One of the most common arguments regarding Australian electoral systems concerns whether voters are required to number every box on their ballot paper in single-member elections. There are principled arguments for two different methods, but often these arguments are backed up by the perceived self-interest of someone’s preferred political party. For today’s post, I am going to present the case for why both systems have downsides, and why we should do the work to develop a new system that takes the best of both systems.
Compulsory preferential voting (CPV) is used for most single-member elections in Australia, including for the House of Representatives and lower houses in four states as well as the Northern Territory Legislative Assembly. Voters are required to number every box on the ballot paper, and savings provisions are limited.
Optional preferential voting (OPV) is used for the NSW Legislative Assembly and is also common in some states at the local government level, including for the City of Brisbane. A vote is formal if there is a single clear ‘1’ preference, and further preferences are optional.
CPV has been used at a federal level for over a century, and was originally introduced by the conservative Nationalist government led by Billy Hughes as they faced the problem of split votes with the nascent farmers candidates who would soon coalesce into the Country Party.
For much of the twentieth century, preferences tended to favour these parties more than Labor. Labor governments introduced OPV in New South Wales and Queensland. Not as many preferences flow under OPV, so a party that relies less on preferences from other candidates will do better under that system.
The relative positions of the parties changed with the decline of One Nation and the rise of the Greens in the 2000s. Nowadays you are more likely to find progressives as being strongly supportive of CPV, and conservatives more likely to support OPV. The conservative Country Liberal Party implemented OPV in the Northern Territory prior to the 2016 election, but the new Labor government reversed this decision in the following term. The Queensland Labor government also implemented CPV in Queensland in 2016.
Last Friday’s blog post explored the perceived self-interest of parties in advocating one system or the other specifically in the context of Queensland. I argued that the resurgence of One Nation is making OPV less favourable for the Liberal and National parties. Queensland in particular now looks like a state where OPV favours certain parties in some areas and their opponents in other areas.
I should also state that I am personally not a supporter of single-member electorates in general. Neither CPV nor OPV is proportional, and a proportional system requires some form of multi-member election. The high stakes of the CPV-OPV debate partially reflect that preferences play a large role when electing only one person, and in a party system where many voters are casting their ballots for candidates who won’t make the top two.
In a proportional system, by contrast, a larger share of voters can elect a candidate with their first preferences, so preferences become gradually less valuable as the magnitude increases.
Both electoral methods have downside. Let’s start with the reigning champion, CPV.
Informal voting rates for the House of Representatives are astronomical – at levels that would be shocking in many democracies. 5.6% of votes cast in 2025 were informal.
While some of these votes are likely to be deliberate informal votes caused by compulsory voting (by voters who would simply not vote in a voluntary system) there is substantial evidence that a large proportion of informal voters sincerely try to cast a vote.
The AEC used to conduct regular studies of informal votes to categorise the type of vote. They last conducted such a survey after the 2016 election. At the time they found that roughly half of informal votes appeared to be “intentional informal” and half appeared to “unintentional informal”. To simplify, the former is likely driven by compulsory voting laws, while many of the votes in the latter category are informal due to CPV, and would be formal under OPV (although not every vote would be).
There is reason to think that unintentional informal votes make up a larger share of the informal vote than they did in 2016, and I am looking forward to the AEC again conducting research on informal ballots.
As ballot papers grow in size, it becomes harder to cast a formal vote and easier to make a mistake. There is a clear link between ballot size and formality, with a particular jump when a ballot paper has eight candidates. Voters are instructed to number at least six boxes for the Senate. With a single blank box treated as a last preference, numbering only six boxes would produce a formal vote if there are six or seven candidates, but it would be informal if there are eight or more candidates.
While some informal votes are easily judged to be informal (for example, leaving numbers off or duplicating numbers), sometimes it becomes a question of judging handwriting, as I observed during the Bradfield recount. The AEC does their best to interpret a ballot in a way that ensures it is formal, but the rules are strict.
So a significant number of Australian voters attempt to express their preference but are unable to do so, and their vote is not counted. This is effectively a deterrent to ensure that voters mark every preference, and ensures that most other voters mark preferences that flow. But I don’t think it justifies the cost of those voters not having their vote counted.
I think there are issues with political freedom that affect more voters than just those who vote informally. Requiring voters to number every box reduces the voters’ choice, and can require them to give a preference to candidates they do not wish to support. Indeed it forces most of us to give preferences to candidates we know little to nothing about.
When you number every box on a ballot paper, you are effectively making choices between each pair of candidates. Even if you strongly disapprove of the two candidates who make the final count, your vote ultimately will end up with one of them.
Australians will pat themselves on the back that MPs always win a majority of the vote, but compulsory preferential voting makes it logically impossible for anything else to happen. If you prefer Labor to Liberal in a classic seat, and you want your vote to be formal, you have no option at all. It’s not exactly a mandate for the winner.
The choice to support neither of two options is legitimate, and should be permitted, even if we encourage voters to use their preferences.
Compulsory preferential voting also forces voters to number the boxes of candidates with little or no information about them. Often we have random independents nominate at the last minute with very little information about what they stand for. Under an OPV system, most voters can simply ignore these candidates, but we force voters to number their box under CPV.
When a ballot is particularly long, a voter will sometimes be forced to either follow a party how-to-vote or use some variant of donkey voting to cast a formal vote.
There is also a case against the OPV system. It creates an incentive for parties to actively discourage preferencing, driving up rates of vote exhaustion and making it harder for candidates to overcome a primary vote lead. But there are also other ways in which the use of preferences could be encouraged or discouraged by our electoral institutions.
So why do I think high rates of exhaustion are bad? While I think exhaustion should be a legitimate option for voters, there are also voters who do not understand the significance of that action. It can result in a candidate who is effectively supported by the minority winning because the majority vote is more fragmented. If some voters effectively use their vote and others don’t, that is bad for political equality.
Historically, rates of preferencing was reasonably high under OPV. At the 1995 NSW state election, only 31% of 2PP preferences exhausted in classic seats. At the 1998 Queensland state election, just 30% exhausted in similar circumstances.
But the use of ‘just vote 1’ campaigning drove those exhaustion rates up to over 50% at the next elections in each of those states.
In a single-member electorate system, it is usually easy enough to identify which party would be more likely to benefit from preference flows, and who is likely to do better with weakened preference flows. The party that isn’t likely to get many preferences can then choose to run a ‘just vote 1’ strategy that isn’t really aimed at their own supporters, but actually at their opponents.
I wouldn’t only blame the parties – the official instructions for voters often send instructions that preferences are an optional extra. A NSW Legislative Assembly ballot paper says “You can show more choices, if you want to, by writing numbers in the other squares, starting with the number 2”.
You can see a difference in how voters mark their ballot papers between the NSW Legislative Council, where marking a second box is treated as an optional extra, and the Senate where voters are instructed to mark at least six boxes.
So both systems have significant downsides. But I think it is worth innovating, and trying something that could provide the best of both sides.
I think such a new system should look like this:
- Any vote with a single clear first preference would be counted as formal, with further preferences counted according to OPV rules.
- The instructions on the ballot paper should say something along the lines of “You should number at least 5 boxes”. This would be backed up by clear signage and advertising from the Electoral Commission encouraging voters to number at least five boxes.
- Messaging along the lines of ‘Just Vote 1’ will be prohibited. Candidates and parties would not be required to recommend 5+ preferences on how-to-vote cards, but if they provide instructions on how to vote it should say “You should number at least five boxes”, and a full ballot paper mockup should only be used if it features at least five boxes.
The first two points in this plan echo the approach taken for Senate voting since 2016: very generous counting provisions, but ballot paper instructions that assume a higher threshold of preferencing.
So why do I think the third point is necessary? It is because the incentives are different in a single-member contest. In a Senate contest, all parties can benefit from preference flows, and no-one is sure who would be the beneficiary of a high rate of exhaustion. But in a single-member contest, it is often easier to identify who has the primary vote lead and who is relying on preferences to catch up. The former candidate then has an incentive to encourage exhaustion. So I am suggesting policies that would limit what sort of advice candidates can offer voters, without forcing them to recommend preferences they don’t support.
This would be a new system, and is not currently used anywhere in Australia. But Queensland seems like a good place to try it out. The current government has promised to switch from CPV to OPV, but there are reasons to be concerned about exhausted votes at the next election. Perhaps this idea could work instead?