 |
| Example of a 2025 election whinge meme seen on social media |
In the unhinging that has followed Labor's massive victory in the 2025 federal election, there has been a lot of scapegoating of preferential voting. Some of this may be because the landslide seat result was unexpected. In polls this mostly looked like a close election in terms of whether Labor could get a majority or not. Many voices in the media made it worse by claiming Labor definitely or very probably would not get a majority, and continuing to claim it after the polls (such as they were) no longer supported that view.
Labor won 94/150 (62.67%) of seats with a primary vote of 34.56%. Many people are saying this was caused by preferential voting. In fact, it mostly wasn't. This article explains how this 28.11% gap between Labor's seat share and their vote share was mostly caused by other factors. I find it deeply unfortunate and concerning that many people are in response attacking our very fair voting system and supporting instead the pointless abomination that is first past the post without bothering to understand the arguments in favour of preferences and the extent to which the result was caused by other things. If they really care about parties getting vote shares that match their seat shares, they should support multi-member electorates.
What I am going to do here is start from the assumption that Labor should have got about the same seat share as its party vote share and explain, factor by factor, why that didn't occur. As I go I'll build up the share of the seat total Labor is expected to get based on each factor I add into the mix, and show that most of the gap can be explained without needing to consider Labor doing better on preferences. This is a deliberately mathsy look at the nuts and bolts of why Labor won so many seats off such a modest primary vote and has been graded Wonk Factor 4/5.
There's Always Been A Gap
Before I start that, I should mention that while the gap between Labor's primary vote share and their seat share gap was unusually large in 2025, in recent decades there has always been a gap - the winner always gets a higher share of the seat tally than they have of the primary vote. The last time this was close was for the Whitlam government in 1972 and 1974. From 1972 to 2022 the average gap was 12%. The second and third largest gaps were for the Coalition's massive wins in 1975 and 1977, followed closely by Labor's win in 2022.
1. Single Member Seats Disadvantage Non-Majors
At the 2025 federal election, candidates from outside the major parties polled a total primary vote of 33.62% (a record high) but won only 13/150 seats (8.67%). This disadvantage is not caused by preferences, which in fact help non-major candidates to beat major party candidates way more often than the other way around (ie if we had first past the post and roughly similar vote shares, the non-majors would do even worse in seat terms, barring massive tactical stand-aside pacts)
It's caused by it being hard for a party with a small share of the overall vote to get enough votes in any particular seat to be able to get 50% after preferences in that seat. In general a candidate needs at least a 20% primary vote (often much more) to have any chance of winning at all, and minor parties can poll several percent nationwide while hitting that range in very few if any seats. As a result, while there have been 31 cases since 1972 of a minor non-Coalition party getting more than 4% of the primary vote, in only one such case (Greens 2022 winning four seats) has such a party won more than one seat, in six cases a single seat and in 24 cases nothing! Non-majors together have always got a seat share that is only a very small portion of their vote share in the House of Reps, and prior to 1990 (the start point of the graph below) they often won no seats at all.
The picture is somewhat brighter for teal and similar independents than for minor parties as their primary vote support tends to be very concentrated in a small number of seats, meaning that they can get somewhere near to a proportional seat result (and they were hugely unlucky not to get more seats in 2025). But still, the disadvantage for non-majors alone has become a bigger factor in recent years in major parties overperforming on seat share - as the non-major vote increases, the pool of votes that ends up with the majors after the elimination of non-majors is rising.
Of the two major parties, Labor led by 2.74%. If we imagine preferences splitting completely evenly (and with no leakage on Coalition to Coalition transfers) that would have been a 51.37% 2PP, which as a proportion of the seats won by non-majors would have given Labor 46.92% of the seats (about 70 seats). And that's already explained close to half the gap!
2. The Leading Major Tends To Overperform On Seats
Single-member-per-seat systems are not designed to be proportional, even between the two leading parties. They're designed to elect local representatives, and it has happened over time that those local representatives have tended to organise themselves into parties or coalitions, with one side or the other nearly always getting a majority. If every seat in a single member system voted the same way, the same party would win every single seat, unlike in proportional systems where the parties would win seats in close proportion to their support.
As it happens there is a lot of variation in voter behaviour between seats in single-member systems, which is nearly always enough to see more than one party win a substantial number of seats, unless the vote share result is extremely lopsided . But this variation generally doesn't result in the loser winning a proportional share of seats either. In essentially two-party first past the post systems this pattern is often known as the "cube rule" - the ratio of the winners' seat share to the loser's is often something like the ratio of the winner's vote share to the loser's vote share, cubed. So if party A beats party B 52-48 in a two-party contest (a ratio of 1.083), the cube rule predicts the seat ratio will be 1.083^3=1.271, meaning party A gets about 56% of the seats. Various things can mess with this rule in parts of the world, including personal vote effects, campaign tactics, gerrymandering etc (some parts of the US use gerrymandering not to cheat but to try to stop heavy minorities from going unrepresented) but on the whole something like this is common.
In Australia, something similar to the cube rule operates (the cube of the ratio of the 2PPs is a decent predictor of the ratio of major party seats won, with occasional hiccups such as 1998) but no need to graph the cube rule when a linear relationship has about the same predictive power.
This graph predicts that if Labor won 51.37% of the 2PP, which it would have done had preferences done nothing, Labor would have won about 54.4% of the major party seats. That puts Labor up to 49.68% of seats, or 74-75 seats, almost a majority.
3. Labor's Vote Was Way Better Distributed
There's a big gap, however, between the 74-75 seats expected above and the number Labor actually led on primary votes in, which was a massive 86 seats (though that included two where a Labor candidate led two Coalition candidates individually but not combined). This number 86 is the fundamental thing those complaining about preferences do not get. Winning from behind on preferences added only a net eight seats (or if counting Bendigo and Bullwinkel ten) to the seats where Labor was leading on primaries already, and would have won even with an even preference split. So Labor would have won a large majority on the votes cast even had preferences done nothing, or if we had first past the post and everyone voted the same. It seems quite amazing that Labor candidates led on primaries in 86 seats (to Coalition 57 others 7) when the primary vote gap between the major parties was so small.
The answer is that Labor's primary vote was much, much more favourably distributed than the Coalition's. One measure of this is the median. While the national total primary vote gap between the majors was 2.74%, the median gap was 5.22%. Labor's median primary was a pretty healthy 36.84%. A small part of this is that the mean primary vote gap per seat was actually 3.16%, as a result of Labor doing better in seats with smaller formal primary vote totals (such as the Tasmanian and NT seats and also Western Sydney seats with high informal rates). The larger part of it is that Labor just didn't waste as many primary votes in seats it was uncompetitive in. Labor had ten seats where its primary vote share was under 15%. All of these were Coalition vs Independent seats, seven of which the independents won (though Labor even won the indicative 2PP in two of those). The Coalition had only four such seats, two of them also won by independents.
Comparing Labor's worst 50 seats on primary votes with the Coalition's, Labor averaged 0.36% worse on primaries in these than the Coalition did in its worst 50. In Labor's best 50, Labor averaged 4.77% better than the Coalition's best 50. In the middle 50, Labor averaged 5.06% better.
The table below shows the two sides by primary vote bands (where the Coalition ran two candidates I have over-generously added them together).
An especially striking comparison occurs at the 30-35% range. Labor polled below 35% primary in just 67 seats, the Coalition in 92, meaning that Labor had an 83-58 lead in terms of how many seats there were where its primary vote was over 35%.
The focus on Labor's modest primary vote and overall small primary vote lead ignores the fact that in the middle ground seats Labor was beating the Coalition convincingly on primary votes, even though it would have still won the election easily had it only broken even in them. Labor's national primary was probably deflated by approaching 2% by competition from teals and tealoids in seats Labor mostly wasn't winning anyway. There is a myth that increased competition from right-wing minors also splintered the vote on the right causing the Coalition to lose the national primary vote, but in fact the right-wing minors had a primary vote swing against them too (albeit only about 0.5% depending on who you count), and had a lower combined primary vote than left-wing minors.
Factors playing into Labor's superior primary vote performance in the mid-pack seats would include personal vote effects (which seem to have been unusually strong this year). In seats in which Labor first term incumbents defeated Coalition incumbents in 2022 especially, Labor would have gained a personal vote advantage. But it's also about the Coalition's failed strategy in which its messaging was aimed too narrowly at outer suburbia, where it didn't do all that well anyway.
Even after allowing the Coalition to count Bendigo and Bullwinkel as primary vote leads (based on two candidates against one), Labor's better distributed primary vote still put it in the outright lead before preferences in 84 seats (56% of seats). So we've now got from 34.56% of the vote to 56% of the seats even assuming that preferences split evenly.
4. Wins On Preferences That Would Have Been Wins Anyway
As often noted here before, it's clueless to assume that if Australia had had first past the post at any election then all the voters would have voted the same way. In particular, if the Greens are not competitive in a given seat, then a lot of Greens voters under first past the post would be likely to vote tactically for Labor - an issue that suppresses the overall Greens vote in first past the post systems (though they do still win seats in some of them). Not all would do so - but the proportion who would is probably higher than the same for One Nation voters, who are more likely to dislike both majors and hence to grab the opportunity to waste their vote by voting One Nation under FPTP. If we had FPTP, Labor and the Greens might even make stand-aside deals to reduce the number of seats where they wasted votes by competing with each other.
The only seat where Labor led on primaries but lost was Fowler, which I think they also would have lost under FPTP.
There were eight seats where Labor trailed the lead Coalition candidate on primaries but won - Aston, Dickson, Petrie, Brisbane, Banks, Solomon, Deakin, Menzies. They also trailed Adam Bandt (Greens) in Melbourne and won that. Of these, in Aston and Dickson at the very least the Coalition's lead was small compared to the 2PP margin; I have no doubt Labor would have won these under FPTP. I also think Banks and Petrie would be rather promising. Solomon, Deakin and Menzies seem more difficult unless the Greens decided not to run. In Brisbane the LNP were thumped on 2PP; the only way they would have won under FPTP is if both Labor and Greens voters refused to vote strategically and thereby split the vote, that said, this does sometimes happen in FPTP which is another reason why FPTP is garbage. In Melbourne, the Greens had a large lead and were beaten on mainly Liberal preferences; the question under FPTP would be whether enough Liberal voters were willing to vote tactically to defeat Bandt, maybe not. I'm assuming the Coalition parties would be cute enough for one to step aside in Bendigo; Bullwinkel is less certain with the Nationals not part of the Coalition in WA and thinking that they were in the mix.
I'd therefore put Labor on at very least 87 seats (58% of seats) under FPTP. The maximum share of Labor's victory that I think can be put down to preferential voting is therefore 7 seats (4.67%), which is about a sixth of the gap between Labor's primary vote share and Labor's seat share.
Labor getting a seat share way higher than their vote share was therefore mostly not caused by preferential voting. And we can see that the same thing happens in FPTP systems. In the UK 2024 Labour (albeit with a much larger primary vote lead over their main rival) got 411/650 seats (63.2%) off 33.7% of the vote. Leading parties winning a far higher vote share than their share of the primary vote is a general property of single-member seat systems. It is silly to blame preferences for that, and it needs to stop.
