At around 9:30 p.m. Saturday night on Australia’s east coast, Antony Green, chief elections analyst at the ABC, made his announcement to a half-stunned, half-delirious nation. “At this stage, we think the Morrison Government has been reelected.” Prime Minister Scott Morrison’s conservative Coalition, widely expected to lose to Bill Shorten’s Labor Party after six turbulent years (and three Prime Ministers), instead won a majority government in the House of Representatives. For Labor, it was a crushing and baffling defeat in an election many thought un-losable. For the polling industry, however, the Coalition’s triumph was catastrophic, the latest high-profile miss over an abysmal and embarrassing three-year span.
Since the Coalition’s last Prime Minister, Malcolm Turnbull, squeaked by with a one-seat majority against Shorten three years ago, Labor had led two-party preferred polling nearly wire to wire. Turnbull’s inability to turn the ship around, or even gain any ground against Shorten, was a key reason his Liberal Party dumped him overboard for Morrison last August. The outcome didn’t seem any less uncertain as polling day approached. Popular gambling website Sportsbet opted to pay out Labor-backers two days early. Even exit polls that night predicted a comfortable Labor win. Only when the first actual votes were counted did it become apparent that things would not go as planned.
“At the moment, on these figures, it’s a bit of a spectacular failure of opinion polling,” Green remarked. To certain folks who keep track of this stuff, it’s at least the third such fail— Brexit and Trump included— in three years. So, is polling broken? Experts Down Under have echoed their counterparts in America and elsewhere: cost-cutting measures and changes in technology have thrown off the results. While voting may be compulsory in Australia, answering a telephone poll is not, especially when the voice at the other end is a robot. “Not everybody has a landline and the numbers that are published are incomplete,” added Martin O’Shannessy, former head of Newspoll.
For what it’s worth, one man did predict Scott Morrison’s win— and Brexit, and Trump. Data mining expert, Professor Bela Stantic of Griffith University in Queensland, analyzed two million social media comments over the course of the campaign, according to the ABC. He was able to surmise, correctly, that Labor wouldn’t be able to make the gains they needed to win. “I must [just] be careful of fake news,” Stantic told them. Or, he might’ve just gotten lucky (again). Since only public pages can be evaluated, for example, many issues remain before social media can be counted on as a reliable reader of the tea leaves.
Back in Australia, marketing and polling companies have been lobbying the government for access to the Integrated Public Number Database, so as to obtain more reliable results. While obtaining the listed and unlisted numbers currently available to emergency and law enforcement services might enable pollsters to cast a wider net, it still might not get more respondents to bite. The recent comedy of polling errors around the world has led many to question whether they could ever really trust pollsters again. “Perhaps we will see a change in how many polls are done in the future,” Green says.
Or, we could just wait for the voters to speak.