The first UK General Election I ever voted in was 2010, and in Scotland it was an extremely boring affair. Every single seat was won by the exact same party that had won it in 2005, resulting in a very rare no change election – even whilst changes elsewhere in the UK helped see in the first Coalition Government since WWII. It really feels like every General Election since then has been making up for that boredom, and yesterday’s dramatic results were no different.
It was a complete shock for most people involved in analysis of the campaign to see the SNP projected to win 55 seats when the exit poll was released. Although they didn’t quite reach that number in the end, they did still do better than pretty much anyone was expecting. Having suffered a substantial reversal in fortunes in 2017 in the face of resurgent Conservatives, clever targeting by the Lib Dems, and a small Labour fightback, the SNP inflicted stunning defeats on all of those parties yesterday – most prominently by unseating Jo Swinson, the Lib Dem leader.
Scotland’s 59 seats split as follows:
- SNP – 48 (+13)
- Conservative – 6 (-7)
- Liberal Democrat – 4 (nc)
- Labour – 1 (-6)
In terms of votes, it was:
- SNP – 1,242,380 (45.0%, +8.1)
- Conservative – 692,939 (25.1%, -3.5)
- Labour – 511,838 (18.6%, -8.5)
- Liberal Democrat – 263,417 (9.5%, +2.8)
- Green – 28,122 (1.0%, +0.8)
- Brexit – 13243 (0.5%, +0.5)
- UKIP – 3303 (0.1%, -0.1)
- Others – 3819 (0.1%, -0.1)
I’ll get into the specifics of vote shares and which seats changed hands shortly, but on headline figures this is another remarkable election for the SNP. They won 81% of Scotland’s seats on 45% of the vote, which although not topping their own 2015 tsunami is still the second highest share of seats in Scotland at any election in the democratic era, ahead of Labour’s 78% from both their 1997 landslide and in 2001. The Conservatives may not much enjoy having lost more than half their seats, but they at least remain well clear of wipeout. Meanwhile, Labour’s relapse to the single seat they held in 2015 is a further electoral blow for a party that hasn’t been short of those in recent years, and what should be the relief of a static result for the Lib Dems is marred by losing their leader.
Note that for the purposes of analysing results, I am counting Dunfermline and West Fife MP Neale Hanvey in the SNP column. He remains suspended from the party, which withdrew support from him following an antisemitism scandal. However, he remained on the ballot identified as an SNP candidate. On my UK Parliament page he is identified as an independent. It’s an awkward one to accurately convey, to be sure.
How accurate were the polls?
In my prediction post, I also included the average from the last poll taken by three different pollsters – Panelbase from the 3rd-6th, YouGov from 4th-10th and Survation from 10th-11th. So how did that compare to the actual result? Here’s the difference between each party’s result and the polls:
- SNP +4.0%
- Conservative -2.9%
- Labour -1.7%
- Liberal Democrat +0.5%
- Green +0.2%
- Brexit -0.3%
Basically we can say that for the Lib Dems, Greens and Brexit the average was pretty much spot on. Survation were notably far off with only 7% versus the 10% the other two found for the Lib Dems, however. However, every pollster underestimated the SNP and overestimated the Conservatives and Labour. It’s worth at this point briefly saying there were folk in Ballot Box Scotland’s mentions before the election pointing out the opposite had happened in 2017 – a timely reminder then never to assume what pollsters got least right last time will be repeated the next time.
Panelbase fared the worst out of all pollsters, getting everyone out of the big 4 the most wrong except for the Lib Dems, but since their fieldwork was a week before the vote, they may have fared better with a later one to pick up last minute swings.
Seat Swings and Roundabouts
Now let’s get down to the nitty gritty of which seats changed hands. Again if you refer back to my prediction post, I had a list of 17 seats I thought were the most likely to change hands. A few people were very grumpy with the level of certainty in my prediction, though some wished it had been more certain and others less. However, given that 14 of those 17 seats then did change hands, I feel somewhat vindicated! Only one seat I didn’t think likely flipped, and I’ll take that level of predictive success.
Going from the Conservatives to SNP, there were 7 changes:
- Stirling, from Con lead of 0.3% to SNP lead of 17.6% (swing 9%)
- Gordon, from Con lead of 4.9% to SNP lead of 1.4% (swing 3.2%)
- Ayr, Carrick and Cumnock, from Con lead of 6.0% to SNP lead of 5% (swing 5.5%)
- Ochil and South Perthshire, from Con lead of 6.2% to SNP lead of 7.8% (swing 7%)
- Angus, from Con lead of 6.6% to SNP lead of 8.7% (swing 7.7%)
- East Renfrewshire, from Con lead of 8.8% to SNP lead of 9.8% (swing 9.3%)
- Aberdeen South, from Con lead of 10.7% to SNP lead of 8.8% (swing 9.8%)
I hadn’t expected absolutely all of these seats to go from Con to SNP despite identifying them all as potential, but that’s exactly what did happen, and is a large part of why my lower seat estimate for Conservatives turned out to be 2 too high and SNP 2 too low. Angus is a particularly notable result here as that was a longstanding SNP seat before they lost it in 2017, so regaining it will have been a great comfort. Conservative losses in these seats effectively wipes them back out of the Central Belt, confining them to the South and North East.
Huge swings in urban or heavily Remain areas like Aberdeen South, Stirling and East Renfrewshire were never likely to be matched the the more Leave and rural areas like Moray or Banff and Buchan, and my belief those weren’t likely to change hands proved spot on. The losses are substantial, but this is still their second best result both on seats and votes in decades.
From Labour to the SNP, there were 6 changes:
- Rutherglen and Hamilton West, from Lab lead of 0.5% to SNP lead of 9.7% (swing 5.1%)
- Kirkcaldy and Cowdenbeath, from Lab lead of 0.6% to SNP lead of 2.6% (swing 1.6%)
- Glasgow North East, from Lab lead of 0.8% to SNP lead of 7.7% (swing 4.3%)
- Midlothian, from Lab lead of 2.0% to SNP lead of 11.8% (swing 6.9%)
- Coatbridge, Chryston and Bellshill, from Lab lead of 3.5% to SNP lead of 11.6% (swing 7.6%)
- East Lothian, from Lab lead of 5.5% to SNP lead of 6.7% (swing 6.1%)
I’m far less surprised to have seen Labour almost completely wiped out – and will all due respect to YouGov’s infinitely greater expertise on polling and projection, I think this result emphasises that their MRP model was a bit off for Scotland. Given Labour lived up to their “worst ever GE result” polling here, it was incredible to me that anyone could think they’d hold most of their seats, whatever a model might say!
Note that despite this catastrophic result, none of the SNP’s swings from Labour are near the clutch of 9%+ swings from the Conservatives. Labour did do better in their seats than they did overall, even if that’s a very faint silver lining. The closest result came in Kirkcaldy and Cowdenbeath, where the SNP actually lost votes by virtue of that suspended candidate. It’s not a brilliant comparison as a result of that confused situation, but that’s how things can sometimes end up with First Past the Post.
Between the SNP and Lib Dems there were two changes, one each way:
- North East Fife, from SNP lead of 0.005% to Lib Dem lead of 2.9% (swing 1.5%)
- East Dunbartonshire, from Lib Dem lead of 10.3% to SNP lead of 0.3% (swing 5.3%)
It was completely unsurprising to me that the Lib Dems took North East Fife – just as it’d have been unsurprising if the SNP held it. The most marginal seat in the UK, there was really no way in this campaign to claim certainty for either side here, with both parties polling better than in 2017. That puts Stephen Gethins in the unenviable position of being the only member of the SNP’s 2017 Westminster group not to be returning to that group this time.
On the other hand, the loss of East Dunbartonshire completely astonished me. Not only am I eating crow personally having spent two months declaring it was hugely unlikely she’d lose her seat, but it has been an incredible shock for a major UK party leader to lose their seat. That certainly hasn’t happened in my lifetime – the last time I would assume it might have was in the collapse of the Liberals in the early to mid 20th century. To lose votes in your leaders seat whilst gaining them almost everywhere else in Scotland just boggles the mind.
It perhaps raises a quite serious question – is it now possible for a Scottish MP to head up one of the three major GB-wide parties? Few Scottish seats are truly safe from prospective SNP gains, and it’s easier to present yourself as an accessible and likeable local candidate when you don’t have wider responsibilities across the UK keeping you away from your constituency.
The Smaller Parties
The final part of my prediction yesterday related to how well the Greens, Brexit and UKIP might do. I thought the Greens had the potential to retain a handful of their deposits (that is, win more than 5% of the vote) but even in their previous strongest seat, Glasgow North, they came up short. What is a well-evidenced reluctance to vote for smaller parties under First Past the Post was bolstered in this election by what many voters perceived to be a serious need to vote tactically, and that put a huge squeeze on all of them.
For these three parties it’s important to remember the national vote share is skewed by their number of candidates. If we look just at the seats they contested;
- Green, 22 seats, 2.6%
- Brexit, 15 seats, 2.0%
- UKIP, 7 seats, 0.9%
- Taking Brexit and UKIP together as a core hard line Pro-Brexit vote, 22 seats, 1.6%
These are still not exactly astronomical vote shares, but they do give slightly better context for the smaller parties performances.
And that’s the first part of Ballot Box Scotland’s General Election 2019 analysis out the way! Over the next few days I’ll be adding the full results to a dedicated 2019 Westminster page, having a look at some of the other fallout vis a vis prospective council by-elections, and of course re-calculating the whole blasted election UK wide using a Proportional Representation model.
As one final note though I want to express my huge appreciation to the three folk on my count night team – KB and DM who were on charting duty all night, and NP who kept the live map graphics on the website running smoothly. As I discovered in May’s EU elections, covering count night is a massive undertaking, and I couldn’t have done it without them. If you find/found all this coverage useful or interesting and can afford to do so, please pop in a donation to the project, which will be shared out between me and the team.