The Scottish Independence Referendum becomes a clash of polling methodologies
The Scottish Independence Referendum: polls from the last two months (Wikipedia) http://t.co/EdVHyAIdls pic.twitter.com/7po7alHTbw
— Mike Smithson (@MSmithsonPB) July 2, 2014
YouGov’s Peter Kellner fires the opening salvos
There was a time when you asked YouGov’s Peter Kellner why his figures were very different from other firms he would respond by saying that he never commented about how other firms operated. No more.
Last night Peter launched a sharp attack in a long blog post on some other pollsters, particularly Survation, over how they are carrying out their IndyRef surveys.
As can be be seen the firms polling Scottish opinion ahead of the September 18 vote split into two camps: those that have the outcome relatively close and those that have No with a substantial and apparently insurmountable lead.
Survation, Panelbase and ICM are in the former while YouGov,TNS-BMRB and Ipsos-MORI are in the latter.
The key point is how you politically weight the sample. Peter’s argument is that going back to the 2011 Holyrood elections enhances the SNP’s position. In that election, it will be recalled, Alex Salmond’s party pulled off a stunning victory because the then Scottish LAB leader did not come over as a convincing alternative first minister.
- The essence of Peter’s post is that many of the Labour voters who switched to the SNP in that election did so because of the choice of first minister not that they were committed to the nationalist cause.
To deal with this YouGov now has a separate weighting category for this specific segment thus deflating to some extent the SNP effect.
Will he be right? If one of the pollsters under attack from Peter wishes to respond then I would be delighted to publish that on PB.
In the meantime it won’t just be the future of the union that’s decided in September but the reputations of leading political pollsters could be on the line.