This is the latest in a string of articles that I write after each Senate election tracking certain themes in the Senate races. Previous volumes in the series were called Senate Reform Performance Review, referring to the 2016 Senate changes that got rid of Group Ticket Voting. I think now we've reached the point after four elections where it's very clear that the new system works very well indeed and needs no longer to be considered on probation, hence the shorter title. For previous instalments see 2016 part one, 2016 part two, 2019 (single article) and 2022 (single article). On the agenda for this issue are: proportionality, winning vote shares (with a focus on the One Nation wins from behind), preferencing impacts, and the curse of Inclusive Gregory. Part two covers Senate 2PP, How to Vote cards, just-voting-1, exhaust, informals, below the lines, and the fun bit about people who we wonder why they bother. And yes that includes the ACT Liberals!
I've decided again to split the article into two because the volume of material this time is a bit much for one go. At least for my own feeling that I'm spending a lot of time on a single article that I haven't released anything from yet.
In this article I treat Labor, Greens and Pocock as comprising the left of the Senate (in relative terms, this should not be taken as me declaring Labor to be an outright left-wing party) and Coalition, One Nation and UAP as the right (with Jacqui Lambie treated as neither though these days there is a case for treating her as left if anything).
Proportionality
The late wins by One Nation were great for proportionality! As usual, firstly here is the national votes to seats conversion (though this is a silly yardstick because the state-based malapportionment and the territories having only two-seat contests):
There were several joint minor right party tickets involving Libertarians, GRPF, Heart and Great Australians. I have lumped these by lead party since obviously nobody else was getting a seat.
Overall on a national basis this one shows Labor, the Greens, Coalition and to a lesser degree One Nation punching above their weight as a result of collecting preferences from parties with too little support to be in the fight anywhere. Legalise Cannabis is hard done by because its support was too even to manage a win anywhere and there were not the same favourable preference sources available to One Nation. On a simplified left-right basis the primaries are something like 56-44 depending on how one treats a few debatable cases and the left has done a little better but not much. Even on this basis this is hardly a disproportionate outcome.
Now, the average of primaries in the six state contests compared to state wins (a better yardstick):
With this thrown in, disregarding the not very classifiable Lambie, there is basically no left-right imbalance; Labor and the Greens combined overperform on average by virtually the same amount as the Coalition and One Nation.
Obviously the right holding only 15/36 state seats will be a big problem for the Coalition should it somehow win the next election and I expect it would be in an even bigger rush to a double dissolution in its first term than it would have been had it won this one.
Had this election been a double dissolution I very roughly estimate the seat results at Labor 31 Coalition 24 Green 11 One Nation 6 and one each for Legalise Cannabis, Rennick, Lambie and Pocock. However Legalise Cannabis were very narrowly behind Labor in NSW and Vic on my projection so might do better. My DD estimates by state are below. (Note that one cannot use
ConcreteSTV when available to reliably simulate DD results because the major parties run out of seats.)
Even in this landslide election, a DD would only give Labor an extra route to passing things that goes through One Nation and others. For Labor the only plausible point in a DD in this term could be to use the joint sitting mechanism to resolve bills blocked by the Coalition and Greens. I'll be surprised if we see that.
I don't see a need to again state in much detail the point that if we still had the Group Ticket Voting nightmare, close to a third of the seats at this election could have been won by unaccountable randoms, and the Greens and One Nation with a combined 17.4% of the vote between them would only be guaranteed one seat in a supposedly proportional system!
Nor do I need to take too long pointing out that under the Senate system that was supposed by Sam Dastyari and co to deliver a permanently blocked Senate, the right (Coalition/One Nation/UAP) now has only 32/76 seats, six short of being able to block things care of Pocock, two JLN wins in Tasmania and left-right 4-2s in WA 2022 and Vic and SA 2025. In fairness even the hack faction of 2016 federal Labor should not have been expected to think their party would ever have a win like this.
Winning Vote Shares
Of the 36 state seats, 22 (down 1) were won on raw quotas. All seven leading candidates (in the top six after surpluses) with 0.7 quotas or more won easily. Everyone with below 0.4 quotas lost. In the range 0.4-0.7 quotas there were ten candidates of whom seven won and three lost. The breakdown here was:
Labor: two wins and three losses (leading in both wins and two of the losses)
One Nation: three wins (leading in one)
Liberal: one win (leading)
JLN: one win (leading)
The two wins were by One Nation, beating Labor after starting with 0.411 Q vs 0.53 Q in WA and, more incredibly, 0.422 quotas vs 0.630 quotas in NSW. They were not by any means low vote shares (5.87% and 6.06%), and indeed were similar to Labor's winning excess in Victoria and higher than Ralph Babet's 4% primary in 2022, though Babet was
also elected fair and square. Which brings me to...
Preferencing Impacts
The big story of this Senate election was the improved performance of One Nation on preferences, even above their strong performances in previous elections. Prior to this election based on the Reps polling I thought One Nation were a serious chance to win in all five mainland states. Labor massively overperformed their Reps polling and One Nation underperformed theirs but One Nation were still able to grab two new seats from behind anyway - in the same two states where they had previously only won off lower vote shares at the 2016 double dissolution.
Four races (NSW, Victoria, South Australia, Western Australia) finished as Labor vs One Nation contests with Labor leading in all but holding on only in Victoria and SA. However these contests varied in structure because the Coalition was short of its second quota in SA (and also in WA though there not if one includes WA Nationals) and because the Greens had further to go to reach quota in NSW.
Here is a chart that shows how these counts fared, albeit in very wonky quota numbers form:
ALP, ON: ALP and ON start quotas
ALP+, ON+, GRN+,L-NP+: quotas gained by each of these in the count (set to 0 if already over quota)
EXH: exhaust
LW, RW: quotas held by left wing and right wing preference sources (RW includes Coalition in NSW and Victoria)
LG: quotas gained in count by Labor + Greens
RG: quotas gained in count by One Nation + Coalition (Coalition loss of surplus ignored as they are treated as a right preference source.
What happened in these contests overall was:
New South Wales: There were slightly more right preferences than left and the gain by One Nation on these was only slightly better than Labor and the Greens combined, but Labor and the Greens were splitting votes two ways almost all the way, which made it easier for One Nation to catch up. On 2022 preference flows One Nation wouldn't have got nearly so much as the Greens and Labor combined but this time the flows from minor right parties to them in this three-way split were much stronger and they did.
One of the factors here is the Libertarian/HEART/GRPF ticket, which replaced the Liberal Democrats. In 2022 Liberal Democrats preferences split 32.8-16.5-29.6 Labor/Greens/One Nation. In 2025 the Libertarian/HEART/GRPF ticket headed by Craig Kelly did far better than the Libertarian tickets outside NSW (which crashed and burned losing more than half their 2022 vote) and its preference split was radically different: 6.8-6.6-75.0. Similar flows were seen in other states with GRPF tickets, but Libertarian tickets were generally not quite that strong. There has been a lot of talk about the weak flow off Legalise Cannabis to Labor at the end (partly a result of weaker Legalise Cannabis preferences in the 3CP split compared to 2022 and more strongly as a result of other votes that reached Legalise Cannabis) but by the time Legalise Cannabis were excluded, even a repeat of 2022 flows off them would not have quite saved Labor.
In NSW the ALP primary was incredibly high, with a swing to Labor of 7.2% although the Reps primary vote swing was only 1.8% and even the Reps 2PP swing was only 4.3%. There were Senate primary vote swings to Labor in the low double digits in the teal vs Coalition seats. However, on a 3CP split with the Greens and One Nation, Labor's performance relative to One Nation weakened off all eleven parties that recontested from 2022. Labor was also underwater on this split off all four new parties added in 2025, whereas in 2022 it had benefited from four of the eight that did not recontest. It is as if Labor so succeeded in maxing out its primary vote that there was nobody left to get preferences from. (Here I have treated UAP as the same as Trumpet of Patriots, and Liberal Democrats and Libertarians the same).
Victoria: The difference between right wing and left wing preference sources was enough to beat Labor if the flow to One Nation was strong enough given that the Greens were soaking up votes, but the flow to One Nation was too weak; indeed the combined left parties held their side's votes as preferences much better than One Nation did.
Western Australia: Here the Greens and Liberals were almost equally short of quota and there were a lot more right-wing than left-wing preferences; the question being would the flow be strong enough for One Nation to close the gap. It was, and enough spare to deal with the Liberals being slightly shorter of quota than the Greens. A key shift here was the Australian Christians' preferences, which were far more favourable to One Nation than in 2022. However, there were also shifts in the opposite direction with Trumpet of Patriots and especially Great Australians being less strong for One Nation than last time. There were also new players like Gerard Rennick People First who were very strong for One Nation, so it's not clear that shifts in preferencing behaviour among recontesting parties alone caused this result.
South Australia: Again the Greens and Coalition were about equally short of quota. The preference sources were more right wing than left wing but not massively. The right didn't gain that much on preferences and the primary vote gap was too large for One Nation with the Greens close to quota.
The other interesting preferencing contest was in Tasmania, with a three-way race between the Liberals, Jacqui Lambie and Labor for the final seat. The Liberals started well ahead of the others. On 2022 preference flows Lambie was expected to overtake the Liberals who would have a relatively narrow win over Labor. Lambie performed accordingly despite some expectations she might not, but the Liberals won far more easily over Labor than projected (more than 0.2 of a quota) largely because the One Nation preferences benefited the Liberals far more in Lambie/Liberal/Labor split than in 2022.
Inclusive Gregory Must Go!
Australia's Senate voting system is excellent and the counting system is mostly excellent but has a relic defect called unweighted Inclusive Gregory distortion that will someday elect a Senator who is not the voter's choice. Parliament needs to stop this before it happens and not after! For more on this problem that has been known for many years see Antony Green here.
Under the system for distributing surplus votes in Senate elections, each vote that has reached a candidate who goes over quota during the count is given the same value going out, irrespective of the value it had going in. In NSW votes that flowed through the surpluses of both Andrew Bragg and Jessica Collins had a value going forwards of about 0.022, meaning nearly all their value had been used electing Liberals. There were 1.45 million of these votes. About 90,000 of these that reached Mehreen Faruqi (Green) were given the same weight in her surplus as votes that had to that point never reached anyone. This meant that the Coalition surplus votes that had contributed about 0.26% of the value of Faruqi's total suddenly became worth over 10% of her 0.069 quota surplus and actually more than doubled in ongoing value! These votes would have flowed more weakly to Labor than Faruqi's own votes and help explain why the flow off Faruqi's surplus was as weak as it was - though a better system would not have stopped One Nation winning. When I'm in a position to quantify the exact impact of this distortion on the NSW and also WA counts I will add a note on it here.
The Inclusive Gregory system for surplus transfers must be replaced with Weighted Inclusive Gregory before the next election. It is way past time. I don't want to ever have to say "we told you so" about this one.
I expect to post part two in the next few days. In advance I can say that voters following the Coalition's how to vote cards did not cause One Nation's win in either NSW or WA, although contributing in both cases.
Hi Kevin, any chance you could point me to a reminder of what Dastyaryi's proposed Senate voting system was? Was recently reminded of Labor's Stephen Conroy trying to butt heads with Antony Green in a Senate hearing on the same issue and went to get familiar with all of Labor's howlers on the topic.
ReplyDeleteDon't believe they had any alternative; they wanted to keep Group Ticket Voting. I kept a list of wrong predictions they made during the 24-hour nonsense debate in the Senate here https://drive.google.com/file/d/0B2jzYRL4laUYN1lMMzFnVkcxQ28/view?resourcekey=0-h1GvKyk5Zvi6cIODZljTJw
DeleteAh my mistake - misread the line about Dastaryi, thought I saw a "proposed" but it was a "supposed". Thanks for the link.
Delete