Songful joy for Sturgeon as SNP reach 62% in melodious TNS poll

This month's TNS poll of Scottish Parliament voting intentions is out today, and there's no sign of the SNP's honeymoon ending yet.  If anything, the marital bliss is just growing stronger.

Constituency vote :

SNP 62% (+2)
Labour 20% (n/c)
Conservatives 12% (-2)
Liberal Democrats 3% (-2)

Regional list vote :

SNP 54% (+3)
Labour 20% (-1)
Conservatives 12% (-1)
Greens 8% (+1)
Liberal Democrats 4% (-1)

This is the third TNS poll since the Westminster general election in May, and it's comfortably the best showing for the SNP so far - on the constituency vote they were on 60% in both of the previous polls, and on the list vote they were on 50% and 51% respectively.  The increase could just be an illusion caused by sampling variation, but if so that would require us to revise upwards our 'central estimate' of the party's support over recent months.  Alternatively, the SNP may genuinely have become even more popular since last month's poll, although I can't think of any obvious reason why that would have happened.  It would be easy to blame it on Labour's antics (the welfare abstention and so forth) but in fact Labour's support hasn't really dropped any further - they've slipped 1% on the list vote, but that still leaves them 1% higher than where they started two months ago.

Weirdly, the record-breaking SNP numbers have come about in spite of TNS having to massively weight down respondents who say they voted SNP in the general election - they've been reduced in number from 456 to 366, which in turn has led to respondents who plan to vote SNP next year being weighted down from 434 to 366.  Among the raw unweighted sample, the voting intention numbers were : SNP 67%, Labour 17%, Conservatives 10%, Liberal Democrats 3%.  Of course the adjustment made is entirely appropriate, but it's disconcerting that it had to be done, because prior to the general election TNS typically had to do the opposite - ie. downweight people who recalled voting Labour in the past.  Are Labour voters suddenly keeping their heads down when pollsters come chapping on their door?  Or are some of them following the prevailing fashion and falsely claiming to have voted SNP?

We should count ourselves lucky that TNS are persevering with their monthly polls.  Since the humbling of the polling firms in May, voting intention polls have become unbelievably scarce, to such an extent that the TNS series alone could almost be enough to leave us with as many Scottish voting intention polls as Britain-wide ones.  (That's an exaggeration, but not a huge one.)  The snag is, though, that TNS and Ipsos-Mori seem to have established themselves as the most SNP-friendly firms in recent months - an ironic turn of events, given that they were both among the most No-friendly firms during the referendum campaign.  So if we become too reliant on TNS to track the Holyrood race, we could easily end up with an over-inflated sense of how well the SNP are doing.  But with a bit of luck we might at least get semi-regular Survation and Panelbase polls to provide a little balance and perspective.