Given that John Howard was, in 2007, the first prime minister to contest a marginal seat it could be argued that his defeat in this election was no particular humiliation for him. Indeed I used to argue that proposition myself – until I did some statistical tests on the Bennelong result.

To do such tests I went to the AEC classification of seats and found that they classify the following Sydney seats as inner metropolitan – Banks, Barton, Bennelong, Blaxland, Bradfield, Cook, Grayndler, Kingsford Smith, Lowe, North Sydney, Parramatta, Reid, Sydney, Warringah, Watson and Wentworth.

With the exception of Wentworth (where there was a swing to Liberal of 1.2 per cent) every one of these seats recorded a swing to Labor. The average for the 16 seats was a pro-Labor swing of 4.6 per cent. In Bennelong it was 5.8 per cent.

However, Howard’s performance is actually worse than that – for this reason. A number of these seats also recorded swings to Labor in 2004. Consequently all of the following seats recorded swings to Labor both in 2004 and again in 2007, Barton, Bennelong, Bradfield, Cook, Grayndler, Kingsford Smith, North Sydney, Parramatta, Sydney and Warringah.

The cumulative swings over the two elections is 10 per cent in Parramatta, 9.2 per cent in Bennelong, 8 per cent in North Sydney, 6.5 per cent in Bradfield, 6.4 per cent in each of Barton and Cook, 5.5 per cent in Grayndler, 4.9 per cent in Kingsford Smith and 4 per cent in each of Sydney and Warringah.

As Crikey readers will understand I am currently analysing all the figures thoroughly. The more I study them the more I realise that the Howard Government was not as badly defeated as election night impressions created. The Rudd Labor Government will probably have a majority of only 18 seats in the House of Representatives. By contrast Howard himself was more badly defeated in his own seat than people seemed to realise on election night. His defeat in Bennelong was thoroughly ignominious.