ICM poll confirms that a brilliant Alex Salmond stormed to a decisive win in tonight's crucial TV debate

The results of ICM's instant post-debate poll are as follows :

As far as you are concerned which one of the two leaders do you think won the debate?

Alex Salmond 71%
Alistair Darling 29%

There were red faces at both the BBC and the Guardian tonight, as the disconnect between the above numbers and the initial spin that both organisations attempted to place on the debate illustrated vividly just how out of touch they often are with the instincts and concerns of the electorate in Scotland.  Andrew Sparrow of the Guardian hilariously called it "a draw" (yeah, that Brazil v Germany match was pretty close as well, wasn't it, Andrew?), while Allan Little of the BBC bizarrely observed that this debate was "much closer" than the previous one.  Thankfully, good old Brian Taylor partly saved the day for the BBC by popping up afterwards to insert a dose of realism, and to declare that Alex Salmond would be the happier of the two men.

I probably shouldn't be too hard on Allan Little, who has consistently been the best and most even-handed BBC network reporter on the referendum.  But just to emphasise how absurd the "much closer" claim is, the ICM poll after the first debate gave Alistair Darling a mere 12% advantage, which wasn't all that far outside the standard margin of error (and was probably inside the real world margin of error, given the severe upweighting of some groups of respondents).  Tonight's ICM poll gives Alex Salmond the win by a whopping 42% margin.

Would it have been better for Yes if Salmond could have produced these kind of dream numbers in both debates?  Perhaps, but there's also a lot to be said for being on the right side of a "comeback kid" narrative, and if any of us had to choose which of the two debates we'd rather have seen result in a clear Salmond win, it would undoubtedly have been this one, with postal votes just about to be sent out.

If the poll results had been slightly closer, I have no doubt that the unionist media would have attempted to maintain the risible line of "it was a draw".  As it is, they'll be unable to completely ignore the elephant in the room, although I suspect their back-up argument will be "Salmond won, but it doesn't really matter".  To be fair, they can pray in aid the fact that ICM have before-and-after numbers for voting intentions, which on the face of it suggest that Salmond's win had only a minimal impact -

Should Scotland be an independent country? (Post-debate figures, with changes from pre-debate figures in brackets)

Yes 49.5% (+0.6)
No 50.5% (-0.6)

However, exactly the same thing happened in ICM's instant poll after the first debate, and yet the majority of the polls since then have shown movement (mostly towards Yes).  So the true impact may not become clear until voters have had a chance to reflect on the debate, talk about it with family and friends, and digest the media spin on it - which ironically will be largely driven by the ICM poll itself.  And it should also be remembered that ICM's respondents may have wanted to avoid looking fickle by admitting that the debate had directly changed their voting intention.

Now.  Ahem.  You'll probably have noticed that, regardless of the fact that they didn't alter much after the debate, the voting intention numbers listed above are rather wonderful for Yes.  Should we be excited?  Only up to a point, because ICM have once again been at pains to point out that this is not a normal representative poll, and was by definition restricted to people who were planning to (and who actually did) watch the debate. Nevertheless, the results were still weighted to reflect the demographic profile of the Scottish population, so what we've ended up with is a curious 'half-breed' poll, which doesn't attempt to directly reflect either the population as a whole or just the fraction of the population that watched the debate, but instead aims to artificially hover somewhere in between those two possibilities, which you'd think really ought to be mutually exclusive.  So although the voting intention numbers can't be regarded as fully reliable, they shouldn't be dismissed out of hand either.  At the very least, we can certainly use them to help measure how opinion has changed between the two debates, because ICM performed exactly the same exercise last time around.  On the pre-debate figures, these are the changes since the 5th of August -

Yes 48.9% (+2.0)
No 51.1% (-2.0)

And on the post-debate figures...

Yes 49.5% (+2.7)
No 50.5% (-2.7)

Although the swing to Yes is within the margin of error, it's entirely consistent with what we've been seeing in the last three 'proper' polls, from ICM, YouGov and Panelbase respectively.

And so much for Salmond's much-vaunted "woman problem", by the way.  ICM found that female watchers of the debate were significantly more likely than men to say that Salmond had a more appealing personality than Darling, and had the better arguments.  Having said that, both genders agreed that Salmond won on those counts.