Week zero

The Coalition prepares to choose or confirm leaders, Section 44 rears its head once again, and a look at the aggregate two-party preferred numbers.

To allow for a separate thread for the late election counting, which can be found here, here goes my first post-election summary of relevant news to kick off a general discussion thread. Which is naturally less easy to do now that there are no polls or election horse race scuttlebutt to relate. Here’s what I’ve managed:

• Peter Dutton now appears certain to be elected unopposed as the new Liberal leader at the first meeting of the party room after the election winners have been confirmed. There appears to be strong support for the notion that the deputy position should go to a woman, names mentioned including Karen Andrews, Bridget Archer, Sussan Ley, Anne Ruston and Jane Hume. There were some suggestions that Andrews might seek the leadership, together with Dan Tehan, although it always seemed clear Dutton had the numbers.

• The Nationals party room will meet on Tuesday, which could see a challenge to Barnaby Joyce’s leadership from David Littleproud or Michael McCormack. However, the ABC reports it has been put to McCormack that it would be preferable to have a “fresh start”. Mike Foley of the Age/Herald reports Keith Pitt might put his name forward on the “off chance” that Joyce declines to stand, positioning himself as the heir to Joyce’s skepticism on net zero carbon emissions.

• Following her win over Labor’s Kristina Keneally as an independent for Fowler, it has been noted that Dai Le asserted on her Section 44 checklist as part of her nomination for the election that she had never been a citizen or subject of a country other than Australia. Queried by The Australian, constitutional law expert Anne Twomey offered the inuitively obvious point that this seemed unlikely given she was born in Vietnam in 1968 and remained there until her family fled in 1975. However, while a nomination may be rejected if a prospective candidate does not complete the checklist and provide supporting documentation is required, it would not appear a nomination is retrospectively invalidated if the information provided was later shown to be incomplete. The sole point at issue is whether Le does in fact have Vietnamese citizenship, which would appear unlikely based on the account of Sydney barrister Dominic Villa.

• The projections of both the ABC and myself are that Labor will win the final two-party preferred count by 51.8-48.2, from a swing to Labor of 3.3%. This is derived from two-candidate preferred counts between Labor and the Coalition in seats where one is available and estimates of other parties’ preference flows where they are not. I have Labor winning by 51.3-48.7 in New South Wales, a swing of 3.0%; 53.9-46.1 in Victoria, a swing of 0.8%; the Coalition winning 54.3-45.7 in Queensland, a swing to Labor of 4.1%; Labor winning 54.7-45.3 in Western Australia, a swing of 10.2% (their first win in the state since 1987 and best result since 1983); 53.9-46.1 in South Australia, a swing of 3.2%; and 53.8-46.2 in Tasmania, a swing to the Coalition of 2.1%.

Late counting: rolling coverage

Progressively updated commentary on late counting of the results from the 2022 federal election.

Click here for full federal election results updated live.

Saturday, May 28

As I should have noted yesterday, the AEC has published all-important three-candidate preferred counts for Brisbane and Macnamara, where the results hinge on who will finish second and third out of the Labor and the Greens. With “88% of the total expected ballot papers” accounted for in Brisbane, the Greens lead Labor by 29.55% to 28.56% (with the LNP on 41.89%), which is sufficient for the ABC to have called the result.

Macnamara on the other hand is exquisitely close three ways, with the Liberal candidate on 33.56%, Labor member Josh Burns on 33.50% and Greens candidate Steph Hodgins-May on 32.93%. This suggests 46.9% of minor candidate preferences are going to Liberal, 35.9% are going to the Greens and only 17.2% are going to Labor. For the Greens to win, two things need to happen after the remaining 6500 or so outstanding votes are counted: they need to close their 495 vote deficit against Labor, and the Liberals need to not fall to third.

The first of these seems entirely possible. If the outstanding batches behave like those already counted, they should make up around 200 out of 3300 pre-polls and 100 out of 1100 declaration pre-polls. If the 300 or so provisionals behave like they did in 2019, they should make up a further 40 or so. Then there’s the COVID votes, of which there are about 1000, and which are apparently also expected to favour the Greens. Postals are diminishing in number, but each batch continues to be better for the Greens than the last, to the extent that today’s was the first on which they gained on Labor.

As for the Liberals, the 3CP count has them 545 ahead of the Greens and 50 ahead of Labor. There seems no particular reason to think they will either gain or lose in large degree relative to Labor out of any of the remaining vote types, so everything that was just said about the Greens relative to Labor applies to the Greens relative to Liberal as well. The question is whether the chips just happen to fall in such a way that Labor gains 50 votes or so at their expense. I am flying blind here with respect to the COVID votes, and also with how many postals can be expected – postals can arrive until Friday, and I can’t really tell you how many tend to trickle in in the second week.

All of this amounts to bad news for Labor in its quest for 76 seats, with Brisbane now out of the picture and the odds most likely leaning against them in Macnamara, which I for one thought they had in the bag earlier in the week, and which my results system is continuing to call as Labor retain due to its inability to think in three-party terms. The likely retention of Lyons only gets them to 75: they now need a late break in their favour in Gilmore or, less likely, Deakin. A great deal hinges on the absents, declaration pre-polls and COVID votes in these seats, on which we remain none the wiser, with no progress in either count today.

Friday, May 27

My results system is now registering Ryan as a Greens gain from the LNP, as the fresh two-candidate count finally advances enough to tip the probability dial over 99%. Similarly, Wannon has been restored to the Liberals after further booths were added in the Liberal-versus-independent count that was begun yesterday, and Nicholls is now being called for the Nationals as Rob Priestly’s independent bid falls short.

Labor aren’t dead yet in Deakin, where rechecking today gained them 138 on postals and 37 on ordinary votes, while costing the Liberals 169 and 30. However, there was little in it in today’s batches of absents, which broke 236-232 to Labor, and declaration pre-polls, which broke 463-462 to Liberal. Labor will perhaps need about 55% of what’s outstanding to reel in a Liberal lead that today shrank from 1032 to 655.

The contention related yesterday by Antony Green that late counting would favour the Greens in Macnamara was borne out in that the latest batch of 1951 postals were much stronger for them than earlier batches and they also performed well on the first 965 declaration pre-polls, particularly relative to Labor. Their situation will apparently continue to improve from here, but the secret of the final result remains hidden in the preference flows, on which I can offer no hard information.

It seems rechecking of ordinary and postal votes turned up 157 extra ordinary votes for the Liberals in Lyons, but little else. The first absents from the seat broke 696-621 to Labor. That leaves Labor’s lead at 678, down from 784 yesterday. For the second day in a row, the only progress in Gilmore was rechecking, which cost the Liberals 41 votes and Labor five. Counting will continue over the weekend.

Thursday, May 26

The ABC is calling Lyons for Labor after the correction of errors gave Labor a 582-vote fillip on ordinary votes. My system isn’t quite there yet though, in part because each batch of postals so far has been better for the Liberals than the last, the latest breaking 1508 (56.1%) to 1178 (43.9%). However, it may also be because the ABC has formed a more considered view than I have as to how many outstanding votes remain.

I have been assuming for some time that Labor will win Macnamara, which together with Lyons would put Labor over the line to 76 seats and a majority. However, Antony Green has dropped by in comments with an account of his own decision to hold off on such a call, informed by Labor member Josh Burns’ own lack of confidence. Specifically, both Labor and the Greens believe the Greens will enjoy a surge when absents and “an estimated 1000 COVID votes” are added to the count. On top of anything else, this is the first intelligence I have received as to how many COVID votes might be expected, here or anywhere else.

The very first absent and declaration pre-polls were added to the count today, mostly in Deakin, which respectively got 455 and 449 (there were also 787 absent votes added in Lingiari). The absents in Deakin broke 251-204 to Labor, but the pre-polls went 249-200 against. However, there was a turn in Labor’s favour on postals, which broke only 1493 (51.1%) to 1427 (48.9%) in the Liberals’ favour, compared with 58.8% to 41.2% on the previous batches. With the Liberal lead at 1032, Labor will need everything to go right on the remaining postals (perhaps about 3000), outstanding absents and declaration pre-polls (seemingly around 3500 apiece), COVID votes (around 1000, I guess) and provisional votes (a couple of hundred).

I’m not sure of the details, but rechecking of ordinary votes in Gilmore today was to the advantage of Andrew Constance, who gained 25 while Labor lost 149. No new postal votes were added to the count. Another 4095 postals in Brisbane didn’t fundamentally change the situation described here yesterday, with Labor continuing to hold a 0.7% lead over the Greens on the primary vote that I don’t think will be quite enough for them when preferences are distributed. Liberal member Celia Hammond conceded defeat to independent Kate Chaney in Curtin, perhaps because the latest batch of postals sharplhy reversed earlier form in breaking 1955 (52.3%) to 1780 (47.7%) to Chaney.

My system has withdrawn Wannon as a confirmed Liberal retain, not because such a result has become objectively less likely, but because the AEC has concluded Labor will run third and begun a fresh two-candidate count between Liberal member Daniel Tehan and independent Alex Dyson. This has so far accounted for only about 10% of the vote, and is presently giving Tehan a fairly modest lead of 52.2-47.8. Based on the relationship at polling booth level between the Liberal primary vote and the share of preferences, I am expecting this to inflate quite substantially and do not believe Tehan is in trouble.

Wednesday, May 25

My results system today called Dickson for the Coalition, bringing them to 51 seats, to which I think it more than likely that Casey, Menzies, Cowper, Nicholls and Moore will shortly be added. Labor remains on 74, and I don’t think there is any real doubt they will further gain Bennelong. Batches of postal votes are being added to the count in diminish number, but we still haven’t seen any absents or declaration pre-polls, which we can at least make broad guesses about based on past performance, or the COVID-19 electronic assisted voting results, which are anyone’s guess.

Here’s the latest from the seats that may add further to the Labor count, any one of which will get them to a majority:

Brisbane. The result remains at the mercy of unavailable information on how minor candidate preferences are flowing between the Greens, Labor and the LNP. A source familiar with the matter has passed on an informal tally based on observation of pre-poll and postal vote counting that suggests around 50% of preferences are flowing to the LNP, 32% to the Greens and 17% to Labor. If this is accurate, Labor will need a primary vote lead of about 1% when all the votes are in to remain ahead of the Greens during the preference distribution. Currently the gap is 0.7%, having increased today from 34 votes to 528 following a batch of postal votes that was actually weaker for Labor than the previous. However, that’s likely to be sent down rather than up by absent votes, on which Greens do well. So it would appear the Greens remain favourites.

Gilmore. Labor had a better batch of postals today, which only favoured the Liberals by 754 (51.3%) to 715 (48.7%) compared with 5895 (54.5%) to 4913 (45.5%) from the previous batches. With further unfavourable adjustments from ordinary vote rechecking, that increased Andrew Constance’s lead only from 104 to 112, with postals now set to slow to a trickle. This isn’t quite the lead Andrew Constance would have wanted ahead of what’s likely to be a Labor gain when absents are added.

Lyons. There seems to be an improving trend here for the Liberals on postals, the latest batch of which broke 1103 (55.6%) to 882 (44.4%) their way compared with 2966 (50.9%) to 2857 (49.1%) against them previously. If the remaining postals break the same as this latest batch, there’s going to be next to nothing in it.


Curtin. Liberal member Celia Hammond has brought Kate Chaney’s lead on the raw count inside 1%, but today’s postals were actually a bit weaker for her than previous batches, favouring her by 1075 (55.3%) to 868 (44.7%) compared with 6729 (59.4%) to 4605 (40.6%) previously. They were also notable fewer than number, and have still left her 1640 behind. As noted here yesterday, the 2019 result suggests absents are unlikely to favour her. The dynamic may be a little different this time given the redistribution has pushed the boundary substantially northwards, and many absent votes are cast just outside an electorate’s boundaries, though I can’t specifically think why this would make them a whole lot more favourable to the Liberals.

Federal election live: day five

The three seats that might potentially get Labor over the line to a majority remain up in the air, as more distant prospects for them fade further from view.

Click here for full federal election results updated live.

My system today called Bass and Wannon for the Liberals and Wentworth for Allegra Spender, the latter being the first gain called for the teal independents, although I don’t doubt there will be four and probably five to follow. Postals continue to be added in large numbers, although they will start to diminish henceforth. As noted below, one of the biggest developments today arose from rechecking. Tomorrow we are apparently see numbers from electronic-assisted telephone voting added, which is exciting because I have absolutely no idea about their partisan tendency and how many there will be.

The latest from the three seats that could potentially push Labor over the line to a majority:

Brisbane. Kevin Bonham’s post-count post suggests the AEC is conducting an unusual indicative three-candidate preferred count to determine which out of Labor and the Greens will drop out and deliver the seat to the other. However, I’ve heard no official word on this. Based on the preference distribution in 2019, my earlier assessment was that Labor would need a buffer on the primary vote to hold out against preferences to the Greens from Animal Justice, and even to some extent from the right-wing parties, more of whose preferences went to the Greens than Labor (though a great deal more again went to the LNP). However, as with one or two of my other early assessments, this may have failed to fully account for the substantial increase in postal votes this time, which are being true to form in being weak for the Greens. Labor now leads the Greens on the primary vote, but it will need to further boost the margin if my surmise about preference flows is borne out.

Gilmore. Labor had a very handy boost of 382 votes in rechecking that was mostly down to the Gerringong booth, where the two-candidate figures had been entered the wrong way around. This apparently put Labor in the lead briefly on the raw count, but the Liberals recovered it when a small batch of postals favoured them 701-521, with Andrew Constance currently 104 votes ahead. Postals will no doubt continue to favour Constance, but the bulk of them are now out of the way. Still to come are declaration pre-polls, which should break about evenly; absents, which should boost Labor by maybe 300; provisionals, which should add a couple of dozen for Labor; and electronic-assisted votes, which I continue to have no idea about.

Lyons. This is the first result I’ve looked at where the second batch of postals was observably different from the first, going 1024-910 to Liberal compared with 2966-2857 to Labor. If the outstanding postals break like the latest batch, Labor’s current lead of 703 votes will be cut in half. That makes it very close, but there is no specific reason to expect the other outstanding votes will move the dial in either direction.

Elsewhere, Labor continues to be buried on postals in Deakin, the latest batch breaking 3715-2584 to the Liberals. Yesterday I asserted that outstanding postals should add around 1000 to Michael Sukkar’s lead, but this batch alone adds to 1131. From here Labor will need stronger than anticipated absents and/or declaration pre-polls, and/or for the enigma of electronic assisted voting. I would personally call Menzies for the Liberals now even though my system doesn’t yet have it past the 99% threshold, yesterday’s postals having broken 3715-2584 in their favour.

After a quiet day in Curtin on Monday, a second batch of postals were added that favoured Liberal member Celia Hammond 4464-2950, a similar proportion to the first batch. This suggests the outstanding postals will bite a further 1000 or so out of independent Kate Chaney’s 1842 vote lead. However, the Liberals were relatively weak on absent votes in the seat in 2019, and there’s little reason to think out-of-division pre-polls will be particularly favourable to them.

Federal election live: day four

What now seems a certain Labor win in Bennelong leaves them one short of a majority, with a further three in-doubt seats as candidates to get them over the line.

Click here for full federal election results updated live.

The count failed to progress yesterday in many of the seats I rate as in doubt, but my system yesterday called Lingiari for Labor and Bradfield for the Liberals. It is clear Bennelong won’t be far off, with the second batch of postals reducing the Labor lead at the same insufficient rate as the first. That will leave Labor needing one further seat to get a majority, which might (or might not) be provided by Lyons, Brisbane and Gilmore, on which we are today none the wise.

The fresh two-candidate count in Cowper has dispelled any doubt that Nationals member Pat Conaghan will hold out against independent Caz Heise, whom he leads with 53.2% of the two-candidate vote. I’m projecting that come down to around 52-48 when the two-candidate count has caught up with the primary votes. The fresh count in Ryan records a slight lead for the LNP with about 12% completed, but this is because the booths counted so far lean conservative. My projection of a 2.6% winning margin for the Greens is based on the fact that preferences in the booths added so far are breaking nearly 70-30 in their favour. It is by the same logic that an 11.2% Greens margin over the LNP is projected in Griffith.

New batches of postal votes further shortened the odds on Liberal wins in Deakin, where Michael Sukkar has opened a 55-vote lead; Menzies, where the Liberal lead increased from 624 to 1748; and Sturt, where it increased from 723 to 982. My projection that Labor will ultimately win a squeaker in Deakin fails to properly account for the clear trend on postals, about 40% of which are still to come. That should add around 1000 votes to Sukkar’s margin, only about half of which Labor is likely to recover on absents. I should acknowledge though that I have no idea what the electronic assisted voting results have in store, which will include those in COVID-19 isolation, but my best guess is that they will be few in number.

Federal election live: day three

Ongoing coverage of Labor’s search for a path to 76 seats, which it may or may not reach.

Click here for full federal election results updated live.

There were two developments in counting yesterday, one of which was that batches of postal votes were counted in every seat. Taken in aggregate they recorded a bigger swing to Labor than booth and pre-poll votes, consistent with the notion that the 70% increase in applications would make the postal voter population more representative and less conservative than in past years. However, this doesn’t mean Labor can expect a surge in its favour in late counting, since the swing is not so big as to completely eradicate postal voting’s conservative lean and they are in greater in overall number now.

The other development was that fresh two-candidate preferred counts were commenced in three interesting (Griffith, Ryan and Cowper) and seven uninteresting (Bradfield, Calare, Sydney, Hinkler, Maranoa, Melbourne and Grey) races. None of these counts is very far advanced, and for several it was just a case of throwing to new pairs of candidates during today’s counting of postal votes. However, we can presumably expect them to go back through the ordinary votes and publish fresh two-candidate preferred results in fairly short order.

There are another four seats where it is clear the wrong two candidates were picked for the candidate on the night, but in which new counts have not been commenced since it is not clear which candidate will drop out before the final count, of which I rate one to be very much in doubt (Brisbane) and three not so (Richmond, Macnamara and Wannon). We won’t know exactly what’s happened in these races until all the votes are in and the full distributions of preferences are conducted.

My system is definitively calling 72 seats for Labor, 47 for the Coalition, three for independents, two for the Greens and one for Bob Katter, but there are a number involving independents that it is being too slow to give away, which I should probably do something about. These include Mackellar, North Sydney, Wentworth, Goldstein and Kooyong, where the Liberals can hope for no more from postals than to reduce the teal independents’ winning margins, and Fowler, where Kristina Keneally seemingly can’t even hope for that much. Conversely, the presence of independents in the race is making the system too slow to call Bradfield, Nicholls and Wannon for the Coalition.

The only one of the six main teal independent targets I would still rate in doubt is Curtin, and even there Liberal member Celia Hammond will need to pull a rabbit out of the hat. Postals are favouring her, as they are with the Liberals in all such contests, but they are on track to bite only around 2000 out of Chaney’s existing 3350 vote margin. That would leave her needing some dynamic on absent and out-of-division pre-polls to favour her, the nature of which wouldn’t seem clear at this stage.

Then there’s Cowper, where my system is crediting Nationals incumbent Pat Conaghan with a slight advantage over independent Caz Heise. We’re in the very early stages of a two-candidate count between the two in which the only substantial result is the postals, on which preferences are flowing to Heise 67.3-32.7. Applying that split over the projected primary votes, which have Conaghan on 39.6% and Heise on 26.7%, Conaghan would hold on by a margin of 0.6%, which is closer than the 2.0% being produced by the crude estimate of preference slows in my system.

The system is also being slow to call Ryan for the Greens, but I expect that to resolve when the fresh two-candidate count there reaches a sufficient stage that I stop relying on my preference estimates, which cause me to impose a bigger margin of error. There has been some talk of the Greens making it as high as five, but this includes Macnamara which I now can’t see happening. Labor had a very strong result on the first batch of postals, which swung 9.0% in their favour on the primary vote, making it very unlikely they will drop out ahead of both the Greens and the Liberals, which is what it would take for them to lose. The remaining issue is whether Brisbane gets them to four, on which more below.

I don’t imagine my system is too far off calling Lingiari for Labor and Casey, Dickson and Bass for the Coalition, though I’d keep at least half an eye on the latter. Throwing those on the pile, we get Labor to 74, the Coalition to 54, independents to ten and the Greens to three, with Bob Katter still on one, Cowper to either stay Coalition or go independent, and a further eight outstanding from which Labor might get the two extra they need to make it to 76. As I see it, these are, in roughly descending order of likelihood:

Bennelong. Yesterday’s 5760 postals broke 3131-2629 to Liberal, but this marked a 12.2% swing to Labor compared with 2019 and suggested postals will not give Liberal candidate Simon Kennedy the lift he needs to close what remains a 1749 vote Labor lead. Whatever they Liberals gain on remaining postals seems likely to be approximately matched by advantages to Labor on other types of outstanding vote.

Lyons. Labor’s Brian Mitchell had a very encouraging first batch of postal votes here, breaking 2056-1833 his way. Mitchell leads by 0.6% on the raw count, but I’m projecting this to come out at just 0.1% based on an assumption that the outstanding postals will lean conservative, as they did last time. If further batches of postals put paid to that idea, he can start to breathe easier.

Brisbane. As I noted in yesterday’s post, what we need here not a two but a three-candidate preferred count to establish who will out first out of the Greens and Labor, as the seat will go to whichever survives at this point. While I am projecting the Greens to hold a slight lead on the primary vote, and they should get a fillip in the preference count when Animal Justice are distributed, the postals give Labor more than a shred of hope. Postals are always weak for the Greens, but the first batch has only recorded a 1.3% primary vote swing for them compared with 5.6% overall. If that’s maintained over the rest of the postals, it’s likely to be very close. So unless the AEC does something innovative here, this will have to wait until all the votes are in and the full distribution of preferences is completed.

Gilmore. A similar story to Bennelong insofar as a weaker than anticipated showing for the Liberals on the first batch of postals suggests the final result will come in roughly where it is at the moment. That means lineball in this case, with Liberal member Andrew Constance leading by 306 votes.

Deakin. Here on the other hand postals were favourable to the Liberals, swinging 3.7% to Labor compared with 5.0% overall. While my projection still has Labor 0.5% ahead, if the postal count so far continues over what should be at least 12,000 more yet to come, Michael Sukkar will retain the seat.

Menzies. The postals swung similarly to the overall result here, which is good news for the Liberals because the increased number of them means their natural lean to the Liberals should cause the gap to widen as more can come in, by a greater amount than Labor can hope to reel in on other types of vote.

Sturt. The swing on postals here was in line with the overall result, so I’m satisfied with my projection of a 0.5% Liberal lead, which happens to be very close to the raw count. However, there are enough votes still out there that it can’t be given away yet.

Moore. Labor recorded a below par swing on the first batch of postals, suggesting Liberal member Ian Goodenough’s 1138-vote lead is more likely to widen than shrink, and that Labor will have to make do with four gains in Western Australia rather than five.

Federal election live: day two

An explanation of the Poll Bludger results system’s reading of the situation, and some observations on the likely make-up of the Senate.

Click here for full federal election results updated live.

Sunday updates

4pm. I gather that in seats where the AEC deemed it had picked the wrong candidates for the TCP count, it has picked new candidates and is throwing to those with the postal vote counts it is conducting today. Presumably it will do new TCP counts for the ordinary votes in these seats in due course, but for now the only TCP figures in these seats are for postals.

That at least is the situation in Bradfield, and I’ve rejigged everything for that seat so the page shows whatever results are available from the fresh TCP count rather than the redundant Liberal-versus-Labor one. The independent contender, Nicolette Boele, hasn’t done nearly as well on the postals (13.8%) as the pre-poll (24.6%) and polling booths (23.1%). She has however received nearly three-quarters of the preferences from the postals, which leave her short by around 52-48 if applied to the overall results. However, she will presumably lose further ground as more postals are added.

I’ll now to through all the other seats where I’d have thought this might be happening and will add commentary as I go.

End of election night

A lot more naturally remains to be said, but as was the case with the commentary I offered here last night, I’ll let my results system do the talking and mostly limit myself to explaining what it’s up to. First, I should note that my reading of the national two-party preferred – 52.8-47.2 in favour of Labor – is 1% stronger for Labor than the ABC’s, which is quite a lot closer to Nine’s projection, and I’m not going to pretend you should take my word for it over theirs. I’ll have a closer look at this tomorrow, but given the state of the primary votes and the large number of votes still outstanding, a lot of this is necessarily based on projection.

As you can see at the top of the entry page, I’m currently calling 71 seats for Labor, 46 for the Coalition, two for the Greens, three for independents and one for Katter’s Australian Party. However, this largely reflects the fact that the system is very cautious in making calls in most of the seats where independents and the Greens are looking formidable, for reasons that will be explained shortly.

Bennelong, Gilmore, Deakin, Menzies, Bass, Lyons, Sturt, Lingiari, Moore and — to stretch the elastic a little — Monash, Casey and Dickson are the old-fashioned kind of in doubt, with close races between Labor and Coalition candidates that could go either way on late counting. Projections here are based on comparison of how the votes in so far matched up with the equivalent votes last time, and an assumption that postals, absents and out-of-division pre-polls will record broadly similar swings (there are also a fair few pre-poll voting centres around the place that didn’t complete their counting last night, despite the head start they had for the first time in sorting their envelopes from 4pm).

However, it’s possible that postal votes in particular will behave quite a bit differently this time in swing terms due to a roughly 70% increase in the application rate. If that results in the postal voter pool being more demographically representative, these votes could be less conservative than usual. For the time being though, this is only a hypothesis.

Most of the seats that are listed as in doubt are ones where there is no historic data by which this result can readily be compared with the last, as is the case in the many seats where non-incumbent independents are in the race. My system applies a wider margin of error in these cases, meaning a fairly substantial lead is required before the 99% probability threshold is crossed and the seat is deemed to be called. For example, no human observer doubts that Allegra Spender has defeated Dave Sharma in Wentworth, but my system isn’t all the way there yet.

The AEC has pulled its two-candidate preferred counts in 15 seats where it deems it picked the wrong candidates, although it seems this hasn’t caused the results to disappear from my pages, which I believe is a happy accident. My two-party projections in these seats work off my estimates of how preferences will flow to the candidates deemed likely to finish first and second. There are a few here that warrant explanation or discussion:

Brisbane. While it is clear LNP member Trevor Evans will lose this seat, it is not clear to whom he will lose out of the Labor and the Greens candidates. The AEC conducted a traditional two-candidate count between the major party candidates, but has discontinued it because Labor is running third. However, it’s far from clear that it won’t prove to have had it right the first time after all the votes are in. There would not be much point in a fresh two-candidate count here as it is already clear who would win between the Greens and the LNP. It would be more instructive to determine how many preferences from the minor candidates are going to Labor and how many to the Greens. Most likely we will not be sure of the result here until the full preference distribution is conducted, which can’t happen until all the votes are in.

Griffith. Here too Labor is running third, but in this case it’s the Greens first and LNP second rather than vice-versa. Labor’s Terri Butler could theoretically make it over the line on preferences if she moved in to second, but the flow of One Nation and United Australia Party preferences to the LNP are likely to put paid to that.

Ryan. My projection of a 52.6% Greens two-party vote over the LNP is based on an estimated 85-15 split of Labor preferences, which the ABC seemingly expects to be even wider, because it has it at 53.7-46.3.

Macnamara. Labor, the Greens and the Liberals are very close on the primary vote, and to the extent that the Liberals are slightly behind, I’m projecting them to make up most of the gap on late counting. So any three of them might end up being excluded before the final count. I would have thought Labor would win on either scenario where it clears this hurdle, since Liberal preferences favour them over the Greens fairly solidly when they direct their preferences that way. However, the ABC is projecting a lineball Labor-Greens result, so perhaps I’m wrong. If so, a fresh two-candidate count between Labor and the Greens would be instructive, but it would only apply if the Liberals did indeed go out before either of them. Certainly the Greens will win if it’s Labor that gets excluded.

Cowper. Independent Caz Heise landed well clear of Labor, and with Nationals member Pat Conaghan well short of a majority at 40.4%, my projection is that she will take it right up to him after preferences. We will need a two-candidate count to see exactly how accurate that is, which presumably the AEC will be forthcoming with fairly shortly – perhaps as soon as today.

Bradfield. A similar story here, with yet another independent, Nicolette Boele, outpolling Labor to finish a clear second, while Liberal member Paul Fletcher is on less of the primary vote than he would like. My preference estimates suggest Boele won’t quite get there, but here too we will need a new two-candidate count to see if I am right.

Now for a very quick look at the Senate before I collapse altogether. It seems to me there is a fairly strong possibility of what Labor would regard as a rather happy result where they and the Greens between them have half the numbers, and can get the extra votes needed to pass legislation from two Jacqui Lambie Network Senators or ACT independent David Pocock. Each of the six states seem to be looking at results where the first five seats have gone two Labor, two Coalition and one Greens, with the last seat up for grabs. To deal with the latter situation in turn:

New South Wales. Most likely a third seat for the Coalition in New South Wales, unless right-wing preferences lock in strongly behind One Nation.

Victoria. Probably the United Australia Party unless preferences flow strongly to Legalise Cannabis, a possibility I hope to be able to shed more light on after running an analysis on past ballot paper data.

Queensland. Probably Pauline Hanson but possibly Legalise Cannabis, who have been something of a surprise packet across all Senate races but particularly here at 6.7%. Neither Clive Palmer nor Campbell Newman are in contention.

Western Australia. The strongest possibility would seem to be a third seat for Labor, something it has never managed before in Western Australia and indeed hasn’t managed anywhere at a half-Senate election since 2010. However, I will also investigate the possibility that One Nation and Legalise Cannabis might be in contention instead.

South Australia. Very likely One Nation’s Jennifer Game, whose daughter Sarah Game won the party’s seat in the state Legislative Council in March, and has quickly emerged as something of a surprise packet. Out of contention is Nick Xenophon, whose “Group O” managed only 2.7%.

Tasmania. A second seat for the Jacqui Lambie Network, whose Tammy Terrell is on 8.1%, almost matching the 8.3% the party managed with Lambie at the top of the ticket in 2019.

Then there’s the Australian Capital Territory, where in yet another triumph for teal independence, David Pocock seems assured of unseating conservative Liberal Zed Seselja. The ABC projection says otherwise, but it seems to me that Seselja will assuredly be buried by the flow of preferences to Pocock from the Greens and Kim Rubenstein.

Federal election live

Live commentary and discussion of the election night count for the 2022 federal election.

Click here for full federal election results updated live.

10.39pm. My system’s call of Durack as “ALP ahead” is a puzzle to me. Neither Nine or the ABC thinks so. Perhaps this is because we have different ways of processing an extraordinary result at the large Broome pre-poll voting centre, which looks to me like a 30% swing to Labor. I’ll be interesting to see what happens when it reports its two-candidate preferred numbers.

10.22pm. Now that the mad rush is past, if you’ve been enjoying my live results, perhaps you make care to make a small contribution through the “become a supporter” button at the top of the page. Not that its work is done by any stretch – it will continue to chalk up new results and maintain projections and probability estimates over the fortnight to come.

10.07pm. Having worked out what was wrong with them, I’ve now restored the aggregated national and state two-party preferred projections to the results display, which suggest Labor will win the two-party preferred by around 52-48. This may inflate a bit as the Western Australian count catches up with the rest of the country, but it might also do unpredictable things as pre-polls and postals come in, as indeed may the seat totals.

9.54pm. My system has Labor ahead in 77 seats, but it’s still treating Brisbane as a Labor-versus-LNP contest when there’s a high probability that Labor will actually finish third and the seat will be one of what could be as many as five Greens seats.

8.31pm. The AEC have, as they say in the trade, declared Calare and Cowper “mavericks”, which is to say acknowledged they have picked the wrong candidates for the two-candidate preferred count and withdrawn those results. Quite a few more will follow. I’m pleased to say this hasn’t made my results pages fall over, which are now projecting them properly as Nationals-versus-independents races with TCP results based on my estimates. Cowper is merely down as “NAT AHEAD”, but this is because I apply a wide margin of error in these cases – I’m in little doubt they will ultimately win the seat.

8.10pm. I’ve now done a manual override to make Ryan a Greens-versus-LNP contest, as I did earlier with Griffith. To be clear, there’s a lot of seats where I should be doing this sort of thing, but it’s a somewhat delicate operation and I’m unable to devote my full attention to this.

7.53pm. My system is calling Bradfield for Labor because, I am quite sure, they have entered Labor and Liberal the wrong way around for TCP in the North Wahroonga booth. I expect this to get drowned out as more votes are added.

7.41pm. There are a number of issues with seats where the AEC has picked the wrong candidates, notably Griffith and Ryan, which it would be nice if I had time to attend to which isn’t available to me. I’ve bitten the bullet though and added an override in Griffith which makes Labor finish third and their preferences decide the seat in the Greens.

6.41pm. I’ve had to pull by aggregated two-party preferred projections – there are issues with them that I’ll be too busy to address. Everything else is working as it should though.

6pm. Welcome to the Poll Bludger’s live coverage of the federal election count. My coverage will basically consist of the above link – I am unlikely to have much to offer this evening in the way of blow-by-blow commentary of the count, which will be in ample supply elsewhere. My comments will be probably be restricted for the most part to pointing out bugs and errors in my live results facility, which I don’t doubt there will be. Though I’ll start with the curious fact that the Norfolk Island booth, in the ACT seat of Bean, was already in before polls closed on the eastern seaboard.

As you can see from the link though, the live results feature should have a huge amount to offer even in imperfect form – still more so now that the style sheet has been expertly pimped by new-found friend of the site Steven Tan. The link above opens to a landing page that summarises the current state of the count in all 151 seats, and includes national and two-party preferred projections that I dearly hope end up doing their job.

Further down are links to individual results pages for every seat including projections of the final two-party result and win probability estimates, a table of booth results encompassing two-candidate preferred and the primary vote with tabs allowing you to toggle between raw votes, percentages and swings. Unless the Australian Electoral Commission and other media outlets have changed their practices, I believe this will be the only place on the internet where booth results will be available for most of this evening. At the bottom is a button which, if clicked, opens a booth results map with colour-coded dots and numbers pointing to the results of each booth that you can click on for a full set of results.

Obviously this involved a good deal of effort, so if you find the effort useful or admirable in any way, please consider making a contribution to the upkeep of the site using the “become a supporter” link at the top of this page, or the “become a supporter” button to be found on the top right of each page of the results pages.

Newspoll: 53-47 to Labor

Newspoll closes its account with a poll result which, if bang on the money, would surely deliver Labor a majority.

The Australian reports the final Newspoll of the campaign has Labor leading 53-47, which compares with 54-46 a week ago (and 51.5-48.5 in the erroneous pre-election Newspoll in 2019). The primary votes are Coalition 35% (steady), Labor 36% (down two), Greens 12% (up one), One Nation 5% (down one) and United Australia Party 3% (steady). The two leaders are tied at 42% on preferred prime minister, after Scott Morrison led 43-42 a week ago. Scott Morrison is down one on approval to 41% and up one on disapproval to 54%, while Anthony Albanese is up three on approval to 41% and down three on disapproval to 46%. The poll was conducted Friday to Thursday from a larger than usual sample of 2188.

This is presumably the final data point to be entered to the BludgerTrack poll aggregate, which continues to credit Labor with a two-party preferred lead of 53.5-46.5, which I’ll believe when I see. Its primary votes are Labor 36.3%, Coalition 34.9%, Greens 11.7% and One Nation 4.0% (and United Australia Party 3.2%, but I don’t include that in the display). Both leaders have net negative approval ratings on the leadership trends, Scott Morrison at minus 7.3% and Anthony Albanese at minus 5.9%, with Morrison at plus 5.4% and Albanese at minus 5.4% on the all-or-nothing preferred prime minister result.

We have also had today a report from Roy Morgan in which they relate they have “continued interviewing throughout this week but there has been no evidence of a swing to the L-NP seen in previous weeks continuing during the final week of the campaign”. The report then proceeds to reiterate the results of the poll that was released on Tuesday, which fooled me into thinking this was a new set of result, but apparently not. Hat tip to Adrian Beaumont for pointing this out and apologies to anyone who misled by the original version of this post.