Evan Williams wants to talk to you about how voters switched in the 2017 general election in Scotland, and would quite like to do so over a pint.


There has been quite a lot of discussion about tactics and campaigns and the impact of the Tory campaign, Brexit, indyref2 and all that stuff. I have some thoughts about those things, and would be happy to discuss them over a few beers. But I thought it would be interesting to try and take the Scottish results and some of the helpful analysis that has been done by YouGov and the British Election Survey to take a stab at working out how voters in Scotland switched between 2015 and 2017.

The first non-trivial caveat is that 2015 was an unusual election in Scotland in lots of respects. In 2015 the SNP had a stellar campaign that got traction across the UK and the Tories used attacking the SNP as a means of shoring up their vote in the rest of the UK while essentially writing off the prospect of seats in Scotland. Comparing 2017’s snap election could be just comparing one odd election with a different odd election in Scotland. Also, comparing what happened in Scotland with England as if that tells us anything meaningful about the respective campaigns is a disservice to numeracy.

A second caveat is that the data is rarely available at a Scotland-only level, although swings from the SNP are identifiable even in national results because they helpfully only stand in Scotland, but that leads to very small sample sizes and we all know what small sample sizes mean.

I expect you will have noticed the rather nice diagrams people have been using in election analysis like the one I have created below. They are known as Sankey diagrams and I’m a big fan. Check out sankeymatic.com to do your own.

A third caveat, I have included in my analysis a dod of people who did not vote (DNV) and have slightly artificially tried to keep the vote shares as % against each party, so there is a little sleight of hand with the numbers, especially for DNV, but that shouldn’t stop you getting the sense of movement that the diagram conveys. I have preserved the colour of the lines from the 2015 vote to make it easy to track where 2017 votes came from.

What I think the evidence shows us it that there were perhaps four important movements of people

  1. SNP to Conservative switchers – and these seem to have been concentrated in areas that might have been thought of as Conservative in the 80s and 90s.
  2. Labour to Conservative switchers – not huge, but a noticeable number of 2015 Labour supporters making that move. My guess is that these were people for whom the Union is a bigger issue than left-right politics.
  3. SNP to Labour switchers – slightly more than the Labour to Conservative switchers, presumably predominantly people who switched to the SNP from Labour some time from 2007 onwards.
  4. SNP to did not vote – predominantly people who are reluctant voters but helped to account for the very high turnout in 2015.

The DNV to Labour switch observed in rUK seems to have been a bit muted in Scotland although that could simply be because those people did vote in the unusual circumstances of 2015.

None of these movements seem to be the kind of more or less uniform swings we are used to at election time. They are also in part quite geographically specific movements, which is in part why everyone had so much difficulty in predicting the results, especially if they were using a 2015 baseline and a swingometer to do so.

It is tempting to suppose that most of those movements are fairly permanent, and you can easily construct a narrative that says former Tories who lent their support to the SNP in the 80s and 90s stuck with the SNP through to 2015 but some combination of Brexit, Indyref, Ruth Davidson and the leftward pitch of nationalist rhetoric made them feel comfortable switching back to their more traditional comfort zone. Equally you could argue that SNP to Labour switchers have “seen through” the SNP and are coming back home, and that the Labour to conservative switchers will come back to Labour if the issue of the union / indyref ceases to be the dominant frame in Scottish Elections. However, I don’t think we should suppose anything about the way people voted in 2015 or 2017 elections is set in stone.

First past the post elections have significant tipping points depending on whether an election is a 2, 3, or 4 way contest. In a fairly even Scottish 3 way contest Labour does well through a concentration of votes in the central belt, the Tories do well in the north east and the borders and the SNP struggle.

A lot of people will be doing hugely optimistic analysis about the small swings needed for Labour in Scotland, and its good to be optimistic, but there’s a really important caveat to such calculations: for every one of those SNP to Tory switchers who the SNP get back, Labour needs to win an extra vote from somewhere (probably the SNP) just to stand still.

The SNP have shown significant tactical nous over the years and I would hesitate to suppose they won’t be applying their very considerable political skills to what they need to do to get some of those voters back. The Tories must be working out how best to keep what they have and get even more.

Anyway, I’m off to support Neil Bibby’s Tied Pubs bill – if you want to discuss how Labour builds from here I’m happy to do so over a pint.