There’s a very interesting piece of analysis by “Ian and Charlie” comparing aspects of the three referendums held in Scotland, in 1979, 1997 and 2014. It can be found at this link.

I’m not completely convinced by all of the arguments made, particularly because the 2014 referendum was different in a pretty fundamental way from the 1979 and 1997 ones, in that the potential outcome was a far greater step to take and irreversible as well.

Still, it’s well worth having a look. Decent set of comments added as well.

I made a comment myself, on the matter of the 1979 40% rule, which I reproduce here:

“Hi,

Good work here, thanks.

Just one comment regarding the 1979 40% rule. The issue was never the simple fact of having a hurdle, albeit that tacking it on as an afterthought did seem sneaky at the time.

The issue was that the electoral roll of the time was not designed to be used in such a way. Following the introduction of the Poll Tax, the roll was effectively rebuilt from scratch and with the subsequent passage of time it has become a decent register of those eligible to vote, particularly now that we have 97% registration. In 1979 it was a different animal, however.

It wasn’t that the register wasn’t updated regularly; it was more that the approach taken was to err in favour of allowing a vote rather than the reverse. Being on the register twice wasn’t viewed as a big issue.

At that time people were added to it as they became 18 or as they moved to a new address. What didn’t happen anything like as diligently, however, was removal of those who had died or of people from their old addresses. Newspaper reports of the time highlighted whole streets of tenements in Glasgow that no longer existed yet still appeared on the roll, complete with electors. I knew of Yes supporters who complained that, because they were on it in two places, they could at best register only a neutral effect.

So, you need to see the 1979 result in that context. There was an adjustment made of about 100k (I think) to the overall population figure, but the talk at the time was that many times that would have been more like it. We may never know what the correct figure should have been (though – in theory at least – it may be possible to compare the 1979 register with that from, say, the late 1980s).

That had the effect of inflating the number of Did Not Votes, so depressing the perceived percentages of those who did vote, i.e. both the Yes and No counts. From memory, the motion would likely still have been lost but not by much at all.”