John Curtice: Brown needs not just a victory but a Tory rout

The party conference season is finally over. Now Gordon Brown has to decide whether he is going to fire the starting gun for an election. And at first glance it seems inevitable that he will.

There certainly appears to be no danger that Labour could actually lose an early election. Virtually any lead at all over the Conservatives should be enough to secure the party a fourth overall majority. The average lead in the most recently conducted polls is no less than eight points.

Moreover, apart from one short spell at the end of August, Labour has enjoyed a comfortable lead over the Conservatives ever since Gordon Brown became Prime Minister at the end of June. The "Brown bounce" does not look like a temporary blip of the kind that disastrously persuaded Harold Wilson to call a sudden election in June 1970. Instead it looks solid and potentially durable.

However, in calling an early election Mr Brown is not simply hoping to win. He wants to win well. His aim will be to secure a "personal mandate" that would mean he was no longer simply be the "heir to Blair".

To be seen to have secured that mandate he has to do at least as well, if not better than Tony Blair himself did in May 2005. Then Labour secured an overall majority of 66. To reach that target Mr Brown needs not just to come ahead of the Conservatives, but well ahead.

Exactly how far ahead is inevitably a bit of guesswork given the vagaries of the first-past-the-post electoral system. But we do know the new parliamentary boundaries on which the election will be fought in England and Wales are to Labour's disadvantage. According to authoritative estimates provided by Plymouth University, Labour would have emerged with a majority of 48 in 2005 if the new boundaries had been in place in 2005, not 66.

So to some extent Gordon Brown has to advance just to stand still. To secure an overall majority of 66 seats he could need a lead of around four points, rather than just the three points Labour had in 2005, though a fall back in the Liberal Democrat vote could reduce that target a little.

Still, Labour's current poll average poll lead of eight points is clearly more than four. Indeed of the 30 or so polls conducted since the middle of July only half a dozen put Labour's lead at below that figure. Mr Brown would seem almost certain of achieving his objective.

Yet there are reasons for caution, although the widely discussed concern that Labour voters would be particularly likely to stay at home in the dark and gloom of an autumn election is not foremost amongst them. The last two elections to be held in the autumn – in 1964 and 1974 – were both won by Labour (if narrowly). And in any event if Labour voters in safe Labour seats stay at home – as they evidently already have been doing in past elections – it makes not a jot of difference to the outcome.

Instead Mr Brown has to worry about three dangers. First, the polls have consistently overestimated Labour's strength at recent elections.

True, their average error was small in 2005, but the final polls still anticipated a five-point lead rather than three. So perhaps the eight point lead might really only be a six-point one.

Second, in each of the 47 seats that Labour lost at the last election, new Conservative (and Liberal Democrat) MPs have been working away trying to impress local voters. The evidence of past elections suggests such activity can make a difference, helping a new MP develop personal vote. That vote may not be large, but in a key marginal one per cent can make all the difference.

So Mr Brown may find any swing to Labour proves to be rather less in the crucial marginals he will need to capture to keep ahead of his 66 target.

Finally, there is what might happen in the campaign. Mr Brown's bounce may look durable but no party leader can be sure his party will not slip a couple of points during the course of even a short election campaign.

Together these three dangers mean that Mr Brown cannot at present be certain of securing his "personal mandate". As a result he will doubtless be looking anxiously at this weekend's polls. If Mr Cameron's speech yesterday has helped to narrow that eight-point lead at all, then Mr Brown's famous caution might yet prevail. If, on the other hand, they show Labour with a double-digit lead, the temptation to call an election will probably prove irresistible.

The author is Professor of Politics, Strathclyde University