The government will be hoping opinion poll history repeats. But which piece of history?

Below are graphs of Newspoll leading up to some recent federal elections. (The two party preferred estimates are mine, from Newspoll primary data.)

The first one should look familiar: from December last year, when Kevin Rudd took the Labor leadership, to today. It has had Labor’s lead fluctuating between huge and massive.

So which previous Newspoll graph can the government take heart from?

Scenario 1: The sweetest victory of all (1993)

Some are clutching at the March 1993 election, when the Keating government rose from the political dead.

It is true that Keating made significant inroads during the campaign. But as the graph below shows, the results for the year leading to the campaign were patchy, with Labor even opening up large leads late in December 1992.

That is, the Hewson-led Coalition did not enjoy a constant, large lead as Labor has this year. Little to take heart from there.

Scenario II: Tampa revisited (2001)

The 2001 result has also been suggested. But as the graph shows, back then Coalition support had been gradually rising since March 2001; it peaked after Tampa and actually went south during the campaign.

Some Beazley supporters have maintained that with another week they would have gotten over the line, and perhaps they have a point.

In any case, the opinion poll direction in the 2001 campaign is all wrong for the government.

Scenario III The Latham experiment (2004)

What about 2004? This is actually trickier. Newspoll’s preference strategy was not great, and overstated Latham’s two party support all year. Their last published two party preferred numbers before the campaign was called were 52 to 48 in Labor’s favour, and their final data, in the last week of the campaign, was 50 50. On election night the real numbers fell at 52.8 to 47.2 in Howard’s favour.

So the government closed the Newspoll gap during the campaign, but only by two points.

The Coalition can take heart from Newspoll’s under-stating of their support back then, but in 2007 the pollster’s preference allocation is more realistic.

In any event, Latham’s (inflated) leads were never anything like Rudd’s are today.

IV And the winner is … comfortable and relaxed (1996)

Sadly for the government, the graph that most closely resembles that of 2007 is the 1995-6 one. Apart from one rogue poll, the Coalition was consistently ahead by large amounts, and the calling of the campaign had little effect.

In fact, Rudd’s leads in 2007 are significantly bigger than Howard’s were in 1995-6.

Of course, every election is different, and it is possible that, because of technological and lifestyle changes, the pollsters are uniformly mis-measuring voting intentions in 2007. Those who want to see a close election should hope so.