Opinion polls are perhaps one of the noisiest parts of a federal election campaign. It seems like every other day, there’s a new poll accompanied by a splashy headline making a bold prediction about the fate of the nation. But can you trust it? And is polling able to capture the vibe shift towards minor parties and independents?
We asked Tasmanian psephologist (someone who studies elections and voting) Dr Kevin Bonham for some advice on interpreting the polls in 2025 – and why it’s a little different to the last election.
What do polls actually mean?
Polls are a snapshot of how a section of voters are feeling about the election, the political parties and its leaders at a given moment. They are an indication of the current mood – and nothing more than that.
Kevin stresses: “Polls are not predictive.”
While the final polls done in the last few days of a campaign “get closer to the actual result”, anything released before then is “a snapshot of what is going on at the time it was taken, and not a prediction.”
Now that we’ve cleared that up, how should we understand the polls that are published throughout the campaign? What’s really being measured?
The two-party preferred polls (2PP) are “a measure of the primary vote support for the parties, and the estimated two party margin between Labor and the Coalition.”
Preferred prime minister and approval ratings are “standard leadership indicators for how popular the leaders are, and how popular they are in comparison to each other. But [preferred PM polling] is a bit fishy because it tends to skew towards the incumbent.”
What does this mean in the era of major party decline?
Measuring the comparative popularity of the major parties is all well and good. After all, only Labor or the Liberals (supported by the National party) can actually lead the government. But is this kind of polling still helpful in an era when support for either major party is in serious decline?
The primary vote (i.e.: people give them the #1 preference) for Labor and the Coalition is at an all-time low. In 2022, Labor’s primary vote was just 32.6% and the Coalition 35.7% – these are very low numbers, once upon a time the major party primary vote as was almost as high as 50% in the 1930s. This is being driven by young voters from Millennial and Gen Z generations becoming eligible, as both groups have very weak political affiliations compared to older generations.
Is there any type of polling that can capture this shift?
“There are some struggles with that,” Kevin explains. Because voting independent is such a hot topic, Kevin says a lot of the current polls are showing independent voting intention a little bit higher than the final result is likely to be. “We are seeing polls showing independent at about 10% [of the voting intention] – some of that is voters wishfully thinking they had a good independent to vote for.”
The presence of independent candidates is also very local and seat-specific, which is a problem for nationally run polls taking a sample of voters from across the entire country or state. “You need to know where support is concentrated enough for independents to win a seat – we’ve seen in state elections independents [in polling] can get, say, 7% across the state. Depending on which state it is and how their vote is split around, that might be no seats at all, or it might be several seats.” Local polls are usually too unreliable to fill in the blanks.
The push to ‘ditch the duopoly’ in the 2025 election is also having an impact on the clarity of the two-party preferred polls. “Because there are more preferences of minor parties and independents, the two-party vote becomes more dependent on how those preferences flow, and so it becomes harder to get the two party vote.”
New bombshell enters the villa: MRP poll modelling
There is a new polling technique that attempts to factor in the rising popularity of independents, minor parties by modelling the responses to show a result for every individual seat. It’s called multi-level regression with post-stratification, also known as MRP modelling. You might have seen news about YouGov’s MRP results from the end of March, which showed Labor just one seat short of a majority government.
“MRP is an attempt to model every seat off breakdowns based on small samples in the seats that are networked together. It’s quite complex to explain, but they switched to this sort of bulk model method away from individual seat polls, because individual seat polls have so many problems.”
“It’s still pretty experimental in Australia. There, there was one at the last election, I think there was one for the voice. That is something these MRP models are trying to help us with, but they’re still struggling a bit with seats where the independent is a local independent who is not part of a set pattern.”
Advice for how you should interpret the polls
So, with all that said, how much attention should you pay to election polls? Kevin’s golden piece of advice is to look for the polling trend rather than at individual poll results. “Look at results across a range of polls – don’t place too much stock in any one polling result or in any one pollster. It’s better to look at results over time, rather than focusing too much on any one poll that might be an outlier or might have some kind of method issues.”
Some bonus advice from Zee Feed: no, the election polls published by the Australian Financial Review (by Freshwater), The Australian (Newspoll), Guardian (Essential), or the Sydney Morning Herald/Age (Resolve Monitor) are not biased. The polls are run by polling companies that are independent of those media outlets, and do not survey subscribers of the publications. The relationship between the media outlet and pollster is around who gets to exclusively publish the poll results – that’s it.
And make sure you focus more on the poll data itself, rather than what a news headline says about a poll’s results. As we’ve written before, they don’t always match!
Comments are closed.