Also in doing 2PP pendulums one finds out things - such as that the Coalition is in even bigger trouble for the next election than the scale of the 2PP disaster makes obvious. The inflated swings to Labor in marginal seats at this election have created a skewed pendulum where Labor could lose the 2PP and still win a majority.
At this election claims of the demise of 2PP swing as a predictive tool were even harder to get away from than in 2022 ... and even less correct! The overwhelming story of the election was the 13 classic seats that switched from the Coalition (ignoring defections) to Labor. The six seats switching from a major party to a non-major candidate or vice versa were a sideshow, especially as for totals purposes two of them cancelled out. There is a lot of hype about how "no seat is safe any more" but for all of that no safe seat held by a major party fell and the only 2CP-safe seat that fell at all was a Greens seat (Griffith) that was clearly marginal on a three-candidate basis. And the odd 2CP-safe seat falling is nothing new.
Nonetheless in this day and age the story of what seats are marginal and how they are marginal is getting more and more complicated. A decade or two ago preparing a pendulum that noted the most relevant non-classic stuff was twenty minutes' work to do a cut and paste from the official results and annotate the few edge cases. This year I've put notes on 33 of the 137 seats won by major parties, a massive jump from 16 out of 135 in 2022. I've also added in a Coalition vs tealoids side pendulum with some comments about such an endeavour.
The following are the conventions I use in the main pendulum:
1. If a seat was won by a major party it is shown on that major party's side and primarily by its 2PP margin, even if somebody else made the final two. The margin vs the 2CP loser is then noted in brackets.
2. If a seat was won by a non-major candidate it is excluded from the main section of the pendulum and appears below the Opposition seats. The 2PP winner and margin is shown in brackets.
3. Three-candidate margins are shown in the following categories, designed to shed light on all the brave new forms of marginality we have to think about these days:
* 3CP: Where a 3CP swing of 6% or less from the seat winner to some other force would be expected or known to result in that other force winning the seat instead.
* L3CP (losing 3CP): Where a 3CP swing of 6% or less from the 2CP loser to some other force would be expected or known to result in a different 2CP loser who would probably lose the 2CP by less than 6%.
* S3CP (survival 3CP): Where a 3CP swing of 6% or less from the 2CP winner to some other force would result in that other force making the 2CP but being expected to lose by less than 6%
* E3CP (escape 3CP): Where a 3CP swing of 6% or less from the 2CP loser to some other force would result in that other force making the 2CP and being likely to win the seat. This means that the seat winner appears to have been lucky that the 2CP opponent they actually faced was not the one who could have beaten them.
* K3CP (knockout 3CP): Where a 3CP swing of 6% or less from the 2CP winner to another force would knock out the winner by them failing to make the 2CP, but the replacement in the 2CP would not be competitive.
(There's also the case where a 3CP swing within the marginal range would replace a competitive 2CP loser with an uncompetitive one, but I haven't annotated those
4. In the case of Bradfield, for this article I use my own 2PP estimate of 50.64 to Labor, pending any actual 2PP number the AEC will I hope sooner or later produce. The AEC was forced to use an impartial but inaccurate estimation method to finalise the 2025 Tally Room for archiving without disturbing the ballot papers ahead of a possible court challenge, but I expect Labor did substantially better than their method. My method is based on the national swing in independent preferences to Labor (after removing Bradfield from both 2022 and 2025 flows as best I can). Ben Raue gets 51.1 to Labor based on the average flow change in Wentworth, Warringah and Mackellar which is another reasonable method.
Here 'tis then (Click and open image in new tab if needed for larger clearer version). (May still contain some typos or other errors, I'm posting a new version as they are spotted).
The seats of Forrest and Grey have attracted attention as seats where an independent narrowly missed the final two and might have won had they made the final two, suggesting that these indies may have been beaten Condorcet winners. (Labor were beaten Condorcet winners in Ryan.) During the Grey postcount I received some detailed impressions from a scrutineer whose view was that Anita Kuss would have won the 2CP against the Liberals. Ben Raue's estimates have the independent winning the 2CP for both seats, albeit very narrowly. I have therefore marked both seats as E3CP.
Pending any scrutineering information I have also so marked a third that has received less attention: Bullwinkel. Labor barely won this three-cornered seat against the Liberals, and Liberal to National preference flows are frequently much stronger than National to Liberal. In Bullwinkel, the Liberals received 80.7% of 1 Nationals preferences, and got 74.9% of 3CP preferences on the Nationals exclusion. The Nationals' Mia Davies would have needed 80.8% of preferences on the Liberal exclusion to win the 2CP vs Labor if the Liberals were excluded first. I suspect this could have happened and that Labor were lucky to face the Liberals and not Davies in the final two.
Calare saw two independents finish in the top three. At the 3CP stage Kate Hook would have needed 66% of Andrew Gee's distribution to win the 2CP against the Nationals. The 3CP flow from Hook to Gee was 80.8% but I've assumed it would have been much weaker the other way around, with Gee being a defecting former National.
One of the uses of a pendulum is looking at what numbers, with a uniform swing and no change in the non-classic seats, the loser would have needed for a different result. Here Labor won 94 seats to the Coalition's 43, off a 2PP (using my Bradfield estimate) of 55.26 to ALP. On this basis:
* Labor loses its majority at its 19th least safe seat, Whitlam (6.25% swing, 49.01% 2PP by uniform swing, so Coalition needs 50.99%)
* The Coalition becomes the largest party on gaining Labor's 26th least safe seat, Braddon (7.2% swing, 48.06% 2PP, Coalition needs 51.94%)
* The Coalition wins a majority on gaining Labor's 33rd seat, Brisbane (8.96% swing, 46.3% 2PP, Coalition needs 53.7%).
The comparable figures for Labor 2PPs in the 2025 pre-election pendulum, treating Aston as on an average of its 2022 and by-election results, were 51.21, 48.78 and 46.13. The Coalition's two-party vote in 2025, abysmal as it was, was even less efficiently distributed in 2025 than it was in 2022. With a uniform swing from the actual results Labor could actually have lost the 2PP and still retained majority government! In fact the threshhold for that to occur turned out to be more than two points lower than it appeared to be, and on that basis Labor has won this election by even more than the two-party thumping suggests. There are similarities with Victoria 2022 in terms of just how bad the resulting pendulum is for the opposition.
While this is just the post-election pendulum and redistributions will have an impact, I suspect the overall pattern of the pendulum strongly favouring Labor will still be with us, though the Coalition can counter that if it does a better job of appealling to a wider range of voter types across a wider range of seats.
Teals Side Pendulum!
There are now so many Coalition vs tealish Independent 2CP seats that it's possible to construct a side-pendulum including just these seats and to look at how swings on this side-pendulum behave. Ten seats that finished as Coalition vs such an independent in 2022 did so again in 2025 (in all ten cases the same independent; here I exclude Calare). Another three joined them. Here's the post-election side pendulum showing these seats.
This oversimplifies the Coalition vs teals story because of another eight seats where such independents would have been marginal on 2CP but didn't make the final two. But it will be interesting to watch if this continues to be a common 2CP pairing.
What's interesting here is that this side-pendulum did behave so much like an actual pendulum in 2025. The effective 2CP swing in the ten repeat pairing seats was about 0.2% from Coalition to tealoids, using Ben Raue's post-redistribution estimates. For whatever reasons - possibly including the impact of redistributions - there was not a Coalition to tealoids swing. The expected outcome was no change and this is what happened in terms of the total, although each side very narrowly gained one seat from the other (pending any challenge in Bradfield). The low standard deviation on the swings (2.2%) is interesting. There was only one with a swing as large as 4%, but unluckily for the teal side that was just enough to account for Goldstein. The challenge with a Coalition vs tealoids pendulum is how do we find a method of projecting what is going on in these seats from national primary vote figures? I would have expected with such a terrible national primary vote the Coalition would have struggled in these seats but it was not so much the case.
One can do the same thing for Labor vs Greens seats as well but it's not so informative as only a few seats on that one are close (most of the close Labor vs Greens contests were the Brisbane seats that depended on 3CP exclusions). In the six repeat Labor vs Greens 2CP contests the average swing was 3.3% to Labor but with a very high standard deviation of 4.9%. In the same election the Greens lost Melbourne on 6.9% and came pretty close to winning Wills on 4.6% to Labor.
This is one of a large number of goodies I'll be rolling out from the 2025 Reps election results as time permits.
