Australia’s major pollsters are all tipping Labor to beat the Coalition on election day. But one data firm says it has picked up a trend showing the contest is much closer than it seems.
Elisa Choy is the founder and managing director of Maven Data, a company that uses a form of artificial intelligence called natural language processing to glean insights from open source data, including social media, websites and blogs. Essentially it slurps up as much information online that it can, then applies a secret sauce so it can find trends about what people are thinking about and how they’re thinking about it.
In the past, Maven Data has used its tools to extract insights on subjects ranging from Australians’ attitudes towards migration, the wellness industry globally, and the 2020 US election. Choy even brags it has been able to predict the past three winners of the reality television show The Voice.
Now it is turning its gaze towards the Australian federal election.
What it’s seeing, Choy says, is that the election is still neck and neck. Maven Data’s analysis looks at the amount of engagement and sentiment towards the subject. Choy says the two major parties have similar engagement, but the Coalition remains a stronger brand.
However, she cautioned that Scott Morrison’s reputation appears to be holding back the government’s chances: “Sentiment towards him is very negative, especially on issues including women, integrity and climate change.”
Maven Data’s analysis also looks at some of the minor parties. It found the Greens have a very strong sentiment — “almost love” — but don’t have a strong engagement across the country. Plus their pillar issue of climate change has not featured during the campaign.
The United Australia Party on the other hand has strong engagement across the country, as issues such as “freedom” and ending COVID-19 restrictions continue to resonate. Choy was surprised that Palmer’s party is similar in terms of engagement to Labor and the Greens.
Beyond the election, Choy says Maven Data has picked up a tainting of Australia’s brand linked to the federal government’s border decisions during the pandemic: “From a business and industry standpoint, an election will pass but business continues. The world is watching.”
Does Maven look at things on a seat by seat basis, or just the general vibe?
Australia has compulsory voting, so it’s different to voting for The Voice, where only the engaged are going to vote (doesn’t say how accurate they were on US election).
Not seeing too much difference for isentia doing media tracking from the article.
My take as well, Crikey should make some effort to update this article – local context is important. It also don’t make any clear measure of degree and speed of changes in opinion. Not much to lose for Maven, a free advertorial, was my strongest reaction.
I agree, the contest is marginal seats not overall public sentiment, and no mention in the article. Not the standard we have come to expect from Cam.
Cam, could you do an article comparing poll methodology and which polls are most helpful considering marginal seats?
There are many people who don’t do facebook, tik tok and twitter, etc., the things that they measure.
And responses on Facebook etc can be stacked. Get lots of comments in support but not necessarily from many people.
Although it is a Liberal loss that I am desperately hoping for (rather than a ‘Labor’ win), I remain very skeptical about what might, or might not happen on the evening of May 21st. I would merely urge/encourage all non-conservative voters to get out there now and letter-box, handout how-to-vote leaflets or help in some way to try to ensure that Morrison and his disgraceful government are removed from office on Saturday week.
The high level of engagement for the UAP really does highlight the stupidity of the electorate and my level of naivete.
Saturation national coverage via $100m
But that UAP ‘engagement’ level – whatever that is – is what makes me wonder why Cam bothered to write this. Has this Maven crew got any kind of reputation for election foretelling, based on its ‘gleaning insights… slurping up information… applying a secret source… finding trends’? I’m not impressed by someone’s having predicted ‘The Voice’ winner. I can’t discern anything of substance here.
How is engagement measured, and is it necessarily positive? I read a lot of derision posts about UAP – is the AI smart enough to tell the difference between those and supportive ones?
I am not all that confident that Albanese and the Labor Party will defeat Morrison and the LNP next week. Think the best we can hope for is a hung parliament The Murdoch media is painting a ‘80 seat Labor win’ prediction in an attempt paint the LNP as the underdogs. 2019 all over again
yep…never underestimate rupert’s agit-prop crew’s cheerful willingness to risk ‘looking wrong’ in order to maximise the chances of them getting the result they want. you watch the ‘hand on heart’ polling ‘tighten’ in the next week, accompanied by the studiously ‘fair’, ‘earnest’ Op Edding – more in sorrow than anger, don’tcha know – that the electorate in its no-nonsense tribal wisdom is adjudging that, nice bloke though he be, dear old plucky-effort Albo just isn’t up to the tricky economic and foreign policy jobs ahead…
i know i sound paranoid Fez but…t’is a bitterly hard-earned paranoia!!! 🙂