Scottish Independence: from polling to voting

John McTernan writing in The Scotsman on 7 March gave his article, What next for Scots after No vote?, over to an examination of the politics of Scotland after September’s independence referendum. He made the assumption that the No vote will prevail. Nothing in the opinion polls suggests he could be wrong, but I thought it might be enlightening to look at a recent opinion survey in some detail to see what might underlie the way it was reported in the media. For example, on 4 March the Daily Telegraph, under a headline, Independence support drops after George Osborne's Scottish pound warning, compared the latest Ipsos MORI poll with an earlier one and concluded:

Overall, support for leaving the UK was down two points at 32 per cent compared to an identical survey conducted in December, while opposition to independence remained unchanged at 57 per cent.

Ipsos MORI have made the details of their survey available to anyone who wants to look at their Tables of data which are broken down as usual under various headings, eg age, gender, social class and so on. (Pollsters tweak their sample (weighting) to match what they think is the profile of the actual electorate – such adjustments are small and their effect won’t be discussed here). The interesting thing about a referendum is that on the day the votes can only be YES or NO, the only other options being to spoil your vote or not go to the polling station at all. But drawing conclusions from the responses to a survey like this one of 1001 people seven months beforehand is more complicated and I’ve attempted to indicate why in the diagram below:

At the top are the responses of the 1001 and, at this stage in the referendum campaign, as well as Yes and No, people quite reasonably can be Undecided. Also, some of the Yes’s and No’s have not made up their minds definitely (percentage figures from Table 3) and similarly some of those who feel Undecided may, nonetheless, have an inclination towards Yes or No (Table 22). The arrows indicate the changes in opinion that these uncertainties could lead to.

The survey asked the 1001 people how certain they were to vote on a scale from 1 (absolutely certain not to) to 10 (absolutely certain to) – 12 didn’t know (Table 24)! There were 779 who were 10s and in the middle layer of the diagram these are again broken down into Yes, No and Undecided. Again, some of these people, albeit definite voters, may yet change their minds (Table 8). It is the nominally Yes and No percentages here of 32% and 57% which the Daily Telegraph quoted. Again there are some arrows to indicate how these opinions might change as well.

779 in 1001 voters definitely voting would be a turnout of 78% of the survey. The table below shows the actual turnout by Scottish voters in elections and referenda since 1979*:

So 78% looks a bit optimistic judging by the last 15 years of turnouts. However, as the diagram indicates, it currently appears that for YES to secure more than 50% of the votes in September, one or more unlikely events would have to occur, for example a substantial failure to turnout by those currently intending to vote NO, or all the currently Undecideds turning out and voting YES, or substantial changes of mind by those who currently consider themselves committed to NO.

At which point, back to McTernan’s article.


*Data up to 1997 from Wikipedia, subsequently from Scottish Parliament.