Monthly Archives: March 2015

Did Malcolm Fraser briefly shift the divisions of Australian politics?

Today we lost Malcolm Fraser, Australia’s fourth longest serving prime minister (1975-83) and a man truly dedicated to public service. He may have also (briefly) changed the divisions of Australian politics in ways consistent with his later socially progressive activities.

Much has been written about his time in office, as well as his contribution to public life since he left parliament. One of the more common statements made upon his time as prime minister, and his activities since, has been that he cannot be easily pidgeon-holed in ideological terms.

Although he was pretty much the Cold-warrior personified during the 1960s and 1970s (including being a key supporter of Australia’s involvement in the Vietnam War), it has been noted he was a strong supporter of Australia accepting large numbers of Cambodian refugees in the late 1970s (more so than the centre-left Labor Party at the time) and a major supporter of a multicultural Australia. He also campaigned for human rights abroad; in particular championing a boycott of apartheid South Africa (and similar actions against white rule in Rhodesia (Zimbabwe)).

He continued his advocacy for multiculturalism and humanitarian policies once he left office (which earned him the enmity of many of his former partisans in his later years).

A good overview was provided by The Age this morning. Robert Gottleibsen has a different (but equally interesting) take in The Australian (although he is wrong about Fraser not cutting public spending – he did in a number of areas as would be expected of any Coalition Prime Minister – but I suspect he was closer to the mark with his assessment, that Fraser was always a progressive on some issues at least, than many others who claim Fraser moved to the left post-public life). [edit -Paula Matthewson actually covered this really well on The Drum]

Another way to look at the Fraser government is to see how voters reacted to his (arguably) economically conservative but socially moderate or liberal policies when in government.

Luckily, we have some data with which we can do this. As I discussed in this post, we can fit an Item Response Theory (IRT) model to data from the 12 Australian National Political Attitudes Survey (ANPAS) and Australia Election Study AES) surveys covering the 44 years from 1967 to 2010. Doing this provides us with a consistent estimate of voters underlying issue preferences on one or more dimensions over this period, allowing us to measure different kinds of political trends.

These surveys provide information on the policy preferences, voting behaviour and demographic backgrounds of 24,393 Australian voters. The IRT model is fit to 34 issue items using the mirt package (Chalmers 2012) in R (R Core Team 2013) and provides estimates for voters’ positions on two dimensions: one I label economic issues (mostly structured around attitudes towards trade unions, redistribution, taxation and government spending), and the other I call social issues (mostly relating to immigration, law and order and indigenous issues).

I then fit 12 separate logistic regression models to these data to predict the probability of a voter choosing to provide their first preference vote to the Coalition over the Labor Party (minor party and independent voters excluded) based on their economic and social issue preferences, after controlling for their age, birthplace, education and gender.

As can be seen in figure below, voters’ issue preferences – as measured by this IRT model – voters on the economic right are much more likely to vote for the Coalition over Labor (generally 80% or more doing so at each election compared to 20% or less for those on the left), whilst there has generally been little difference in voting behaviour between the social left and right (once we control for everything else, at least).

Predicted probabilities of voting for the Coalition over the Labor Party by issue preferences, 1967-2010. Estimated from the models described above fit to the ANPAS and AES data, each plot shows the probability a voter would vote for the Coalition over the Labor Party, ordered by their preferences from the left to right (with the categories representing voters, from left to right, 2 and 1 standard deviations left of centre, at the electorate mean, and 1 and 2 standard deviations right of centre, respectively). Demographic variables are held at their baseline categories (female, born in Australia, 18-24 years of age, who did not complete high school) and the other issue dimension at its mean value.

In particular, something really interesting appears to have been going on with the relationship between social issue preferences and major-party voting in the 1970s and 1980s. These results suggest that in the late 1970s, after we control for voters’ demographic backgrounds and their economic issue preferences, socially-left voters were more likely to vote for the Coalition than those on the right. This is possibly an impact of Fraser’s socially progressive views and policy positions (on refugees, SBS etc at any rate).

Besides being a true public servant, it appears Fraser may have (briefly) changed the divisions of Australian politics, making the political divide a little more complicated (but also perhaps compassionate).

He will be missed.