This is a tale of two Australian states, each holding two recent general elections. The first is New South Wales where the election dates were March 24, 2007 and March 26, 2011. The second is Queensland where the election dates were March 21, 2009 and March 24, 2012. The recitation of these dates would lead one to think that NSW has a fixed four-year term and Queensland a fixed three-year term. That is correct for NSW but wrong for Queensland, where Anna Bligh has chosen both election dates. All four elections were held on a Saturday.

In the earlier election, the result in NSW was 52 seats for Labor, 22 for the Liberals, 13 for the Nationals and six independents in a Legislative Assembly of 93 members. Labor scored 52.3% of the two-party preferred vote and the Coalition 47.7%. In the case of Queensland, the result was 51 seats for Labor, 34 for the Liberal National Party and four independents in a Legislative Assembly of 89 seats. Due to different counting practices in Queensland, aggregate two-party preferred vote figures are not known but I estimate the 2009 vote to have been 51% for Labor and 49% for the LNP.

I make election predictions. For me in each of these cases the key exercise was to answer this question: “When the Labor government of Kristina Keneally/Anna Bligh is defeated, with how many seats will Labor be left?” For NSW the answer I gave was 20. For Queensland the answer I am giving is 18 seats. I make no secret as to how I do this and William Bowe in Crikey explained it as well as I ever could:

“The conventional means of plotting the likely outcome of an election, as pioneered by Australia’s psephological godfather Malcolm Mackerras, is to observe the trend of the more reliable opinion pollsters and count the number of seats the insurgent party would net in the event of a uniform swing, under the more-or-less safe assumption that the inevitable variations will cancel each other out.”

It is nice to know that the way in which I have always done these things is now considered to be the conventional way.

When I predicted that Labor would be left with 20 seats in NSW, I did not name the seats, for the reason described by Bowe. However, I did predict (wrongly) that Monaro would be one of them. Anyway, the prediction of 20 seats did turn out to be spot on. Likewise, when I predict that Labor will be left with 18 seats in Queensland I am not naming them beyond saying that Ashgrove will NOT be one of the 18.

The NSW statistics are worth noting in some detail. Labor’s share of the two-party preferred vote fell from 52.3% in 2007 to 35.8% in 2011. The Liberal-National share rose from 47.7% to 64.2%. So the two-party preferred vote swing was 16.5%. Writing two months after the election, Premier Barry O’Farrell asserted: “On March 26, the NSW Liberals and Nationals secured the biggest electoral swing in Australian history.” Not having subjected all electoral data in Australian history to the analysis O’Farrell (presumably) has done, I cannot say that he is right. However, the claim sounds right to me.

What I can say is that I have drawn up a table of the electoral support recorded for all defeated Australian governments over the past half-century. I can, therefore, assert quite firmly from my own research that the Keneally government recorded the lowest level of support for any government over the past half-century. The second lowest was the 39% two-party preferred vote recorded for Labor in South Australia in December 1993.

I come back to Queensland. I am predicting that the two-party preferred vote swing will be 9%, producing a division of the vote of 58% for the LNP and 42% for Labor. Checking that against my pendulum I predict that 18 Labor members will be returned on March 24. When that happens, expect to hear some commentators say that “Labor’s defeat in Queensland was as bad as its defeat in NSW”. In terms of seat numbers that would be correct. A party defending 51 seats has performed as badly if it wins only 18 as a party defending 52 seats which retains 20.

However, in terms of votes Labor’s Queensland defeat would leave it with 6% more than in NSW, 42% compared with 36. In my table Anna Bligh’s government would be the third least supported over the past half-century, respectable compared with Keneally. Furthermore it would not even be the worst result for Queensland Labor. In December 1974, Labor retained only 11 seats whereas in March 2012 it retained 18. The point is that the geography of Queensland differs from that of NSW. The latter state has more genuinely safe Labor seats. Consequently in its worst result for more than 100 years Labor was still able to retain 20 seats.

*Malcolm Mackerras is visiting fellow in the Public Policy Institute, Australian Catholic University, Canberra Campus