Political scientist and Macquarie University Professorial Fellow Murray Goot has a very interesting article in the Sydney Morning Herald today:
In the past 10 elections, the gap in the level of electoral support for Labor and the Coalition reported by the polls at the start of a campaign, more often than not, has differed significantly from the gap recorded at the election.
The last election was no exception, but there were three – 1996 with Labor in office and 1998 and 2004 with the Coalition – where the gap in early polls proved a good guide to the outcome.
Where campaigns produced significant changes in party support, the pattern was always the same: the lead either narrowed or was reversed. There were no cases of an initial advantage increasing.
He finds no evidence of either an incumbency or an underdog effect.
If the pattern he discern holds for this election, it makes a fair bit of difference whether Labor really is starting at 50-50 or 55-45.