How the BBC calculates local election results
The local elections in Scotland were fought with new ward boundaries in many council areas across Scotland. This led to an increase in the number of councillors being elected.
To allow a comparison with 2012 results, the BBC used analysis from Prof David Denver of Lancaster University - who has carried out the same process for every election change since 1980 - to work out what the election would have looked like if it had been fought with the new boundaries.
Here Prof Denver explains how he calculated the "notional figures" on which BBC based its results service.
When the boundaries of parliamentary constituencies change, it is clearly vital for analysts to have an idea of what the results of the previous election would have been had the new boundaries been in force and had people voted exactly as they did in their old constituencies.
There is a well-established way of doing this.
Since constituencies are made up of much smaller council wards, we can redistribute votes in old constituencies to newly-created ones making use of local election results. I have done this for Scottish constituencies after every boundary revision since 1980.
- Full council election results published
- Scotland local elections 2017 - BBC News
- What does council vote tell us about general election?
This year, I undertook the same analysis for Scottish local council election wards for which new boundaries came into force in 2017.
The problem is, of course, that there are no smaller divisions of wards for which we have voting figures which could be reassembled in to the new wards.
Of the 354 new wards created, 151 were unchanged as compared with 2012, 34 had very small changes and 68 had what I defined as "minor" changes - involving up to 10% of the electorate.
For these, the only differences between actual and notional results registered were if the number of councillors to be elected changed.
If the number declined by one then the last candidate to be elected in 2012 was deducted. If the number increased then the next candidate who would have been elected was added.
Published detailed breakdowns of the various count stages allow this to be done.
For wards which experienced a "major" change (10% to 30% of electorate involved) or could be described as "new", I scrutinised maps of the old and new wards to ascertain which parts of old wards made up the new.
I then transferred votes proportionately from the old to the new (and sometimes using local knowledge and past experience of local elections) to get an estimate of how things would have worked out in the latter.
Here is the full table of the results from 2012 and 2017, giving both notional and actual changes in councillors elected: