Simon Crean and opinion polls don't mix. One is the natural enemy of the other. That's a lasting impression from the political year 2002. But a rummage through the archives of the much-quoted Newspoll surveys makes it difficult to remember why.

There are two popular benchmarks. One is voting intentions, and on this, after preferences, the year was mixed. Post-Bali, it belonged to the government, though not by a large margin; and before October, the opposition led about half the time. The main feature was a curious calm, with both sides attracting low primary votes and neither opening a big two-party preferred lead.

The other measure is "preferred PM", and here John Howard romped home. Crean's leadership began in November 2001 with a Newspoll rating of 19 to Howard's 55, and last month it was 19 to 58.

So voters just can't imagine Crean in the top job. But does it matter? Newspoll data suggest not: preferred leadership is a weak predictor of success.

Take Bob Carr, who trounced NSW opposition leader John Brogden as preferred premier 61 to 16 last month. He trailed incumbent John Fahey all the way to his winning March 1995 election: 27 to 52 a month before the poll, and 32 to 45 the Thursday before.

Victorian premier Joan Kirner led Jeff Kennett 46 to 38 on the eve of her crushing 1992 loss. Kennett in turn never let Steve Bracks closer than 51 to 35. In South Australia last year, premier Rob Kerin went to an election preferred to Labor's Mike Rann 50 to 30, lost it, and promptly trailed Premier Rann 28 to 43. Yet voting intentions in these surveys proved respectably accurate.

Back in Canberra, Crean performs poorly compared with his predecessor. But Kim Beazley was unusual in regularly winning the preferred PM contest while in opposition.

Another much preferred pretender was John Hewson, who was ahead of Paul Keating in most surveys, including those of the losing 1993 campaign. Three years later, opposition leader Howard trailed Keating as preferred PM in every poll of the campaign that the coalition won by 40 seats. (Again, the voting intentions poll consistently showed a big Howard victory.)

So, a negative correlation between preferred leadership and electoral chances? That puts it too strongly; but in rating any opposition leader's chances, we should recall the hapless figures of fun that the above leaders were - until the day they triumphed. Respect and star quality came with the office, not before. Maybe contemporary Australian politics has no place for aspirants with big ideas or big personalities. And perhaps undecided voters, come polling day, feel a final endearment for that ill-defined, uninspiring but stoic trier. Crean could take heart from that.

Peter Brent is the editor of mumble.com.au, a website that looks at electoral behaviour.