A consensus is forming that the nationalists will lose the referendum on whether Scotland should leave the United Kingdom, scheduled for September 2014. This seems to be based in part on recent opinion polls indicating support for Union. But those polls can be discounted.
In 2004 there was a referendum in the north-east of England on whether to establish a regional assembly. For two years every measure of opinion had indicated a solid majority in favour of an assembly. The last poll published ahead of the effective start of the campaign in September 2004 predicted the Yes side would win 64 per cent of the vote. Yet on polling day in November, the Yes campaign managed only a humiliating 22 per cent.
In 2011 there was a referendum across the UK on whether to replace the first-past-the-post method of electing MPs with the alternative vote system (AV). Opinion was erratic (particularly don't knows) but most surveys during 2010 gave the Yes side a lead of around 10 per cent. Polls published just before the formal start of the campaign in February 2011 indicated that the Yes side would get at least 60 per cent. Yet on polling day, May 5, the Yes campaign managed only 32 per cent.
Memory can be selective. Nobody in the commentariat would now advocate regional assemblies for England or the AV system for parliamentary elections. They are obviously silly ideas because of the majorities by which they were rejected. Yet before both referendums, the sensible consensus was that the measures would pass.
It is often claimed that there is an innate advantage for the no-change side in a referendum because voters who are not already committed will choose the status quo. This theory has been heard a lot since the AV referendum. It's a nice idea, but it isn't true with devolution.