Australian politics #

2019 federal election: before #

There is a federal parliamentary election to be held in Australia on May 18. Australian politics has developed a bit of a reputation over the past decade as being ruthless and cut-throat. This reputation is not undeserved: since 2009, when the first rumours of the weakness of Kevin Rudd’s grip on the leadership of the ruling Labor party began, we have had Rudd, his successor Julia Gillard, then Rudd himself again, then conservative Tony Abbott, who was himself deposed by relative centrist Malcolm Turnbull, and finally the current Prime Minister Scott Morrison, who is most commonly referred to as ScoMo.

This is the second election I am experiencing while living overseas, so I’m getting all of my news filtered through a few very specific and biased outlets, and I guess it’s important to say that off the bat. Apart from my Dad and a few of my friends, I don’t talk to any actual Australians about the election or politics. I follow a few dozen Australian Twitter influencers, mostly election nerds closely following poll movements and analyzing campaign gaffes. I read the Guardian’s coverage, flick through the Sydney Morning Herald’s front page every now and then, and read Crikey and the Saturday Paper. So it’s all coming to me in a very digested form, at great distance, and with a considerable left-wing bias: a very different experience to actually being present and part of the campaign, listening to office chat, arguing with distant relatives, and so on.

So there’s some interesting questions here: how accurate a picture am I getting of my home country and its political temperature before this election? I spent quite some time over the winter trying to develop a FiveThirtyEight-style predictive model for the election (called Emma Chisit) and eventually gave up because I didn’t feel like I was close enough or had detailed enough knowledge to really say something interesting. Is this true? Would I be interested in reading an article about politics written by someone who did not live in the country he was writing about and had not for many years?

I don’t really have answers to these questions, other than to provide the insight that elections seem to be of particular importance to expats, desperate as we are to maintain some sort of connection to where we are from. Elections are a taking of the national temperature, and there is a feeling that if I can understand why the country votes the way it does, and the issues it voted on, maybe I can keep my understanding of the country intact, continuous. It’s tenuous, but it’s something, and I definitely feel a little closer to Australia at the moment when I’m reading analysis of polls, or daily roundups of the campaign trails, or just back-and-forth thinkpieces at the frontlines of the never-ending culture wars.

For what it’s worth, I am struggling deeply with the insane ignorance of global leaders, especially in the United States, which has abdicated any sense of responsibility, but also in Australia, of the warming global climate. I can’t help but feel like the entire election is just a bunch of stupid white men with their fingers in their ears, arguing on the deck of the Titanic about how to split the dinner bill while the ship goes down. The fact that there are still people in positions of political power outright denying it is happening is in itself outrageous, but what about the people who accept it is happening but do nothing? How are they not spurred into action by the threat of bequeathing a scorched Earth to their children? And what about Australia’s awesome natural landscape and unique wildlife?

So this is the eye with which I’m reading all news of the elections Down Under: with a definite vested personal interest, but with a broad and nihilistic cynicism and sense of disbelief that this is reality in my distant home.

2019 federal election: after #

I’m reading a lot of people this last week putting the recent Australian federal election result, which saw the incumbent Liberal government, fresh from turfing their second leader in three years, returned in an extremely surprising upset win, in the same basket as Donald Trump’s 2016 presidential victory in the United States, or the result of the 2015 Brexit referendum in the UK.

It’s true that Australian polling firms failed spectacularly at their task of predicting the result. But the magnitude of this failure was much larger these other two, and I am not entirely sure the narrative of Australia falling victim to the same sort of populist politics that drove Trump and Brexit fits the story here, except perhaps in Queensland.

There are many more knowledgable voices than mine speaking about methodological issues in Australian opinion polling, and in the wider world as well (the post by Kevin Bonham below is a great summary of this latest election, and the topic was brought up by every talking head I saw on the ABC on election night). But it is clear that the way in which the world is connected has changed, and the way we communicate with each other has changed, but public opinion polling is still using methodologies build around in-person or telephone surveys.

Are all the assumptions which underlie this sort of information gathering still accurate? Pollsters use a method called reweighting to account for known biased subgroups in their sample. For example if only 400 women were sampled in a population of 1000, the aggregated result needs to be adjusted to account for the fact that women and men on average have different voting intentions (this difference is interestingly smaller in Australia and the UK than America) and the fact that the true population is (about) 50/50. But maybe there are groupings of people which are not properly being weighted for. What about people who don’t answer the phone? What percentage of the population are they, and how do they tend to vote?

I long had a goal of building a poll aggregator, in the style of Nate Silver, or the Poll Bludger in Australia, but have put this project on hold because I just don’t know what is going to happen with opinion polls in the near future. The problem is made even more difficult in Australia, as preferential voting adds another dimension to the inaccuracy. To accurately estimate the swing to the Coalition that occurred in Queensland last Saturday, pollsters would have to accurately estimate

  • the Labor primary vote
  • the Coalition primary vote
  • the United Australia Party (Clive Palmer’s reprehensible attempt to buy a seat in the Senate)
  • the percentage of United Australia Party first preference votes which eventually flow to the Coalition during the transfer of preferences.

Why and what they missed can and will be debated in the coming weeks and months. Comparisons will be made to previous votes, some more valid than others. But what is certain is that the soul-searching among pollsters after their failure will have far-reaching consequences in future elections in Australia. I just hope these lessons are learnt and takeaways applied in a positive way. The worst possible outcome, and one we’re already seeing with the decommissioning of some pollsters in the wake of the election, is that the concept of robust opinion polls is tarnished with the brush of unreliability, and the entire industry fades into obscurity. Because then who knows how politicians will take the national pulse between elections. Who knows how policy will be set without an accurate feedback loop. It’s a scary thought, and I really hope our pollsters can pull their collective fingers out before it becomes an existential problem.

  • Antony Green’s preview of the 2019 NSW election.
  • This summary from Kevin Bonham of the huge polling errors in the recent Australian federal election.