It's official: Savanta ComRes confirm that the Scotland on Sunday "poll" should not be treated as a proper independence poll. Time for (another) correction and apology from Sarah Smith and the BBC.

Savanta ComRes have published the datasets from the now-notorious #Matchettgate "poll", published in the Scotland on Sunday and jubilantly recited by the legendarily "impartial" Sarah Smith on the Marr show.  (In fairness Sophy Ridge and Sky were even worse than the Beeb - they apparently put the numbers up on a big screen!)  There's a very interesting explanatory note provided on the ComRes website - 

"The voting intention was not weighted by likelihood to vote and should not be treated as a headline Savanta ComRes voting intention.  The voting intention was asked to give The Scotsman an indication as to current levels of Yes/No support." 

You can hardly get more direct and explicit than that.  This should not be treated as a proper independence poll and anyone who was initially duped - and that includes myself - should correct the record.  I've already done so, and I hope we can now look forward to the same from the BBC and Sky.  It's particularly important that the BBC issue a prominent correction, because they've been telling us for years that they have a policy of not reporting on voting intention polls, and have used that as an excuse for not keeping viewers and listeners updated on the recent run of Yes-majority polls.  If that policy was casually tossed out of the window because a reporter couldn't contain her excitement at No being back in the lead, and that poll then turns out to be bogus...well, something has self-evidently gone very badly wrong, and it needs to be put right.

Incidentally, this development also puts beyond dispute the observation I made yesterday.  Right now, the last twenty-six legitimate polls have all shown Yes on 50% or higher, after Don't Knows are stripped out.  That may not be the case within a few short hours, because apparently more polls are due tonight.  But as of this moment, the remarkable run of Yes success continues.

*  *  *

I also took a quick look at the Panelbase website to see if the datasets from the Sunday Times poll are out yet, but they're not.  However, what I found instead were datasets for a poll commissioned in January by an organisation called 'Scotland Unhyphenated'.  They may well have been there for several weeks.  This solves the mystery of the unionist propaganda poll with the insanely biased questions that I blogged about at the time.  The results are, to be honest, a mixed bag - with such leading questions it was almost inevitable that they'd at least get some 'hits'.  However, rather amusingly, there are two questions in particular that backfired on the client horrendously...

Should Scots living in the Rest of the UK be allowed a vote in a future Scottish independence referendum?

Yes 44%

No 56%

That'll be a useful one for pro-Yes politicians to have in their back pocket whenever Andrew Neil starts banging on about one of his favourite topics.  Perhaps surprisingly, there's no real difference between Yes voters and No voters on this - 56% of No voters think ex-pats shouldn't have a vote.

Do you consider the SNP contingent of MPs at Westminster to be...

Representative of the Scottish character: 52%

An embarrassment: 48%

Of course the only real embarrassment here was the question.  Whoever this 'Unhyphenated' dude is, he doesn't seem to do shades of grey.