You can't call the election on what's said in June

BEWARE the polls of June. Polling in June is a poor guide to who will win an election due at the end of the year.

In three of our past four election years, the party leading in the polls in June ended up losing the election later that year. And in every case, the victim was the (Labor) opposition.

In those four years, between our Age/Nielsen poll in June and the outcome on election day, on average the Howard government picked up 5.5 percentage points in support in two-party terms.

That's not the fault of our poll which is the only poll that has correctly called the winner of all four elections. (Remember the front page headlines in The Australian on election days? "Late surge to Beazley" it told us in 1998. In 2004 was: "Latham within striking distance".)

If the Rudd government matched the Howard government's average comeback between June and election day, it would wipe out the entire swing to the Coalition in yesterday's poll, and we would end up exactly where we started.

Whether it will or not is another matter. The government's support has been in free fall since Tony Abbott became Opposition Leader. But precisely because Abbott has been so much the Opposition Leader rather than an alternative prime minister, it's a long way to polling day.

It's too early to call the outcome and history backs the bookies' call in making Kevin Rudd the favourite.

Remember 1998? John Howard had gone back on his "never, ever" pledge to commit to a GST. We didn't like it. And there was an election looming.

In June 1998, Labor under Kim Beazley had a commanding 56-44 lead in our poll (and similar leads in other polls).

The Howard government looked like a one-term government with an unsaleable tax. Yet it stuck to its guns, suffered big swings, but it won the marginal seats, and a second term.

In 2001, the GST's debut had led to a spending strike that sent unemployment rising and confidence plunging. The Howard government looked in even worse trouble than in 2001. In the post-budget poll, Labor led 57-43. Beazley looked home.

Then the economy began picking up. Howard told the patrol boats not to let refugee boats land, but take them to Pacific islands. And al-Qaeda attacked the Twin Towers. By September 12, the election was all over.

Then came 2004, and Mark Latham.

We liked him at first, and Labor soared in the polls. It led 56-44 in May, 52-48 in June, the same in July and 53-47 as late as August. But armed with focus group research, John Howard knew something we didn't. He brought on the campaign and boldly declared the issue was trust. He was right: we didn't trust Latham. The government romped home.

What made 2007 different was Kevin Rudd. As usual, Labor led 57-43 at this stage. But Rudd effectively narrowed the issues to WorkChoices and climate change and gave the Coalition nothing to attack. Even so, that big lead in June shrank to 52.7-47.3 by election day.

None of this means that Rudd will be able to turn it around this time. But it means the election is open, and it's in front of us, not behind us.