The past four daily YouGov polls for the Sun and Sunday Times have all put the Conservative lead at 6-7 per cent. That’s firmly in hung parliament territory: or is it?

If we plug the latest figures (Con 38 per cent, Labour 32, Lib Dem 17) into the swingometer produced by my YouGov colleague Anthony Wells on http://ukpollingreport.co.uk/blog/swing-calculator, then those vote shares translate into the following seats:

Conservative 292
Labour 289
Lib Dem 39
Others 30

The Tories would be 34 seats short of the seats they need to secure an outright majority; and it would take only a fractional further swing to Labour for Gordon Brown to emerge as leader of the largest party.

The trouble is, votes might not translate into seats quite like that. Like most such calculations, the one made above assumes uniform national swing. YouGov’s latest figures put the Tories five points higher than in 2005, Labour four points lower and the Lib Dems six points lower. We get our seat figures by applying those vote-changes to every constituency, and seeing who wins each one.

In reality, of course, the movements are not identical in each seat. This does not matter if the variation from uniform swing is fairly random. In that event, seats which “should” change hands but don’t will be offset by the seats that “shouldn’t” change hands but do: the overall seat numbers should be about right.

The problems really start if the variations from uniform swing are NOT random. If marginal constituencies behave differently from non-marginal constituencies, then the votes-to-seats calculation will be wrong.

Here’s my guess. (And it has to be a guess: often people in key marginals make a tactical decision on who to support in the last days or hours of the campaign, so pre-campaign polls in marginals provide only limited help.)
I expect the Conservatives to win 10-20 more Labour seats than would be expected on a uniform national swing: this is partly because I expect some unwinding of the anti-Tory tactical voting that has helped Labour these seats in the past three elections, and partly because of the weight of Ashcroft money in these seats.

However, I expect the Tories to gain around ten fewer seats from the Lib Dems than uniform swing calculations would suggest – in the past many Liberal MPs have often managed to defy adverse national trends, as they build up a personal following.

In other words, I reckon the six-point lead in the latest YouGov poll is more likely to put the Tories 13-33 seats ahead of Labour, which is more likely to result in a minority Cameron government than a mere three-seat lead.

And if the Tory lead widens, it means that they need to to lead the popular vote by 9-10 per cent to secure an overall majority, rather than the 11 point lead suggested by uniform swing calculations.