How Australians fooled the experts (again)

This was originally published on 24th May, 2019, and republished here on 4th January, 2022.


Another year, another election not going the way it was ‘meant’ to go. It’s all getting a little tiresome at this point, having all these votes where we’re all meant to be shocked by the result because the experts, the pollsters, the politicians, the media have assumed that the result would go one way, and it’s turned out in the opposite direction.


This time, it’s Australia’s turn. So, how did the Coalition government get its third term in office, under its third leader? Let’s find out.


(Image: Matt King/Getty Images)

Predicting elections is a bit of a fool’s game, because it requires analysing a mix of data and feelings. An election basically just one big poll, so any polling done beforehand can only sample a select group among the wider population, which may be skewed one way or the other for any number of possible reasons — and that’s the data stuff, the part that’s meant to be easier to grasp, yet pundits still need to try and work out why the polls look the way they do because the data doesn’t necessarily tell them the ‘why?’ on its own.


Feelings then come into play with everything else — the campaign trail, what you hear on the street or at work, your own general sense for what’s happening and what issues people care about. To get a prediction right, you need to be able to remove your own preferences from the equation and think about the bigger picture, examining how a voter in this area or that region may be influenced by particular policies or personalities. What seems to have happened at the Australian federal election is that the polls were giving faulty data, and the experts were applying faulty readings to it as a result.


Polling is, to be fair, nowhere as easy as it once was, with large sections of the population no longer having a landline phone. However, this is still an embarrassment for Australian polling organisations, who have been very proud of their record even in recent years. Australian polling has been highly regarded for its accuracy, and given the length of time that Labor has been leading the polls, it seemed like a sure thing that they would win — right?


Voters don’t really care about any poll but the election itself, but politicians and those who watch them care very much about them all, because it allegedly gives them an indication of where things are headed. This group of poll-watchers also includes the pollsters themselves, who have a habit of herding together to ensure they’re not the outlying poll that gets things badly wrong. The problem with the herding, of course, is that it can end up with all of the pollsters getting it wrong, which is exactly what happened.


For an understanding of how the polls got it wrong, this article by Kevin Bonham, Tasmanian election watcher extraordinaire, is worth a read.


(Image: Canley — CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=72051733)

All this, though, feels like a way for the other ‘experts’ to save face by pointing the finger at pollsters. Both the parties and the media have spent years looking at the polls to tell them everything they needed to know. Parties have their own internal polls when election campaigns come, but they also look at the public polls. Mixed messages are coming out about Labor’s polling — on the one hand, it was nationally wrong and on par with the public polling, on the other hand it indicated that Queensland was going to cause a lot of trouble for them. The Coalition has informed us that their internal polling consistently had them ahead, especially in the seats that mattered most, but even then they got the jitters on election day because of the most recent public polling.


It’s dangerously solipsistic, which sounds about right given the past decade of Australian politics. Polls have been responsible for every single leadership change since 2008 in both major parties. The polls informed Labor they were headed for catastrophe in 2010, so Julia Gillard replaced Kevin Rudd and led them into minority government. The polls had previously told the Liberals twice that they were likewise doing poorly, so Malcolm Turnbull replaced Brendan Nelson, and Tony Abbott later replaced Turnbull.


The polls also said that Labor was going to lose in 2013, so Rudd replaced Gillard (but still lost). The polls then said that the Coalition were heading for defeat under Tony Abbott, so Malcolm Turnbull replaced him (giving us the now-infamous line about losing 30 Newspolls in a row) and led the Coalition to a majority government by the slimmest of margins. After Labor had led the polls for two years, Turnbull was replaced by Morrison.


(Image: cpa.org.au)

Australia has been ruled by polling for eleven years. Politicians, surrounded by a hungry media pack, have spent more time navel-gazing and trying to figure out the psychology of the electorate than they have in presenting policy and sticking to it, and the media has been more than happy to assist them in doing this, as a crisis makes for better news than stability.


The media really deserves the biggest portion of the blame for this, as they are the means through which we, as an electorate, are told of who we should expect to win, what they want to do if elected, why they’re going to win, and so on, and they also have been for many years the means through which politicians communicate with the electorate and hear what the electorate is saying broadly (though these are changing with the advent of social media).


How many in the media were willing to say that the Coalition had a good chance of winning? How many were aware of the potential drubbing Labor would receive in Queensland, and saw it is as a major issue worthy of attention? Nine/Fairfax has just announced that The Sydney Morning Herald and The Age will not be using polling companies for some months, albeit with the caveat of “this isn’t permanent,” but the most interesting part of the article was the realisation that the most accurate reporting was done on the ground. Says Tory Maguire:

“During this campaign our best reporting on the mood of the electorate was done by journalists out on the road (and I don’t mean the carefully controlled campaign busses organised by the political parties). Michael Koziol’s report from Queensland midway through the race, which suggested Labor was in deep trouble north of the Tweed, was informed by old-fashioned boot-leather journalism. Nick Bonyhady detected no dramatic shift to Labor as he travelled from Hobart to Cairns on public transport.”

This isn’t to say that the media weren’t focussing on issues, either. In fact, there was one issue that they generally identified as the major issue of the campaign and were relentless in describing as a game-changer: the environment. Polls came in saying that voters were more concerned than ever about climate change and environmental issues, the media looked at that and the polls telling them that Labor was consistently ahead, put two and two together, and voila! The ALP was destined for government on the backs of the millions of Australians wanting more action on climate change and the environment.


(Image: Australian Labor Party)

Of course, this wasn’t what eventuated. Matt McDonald, writing for the ABC, summed up part of the problem very well: voters feared climate policy more than climate change. However, I would go a step further than he does and say that not only do a lot of voters go with the economic choice if made to weigh up between economics and environment, they actually are unlikely to make a different choice to that unless they see something concrete in their own eyes which makes climate change something that they can personally affect and which must immediately be acted upon. Even saying ‘we’re in a climate emergency!’ is unlikely to sway those who aren’t already swayed.


This is why it hasn’t gained traction as a national issue of first importance, despite being on the agenda since 2007. If people don’t sense that something will truly be different because of their vote, their vote isn’t going to be different from last time. Further, I would suggest that the reason the environment was popping up in polling as a more important issue than in the past was because of the lack of other issues to fill that national space — and the ALP may have caused this to happen, accidentally making a rod for their own back.


Labor was seemingly quite happy about the massive policy platform they put down, covering pretty much every issue that had come up over the past six years of Coalition government (ie. Bill Shorten’s time as Leader of the Opposition). Coal mining, asylum seekers, the Great Barrier Reef, negative gearing, franking credits, taxation reform, and much more besides — a big-spending platform, emphasising wealth redistribution within a socially liberal framework, as is befitting the modern day Labor Party (and their counterparts across the western world).


The three living Labor Prime Ministers. (Image: Dominic Lorrimer)

Unfortunately for them, all they ended up doing with it was creating a massive target that they couldn’t defend from every side. There’s a good chance that every voter in Australia might agree with something in the Labor platform, but then there might be something else which they don’t agree with or are at least concerned about, without Labor really adequately defending or explaining it because they’ve got a tonne of other policies they’re defending or promoting or even announcing. So that voter, unsure of what exactly Labor would do to their sole investment property by getting rid of negative gearing, or to their local coal mine, or to do their independent private school, decides that maybe it’s not worth taking the risk.


Furthermore, creating a large policy platform without a central plank, as the ALP did at this election, is really no better than creating a central policy plank that you can’t adequately defend (as John Hewson’s Liberals did with the GST in 1993). Voters generally don’t have the time, the energy, the knowledge or the interest to absorb every single policy and promise that political parties make. They need to have these platforms synthesised and summarised into something more easily digestible. Labor didn’t really provide that because they didn’t have a central theme to their platform, due to it being that aforementioned collection of issues that have cropped up over the past two terms of Coalition government.


This meant that Labor left it up to voters to make up their own mind about what their policy platform was all about, ably assisted by a Coalition government advertising like it was in opposition, and a Clive Palmer happy to acting like a second government party, spending a fraction of the vast sums of dosh he won from the Chinese government in court. Labor’s decision was not a wise move in any situation, but doubly so given the way that modern politics approaches personality, because the most simple way for voters for summarise a party is to look at who’s leading it — and the parties, the media, and the pollsters all encourage them to do it.


Mr Personality himself, Bob Hawke (Image: ABC/Four Corners)

It was quite a coincidence that Bob Hawke, the man who kick-started the personality obsession in Australian politics, died only days before the election. Ever since his time as Prime Minister, the position has become more and more presidential in image and nature, eating up larger amounts of attention and requiring an army of staffers. Voters will frequently pick their party not on the basis of policy platform, or even of who their local candidates are, but instead of who is leading the major parties. The media will reinforce this by putting the leaders of each party front and centre on anything and everything involving their parties.


I actually think it’s quite likely that part of the reason Labor ran with a big policy platform was to prevent a personality contest at this election, because they knew they wouldn’t be able to compete. Morrison is relatively popular and has the savvy of an advertising guru (which he was before he entered politics). Shorten has never been able to cut through to the electorate, although he is perhaps a more sympathetic figure now than he was six years ago. If we look back at the polls, which consistently put Labor ahead, those same polls also had Morrison as the preferred Prime Minister every week except his first as PM.


Before him, Turnbull had been preferred PM every single week, even as his party’s polls were heading south. Preferred PM has been a fairly good indicator for a long time now of where undecided voters may end up going, and where the general mood of the election lies, and in 2019, both major parties knew that Shorten as leader was a liability. Ironically, the biggest thing the ALP did to combat that ended up forcing voters to weigh the leaders against each other by not providing a clear summary of what they stood for.


A good example of what that looks like would be the 2013 Coalition campaign: “Stop the boats, axe the tax, pay back the debt, end the waste” — simple and effective. Having said that, the man most associated with that campaign also lost his seat at this election. Labor was hoping would happen, although Tony Abbott didn’t lose his seat to the ALP, but instead to independent Zali Steggall (formerly of Olympic fame) following a huge campaign from a raft of organisations, led by GetUp. At the time of writing, Abbott received a touch under 40% of first preferences in a seat that has been exceptionally safe for the Liberals. A similar campaign to unseat Peter Dutton in favour of a Labor candidate failed miserably, with Dutton getting a swing towards him.


(Image: Lukas Coch/AAP)

Abbott and Dutton are viewed as peas-in-a-pod, but one suffered an enormous swing against him while the other gained ground and kept his seat. Why? For local reasons. Without an effective national policy issue, voters who weren’t reduced to simply voting on personality might instead choose to vote on an issue that matters to their local area. This is increasingly true in federal politics, where there was once never a crossbench in the lower house, due in large part to the centralisation of the Australian landscape, an issue I have touched on before.


In essence, the effect of centralisation is that more issues become matters of federal government instead of state government — industrial relations, education and health, taxation and so on — and, because voters realise this, they begin to agitate for the federal government to deal with local issues where they once would have looked to the state government instead, and so they are much more willing to elect MPs from outside the major parties. Couple this with an increasing disillusionment with the major parties in general, and you have a recipe for Steggallomania.


This also means, though, that even if voters aren’t switching to minor parties and independents, they are more likely to vote according to state or regional issues than they used to be, for the same reason (ie. the states can’t deal with the issues anymore because of centralisation). As a result, you end up with a localised and regionalised election, that just happens to take place on a national level. The ALP nearly overtook the Coalition on first preferences alone in Victoria, and yet in Queensland their primary vote is at 27%, some 14% behind the LNP. Zali Steggall is, as of writing, on 57% in an inner-city Sydney seat, but former independent MP Rob Oakeshott lost by a bigger margin than last time in rural NSW.


Given that elections in Australia are fought across 151 seats, a national poll is not necessarily going to indicate what each of those individual contests look like, and given what Queensland as a whole was doing, that spelt disaster for the ALP. They needed Victoria and NSW to swing massively, but each of the key seats in those states did something different, because of local issues.


This strongly suggests that the weakening federal structure needs to be examined and fixed, something that has been identified by a group of professors from Griffith University in this article, which is worth reading for an understanding of Queensland politics more broadly. Given the intended structure of Australia according to its constitution, a localised and regionalised federal election defeats the purpose of the constitution, and muddles which issues are dealt with at which level.


(Image: Wes Mountain/The Conversation)

But regionalism and local issues are not absolute. There are always going to be voters in every seat that disagree with the majority, increasingly for reasons of identity. The relationship between identity and voting has been explored for many years, and possibly reached its zenith at the 2016 United States presidential election, which was following on from the British EU referendum earlier that year, and was also followed by a British general election in 2017 which showed signs of the same patterns emerging: working class areas voting for traditionally conservative parties, and richer areas voting for more liberal parties. Again, these are not absolute — there are many, many voters who have been ‘rusted on’ to their party for generation after generation in Britain and the United States alike.


But the tide is rising, and now it appears some Australians are taking a leaf out of their overseas counterparts’ books and voting the opposite direction to what they would normally be expected to do. Separate analyses of Saturday’s results from the Australian National University and the Grattan Institute have found that poorer voters were switching to the Coalition, while richer voters were moving to Labor. The three groups most identified with moving away from Labor are Christians, blue-collar workers and families with children, mirroring the US and British results to date very closely, and speaks to broad policy areas the Labor touched on throughout the campaign.


While the latter two may be broadly explained through the aforementioned economic issues, the questions of religious freedom and issues concerning life and morality (particularly abortion) were rarely brought up during the campaign, but were of real concern to many Christian voters. Given that Christianity was already a far weaker voice in the Labor Party before this election than it once was, the fact that those who had remained with Labor still fled in large enough numbers at this election to be one of the party’s three biggest demographic losses speaks volumes about how significant this issue was to that group.


Regardless of whether the concerns with Labor policy were accurate or not, the party was unable to effectively combat them. Without a readily identifiable central plank to their platform, Labor struggled to get across to voters what they were about; with a broad range of policy proposals, they left themselves open and were stretched too thin in defending themselves; with a leader that most voters didn’t especially like, identify with or trust, the chances of explaining and defending their policy platform diminished; without any central issue to grab hold of, local issues combined with identity politics to create a whole gamut of reasons for voting one way or the other.


Ultimately, voters were left to come to their own conclusions about what an ALP government would mean for them. Once they weighed it against a government promising to make precious few changes to anything, the majority stuck with the devil they knew best.


(Image: Dean Lewins/AAP)

A mix of years-long inaccurate polling, a dearth of national issues in favour of local ones combined layer with party identity shifts, and good old personality politics came together to result in a ‘shock’ Coalition victory, almost completely blindsiding the media and associated experts.


In the short term, there might be some changes to the way politics is seen and reported by those who got it so wrong. The polling companies won’t have a choice but to change their methods and the politicians have probably had enough of leadership changes for the time being (because their electorates are fed up with it). But the media, who do more to drive the news cycle than anyone else, are so dependent on having polls to drive the cycle that I wouldn’t bet on them learning their lesson in the long run. Betting is, after all, even more of a fool’s game than predicting.