The vote count is now getting to a point where pretty much all primary votes have been counted, even if we don’t have all of the preference counts yet. So that means we can now begin to analyse the informal voting trends – where it went up or down, and what some of the causes could be.
I also wanted to look into a story that appeared yesterday in the Sydney Morning Herald which claimed that there had been a surge in informal rates in voting at special hospital booths in 2025, but seemed to be mostly focused on a handful of seats.
Right now the informal rate has increased from 5.2% in 2022 to 5.6%. This would be the second highest rate on record, but only just ahead of the rates at two other elections.
The first thing to check is how the informal rate compares to the size of the ballot paper.
There is a very clear relationship, but there is some nuance worth noting.
There is obviously variation at each level – ballot size doesn’t explain all differences in informality. The most obvious difference is that NSW has higher rates of informality for any given ballot size, and that is particularly true for seats in Western Sydney. Past research has identified communities with more voters of non-English speaking backgrounds as having higher informal rates. The chart above shows how McMahon, Macarthur, Blaxland, Fowler, Chifley and especially Werriwa and Watson stand out from the pack.
The previous record for the highest informal rate in a seat was Scullin in 1984, with 14.09%. Fowler came close to this level, and Werriwa and Watson easily broke the record.
The uptick in the informal rate particularly appears when you get to eight candidates on the ballot. This has been noticeable since the change in the Senate electoral system in 2016, which introduced the recommendation that voters number at least 1 to 6. If you follow those instructions in the House of Representatives, your vote will be formal on a ballot paper with no more than 7 candidates, but informal on a larger ballot.
Next up, I wanted to look at how the change in the informal rate compares to the change in the size of the ballot. In the case of redistributions, I have weighted the 2022 ballot size to the parts of each old electorate. So, to take an example, the seat of Bradfield had 7 candidates, and North Sydney had 10 candidates. Just over a quarter of the new Bradfield came from North Sydney, so I give it a score of 7.8.
The relationship is even stronger than with the current size of the ballot paper and informal rate. Overall the national enrolment rate increased, but so did the average ballot paper size. Four Western Sydney seats stand out as having even more of an increase of their informal rate than the change in ballot size would suggest.
If this theory is correct, there is not necessarily any change to voter behaviour driving higher informal rates – it’s a natural consequence of increasing numbers of candidates making the act of voting under our electoral system more difficult. But that clearly doesn’t explain everything.
With Bradfield being so close, there has been a lot of interest in that seat. This led me to mapping out the change in informal vote across Bradfield, and for reasons that will become clear in a minute, to also do the same for the two other seats which took in parts of North Sydney.
Bradfield and Mackellar each had seven candidates in 2022, in the range where a 1-6 vote is formal. Bennelong had eight candidates, and North Sydney had ten. All three seats had eight candidates this year.
This means that all three seats have taken in areas that previously had an 8+ ballot paper, while the remainder of Bradfield and Mackellar had an under-8 ballot paper.
This ends up being very obvious on the map. I’ve highlighted the former boundaries of North Sydney in light blue.
When you look at the booths of Bradfield, there is a clear trend where the informal vote has increased in the areas already part of Bradfield, but not in the former North Sydney. The same trend can be seen in Warringah, but less so in Bennelong which already had a ballot of eight in 2022.
There is one interesting theory about some of these increases in informality, which is due to the emergence of new groups publishing how-to-vote cards, whether it is a teal independent like Nicolette Boele, who issued a how-to-vote card that would be informal if a voter copied it exactly, or added complexity in seats like Blaxland, Watson and Werriwa where Labor’s domination has been challenged. I can’t prove this theory, but I think it’s interesting.
The AEC was asked about informal votes in yesterday’s media briefing and they confirmed that there will be a survey of informality following this election. I believe this was last conducted following the 2016 election, and yielded very useful data on the exact ways in which voters marked ballots that ended up informal.
Finally, I have attempted to holistically analyse informal rates amongst special hospital teams – how high the rates typically are, and how much they have changed. The SMH story appears to have focused on particular electorates, but with such small numbers of voters you could expect particular booths to bounce around even if the national trend doesn’t change much.
The first thing to note is that hospital voting basically did not exist in 2022. While about 80-90,000 votes were cast in 2016, 2019 and 2025, it was just 4,000 votes in 2022. So it is fairer to compare 2025 to 2019, which the SMH story did.
Special hospital teams has historically been one of the types of voting with a higher rate of informal voting. The rate has increased, from 7.3% in 2019 to 10.5% in 2022. That rate has been repeated in 2025. So yes, the informal rate has gone up, but by nowhere near the rates suggested in the Herald story.
Bradfield in 2019 had over 1,000 hospital votes cast, with an informal rate of just 1.6%. This is compared to an informal rate of 16% in 2025.
But that 2019 statistic was an extreme outlier. Excluding two seats where a very small number of hospital votes were cast and no informal votes were included in them, Bradfield had the lowest rate in the country. The 2025 rate is also quite high, but not right at the top. Twenty other seats had a higher informal rate than Bradfield amongst the hospital booths in 2025.
So yes, there has been an increase in informal voting in the hospital booths, but most seats have not seen an increase anywhere near as big as Bradfield, which partly looks bad because it had such a remarkable result in 2019.
When a race is coming down to the wire like it is in Bradfield, any factor can be enough to tilt the result. But I don’t see a reason to think that informal voting in the hospital booths is a national problem or is significantly worse than it was in the past.
Hi Ben,
Thanks for your analysis but it doesn’t mention a likely influence in NSW because of the difference in preferential voting in NSW State and Local Government elections. In those elections it is optional preferential but for Federal it is compulsory preferential where a number needs to be placed in every square. As a Labor volunteer it is incumbent on us at polling places to remind people that they need to complete every square in the House of Reps ballot paper. Maybe the AEC survey of informality will shed some light on this.
Personally I believe that all elections should be compulsory preferential as this would make the will of the voter clearer. In the Sept 2024 Local Government elections for Mayor in the Bega Shire this was a major contributor to the ALP candidate falling short by 35 votes to beat the Lib candidate on 2PP as so many votes for the greens candidate exhausted because there just a one in the greens square. The ALP candidate won the primary vote quite clearly.
I would appreciate your thoughts on this.
David
Yes, I did mention that NSW has a higher rate than other states and that is obviously because of the use of OPV for state elections. To be honest I didn’t go there because it’s not a new finding.
I don’t know how you can look at all the damage compulsory preferential voting does in throwing out so many votes and think that the problem is with the one state that doesn’t do that.
This sort of analysis can drive funding for information and education into particular electorates at election time.
In addition, perhaps extra information should be provided to those electorates with 8+ candidates standing.
Describing CPV as the “will of the voters” is a bit of an odd one. Maybe those Greens voters in Bega Valley didn’t want to preference Labor? I do agree that some voters are sometimes misled into voting just ‘1’ when they have a genuine preference but there’s plenty of others who don’t want to and are forced to mark preferences so their vote counts.
Optional preferential sounds the way to go, that’s essentially the Senate voting system anyway.
What’s wrong with the voter picking the one theywant and leaving it at that?
The only beneficiaries of compulsory most times are the major parties, which 33% don’t want anyway.
I think CPV also means elections are a contest of both parties trying to win people who don’t really have a preference either way, instead of putting out actual ideas for engaged people to debate.
The increase in informal voting also correlated with:
1. The increase in turnout. This means some voters didn’t vote last federal election, including first-time voters.
2. Teal independents. Most of them have a 1 next to their name on the HTV card and that’s it.
3. Increased appetite for non-major parties and disengagement.
“I don’t know how you can look at all the damage compulsory preferential voting does in throwing out so many votes and think that the problem is with the one state that doesn’t do that.”
Given that Federal elections are CPV, the inconsistency of one state having optional at state level obviously causes more formals within the CPV system, that’s not an unfair point.
Your point that these votes would not be informal in the first place with OPV is also fair. That said, the informal rate at NSW state elections is not that far below the informal rate in federal elections, and that’s with the informal rate in federal elections being bulked up by this NSW confusion issue. I wonder- if NSW seats had the same informal rate as Victorian seats, what would the federal formality rate be?
Thanks Ben for this research. It suggests the AEC needs to do targeted civic education in Western Sydney, and also that NSW’s inconsistent voting system at state level is a problem. NSW needs compulsory preferential voting for consistency’s sake as much as for the other reasons for it. The best-of-both-worlds option would be consistent use of CPV with a savings mechanism, as this system satisfies both the interests of deciding the election upon the preferences of as many voters as possible as well as the interests of saving partially completed votes.
Also, the theories about independents’ how-to-vote cards appear unfounded. In NSW, Wentworth, Warringah, Mackellar, and Bradfield are all low for informal votes relative to other NSW electorates with the same numbers of candidates. It’s a similar picture for Clark, Curtin, Kooyong, and Goldstein relative to all federal electorates with the same numbers of candidates. Meanwhile Indi and plenty of others like Wannon and Dickson are middling to low too. And the cluster of western Sydney electorates appears consistently relatively high for informal votes no matter whether there are prominent independent MPs or candidates (as in Fowler, Watson, Blaxland) or not (as in Werriwa, Chifley, Macarthur). There doesn’t appear to be a correlation between independents and informal voting.
I remember working a state election as a polling booth official. We had an aged care facility nearby, and they had had problems getting to patients. Some turned up at the booth, and I took their votes. Fortunately it was optional preferential voting so they could just vote 1. However, it would have been very hard work in the Hospital and Aged Care homes, especially where all squares are meant to be numbered. My only surprise is that the informal vote is lower than I expected from the article.
I strongly disagree that NSW needs to change its system to deal with this problem. It’s the federal system that is excluding votes for unnecessary reasons. Why should NSW need to change?
Insisting on consistency between levels of government is a recipe for never innovating or trying new things.
CPV says that if you do not express all of your preferences, then none of them will count – your vote will be thrown out. A voter who votes “just one” has very clearly expressed a preference for that candidate, yet this is discarded – not to mention a voter who numbers six of eight candidates! My view is that this is tantamount to institutionalised voter suppression. I strongly support preferential voting and understand the desirability of voters numbering as many candidates as they can. But disenfranchising voters to this end cannot be justified.
Ben Thanks for this. A proper study on informal is a prelude to reform before the next election. In other words AEC should be told to get on with it quick so the polls can think about it quick.
Same trend in Whitlam with 10 candidates the informal vote was 9.81% compared to 6 candidates in 2022 the informal vote was 4.9%. Ballot size obviously matters. Whitlam is also interesting because it covers 4 different distinct regions:
Shellharbour region
Dapto region (both state seat of Shellharbour)
Albion Park/Calderwood/Tulimbar (AP region State seat Kiama) which has now broken away socio/economically from the Shellharbour region with huge new private developed housing estates &, of course,
The Highlands.
The divergence in informal vote is huge with the Highlands in the range of 6 – 7.5%, AP in the range of 10 -11.5%, Dapto region in the range of 11 – 15.9% & Shellharbour in the range 13.2 – 15% (Barrack Heights 18.10%, Mt Warrigal 14.25%, Blackbutt 13.8%, Oak Flats PS 13.3% & an outlier Oak Flats/Balarang 11.28%).
Needless to say the informal is highest in safer/safest labor areas. I do not hasten to say lower socio-economic because you could drive around Shell Cove & Oak Flats & Lake Macquarie waterfront & see better housing & views than Cronulla. Resident just don’t know it?
Obviously some rules about say making a vote with six numbers formal would help. (Incidentally to diverge I see that the conservative leader lost his seat in Canada after being opposed by 79 other candidates in what I think was a protest for electoral reform). Instead spending of $ 80 million on advertising it will surely occur to Clive Palmer that he could gum up the works against Labor more easily in mixed geographic/regional seats like Whitlam be paying nomination fees for heaps of candidates in each seat.
& he can evade the new limitations on advertising funding?
I believe from careful observation as a scrutineer that more candidates encourages voters to simply spoil their paper or not touch it all rather than failled attempts to vote formal. The AEC study will show if this is general, socio-economic general or my booth is just unique?
.
Ben a final thought which you may or may not want published.
A final local thought that concerns a local phenomena which is the plethora of Facebook My this & My that (over 50) alone in Shellharbour/AP region which are often just toxic in content. This is not a phenomenon seen in cities (I think). I certainly don’t remember a “MY Rushcutters Bay” or “MY GLEBE”? Almost all of these semi-closed sites do not follow the constitutional law & allow freedom of political communication & I think that this is another example of the internet eroding democracy in that they encourage contempt for all politics in the end & informal voting. For instance one site had a comment “I voted for a penis without balls on my ballot paper and my candidate won”. (a reference I think to Albo not Carol Perry)
The constitution does not oblige internet forums to allow anyone to post anything on their pages. I don’t see anything undemocratic about that, just as I am free to limit what is posted here.
My seat of Cowper went from 4%+ informal 2022 to 9%+ 2025, with an increase of candidates from 8 to 11. Almost successful Teal had an open ticket – ie. htv was number 1 then number all squares – which would not have helped, but she did this in 2022! Go figure – does seem counting / deciding who is palatable, past say 6 is a problem. The Nats tagetted negative campaign (‘preferences’ and ‘rich greenie bloke’ finding her millions $) made it very hard for her to decide to mark preferences.
I had noted the big increase in informal in my electorate (Mackellar) from 2022 to 2025. Close to doubling being up 3.65 % points.
I am one who wonders about the open HTV card, the same as in Bradfield. If only we could see who the in-error informals “voted” for …
It didn’t hurt Scamps who increased her vote (she got a bit of redistribution help) but may have cost Boele. Informal votes also rose a lot from 2022 to 2025. Boele put question marks in the boxes in 2022, not this time I think.
Those talking about optional preferential voting…I say no. If it’s a preferential system, then it should be every box ie decreasing preference.
There’s a loophole in optional pref voting the Liberals exploited at the NSW election where they faced teal independents. It could apply other places where the threat was an independent. They told everyone (signs, exhaustively on the day) in Pittwater and I believe on the north shore that they only needed to number one square.
This is true, but irrelevant for Liberal voters as the candidate will always finish first and at worst second, so their preferences aren’t used. It was a gambit to encourage Labor and Green voters to exhaust their votes rather than give prefs to the teal candidate Jacqui Scruby and it worked. They won narrowly by far less than the exhausted votes x % pref flow to Scruby.
She was subsequently elected at the by-election after the resignation of the Liberal MP as it happened, but that doesn’t change the Liberal cynical gambit.
Is there any suggestion or sign that the increase in informal votes was a concerted ‘donkey vote’?
The worst effected seats in Western Sydney seem to correlate with the groups worst affected by COVID lockdowns and containing the highest Muslim vote likely to be influenced by Gaza and I wondered if there had been any signs of a public push to disengage with the compulsory voting process, rather just than accidentally informal votes?
With our voting system it seems very difficult to disentangle deliberate disengagement from accidental and the assumption appears to be the latter but it doesn’t seem beyond question that effective absenteeism (while avoiding the fine) would be increasing alongside the shift away from the majors. The potential remedies would appear quite different where this to be the case.
@Ben re: your 1:10pm comment:
“It’s the federal system that is excluding votes for unnecessary reasons.”
That’s an ideological rather than an evidence-based argument. Your article is great, but it doesn’t include any data on the numbers of votes that were excluded due to numbering errors verses the number of votes that were excluded due to voters deliberate choices to vote informally, so there is no data here that addresses the issue of how many votes were excluded due to mistakes.
“Why should NSW need to change?”
Because there clearly is data here that evidently says that NSW is over-represented in informal voting. A NSW-specific problem requires NSW-specific solutions.
The data is also showing us that the numbers of candidates is also linked to rates of informal voting, and that could also be addressed with specific solutions, such as Julian’s excellent suggestion for targeted education in electorates with 8+ candidates.
“Insisting on consistency between levels of government is a recipe for never innovating or trying new things.”
Sure, but personal preference for inconsistency is cold comfort for the people in 11 electorates where more than 1 in 10 people voted informally, and where 10 of 11 of those electorates were in NSW.
It’s an education problem. We don’t say we need the same candidates on all levels so voters don’t get confused, we don’t need the same system either. Voters aren’t stupid, they just don’t know how to vote, so the solution is to tell them, not to dumb down our democracy.
OPV proved itself at the ’12 & ’15 Qld elections, imo.
In 2012, voters had had enough of 14 years of Labor and responded to Just Vote 1.
3 years later, LNP ran with that slogan again and lost 35 seats, it was all over at 6:20 pm. Here’s AG’s analysis: https://www.abc.net.au/news/2012-11-28/the-impact-of-optional-preferential-voting-at-the-2012-queenslan/9390104
Thanks for your insightful writing.
From the perspective of someone who scrutineers, the largest number of informal ballots are blank, followed by ‘1 only’, then duplicate numbering, followed by indecipherable numbers.
Very few are 1-6 – under the assumption the voter was mistakenly using Senate numbering rules in HoR.
A driver of informals not discussed is a product of HTVs, when a voter decides they wish to alter part of their preferred party’s HTV. This is the most common driver of duplicate numbers, not voters consciously indicating a tie.
Obviously the benefit of HTVs is fewer informals from low engagement voters, and those with language / comprehension difficulties. However more could be done to alert voters to potential change of numbering issues.
I looked at two historical examples – Lowe in 1972 when there were 9 candidates in Billy McMahons seat when he was PM and Werriwa in 1974 when there were 13 candidates or 12 running against Gough Whitlam. Both were exceptional candidate numbers at the time. In Werriwa the informal vote was 2.5% and in Lowe 2.9%. Were voters more numerate in the 1970s? In both those cases, beyond the ALP and Liberals the other candidates polled very poorly. Was that a factor?
There seems to be periodic spikes in informal voting. Why in Werriwa was there a spike from 3.0% in 1983 to 7.9% in 1984? And in Lowe there was a similar spike in 1984 as well. I do recall there was a similar issue in the 1980s particularly in Western Sydney and voter education seemed to be the answer.
In those electorates with a high CALD population and high informal votes it would be interesting to know if there is an issue with the numbers 1 and 7. These are two numbers that people in various countries are taught to write differently particularly if they put a stroke on the 1. It is possible that polling officials see two #1 or two #7 or can’t differentiate which is meant to be a 1 or a 7. Hence confusion past 6 candidates.
European 1s and 7s are recognised as valid, as well as roman numerals.
I’d say more people are expressing an opinion on a voting system that makes no sense to them. Reading earlier comments, it sounds like the will of the voters in Bradfield was to elect Boele, so she’d be entitled to feel dudded if she loses, imo.
It is a statement of fact that the *federal* electoral system is the one that excludes votes for not being complete even when a first preference is clear.
“Sure, but personal preference for inconsistency is cold comfort for the people in 11 electorates where more than 1 in 10 people voted informally, and where 10 of 11 of those electorates were in NSW.”
Sure, so the jurisdiction that is excluding those votes should fix it. NSW is not the jurisdiction doing this.
I think education has a role to play but there is also a question of practicality. When you have to number nine boxes, some people are going to make mistakes no matter their education levels.
I do think the higher levels of how-to-vote obedience in the 1970s probably helped too. A lot more voters mark their own preferences now and some of them stuff it up.
“Your article is great, but it doesn’t include any data on the numbers of votes that were excluded due to numbering errors verses the number of votes that were excluded due to voters deliberate choices to vote informally”
We know the data on this, but there is no fresh data. I will analyse them when the informality survey is conducted. The last time there was a survey in 2016, roughly half of the informal votes appeared to be deliberate, the other half accidental.
The other issue is that maybe voters don’t know that they can go back and get another ballot paper if they make a mistake or they may be intimidated by queues etc. In the days before 1-12 I went back twice once because I had made a mistake voting below the line.
The preference issues are more complex than the NSW Optional Preferential System and the compulsory preference systems used everywhere else in single-member lower house elections.
The NSW system is unfortunately embedded in the NSW Constitution. It requires a referendum to change it, but there should be scope to get the ballot instructions to be more encouraging of additional preferences. The current wording can be found at https://elections.nsw.gov.au/candidate-handbook-nsw-state-by-elections/nomination/legislative-assembly-ballot-paper – it could be more like “Number the candidates in the order of your choice, starting with 1, then 2, 3 and so on. You may leave some boxes empty if you wish.”
The full-preferential systems can be improved by moving toward the current Senate arrangements, which require ballot instructions and electoral material to indicate the need to number at least 6 boxes above the line (or 12 below), but where savings provisions ensure that ballots with fewer numbers are still counted. A possible alternative would be to require instructions and electoral material to specify that at least 6 (or maybe 5) boxes be numbered, but with savings provisions to count ballots with fewer boxes numbered.
There will be some exhaustion of votes, but this would almost certainly be balanced by the inclusion of significant numbers of presently informal papers.
Antony Green wrote a paper on the political impact of OPV in the 2019 NSW election – https://antonygreen.com.au/preference-flows-by-party-2019-nsw-election/
It is absurd and unjust to demand that NSW changes to CPV merely because it leads to confusion in federal elections. Western Australia since the mid 1990s at Legislative Assembly elections allows a break in numerical sequence, so long as there is a 1 and numbers in all but one square. This was introduced to prevent the injustice of a vote being invalidated just because there was a mistake in sequencing multiple candidates. It may also allow voters to deliberately negate preferences “2..2..2” but frankly who cares. The number of exhausted votes, even though excluding the votes of the two finishing candidates, is very small.
I noticed that a lot of postal votes (1500 or more in some seats) were not returned to the AEC by the deadline. Is this a growing problem? What might be driving this?
@Warren,
That is not a problem. Not every postal vote application turns into an actual vote. Those voters may have voted in person. It is perfectly normal for the number of applications submitted to exceed the number of votes returned.
@redistributed – I would suggest in the 70s people may have been far more likely to just take a how to vote card and follow it.
@ben – the fact that NSW still has over 3% informal with OPV still raises the question of how many votes are really being thrown out by CPV which wouldn’t be thrown out by OPV, especially if we didn’t have the NSW confusion issue inflating the numbers. If we had some better savings provisions, it might be a vanishingly small problem that doesn’t require the entire country to change to OPV to solve.
My problem with OPV is it encourages a just vote 1 mentality with no preferences which is effectively a migration back towards FPTP. It might be hard for anyone to vote informal in FPTP that doesn’t mean to but that doesn’t make it a better system.
It’s on the basis of the AEC’s historic data that informality in special hospital team booths at least appears to have tripled. Look at the informal swings recorded for SHTs on the AEC site, particularly in NSW, and you’ll soon see what I mean. Presumably the issue here is how the AEC determined its historic informal data, which in turn presumably has something to do with the fact that they serviced fewer hospitals in 2022 because of the pandemic (though beyond that I can only speculate).
CPV is the only way to get voters to pay attention. In a democracy, the ability to vote is a right, but it is also a responsibility. In the old days, when the two parties collected 80-90% of the primary vote, you could be a bit lazy about things because it was mostly one or the other. These days, some seats, such as Calwell, where the two-party primary vote was 46%, are like this, where it takes some time into counting before we know the top two candidates. As such, it is doubly essential for candidates (and their parties, if they have one) to issue an HTV card and for voters to follow it. At the very least, to fill out the ballot paper correctly. It is not as though this is a new concept unless you are a first-time voter. There may be a tendency for an enthusiastic teal or Greens supporter (to use two examples) to single-shot vote https://en.wikipedia.org/wiki/Bullet_voting without preferences, thinking that by doing so, they will maximize their effect, but to no avail. Preferences will likely only continue to grow in importance in the future, as the first preference curve continues to flatten.
In Bradfield, I would love to review the informal votes that preferred the teal candidate as #1 but were discarded because they were incomplete.
While I support every available means to ensure access to the vote, ultimately, it is the voter’s responsibility to do their due diligence.