The latest Newspoll would suggest the Coalition is in dire straits, with Labor leading 56-44 on a two-party preferred vote.
Those numbers make another victory for the Coalition seem unlikely, especially this close to an election. But there’s a general distrust in polling since Donald Trump and Brexit proved them wrong and, closer to home, the 2019 election result.
So just how bad is this pre-election poll compared with others? Could the Morrison government go on to win anyway?
Keating 1993 (election held March 13)
In a precursor to 2019, this election was one where the pollsters simply got it wrong. Keating was polling at 46 points two-party preferred in late February, and by election eve was still trailing the Coalition at 50.5 points. It was always supposed to be close, but no one expected Keating would pick up new seats. Similarly to 2019, Liberal leader John Hewson’s heavy policy agenda — including a proposal to introduce GST (plus the birthday cake thing) — are often blamed for the unexpected support for an incumbent.
Howard 2001 (election held November 10)
In early 2001, Labor was holding strong on two-party preferred, and at its lowest point in March, the Coalition polled at 43 points. The Tampa affair (mostly credited with winning the election and birthing the “tough on borders” election ploy) provided a two-point boost. But it was really 9/11 that secured the victory, with the shock providing a five-point boost in the aftermath of the attacks, and had the Coalition lead the two-party preferred in for the first time that year, two months out from the election. The polls narrowed before polling day, but the two boosts proved enough to deliver a turnaround victory for the Howard government.
Howard 2004 (election held October 9)
Howard was again able to turn polls in his favour in the lead-up to the election after recording two consecutive low numbers — 45 points in March and 46 in August. The Coalition eventually found the support it needed, and the polls went with it again — just in time to take out the election in October.
Morrison 2019 (election held May 18)
In a result reflective of yesterday’s poll, Morrison recorded a two-party preferred result of 44 after the Lib spill in August/September 2018. According to polling, the two-party preferred vote narrowed but remained in Labor’s favour — but obviously this is not how the election panned out. 2019 repeated the surprise results of 1993 where the polls simply didn’t see the result coming, and battleground seats ended up delivering the election.
Morrison 2022 (election TBA)
The latest results suggest Scott Morrison has another uphill battle ahead, but the relevance of two-party preferred polling has been questioned. The Sydney Morning Herald and The Age have recently reentered the polling space, and have done away with recording two-party preferred results, claiming the “horse race” it creates distorts the actual situation. Another misread result may just do away with two-party preferred altogether.
Do you have any faith in the polls? Let us know your thoughts by writing to letters@crikey.com.au. Please include your full name if you would like to be considered for publication in Crikey’s Your Say column. We reserve the right to edit for length and clarity.
I have some faith in the Two Party Preferred polling, but it would be more accurate if pollsters actually asked who people intend to preference, using last time’s preference distribution is clearly a short cut with limited statistical validity. Why on earth try and predict this year’s result on what people were thinking three years ago? It doesn’t make any sense at all. But it does save money.
Preferred Prime Minister is simply a load of tosh. The incumbent always gets an enormous boost from media coverage which totally distorts the perception. But I guess right now it’s the only thing NewsCorp have to go on “LOOK!!! LOOK!!! Our boy is preferred by a whole 1% more of the population. Isn’t he doing a fabulous job, and these figures show it. He must be a shoe in”
I broadly agree with this, but the standard industry response is that people may claim they will preference a certain way, but on election day, they will follow the how to vote card. This is alleged to have been a big factor in underestimating the flow of Palmer preferences to the government at the last election. It would be useful if the stenographers (oops, sorry, press gallery) could give some informed attention to how this may play out this time.
“Howard 2001” …. “Children Overboard”?
Those lies (while those that were there and their superiors were trying to convey the truth of what happened to those perpetrators of mendacity – Reith, Howard, Downer, Ruddock, Hill et al, more interested in political canard than the truth and being honest with voters) – and the subsequent protection provided by Halton and other actively political public servants?
The base strumming of such xenophobia.
It sure would be nice if the MSM spent less time on poll-fuelled speculation, and more time on policy or the lack of it, and counting broken promises.
… You mean work at earning their salt?
We must not forget the Murdoch monster setting the daily agenda that many of the smaller lickspittles follow blindly. This was recently mocked in the UK, where a commentator said whereas we have a Murdoch problem Australia is stage 4 terminal.
The thing about opinion polling is that (assuming the sample size is reliable enough) there are two distorting factors – honesty and representation.
People often lie to pollsters because they are embarrassed to reveal how they vote – so some Trump and Brexit voters lied about their intentions because they didn’t want to admit it (to avoid being judged by the pollsters).
Representation is a problem if the method of polling isn’t neutral – such as via landline getting more older people, and online getting more internet-savvy people. And those people refusing to answer may go mostly one way.
The question is which way will the opinion poll results be skewed this time around?
To really understand the data, the polls need to be in real time, on a tv/internet channel set up for polling, and people can be called, call in, internet in, SM in their response. Then as well as total on TPP, the other question that is asked is what is their main media channel for political news. 7 9 10 Sky commercial radio vs ABC SBS vs social media.
I know that the embarrassment factor theory is widely supported, but it does seem at odds with the aggressive pride of Trump supporters. Why would they be embarrassed to reveal their support for their hero, surely a chance to stick it up the pollster elites?
Pollsters ask questions but do not listen to the “vibe”. The two people I found to have predicted the outcome of the last election were my Physio and my Barber both good listeners and conversationalists. Both these men were very close to the final result and both based their predictions on real cross-sections of society.
Polling is asking mainly set questions People vote on many issues not polled.
But did your physio and barber actually talk to a real cross-section (how many women did the barber speak to, for example) or did they just guess? How detailed were their predictions?
What happened to the good old taxi driver, the ultimate barometer of public opinion?