SA government proposes optional preferential voting


The South Australian government announced earlier this week that they plan to introduce legislation that will change the voting system from compulsory preferential voting (CPV) to optional preferential voting (OPV).

There are principled arguments for such a change, as it would likely ensure more votes are counted and the informal rate drops, but it is also something which would likely help the Liberal Party in the short term against Labor – although OPV has sometimes benefited Labor over the Coalition in the past.

The current system requires voters to number every box on their ballot paper (although savings provisions do allow some votes to be counted if the voter doesn’t number every box). CPV is used to elect the House of Representatives and lower houses in Queensland, Victoria, Western Australia, South Australia and the Northern Territory.

Optional preferential voting would mean that votes would be counted as long as there was a single ‘1’, even if there were no further preferences. Voters would still be able to mark preferences, but they would not be necessary for their vote to count. OPV is used to elect the New South Wales lower house, and has been used until recently in Queensland and the Northern Territory.

There are principled arguments in favour of OPV, but you can’t look past the political self-interest that motivates the Liberals to propose this change.

Under OPV, preferences are less likely to flow, and this tends to favour the candidate who is leading on primary votes. It’s harder to overtake a leading candidate when some preferences exhaust, and reduce the pool of preferences.

Labor tends to do better under CPV, primarily because of Greens preferences. Most Greens preferences flow to Labor when they are required to mark preferences, but a lot of Greens voters instead choose to exhaust when that’s an option.

Antony Green also points out that, in the South Australian context, compulsory preferences have helped independents win seats off the Liberals, usually with the benefit of Labor preferences. Of the 26 contests in South Australia since 1982 where a candidate trailing on primary votes went on to win, 14 were won by Labor, 11 were won by independents or minor parties, and just one was won by the Liberal Party. A number of those independents went on to support minority Labor governments after the 2002 and 2014 elections.

There is nothing permanent or unchanging about OPV benefiting the Liberal Party. Party systems change over time. Labor once supported OPV in an era when Democratic Labor Party preferences benefited the Coalition. The Whitlam government attempted a switch to OPV, and the Wran Labor government introduced OPV in New South Wales in 1980.

After OPV was implemented in Queensland in 1992, the system proved very helpful to Labor at the 1998 election, when the conservative vote split badly between the Nationals, the Liberal Party and One Nation. OPV was seen as being particularly difficult where the Nationals and Liberals ran against each other, but the NSW Coalition solved that problem by pretty much eradicated “three cornered contests”.

Queensland Labor changed the voting system back from OPV to CPV after the 2015 election, just in time for One Nation to re-emerge and make CPV seem much less advantageous for Labor than might have been thought.

Still, we can generally assume that as long as the Greens are the biggest minor party, and independents are a stronger presence in conservative seats, that Labor will benefit from CPV and the Coalition will benefit from OPV.

But let’s put aside the partisan advantage: what are the arguments for OPV?

The compulsory preferential system tends to produce a lot of informal voting – about half of all informal votes are attempts at casting a vote, but don’t meet the strict requirements of consecutive numbering of most or all boxes on the ballot paper.

Casting a formal vote becomes harder as more candidates are added to the ballot, leading to higher informal rates when lots of candidates run. Informal voting also tends to be higher where there are more voters who speak languages other than English. Incidentally, Labor tends to believe they would benefit if these votes were counted as formal, but the evidence is not clear.

On the contrary, CPV maximises the number of votes which contribute to the final outcome of the race, when the race dwindles down to two candidates. Under OPV it’s possible for candidates to win with substantially less than 50% of the total vote. Yet many of the votes rendered informal by CPV’s strict formality rules wouldn’t actually exhaust – votes for Labor and Liberal in most seats are never at risk of exhausting, yet those voters can’t simply just cast a ‘1’ preference.

In theory, there is nothing stopping voters choosing to indicate preferences under an OPV system, but it does create space for parties to encourage ‘just vote 1’ campaigns when they don’t expect to benefit from preference flows.

So you end up with a dilemma: do you ensure maximum numbers of votes are formal, or do you use a strict rule to force more of those voters to number enough preferences to prevent their vote exhausting?

I personally come down on the side of OPV on principle – I don’t think you can justify rendering votes informal when someone has indicated a clear preference.

But that doesn’t mean we should have a system which treats a simple ‘just vote 1’ as a default.

I think we’ve seen a way forward in the Senate reforms introduced prior to the 2016 election. The instructions on the ballot paper (repeated by AEC staff and advertising) tell voters to vote 1 to 6 above the line or 1 to 12 below the line, but the threshold for formality is lower: just 1 above the line or 1 to 6 below the line.

This was backed up with a strong advertising push in 2016: the AEC took an active position in support of preferences (although this wasn’t perfect – in many cases this advice was seen as telling voters to number 6 boxes, instead of numbering at least 6 boxes).

I think such an approach could also work for a single-member lower house election. I would think you would give official advice to voters to number multiple preferences (maybe “at least 5” is enough – if there’s 20 candidates I don’t think there’s any sense in telling voters to number every box).

But you would need a solution to the ‘just vote 1’ campaign problem. Such a campaign doesn’t work in a Senate election, where all parties can benefit from preferences. The likely beneficiary of preferences is much easier to predict in a single-member election, and thus there’s an incentive for the other side to try to undermine preference flows. Perhaps we just need a rule requiring campaigns to recommend at least as many preferences as the official advice.

Some work needs to be done, but I think it should be possible to encourage high rates of preferencing without throwing out perfectly good votes.

Elsewhere: Antony Green and Kevin Bonham have also written long and thorough posts tackling different angles of this story.

Liked it? Take a second to support the Tally Room on Patreon!


  1. Great post Ben. I support fully optional preferential voting, and suggest that the solution is to align all elections (federal, state, local) to this method, and support it with consistent messaging, such as “number as many candidates as you are willing to support”. Even better would be to align all elections (federal, state, local) to use multi-member electoral districts, so voters become accustomed to a single process, e.g. numbering a few candidates.

  2. I wrote a working paper on OPV for the Senate in 2013, at, and pages 12 to 18 discuss the relevant principles in some detail. Perhaps the most important objection to compulsory preferential voting is that it forces many voters to lie on their ballots by professing preferences which they don’t actually hold for candidates of whom they have never heard, and about whom it’s almost impossible to find anything out. (This requirement probably makes the ballot paper the only government form in Australia on which it’s compulsory to lie.) This situation reached its nadir at the Bradfield by-election in 2009, where there were 22 candidates and almost 7,000 informal votes, far more than ever before in the seat. Most of those informal votes resulted from deficiencies in the later preferences shown by voters; in a seat where preferences have never had to be counted to find out who had won. If gangs of thugs had invaded polling places and set fire to ballot boxes, destroying 7,000 ballots, that would make news all over the world; we should be just as outraged when it’s the law, rather than thugs, which is depriving people of their say.

    Regarding the South Australian initiative, I would only comment that while OPV may well be in the Liberals’ interest in the short-run, that’s not a problem in principle, since even the best electoral reforms tend to progress only when notions of good policy coincide with the self-interest of whoever is in government.

  3. I’m a lifetime Liberal voter and support OPV on principle – even though compulsory preferences only became a thing after the Country (now National) Party formed.

    I look forward to WA introducing it one day.

  4. Optional Preferential; is the go compared with the currrent voting system. Either number every box consecutively or just vote for the candidate or selected candidates you prefer but not all candidate if you wish. That is real democracy almost. However the most democratic voting method is a one in the box for your candidate only like in the UK.

  5. I support optional preferences for reasons both pragmatic (minimising spoilt ballots) and principled (voter choice).

    What’s weird to me is how we ended up at a point where full preferential is now the default. Preferential voting is an elegant solution to the obvious deficiency in first-past-the-post: the spoiler effect. Full preferential tries to go further to solve a different problem: requiring elected candidates to receive an absolute majority of the vote. It does so in a self-defeating way; by invalidating votes at the beginning of the count. A case of the cure being worse than the disease.

    In places where preferential – or “ranked choice” – voting is starting to be implemented, I doubt we will see full preferential take hold. Telling a Democratic voter in Maine that their ballot was invalid because they failed to rank the Republican, Greens, and Libertarian candidates would probably strike them as very puzzling.

  6. should one party out poll the other party the by amt of 2% or more but not do good on preferences of other parties then opv benefits that party…./eg East Hills last election the libs outpolled labor by 2% on primary votes…. other parties mainly left of centre polled 12% most of that 12% exhausted and only about 1.5% flowed to labor. opv becomes a quasi first past the post system. With a savings provision based on full preferences and formality based on clear voter intention there is a choice of opv voting aNy way. 1, 22223 etc

  7. Optional preferential in theory seems great but in practice has a habit of reverting to fpp.
    Something none of us want

  8. Ah yes, the old OPV is quasi first past the post. By that argument Lawrence Springborg would be Premier of Queensland, not Annastacia Palaszczuk. Labor in 2015 won 9 seats from second place, one in five of the seats Labor won, as voters suddenly found a reason to give preferences.

    And here’s the trailing first preference margins, 0.9, 1.1, 1.2, 1.8, 2.2, 5.0, 5.3, 6.6, 10.2

    I don’t get what you mean with 1,2,2, etc. That’s informal under full preferential voting. The “clear intent” rule saves almost no votes under full preferential voting.

  9. Janelle Saffin sits in state parliament despite losing the primary vote in Lismore by more than 14 percentage points.

    The notion that optional preferential is de facto first-past-the-post is spurious nonsense.

  10. a 1 alp 2 np 2 gr 2 onp 3…….. shows a clear intention to vote labor but after that there is no clear preference indicated…… this would be a defacto opv vote and a formal vote … under rules that made the vote formal to the extent that you could see what a voter intended. Changes to opv from full preferential voting were all made by parties riding high on first preference votes. An alternative is to this is side by side to have a proportional system to top up the inadequacies of single member electorates. can some one do the sums to see what would happen in sa based on opv…. I suspect it would give the libs a couple more seats.

  11. Mick, and that’s why the Langer provision was abolished in 1996. Because it was becoming de-facto OPV. 25 years after Langer voting was abolished, I still get asked about 1,2,2 etc sequences. I even get asked about it under OPV when it is completely unnecessary.

    MPs and the AEC still want nothing to do with such a provision given the legal action that followed from the original Langer provision.

    You could do what I call progressive informality. A vote remains formal to the point where it is inspected for preferences that don’t exist, at which point it becomes informal. The problem with the idea is that most of the votes declared informal by such a provision would be for minor parties.

  12. This is the same Mick Quinlivan who had the ALP winning Kiama, Oatley and Bathurst at the 2019 state election. So psephology for dummies is probably the starting point here…

  13. I wouldn’t mind OPV, as a voter. Choosing between the largest three parties is easy enough, but having to decide who gets to be in spot 6, 7 and 8 out of Socialist Alternative, the Christian Democrats and some no-name kook who wants to abolish parking meters is a pain. My lack of opinion between two different parties which will get <2% each shouldn't invalidate my vote.

    The SA Libs don't tend to win seats from behind on preferences because there's no minor right-wing / conservative parties with an appreciable vote. Other states have the Nats, One Nation, the Shooters or the CDP. (That one example from 1989 featured Call To Australia, a forerunner of the CDP.) SA only had Family First, and they're dead now.

  14. My choice would be that the instruction was to number all squares….. but with a savings provision…. based on identified preferences to limit informal votes. There would be no accidental informal votes. The obsession about the “langer provision” was wrong as very few would have voted that way.

  15. I don’t like OPV in the same system that has compulsory voting. It doesn’t make sense that we want every vote to count at the same time as having a system where many votes don’t. Strategies built around ramping up and curtailing the exhaust rate are hallmarks of OPV elections. These are strategies based on gaming the system, not on convincing people of the merits of candidates.

    They don’t always decide the election (as Antony Green points out), but they do result in a system where many votes don’t count. It definitely isn’t a well informed decision to have one’s vote exhaust. or else those official looking “just vote 1″/”number every square” signs wouldn’t matter. CPV isn’t perfect with its preference deals (which are also a thing in OPV), but at least involve following the how to vote card from the party you are voting 1 for (i.e. there’s some level of buy in, or trust of the entity saying the preferences). “Just vote 1” and “Remember to number every square” materials are designed to influence the voting patterns of voters who are NOT following a how to vote card and are more effective the less people understand our electoral system.

    Whatever system ends up being in place, there needs to be far better education, and far harsher penalties for entities that misrepresent the system. It’s far too common for politicians and media to talk about “splitting the vote”, which isn’t a relevant factor in preferential voting, and not be called out on it.

  16. SA would do well to move to Hare Clark.

    South Australia’s electorates are quite small in population and are approaching 5 per federal MP. Only 3 more MHAs would be required to get 5 MHAs per federal seat.

    These electoral boundaries are likely to be far more stable than the ones at SA state elections, which end up being redistributed often as part of the fairness provision. It also makes it more likely (though not certain) that the party with more votes will win more seats, which is very relevant in the recent history of SA politics.

    SA has multi member electorates in its history so this would not be an unprecedented move.

  17. I see no contradiction in compelling people to vote and not making them choose between the Communist Party and the League of Rights.

    Conversely, I see a massive contradiction in compelling people to vote and then rejecting their ballot due to absurdly harsh formality rules.

  18. Yes there is a contradiction in compelling people to vote and their vote not affecting the outcome of the election – whether through rejected ballots, or through misleading “just vote 1” signs. Whichever has a higher rate of voters involved in the final 2 candidate runoff is the ideal system IMO.

    It may come down to reforms in the space rather than the system itself though.

  19. “South Australia’s electorates are quite small in population”

    Unfortunately, not always in area. Hare-Clark works in Tas because the electorates are relatively small (in both population and area). In most of the mainland states, they’d be unreasonably huge. You wanna merge Flinders, Giles, Stuart and Frome into some mega-electorate electing 4 or 5 MPs? Because that’s Grey.

    A possible solution could be varying the number of MPs per electorate – keep the big outback ones as 1 seat, while making the compact inner metro ones 2 or 3 seats. That comes with its own issues, though.

  20. Another thought: OPV and the fairness provision don’t mix well at all. It’s difficult enough to figure out an underlying Labor vs Lib 2pp with independent seats as it is – the Boundaries Commission has sometimes had to interview people like Bob Such and Karlene Maywald to figure out whether their seats “should” be Labor or Liberal.

    Under OPV, if an independent attracts more supporters from one major party, and those people then just vote 1 for the independent, it skews the underlying 2pp in favour of the other major party, and also produces an artificially high exhaustion rate. That happens in NSW, with strange examples like Northern Tablelands in 2011, won by an independent with a huge margin. Labor got just 3.4% of the formal vote, yet they got 23.9% of a highly artificial 2pp, with a huge 39.6% exhausting. The actual exhaustion rate was 3.4% (about 40% of the 8.3% voting for someone other than Torbay or the Nats). There’s similar quirky results like the Libs “winning” Sydney or Labor “winning” Wagga Wagga.

  21. The solution to having big electorates is to allow the local mp better resources… 2 or 3 electorate offices, light plane travel petrol allowance etc

  22. Full preferential voting makes no sense with voluntary voting. But let me assume someone has been mad enough to implement it.

    Back in 2005, my local electorate had a by-election. It was a Labor v Green contest and no Liberal candidate was nominated. So under voluntary voting, a Liberal supporter wouldn’t have to vote.

    But come the next state election, there was a Liberal candidate. A Liberal voter would be forced make a preference choice between Labor and the Greens if they wanted to vote for the Liberal candidate, a choice they didn’t have to make when there was no Liberal candidate. It makes no sense.

    Voluntary voting and full preferences gives you the choice of voting for all candidates, or no candidates. You can’t vote for some candidates. Compulsory voting and optional preferences requires a voter to vote, but the voter can vote for as few or as many candidates as they like.

  23. I favour OPV for reasons of principle: that you shouldn’t be required to lie about a preference you don’t have, in order to express a preference you do have. Taking that as the principle means no minimum number of preferences, and no ballot instructions suggesting that more than one preference is required.

    The concept of a gap between ballot instructions and savings provisions also re-opens the old chestnut about whether it should then be illegal to advocate a vote that is valid, but contrary to the ballot instructions. For two Federal elections (1993 and 1996), a ‘Langer vote’ was still valid, but the then section 329A of the Commonwealth Electoral Act made it illegal to advocate one. I don’t believe there is an equivalent in relation the current Senate savings provisions, so you could in theory run a “Just Vote 1 above the Line” campaign.

  24. We need to ensure that there is no Just Vote 1 Campaign. All monor parties shoul respond to aJust Vote 1 campaign by dropping the party off there preference list . The Two major parties have a deep ingrained feeling of entitlement and an attitude that the 1/3 of population who reject both Lib and ALP have no rights.
    The Electoral Acts of Australia need to be printed documents which are near unchangable.
    Any change requiring 2/3 majority inn Parliament and not taking effect for five years after passing.

Comments are closed.