Australian federal election 2022: Can we trust the polls?
As the calling of the next federal election for May this year looks imminent, Yahoo News Australia is analysing all the aspects that will contribute to the outcome of the vote.
With many tipping the first Labor victory in four elections, that would send the more conservative Coalition back into opposition for the first time in nine years.
The latest Newspoll suggests this outcome with the combined Liberal-Nationals primary vote falling two points to 34 per cent.
The poll also shows that Labor would win 56-44 on a two-party-preferred basis.
But with the memory of a 2019 Labor election defeat still fresh in everyone’s minds, when polls overwhelmingly had them winning, can we trust the polls this federal election?
Australian National University professor of political science and polling expert Ian McAllister’s has spent many years studying polls across the world.
They are affected by many factors, and can be relied upon only for a short period of time, he says.
“Polling is generally an accurate reflection of people’s voting preferences at one point in time when the poll was conducted,” Prof McAllister told Yahoo News Australia.
“If something changes, events change, attitudes change, policies change, then people’s preferences can change.
When is the next federal election? Everything you need to know
Scott Morrison stunned by reporter's text message question: 'Horrible person'
Scott Morrison responds to 'psycho' text message report : 'No one cares'
“The other thing that is most interesting about this – and I’ve done a lot of research about this in various countries – what generally happens at the start of an election campaign is the voters are generally apart in one way or another.”
But as the election draws nearer, the race usually tightens.
“Generally they are five or seven percentage points or even more, and then what you see is a convergence towards the polling day itself, and that happens virtually in all countries.
“You can get late swings during campaigns that occur maybe even in the last two or three days and in Australia it’s always a particularly tight contest between the two major political parties.”
The 2019 polls that got it so wrong
So what happened during the 2019 election campaign that pollsters got so wrong?
A report prepared by the Association of Market and Social Research Organisations (ASMO) from an inquiry into the failure of the polls from the 2019 election suggested a number of issues contributed to the faulty projections.
One of the findings was that one poll was deliberately withheld from the public as it was “out of alignment” with the other polls that all pointed to a Labor victory.
However, the difficulty in obtaining information from the organisations responsible for the polls prevented the inquiry from reaching a specific conclusion as to why the public polls got it so wrong.
Prof McAllister believes the answer may lie with the way the polls were conducted and the reliability of the participants involved in any particular survey.
“How the polls are conducted, the methodology, is really quite important so the sampling methods they use can be quite sensitive, particularly in our case where you’ve got very close gaps between the two major parties,” he said.
Polls typically omitted the undecided, or 'don't know' voters, who tend to be the cohort which ultimately decide the election outcome.
“A big thing is how they deal with ‘don’t know’ respondents in surveys, which is typically five to eight percent of the people being sampled," Prof McAllister explained.
“The inquiry by ASMO found that all the polls gave the wrong answer and a lot of it came down to how ‘don’t knows’ are allocated.
“If you’re saying there’s a gap of four or five percent between the parties, that’s usually more than the ‘don’t know’ respondents.
“If you assume the 'don’t knows' are randomly distributed across the electorate you can reliably exclude them, but if they are systematically biased in any way, such as according to economic background or political predisposition, it can make a difference.
“You’ve got to be very careful with online polling because there’s about 20 per cent of the people who won’t do anything online and you have to have some methodology for picking them up as well.”
Some polls, such as the Guardian Essential poll, have since tweaked their methodology and how they report their poll results, to better account for the undecided voters.
The 'Spiral of Silence'
Prof McAllister went on to explain that many of the ‘don’t knows’ may simply be too embarrassed to admit they are supporting certain parties due to a stigma that may be attached to them.
“I did some polling in Northern Ireland when Sinn Fein was the political rep of the IRA in the 1990s (a paramilitary group known as the Irish Republican Army) and we had a damned of a job in surveys getting people to admit to supporting them,” he said.
“You’d find in the surveys 8 to 10 per cent of people would say they supported Sinn Fein and then in an election you’d get 25 to 30 per cent of people who voted for them.
“It’s what we call in polling the ‘spiral of silence’ – people won’t admit to voting for parties that they regard as not socially acceptable.”
As trust and support for the incumbent federal Coalition wanes under prime minister Scott Morrison’s leadership, Labor leader Anthony Albanese look set to take up residence at the Lodge.
But will the polls yet again prove misleading? It's possible.
“When you actually look at people who support the minor parties the number is not great in the surveys, but what you find is about half of them are former Labor voters, which suggests that they are basically protest parties,” Prof McAllister said.
“People are jacked off with the two major parties.”
How that plays out this time around remains to be seen but the pollsters have a major chance to redeem themselves.
Do you have a story tip? Email: newsroomau@yahoonews.com.
You can also follow us on Facebook, Instagram, TikTok and Twitter and download the Yahoo News app from the App Store or Google Play.