Category: Federal Election 2025

Roy Morgan has its (and anybody’s) second federal poll since the election, showing Labor’s two-party lead unchanged from the first such poll three weeks ago at 58-42. The primary votes are Labor 37.5% (up half), Coalition 31% (steady), Greens 12% (up half) and One Nation 6%. The poll was conducted June 2 to 22 from a sample of 3957.

Also of note was the Australian Electoral Commission’s publication of preference flow data from the election, including aggregated measures of how each minor party’s preferences split between Labor and the Coalition. A large component of the pollsters’ failure to credit the 55.2% two-party vote share Labor ended up recording lay in an expectation, fuelled by recent state elections, that preference flows to Labor would not match those of 2022. In aggregate, Labor’s share of all minor party and independent preferences ended up being all but identical to 2022, vindicating the determination of RedBridge Group (together with late newcomer DemosAU) in their persistence with 2022 election flows in determining their two-party preferred headlines.

However, the preference flow by party data shows that beneath the surface stability was a continuation of an apparent polarisation in minor party preferences, reflected in record highs for both the Greens flow to Labor and One Nation flow to the Coalition. Pollsters were thus vindicated in revising upwards the flow of One Nation preferences to the Coalition – but none correctly apprehended that Greens preferences would continue to trend the other way, and at least one did the opposite.

Click here for full display of House of Representatives election results.

Wednesday

The Bradfield recount has been completed with Nicolette Boele 26 votes ahead. Importantly, the AEC reports that its investigation into cases of multiple marks against names on the electoral roll found that only two voters were “likely to have had a second vote admitted to the count”. Should the Liberals (or, in theory, any voter in Bradfield) wish to pursue a legal challenge, they will have to persuade the court that at least 23 errors were made, either to their disadvantage with respect to adjudication of ballot papers or with voters having been wrongly allowed or refused votes.

Tuesday

Nicolette Boele ended the day 27 votes ahead in Bradfield, one down on yesterday. A good result for Boele from the St Ives Chase booth at the start of the day, with six votes knocked out for Gisele Kapterian, was cancelled out later on when Kapterian gained six votes from the Warrawee booth, the first revision in her favour affecting more than one or two votes. That just leaves a handful of votes to be accounted for, which will assuredly be wrapped up at some point today. The largest booth outstanding is Wahroonga with 693 votes – I am told that this is in fact mostly done, with no indication that the result will be substantially revised. There are also the very small Ultimo and Wynyard booths, with 51 votes between them, along with around 185 postals, 160 provisionals and 150 absents.

In short, it seems likely that the present margin will undergo only minor change, in which case the AEC will shortly declare Boele the winner. The question will then arise as to whether the matter ends up in court. The AEC itself can refer the matter to the Court of Disputed Returns (meaning the High Court or the Federal Court if it chooses to delegate the matter), but a media briefing conducted yesterday suggested this was unlikely. The most obvious basis for it to do so involves the incidence of multiple voting being greater than the final margin. Typically the number per electorate has been around 15, but an AEC spokesperson suggested it would be lower than that due to the increased use of electronic certified voter lists, through which it can be determined in real time if a prospective voter has been marked off already.

If so, it will be a matter for the defeated candidate (or perhaps more to the point, their party) to determine if a challenge is worth pursuing. Doing so would almost certainly involve disputing formality rulings, as was done without success by Labor candidate Rob Mitchell in the seat of McEwen in 2007 (who went on to win the seat in 2010 and has held it ever since). After winning the initial count by six votes and losing the recount by 12 (seemingly the only time a recount has reversed the original result at a federal election), Mitchell argued there were 40 ballot papers ranking him higher than Liberal rival Fran Bailey that had been wrongly rejected, along with one where the opposite happened. After reaching its own conclusions concerning 643 ballot papers that had been reserved for the adjudication of the returning officer, the court revised Mitchell’s losing margin up to 27, and duly rejected his appeal.

The precedent of a court determining a declared result to have been out by 15 suggests the Liberal Party would be strongly tempted to pursue the matter further – although its Queensland equivalent chose not to do so after falling 37 votes short in Herbert in 2016. While the chances of a court declaring Kapterian the winner outright would seem rather slim, it might conclude the proper margin to be below than the number of observed irregularities, causing it to void the result and have the election held afresh.

Monday

Nicolette Boele had her best day yet in the Bradfield recount, the 12 vote lead she opened yesterday widening to 28. Rechecking of nine booths, four of them in the Liberal stronghold of St Ives (reflecting the fact that the recount has proceeded more-or-less alphabetically), knocked 14 votes from Gisele Kapterian’s tally and four from Boele’s, while batches of absents and postals cut five from Kapterian and added one for Boele. This leaves the recount nearly 80% done – the 12 booths remaining to be rechecked (only one of which is a pre-poll centre, and that a rather small one) account for 12,056 votes, on top of which I am told that about 3700 postals, 1500 declaration pre-polls and 500 absents are still outstanding.

Saturday

Good news and bad news for tealdom today, the former being a breakthrough for Nicolette Boele in the knife-edge Bradfield recount. After slowly slipping in counting through most of the day, her one-vote deficit at the start drifting out to seven, Boele’s situation was transformed by the Turramurra pre-poll centre, which knocked out 16 of Gisele Kapterian’s votes and none of her own, pushing her to a 12-vote lead. In total, the recount has cut 82 votes from Kapterian’s tally after preferences and 62 from Boele’s. Out of the grand total of 118,851 votes, still to be recounted are 20 out of 52 election day booths, accounting for 20,322 votes, and two out of 13 pre-poll booths, accounting for 975. It’s a lot harder to say where we are with non-ordinary vote types, except that all but the fairly insignificant provisionals category have been revised, postals apparently on six occasions (though I remain unclear if this encompasses the early and especially strong batches for Kapterian).

Proceedings in Goldstein are finally at an end after the partial recount ended with Tim Wilson 175 ahead, in from 270 at the start of the process, prompting Zoe Daniel to concede defeat. It turns out the 50-vote error alluded to yesterday related to a batch being double-counted (which wasn’t the only time such a thing was found to have happened in Goldstein), rather than the maximal scenario of votes for Daniel having been attributed to Wilson.

Friday

The resolution of the last Senate result today in New South Wales turned up the first genuine surprise, with One Nation’s Warwick Stacey winning a seat that I (and to my knowledge everyone else) had reckoned a sure thing for Labor’s third candidate, Emilija Beljic. After Labor (Tony Sheldon and Tim Ayres) and the Coalition (Andrew Bragg and Jessica Collins) elected their top two candidates with full quotas, the remainder of the pack was whittled away until Mehreen Faruqi of the Greens crossed the threshold to win the fifth seat leaving Stacey and Beljic competing for the sixth. At that point, Stacey led Beljic by 0.886 quotas to 0.823, which preferences from Faruqi’s 0.069 surplus were insufficient to close, Stacey winning at the last by 0.891 quotas to 0.867.

As this simplification of the distribution illustrates, Stacey passed Beljic with the exclusion of Legalise Cannabis one step before the election of the Greens (UPDATE: Kevin Bonham in comments notes it was actually two steps, the previous transfer from Family First’s exclusion having put him ahead), a transfer that included 0.2446 quotas worth of first preferences for Legalise Cannabis and 0.1510 in preferences picked up along the way. My model based on preference flows in 2022 had Labor, One Nation and the Greens gaining very similar shares of preferences with the exclusion of Legalise Cannabis: instead, One Nation gained 0.146 quotas, the Greens 0.123 and Labor only 0.066, the 0.0797 gap between the One Nation and Labor shares pushing One Nation to their 0.024 quota winning margin. However, that is only part of the story of how my model’s projection of a 0.112 quota winning margin for Labor proved out by 0.136 quotas: up to the point of Legalise Cannabis’s exclusion, Labor under-performed the model by 0.046 quotas and One Nation over-performed it by about the same amount. Further insights are available to be gleaned from the full ballot paper data that has already been published by the AEC, though this will have to wait for now.

In the Bradfield recount, Nicolette Boele hit the lead today – but not for long, her one-vote advantage being reversed in the last updates for the day. Debate rages as to whether there is any underlying pattern within changes that have collectively reduced her deficit so far from eight votes to one, with what I would guess to be about 40% of the recount remaining to be done, continuing into today. Updates that looked promising for Boele based on the theory that large numbers of first preferences for Gisele Kapterian meant opportunities for them to be knocked out as informal have in some cases failed to deliver, notably today’s revision for the St Ives pre-poll centre, which reduced both candidates’ totals by one. Thirty-seven of 66 ordinary booths have been recounted, plus I assume all of the absent votes and an unknowable but obviously significant share of the 14,666 postals, but none of the 3405 declaration pre-polls.

The theory of high primary vote disadvantage in recounts has looked good in other contexts, including the Bradfield preference distribution where it played to Boele’s disadvantage by harming her main sources of preferences – and also in the partial recount in Goldstein, which seems to be almost finsihed. Tim Wilson has lost a net 63 votes since the start of the recount to Zoe Daniel’s 36, reducing his lead to 233. The AEC advises we can expect a correction tomorrow involving, among other things, a “change of 50” in favour of Daniel – which could mean a reduction in her deficit to around 183, or a transfer from Wilson to Daniel that would get it all the way to 133. Neither would get her as far as the 100 vote threshold she would need for the AEC to determine if it will keep the ball in play by proceeding to a full recount, though the latter might just about get her within striking range if the tide kept flowing in her favour.

Click here for full display of House of Representatives election results.

Thursday

As suggested here yesterday, today’s Queensland and Western Australian Senate distributions both produced results of two Labor, two Coalition, one Greens and one One Nation – predictably in Queensland’s case, less so in Western Australia’s. It was noted that my model based on 2022 election preference flows got the One Nation candidate in WA to a winning margin over the third Labor candidate of 0.013 quotas, and that the party had over-performed this in similar circumstances by 0.046 quotas in South Australia and 0.032 in Victoria. In the case of WA the improvement was 0.023 quotas, the margin at the final count being 0.895 to 0.859. That just leaves the New South Wales count to be finalised, which is scheduled for 9:30am tomorrow, and looks like a clear-cut result of three Labor, two Coalition and one Greens.

In the Bradfield recount, Nicolette Boele’s momentum yesterday failed to carry over to today: she began proceedings two votes behind and ended three votes behind, with a net 22 votes being knocked out for Kapterian and 23 for Boele. Much of today’s effort was seemingly spent on the 9589 votes of the Willoughby pre-poll centre, which disappointed for Boele in yielding only a net gain of one vote, despite the high Liberal vote there. Another part of today’s recounting was postals, which I presume wasn’t all of them given the modest scale of the changes, with two primary votes knocked out for both leading candidates. This unknown factor means it’s no longer possible to precisely calculate how much of the recount has been completed: ordinary votes amounting 41.1% have been accounted for, together with however many of the 12.3% of the total that were postals have been accounted for. The Liberal favourability deficit out of what’s been counted will also have narrowed, though not closed (remembering that the relative Liberal strength out of what’s been counted will likely be to their disadvantage, since it means more opportunities for their votes to be knocked out).

It seems clear now that the Goldstein recount will not pull any rabbits out of the hat for Zoe Daniel: Tim Wilson’s lead remains unchanged from yesterday at 263.

Wednesday

The Victorian Senate result was finalised today, producing the anticipated result of Labor three, Liberal two and Greens one: in order, Raff Ciccone (Labor), James Paterson (Liberal), Jess Walsh (Labor), Jane Hume (Liberal), Steph Hodgins-May (Greens) and Michelle Ananda-Rajah (Labor). Ananda-Rajah won at the final count with 0.869 quotas to the One Nation candidate’s 0.814, which was 0.032 quotas narrower than anticipated by my model based on preference flows from 2022. Something similar was observed in South Australia, where the final count likewise pitted One Nation against Labor’s successful third candidate, the final result being 0.046 quotas narrower than anticipated by my model.

This is encouraging for Tyron Whitten, One Nation’s candidate in Western Australia, where the count will be finalised tomorrow at 3pm eastern time. My earlier modelling of the result gave the third Labor candidate a narrow win over Whitten of 0.869 quotas to 0.852, but this was before Labor’s vote share fell back on late counting – re-running it with the final results, I get Whitten winning by 0.862 to 0.849. Half an hour later, the Queensland result will be finalised – here a result of two Labor, two Liberal National, one Greens and one One Nation looks assured.

Gisele Kapterian’s lead in Bradfield was slashed today from 14 to two, as proceedings went more as I expected them to go initially: votes for both candidates knocked out as informal, but the process favouring Nicolette Boele by virtue of the recount mostly affecting first preference votes for the two leading candidates, of which Kapterian has more to lose. On Monday and Tuesday, when as many previously informal votes were being deemed formal as vice-versa, Kapterian’s lead climbed from eight to 14 – today, Kapterian had a net loss of 29 votes compared with 17 for Boele.

The process has now resulted in the recounting of 30,357 votes out of 118,856, or 25.5%, encompassing 16 out of 52 election day booths and four out of 14 pre-poll voting centres. Non-ordinary vote types, including over 14,500 postals, are yet to be examined. Out of the votes examined so far, 33.5% were first preferences for Kapterian and 28.2% for Boele, compared with 38.1% and 27.0% out of the total count, with the two-candidate count being 53.9-46.1 in Boele’s favour compared with 50.0-50.0 overall. For reasons noted in yesterday’s update, this indicates the votes to be rechecked lean to Kapterian.

If today’s dynamic holds, with votes being knocked out as informal providing most of the changes, it seems very likely that Boele will soon pull ahead and stay there. However, I have no reason to be sure that we won’t see a re-emergence of the earlier dynamic of as many votes doing the reverse, in which case Kapterian may still be a show. Either way, the margin looks like being fine enough to raise the strong possibility of a legal challenge. Ben Raue of The Tally Room has done us all a fine service in attending the count as a scrutineer and recording his observations.

The partial recount in Goldstein began today, and got about 15% through without bringing any joy for Zoe Daniel, whose deficit against Tim Wilson is out from 260 to 263.

Finally, the AEC is now well into finalising two-party Labor-versus-Coalition counts for “non-classic” contests, which will ultimately allow for a national two-party preferred result. This will settle somewhere between 55-45 and 55.4-44.5, indicating that pollsters who were revising preference models based on the 2022 result to make them less favourable to Labor would have done better to have let them be. This under-estimation of Labor extends to the estimates I was using for non-classic contests to produce the national two-party preferred on my results page – I have revised these upwards, though probably not far enough.

Tuesday

End of day update: In the second day of the Bradfield recount, Gisele Kapterian again widened her lead over Nicolette Boele, which has gone from eight to ten to fourteen. Substantial revisions have been made in the Artarmon Central (17 informal votes reclassified as formal) and Gorton (14 votes going the other way) without appreciably advantaging one candidate or the other – the changes arise from another 13 booths that have been rechecked, five producing revisions in favour of Kapterian against one for Boele. Boele can take at least some comfort in the fact that these booths recorded a relatively narrow 29.6% to 27.6% advantage for Kapterian on the primary vote, compared with 38.1% to 27.0% overall, on the principle that opportunities await for challenges to the formality of Kapterian votes. So far though, it seems that as many informal votes are being deemed formal as the other way round, contrary to the experience of the preference distribution.

Earlier: The Senate distribution for Tasmania produced a result of two Labor, two Liberal, one Greens and Jacqui Lambie, in the order of Carol Brown (Labor), Claire Chandler (Liberal), Nick McKim (Greens), Richard Dowling (Labor), Jacqui Lambie (Jacqui Lambie Network) and Richard Colbeck (Liberal). With four candidates chasing two seats at the second last exclusion, Lambie had 0.82 quotas, Colbeck 0.80, third Labor candidate Bailey Falls 0.73 and Lee Hanson 0.57. Hanson’s exclusion put both Lambie (1.05) and Colbeck (1.01) over the line for a full quota, leaving Falls holding the bag with 0.80. I should have had more faith in my model based on 2022 preference flows in last night’s update, as Lambie got more preferences this time from lower order and mostly right-wing candidates, outperforming my model’s projection of 1.00 quotas for her. Colbeck outperformed his projected 0.92, while Falls did weaker than the anticipated 0.87, again contrary to what I suggested might happen last night.

The distribution for the South Australian count was published today, showing that Labor’s third candidate, Charlotte Walker, recorded 1.00 quotas at the final count ahead of 0.80 for the One Nation candidate, against which my model’s projection of 1.00 to 0.75 stacks up quite well. Also finalised today was the Northern Territory Senate count, confirming the formality of Malarndirri McCarthy (Labor) and Jacinta Price (Country Liberal) winning the two seats. The button-press for the Victorian Senate count is scheduled for 9:30am tomorrow – the evidence so far offers no encouragement for One Nation that they will be able to close what my model projects as a deficit of 0.85 to 0.76 in the race for the final seat against the third Labor candidate. This will shortly be followed by the foregone conclusion of the Australian Capital Territory count.

The preference distribution has been finalised in Calwell, Labor’s Basem Abdo emerging a comfortable winner with 49,481 votes (55.1%) to independent Carly Moore’s 40,350 (44.9%).

Monday

Today’s developments:

• The preference distribution in Calwell has all but confirmed a win for Labor, who received more than two-thirds from the exclusion of Greens, putting Basem Abdo on 48.0%, independent Carly Moore on 29.7% and the Liberal candidate on 22.3%. This leaves Moore needing an all-but-impossible 91% share of the preferences with the imminent exclusion of the Liberal.

• The button was pressed on the South Australian election Senate count, confirming the anticipated result of Labor three (Marielle Smith, Karen Grogan and Charlotte Walker), Liberal two (Alex Antic and Anne Ruston) and Greens one (Sarah Hanson-Young). We must await publication of the preference distribution for further detail.

• The Australian Electoral Commission advises that we can expect the buttons to be pressed tomorrow morning for Tasmania and the Northern Territory. The latter is a foregone conclusion, but the former is likely to find a three-way battle for the last two seats between Jacqui Lambie, the third Labor candidate and the second Liberal candidate. My earlier modelling suggested Lambie was very likely to be re-elected, since substantially different preference flows from the last election would be needed for her to fall behind both Liberal and Labor. On reflection though, the former seems more plausible than I was allowing, given the observable impact of Lambie’s opposition to salmon farming on the geographic distribution of her primary vote, and the fact that most of the preferences being distributed are from right-wing parties. The latter might arise as a corollary of Labor’s stronger performance overall.

• The AEC announced today responded to Zoe Daniel’s request for a recount in Goldstein by announcing a partial recount that would consider first preference votes only, which in fact account for about three-quarters of the total. Substantial revisions were made to the totals during the course of the preference distribution, a process which examined only the remaining one-quarter of the vote, ultimately with the effect of increasing Tim Wilson’s margin from 129 to 270. The recount will begin on Wednesday and is expected to take about four days.

• The first day of the Bradfield recount, which may take as long as two weeks, increased Liberal candidate Gisele Kapterian’s lead over independent Nicolette Boele from eight votes to ten.

Click here for full display of House of Representatives election results.

Saturday

Tim Wilson has finished the preference distribution in Goldstein 260 votes ahead of Zoe Daniel, after a series of late revisions that drove first drove his lead up yesterday from 129 to 444, before today cutting it back to 170 and then settling on the final margin. These convulsions presumably loom large in the request Daniel has submitted for a recount, but experience suggests the AEC will stand by the 100-vote threshold it set in place in 2008.

Friday

Yet another twist in the saga of Bradfield, which ended with Gisele Kapterian taking the lead at the last and finishing the scrutiny with an eight-vote lead over Nicolette Boele. The Australian Electoral Commission promptly confirmed that this would be subject to a recount, as it does automatically when the margin is inside 100 votes, which will begin on Monday and is “expected to take up to two weeks”. Twenty-two out of Kapterian’s 48-vote gain during the preference distribution came with a second correction from the St Ives pre-poll centre, which had put Kapterian in the hunt the Monday after the election with the addition of hitherto unreported votes to its tally, booting her by 440 votes. The issue this time was apparently a transpositional error in the record of preference flows, causing 11 votes to shift from Boele to Kapterian.

The rest of the movement largely resulted from ballots previously admitted to the count being deemed informal, a process that favoured Kapterian because only the third or so of the vote that was cast for excluded candidates was under consideration, around two-thirds of which went to Boele as preferences. Boele’s hope lies in the recount revisiting the two-thirds of the vote that was cast for the two leading candidates, where the same dynamic is likely to work against Kapterian, who has 38.1% of the primary vote to Boele’s 27.0%. If these votes are excluded in roughly the same proportions as those of the other candidates during the preference distribution (during which Labor lost 23 votes, the Greens 14, One Nation 8, independent Andy Yin 7 and the Libertarians 6), Kapterian will lose about 65 votes to Boele’s 45. The distinctions are fine enough that clearly nothing can be said with certainty – and even if Boele were to emerge with the slender lead implied, there would be a strong chance of a court finding enough routine irregularities to void the result and force a by-election.

In Calwell today, the preferences of independent Joseph Youhana were excluded, nearly 60% of them favouring independent Carly Moore, pushing her well clear of the Liberals into second place. Labor holds a lead of 36.6% to 25.6% that Moore needs to chase down with the successive exclusions of the Greens and the Liberals, on 16.7% and 21.2%, which scrutineers’ reports related through the media suggest is unlikely to happen. Also today, whatever lingering hope there may have been for Zoe Daniel in Goldstein was snuffed out by dramatic revisions that blew Tim Wilson’s lead out from 129 to 444.

12.30pm. My assessment of last evening was evidently too sanguine with respect to Nicolette Boele, whose margin is dropping fast – now down to five votes. Almost all of the correction so far today is down to the St Ives pre-poll centre – the same one whose result was dramatically revised in the Liberals’ favour in the early stages of the check count – where 11 votes have been shifted from Boele to Kapterian. The broader dynamic is that the distribution of Labor preferences and their strong flow to Boele means that votes successfully being contested on grounds of formality are mostly for her.

Thursday

The Calwell preference distribution turned up its first real surprise with the exclusion of independent Sam Moslih, with fully 61.3% of the distribution going to the Greens ahead of the other remaining contenders, namely Labor, Liberal and independents Carly Moore and Joseph Youhana. This pushes the Greens ahead of Youhana, who will be the next candidate excluded. Kevin Bonham suggests this reflects a strong influence of Moslih’s how-to-vote card and/or that of Muslim Votes Matter, which is good news for Labor because both favoured Basem Abdo over Moore. Moore presently holds a 17.5% to 15.3% lead over the Greens, which seems unlikely to be closed with the exclusion of Youhana, given he favoured Moore on his how-to-vote card and the general tendency of independent votes to favour other independents. Assuming that’s so, we are now likely to see Greens preferences push Moore ahead of the Liberals, whose preferences will then produce a final result between Labor and Moore. Moore will need around 67.5% of the preferences shortly to be distributed from Youhana, the Greens and the Liberals.

Proceedings today in Bradfield wore Nicolette Boele’s lead down from 41 votes to 28, with three added to Gisele Kapterian’s tally and ten subtracted from Boele’s. A source familiar with the matter in comments indicates we should now be a good way into the last phase, namely the distribution following the exclusion of Labor with only Boele and Kapterian left standing. If the apparent pattern of movement in favour of Kapterian looks unlikely to eliminate the margin altogether, it does remove whatever doubt there may have been that the it will fall inside the 100-vote threshold for an automatic recount.

Wednesday

Calwell proceeded today through to the eighth count, leaving a remaining field of Labor, Liberal, the Greens and three independents. Carly Moore’s lead over the other independents has widened, and seems likely to be maintained through the imminent exclusions of the Greens and two other independents, together with the elimination of the current 18.5% to 16.4% gap between the Liberal candidate and Moore. Between now and the final count, Moore would need two-thirds of the preferences to overtake Labor.

Today’s preference distributions added 15 to the informal counts in both Bradfield and Goldstein, respectively cutting Nicolette Boele’s lead by four to 41 and increasing Tim Wilson’s lead by one to 129.

Tuesday

End of counting. We’re now six counts into Calwell, with another six exclusions to come. Candidates accounting for 9.1% of the primary vote have now been excluded, with results that probably don’t tell us all that much. Next out will be Legalise Cannabis and One Nation, who will perhaps go relatively heavily to established parties rather than independents, followed by Sam Moslih, whose how-to-vote card had Labor higher than the remaining independents. Most likely, the issue will then be whether preferences from Joseph Youhana, the Greens and the Liberals favour Moore enough to get her ahead of Basem Abdo.

Revisions arising from the preference distribution in Bradfield today have added 11 to the informal vote tally, costing Gisele Kapterian eight votes and Nicolette Boele two, increasing the latter’s lead from 39 to 45.

5pm. The Goldstein count has ended with Tim Wilson up by 128 votes. The AEC relates that the votes still in the system as awaiting processing have Senate ballot papers only. The preference distribution will now proceed, to be followed only by an automatic recount if the margin comes in below 100, though the discretion remains to conduct one even if it doesn’t. Arguments have been made that the population has increased since the 100-vote threshold was established about 15 years ago.

2.30pm. The Australian Electoral Commission will helpfully be publishing updates from Calwell in the form of progress preference distribution results that will presumably be updated with each exclusion. These are a bit hard to read, so I offer the following summary below, showing us up to count four out of twelve. This looks promising for independent Carly Moore with respect to her prospects of making the final count, with 21.5% of the preferences from the first three exclusions having gone to her. However, the marginal nature of the candidates excluded so far is such that these figures are unlikely to offer much insight as to whether Labor will receive enough preferences to get from their 30.5% primary vote to 50% at the final count. If it is Moore who comes second, she will need about two-third of preferences (and Labor one-third) from all other candidates.

Monday

The last batches of votes in Bradfield kept true to the contest’s epic form, with independent Nicolette Boele taking the lead at the last to end the scrutiny 39 votes ahead. But it doesn’t end there: the formal distribution of preferences will proceed throughout this week, almost certainly to be followed by the recount that proceeds automatically when the margin is inside 100 votes, so Liberal candidate Gisele Kapterian has at least some hope that proceedings turn up errors substantial enough to reverse the result. Kapterian began the day 43 votes ahead, then moved to 50 ahead when absents broke 29-22 her way. Boele’s breakthrough came when postals broke fully 125-56 her way, consolidated when declaration pre-polls favoured her 111-90.

A recent recount precedent missing from yesterday’s summary was Clive Palmer’s win in Fairfax in 2013: at 36, his margin on the indicative two-candidate count was very close to Boele’s, but it was reduced to seven during the preference distribution and then inflated to 53 on the recount. An informed source in comments notes that recounts have become less prone to produce changes since the initial recheck became a routine part of the procedure in 1984, and court rulings established legal precedents about formality, most notably in relation to the seat of McEwen in 2007.

In Goldstein, Tim Wilson’s lead is down from 254 to 206 after postals broke 94-60 and absents 76-62 to Zoe Daniel. The AEC records 332 envelopes awaiting processing, of which Daniel would need two-thirds to land in her column to get to automatic recount territory.

Sunday

With the deadline for the arrival of late postals having passed on Friday, there are two seats that can still be regarded as in doubt, barring extraordinary late developments. One is Bradfield, where today’s counting will account for 260 declaration pre-polls, 104 postals and 66 absents (UPDATE: Outstanding postals revised up to 191). Some of these will be deemed invalid and a handful will be informal, but as many as 380 will be added to a tally on which independent Nicolette Boele trails Liberal candidate Gisele Kapterian by 43 votes.

The counting of these votes will be followed immediately by a full distribution of preferences. Should the margin land inside 100, as seems extremely likely, this will be followed by an automatic recount. A review conducted for the Australian Electoral Commission in 2014 helpfully reviews the history of recounts, which provides at least some level of information on how much the dial was moved by 11 recounts going back to 1958 (see pages 24 and the very last page). A recount for Bass in 1998 was something of an outlier in increasing Labor’s winning margin from 16 to 78. Including the one recount conducted since – for Herbert in 2016, which increased Labor’s margin from 8 to 35 – a typical recount seems to make about 20 votes’ difference to the final result.

A recount would seem to be the only remaining chance for Zoe Daniel in Goldstein, who trails Liberal candidate Tim Wilson by 206 votes with 332 remaining to be processed: 172 declaration pre-polls, 100 postals and 39 absents, plus 21 provisionals that may all be disallowed. Even getting to the 100-vote threshold requires stretching the arithmetic here, but the returning officer can use their discretion to require a recount even if the threshold isn’t reached.

The other unknown is the seat of Calwell, which I have not been making the effort to follow on a blow-by-blow basis, since the point at issue is that there’s no way of knowing which out of as many as four candidates will make the final count along with Labor. The only thing that can be said for sure is that Labor win the seat if it’s the Liberal candidate, but it’s quite a bit more likely to be an independent. Such questions can only be answered by a full distribution of preferences. With only 154 votes remaining to be processed, this will presumably begin later today.

Then there’s the Senate, where the pressing of the button on the final results is still as much as a fortnight away. I have a post below with my latest updated assessments on how that is likely to play out.

I have conducted a repeat of the exercise from my previous post on the Senate results, which involved taking random samples of the ballot papers from the 2022 election, weighting them to match the various players’ shares of the first preference vote at this election, and simulating preference distributions in which flows behave as they did in 2022. This is a simplified approximation of the process, so what might be referred to below as “Count 72” would be a lot higher by the AEC’s reckoning, but more than adequate for current purposes.

There are two important differences from the first run, the most obvious being that it’s based off a more advanced stage of the count. The other is that I have factored in changes in how-to-vote cards at this election, at least for the major parties (minor party how-to-vote cards are rarely followed, and changes unlikely to amount to much when considered in aggregate). This was mainly deemed necessary because the Coalition was a lot more amenable to One Nation than in the past, although the (electoral) significance of this should not be overstated — impacts are trivial where Liberal candidates are elected with only small surpluses, as is invariably the case when they themselves are elected off the preferences of other parties, and non-existent when they remain to the final count. The former applies in New South Wales and South Australia, where the Coalition will respectively win two seats from a shade over and a shade under two quotas on first preferences, and the latter applies in Tasmania.

My overall assessment is unchanged, the modelled result being Labor 30, Coalition 27, Greens 11 and One Nation two, plus Ralph Babet, Jacqui Lambie, Tammy Tyrell, Lidia Thorpe, Fatima Payman and David Pocock. However, the already live possibility of One Nation taking Labor’s third seat in Western Australia now looks stronger. I don’t believe the boost to One Nation from Coalition preferences puts them in serious contention in Victoria, and it affects only the size of their winning margin in Queensland. Nor do I think it likely that Jacqui Lambie will lose her seat, notwithstanding The Australian’s contention yesterday that her “folly” in attacking Tasmania’s salmon farming industry “may cost her political career”.

New South Wales

	Count 1	Quotas	Swing	Count 72	Count 73
ALP	37.77%	2.644	+7.34%	0.803	0.892
LNC	29.63%	2.074	-7.09%
GRN	11.16%	0.781	-0.30%	0.973	1.068
ON	6.03%	0.422	+1.91%	0.692	0.780
LC	3.39%	0.237	+0.78%	0.379

My earlier projection here of three Labor, two Coalition and one Greens doesn’t seem to be in doubt. Since the Coalition scrapes over the line for a second quota on primary votes, what happens with their preferences matters little — where previously I had Labor winning the last seat ahead of One Nation by 0.938 quotas to 0.771, I now have it at 0.892 to 0.780, and I expect most of the change is due to shifts in party vote shares over the past week.

Victoria

	Count 1	Quotas	Swing	Count 73	Count 74	Count 75	Count 76	Count 77	Count 78
ALP	34.79%	2.435	+3.34%	0.506	0.515	0.572	0.606	0.692	0.850
LNC	31.77%	2.224	-0.52%	0.292	0.293	0.308	0.342
GRN	12.31%	0.862	-1.54%	1.019
ON	4.44%	0.311	+1.53%	0.412	0.412	0.426	0.558	0.683	0.763
LC	3.56%	0.249	+0.55%	0.319	0.320	0.370	0.399	0.453
ToP	2.52%	0.176	-1.49%	0.244	0.244	0.260
AJP	1.56%	0.109	+0.04%	0.174	0.179

After the election of two Labor, two Coalition and one Greens, I now have the third Labor candidate’s lead over One Nation at 0.850 to 0.763 with the amendment of Coalition votes that followed the card, as compared with 0.740 to 0.608 in the first run. One Nation now gains 0.125 when the Liberal is excluded, compared with 0.060 previously. However, Labor has gained nearly as much since last time, for one reason or another, and a One Nation win would have to be rated unlikely. It’s true that One Nation’s 4.44% on first preferences is higher than the 4.01% from which Ralph Babet scraped home for the United Australia Party in 2022, but the two major parties had between 2.2 and 2.3 quotas on that occasion – this time Labor is up to 2.435, reducing the chances of two seats going to minor parties.

Queensland

	Count 1	Quotas	Swing	Count 78
ALP	30.67%	2.147	+4.01%
LNP	31.36%	2.195	-6.68%
GRN	10.29%	0.720	-3.09%	0.993
ON	7.05%	0.494	-0.94%	0.748
GRPF	4.63%	0.324		0.468

This appears clear cut, with both Labor and the Coalition a bit above two quotas and no chance of a amassing a third, leaving the other two seats to go to the Greens and One Nation. A joint ticket for Gerard Rennick and Katter’s Australian Party scored a solid 4.63%, which I dealt with by substituting it for the United Australia Party in the 2022 party data, such that it gave and received the same preferences flows. No doubt this is imprecise, but the margins involved are such that it doesn’t matter much.

Western Australia

	Count 1	Quotas	Swing	Count 54	Count 55	Count 56	Count 57
ALP	36.49%	2.555	+1.94%	0.642	0.654	0.720	0.869
LNC	30.31%	2.122	-1.36%
GRN	12.78%	0.895	-1.47%	1.017
ON	5.87%	0.411	+2.38%	0.625	0.625	0.746	0.852
LC	3.94%	0.276	+0.56%	0.399	0.402	0.416
AUC	2.64%	0.185	+0.47%	0.253	0.253

I continue to project a result of Labor three, Liberal two and Greens one, but incorporating One Nation into the Liberal how-to-vote card order makes a close race even closer. The operative margin between the third Labor candidate and One Nation at the close is now 0.869 to 0.852, in from 0.872 to 0.825 the first time. I have factored in that Liberal how-to-vote cards varied from seat to seat in 2022 depending on how it was thought a recommendation for One Nation would play with local voters (a source of much thundering outrage from elements of the media when Labor did something similar with the Greens at this election), and am indebted to Kevin Bonham for recording the variants from 2022.

South Australia

	Count 1	Quotas	Swing	Count 44	Count 45	Count 46	Count 47	Count 48	Count 49	Count 50
ALP	38.38%	2.687	+6.12%	0.704	0.706	0.723	0.730	0.852	0.944	0.995
LIB	27.77%	1.944	-6.16%	1.013
GRN	12.73%	0.891	+0.78%	0.996	0.998	1.011
ON	5.26%	0.368	+1.25%	0.415	0.420	0.470	0.470	0.504	0.551	0.749
ToP	2.80%	0.196	-0.23%	0.228	0.228	0.262	0.262	0.286	0.330
LC	2.77%	0.194	+0.45%	0.232	0.233	0.240	0.242	0.263
JLN	2.71%	0.189	+0.63%	0.219	0.220	0.233	0.233
FFP	1.99%	0.139	+1.56%	0.157	0.158

Here the Liberals scrape over the line to a quota at a late stage of the count, meaning their preferences for One Nation are of limited consequence. Previously I had the third Labor candidate’s winning margin over One Nation at the final count at 1.029 quotas to 0.743, now it’s 0.995 to 0.749. Here I used Bob Day’s independent candidacy in 2022 as a stand-in for Family First and the Rex Patrick Team for the Jacqui Lambie Network.

Tasmania

	Count 1	Quotas	Swing	Count 36	Count 37
ALP	35.53%	2.487	+8.55%	0.761	0.874
LIB	23.67%	1.657	-8.28%	0.793	0.922
GRN	16.14%	1.130	+0.70%
JLN	7.25%	0.508	-1.37%	0.785	1.001
ON	5.08%	0.356	+1.21%	0.596

Beyond a clear two seats for Labor and one each for Liberal and the Greens, The Australian rates this a “tight, four-way preference contest” between Labor’s third, the Liberals’ second, Jacqui Lambie and Lee Hanson of One Nation. I consider this generous to Hanson, who does not stand to benefit from Liberal preferences as she will be excluded while the second Liberal remains in the count. I then have the third Labor candidate losing the game of musical chairs to the other two, but if the primary vote swings are reflected in preference flows stronger for Labor and weaker for Liberal, the modelled gap of 0.922 to 0.874 is narrow enough that Labor might win a third seat at the expense of the Liberals’ second (conversely, a weakening in support for Lambie among conservatives might mean stronger flows from One Nation to Liberal). The Australian makes some notable points about the pattern of Jacqui Lambie’s 7.3% vote, which is down from 8.9% when she last ran in 2019 and the 8.6% her party’s ticket scored in her absence in 2022. Support for Lambie went up in the city and down in the country, and slumped in salmon farming towns. However, my model has her on over a quota in the three-way race with 1.001 quotas, meaning she would have to fall behind both Liberal on 0.922 and Labor on 0.874 to lose. This doesn’t seem likely even allowing for the principle that a lower primary vote means weaker preferences, particularly considering such an effect would harm the Liberals at least as much.

Click here for full display of House of Representatives election results.

Friday

No counting today in Bradfield — apparently everything outstanding will be knocked off on Monday, followed by a distribution of preferences. Arriving today were 106 declaration pre-polls, which were the last of those unaccounted for, and the last postals before the deadline, of which there were 48. This is in addition to those that were already awaiting processing, of which there were 66 absents, 154 declaration pre-polls and 83 postals. That’s 430 overall, of which about 10% will be deemed invalid or informal if consistent with the overall trend, but I’m persuaded it might be a bit higher with the final batch. If not, my estimate yesterday of 380 will hold, as will my assessment that Nicolette Boele will need about 56% to win or 43% to quality for an automatic recount, the latter of which at least seems extremely likely.

In Goldstein, Zoe Daniel continues, with astonishing regularity, to whittle away at Tim Wilson’s lead at a rate not quite high enough to overturn what looked like a comfortable lead this time last week. Today it was a 104-102 break in declaration pre-polls that reduced her deficit to 254. However, there are only 332 remaining to be processed with no more to follow, of which Daniel would need nearly three-quarters to get to an automatic recount even if none of them were disallowed or informal.

Thursday

Bradfield. Gisele Kapterian’s lead was cut from 80 to 43 due to postals breaking 112-85 and out-of-division pre-polls breaking 149-138 to Nicolette Boele, while absents went 46-45 to Kapterian. Awaiting processing are 154 out-of-division pre-polls, 83 postals and 66 absents. A further 105 out-of-division pre-polls are yet to arrive, but will be admitted to the count when they do, as will whatever postal votes arrive tomorrow, of which there were 33 today. With about 90% of likely to become formal votes, that leaves about 380, of which Boele will need about 56%, or 43% to stay below the automatic recount threshold of 100 votes. I believe the last recount was held for Herbert in 2016, and it moved Labor’s winning margin from eight votes to 35. The only recount ever to change the result was in McEwen in 2007, when a six-vote Labor margin became a 12-vote Liberal margin. Labor succeeded in getting the Court of Disputed Returns to review the formality decisions, but its determinations in fact increased the Liberal margin to 31.

Goldstein. Tim Wilson’s margin was cut today from 401 to 292, as out-of-division pre-polls, absents and provisionals respectively broke 248-238, 158-92 and 98-65 to Zoe Daniel. However, Daniel’s vague hopes of getting below the automatic recount threshold have likely been dashed: I’m estimating another 380 out-of-division pre-polls, 225 postals, and 175 absents, of which Daniel would need 62%, increasing to 68% for her to actually chase down the margin.

Wednesday

Bradfield. A less fortuitous day today for Nicolette Boele, with absents breaking 105-79 against her, postals defying the strong earlier trend in breaking only 145-143 her way, and out-of-division pre-polls going a lot less well for her than yesterday’s in favouring her only by 284-281. That increases Gisele Kapterian’s lead from 59 to 80 and blows most of the chances Boele had to chase it down. My assessment of what’s yet to come: about 400 out-of-division pre-polls, 150 absents and 350 postals (262 of the latter are awaiting processing, which will be supplemented by new arrivals tomorrow and the next day, of which there were 144 today and 248 the day before). Boele will need around 54.5% of them to take the lead and 49% to make it to an automatic recount.

Longman. LNP member Terry Young has finally shaken off the challenge here, thanks to a large batch of out-of-division pre-polls breaking 1299-1111 his way, together with a 248-241 advantage on postals and a 510-488 disadvantage on absents. This boosts his margin from 162 to 335 with maybe 700 votes to go.

Goldstein. By popular demand, I’ll note that Tim Wilson’s lead is now down to 401, having started the week at 1472. Zoe Daniel won today’s batches by 296-189 on absents, 554-422 on out-of-division pre-polls and 251-231 on postals. Likely still to come: 750 declaration pre-polls, 500 absents, 150 postals and 100 provisionals, of which Daniel would need at least 64%, compared with the 62.9%, 55.2% and 55.7% she’s received across the three days this week.

Flinders. Of note because the AEC are doing a fresh count between Liberal member Zoe McKenzie and independent Ben Smith that’s now mostly done, on which McKenzie leads 53.3-46.7, which I’m projecting to come in at 52.6-47.4.

Tuesday

The AEC has pulled a number of its three-candidates, indicating it it satisfied it has the right candidates in its two-candidate counts. This is of consequence in Ryan, as it means the Greens rather than Labor will defeat the LNP at the final count; Grey, as it rules out independent Anita Kuss finishing ahead of Labor and being a shot at defeating the Liberal; and Monash, where the same applies for Deb Leonard. Elsewhere:

Bradfield. The big development of today’s counting was the return of Bradfield to everyone’s “in doubt” column after a small but potent batch of out-of-division pre-polls broke 298-179 in favour of Nicolette Boele, slashing her deficit from 178 to 59. For my part, I have also determined (here as elsewhere) that I over-corrected yesterday in reducing the number of outstanding votes, which as much as the size of the Liberal lead was a factor in my system calling it for Liberal from last night until the new numbers were reported this afternoon. My assessment is now that that there are likely to be:

• About 1000 out-of-division pre-polls. Since Boele has received 51.8% of these overall, she will gain about 35 votes if those outstanding prove typical. Of the four batches that have reported, she has successively received 49.6%, 46.0%, 50.4% and now 63.5%.

• Perhaps as many as 700 postals. There are 421 listed as awaiting processing, and the acceptance rate so far has been 97.7%. This includes 86 that were received yesterday and 248 today – for purposes of my calculations I’m conservatively estimating another 100 a day up to the deadline on Friday. Here the trend to Boele has been clear, her share of successive batches being 41.0%, 43.1%, 43.4%, 48.6%, 49.5% and 54.7%. If that last batch is repeated, Boele will gain about 60 to 70. Should they suddenly revert to the mean, Kapterian will gain about 100.

• About 350 absent votes. Of the three batches that have reported so far, Boele has progressively received 47.2%, 49.6% and 53.9%. If the remainder break like the last batch, Boele will gain about 25 votes. If they break as absents have done in total, they will make next to no difference.

Which collectively suggests a trend to Boele that my projected 27.6% win probability for her isn’t factoring in.

Longman. After an interruption yesterday, Labor’s very slow and steady progress in chipping away at the LNP lead resumed with favourable breaks in out-of-division pre-polls (296-262) and absents (415-414) and a net gain of 59 on rechecking. This brings the LNP lead in from 256 to 162. My estimate of what’s to come is about 2400 out-of-division pre-polls, 1250 absents, 700 postals and 100 or so provisionals. If these behave as such votes have so far, Labor will make up about 100 votes. Absents and postals have been trending in their favour, but not overwhelmingly so.

Goldstein. Zoe Daniel’s late charge continues, today’s out-of-division pre-polls (1084-886) and absents (547-409) reducing Tim Wilson’s margin from 963 to 660, after yesterday’s counting reduced it from 1472. However, even allowing for my increase in estimated outstanding votes to about 3400, this leaves her needing a formidable share approaching 60%. This is slightly higher than the 58% I estimated yesterday, and well clear of the 55.2% she got today.

Fisher. Yesterday I wrote that the AEC’s three-candidate count “makes it clear independent Keryn Jones will make the final count”. Evidently I shouldn’t have, because it now has her falling to third with 30,485 votes to second-placed Labor’s 30,672. That count is 1513 behind the primary vote count, and I’m estimating 6000 still to come, so the identity of the final two candidates remains an open question. If it’s Labor, Andrew Wallace of the LNP will win comfortably; if it’s Jones, he will probably win quite a lot less comfortably. My system is presently giving Jones a 7.7% chance.

Bean. Labor now leads by 354, and I only rate that there are about 1600 still to come, leaving independent Jessie Price needing over 60% after every category of vote has been running against her.

Calwell. The Guardian explains this so I don’t have to.

Monday

I’ve finally made the effort to revise how many votes are outstanding, which has until now erred well on the high side, causing my probability estimates to be generous to trailing candidates, and also used the AEC’s three-candidate counts to revise preference flows in seats close enough for it to be worth the effort. I’ve also done some pretty serious rewriting of the code that handles preference flows, which was at once too clever and not clever enough — if it’s doing anything strange, that’s where the explanation will lie.

Bradfield. I’m calling this for Liberal now, but a late surprise is at least mathematically possible. Nicolette Boele’s deficit narrowed today from 227 to 178 after she got the better of absents (173-148), out-of-division pre-polls (143-141), provisionals (78-75) and re-checking (a net gain of 19). There will be maybe 1300 to 1400 more votes admitted to the count, of which Boele will need at least 56% – even on a good day like today she managed only 52%, rechecking aside, which is probably complete now. A bit under 53% would get her as far as an automatic recount.

Kooyong. This is well and truly over now: Monique Ryan won today’s postals 1420-1006 and its out-of-division pre-polls 253-232, and now leads by 1128.

Longman. The momentum to Labor here has slowed and now stalled: today, absents broke 142-139 to the LNP, who also made a net gain of 34 on rechecking, increasing the lead from 219 to 256. There’s a lot still out there though: over 2000 absents, approaching 3000 out-of-division pre-polls, and about 400 postals and provisionals, with absents in particular likely to be favourable to Labor.

Ryan. This is as good as called for the Greens now, which is to say that the three-candidate numbers give Labor essentially no chance of making the final count at their expense, since there’s no question but that the LNP will come second.

Goldstein. Since my system doesn’t call a seat until it rates the probability at 99%, it isn’t entirely writing Zoe Daniel off — some good results on absents and rechecking have brought her deficit to inside 1000, and there’s a lot still to come, including another 2000 absents at least 3000 out-of-division pre-polls, and about 750 postals and provisionals. Of this she needs about 58%, which is remote enough that I’m not going to comment further unless something surprising happens.

Fisher. I haven’t been commenting on this one, but it’s a wild card — the three-candidate count makes it clear independent Keryn Jones will make the final count, and as the AEC’s two-candidate is still between LNP and Labor, it’s only on the basis of preference estimates that my own projection has Jones’s chances at as low as 3.2%. A lot of votes are as yet uncounted, but the real variable of consequence is preference flows, which we won’t know about until the full distribution is conducted.

Bean. This has been trending away from independent Jessie Price, who now trails by 450, and with the revision of outstanding votes my system is a fraction away from calling it.

Grey. My model looks like it was doing its job in rating it unlikely that independent Anita Kuss would reach the final count at Labor’s expense, because she’s fallen into third place on the AEC’s three-candidate count now the Whyalla booths are in. My model currently rates her a 10.0% chance of getting ahead and then making it home over the Liberal at the final count.

Forrest. The ABC is calling this for the Liberal, but I’m not writing off independent Sue Chapman quite yet. She’s third behind Labor in the AEC three-candidate count by 30.1% to 29.1%, which my projection narrows to 30.0% to 29.3%, giving her an 11.1% chance of edging ahead and then doing well enough on preference to win at the final count.

Saturday

Time for a fresh post of daily updates on counting for House of Representatives seats, which will linger through the coming week as late postals arrive.

Grey. I’ve managed to get my system to stop calling a clearly in-doubt seat as a win for Liberal candidate Tom Venning by having it treat Anita Kuss as a generic rather than a teal independent for the purposes of the ad hoc two-candidate preferred count between the two, which is necessarily speculative because there is only hard data for Liberal-versus-Labor. However, my projection still rates it as more likely that Kuss will fail to make the final count ahead of Labor — Kuss now leads Labor on the AEC’s still incomplete three-candidate count, but the booths this is based on are weaker for Labor and stronger for Kuss (by virtue of under-representing Whyalla), and my system is correcting for the difference.

Bradfield. With the clock running down, Nicolette Boele’s deficit went from 209 to 227 today, as absents broke 684-672 and out-of-division pre-polls broke 250-213 in favour of Gisele Kapterian, cancelling out a 31-vote gain for Boele on rechecking. My best guess is that about 1750 votes of various kinds are still to come, of which Boele needs upwards of 56%.

Kooyong. My system is calling this for Monique Ryan now — Amelia Hamer was relying on a borderline-implausible flow of out-of-division pre-polls, but the first batch of them has come in at 1260-1183 in favour of Ryan. A batch of postals broke 511-467 to Hamer, limiting Ryan’s gain since yesterday to 661 to 693. Likely still to come: about 1500 more out-of-division pre-polls, 1000 absent, and few hundred each out of provisionals and late-arriving postals, of which Hamer needs at least 70%.

Bendigo. The Labor-versus-Nationals two-candidate count has caught up with the primary votes now, putting Labor ahead 51.4-48.6, which my projection is generously (for the Nationals) revising down to 50.7-49.3, but there’s no real doubt Labor have made it over the line.

Longman. Yesterday I noted that Labor would sneak home if absents and out-of-division pre-polls continued at their existing rate, so it was good news for Terry Young that a batch of the latter broke 509-458 his way, where the first batch had gone 360-321 against. Conversely, he had his first unfavourable batch of postals to date, going 499-473. He also lost a net 37 on rechecking, leaving him with a lead of 219, in from 231 yesterday.

Bullwinkel. Labor’s lead here is out from 634 to 990, and I’d say that’s your lot.

Finally, turning to the Senate. Below are the results of an exercise in which I have sought to model the count using ballot paper data from the 2022 election. The approach involves weighting ballot papers to reflect the extent to which the various parties’ votes shares increased or decreased at this election, and to conduct simulated preference distributions that effectively assume that each parties’ preferences will flow in the same way this time as they did in 2022. To make this doable in a manageable time, I have sampled every twentieth or every tenth ballot paper, depending on how big the state’s population is.

The only real fly in the ointment I can see in relation to this approach relates to changes in how-to-vote cards, though I only think this likely to be consequential in relation to the Liberal Party directing its preferences to One Nation ahead of Jacqui Lambie on this occasion in Tasmania, reversing its recommendation from 2022. However, this turns out not to matter since, as will be detailed below, the second Liberal will remain until the final count, meaning the preferences of those who followed the Liberal how-to-vote card will not be distributed. The other point to be noted is that I haven’t got around to doing Western Australia yet, but I’ll be on to that tomorrow.

To cut a long story short: in New South Wales, Victoria, South Australia and, it seems, Western Australia (see also the analysis of Adrian Beaumont at The Conversation) the results should be Labor three, Coalition two and Greens one. In Queensland, the result will be two each for Labor and the Coalition and one each for the Greens and One Nation. The closest contest looks like being Tasmania, not between Jacqui Lambie and Lee Hanson as I might have figured, but between the third Labor and the second Liberal, with the latter favoured. The results in the territories will be one each for Labor and the Coalition in the Northern Territory, and one for Labor and one for David Pocock in the Australian Capital Territory. Together with Senators carrying over from the last term, that will likely mean Labor 30, Coalition 27, Greens 11 and One Nation two, plus Ralph Babet, Jacqui Lambie, Tammy Tyrell, Lidia Thorpe, Fatima Payman and David Pocock.

New South Wales. I’m projecting a result here of three Labor, two Coalition and one Greens. The relevant players start out as follows: Labor 2.6097 quotas (37.86%), Coalition 2.0487 (29.77%), Greens 0.7661 (11.13%) and One Nation 0.4122 (6.01%). That gives Labor and the Coalition two each off the bat, and the Greens eventually reach a quota at a point where the third Labor candidate has 0.8934 and One Nation has 0.7686. The Greens surplus then flows mostly to Labor, who win the last seat with 0.9377 quotas to One Nation’s 0.7711.

Victoria. The same basic story here, the relevant starting point being Labor 2.4129 quotas (34.82%), Coalition 2.1919 (31.54%), Greens 0.8635 (12.42%), One Nation 0.3064 (4.38%), Legalise Cannabis 0.2478 (3.58%) and Trumpet of Patriots 0.1710 (2.47%). From there the Greens accumulate enough preferences to get a quota, after which things unfold as follows:

LIB #3	LC	ALP #3	ToP	ON
0.2501	0.3186	0.5021	0.2113	0.3622
0.2782	0.3406	0.5303	Excluded	0.4761
Excluded	0.3750	0.6011		0.5366
	Excluded	0.7398		0.6082
		Elected

I note that Labor got slightly more preferences upon the Liberals’ exclusion than One Nation, and would think that unlikely to happen this time. But I’d doubt it will be enough to account for the size of the projected margin. UPDATE: I said that not realising the Liberal how-to-vote recommendation included One Nation this time and not last time, a fact I am now alerted to by Kevin Bonham in comments, so that was undoubtedly an understatement — though probably not enough of one to account for the final gap of 0.7398 to 0.6082.

Queensland. Here there is the challenge of a 5.16% vote for a joint Gerard Rennick/Katter’s Australian Party ticket, which has no precedent from 2022, and which I’ve dealt with by substituting it for the United Australia Party, thereby writing Trumpet of Patriots out of contention, which matters not because they aren’t competitive anyway. The starting points are Labor 2.1138 quotas (30.80%), Coalition 2.0971 (30.89%), One Nation 0.4754 (6.98%), Greens 0.7174 (11.25%), Gerard Rennick 0.3349 (5.16%) and Legalise Cannabis 0.2389 (3.70%). After the Greens get over the line leaving one seat to go, One Nation has 0.8204 quotas, Rennick 0.4962 and Legalise Cannabis 0.4733. With the exclusion and distribution of the latter, One Nation wins 0.9145 to 0.5732, a margin that would presumably widen if Legalise Cannabis edged ahead of Rennick and the latter’s preferences were distributed instead.

Western Australia. Forthcoming, but pretty clearly Labor three, Liberal two and Greens one. UPDATE: I speak too soon. This in fact looks to be a very close race at the last between Labor’s third candidate and One Nation. The relevant players start at Labor 2.5355 quotas (36.55%), Coalition 2.0899 (30.03%), Greens 0.8949 (12.98%) and One Nation 0.4044 (5.82%). The Greens accumulate enough preferences to win the fifth seat, after which Trumpet of Patriots and then Australian Christians go out. This leaves Labor’s third candidate on 0.7264, One Nation on 0.7257 and Legalise Cannabis on 0.4071. The exclusion and distribution of the latter then gets Labor home by 0.8723 to 0.8246. A complication here is that the Nationals had a ticket this time but not last time, which polled 3.57%. I’ve dealt with this by merging Liberal and Nationals in determining the weighting applied to Liberal votes from 2022, which may mean One Nation gets stronger flows out of the Coalition than I’m crediting them with, since I assume they will do better out of the Nationals than the Liberals. But I’d doubt there’s much in it. It also means I’m not accounting for whatever share of the Nationals vote fails to pass on to the Liberals as preferences, but since they are assured of two seats at a fairly early stage in the count and nothing further, I don’t think this will matter.

South Australia. The starting points here are Labor 2.689 quotas, Liberal 1.9164, Greens 0.8885 and One Nation 0.3606. The Greens and the second Liberal accumulate enough preferences to get elected at around the same point in the count, after which the third Labor candidate wins the last seat like so:

ALP #3	JLN	ON	LC	ToP
0.7584	0.2337	0.4600	0.2406	0.2628
0.8850	Excluded	0.4983	0.2600	0.2839
0.9748		0.5452	Excluded	0.3325
1.0290		0.7426		Excluded
Elected

Tasmania. The likely outcome here is two Labor, two Liberal, one Greens and Jacqui Lambie, but I’m leaving open the possibility of three Labor and one Liberal. The relevant starting points are Labor 2.2521 quotas, Liberal 1.4012, Greens 1.1027, Jacqui Lambie Network 0.4744 and One Nation 0.3478. Two Labor, one Liberal and one Greens will be elected off the bat, with the last two seats to play out following exclusion of sundry also-rans:

ON	ALP #3	LIB #2	JLN
0.5897	0.7395	0.7983	0.7791
Excluded	0.8522	0.9183	1.0011
	0.8523	0.9184	Elected
	Excluded	Elected

I’ve been too consumed by the minutiae of the count to have accumulated any deep thoughts about the result, or even the polls. But I can recommend the assessment of Matthew Knott of Nine Newspapers, for identifying the relevance of Nate Silver’s axiom that “almost all polling errors occur in the opposite direction to commentators’ predictions”. A case in point being my own instinct that Labor couldn’t possibly doing as well as as BludgerTrack’s end-of-campaign reading of the situation, which gave Labor a lead of 53.2-47.8. My own projection of the national result currently has it at 54.2-45.8, but this is generally reckoned (in both senses) too conservative — a display at The Guardian appears to have it at around 54.6-45.4. On either reading, this is Labor’s best result since 1943, and was exceeded in this time by the Coalition only in 1966 and 1975. The primary vote, of course, is another matter — Labor’s currently stands at 34.7%, compared with BludgerTrack’s 32.6%; the Coalition is at 32.2%, by far the worst result in its modern history, compared with BludgerTrack’s 32.9%; while the Greens and One Nation are at 11.8% and 6.3%, compared with BludgerTrack’s 12.5% and 8.0%.

Not everyone agrees with me about this, but I don’t think it can reasonably be described as any sort of failure on the part of the Australian polling industry. In dismissing the notion that even its 2019 performance counted as such, Nate Silver pointed out that its roughly 3% error was exactly normal by international standards — though this rather glossed over the extent to which the industry’s failure on that occasion lay in the herding-related uniformity in the size and direction of its error. There were at least a few suspicions abroad that something similar was happening this time, and a general reluctance to believe what some polls were seeing — including some mid-campaign blowouts from Roy Morgan that were hardly credited by anyone at the time, but which proved about on the money — may have helped prevent polls and their aggregates from landing nearer the mark.

As it stands, the measure of any given pollster’s accuracy relative to its rivals was a simple function of how high it came in for Labor. As The Guardian’s display shows, line honours were shared by Resolve Strategic, RedBridge Group and Roy Morgan, and trailing the field was the Coalition’s hapless internal pollsters, Freshwater Strategy (apart from an Ipsos poll that was apparently half a point worse in having Labor’s lead at 51-49, which is news to me — the only polling I’m aware from it at any point during the term was limited to leaders’ ratings). Not included in this assessment was RedBridge Group/Accent Research tracking polling, which did very well for much of the campaign in pointing to a 3.5% Labor swing across 20 marginal seats that ended up swinging 4%, only to fall short with a 2% swing at the last. As an indication of how much better the polling industry performed than certain other areas of the media-political complex, the publisher that commissioned this polling persistently instructed readers to share its delusion that it pointed to a Labor minority government.

Another aspect worth noting of the news media’s horse race coverage was its acceptance of the Coalition’s claim that polling was failing to measure a revolutionary transformation in preference flows, such that the precedent of 2022 offered no guide on this score. Pollsters did in fact tweak their preference models in anticipation of weaker flows to Labor — BludgerTrack’s relatively good performance on two-party preferred had a lot to do with its persistence in applying 2022 election flows, which I must confess was more down to indolence than insight. We won’t actually know the true story here for a couple of months, when the Australian Electoral Commission will provide two-party preferred data from the unprecedented number of seats where the two-candidate counts include independent or minor party candidates, and — better yet — preference flows broken down by party.

We do, however, have 118 seats where the two-candidate counts are between Labor and the Coalition, and from which we can observe how well applying preference flows from 2022 would have done in projecting the two-party preferred. And they do in fact suggest that flows to Labor were quite substantially weaker than last time, such that Labor would have scored 55.4-44.6 across these seats on 2022 preference flows but in fact managed 54.5-45.5.