presaged losing seats in the 2017 elections. That’s why in addition to Conservative-held Aberdeenshire West, Angus North and Mearns plus Aberdeen South and North Kincardine are in play. Banffshire and Buchan Coast is also a possible win for the Conservatives, though I’d rate everything except Aberdeenshire West as likely SNP – that is instead a tossup.

Labour could lose one of their list seats here on a bad day, whilst the Greens are in with a good chance of closing the very small gap here to pick up a seat, and the Lib Dem seat should be safe enough. A spanner in the works may be Alba, with this being the only region I’ve got them with a clear chance of a seat – they could surprise elsewhere, but emphasis on “surprise”.

South

Projected Seat Ranges

SNP ~ 6 to 8 seats
Conservative ~ 5 or 6 seats
Labour ~ 2 or 3 seats
Green ~ 0 or 1 seat
Liberal Democrat ~ 0 or 1 seat

Summary

To a certain degree, South is the anti-Lothian, as it has the least prospect of SNP overhangs distorting proportionality. That’s not to say there isn’t a risk, as if they managed to take 8 seats here they’d almost certainly be over allocation, but that seems a stretch.

They have their best chances at constituency gains in Labour-held East Lothian and Conservative-held Ayr, which I’d rate as tossups. Dumfriesshire and neighbouring Galloway and West Dumfries lean more Conservative by my reckoning, though the former more than the latter. Labour could lose one of their list seats here if they are having a bad day, whereas the Greens and Lib Dems are in the hunt for gains, with the former more likely than the latter.

West

Projected Seat Ranges

SNP ~ 8 to 10 seats
Labour ~ 3 or 4 seats
Conservative ~ 3 or 4 seats
Green ~ 1 seat
Liberal Democrat ~ 0 or 1 seat

Summary

West is another region with some substantial possible overhangs, as if the SNP gain either or both of Dumbarton from Labour or Eastwood from the Conservatives, they could be over their share. I have both seats as tossups.

Whereas the Greens would have been the victims if that happened in 2016, they should be somewhat more comfortable this time given their polling, so I reckon that list seat is more secure. This could be the Lib Dems best chance at a gain were it not for prospective overhangs, and if they have a strong result they could break back through here anyway.

And that is, I promise, the end of Ballot Box Scotland’s pre-election coverage! Remember to check back regularly tomorrow for the latest updates from the count – and most importantly, remember to get out there and vote! Polls are open until 10pm, and make sure to give yourself plenty of time given the pandemic may result in queues.

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)

1 Comment

  1. Just a quick technical question- are the quoted ranges the standard 95% confidence intervals?

Leave a Reply

Your email address will not be published.


*


This site uses Akismet to reduce spam. Learn how your comment data is processed.