Australia’s major opinion polls all accurately predicted Saturday’s Labor election victory, in a boost to pollsters’ reputations after the industry-wide failure at the 2019 election.
Surveys measuring voter sentiment throughout the campaign pointed to a Labor win, with the polls tightening in the final week before Saturday’s election.
However the surge in support for the Greens and independents showed that the national two-party preferred measure that has traditionally been used to gauge an overall winner will be less able to paint a comprehensive prediction in elections where the most hotly contested seats are not between Labor and the Coalition.
Labor had 73 lower house seats called in its favour by Monday, according to Guardian Australia’s electoral analysis, and is likely to gain a further three to secure a parliamentary majority.
The Coalition has 51 seats so far, while independent candidates and the Greens were hoping to pick up as many as 10 and four seats respectively. Centre Alliance and Katter’s Australia party retained a seat each.
While votes were still being tallied on Monday, results showed the Coalition received 35.7% of the primary vote nationally, while Labor received 32.8% and the Greens 11.9%.
Other candidates – including independents, the Centre Alliance and Katter’s Australia party – received 10.5%, with that vote translating to as many as 13 seats. The 4.9% of primary votes that One Nation candidates received and the 4.2% for the United Australia party did not help either win a single lower house seat.
In the six major national opinion polls run in the final 10 days of the campaign – conducted by market research firms Essential, YouGov, Ipsos, Resolve and RoyMorgan – all measured a Coalition primary vote of between 34% and 36%, but were more varied on Labor’s primary vote, with results between 31% and 38%.
On a two-party preferred measure, these polls varied from 54-46, to a 51-49 result for Labor. By Monday afternoon, the national two-party preferred result appeared to be about 52-48 in favour of Labor.
Pollsters’ uniform accuracy in this election is a total reversal from 2019, when every major poll failed to predict Scott Morrison’s re-election. That result sent shock waves through the polling industry, and kicked off a period of reflection, innovation and transparency.
After an inquiry into 2019 election polling which found the respondent samples were unrepresentative of the public and not adequately adjusted, all pollsters tweaked their methods.
There was a move away from so called “robopolling” – phone calls with an automated voice listing the voting options and asking which demographic categories you fall into. Robopolling was only used in individual seat surveys this election.
All major polling companies conducting national polling now use online surveys, and pay respondents an incentive, and some also incorporated telephone interviews.
The executive director of Essential Media (which conducts polling published by Guardian Australia), Peter Lewis, said that while continuing vote counts could change the final figures, there was a sense of relief that pollsters weren’t the main news story the day after the election.
“We’re pretty happy with the way our poll charted the mood all the way up to the election,” he said.
“The big story here is that after 2019 different pollsters changed their methodologies to see if they can do things better and what we’re seeing is most can say they did in fact do better.”
However Lewis stressed that national two-party preferred polls weren’t designed to capture the performances of minor parties and teal independents in specific seats, and said polls should still be “taken with a grain of salt”.
“I’d hate to think just because polls were close to the real result close this time that people start treating polls as gospel,” Lewis said. “It should be a navigator, but it’s never a perfect map.”
Murray Goot, an emeritus Prof of politics and a leading polling expert, said that while early results on Saturday night suggested polls had overestimated the Coalition’s primary vote, updated tallies had levelled this out. But he said most polls appear to have again overestimated Labor’s primary support, as they did in 2019.
Goot pointed to the election pendulum concept – which lists seats held by each major party based on their winning margins at the last election, with the closest seats nearest the centre – and said that most surveys correctly had Labor above 51.8% of the two-party preferred result, which was the point on the pendulum at which Labor would have a parliamentary majority.
Regarding the success of Greens and independent candidates, Goot said there were shortcomings of national primary polling when looking at minor parties.
The UAP, whose efforts were spread across all 151 electorates it ran candidates in, gained 4.2% of primary votes but no seats, while Greens and independent primary votes, at 11.9% and 10.5%, were “clustered” in just a handful of electorates so better translated to seat gains.
Goot said that to predict outcomes of any seat, individual electorate polling is needed. For this reason, it was difficult for national polls to predict that the Greens primary vote would be clustered enough to translate to the four lower house seats it could gain.
Lewis said that individual seat polls – most of which were commissioned by candidates and parties – “largely got it right”, with several surveys toward the end of the campaign pointing to teal independent victories in Wentworth, Kooyong and Goldstein and tight contests in others.
The small but significant primary vote for UAP in the lower house dispersed across the country, that was detected by polls, was seen in the senate, with the party on track to win an upper house seat in Victoria.
Senate polling is rare in Australia, Goot said, in large part due to a lack of interest in the upper house in the context of media coverage of who will win government. He also said it was difficult to poll for, given the importance of preferences in deciding how the fifth and sixth seats (in states) are distributed.
