Rob Johns is professor of politics at Essex University and an election night analyst on STV. In the first of a series of blogs, he looks at how Scotland might vote on December 12.
Is there everything to play for at this election? It looks that way in Scotland.
Defining a marginal seat as one which would change hands with a 5% swing to the second-placed party, no fewer than 46 out of 59 Scottish seats are marginal.
Since the last two Westminster elections in Scotland have seen much bigger swings than that, MPs travelling from London to their constituencies for the campaign might well have wondered whether it was worth booking a return ticket.
So what's going to happen? It is harder to say than usual because Scotland has been quite 'under-polled' in recent months. While the opinion polls have hardly excelled recently, they remain the worst way of predicting elections except for all the others.
Pending a flurry of Scottish-only polls later this month, we are reliant on Plan B: combining the Scottish subsamples (usually between 100 and 150 voters) from Britain-wide polls. Pooling those samples across eight such polls, all conducted between October 29 and November 6, we can draw pretty reliable conclusions about how things look now compared with the previous UK general election.
The graph comes with caveats. One concerns turnout. The SNP's eventual eight-point advantage over the Conservatives in 2017 was several points smaller than suggested by pre-election polls, not because of a late swing but because the SNP has more difficulty getting its vote out.
The same is likely to happen again, not least because bad weather tends to exacerbate the usual social and demographic biases in turnout. Another concerns small parties. Some of those expecting to vote Green or for the Brexit Party may find no such option on their ballot paper (although both are set to stand many more candidates than in 2017).
But the bigger picture is clear: the SNP is in a better place and the Tories and Labour in a worse place than two years ago. So the marginals held by the SNP look safer than the often wafer-thin majorities suggest.
Of the nine seats in which the party squeaked home by fewer than 1000 votes, only one would be in play given swings like those in the graph above. That is the most marginal of all, North East Fife, won by just two votes ahead of the Liberal Democrats in 2017.
That party's resurgence has been more limited in Scotland than elsewhere in Britain. Still, with strong local roots and the potential to squeeze the Tory vote in a seat estimated to have voted 64% Remain, the Lib Dems are reasonably the bookies' favourites there.
Meanwhile, the SNP looks likely to retake much of what it lost in 2017. The table shows where the party would regain control if the trends in the graph above were replicated Scotland-wide. As noted, stronger turnout among Conservatives means that this probably overstates how many blue seats will turn yellow. And this type of simple calculation has always been better at predicting how many seats will change hands than which seats will go.
Uniform national swing is a thing of the past. Individual results can be swung by popular incumbents, big local issues or the intervention of a third party. Still, the message of the table is plain. Current polls point to something more like the yellow-wash of 2015 than the narrower SNP majority of 2017.
If 2019 marks something of a return to 2015, it is worth thinking about who shifted between 2015 and 2017. There were four main types of switching between those two elections.
By looking at different types of voters and tracking their attitudes toward the parties, we can get a sense of whether and why they might switch back.
The graphs below are based on data from the British Election Study's internet panel, showing parties' average likeability scores on a scale from 0 to 10.
From SNP to Did Not Vote
Turnout increased across the UK in 2017 but fell by five points in Scotland as the referendum effect slowly wore off. The SNP in 2015 profited from unusually high turnout among Yes voters in socioeconomically deprived areas. Many drifted back into abstention in 2017 and it is hard to see that changing in 2019. This is one reason why the SNP will struggle to get anywhere near the 50% that they polled in 2015.
From SNP to Labour
As the Corbyn fightback gathered pace in 2017, many left-wing voters in Scotland refocused from the independence question to the Labour-Conservative battle. The graph below is based on all those placing themselves in positions 0 to 3 on a left-right scale from 0 to 10. As it shows, those voters warmed to Labour in 2017 - but have cooled again since. The SNP is now again the favoured party among this group, and without a repeat performance from Corbyn it will sweep away Labour's gains in the central belt.
With the overall results showing an SNP slump and a Conservative surge, it looked as if many voters switched directly from one to the other. In fact, this was probably the least common of the four forms of switching. For every Leave voter that went to the Tories, three more stayed with the SNP, reflecting the fact that most Scots have stronger views about the UK than the European Union.
However, a concentration of stauncher Brexiteers in the north east helped the Conservatives to some huge swings in 2017. The graph, based on those who voted Yes in 2014 and then Leave in 2016, shows just how far the SNP's Remain stance has lost the party sympathy among this group.
Yet the Conservatives have made no headway and, while Boris Johnson's Brexit deal might help, his pledge never to agree to a second independence referendum will have the opposite effect.
From Labour to Conservative
Given the polarisation of opinion on independence, there is electoral profit in being seen as the most unionist party. Ruth Davidson's 'we said no and we meant it' slogan helped capture that ground for the Conservatives. With the SNP now the primary enemy for many Labour supporters, they could think the previously unthinkable and vote for the Tories - who thus won almost 50% of the unionist vote in 2017.
Will they do the same again? The graph, showing sympathy for the unionist parties among 2014 No voters, suggests that many will. On the one hand, Davidson has gone and the 2019 Tories look harder to swallow for Remainers. Yet Labour's ratings have slumped even faster - they are now barely more popular among No voters than among Yes voters.
There is much talk in England and Wales of pacts on either side of the Brexit divide. In Scotland, it is the Yes/No divide that matters most and the outcome will depend on how effectively the No vote is coordinated. With the collective popularity of the unionist parties on the slide, such coordination is all the more crucial.
As ever with tactical voting, there are two stages: 1) working out which party is the strongest challenger to the SNP; 2) being willing to vote for that party. Both can be difficult. Only with a lot of calculation and a lot of voting through gritted unionist teeth will the SNP be denied the majority that is the cornerstone of its demand for indyref2.
You can follow Rob on Twitter: @robjohns75