Game-changing Ipsos-Mori poll suggests Scotland would vote for independence in second referendum

This is truly historic - the first time that Ipsos-Mori have ever shown a majority in favour of independence.

Should Scotland be an independent country?

Yes 53%
No 44%

It's the first post-referendum poll on independence from Ipsos-Mori, so it doesn't necessarily point to a sudden Yes surge out of nowhere.  To some extent, it could even be said to be in line with recent polls from other firms showing a slender No lead, because all pollsters are agreed that there has been an increase in Yes support since last September.  But Ipsos-Mori's distinctive methodology has seen that increase translate into an outright Yes lead, and one of sufficient size to qualify as "decisive" in BBC terms.

Ironically, Ipsos-Mori were one of the most No-friendly firms throughout the long referendum campaign, which meant they showed a much bigger swing to Yes when the 'Great Convergence' occurred at the death.  But it now seems that was simply a waystation on an unlikely transition, and that Ipsos-Mori are emerging as by far the most Yes-friendly firm (of those that have reported recently).

The biggest issue is probably that they don't weight by recalled referendum vote, or indeed by any recalled vote at all.  Instead, they put all their eggs in the basket of demographic weighting.  That carries the risk of there being too many pro-independence voters in the sample, because those may be the people keenest to answer political questions down the phone.  But that can't wholly explain such a favourable result for Yes - remember that Ipsos-Mori's final poll of last year's campaign (published on the morning of polling day itself) overstated the Yes vote by just 2%.  And the early post-September polls from YouGov, carried out before recalled referendum vote weighting was introduced, showed a Yes lead but a much narrower one than we see today.  The most positive interpretation is that today's poll is a very rare telephone poll, and perhaps a 'real world' approach to data collection is picking up a Yes majority that polls reliant on volunteer online panels are missing.

Either way, the very fact that it's a telephone poll makes it that bit more difficult for opponents of independence to dismiss.  Back in the days when Panelbase were the most Yes-friendly firm, Labour people used to make daft allegations that the Panelbase panel had been "infiltrated".  It'll be a bit difficult to make that claim about Ipsos-Mori, who seek out a completely fresh sample of respondents every time.

What we could really do with now is an independence poll from TNS - the only other regular 'real world' pollster on Scottish political affairs.  They use the face-to-face approach rather than telephone, and if they come close to replicating Ipsos-Mori's finding, it could be hugely significant.  (If they don't, of course, it increases the chances that the online pollsters are closest to the mark.)

Make no mistake, though - today is a game-changer.  Until now, it's been possible to make the claim that the polls are broadly showing a small No lead, and that the occasional exceptions have been caused by normal sampling variation.  Today's poll can't be explained away like that - the Yes lead is too big.  So the new reality is that Yes could be in the lead, or No could be in the lead - we can't say which, not even on the balance of probabilities.  That will obviously have an impact on the debate over the timing of a second referendum.

Apologies for taking so long to complete this post, and also for the fact that it's a bit shorter than usual.  As you've probably gathered by now, I'm away on holiday at the moment.  I blogged about independence polls in some unexpected locations last August, but I think today's was the most exotic so far!  If I ever get rich, I'll aim to blog about a 30% Yes lead from a beach in Goa...