We are now getting to the point in the election count where all of the primary votes appear to have been counted. We don’t have all the preference data yet – Bradfield is obviously still in play, the two-party-preferred count is not quite complete, and we don’t have 3CP or distribution of preferences data outside of Calwell. But now that we know how many votes were cast, we can analyse the level of turnout, and how many people voted early, and voted with a particular voting method.
People always like to make pronouncements about these statistics before the numbers are finished. It’s pretty common to see people taking the raw turnout figures long before the vote count is finished and then despairing how low they are.
In short, turnout is up a bit. Once you factor in the increased completeness of the electoral roll, the level of voter participation is the highest it’s been in decades, and likely much longer. More people have voted at pre-poll than ever before, while postal voting is down slightly from the 2022 peak.
Let’s start with the turnout. I last explored this topic in 2019, when the official turnout statistic dropped slightly (although, as is common, it dropped by far less than the journalists expected when they prematurely analysed incomplete data).
Turnout is defined as the number of votes cast as a share of the electoral roll. But there are also people who are eligible to vote who aren’t enrolled, and in recent years the AEC has become much better at enrolling voters who are difficult to get on the roll, so the enrolment rate has climbed from 90.9% in 2010 to 98.2% in 2025.
The kind of person who is difficult to enrol is also less likely to come out and vote, but not entirely unlikely to do so. So as the roll gets more complete, the turnout figure drops.
But there is an alternative figure, the participation rate, which calculates votes cast as a share of the voter-eligible population (VEP). And that figure has climbed from 84.7% in 2010 to 89.0% in 2025.
2022 was definitely a bad election whichever way you cut it. The enrolment rate dropped slightly, as did the turnout, so the participation rate reverted back to 2016 levels. But in 2025 we’ve seen turnout and enrolment figures increase.
And then let’s look at how people got to voting.
Prior to the election, it seemed clear that pre-poll voting was breaking the records from 2022, while postal voting seemed to be running slightly under the 2022 records. And that is exacty what has happened.
Postal voting has dropped from 14.6% of total turnout to 13.6%, while pre-poll voting has climbed from 36.5% to 41%. On this chart it looks like pre-poll has now overtaken ordinary voting but this is probably unfair because I have lumped together ordinary pre-poll and declaration pre-poll, while the election day equivalents are separate. 43.6% of people voted either absent or ordinary vote at a typical election day booth. That is still slightly higher than pre-poll voting.
If you simplify the picture to those who voted on election day and those who voted early, the early vote already exceeded the election day vote in 2022, but it has grown further this year.
The early vote reached a majority in 2022. It has continued to grow, but at a slower rate. 55.7% of people voted early via remote, pre-poll or postal voting. 43.6% voted on election day, with the remainder using a different method which can’t be categorised as early or election day.
Finally, I have produced a map showing the share of the electorate who voted early or on election day by electorate.
The seats with the highest rates of voting on election day are in South Australia, Tasmania and parts of Sydney. 58.8% of Tasmanians still vote on the day, as do 52.95% of South Australians. Over 70% vote early in the NT, which mostly reflects the big remote vote in Lingiari. Other than the NT, Queensland has the highest early vote, at 63.85%.
This map does show the 2PP by vote timing if you click through to a seat. Bear in mind this isn’t complete and is unavailable for some seats. I’ll return to this topic once the 2PP is finished.
Hi Ben, thanks for the great analysis – are you able to provide any insight from the data into whether the AEC had any success increasing turnout in remote Aboriginal communities – something they were focussed on achieiving? As you say this would be particularly relevant in Lingiari but also potentially Leichardt and Durack
It looks like the raw number of votes cast in Lingiari has dropped compared to 2022. So no, I don’t think so.
“and we don’t have 3CP””.
By this I take it you mean you can’t yet compute a National “3CP” figure.
How do you calculate the National 3CP figure?
Is it by adding up all the 150 electorates 2CP figures? Or is it by delving into the result and preference distributions in each electorate to get the 3CP in each electorate – and then adding up those figures.
Have you done both before? If so, how different are they?
However you do it, do you have a history of 3CP results? I would be interested to see back to 1996, and especially since 2007 & 2010 when “Other” began it’s current rise.
We don’t have 3CP data for any seat except for Calwell. We will have it when we get the full distribution of preferences dataset. The 3CP data is the second-last round of the count.
I don’t know why you would try to calculate a single national 3CP figure. Just like a single national 2CP figure, it would be meaningless because the combinations would be different in different seats. You could do it on the Senate vote, I suppose.
You could collate 3CP in all three seats as far back as 1984. Prior to that they only did a complete DoP when necessary. I’ve got a spreadsheet with all 3CPs back to 2004.
I wrote this piece in 2022.
The reason you would do it would be to see the changes over time in the 3CP. It’s a proxy for the support for non-major parties.
It’s really not.
One of your sentences seems to be missing an ending? You wrote “If you simplify the picture to those who voted on election day and those who voted early, the early vote already” … and then trailed off.
Thanks Grace, fixed now.