Australia has decided. And the major question in the aftermath of Labor’s devastating defeat is not how did a guy who gave himself his own nickname and (allegedly) shat himself in an Engadine Maccas become Prime Minister, but how did the nation’s supposed experts and polling agencies get the result so wrong?
Bill Shorten, as unlikely and unlikeable as he is, was understandably disappointed with the results. He’d been assured by practically every poll over the past few years and every single poll during the election that it was a lock. But alas, it was not.
In the past, opinion polling has been an accurate barometer of the Australian political zeitgeist. As opposed to the United States, where their laisse faire approach to voting means pollsters need to divine a voter’s intent to vote, compulsory voting in Australia means our researchers only need focus on how people want to vote.
And they are usually very accurate. Newspoll, Australia’s most famous and reliable political oracle, has never incorrectly called an election.
According to their website:
“Since 1985 Newspoll has conducted regular public opinion polls looking at a variety of social topics, as well as voting intentions and perceptions of Australia’s political leaders. In every instance, 59 state and federal elections to date, Newspoll has called it correctly.”
Those are some pretty impressive stats. Up until now. So what’s going on here?
Two things are at hand, according to most media reports.
First, the overreliance on landline telephones to engage voters.
In a recent ABC news article, former Newspoll boss Martin O'Shannessy blamed people’s changing telephone habits for the flawed results.
"The reason that it's hard to do good telephone polling is because the old White Pages — the phone book — doesn't exist anymore," he explains.
"Not everybody has a landline and the numbers that are published are incomplete."
The other reason is a form of number fudging. As Nobel Laureate in physics and the vice-chancellor of the Australian National University, Professor Brian Schmidt argues in a recent article for the Guardian, sampling errors come about, like in the case of flipping a coin ten times. Logically, this results in five tails and five heads — but mathematically, it occurs in such a small sample only 25.2 per cent of the time.
He describes how this relates to polls:
“If you do a similar calculation for the 16 polls conducted during the election, based on the number of people interviewed, the odds of those 16 polls coming in with the same, small spread of answers is greater than 100,000 to 1.”
In other words, the polls have been manipulated, probably unintentionally, to give the same answers as each other.”
However, this statistical smoothing may not be so unintentional. In a recent report by the New Scientist, independent analyst Kevin Bonham suggested polling companies may somewhat intentionally rejig their results to present the same data as their competitors. He points to an Australian example in the past where a polling company broke rank with their competition and called the wrong election result, losing them their contract.
While polling through ridiculously outmoded landline telephones and the fidgety nature of data collection are certainly major reasons why the polls got it so wrong — and most likely will continue to get it more wrong in the future — they don’t paint the complete picture.
One issue, picked up in the Spectator, is something called the Bradley Effect.
The Bradley Effect is a theory that accounts for the difference in opinion polling and election outcomes in the United States where a white candidate is up against a non-white candidate. The theory proposes that when asked, voters are more likely to tell pollsters they will vote for the non-white candidate, or that they are undecided, because they feel like it’s the socially desirable thing to do.
Once these individuals are alone, however, they vote the way they really want.
In an election dominated by highly emotive coverage around issues like climate change and income inequality, the Bradley Effect might explain why not only polling companies wrongly called the election, but why exit polling was similarly affected.
Our own version of this was perhaps best captured by former Labor leader/former Liberal Democrats candidate/current One Nation MP/potential reptile, Mark Latham. On the night of the election, while talking with Sky News host Paul Murray, he proposed a hidden electorate that doesn’t show up in polling data.
“With the polls, there is perhaps a 3 to 4 per cent hidden voter factor out there that don’t always manifest themselves, and you don’t find out they exist in any particular election until the votes are counted,” says the former member for Werriwa. “But that 3 or 4 per cent, maybe it’s bigger – there’s certainly a constituency out there, I’ve gotten to know them better through One Nation work – but they’re people who don’t like the system, don’t trust the media, would never answer a poll, would be dead-set against the things Bill Shorten was trying to do to the country and they just don’t show up in the data.”
Never missing an opportunity to take a dig at his former party, he argues Labor, who according to Latham, excessively indulges leftist identity politics, no longer represents the values of regular, working-class Australians.
The corollary: there is a small portion of disenfranchised voters that detest the Labor Party and, more generally, left-wing politics, and it isn’t being accounted for because they won’t answer polls either out of mistrust or out of fear of being judged negatively for openly admitting their politics.
His own likely bias aside, perhaps Ole Marky Mark is right?