(UPDATE: This was written on Sunday to try and explain how the count would unfold after election night. After a few days rest and recovery, I’ve started another post with rolling updates of the result which I will update at least once each day. For more regular updates I suggest you use the ABC election site which will update results every time the Electoral Commission produces new data. I’m doing work on the site weeding out inconsistencies in the Electoral Commission’s data.)
I’ve been in Adelaide this week working for the ABC on the state election coverage. I may have retired from on camera work, but I’m still involved in analysis off camera.
And last night was one of the most challenging coverages I’ve worked on. I estimate that half of the state’s 47 electorates had a choice of preferences that did not reflect the primary results that arrived on election night. One Nation finished in the top two in around half of the state’s seats. This was very many more seats than the Electoral Commission had prepared for.
That’s not a criticism of the Electoral Commission. It must nominate a pairing of candidate before the count is known, and is guided by history as much if not more than by opinion polls.
But that’s all about last night. Let me describe how the count will unfold from here.
First, the Electoral Commission has abandoned preference counts in seats where it clearly undertook a count between the wrong pairing of candidates.
Second, there are many early voting centres that did not report their first preference counts last night. These counts should be reported today.
Third, once the early voting centres are added to the count, the Commission can make a more reasoned decision on which seats need to have their preference counts re-thrown. And if should do them at this stage when a winning candidate is already clear.
Finally, if it is still not clear who will be the final pairing, the Commission may take further steps to clarify what should be counted.
On a seat by seat basis, this could mean moving to counting postal votes, or undertaking three-candidate preferred counts to determine a final pairing, or abandoning indicative preference counts and leaving the finalisation of the result to the formal distribution of preferences next weekend.
I have no news on which of those options will be adopted by the Electoral Commission for the ultra-complex counts.
But today will be about updating the current first preference counts with unreported early votes. As I understand it, any re-throw of preferences will be considered today but not started until Monday.
With so many complex counts, the first things that needs to be done is the votes counted in polling places on election night need to be received by each district’s Returning Officer. There will need to be some reconciliation of votes against roll mark-off. and a check-count of ballot papers, before re-throws of preferences will be undertaken. That will be Monday’s job.
Everyone should remember, Electoral Commissions have two legal obligations.
- (1) – to undertake an accurate count of first preference votes.
- (2) – conduct a full distribution of preferences to determine the winner, after the primary count is finalised.
The Electoral Commission may undertake other counts to aid in understanding the progress of the count. But no Commission will undertake multiple counts for information purposes if they would interfere with the Commission’s two legal obligations.
The full distribution of preferences in each district are due to be conducted next Saturday. Finalising the primary count before then is the legal obligation of the Commission.
It is in no one’s interest to undertake multiple information counts if they result in a Court of Disputed Returns case that overturned the result.
Thanks Antony. Good clear explanation.
100% on clarity. But if ON received more preference votes than LNP. Should that automatically make ON a two party preferred now and LNP pushed aside as a minor party?
COMMENT: I’m not sure ECSA will do separate two-party counts.
Preference voting works by removing the candidate with the fewest votes first, and redistributing those votes by their preferences. This process continues until a candidate has a majority of the votes. Whilst ON may receive a higher primary than LNP, preferences may actually push the LNP candidate above the ON candidate when they get to the final 3, in which case the ON votes are redistributed.
COMMENT: Labor leads on first preferences in 35 seats, the Liberal Party seven, One Nation three and Independents two. The number of members elected will not be much different after the distribution of preferences.
I love your work. Always clear, concise, and cuts through the media confusion.
So, is it possible for One Nation to be considered as viable candidates in more than the two seats currently viewed in that light, given the focus on ‘main players’ has been incorrect.
COMMENT: One Nation is a strong chance in two seats, and possible in another two. Resolving it requires more counting and possibly re-throws of indicative preferences.
very clear outline of the current election process. Anthony Green is incredibly knowledgeable and its great he is still involved, having 3 parties sharing the primary vote is a new complex. It would be nice to know the overall preference distribution by party.
In many ways the expected runaway victory for Labor has provided cover for this process in this election since government isn’t hinging on the count. The Victoria state election in the fall will be the test when the very essence of government could be tested on seat by seat results. Which makes me worry that the actual counting process may become politicized then creating a real mess.
So much of the counting work is done by hand for obvious and less obvious reasons and to provide opportunities for scrutineers to observe the process on behalf of the candidates. That said, at some point someone will ask the question as to why we don’t build AI code that essentially automates the entire process including the logic for the pairing. A computer could accept the data and perform all of the pairings simultaneously as it counts and produces a result once all of the ballots are entered. Informals and such would be handled manually. I could easily see the day when there will be a demand for a fully electronic voting system that seeks to both automate the counting process and try to reduce as much ballot spoilage as possible. This also means the counting process becomes as much the story as the candidates.
I expect there will be strong demand for full distributions since the will be much analysis going on about voter behaviour. This puts extra pressure on the Electoral Commissions and their work as this process becomes slower as their workload expands.
As always much thanks to the work that you and the Electoral Commissions staff do to provide us with accurate and informative elections!
COMMENT: You’ve got that wrong. It doesn’t need AI and it doesn’t pairings. You simply scan or data enter the ballot papers and you can do a distribution of preferences at any stage of the count. NSW already has the technology to do this and publishes the data during check counts. The ACT already produces interim distributions of preferences under Hare-Clark on a daily basis.
But technology won’t overcome the election night problem. Counting is still done manually in polling places. It would be very expensive to have the technology in individual polling places, and then there is the issue of training staff to use it.
NSW does the data entry and scanning at the check count when all ballot papers have been returned to the Electoral Commission’s central scrutiny centre. As there are no plans to replace polling places and manual counts, this is where technology can play a roll. If you check the NSWEC site, you can do distribution of preferences by polling place by any pairing of candidates you like.
Can you explain about “pairing”?
COMMENT: For the information or indicative preference count, they nominate two candidates as the likely final pairing, and do a provisional distribution of preferences from all the other candidates to the nominated pairing. Sometimes, as in around half of all seats at the 2026 SA election, they nominate the wrong pairing which means the indicative count is not helpful in determining the winner.
The actual winner is always determined by the full distribution to be done next weekend. But the indicative count is enough to resolve all but 6-8 seats at this election.
The Election counts always appear Very Slow!
Pre poll votes should always be counted on the night of the Election.
COMMENT: They are. But sometimes there are so many they can’t finish them.
Pre Poll voting closes at 5 or 6pm on the Friday why cant counting of pre poll start on election day with a strict embargo on the release of numbers until close of polls on election day.
COMMENT: I’ve appeared before a number of parliamentary committees on this subject. There has been extreme reluctance to actually count results before 6pm. The compromise is to open the early vote ballot boxes before 6pm to unfold and sort ballot papers to allow counting to start immediately after 6pm.
New Zealand counts what they call advance votes on polling day. But New Zealand does not allow campaigning on election day, so no party workers outside polling places, and there is an absolute ban on reporting election news or who people voted for on election day. That leaves party workers free to be scrutineers for their party observing the count on the day. Trying to ensure there are scrutineers available is one of the reasons why Australia does not count on the day.
Have they counted postal votes yet? Any idea how many there are?
COMMENT: Postal votes will begin being counted on Monday afternoon. There were around 174,000 postal vote applications which will translate into about 130,000 votes.
Antony, Love your website and your expert opinion on our complex Electrol System.
Just a point from me, the average voter. I received the Electrol Commission SA’s (ECSA)”Official Guide” before early polling began. The copy I received in the post differed from the “final” copy on the ECSA’s website today. I was mislead by the one I received and I went to Turramurra Rec Centre(TRC) Highbury (16/3) to vote (as I understood from the ECSA card that TRC had opened early and anyone could vote). When I checked with reception at TRC they advised the Centre was not open for voting until 21/3 and also mentioned that a huge number of people had attended there seeking to vote. I understood that the Campbelltown Uniting Church Polling Booth was opened so I went there to vote. On arriving, the various “party Supporters” gave me their how to vote leaflets, I took them all into the voting booth with my favorite on top. I hadn’t been given any extraordinary instructions from the ECSA team at the booth. To my dismay when there I realised I wasn’t in my “Morialta Voting Electorate”, I could vote there, but I didn’t have any pertinate voting guide as they were all for the wrong electorate. I went ahead and voted for my “party” but I just chose other candidates that I thought might give my party their best chance to succeed.
As the ranking of candidates votes impacts all parties I believe my vote has been spoilt, to my party’s detriment. I wonder how many other voters have experienced the same outcome.
COMMENT: If you filled in all the squares on your ballot paper with a sequence of preference numbers, the vote would be counted as you completed it. Parties have no say on preferences, your vote is counted as you completed it.
Tom, at the risk of sounding condescending – I don’t mean to be – you should simply list your preferences in order of who you like the most (or dislike the least). There is no point in worrying about giving your preferred party ‘their best chance to succeed’ because if your preferences are being counted, your number 1 choice has already been eliminated from the count and your vote goes to your number 2 choice. If number 2 gets eliminated, then your vote goes to number 3 and so on.
In the electorate that I live in, it always comes down to the two major parties. I vote independent and put my preferred major party second, knowing that my vote ultimately goes to them but that the independent receives the funding from my primary vote.
Even in our system of preferential voting there is room for tactical voting though. Take the case of a Liberal heartland seat with a strong teal independent. Many labor voters may choose to preference the teal independent first in order to ensure that they reach the 2CP and possibly defeat the Liberal candidate.
COMMENT: I wrote about that several weeks ago in this post.
Thank you, as ever.
A hypothetical.
If both the Liberals and One Nation finally got 4 seats in the lower house, how would the official Opposition be determined?
At 4 seats is that below official party status in SA?
COMMENT: A lot of this is governed by convention. I don’t know of any specific SA rules, but the opposition leader is usually the person who the Speaker recognises as Leader of the Opposition. The Assembly can dissent from the Speaker’s ruling. That happened in the NT a decade ago when a mid-term change in party numbers saw the Speaker recognise a different Opposition Leader. Dissent was moved and in a ballot involving only non-government members, the original opposition leader was restored.
I don’t know the rules on parties being recognised in the SA parliament for funding of staff etc. But in Queensland in 2001, when the Liberal Party dropped to 3 members, the government granted funding for them to still be recognised as a party. I would think a similar waiver would and should be granted here.
Great explanation. Would it be better for the Commissions to focus their election night effort at counting as much as possible of the first preference vote so we can get a clearer idea of the tally? As I review their website at 4:30p in Melbourne, the ECSA has counted 96.5% of polling booths and 79.7% of early voting centres with a total of 58% of enrolled votes counted. Could we increase that % significantly by ignoring the preference flow entirely?
We could then supplement first preference voting with a sampling of distributions across the state, randomly selecting a sample from each seat. So for example, 80% of Greens voters preference Labor over Liberal, 80% preference Liberal over One Nation, 70% Independent over Liberal. Experts could then use the state sample and apply that to specific seats. Therefore in Seat X if Greens voter preference Labor at the state sample average, Labor will be within 100 votes of 2nd place so too close to call, but in Seat Y that would put Labor 1,000 votes ahead of Liberals so will be safely in 2nd place. Using a state sample would not be perfect but would better indicate the uncertainty and stop us wasting time re-throwing distribution.
Look at Heysen: currently Other 4.8%, ONP 14.5%, Green 23.0%, ALP 24.4% & Liberal 33.4%. I wonder if ONP voters will preference Greens ahead of the “uni-party” allowing them to move into the 2CP and win the seat. If we had a sample of ONP Distributions we might have a better estimate of that. It would be helpful to know today what % of Liberal voters preferenced Labor ahead of One Nation ignoring the HTV cards.
This would not solve a situation like Calwell in 2025, in that case waiting for a full distribution is the only solution.
COMMENT: There is no reason preference counts should interfere with primary counts. The primary count is done first. After that an indicative preference count is done. Why so many primary totals were not reported is a question to be answered. There were very few early voting centres reported on election night. I understand many were not finished. There will be questions asked whether there were enough staff to complete counting give the huge volume of early votes.
Some sampling has been done since Saturday in certain seats to re-set results for a changed indicative preference count.
So One Nation is coming second by they don’t have any seats won, why?
COMMENT: One Nation is second in the state vote. It may win up to four lower house seats. The party has polled well in a lot of seats where it has finished well back in second place, hence lots of votes and fewer seats.
It is doing better in the upper house where proportional representation is used and the party’s vote is translated more directly into seats.
.
Glad to see you’re still around Antony, explaining the logistics of the election for us lay folk.
Given the large (and seemingly increasing) proportion of early voting, do you know if the AEC has any plans to adapt to the new dynamic?
COMMENT: Discussion continue. The ACT has moved to electronic pre-poll voting where there is an advantage due to the use of the Hare-Clark electoral system. But there is resistance to it elsewhere. I see electronic early voting restricted to early voting centres in the future.
There has been a large increase in the number of “non-traditional” 2CP counts with One Nation appearing in a majority. One thing that has been bugging me in election coverage (including on the ABC website) is that many commentators are referring to a 2CP “swing” in these contests, e.g. the 2022 result in a seat was 55% ALP/45% LIB, the 2026 result is 60% ALP/40% ONP, therefore there was a 5% “swing” to the ALP. In my view talking about a “swing” in this situation is just statistically wrong as they are totally separate metrics. If we were talking about apples and oranges last year you can’t talk about apples and bananas this year. Do you agree it’s incorrect to talk about a swing in these circumstances if the 2CP parties have changed since the last election and any thoughts on how to correct this.
COMMENT: When I produce election publications I always deal with this by showing three totals in the after preferences tally. The candidate who dropped out of the 2CP has a huge negative swing to zero. The new candidate in the 2CP has a massive positive swing up from zero. And the candidate present in both counts has the useful swing. That is the 2CP swing shown in television coverages. Television coverages can choose to show no swing or this swing and the choose to show the swing. The world is not a perfect place, especially in the brief hours of an election coverage.
Thanks Anthony. Glad to know you’re on the job even if not in front of a camera. On the night your mix of competence and quiet enthusiasm is missed nonetheless!
NSW and my local council do electronic distribution of preferences, Federal does manual distribution of preferences, what does South Australia do?
COMMENT: In the lower house of all states except NSW lower house preferences are distributed manually. All state upper houses are scanned and data entered and preferences distributions done the electronically.
NSW is the only state to data enter lower house ballot papers for distribution. Optional preferential voting makes the data entry process easier as single ‘1’ votes are entered as a bult total. Victoria has experimented with data entry doing 7 lower house electorates in 2022.
If South Australia used the Hare-Clark Electoral system, would the number of seats won by each party this year be dramatically different?
COMMENT: Hare-Clark is a form of proportional representation. It is designed to produce proportional results. Our lower house single member electoral systems are not designed to be proportional so yes, Hare-Clark would produce a different result. Just look at the upper house result which uses a related form of proportional representation.
Excellent. Keep up the good work. ABC Live coverage is not the same any more.
Thanks Antony. I had to do some reading about the candidate pairing and found a very good description on the AEC website under “Complex Count Scenarios Explained” tab. I’ll be very interested to see the pairings list when it’s available. Love your work.
COMMENT: The pairings are available and have been since 6pm on Saturday night. Every two-candidate preference count you see on the ECSA site has been produced by a preference throw to a pairing chosen before Saturday night. With first preference counts now available, those counts will change today to reflect the primary votes. In particular there will be an increase in the number of indicative preference counts involving One Nation candidates.
In law none of these counts will determine the winner, merely indicate the winner. By law it is the full distribution of preferences, where candidate are excluded one by one, to be carried out this coming weekend, that will in law determine the winner. The indicative preference counts are only done to provide a better understanding of who will in advance of the official count.
Glad I found your blog. Your explanation adds clarity to our discussion thank you.
We missed you on Saturday night on our televisions, but so glad you were there in the „back room“. Casey nailed it out front. Our family commented that you would have been loving every second of the democracy in action and the chaos which it threw the system in to. Especially loved the ABC tech issue in the throw over to Pauline Hansen, the cheeky capture on return to the studio before crossing back to the live coverage at ON HQ. We were laughing all the way to the chaos of the calls of the seats in the balance!!!!!
Thank you for succinct concise reporting that the layman can understand.
Please add my thanks to you for how informative you are, Antony. One thing that concerned me was that before the election there was no facsimile of the Legislative voting sheet, just a list of the candidates according to the column they were in. If you’re voting above the line, this is not a problem, but if you want to research for yourself each candidate’s/party’s policies and then number them below the line, there’s no easy way to do this. You have to make up your own list of how to number each candidate instead of just filling in a facsimile of the voting sheet and taking that with you into the polling booth. To me the Commission’s withholding of a copy of the voting sheet is a disincentive to vote below the line and, as such, a diminishing of democracy.
COMMENT: The Commission had a version of the ballot paper on its website, the first time I have seen an Electoral Commission produce one.
To Antony – yes, it has the start of a sample ballot paper, but it did not have a specific one for this election.
Thanks Antony, so great to know you’re still working behind the scenes at the ABC.
Just wondering what you think of the Greens’ chances in Heysen? It seems to me the AJP and Independent preferences should push them into second place above Labor, and then it will come down to Labor preferences in the 2CP? Do you have any thoughts on whether Labor preferences are likely enough to get the Greens across the line? Many thanks!
COMMENT: The issue is about the Greens getting into second place before anything else is worth considering.
Thanks sincerely for your analyses – always much appreciated; love your work.
Pardon my ignorance, but why have you used that verb “throw” ?
Is a ‘re-throw’ a more formal re-counting of actual second preferences (rather than some [mistaken] assumptions about the preference trends from previously), or is it perhaps a revised [guess]-estimation, pending the full final count by next weekend ?
Whatever, I am not clear how such Electoral Commission “throwing” is either simply a source of delay in formal counting, or a less than helpful distraction from that final count declaration ?
Cheerio.
COMMENT: It’s a scrutineering term. To calculate two-candidate preferred numbers between a final pairing of candidates, scrutineers used to look closely at votes for all but the final two and estimate the number of preferences to the two final candidates. Since 1993 Electoral Commissions have done this after finalising the primary count.
A final pairing of candidates is nominated beforehand and all other ballot papers examined and a two-candidate preferred tally calculated. This process is often called a throw as the votes for other candidates are placed into separate piles before being put back in the bundles they came from.
Antony Green,
I have made an informal study of Australian, British and American politics for around 30 years, and taken an interest in Asio-Pacific geopolitics for at least the last ten. I was co-president of Parliamentary friends of South Pacific Nations, served on 3 committees and responsible for several pieces of Vic legislation. In all that, there has been only one voice that I could definitively, absolutely rely on. One person who’s knowledge of every state and federal system is unsurpassed and unbiased.
That sir, is you.
My admiration for your work is deep, and I sincerely hope you will continue this work, as I see you as an “Essential Service”. If I may be a bit cheeky, we should perhaps have you enshrined in the Essential Service legislation, so that the country has a continuous source of intelligent commentary.
With so many people now voting early and increasing numbers moving away from the traditional Labor and Liberal/National parties – and even those who vote for them less likely to follow preference distributions – we need to have a new way of predicting final outcomes.
COMMENT: You mean a method that doesn’t involve counting the primary votes and counting out preferences?
The existing Electoral Commission process to nominate a pairing of candidate before the count is known, is no longer relevant in a 3-4 major parties Australia. Labor and Liberal are no longer the automatic defaults. We should only get the primary votes at the night of the election and quickly (not a week later) progress to fully count the preferences with winners only announced at the end of the process.
COMMENT: Winners are already only announced at the end of the process. The indicative count was introduced so that people had official figures to use rather than rely on figures from scrutineers. The South Australian election is unusual in that so many pairings were wrong because One nation came from nowhere. At most elections the correct pair is chosen in 90% of seats so there is no reason to abandon the process.
So it seems likely that there will be a total of 5-7 ONP members in the new SA Parliament. I recall a piece you did a few years ago which showed a massive drop-out rate of over 70% of ONP MPs leaving the party for various reasons before the completion of their terms. It will be interesting to see if history repeats in SA!
In the seat of Ngadjuri, it looking likely the ONP candidate will be elected, knocking out two sitting MPs. Have two sitting MPs ever been knocked out in the same seat before?
COMMENT: I don’t know.
Thank you Antony for still providing your wisdom. The ABC coverage was very good as usual but the viewers missed you.
BTW the interview with the McBride bloke was a window into delusion if there ever was one. Yes, we live in a society where we have the right to be tried with a jury of the public and the presumption of innocence but to hear this bloke saying he would have done much better if these `issues` he`s been involved with hadn`t been so distracting was something to behold. I suppose being charged by the police on serious charges like DV is some kind of inconvenience when running for elections in McBride`s world.
I`ve been intrigued by the reference to the Hare Clark preferential voting system Antony – I remember it was the voting method for student polly elections at university. Can you give us any insights into what general differences it could have made to this 2026 state election result?
COMMENT: Hare-Clark is a proportional electoral system that produces results where members elected are roughly proportional to votes received. The current single member electoral system is not proportional. So Hare-Clark would have produced a more proportional result than we saw at this election. See the Legislative Council which uses proportional representation.
Second can be viewed in how far behind you are First… in many seats there was a swing to the Government putting more and more distance between First and Second from the last election.. Maybe we should do it by centiles (1st to 9th).. in many seats ON came 7th or 8th in 1PC
Are records of informal votes kept? How many of these are empty ballots and how many are dick picks?
COMMENT: There is a report done on informal voting after every SA election.
I understand the Indicative count was introduced as a means of getting a “quick” indication of the results. Wouldn’t it be better to just get on with the real count rather than waste time on someone that isn’t necessarily correct.
COMMENT: You can’t do the real count until all primary votes are counted, and that has to wait for the close of postal voting, after 1 week in SA or 2 weeks for federal elections. The counts are done otherwise scrutineers will do it and these unofficial counts circulate in place of Commission data.
Antony
re electronic pre-poll. This was trialled in 2021 council elections & produced lots of stuffs & 2 by-elections in close results. The reported case is:
NSW Electoral Commission v Kempsey Shire Council (No 2) [2022] NSWSC 282
COMMENT: No, that was iVote, internet voting. That is why iVote was abandoned. Unreliable, not in-person voting and required running software on platforms that weren’t under control of the Electoral Commission. The government instructed the Electoral Commission to extend the use of iVote for the local government elections to cover absent voting, not its design as a replacement for postal voting. It caused all sorts of problems with people trying to register very late causing problems in connecting the registration and the enrolment system resulting in people being unable to vote.
Electronic pre-poll would be in-person voting in early voting centres using software and hardware under the control of the Electoral Commission. Very different to the iVote experiment.