Savanta polls: figures don’t add up


By Russell Bruce

Savanta have done two polls for the Scotsman. Both polls calculate that the Tories will hold their 6 main seats in Scotland which does not tie up with either Scottish election history or the UK wide trends indicating that Labour is set to secure a massive majority south of the Border on Thursday. We all know that elections can spring some surprises but against the wider run of polling the Tories holding on to all six Scottish seats they won in 2019 is beyond a believable.

The consensus emerging from the breadth of polling suggests the SNP will win Aberdeenshire West & Kincardine, and Dumfries and Galloway both of which the YouGov MRP poll of 15th January indicated would be won by the SNP albiet by narrow margins of 3% and 1% respectively which we wrote about in January. Electoral Calculus is among those suggesting the SNP will now win Aberdeenshire North and Moray East where the candidate is the controversial Douglas Ross. Electoral Calculus gives the SNP an 11% lead over Labour with Ross reduced to 3rd place. In Gordon and Buchan Electoral Calculus give the SNP a 10% lead over the Tories. As Electoral Calculus is not alone in predicting 4 of the 6 seats the Tories won in 2019 are set to be SNP wins this seems a likely probability.

That leaves two South of Scotland seats – Dumfriesshire, Clydesdale and Tweeddale held by David Mundell since 2005 when the seat was created. Electoral calculus thinks Labour will take this seat with 32.1% of the vote over Mundell on 28.9% with the SNP third, down to just 23.9%. That would be a major drop in the SNP vote when for the last 3 elections they have been in second place with Labour a poor third. Emma Harper for the SNP came within 0.5% of Mundell in 2015. The Labour vote collapsed to just 8.5% last time and the SNP candidate increased their share of the vote by 8% in what was a direct move from Labour to the SNP in 2019.

That takes us to the last Tory seat, Berwickshire, Roxburgh and Selkirk which Electoral Calculus believe is a Conservative hold for John Lamont who has held the seat for the last 7 years. They predict Lamont’s victory will be a massive 1%. SNP candidate David Wilson is predicted to gain 26.7% with Labour coming third on 26.3%. Lamont appears to be panicking and had a team out at the weekend delivering yet another leaflet this time stressing the result was on a knife edge.


Labour are going to win back seats in Scotland. They are also going to win big in England and Wales and do not need that same level of support here. Today’s Labour party is not what people hope it will be other than it is not the Tory party of May, Johnson, Truss or Sunak. There is a degree of consistancy in some polls although some outliers like Savanta’s for the Scotsman and their UK wide poll for the Sunday Telegraph cut Labour’s lead to 15%. A Delta poll carried out in the same period maintained Labour’s well established 21% lead and predicted a rout of the Tory party in England as Labour surge into many formerly long held very Tory safe seats.

I am not a fan of MRP constituency polling predictions. (MRP multilevel regression and post-stratification) They might get a fair number correct of the basis of the overall national picture but do come up with a number of howlers because of a lack of detailed local knowledge in their modelling. I am also sceptical at the predicted size of the Labour vote in parts of Scotland. It is not surprising that Scotland is again desperate to get rid of a Tory government it did not vote for but the increase in suggested size of the Labour vote in Tory seats where the SNP is the challenger is strange. How anyone could miss the Tories blasting away for months on end that only the Tories could beat the SNP and therefore only the SNP could beat the Tories in these constituencies is weird.

John Curtice has pointed out that Labour’s vote concentration gives them an advantage in urban central belt seats. True but also that the SNP has greater concentration in more rural areas with a long history of out polling Labour. Hence my doubt that the Labour vote is as strong as the polls are suggesting in more rural areas like Berwickshire, Roxburgh and Selkirk. The 26.3% for Labour quoted above has no precedent in the history of this constituency. I would not be surprised if it increased to around 12% this time but a 560% increase from 4.7% in 2019 would be extraordinary.

The numbers in Scotland

The expectation is many results will be very close. Will we get a few penny majorities involving multiple recounts? Despite the number of polls, getting a real handle on who wins what in Scotland is tricky but I am going to provide my judgement, or best guess, on the range for each party. Labour 20-26, SNP 19- 25, Lib Dems 4-5, Conservatives 0-2. Reform, Greens and Alba will not win any seats in Scotland.

UPDATE Wednesday noon – Savanta’s new poll puts SNP ahead

Conducted betwenn 28th June and 2nd July this poll puts the SNP on 34% (unchanged) and Labour on 31% (down 3%) for the general election tomorrow. Commenting on the change John Curtice said ” this would see the SNP return 24 MPs, Labour 22, the Conservatives 6 and the Liberal Democrats 5.

Perhaps the most important reference in the report on Savanta’s polls for the Scotsman is that since mid June Labour’s vote has dropped by 7% and just topped up from 4% to 7% in this latest poll. It is good news for the SNP but all these figures are within the margin of error. Undoubtedly the progressive peel back in support for Labour is the big finding here. I have a lot of respect for John Curtice but still question his calculation that the Tories will hold on to 6 seats. The naked truths on a lot of surprise results across the UK will be revealed in little over 40 hours by 6am Friday.

Savanta’s latest calculations for all other parties general election votes in Scotland are Conservative 15%, Lib Dem 9% Reform 6%, Green 3% and Others 2%.

UPDATE 23.00 3rd July

Nigel Marriot is a Chartered Statistician whose prediction for the 2019 General Election was the most accurate beating all the polling companies and even John Curtice. His final forecast for the 2024 General Election is Conservatives 130 seats, Labour 420, Lib Dems 50, SNP 25, Plaid 2, Reform 2,