Readers may be interested in my take on the forthcoming by-election for Bradfield. So I begin with a straight forecast. I predict that the two-party preferred vote will split 59 per cent for the candidate chosen by the Liberal Party and 41 per cent for whoever stands for the Greens.

To explain such a prediction requires some contested psephology. Before I do that, however, I wish to dispose of a myth. It is that by-elections normally record good results for Opposition parties. In the good old days that used to be so. However, in those days by-elections were typically caused by the death of a member. These days, by contrast, they are almost always caused by the greed-driven resignation of the sitting member.

Since our first federal general election in March 1901 there have been 144 federal by-elections, of which 108 occurred in the period from 1901 to 1981, inclusive. Only 37 of those 108 were caused by the resignation of the member. By contrast, in the period from 1982 to 2008, inclusive, there have been 36 by-elections, of which 33 were caused by resignation, one by a natural death, one by a suicide death and one by the court of disputed returns voiding the result in a seat at the previous general election.

My assessment is that in modern circumstances voters have become sick and tired of greedy politicians resigning their seats. So they take it out on the party of the resigning member. For that reason it would be astonishing if there were not a swing against the Liberal Party in Bradfield.

Three state by-elections have so far been held this year. All three were caused by greed-driven resignations. All were lost by the party holding the seat. In January the Liberal Party lost Frome (SA) to a pro-Labor Independent. In May Labor lost Fremantle (WA) to the Greens and in August Labor lost Pembroke (Tasmania) to the Liberal Party. Okay, they were state by-elections and Bradfield is a federal seat. So let me look at the three federal by-elections for this parliamentary term. All three were caused by greed-driven resignations as, of course, is Bradfield.

As things turned out in 2008 the Nationals retained Gippsland and the Liberals Mayo. Mark Vaile, by contrast, threw away his safe seat of Lyne to a hostile Independent. As things turned out the Nationals performed well in Gippsland, as I predicted would occur. The reason is sometimes an over-riding national issue takes over. In the case of Gippsland that issue was climate change. For a seat like Gippsland the Nationals were very much on the right side of the dominating issue. It over-came what would otherwise have been the negative of the greed of the resigning member, Peter McGauran. It was a good result for Warren Truss. Both Gippsland and Mayo were bad results for the then Liberal leader, Brendan Nelson.

The Mayo by-election in September 2008 constitutes a good piece of analysis for Bradfield. Both are blue-ribbon Liberal seats but the ribbon of Bradfield is bluer than that of Mayo. Since Mayo was a contest between Liberal and Greens I have no doubt Bradfield will also be that. In mayo the Liberal candidate was Jamie Briggs and the candidate for the Greens was Lynton Vonow. The two-party preferred vote was 39,381 for Briggs (53.03%) and 34,879 for Vonow (46.97%).

I lack the space to demonstrate the correct statistical status of my next statement so I shall just assert it and wait for events to prove me right or wrong. I assert that a transposition of the Mayo votes to Bradfield yields the 59-41 prediction made above.

In Bradfield the November 2007 two-party preferred votes were 53,512 for Brendan Nelson (63.45%) and 30,819 (36.55%) for the Labor candidate, Victoria Brookman.

When the votes are cast and counted I shall make my assessment as to what it shows for the leadership of Malcolm Turnbull. So why should I not do it now? I say if the Liberal share exceeds 60% of the two-party preferred vote then that is a good result for Turnbull. If it falls short of 58% then it is a bad result for him.