How should we interpret the early results in the Scottish referendum?

One challenge for those of us trying to interpret the early results (whether you’re trading, spinning or just really want to know) is benchmarking each result as it comes in, given that there are likely to be significant regional dispersions. If the first local authority that reports goes 60-40 one way, which side is that good news for?

One thing that strikes me is the uniformity of swing between the 1979 and 1997 referenda, despite them being 18 years apart and the fact that the swing was very large (22.7 points). See my earlier post for the complete numbers.

Clearly independence and devolution are two very different things, but given that some areas of Scotland will be markedly more nationalist than others, and there are no other proxies available for this variation, I thought it would be interesting to examine what a Scotland-wide 50-50 tie would look like in each area if the regional dispersion were the same as in the 1997 devolution referendum. Equivalently, this represents what would happen if we took the 1997 devolution result and applied a uniform swing of -24.3 (to make it a 50-50 tie across Scotland as a whole). The local results generated could then serve as a benchmark for early results in this referendum.

Some health warnings:

  • As mentioned already, the question is different and the implications of that might not be uniform across Scotland (banks, Trident, etc)
  • 17 years, obviously, is a long time
  • The new local authority areas are much smaller than the ones used in 1979 (and hence also in my 1979 vs 1997 comparison), so there could be a greater variety of swings. Additionally, smaller local authority areas might be more sensitive to demographic shifts in the intervening period
  • A very high turnout (particularly by modern standards) is widely expected, and these increases might not by evenly spread geographically, which could be significant if the “don’t normally vote” group lean one way

Nevertheless, for the purpose of interpreting early results, it is important to get some feel for the kind of way those votes could be spread around.

So, with all those words of caution in mind, here is that uniform swing example:

If this is representative, it is clear both sides would need to win big in their stronger areas to have any chance of winning overall.

This analysis is intended as a starting point. If anyone has suggestions on how to improve it, feel free to comment.

For further updates, pleaseFollow @NCPoliticsUK





About The Author

Related Posts

  • Timothy (likes zebras)

    Possible improvement, or at least a sanity check. If you can aggregate your figures to match the regions used in the Survation polling then you have some level of check as to whether the pattern is still roughly the same.

    You would probably want to aggregate the results from several Survation polls to do this – you can find the data tables linked from https://en.wikipedia.org/wiki/Opinion_polling_for_the_Scottish_independence_referendum,_2014#2014

    Thanks very much for what you have done already. Do you have any idea which local authority area is likely to declare results first?

  • http://www.blogger.com/profile/15819413906544791899 Andy JS

    My current prediction based on lots of factors including the 1979 and 1997 referendum results, as well as other indicators such as demographic data and, to a lesser extent, recent anecdotal reports:

    https://docs.google.com/spreadsheets/d/txcTnTqEF6hmKvzevjiZyZw/htmlview#gid=0

  • http://www.blogger.com/profile/15819413906544791899 Andy JS
  • http://www.blogger.com/profile/01575922859885441220 Number Cruncher

    Timothy, thanks for your comment, and it’s my pleasure! I’ll take a look at the regional poll breakdowns, but I’m always wary of crossbreaks becuause they often aren’t demographically weighted at that level. But it’s worth a look and I’ll probably do that.

  • Anonymous

    I would think that Dundee,Angus and Perth are likely to be atleast 10% higher for yes than the figures that you have. (They all return SNP MPs,MSPs and control the local councils) (Moray being Salmond’s own area is likely to be pro Yes aswell)
    On the other hand I think that the 3 main cities are all likely to be 5% lower than what you have got. (Highland aswell is likely to be lower)
    They way that things are likely to declare time wise Yes will probably take the lead and it will only be at the end when the stronger No areas of Edinburgh,Borders and Aberdeen are likely to declare.
    If I had to guess of a region likely to mirror the Scottish vote overall someplace like Fife would be good indiactor. It has a good mix of types of votes from the more affulent NE area which has returned either liberal or Conservative MPs for last 30+ years and then the more SNP/Labour areas of KIrkcaldy.Glenrothes and Dunfermline. (It is a pity that it may be one of the last to declare…)
    James

  • Anonymous

    https://mobile.twitter.com/cllrdmeikle/status/510385027721736192/photo/1
    For declaration times (and rough guide to how likely places are to vote Yes)
    James

  • http://www.blogger.com/profile/01575922859885441220 Number Cruncher

    Andy, thanks for the spreadsheet… I’m actually working on a different version based on party representation which seems to be giving very similar results to yours (in fact, everywhere besides Borders and the islands the two models are within 5 points of each other).

    James thanks for the comments, the new model I’m working on has different results that are broadly in line with what you thought, with Fife as a bellweather and Angus, Perth and Moray all much higher. Dundee is also higher but not by as much. The big 3 cities collectively are lower (though not by a huge amount.)

    I’ll post the revised model tomorrow.