TNS-BMRB poll shows swing in favour of independence

Some very welcome news on the referendum front today - the latest TNS-BMRB poll has shown the No campaign's lead dropping by 3%, the equivalent of a 1.5% swing in favour of Yes.

Should Scotland be an independent country?

Yes 25% (-)
No 44% (-3)

Of course the headline story of the last TNS-BMRB poll was the dramatic and unexplained increase in the number of undecided voters, and that trend has continued in this poll, with the Don't Knows now comprising an extraordinary 31% of the electorate.  In every sense, then, the message that the Yes campaign should take away is that there is all to play for.

TNS-BMRB's methodology was called sharply into question after the last poll, due to their failure to weight the figures to take account of the fact that their respondents' recalled Holyrood vote from 2011 bore no resemblance to the actual outcome of that election.  I can't yet tell you if that problem has persisted, because I received an error message when I tried to download the full datasets from the TNS website.

A supplementary finding from the poll that leapt out at me is that a significantly higher proportion of No voters (40%, compared to 31% of Yes voters) feel that they don't yet have enough information.  That could suggest that the No vote is softer, and supports the evidence from previous polls that voters are more likely to turn to the Yes camp if they are better-informed.  And one thing is for sure - voters will indeed become better-informed as polling day approaches.

One other interesting nugget - a parallel poll of English and Welsh respondents was carried out, which discovered that slightly more people thought that the UK would be worse off without Scotland (26%) than thought it would be better off (23%).  Is the "subsidy junkie" mythology finally starting to wear a bit thin, even south of the border?

UPDATE : I should have read to the bottom of the page - TNS-BMRB have announced they are now coming into line with other pollsters by weighting according to past voting behaviour.  The wording is slightly ambiguous - I'm hoping they're following the lead of ICM and Panelbase by weighting solely according to recalled Holyrood vote, rather than the YouGov practice of weighting by recalled Westminster vote (which is much more likely to be subject to false recall).