Following the agreement between David Cameron and Alex Salmond, Scotland's First Minister will be able to say he has kept his promise to hold a referendum on Scottish independence in 2014. Given the low regard in which politicians generally are held, keeping such a major promise is no trivial thing.

However, from Salmond's point of view, it is about the only thing going for him. Indeed, were he to be given a truth drug, he might well curse the fact that the SNP won last year's Scottish elections outright, and thus found himself in a position to keep his promise. He would surely have been much happier remaining the leader of a minority government, unable to get his independence legislation through Holyrood. Then he could have railed against the Scottish satraps of the Britain - wide parties for silencing the voice of the Scottish people.

Instead, by winning an outright majority, he has shot his own fox. Rather than shed crocodile tears for his inability to call a referendum, he must now put the issue to the test. As a shrewd and intelligent man - indeed, one of the shrewdest and most intelligent in British politics - he must know that his mission is impossible, that in two years time his country will vote to remain part of the United Kingdom, and that far from being achieved, independence will be deferred for at least a generation.

All YouGov's evidence from the past four years is that independence is a minority passion north of the border. Even as the SNP was surging to victory last year, Scots told us by two-to-one that they wanted to remain within the UK. The SNP won because most Scots thought it had governed their country well, because they liked Salmond, and because they thought the Scottish Labour Party was useless - not because they wanted to sever links with London.

Our latest Britain-wide poll for the Sunday Times suggests that independence remains a very hard sell. The key issue is whether Scotland would be economically stronger on its own. The Scots say no, by two-to-one. To be sure, there are just under 200 Scots in this sample, so the margin for error is large; but our voting figures are in line with other recent Scottish polls (Labour is back in a comfortable lead when people are asked how they would vote in a British general election), so there is no evidence that this particular sample under-represents the kind of people who think independence would be good for Scotland's economy.

Nor does it seem likely that this picture will be radically defaced by the two things Salmond has reportedly achieved in his talks with Cameron.

First, he appears to have got his way on the wording of the referendum question. Salmond wants it to say: 'Do you agree that Scotland should be an independent country?' Critics have said this is a loaded question. They are right. We pollsters prefer more balanced questions. Instead of offering a Yes/No choice to a single proposition, we would normally offer two alternatives - in the case, independence or staying part of the UK - or, if testing a specific idea, ask people whether they support or oppose it.

However, I don't think the loaded nature of the question will make any difference to the outcome. In a poll, balance matters because we are generally springing on people questions to which they may not have given much thought. In these circumstances, wording can affect the results we obtain. But when people vote in a referendum, they do so having been bombarded with arguments for and against. Voting will be a deliberate act. People will have decided whether they are for or against independence. This means that the precise words on the ballot paper won't matter. Cameron is right to have conceded on the wording. In practice he has conceded nothing that will affect the outcome - and, if the Scots vote to stay in the UK, Salmond won't be able to complain they were asked the 'wrong' question.

Salmond's second 'success' has been to get a green light to lower the voting age for the referendum to 16. However, if he thinks this will make anything more than the tiniest difference, he is uncharacteristically wide of the mark. Even if all 16 and 17-year olds who are entitled to vote are added to the register, this will add 3-4% to the electorate. All the evidence is that the younger the elector the less likely s/he is to vote. And those who do vote do so in roughly the same shares as their parents.

This means that the most Salmond can realistically hope for is that reducing the voting age will increase the number of people who vote by 2%, and that the pro-independence vote among 16 and 17-year olds will be ten percentage points higher than that of the rest of Scotland's voters. The overall effect will be to increase the share of the Yes vote by 0.2 percentage points. And that's an optimistic estimate: changing the voting age might make no difference whatsoever, or even reduce the Yes share fractionally. Again, Cameron has done the unionist cause, no real harm by conceding the point.

In order to win the referendum, Salmond will have to defy history and deploy his great campaigning skills to truly remarkable effect. As I have argued before, when referendums are used to resolve national divisions, the status quo tends to prevail. Normally it requires a consensus for change for a referendum to produce a Yes majority. No such consensus currently exists in Scotland and none seems likely. And on the biggest single influence on attitudes - whether independence would strengthen Scotland's economy - most Scots are fearful of going it alone.

Salmond deserves credit for ensuring that, 314 years after the Act of Union, Scots will finally be asked whether they want to stay joined to the rest of the UK; but he is most unlikely to like the answer.

See the latest Sunday Times results and full details here

Follow Peter Kellner on Twitter: www.twitter.com/@YouGov