Federal election 2019: Why opinion polls didn’t predict Scott Morrison’s shock victory

Australian pollsters are scrambling to explain their ‘disastrous’ attempts at forecasting the federal election, after every major opinion poll in the country failed to predict Scott Morrison’s ‘miracle’ win.

Australian pollsters are scrambling to explain their “disastrous” federal election after every major opinion poll in the country failed to predict Scott Morrison’s “miracle” win.

News Corp Australia has learned that four of Australia’s biggest pollsters, Newspoll, YouGov Galaxy, Ipsos and Essential, have launched “major” reviews of their election polls after the Coalition’s shock triumph on Saturday.

Commentators have referred to it as Australia’s version of Brexit or US President Donald Trump’s surprise 2016 election win, which overseas pollsters also failed to predict.

“None of us called it for the right team. It wasn’t good enough,” Ipsos Australia’s Jessica Elgood told News Corp yesterday.

Prime Minister Scott Morrison was re-elected in a shock result on Saturday. Picture: AAP

YouGov Galaxy managing director David Briggs also admitted the election was a “disaster” for the polling industry.

WHO WILL REPLACE BILL SHORTEN?

Every key poll throughout the election predicted a Labor win before all voter research settling on Bill Shorten’s party holding a 51-49 per cent lead in the week prior to polling day.

In contrast, latest Australian Electoral Commission results last night showed the Coalition won the election with 50.92 per cent compared to 49.08 per cent on two-party terms.

A Liberal source last night revealed the party’s internal polling had shown it was on a better footing than the consistent 49-51 per cent two party preferred vote.

The election was a ‘disaster’ for the polling industry. Picture: AAP

But party figures were still surprised by Mr Morrison’s “miracle” win.

Mr Briggs said Galaxy is now reviewing its systems to establish how its national poll could have over-estimated the Labor vote and underestimated the Coalition vote “so badly”.

But he added a series of YouGov Galaxy seat polls had show the Coalition could retain crucial marginal electorates that could prevent a Labor victory.

“As this interpretation was inconsistent with the prevailing narrative, many commentators dismissed these seat polls as being unreliable and argued that the process was fundamentally flawed due to some inherent bias,” Mr Briggs said.

“This assessment can now be seen to be incorrect but the perception blinded some commentators to the true standing of the parties.”

Opposition leader Bill Shorten had been tipped to claim victory, according to inaccurate polls. Picture: AAP Image/James Ross

Ms Elgood said the election’s outcome was within the polls’ margin of error but added: “There are lessons to be learned. We’ll all have to look at what we’ve done.”

She said it was possible there was a late shift to the Coalition, or many were the Australian equivalent of “shy Tories” who didn’t want to say they were voting for the Liberals.

Another explanation was so “location specific” the polls weren’t sensitive enough to pick up the outcome, she said. “There’ll be a thorough review,” she added.

All pollsters dismissed the prominent social media misconception that they weren’t able to predict the result because they were only calling landline phones.

Each company uses a different method but some use an online survey from a pool of thousands of Australian voters.

Prime Minister Scott Morrison was celebrating his victory at PointsBet Stadium. Picture: Brett Costello

Other methods include using the online panel and mobile phones, conducting “robopolling”, or undertaking a 50-50 mix of landline and mobile phone live calls.

This is completed by using random digit dialling to maximise the sample.

Australia’s disastrous polls mirror similar monumental overseas failures that have included a string of wrong predictions.

These included pollsters not predicting that Britain would vote to leave the European Union, that Americans would elect Donald Trump in 2016, or former British Prime Minister David Cameron’s majority victory in the UK 2015 general election.

Bizarrely, one of the only studies to predict a narrow Liberal win was an analysis of social media data and sentiment by Griffith University scientist Bela Stantic.

He predicted Mr Morrison’s victory by analysing millions of social media comments from more than half a million users over the past month.

His Big Data and Smart Analytics lab also predicted Brexit and President Trump’s win.

WHAT THE PRE-ELECTION OPINION POLLS SAID

Result: Coalition 50.92 — Labor 49.08 per cent

Source: AEC

NEWSPOLL

Labor ahead 51.5 per cent to 48.5 per cent on May 18.

Labor ahead 51 per cent to 49 per cent on May 13

Labor ahead 51 per cent to 49 per cent on May 6

Labor ahead 51 per cent to 49 per cent on April 29

Labor ahead 52 per cent to 48 per cent on April 16

IPSOS

Labor ahead 51 per cent to 49 per cent on May 16

Labor ahead 52 per cent to 48 per cent on May 6

Labor ahead 53 per cent to 47 per cent on April 8

ESSENTIAL

Labor ahead 51 per cent to 49 per cent on May 16

Labor ahead 52 per cent to 48 per cent on May 7

Labor ahead 51 per cent to 49 per cent on April 30

Labor ahead 52 per cent to 48 per cent on April 7

GALAXY

Labor ahead 51 per cent to 49 per cent on poll taken May 13-15

Labor ahead 52 per cent to 48 per cent on poll taken April 23-25

Labor ahead 53 per cent to 47 per cent on poll taken March 25-28