Extraordinary from the YouGov poll - the SNP would have led by 24% even WITHOUT the new turnout filter

A number of us wondered last night if the sharp increase in the SNP's lead was a sign that YouGov had introduced a turnout filter in their full-scale Scottish polls, just as they did earlier this week in their Britain-wide polls.  It turns out that they have indeed introduced a filter, but the SNP surge was not a sign of it - because the voting intention numbers for both the SNP and Labour would have been exactly the same without it.  I can't spot any other methodological change that has been made - there's no sign of the "weighting by voting intention in January/February" that's been brought in for Britain-wide polls.

What that means is that it's possible to make a reasonably meaningful comparison between the headline numbers in this poll and previous YouGov polls - albeit with one important caveat, namely that weighting by recalled referendum vote was brought in earlier this year.  In this poll, that's helping the SNP slightly, because people who recall voting Yes have been weighted up from 432 to 458, and people who recall voting No have been weighted down from 584 to 560.  Paradoxically, though, in the early post-referendum polls, there were too many Yes voters in the sample - so if weighting by recalled referendum vote had been applied consistently throughout, the reported increase in the SNP lead since October would almost certainly have been even bigger.

My namesake James has made a reasonable point on the previous thread about an error of logic that YouGov seem to be making - they've sharply upweighted respondents who were born outside the UK to count as a full 9% of the sample.  It seems highly unlikely that as many as 9% of those eligible to vote in Scotland in this election were born overseas, because the rules are not the same as the ones that applied for the referendum, in which all EU citizens resident here were permitted to vote.  To be eligible to vote in a general election, you need to be a UK, Commonwealth or Irish citizen (which excludes 24 of the 28 EU countries, and most importantly Poland).  If YouGov are indeed getting it wrong on that point, it may be leading to a very slight overestimation of the SNP lead, because the small subsample of voters from overseas has the SNP ahead of Labour by a bigger margin of 49% to 10%.  The Tories are in second place on 21%.

I must stress, though, that country of birth weighting is vitally important, because without it there would be far too many English-born people in the sample, which would of course lead to the SNP lead being underestimated.  Among Scottish-born respondents, the SNP are ahead by 50% to 25%, but among respondents born elsewhere in the UK, Labour are actually narrowly in the lead by 30% to 28% (and the Tories aren't far behind on 25%).

Has Nicola Sturgeon succeeded in eradicating the gender gap?  There isn't really one in this poll - the SNP lead by 50% to 24% among men, and by 49% to 25% among women.  And there's also terrible news for London journalists who love to write headlines about how young people are "too cool for separatism" - a mind-boggling 61% of respondents aged 18-24 are planning to vote SNP, which is a much bigger share than the party is scoring among any other age group.  But even among over-60s (by far the most problematical group for the Yes campaign), the SNP enjoy a commanding lead over Labour of 21%.

Possibly the funniest detail in the datasets is that Willie Rennie was regarded as by far the worst performer in the STV leaders' debate by even the tiny number of respondents who are planning to vote Liberal Democrat in May!  Among that group, 28% thought Nicola Sturgeon was best, compared to just 6% for Rennie.  Among the much bigger group of respondents who actually voted Lib Dem back in 2010, the gap is unsurprisingly much bigger - with Sturgeon favoured by 42%, and Rennie by 3%.

The stubborn refusal of voters in the real world to do the bidding of unionist journalists by recognising the "obvious truth" that Jim Murphy won the debate is also reflected in the net approval ratings for various leaders -

Nicola Sturgeon:  +48
Jim Murphy:  -18
David Cameron:  -25
Ed Miliband:  -46
Nick Clegg:  -54

Last but not least, there are voting intention figures for next year's Scottish Parliament election - and because that contest is still a fair distance away, no turnout filter has been applied.  In line with the Westminster numbers, the SNP's lead has increased on the constituency ballot, although that's only because Labour have slipped.  It's pretty much steady-as-she-goes on the regional list ballot, apart from the fact that the Greens have dropped back even further, and are now only barely ahead of the Lib Dems.

Scottish Parliament constituency ballot :

SNP 49% (n/c)
Labour 24% (-3)
Conservatives 16% (+2)
Liberal Democrats 5% (n/c)
Greens 3% (n/c)
UKIP 2% (n/c)

Scottish Parliament regional list ballot :

SNP 42% (-1)
Labour 24% (-1)
Conservatives 16% (+1)
Greens 6% (-2)
Liberal Democrats 5% (n/c)
SSP 3% (+1)
UKIP 2% (n/c)