Sturgeon mustn't become a prisoner of the pollsters

I didn't see Nicola Sturgeon on the Marr show this morning, because I've been recovering from a truly epic journey to Newcastle and back to see Scotland's decisive pool match at the Rugby World Cup (more on that later).  But I gather she was more explicit than ever before that a substantial and sustained shift in public opinion could be a trigger for a second independence referendum.

I suppose in a sense this is to be welcomed, because for the first time it makes clear that the fabled "material change of circumstances" is an event that the Scottish people can have some control over themselves, and won't necessarily have to be something that the London government does or doesn't do.  But what troubles me is that the only measure we have of public opinion is polls.  That means apparent shifts in public opinion can be artificially generated by changes in polling methodology, but it also means that real shifts can be concealed by changes in methodology.  That's already happened to some extent - the real average increase in the Yes vote since last September is markedly higher than most polls have reported in their headline numbers.  The reason is that four of the six active pollsters have introduced weighting by recalled referendum vote, which in most cases hurts Yes.

We have, therefore, already seen a substantial and sustained shift in public opinion.  Perhaps it hasn't yet been substantial enough.  But the fact that we tend to underplay what has already been achieved should serve as a warning of the dangers of appointing the likes of Peter Kellner as our adjudicators.  It's not inconceivable, for example, that the next Yes surge could be hidden from view by a new incarnation of the notoriously convoluted (and secretive) 'Kellner Correction'.