The Australian Electoral Commission conducts a two-party-preferred vote in every seat, which is a count between Labor and the leading Liberal or Nationals candidate. In seats where the two-candidate-preferred vote is not Labor vs Coalition, they do the count later. With 35 seats now “non-classic”, the AEC needs to conduct those notional 2PP counts in more and more seats.
This means that it is possible to calculate the national 2PP. It also means that it is easier to make comparisons over time that cover the whole country. 2CP counts vary from seat to seat so can’t easily be compared, and 2CP pairings often change in a seat. Not so for 2PP.
The national 2PP count has been stalled for some time now, with 149 seats finished their count. But the Bradfield count has barely started. I asked the AEC about this, and it appears that the count remains on hold while they wait to find out if there will be a court case. So it is possible it could be quite some time before Bradfield’s 2PP is determined. So for this post, I wanted to look at the data to come up with some estimates of the likely 2PP, and come up with a number I can use for some other analysis I’d like to do.
The first thing worth noting is that there is very clearly a relationship between the 2PP and the 2CP in these sorts of seats. For this analysis, I’m going to look at data from the urban seats where teals have beaten Liberals in 2022 or 2025: Bradfield, Curtin, Goldstein, Kooyong, North Sydney, Mackellar, Warringah and Wentworth.
This chart shows every polling place in these seats in 2022 and 2025, excluding Bradfield 2025. While there is a high correlation (about 0.85), the independent 2CP is usually higher than the Labor 2PP. The black diagonal line shows the point where the two are identical. Very few booths have a higher 2PP. But that gap is not always the same.
This next chart compares the 2CP and 2PP in each contest.
Warringah, Mackellar and Wentworth had a 2CP 10-12% higher than the Labor 2PP in 2022. On the other hand, the gap was just 2.3% in Bradfield in 2022, 4.2% in North Sydney in 2022, and gaps of 3-4% in Goldstein and Kooyong in 2025.
This could imply a Labor 2PP of 40%, or a 2PP of 47-48%. So what is more likely? And what explains the differences.
I think a big part of the story is the make-up of that 2CP and 2PP. Votes cast for the independent will all flow to the independent in the 2CP, but some of them won’t flow to Labor on the 2PP. Conversely, votes for Labor will partially leak on the 2CP, but not on the 2PP.
This does seem to be part of the story. Bradfield and North Sydney have tended to have higher Labor primary votes compared to other teal seats, and lower primary votes for the teal independents.
So I wanted to come up with a metric to measure this dynamic, and how it might relate to the difference between 2PP and 2CP. I came up with two that seem to both work: the ratio of teal to Labor primary votes, and the proportion of the teal 2CP that started as a teal primary vote. I’ve gone with the second one but they seem to produce similar results.
If you plot out the independent primary share of the 2CP, and compare it to the 2CP-2PP gap, there seems to be a strong relationship. Such a chart implies that Bradfield in 2025 would have a gap of about 4.8%.
The other measure worth considering is the proportion of preferences that flow to the ALP on the 2PP.
In all of these contests, Labor has never polled less than 65% of the 2PP preference flows, and have gotten as high as 79.4% in Kooyong this year.
If the Labor 2PP was to be 45.2% as the previous chart suggested, that would be a Labor preference flow of 59.8%, well below other seats.
If Labor gained a 65% preference flow, that would give them a 2PP of 47.4%. A 70% preference flow would give them 49.5%, almost as good as the independent 2CP.
So I think we should go somewhere between these numbers:
- Judging by the independent primary vote share of the 2CP – 45.2%
- Judging by 65% Labor 2PP preference flow – 47.4%.
So I’m going to go with a figure of 46.3%. I don’t have a great deal of certainty on this – it’s entirely possible the number will be even higher, but this will do.
This would bring Labor’s national 2PP down from the current AEC figure of 55.28% to 55.23%. This is consistent with the ABC projection of 55.2%.
It would also mean that Bradfield would have a swing to Labor of 2.43%, compared to 4.2% in Mackellar and 5.3% in Warringah. So perhaps that is a sign that I’ve estimated the Labor 2PP a bit low, but I’ll stick with it.
It may be a few days before I get around to continuing this analysis – I will be travelling to Canberra tonight for a conference on the federal election book tomorrow, and at some point I’d like to do a podcast about Tasmania – but I will be coming back to this topic to look at how 2PP in each seat has shifted over the last two decades.
Reid is effectively Lowe just renamed to preserve a division named after a PM
@ John
Yes Reid is effectively Lowe Plus Lidcombe (part), Silverwater and the Olympic Park area.
@John pretty much minus the bit at the bottom.
Apologies if I have missed the explanation, but why does Warringah have a higher notional 2pp result for Labor (54.5%) compared to Wentworth (50.6%) when the latter, as stated in the comments, is the most left wing teal seat due to a higher density of Labor and Greens voters?
On an off note, my curiousity’s getting the better of me. Besides Lowe and Reid, are there any other electorates which have retained their old boundaries (or similar boundaries) but have changed their name? I only know of Throsby-Whitlam, Prospect-McMahon, Wilmot-Lyons, Darwin-Braddon, Fraser-Fenner and technically Charlton-Hunter.
@Lurking Westie – I also believe Denison-Clark and Balaclava-Goldstein apply.
James and Lurking Westie – although the current seat of Spence in South Australia is technically the successor to the seat of Wakefield, it actually resembles the abolished seat of Bonython.
Spence, like Bonython, is a purely urban seat centred around Elizabeth and Parafield Gardens. However, it does extend further north compared to Bonython, reaching Gawler whereas Bonython only reached as far as Munno Para.
@Yoh An – thank you for clarifying. That one did slip my mind.
@ Yoh An
That is why Spence is a bit weaker than Bonythorn as Gawler is a more marginal area. However, as Gawler is less of a country town these days it is actually trending Labor as we can see in the state seat of Light.
Agree Nimalan, Gawler is probably like Mandurah in Perth – once a distinct regional town but now seen as part of the greater metropolitan area of the capital city.
Other regional centres like Geelong, Wollongong and Gold Coast also share this trait being seen as extensions of the main capital city conurbation.
@ yoh An
Difference between Geelong and Melbourne is the presence of the Green Wedge which is protected from development so it is outside Melbourne UGB (Urban Growth boundary). Sydney and Wollongong are divided by Royal National Park which gives a seperate identity. Gawler and Adelaide dont have that so will merge over time
@High Street, I’m in an AEC briefing right now. They will publish their own numbers when the final election results are published, and I will suggest a different method of calculating the 2PP when the 3CP data is available, and explain the differences and why the AEC in their role has limits on the assumptions they can make.
Look forward to it Ben.
Did they explain why they are so concerned about just doing the actual count? They have a 6 piles of voter per booth they require to be touched. I’m not sure of the reasoning is that they just can’t go and do it.
Is this the first non traditional contest recount in the 2PP count era? It can’t be, as the Clive Palmer seat in 2013 would have been the same.
On the subject of seat name “changes”, one interesting one is that with the movement of Pearce over time, the boundaries of the new seat of Bullwinkel actually align pretty closely with those of Pearce in the 1990s (which have no overlap with Pearce now).
Lowe (now Reid) and Higgins (now part of Melbourne, Kooyong and Chisholm) are two seats of ex-Liberal leaders that the Liberals wouldn’t be holding right now based on the two seats’ last boundaries before abolition.
Also, Batman became Cooper.
@Lurking Westie
“…why does Warringah have a higher notional 2pp result for Labor (54.5%) compared to Wentworth (50.6%) when the latter, as stated in the comments, is the most left wing teal seat due to a higher density of Labor and Greens voters?”
There are various measures of how left-wing a seat. Nimalan made a list of them.
The type of left-wing in Wentworth might be a lot different to the left-wing in Warringah due the proximity to universities, the Inner West and CBD; younger population, more renters; bougie suburbs like Paddington and Potts Point and gentrifying suburbs like Randwick. Wentworth has more hipsters.
Based on senate primary votes
Wentworth vs Warringah
LNP 37.8% vs 34.0%
ALP 33.8% vs 36.6%
GRN 15.4% vs 15.2%
Warringah had a more dispersed primary vote for minor and mico parties. Nearly every right-wing minor party got more votes in Warringah than in Wentworth – PHON, Trumpet of Patriots, Libertarian and Gerard Rennick People First, Family First, Australian Christians.
@Votante also Zali Steggall got more votes in Warringah since she’s had more time to build up a personal vote thanks to her first two challengers (Tony Abbott and Katherine Deves) being right-wingers.
If it was Labor vs Liberal and no other candidates I think the Libs would win all of the teal seats, but there would be lots and lots of voters who voted for Steggall directly following her HTV card (which has Labor and the Greens above the Libs).
As for Wentworth as I stated even if it has parts that are quite socially progressive it also has Vaucluse, Double Bay, Dover Heights and Watsons Bay, all of which are very affluent blue suburbs.
Furthermore, i think areas like Allambie Heights in Warringah wil have more tradies. Mackellar has a lot of tradies especially in the lower density Forest Dirstrict parts. The Northern Beaches is generally low density apart from Manly.
Mackellar would have lower levels of Education as well
Can I have a fact check here, please. I find it hard to credit that what Nether Portal says is correct, that Steggall had a HTV that recommended Labor and the Greens above the Libs. If it had I feel sure we would have heard about it???
@High Street my mistake this time she didn’t but I swear she did last time.
@Nether Portal, Zali Steggall has had a high profile partly as an Olympic medallist from her first campaign. She has a strong personal vote and won on primary votes at her 3 elections. This would’ve suppressed the Liberal vote. I think North Sydney was trending to Labor before Kylea Tink won it.
on a 2PP measure, North Sydney strongly continued trending to Labor even when Kylea Tink did win it! A 8-8.5% 2PP swing in 2022.
Some of the 2PP and Labor primary vote figures in the western side of North Sydney council are pretty staggering. Waverton is a very good example – 24% Labor but only 21.5% Liberal. Think the 2PP is around 66% Labor. Next election the Liberal voters might be the ones “tactical” voting.
North Sydney lGA along the Railway line is higher density. In the old electorate of North Sydney the bluest parts were Lane Cove and especially Hunters Hill. If the Teals did not exist, it possible though that in the parts of North Sydney closest to the City that Greens will outpoll Labor.
Labor put effort into North Sydney unlike other Teals seats hence a high primary and a lower Teal primary. However, for the time being in seats like Wentworth and Warringah maybe Liberals should tactically vote Teal just to be on the safeside
There is very little evidence that the Greens would outpoll Labor in parts of North Sydney. The 2025 results are generally 2:1 in favour of Labor.
I don’t think the Greens would’ve overtaken Labor in North Sydney in 2025. The Greens did worse in the inner-city electorates (without teals) this election whilst Labor did better.
Labor came second on primary votes at various booths in Crows Nest, Waverton and Wollstonecraft (all now part of Warringah).
@ Votante
The Greens did not do well in Tealish demographics in 2025. However, in Kooyong the outpolled Labor in 2019 and even in Higgins in 2022 it was possible that Greens could have outpolled Labor in 2022. In 2022, Greens outpolled Labor in Ryan and Brisbane which are Tealish demographically so thats why i said it was possible in the past but Greens did were not seen as pragmatic in 2025 so were not popular.
Zali has had an open HTV every single election since she was elected in 2019, continuing in 2022 and 2025.
the 2pp has increased for the libs 63.49-36.51 representing an increase of 7.36%
@ John are you referrring to Bradfield?
@John the 2PP is quite big at the moment only because there’s so few votes counted in the 2PP measure. 7000 something to 4000 so far. Once more preferences are re-thrown then it should tighten.
ok my mistake
Would like to know when the AEC are going to do this “estimate” of the 2PP in Bradfield. Are they going to update the website with the “estimate”. That could actually be very confusing as most people will assume its a real count. Or will it just be available in the downloads file where it can be stated to be just an estimate…
So many questions… so few answers
All of those questions will be answered in my blog post tomorrow morning.
The AEC has emailed to myself and some others their estimate of 2PP based on the 3CP. I’ll explain it all in a blog post but have had other jobs to do today.
All good Ben.
Love your work!