It has been two weeks since Scotland’s independence referendum and we’re not short of analysis of what happened and why. However I think there is an angle that has not been explored in sufficient detail, namely the links between the indyref and the two previous referendums held in Scotland in 1979 and 1997.
For this post I have ploughed through the raw voting data from each of the three referendums. I think my findings turn the received wisdom about each referendum on its head to some extent.
I will begin by recapping the context of each of the three referendums, before sketching out the received wisdom on them. I will then explain briefly how I went about comparing the data from each vote. After that I will argue the following:
• Scotland voted yes with far greater enthusiasm in 2014 than in 1979 across the entire country;
• Support for independence in 2014 was basically as high as support for tax-varying powers in 1997; and
• A staggering 25% of the entire electorate expressed an opinion on Scottish statehood for the first time in 2014 – but they actually voted no.
The three referendums
I’ll begin by noting that I’m using the word “referendums” rather than “referenda”. You’ll have to indulge me this stylistic decision.
Scotland has conducted three referendums on its constitutional status.
The first, in 1979, asked Scottish people to indicate whether or not the provisions of the 1978
Scotland Act (specifically, the creation of a Scottish Assembly) should be put into effect.
The popular memory of this vote was that Scotland voted yes (which it did, by 51.6% to 48.4% of votes cast). However (the story goes) devolution was stolen from a willing electorate by an act of skulduggery by the Thatcher government, whereby an additional voting threshold of 40% of the entire electorate had to be met. It wasn’t, and so the result was annulled and the Scotland Act was repealed. (Incidentally, this is one allegation against the Tories which cannot be borne out, since the 40% rule was inserted to a Labour Act of Parliament by a Labour politician called George Cunningham.)
The second referendum was held in 1997. The Blair government invited Scots to vote on two questions: firstly, should there be a Scottish Parliament, and secondly, should that parliament have tax-varying powers. The popular memory of this vote is of an overwhelming Yes vote.
Finally, as everyone knows all too well, the Scottish electorate was out in force on 18th September of this year to indicate whether they agreed that Scotland should be an independent country. The consensus as I write appears to be that the Yes side whetted an unprecedented appetite for independence, but people lost their nerve at the last minute or were duped by The Vow of home rule. Or, depending on your viewpoint, people made a rational and reasonable assessment that the case for independence had not been made sufficiently to motivate them to vote for it.
So, to summarise:
• Scotland’s sovereign will was ignored in 1979;
• Scotland overwhelmingly asserted its desire for greater powers in 1997; and
• The Yes campaign broke new ground in 2014 but was let down at the eleventh hour by cowards and fools.
Or so the popular wisdom would have you believe in many quarters.
I hope now to convince you that these analyses are all inaccurate and disguise much more interesting trends in Scotland’s attitude to statehood.
How I generated and used the data
Methodology time, guys. I’ll be brief.
I faced two key questions when working with the data. Firstly, the question of how to authentically measure commitment to new powers; and secondly how to compare this commitment across three separate referendums.
Measuring Yes strength
The problem here is that referendum results are presented in terms of votes for and against. However if you ask twenty people a question, to which three say yes, one says no and sixteen don’t answer, you’d have a result of 75% in favour. This doesn’t really give an authentic flavour of that group’s commitment to the contention.
To get over this, I returned to the raw voting figures for each election and recalculated the results in terms of three categories:
• Voted yes
• Voted no
• Did not vote
The figures for each category now reflect the commitment of the entire electorate, rather than only considering those who voted. This, I would argue, gives a more authentic picture of popular support (or lack thereof) for new powers.
To illustrate, the example above had three yeses, one no and sixteen abstentions resulting in 75% yes from a 20% turnout. In my revised system the result would be presented as 15% yes, 5% no and 80% did not vote. This reflects more authentically the extent of popular support for the motion.
Comparing the referendums
This bit was trickier.
In 1979, the result was broken down into the twelve Regions through which Scotland was then administered. In 1997 and 2014, by contrast, the result was broken down into our 32 Local Authorities.
There was no way to recalculate the 1979 result in terms of our present Local Authorities, but it was possible to recalculate the 1997 and 2014 results in terms of the old Regions. So I recreated Strathclyde for the purposes of comparison. And my goodness that was a massive place!
To be clear, seven of the 1979 Regions remain as Local Authorities, so there is direct continuity between their results in the three referendums. They are:
• Western Isles (now called Eilean Siar)
• Borders (now called Scottish Borders)
• Dumfries and Galloway
The remaining five Regions were split into the following Local Authorities:
• Central: Clackmannanshire, Falkirk, Stirling
• Grampian: Aberdeen, Aberdeenshire, Moray
• Lothians: Edinburgh, Midlothian, West Lothian, East Lothian
• Tayside: Dundee, Perth & Kinross, Angus
• Strathclyde: Glasgow, North Lanarkshire, South Lanarkshire, East Dunbartonshire, West Dunbartonshire, East Ayrshire, North Ayrshire, South Ayrshire, Renfrewshire, East Renfrewshire, Inverclyde, Argyll & Bute
The weakness, clearly, is Strathclyde. I didn’t particularly want to aggregate together twelve different Local Authorities but there was no other way round it. However I have included a section at the end of my analysis where I compare the Local Authority results for 1997 and 2014, leaving out 1979. This hopefully does some justice to the nuanced differences within the old Strathclyde Region.
To summarise, then, I have compared the results across the three referendums in terms of the old Regions boundaries, and I have focused on Yes, No and Did Not Vote as proportions of the entire electorate when making comparisons.
As a final point, I should note that for the purposes of argument I am viewing Yes votes between referendums as comparable. Clearly, each referendum asked different questions, and I try to reflect that in my analysis, but as a foundational point I am asserting that a Yes vote in one referendum over additional Scottish powers is worthy of comparison with another. This may not convince every reader.
The main source for my figures from the 1979 and 1997 plebiscites is the following UK Government report:
http://www.parliament.uk/briefing-papers/RP97-113.pdf – and for the 2014 vote I used figures available widely, such as here: http://www.bbc.co.uk/news/events/scotland-decides/results
Due to my methodology I have extended and developed the figures in this report extensively, so any arithmetic errors are my own. The main effort required was to aggregate together the Local Authority votes to calculate figures for the old Regions. These figures were naturally not available in the public domain so I spent a very long time generating them. You can see my notebook if you like, but I hope you’ll take them on trust.
My other half counselled me wisely that the number crunching was an insane exercise. She was right, on this and so many other points. But I finished it nevertheless, so here are the fruits of my strange labours.
Overall Scottish results
The proportion of the overall population voting Yes at each of the referendums was as follows:
1997: 44.7% (Parliament)
1997: 38.1% (tax-varying powers)
The first thing you’ll notice is that support for Yes was considerably stronger in 2014, in a losing cause, than in victory in 1979. Between these two votes, support for new powers increased by a full five percentage points – or one-twentieth of the overall population. That’s a significant rise.
Secondly, support for Yes in a winning cause in 1979 came from less than one-third of the overall population. This was hardly a ringing endorsement. Perhaps the 40% clause, which prevented the Scottish Assembly being set up, wasn’t such a bad idea after all. I’ll say more about this later.
The third point of interest is that the strongest Yes vote, by some distance, was the vote for a Parliament in 1997. However it is perhaps worth noting that considerably less than half of the electorate (44.7%) voted for it.
The final (and for me, most important) point to emphasise is support for Yes in 2014 was basically the same as support for tax-varying powers in 1997. The difference between the votes was just 0.3% of the electorate.
The popular memory of 1997 is of a decisive expression of sovereign will for a parliament with teeth. However there was scarcely any more popular support for tax-varying powers then, than there was for independence last month. Or, put another way: support for independence was comparable to the support for tax-varying powers in the decisive 1997 victory.
But let’s look at the proportion of the electorate who did not vote in each referendum.
1997: 39.8% (Parliament)
1997: 40.0% (tax-varying powers)
As you can see, about two out of every five people on the electoral register (40%) did not bother to vote in 1979 or 1997. However the proportion of non-voters dropped to 15.5% this year. To be clear, a full 25% of the electorate, or one in four people on the electoral register, voted in 2014 when they did not vote in previous referendums.
It should be noted that these are not necessarily the same people – I certainly don’t have access to the sort of the data sets through which this could be tested. But we are talking here about swathes of the electorate, rather than identifiable human beings. And one quarter of the electorate did not vote in 1979 or 1997 but did so in 2014.
If we accept that broadly the same proportion of the electorate backed tax-varying powers in 1997 and independence in 2014, there’s only one place these new voters can have gone.
Here is the proportion of the electorate voting No at each referendum:
1997: 15.5% (Parliament)
1997: 21.9% (tax-varying powers)
As you can see, the proportion of the electorate registering its opposition to new powers for Scotland was dramatically higher in 2014 than in any previous referendum. Indeed, the 46.7% of the electorate who voted No last month was larger than the Yes or Did Not Vote figure from any of the three referendums.
This was a decisive rejection of independence by close to half of the population, when devolution had never been rejected by even one-third of the electorate in previous plebiscites.
If I may hazard some analysis, perhaps the excitement generated by the Yes movement’s success in increasing the size of the electorate, by registering new voters, led many of us to assume that votes entering the electoral bloodstream would usually be Yes votes.
What this obscured was a much greater injection of new votes – the votes of people already on the electoral register but who hadn’t previously voted in referendums. And a decisive number, equivalent to a quarter of the electorate, broke cover to vote No. There is no question that this development won the referendum.
To conclude, a sizeable minority of the population, of almost two in five Scots, has always voted for new powers.
And essentially the same proportion of the electorate who supported a Parliament with teeth in 1997 were mobilised by the Yes campaign in 2014.
But a staggering 25% of the adult population, who had previously abstained in referendums on Scotland’s constitutional status, chose to record a vote in 2014.
And they voted No.
(As a point of clarification, the addition of thousands of people onto the electoral register increased the overall size of the electorate, so more votes were needed to secure the same overall proportion of the vote at roughly 38%. Bearing this out, 1.5m voted for tax-varying powers in 1997, while 1.6m voted Yes in a larger electorate in 2014.
Perhaps this explains where all of the new votes went. The votes of many newly registered voters were sufficient to ensure that the same proportion of a larger electorate voted Yes in 2014 as had supported tax powers in 1997. However they were massively outnumbered by the No votes of people who had previously abstained.)
Casual readers are invited at this stage to skip straight to the conclusion, since the next couple of sections serve to underline the points made above. But you’re welcome to read this bit if you can be bothered with some numerical detail.
I turn now to my analysis of the voting patterns by Region. I should remind you that I aggregated together the Local Authority figures from 1997 and 2014 to make them comparable with the 1979 referendum that was conducted using different administrative units.
The chart below traces the proportion of each Region’s population voting Yes in each of the referendums. I apologise that you have to click on a link to see it – my present level of technological access precludes me from embedding the image in this blog.
Fig 1 – Regional vote by overall electorate
The first point to notice is that in all twelve regions, the Parliament in 1997 commanded the strongest Yes vote out of the various referendum questions.
Secondly, in all twelve regions, the lowest Yes vote was recorded in the 1979 referendum.
Thirdly, and for me the most important point, is that support for independence in 2014 was broadly comparable with support for tax-varying powers in 1997 across all twelve regions.
In fact, support for independence was higher than support for tax-varying powers in six out of twelve regions, often by significant margins (a 5.7% point rise in Tayside for example).
Indeed, of the six regions that recorded a lower level of support for independence than for tax-varying powers, three were within 2% points. The only really notable drop in support was in Lothian (down 4.4% points) and the Borders (down 3.4% points).
It seems reasonable to conclude that:
• support for new powers has grown significantly since 1979;
• support for new powers was maintained between the second question of the 1997 ballot and the 2014 independence referendum; and
• these phenomena are true across Scotland
We can now look at the trends for No votes across the twelve regions:
Fig 2 – Regional No vote by overall electorate
A clear stratification emerges. The No vote was overwhelmingly stronger in 2014 in every region than in any other referendum. The 1979 referendum inspired the second-strongest No vote in each region, while tax-raising powers generated the third-highest No vote and the Parliament the lowest No vote. In each region there is clear daylight between each referendum in terms of strength of opposition.
The pattern of No voting is distinct and suggests that opposition to new powers is nuanced in terms of what is being asked. The Parliament occasioned the least opposition, while independence inspired the most. It may be argued that the 1979 referendum inspired greater opposition than the 1997 ballot since it was the first vote of its kind and was perhaps seen as more of a ‘slippery slope’ to independence than the circumscribed 1997 plebiscite.
Finally, I turn to those who did not vote:
Fig 3 – Regional Did Not Vote by overall electorate
As the chart demonstrates, the proportion of the electorate who did not vote was dramatically lower in 2014 than in any previous referendum, sitting at about 15% in every region.
In the other three referendums, each region (minus Orkney and Shetland) recorded broadly comparable abstention rates of between 35% and 45%. This links to the above point: 2014 mobilised opposition across Scotland where previously there had been much more considerable levels of abstention.
To summarise this section, I would emphasise the following:
• the Yes vote has grown across Scotland’s regions since 1979;
• the Yes vote was largely maintained across the regions between the tax-varying powers question in 1997 and the indyref in 2014;
• there was a significant shift from abstention to opposition (or from did not vote, to no) in 2014 in every region.
Yes areas and No areas
At the risk of getting into a cycle of blame, I will now outline where support for new powers has been strongest and weakest across the three referendums.
I will persevere with the Regions as an analytical category at first, before focusing on the more forensic Local Authority figures from the 1997 and 2014 plebiscites.
It’s easier to start with the Scottish regions that have shown least appetite for new powers.
In 1979, 1997 (tax-varying powers) and 2014, the lowest levels of support for new powers have been recorded in Shetland, Orkney, Dumfries & Galloway and Borders. In the 1997 Parliament referendum they accounted for four of the five lowest levels of support.
In 1979, support for Yes was half the national average in Orkney and Shetland, and three-quarters of the national average in Borders and Dumfries & Galloway.
In 1997, Orkney and Shetland only offered two-thirds of the support for a Parliament delivered in the country as a whole, while Borders and Dumfries were more than four percentage points down on the average figure. Support across the four regions for tax-varying powers was even lower.
Finally, in 2014, the four regions recorded about 30% support for independence against a national figure of 37.8%.
It will hardly escape the notice of budding geographers that Borders and Dumfries & Galloway are the regions nearest the English border, while Orkney and Shetland are Scotland’s most northerly islands. That is to say, they’re the regions furthest from the central belt in either direction.
So, are they the guilty men and women?
Well, I wouldn’t say that. For a start, Orkney and Shetland are recording massively increased levels of support for new Scottish powers these days. In 1979 the proportion of the islanders supporting Yes was remarkably low – fewer than one in six Orkadians and Shetlanders supported the Scottish assembly. In 2014 almost one in three voted Yes. Support for Scottish powers has doubled on Orkney and Shetland over 35 years (albeit from a notably low base) which hopefully insulates them from allegations of backsliding.
Meanwhile, in Dumfries & Galloway and the Borders, the support for Yes in 2014 was largely the same in the 1979 referendum, in the tax question in 1997 and in 2014, at around 30%. This looks more like a stratification than the Orkney and Shetland figures. The vote has, therefore, held up in the south without showing much prospect for growth.
When the No and Did Not Vote figures are examined, an intriguing point emerges for these four regions. The proportion who did not vote in 1979 and 1997 is essentially the same for the two border authorities (at about 35%) and for the two northerly islands (at almost 50%). However in 2014 all four authorities saw their abstention rates fall to around 15%. This added 20% to the No vote in the borders and 30% to the No vote in the northern isles.
When we turn to the regions that have backed new powers at each election, four regions feature in the top six for all four referendum questions over the three votes; namely Central, Strathclyde, Fife and Highland. The remaining places in the respective top sixes are taken by Lothians and Western Isles (three appearances out of four) and Tayside (two appearances). Strathclyde, it should be said, was never out of the top three and Central led the country in 1979 and both ballots in 1997.
The pattern among the Highland, Central, Fife and Strathclyde regions has been a gradual rise in support for yes from around 33% to about 40% across the referendums, while Tayside and the Western Isles reached the same end-point from a lower base.
These six regions recorded a remarkable increase in the No vote between the 1997 Parliament vote and 2014, of 25% points in the case of Tayside, 30% points in the cases of Highland, Fife and Strathclyde, and 35% points in the cases of Western Isles and Central. This means that, in the regions with the greatest appetite for new powers, the proportion of voters who shifted from abstentions to No votes was higher than in the most No-voting regions. This left the Did Not Vote rates in the No-region and Yes-regions looking much the same.
Each region contained a large swathe of Did Not Votes in each referendum prior to the indyref. 15% of the electorate would remain in the Did Not Vote category in each region, while the remainder would break cover to vote No in 2014.
In light of this, it seems churlish to highlight Yes areas and No areas. What the 2014 figures suggest is:
• 15% of the electorate (three in twenty voters) will not vote in referendums in any region;
• about 30% of the electorate in the borders or northern isles (six in twenty) will vote for new powers;
• and 40% of the electorate everywhere else (eight in twenty) will do the same;
• about 20%-30% (four-to-six in twenty) in all regions will oppose new powers in all circumstances;
• this leaves between 25% and 35% of the electorate in each region who abstained in 1979 and 1997…
• and then voted No in the indyref…
There is one final data source to consider – the 1997 and 2014 Local Authority comparisons.
Now, this is less significant in analytical terms since it excludes the 1979 referendum, but may be of local interest.
In the two ballots in 1997, the strongest Yes votes were recorded in the following seven Local Authorities:
• West Dunbartonshire
• East Ayrshire
• East Dunbartonshire
• North Lanarkshire
In passing we may note that Glasgow was placed twentieth on the list for supporting the Parliament and sixteenth for supporting tax-varying powers, while Dundee was nowhere to be seen.
In 2014, the top seven looked like this:
• West Dunbartonshire
• North Lanarkshire
• North Ayrshire
• South Ayrshire
This represents a significant shift in the concentration of Yes support to to the central belt.
In 1997 the lowest levels of support were recorded in:
• Dumfries & Galloway
In 2014, the lowest levels of support were recorded in:
• Dumfries & Galloway
While two of the lowest five Local Authorities have changed, they are still heavily concentrated in the same regions (namely the northern isles, the borders and the North East).
There are only two referendums of data at Local Authority level, however, so it would be ambitious to draw too many conclusions from the above points. The reader is invited to use her imagination if she wishes to.
I think the data make a few things clear, and suggest a few other points in more tantalising form.
Here are the strong conclusions:
• There is a solid block of the overall Scottish electorate that has voted for new powers whenever offered the chance;
• This group of enthusiasts amounts to just under 40% of the electorate, or two in five of the adult population.
• Even in regions of Scotland that have not shown a huge appetite for new powers, such as Shetland and Orkney at one end of the country and the Borders and Dumfries & Galloway at the other, the vote holds up whenever Scotland’s constitutional status is in question.
So the Yes cohort is strong, sizeable and enduring.
• In the 1979 referendum, and in response to both questions in the 1997 referendum, about the same proportion of the electorate (40%) opted not to vote.
• But then, in the 2014 indyref, a staggering 25% of the entire electorate, which had previously been in the Did Not Vote category, voted No. And they did this across Scotland, irrespective of the region.
They are the clear findings from the data which I don’t think can be disagreed with. But that’s not much fun, so here are some more contentious speculations:
One quarter of the Scottish electorate had not been roused to vote in any previous referendum, but they broke cover to vote No in 2014.
The proportion of the electorate that supported the Parliament in 1997 and (especially) the parliament’s tax-varying powers was considerably lower than the proportion that voted No in 2014;
As such it may be contended that the creation of a Parliament with teeth in 1997 was not really a decisive expression of the sovereign will of the Scottish electorate as a whole. Instead, it was the will of the reform-enthusiasts among the population unimpeded by the lurking opponents of constitutional change in the country. The prevailing media narrative of a Yes landslide, and the active support of all major parties except the (at that time despised) Conservatives, occluded this ambiguity.
• In 2014 a comparable share of the population backed independence as had backed tax-varying powers in 1997 – but this time one quarter of the population stepped in to prevent a Yes victory.
As such the story of the 2014 indyref was not one of people losing their nerve – they didn’t. The Yes vote held up. The story was one of a previously silent section of the electorate concluding that independence was a line they were unprepared to cross, and a prospect that they were prepared to break with the habit of a lifetime and vote in a referendum to prevent.
The Yes campaign can therefore be seen as having failed to enlarge its social base beyond the reform-enthusiasts who always vote for new powers. Those of us who were there know very well that the Yes campaign made us all commit much more deeply to the cause than ever before – but we didn’t actually attract new converts. All we did was maintain the roughly 40% of the electorate that support new powers for Scotland.
This might be summed up as a deepening of commitment to new powers, but not a broadening of the support base.
Remarkably, the more electorally-effective campaign was the No campaign, since it coaxed one quarter of the electorate out of their habit of abstention to vote against independence. Now, anyone who saw the Better Together campaign knows this did not happen through its hapless efforts. So how did they pull that off?
Well, it is clear that a fairly unchanging block of the population votes for new powers in every election, and a similar proportion do not usually vote at all. This time, the non-voters overwhelmingly supported the No side.
Is it plausible that they were Yes voters who were taken in by The Vow? That is, reform-enthusiasts who chose to vote No after gaining assurances from the major unionist parties?
I don’t think this seems likely. The same proportion of the electorate who voted Yes in 2014 also supported the much tamer propositions of a Scottish Assembly, a Scottish Parliament and tax-varying powers. It’s more likely that the Yes voters in previous elections backed independence, while the previously dormant abstainers saw something in independence they did not like.
To explain – if about 40% of the electorate voted Yes in 1997 and again in 2014 it makes sense to attribute them to the same category of reform-enthusiasts. The citizens who previously didn’t care but who this time voted No do not seem plausible waverers.
There are two extensions to this supposition.
• Firstly, the Yes campaign did not lose votes at the last minute. They were Did Not Votes at best (although if they hadn’t voted then Yes would have won).
• Secondly, the most convincing explanation for Did Not Votes turning into No votes is that they were motivated sufficiently to oppose independence. They’d never backed extra powers before – they’d just abstained. So it seems unlikely that The Vow of home rule made them vote No.
What this suggests to me is that The Vow, made under conditions of widespread panic on the basis of a single opinion poll, was probably a hasty and unnecessary concession by the unionist powers. It cannot have mobilised people who never vote for more powers to vote for home rule. They don’t care about new powers.
So the home rule powers extracted from Downing Street by Gordon Brown, if they ever materialise, are a massive bonus and we should all have a chuckle about this. The No campaign were going to win by miles anyway, probably thanks to Project Fear more than anything, but we got some promises of additional powers all the same. Nice one guys!
I suppose that would go down as the Better Together campaign’s just desserts. A campaign that ticked every opportunistic, negative and scare-mongering box might yet enable the prevailing English Votes for English Laws (EVEL) mania to rip apart the fabric of the union, on the back of an injudiciously cobbled-together home rule bribe.
And wouldn’t that be funny?
Edit: a follow-up piece summarising the various areas of critique prompted by this essay is now available here