Saturday, December 19, 2020

JSCEM's Recommendation For Optional Preferential Voting

Advance Summary

1. Coalition members of the Joint Standing Committee on Electoral Matters have recently recommended Optional Preferential Voting (OPV), but have provided very little discussion of that recommendation.

2. It is likely the government would have the numbers to pass OPV if both Coalition parties wanted to do so, but unclear as yet if they actually want to proceed.

3. Evidence from NSW state elections under OPV strongly suggests that it would significantly to severely disadvantage Labor if adopted federally. Evidence from Queensland is more confused.

4. The level of likely disadvantage to Labor under OPV could often result in a few federal seats having different results, but could be more severe when Labor holds majority government.   

5.Compulsory preferential voting leads to votes being disallowed because of irrelevant errors.  Reform to reduce or remove this problem is necessary.

6. Optional Preferential Voting has been defended on the grounds of allowing greater voter freedom, but it can also result in misleading campaigns to discourage preferencing and in cases where parties are disadvantaged because voters make incorrect assumptions.

7. Having OPV as a savings provision rather than a direct instruction to voters would solve most of the problems that OPV addresses with less disadvantages.

8. Labor's arguments against OPV in the report are weak.  


I have been trying to write something about the recent Joint Standing Committee On Electoral Matters report on the 2019 federal election, but have decided one issue needs breaking out from the rest as the discussion was getting too long.  

The most headline-grabbing aspect of this year's report is that it has recommended the adoption of Optional Preferential Voting.  There is very little by way of argument for this recommendation - it is mainly defended on the basis of "freedom".  There is some discussion of the impact of the proximity of the 2019 NSW election on the informal voting rate at the federal election.  There is no discussion of the many alternative approaches to trying to reduce the informal voting problem

Will this actually happen?

If the government is serious about OPV it appears to have every chance of passing the Senate, as One Nation and Stirling Griff have expressed support.  With those three votes it could even survive a government Senator abstaining.  The question is whether the government will be serious. At this stage OPV has been recommended only by JSCEM, and JSCEM recommendations have often been ignored in the past.  

Some considerations in whether or not to push for OPV could include:

* The government could be accused of devoting energy to changing the electoral system instead of addressing issues claimed to be more important.

* The issue could give Labor oxygen, which has been in noticeably short supply for the Opposition lately.

* On the other hand, the issue could very easily tempt Labor into campaigning too heavily on the system change and not enough on actual issues.  

* Adopting OPV would probably quickly spell an end to three-cornered Coalition contests, except perhaps in seats where Labor was uncompetitive.  The Nationals, who like to have tilts at seats like Eden-Monaro, might not be impressed with this, but they might also be mollified by the Liberals abandoning similar tilts in, eg, Mallee.

* Perceived advantages and disadvantages in different seats (see below).

* The nature of the debate (also see below).

* Inertia.   More or less compulsory preferencing has existed at federal elections for a century.  The Coalition has done well at federal elections.  There may be some resistance to changing a system that clearly isn't stopping the Coalition doing well. 

The present JSCEM report is unusually light on detailed argument and heavy on what might be called electoral culture war items.  I'm reserving judgement on whether this is serious or a provocation for the sake of baiting or scaring the left.  

The impacts of OPV

OPV has been dubbed part of a "dark [..] authoritarian vision" in at least one Labor MP's response, but if so then it was the dark authoritarian vision of Gough Whitlam, Neville Wran, Wayne Goss and even the Fitzgerald Inquiry.  (The first tried to introduce it, the second did introduce it, and so did the third on the recommendation of the fourth.)  What has changed in the calculus of who OPV benefits is the rise of the Greens since the early 2000s.  

The simple argument for OPV benefiting the Coalition at the moment is that since the Coalition currently suffers on preferences (getting below 40% of them at elections from 2004-2016, and not much above 40% in 2019) it should benefit the Coalition if more preferences exhaust.  However, it is not completely straightforward, because Greens voters are less likely to exhaust their vote than non-Greens minor party voters, and the preferences of non-Greens minor parties tend to favour the Coalition.  Still, that seems to only reduce the impact of OPV on Labor's chances, not eliminate it entirely.  

For instance, if we take the 2019 federal election and assume 40% of Greens preferences exhaust and the rest flow in proportion to under compulsory preferencing, Labor loses about 382,000 votes more than the Coalition.  If we assume 70% of preferences from the seven most popular minor right-wing parties exhaust, Labor has only recovered 196,000.  Applying similar standards to the remaining parties, Labor is likely to suffer an impact of about 0.6 to 0.7% 2PP.

That is, however, assuming that OPV doesn't change the flow of those minor party preferences that don't exhaust, and in the case of the Greens, it appears it sometimes does.  At some elections under OPV in Queensland and NSW, Green preferences that do flow have flowed more weakly than under CPV.  (This is also seen in Tasmania, which has an effective OPV system at party level.)  Labor has to constantly enthuse Greens voters to keep preferences up, and in NSW we've seen campaigns where the Coalition is campaigning on the economy while Labor's banging on about koalas.    

I thought I'd have a look at how Labor has gone in turning preferences to its advantage in NSW and Queensland at both state and federal elections since the rise of the Greens.  The variables I've looked at here are the gap on primary votes and the 2PP gain rate for Labor on preferences, eliminating variation in how many preferences are available.  I also remove the impact of three-cornered federal contests by counting all votes that leak to Labor from them as if they were primary votes for Labor.  There's a well known effect with OPV that where one side is way behind on primary votes, OPV magnifies the lead of the opponent, so keeping an eye on the primary vote difference should screen out landslide elections such as Queensland 2012 and NSW 2011.

New South Wales:

[NB "two-party vote" in the above should read "third party vote".  I'll fix this soon!]

For NSW the pattern is very clear-cut: in the era of high Green votes, Labor has always improved its 2PP position faster on preferences at federal elections under CPV (blue dots) than at state elections under OPV (red dots).  The method of preferencing, and not Labor's primary vote lead, is the overwhelming explaining factor, and even if 2011 is excluded, the OPV factor seems to hurt Labor's gain rate by around 8.5% of the minor party vote share.  That washes out to nearly 2% 2PP, and the reason it is so high is that Labor's performance on Green preferences was not always as good under OPV as it was in the 2019 state election.  


Queensland is of course not so neat!  

Queensland presents one case of Labor doing extremely well on preferences under OPV, and that was 2015.  That election saw nine Labor MPs overturn LNP primary leads, a rate not repeated at two subsequent CPV elections.  The gain rate for Labor in 2015 was equivalent to what would happen if Labor got 64% of all preferences under CPV, which in Queensland doesn't happen.  The 2015 result happened because the Newman LNP government was so detested that even voters for conservative minor parties preferenced against it.  They may well have also done so, or done so even more strongly, had the system been CPV, but we will never know.  

The Queensland pattern is also confounded by the high One Nation primary vote in two CPV elections, the 2019 federal election and the 2017 state election (these are the lowest blue and black dots respectively).  To a lesser degree, a high One Nation vote was also a factor in the 2020 state election (the other black dot).

By the time I take out all the Queensland state and federal elections that were exceptional in some way there are hardly any left to look for a pattern in.  This should make us more cautious about the strength and reliability of the pattern shown in NSW.  It is, however, notable that in Labor's two other Queensland wins under OPV at state level in the sample (2006, 2009) its gain on preferences was very small.  I think this is about Green voters getting jaded with Labor majority governments.  Many Green voters will never preference the Coalition in a fit, but may exhaust their preference if given the option.  This drives much greater shifts in Greens preferencing than are seen under CPV, and was one of the causes of the 2015 Queensland 2PP polling miss.  

Based on the patterns implied by NSW, it's worth commenting on some particular seats - and for now just looking at the patterns from the NSW state election in 2019 and ignoring earlier elections where Green voters under OPV were less keen on Labor.  As William Bowe notes, Macquarie and Lilley would probably have been toast under OPV in 2019 and Cowan and Eden-Monaro would have been wobbly.  There are seats with high Green votes that would have been much closer - for instance Moreton, where Labor could have lost about 80% of its 51.9-48.1 margin.  Labor won Griffith with 52.9% 2PP, but assuming a 40% exhaust rate among Green voters and 70% for conservative minor parties, they actually would have lost it (!) - that said the exhaust rates would probably be lower in this seat.  There would be some seats where the Coalition would be disadvantaged by OPV - for instance Longman, where OPV would wipe out about three-quarters of the Coalition's 3.3% margin.

The most notable recent election that federal OPV could have changed the overall outcome of (based on impacts like those in NSW 2019) was 2010.  In that election Labor made an effective gain rate of .288 votes/vote on preferences.  The election had a high Green vote with low voting for right-wing minor parties.  Under OPV Labor would have saved Denison but been at massive risk in La Trobe, Moreton, Corangamite and Robertson, and the Coalition would probably have won.

Overall it seems that at the moment a shift to OPV would hand the Coalition a substantial advantage, probably good for a few seats per election, perhaps at times more.  

The OPV vs CPV debate

There are two separate aspects of the optional preferential voting vs compulsory preferential voting debate.  The first is that OPV saves voters from having their vote exhausted because of errors.  The second is the argument that voters should not be forced to give preferences if they don't want to.

I explored the first at some length in How Should We Solve The Problem Of Unintended Informal Voting?  A few percent of voters every election are having their votes invalidated because they either make honest errors while attempting to number all the boxes, or are confused between voting systems and number 1 only or 1-6 only.  Unintentional informal voting is often concentrated in areas with a high proportion of voters who primarily speak languages other than English.  While the impact of unintended informal voting at party level is probably weak (especially as it falls mainly in safe Labor seats) we should be uncomfortable about a system that tends in its effect to exclude voters of particular racial and ethnic backgrounds more than others.  For this reason I consider the current system of compulsory preferencing without savings provisions to be unacceptable - any solution that still allows preferencing, including OPV, is better than what we have now.

However, it is not at all necessary to adopt OPV to combat unintended informal voting.  Measures could be taken within the current system to reduce unintended informal voting.  Perhaps the best of them would be "OPV as a savings provision": instruct voters to number every box, but accept any vote that has a unique 1.  This approach already has a sort of parallel in the Senate system, where voters are instructed to number several boxes but a lower number is accepted in some cases under the savings provisions.  

The philosophical issue is the claim that voters should have the freedom not to preference.  Some people say that anyone accepting this should also accept that voters have the freedom not to even turn up, with voluntary voting essentially a fringe position in Australia.  I don't think this is clearcut, as it can be argued that the requirement to turn up at the booth at least provides some counter to factors that could cause some voters to come under pressure to not even bother.  That said I don't think Australia's model of enforcing compulsory attendance (a flat fine) is satisfactory, but that's another story.

The fact that a voter would like to vote a certain way is not by itself a watertight argument for allowing it.  Some voters would like to vote 1 in upper houses above the line and have their party direct their preferences for them, but they should not be allowed this freedom because experience shows it results in preference harvesting, in unaccountable MPs elected from tiny vote shares, in parties being sometimes disadvantaged because they are too popular for other parties to want to deal with them, and in the use of deliberately deceptive preference tickets that confuse voters.   

When it comes to lower houses, here are some reasons why the freedom to make preferences optional might have side-effects:

1. OPV results in parties running misinformation campaigns (often using signs designed to resemble official electoral material) to attempt to cause voters not to direct preferences.  JSCEM makes no attempt to combat this problem in recommending OPV.  JSCEM also, in correctly (in my view) rejecting a push for general truth-in-advertising laws, completely ignores any scope for tightening existing provisions on material that misleads electors in relation to the casting of their vote (eg the Chisholm/Kooyong signs), and seems not to regard the issue as important.

2. OPV can result in disadvantage to parties in cases where supporters of a different party do not give preferences because they falsely believe that their party will not be excluded.  At federal level this is not currently a problem (though it nearly was in Melbourne Ports in 2016) but at state level it is a significant issue, for instance in three-sided Labor/Greens/Nationals contests in northern NSW.  It places a heavy education burden on campaigners for parties compared to CPV.

3. Voters may exhaust their ballots not as a matter of informed choice, but because they do not understand the system and falsely believe that continuing to preference may harm their chosen party or may assist the party they most dislike to win.  This places a major voter education burden on any party that is trying to overcome a primary vote lead using preferences.  

Also, is the freedom to not preference an important freedom?  I'm not sure that it is.  A voter who wants their preferences to over time advantage neither major party, for instance, has the option of randomising their preferences.  A voter might have a moral objection to preferencing between two parties they dislike, but if they actually prefer one of those parties to the other then their objection is misguided, and if they don't then why not randomise?  (One could even randomise the first one then alternate using an A-B-B-A-A-B- etc pattern.)

In Senate elections, semi-optional preferencing is necessary because the alternatives are unacceptably high informality or the preference-harvesting problems of Group Ticket Voting.  But in Reps elections, informality can be addressed in other ways.  One of these, OPV as a savings provision, would be just as effective and also in practice would support the freedom of the elector to exhaust their ballot if that is what they insist upon doing.   We can also see from Senate elections that voters do tend to follow the instructions and therefore that this should lead to a good rate of preferencing.  

In my view, OPV as a savings provision is a superior solution to fully-fledged OPV.   However even OPV as a savings provision, if adopted, should be reinforced with penalties for attempting to mislead electors into not giving preferences.  That is not to say there are never valid arguments for not preferencing. If someone wants to say that the major parties are Tweedledum and Tweedledee and that voters should exhaust their preferences to encourage the major parties to differ, then they should be able to make that argument.  

Labor and Green Responses

Unfortunately Labor's response to the OPV recommendation in the report is weak, and to some degree echoes the unsound line of attack that the party made on necessary Senate reforms.  Labor may have decided on a throwaway response because they think the recommendation is itself a throwaway; if so they'd better know what they are doing.  

The Labor dissenting report claims with no evidence that "The Chair’s recommendation for optional preferential voting is a clear attack on compulsory voting at a time when we need it the most."  Obviously it is not, since many people clearly support "compulsory voting" with optional preferences.  

The dissent then claims that votes are "wasted if a voter’s single preference isn’t elected."  This is not true if the voter only wanted one candidate to be elected and didn't care what the result of the election was if they were not - indeed forcing such a voter to preference distorts their true intention.  No sympathy is shown for voters whose votes were wasted by those votes being disallowed because of irrelevant misnumberings.  Also, given that 40.6% of votes finished up not reaching any candidate who was actually elected, any argument about "wasting" votes is just begging for multi-member seats.

The dissent claims without evidence that "It has been shown that, where given the option, only the most engaged voters distribute all their preferences [..]".  I suspect that not only has there not even been any serious attempt to show this, but that it is not even true, since some disengaged voters may copy How to Vote cards.  It then claims that this " results in the disenfranchisement of a significant number of voters" - no it is not disenfranchisement if a voter who has not been enthused by enough different party campaigns chooses not to give a preference that they were not engaged enough to care about anyway!  Disenfranchisement is stopping people from having their votes cast or counted in the first place (which the current formality rules do), and Labor MPs should be subject to a ten-year internal ban on using the d-word in public because I think that is about how long it will take for them to understand it.

The Greens' opening salvo is too cute by half when it claims that "Compulsory preferential voting is a cornerstone of Australian democracy and helps to promote greater diversity in the MPs and Senators elected to our parliaments. " Both the current Senate and the one before it were elected by semi-optional preferencing, which the Greens have long supported.  The Greens however do at least recommend review of savings provisions, an issue ignored in Labor's effort.  


  1. You pose the rhetorical question: "A voter might have a moral objection to preferencing between two parties they dislike, but if they actually prefer one of those parties to the other then their objection is misguided, and if they don't then why not randomise?" But later you observe that "forcing such a voter to preference distorts their true intention" - which seems to answer the question.

    Having said that, I have some sympathy with the approach you favour, of having savings clauses, but prohibiting advocacy of OPV. That, two-pronged approach would, however, would be somewhat similar to the scheme which applied from 1984 to 1998, which fell from favour somewhat due to the litigation involving Mr Albert Langer to which it gave rise.

    More generally, it's high time that the ALP started to take some of these issues seriously: too many of their dissents over the years, most notably those relating to the Senate reforms of 2016, have been risible. If they want to stop fully fledged OPV they would be better off backing your approach, in the hope of getting the support at least of some crossbenchers, rather than going down with their ship.

    Finally, I find it passing strange that people who complain bitterly, and with justification, about the prospect of people being disenfranchised by GOP-style voter ID requirements often seem quite offhandedly blase about people having their ignored votes because they failed to correctly mark preferences which turned out not to be needed to determine the result anyway.

    1. I wouldn't have an issue with it being allowed to encourage voters to deliberately use the savings provision to exhaust their preferences, only with anyone doing so in a way that misled voters. That would include any material that misled the voter as to the origin of the material or as to who endorsed the material.

  2. I am frankly shocked at the tone of the article and the comments. Electoral reform is about the electors, not the political parties. It is a non-trivial fact that compulsory Preferencing exists only in Australia and not in all of Australia. The question is not which political parties would be advantaged by OPV, but if there is any justification for compulsory preferencing that is consistent with Article 21 of the Universal Declaration of Human Rights.

    1. So Article 21 says:

      "(1) Everyone has the right to take part in the government of his country, directly or through freely chosen representatives.
      (2) Everyone has the right of equal access to public service in his country.
      (3) The will of the people shall be the basis of the authority of government; this will shall be expressed in periodic and genuine elections which shall be by universal and equal suffrage and shall be held by secret vote or by equivalent free voting procedures."

      It doesn't say anything about rules for voting formality. "Freely chosen" is about the right to political participation. I wish that it did because it would be useful for striking out Western Australia's appalling upper house system where the only alternative to Group Ticket Voting is numbering a large number of boxes with zero tolerance for error.

      Note also that my suggested solution, OPV as a savings provision, does allow for complete preferencing freedom. It just doesn't tell voters to exercise it.

  3. The right to vote guaranteed by Article 21(3) must logically include the right to vote a plague on both your houses against the major parties. I voted that way, deliberately, rationally and calmly in a NSW state election. To read Article 21 any other way would imply that voting in a single candidate election complies with Article 21.

    1. If the will of the people guaranteed by 21 (3) included casting a formal vote that was, for instance, 1-2-3 for a bunch of minor parties and then exhausting the vote so as not to preference the major parties, then it would also follow that article 21(3) prohibited first past the post voting, since such a vote is just as incapable of being processed as intended under FPP as it is under CPV. Personally I would love it if the United Nations declared FPP to be not a valid electoral system and a breach of human rights, but I am sure if that was their intention we would have heard of it by now.

  4. It will be interesting to see what the Greens think.

    Greens will be more competitive in contests where Liberals/Nationals run 3rd (as LNP tend to do "just vote 1" cards instead of feeding Labor preferences). I think in NSW seats, the Greens gain a margin on Labor when Liberals are excluded.

    Greens would have won Batman 2016 (where they won first past the post). However they probably would have lost Melbourne 2010, from the era when Greens received Liberal preferences, though they were only 0.35% behind at Liberal exclusion.

    They will be less competitive in seats where the higher of them or Labor needs the other's preferences to win the seat. Nats wouldn't have held on in Lismore 2015 under full preferential. Greens came close to being spoilers twice in that seat; Labor only just lost the 2PP in 2015, and Greens would have lost if they overtook Labor in 2019 when Labor won.

    Most of their federal target seats seem to be the 2nd type (which makes a seat winnable on a relatively low primary) so I think if it were up to them they'd pick CPV.