While opinion polls continue to show the Howard government well behind, hope lies in the "lose the vote but win the election" scenario.

It is indeed likely that the Labor opposition will need more than fifty percent of the two party preferred vote to form government. But how much will they need?

In 1998, the Coalition managed a comfy 12 seat majority with a touch under 49 percent. But that’s not all: Howard still had about a one percent safety margin. That is, if he had received only 48 percent he probably would have hung on.

It is generally the lot of federal oppositions to need more than half the vote. Andrew Peacock lost with 50.1 in 1990, but Labor Prime Minister Bob Hawke’s best seats-per-vote performance was actually three years earlier, at the 1987 election, when he won handsomely with 50.8 but would have survived with just 47.5 percent. 

It comes down to the benefits of incumbency, marginal seat campaigning - and a lot of luck.

The word today is that the government believes that with 48 percent after preferences it can hold on. That’s about what the pendulum says as well, and it sounds reasonable. But we don’t really know: they might need more than that, or they might need less. 

We also keep hearing that the marginal seats are much tighter than the national vote, which is another way of saying the same thing.

But there is one more factor that is largely overlooked. 

The last time the federal ALP received a winning vote was 51.4 percent in 1993. That was a long time ago. At the most recent election, in 2004, they got just 47.2, which represents a net national swing over the decade of a little over 4 percent. But within that swing, some seats have moved massively towards the government – mainly the mortgage belt, in particular Sydney’s outer fringe seats, won by Labor in 1993 but now Liberal-held, whose movements were all well into double digits. 

The flipside is that other, Coalition-held electorates have been under-represented in this shift. They have either moved to the Coalition by only small amounts, or actually shifted towards Labor. Wentworth and Bennelong are the most well-known, but others include Boothby in Adelaide, Ryan in Brisbane and the Treasurer’s seat of Higgins in Melbourne. Plus a few outside the cities.

Most look too safe to matter, but swings are never uniform, and if there’s a large one on Saturday, these seats may be worth watching.

Observers are fixated on electorates that Labor held in office from 1983 to 1996, but every change of government sees a swag of "surprise" wins that the new government has not held in living memory. 

And while 48 percent might just see the government hold on, anything less than that is likely to see the Labor vote bursting through in all sorts of unexpected places.

We should prepare for the unexpected.

Peter Brent is a member of the Democratic Audit of Australia and publisher of mumble.com.au.

Peter Brent is publisher of mumble.com.au