Well, if I had known my 1979 | 1997 | 2014 piece was going to attract so much attention I might have done another draft!
I’ve been enormously flattered by the attention the essay has received. I want to begin this follow-up piece by thanking sincerely everyone who read, shared and/ or commented upon it. I extend particular gratitude to David Torrance for sharing the article with his Twitter followers. He began the social media chain reaction that allowed the piece to find an audience.
I have received some fascinating responses, which frankly out-do the original piece in eloquence and imagination by some distance. In this post I want to gather together the key points from my postbag and offer some measure of response to them, to highlight the interesting counter-arguments as much as anything.
Before I do that, a thought occurs to me.
The social science articles I always find most stimulating are those that provide either the last word on a discussion, or the first.
The last word is generally a work of synthesis, drawing together the literature on a particular debate and finding a foundational truth that enables some sort of reconciliation of competing views.
By contrast, the first word is usually a work that points to uncharted terrain and redirects debate. Such articles are usually full of holes, disputed by all concerned and ultimately superseded. But they have value for being original.
I would never suggest my blog post about the three referendums is worthy of exalted celebration, but I would hope it might merit classification among the latter category of articles. In that spirit, I hope very much that finer minds than mine and more skilled statisticians than me can interrogate the data more thoroughly and see where it takes them. I’m just pleased to have set off some trains of thought.
That said, I have some points to make in reply to the main areas of criticism prompted by my piece (as if the original 5000 words was not enough).
There are four main critiques, which can be summarised as follows:
1) the three referendums all asked different questions – as such the motivations of voters can’t be compared;
2) there were different electorates for each referendum, because they took place several years apart – as such the voters themselves can’t be compared;
3) Labour and Lib Dem voters supported devolution in (particularly) 1997 but voted No in 2014 – so the yes cohorts were different; and
4) opinions polls suggest significant numbers of No voters in 2014 were actually voting for more powers, after “the vow” promised them – so the motivations of the No cohort in 2014 were different from previous referendums.
I’d also like to take the opportunity at this juncture to flag up one of my earlier pieces, After Yes, which offers some more optimistic (and, mercifully, less numerical) reflections on the indyref experience.
Well, here goes. I’ll try to be more brief this time…
Objection 1: the referendums asked different questions
It has been suggested in certain quarters that comparison of the three referendums is methodologically illegitimate, on the grounds that the electorate was not being asked the same question each time. As such I wasn’t comparing like with like.
I think I made it fairly explicit in my piece that the referendums were indeed quite different, but that there were similarities in the behaviour of the electorate each time.
In particular, two-Scots-in-five (or thereabouts) answered “yes” to the different proposals for changing Scotland’s constitutional status.
And then in 2014 we saw the staggering shift of 25% of the entire electorate out of the “did not vote” column and into the “no” column.
So what is interesting to me was that under very different conditions, broadly the same proportion of the electorate keep voting yes, but under the specific conditions of the indyref, a massive constituency of previously unmoved voters chose to register their opposition.
So I wasn’t saying the referendums were the same – I was arguing that the observed behaviours of the electorate carried some features of interest. Things don’t need to be identical to be comparable.
Which leads me to a second area of criticism:
Objection 2: the electorates weren’t the same
A similar methodological critique has been raised in terms of the specific human beings in the electorate being different for each referendum. As such (the critique goes) the behaviour of the electorate cannot be compared across time since we aren’t looking at the same people in each vote.
Well, to be blunt, it’s always a different electorate. It would be difficult to compare voting behaviour at all if we didn’t suppose some measure of continuity over time of a political community.
But at the same time, clearly the electorate in 1979 was very different to the 2014 electorate. I don’t think I tried to suggest otherwise and certainly don’t here.
But again, I return to the figures showing the behaviour of the electorate(s) across the three votes.
Surely what makes the findings interesting is the very fact that the Yes vote does seem to hold broadly constant in three different referendums over 35 years.
This continuity is achieved even with all of the major changes in personnel the electoral role must have seen in that period, not to mention the very different domestic contexts in which the votes took place (economic and legitimacy crisis in 1979, media support in 1997, media hostility in 2014).
I’m happy to acknowledge that this isn’t the most water-tight methodology in the history of the social sciences, but I am confident that the patterns in the observed behaviour of the Scottish electorate makes the comparison valuable.
To summarise, I wasn’t arguing that exactly the same people voted in certain ways to entirely identical referendum questions. I was making almost the opposite point – very different electorates in different referendums in different contexts voted in remarkably comparable ways.
But this leads me to what I think is a much more interesting and valuable critique.
Objection 3: people voted Yes in 1997 and No in 2014
In effect, the argument here is that Labour and the Lib Dems campaigned for a Yes vote in the 1997 referendum, while they were part of Better Together this time around. As such, voters with a strong party loyalty must have voted differently in 1997 and 2014, so the conclusions drawn about the Yes vote are not based on continuity within the electorate.
This must be true to at least some extent, and I recognise that my lack of attention to this point weakened the original piece.
I want to make a couple of points by way of rejoinder, while still acknowledging the usefulness of the critique.
Firstly, the proportion of the electorate identified as Labour or Lib Dem is substantially smaller in 2014 than it was in 1997.
My understanding is that, for the purpose of opinion polls, voters are identified as “Labour” etc by virtue of how they voted in the last election. The election immediately prior to the 1997 referendum was the 1997 UK General Election, in which Labour received 1,283,350 votes in Scotland and the Lib Dems received 365,353 – a total of 1,648,712.
Meanwhile the SNP – by way of comparison – received 621,550 – a full million fewer.
The election prior to the indyref was the 2011 Scottish Parliament General Election. The combined Labour and Lib Dem vote in the first ballot (which I’m using to indicate primary party identification) was 788,175. This is less than half the combined vote they got in 1997.
Meanwhile the SNP won 902,915 first ballot votes – half as many again as in 1997.
The point being that there were considerably fewer Lab/Lib voters around by 2014 – and we might presume that many of the ‘defectors’ headed to the SNP. The consequence for the indyref is therefore a significant decrease in the number of ‘party loyalists’ who would have been duty-bound to support the Labour or Lib Dem plea for a No vote.
(Incidentally, Lord Ashcroft used the 2010 Westminster election as his proxy for party identification, for some reason, so his polling figures aren’t compatible with the point I’m making here. There lies another interesting tale).
A second and related point to be flagged up here is that party and individual voter motivations do change over time. As I’ve argued above, what makes the solid 40% yes block so interesting is that it holds constant despite an ever-changing political landscape.
I’m not suggesting that it’s some sort of essential quality of the Scottish electorate that two-in-five of them must always vote Yes in constitutional referendums. But at the same time it would make sense for ‘reform enthusiasts’ to shift out of a previous identification with Labour or the Lib Dems and head where the action is – such as voting SNP in 2011. And clearly some did.
Therefore we mustn’t conflate the Labour/ Lib Dem constituencies in 1997 with those in 2014.
Despite these two points it’s certainly a fair assumption that a significant number of Lab/ Lib ‘yeses’ in 1997 must have changed to No voters in 2014 – and been ‘replaced’ in the Yes column by different voters. This adds a further layer of intrigue to the post-referendum analysis – and I leave this happily to others to unpick when the ballot box breakdowns are available.
Objection 4: No voters actually wanted more powers
We may find ourselves at times in the realm of the looking glass with this final area of critique, but I’ll try to present the final area of critique as fairly as I can.
Where the previous argument was about the make-up of the Yes contingent, this argument focuses on the No constituency. It is contended that the 2014 No voters were completely different from No voters in 1979 and 1997, on the basis that they weren’t really saying no. Enough of them were actually voting No to secure Gordon Brown’s home rule powers, whatever they were, to render any comparison with previous referendums redundant.
I should stress at the outset that this may indeed prove to be the case, but I don’t really know how we could possibly prove it either way. But let us consider the evidence we have available.
There is some polling data presented with typical skill by Prof Curtice that traces an increase in support for powers short of independence in the run-up to the vote. This would suggest that some voters got interested in the kind of powers promised by The Vow, although it doesn’t necessarily follow that anyone changed their minds on how to vote as a consequence.
I’m not sure how much store to set by opinion polls that proved misleading in terms of the final result, but maybe this was indeed a motivating factor for many No voters.
But can this explain 25% of the entire electorate choosing to vote in a referendum for the first time in order to record a No?
Did the promise of smoother, faster, fairer new powers (or whatever The Vow called them) actually motivate this massive stratum of the electorate to vote No?
As opposed to voting yes, or not voting at all?
Unlikely, I’d suggest.
I raise this because there seems to be a belief in certain quarters that people were swithering on voting yes, but backed out at the last minute after Brown’s intervention and are therefore ripe for turning into future Yes voters. While this is doubtless true of some people – and Curtice’s typically forensic exploration of the polling data is excellent in drawing this out – I find it hard to believe the sheer weight of the 2014 No vote can be accounted for by people who actually wanted more powers.
After all, voters were assured that “No didn’t mean No” (i.e. that more powers would follow if the electorate rejected devolution) in the 1979 referendum too. Yet the “did not vote” and “No” columns are dramatically different in 1979 and 2014.
I don’t want to over-play my hand here, because I can’t get inside the heads of the 2m No voters to truly understand their motivations – but I would be inclined to doubt that the 25% of the electorate who were previously unmoved either way by devolution referendums suddenly found an appetite for the Vow powers this time around. On the basis of Occam’s Razor I would suggest that, for that section of the electorate at least, no meant no.
I want to conclude by flagging up a couple of interesting implications for future Scottish politics that commenters have identified.
Firstly, one commenter on the blog and a couple of other people in the world of Twitter have noted that a huge chunk of the electorate re-connected with voting in 2014, to oppose independence. What if they keep voting?
This is an intriguing thought. Just imagine, for argument’s sake, that 25% of the electorate were stirred from their habitual disinterest in voting in order to reject independence (as opposed to voting to save the Union, which is a subtle but important distinction). Say they are motivated above all by hostility to the SNP. And imagine they continue to vote in future to prevent further SNP gains. That would utterly transform the electoral calculus in Holyrood.
There are a number of “ifs” in that thought experiment and I wouldn’t expect it’s terribly likely. But it’s a scenario worth thinking (or panicking) about.
A related development (which isn’t really relevant to my essay but I raise for interest) is the sudden explosion in political party membership in Scotland. Some of the Yes parties, if we may call them that, have quite democratic policy-agreement mechanisms. The doubling or trebling of their memberships injects a fascinating degree of uncertainty into party agenda-setting in the coming months, since we don’t really know who these people are.
While I would imagine that many of the new party members chose to join up as an expression of solidarity rather than a commitment to seeking influence, some of the members will influence policy. And they might be quite radical. It will be interesting to see how the SNP in particular handle a series of demands from the floor for another referendum.
But I digress.
I want to conclude by thanking everyone again for reading my piece and for engaging with it so passionately. It has been a bizarre and wonderful experience to find one of my essays discussed so widely.
As a WordPress author I have access to a breakdown of the countries in which my pieces are read and the sites from which readers connect with them. It was a very pleasant surprise to learn that my essay found readers in places as far-flung as Peru, Albania and Argentina, as well as being shared on a cycling blog and an Andy Murray forum. The power of Twitter!
And with that I will move on to writing about something less taxing of my rusty maths skills.