On 9 August 2017, a government move to restore to the Senate Notice Paper the Plebiscite (Same-Sex Marriage) Bill 2016 was voted down.  Had the Bill been passed, it would have provided for the conduct by the Australian Electoral Commission (AEC) of a plebiscite with compulsory voting, following electoral procedures substantially similar to those prescribed in the Referendum (Machinery Provisions) Act 1984.

The failure of the Bill in the Senate had been widely anticipated, and there had been media speculation that the government’s fall-back position would be the conduct of a ballot pursuant to arrangements made between the government and the AEC under section 7A of the Commonwealth Electoral Act 1918 (the “Electoral Act”).  Constitutional questions had, however, been raised about the viability of such an approach, for example in an opinion from counsel released on 7 August 2017 by the Human Rights Law Centre.  There was, nevertheless, widespread surprise when it was announced by the Minister for Finance on 8 August 2017, in advance of the Senate vote, that in the event of defeat the government intended to proceed not with an electoral process conducted by the AEC, but with an exercise to be undertaken by the Australian Bureau of Statistics (ABS), variously described as a “voluntary postal plebiscite” and as an exercise of “legal powers of the ABS to collect statistical information”, at an estimated cost of $122 million.  Notwithstanding this ambiguity, it has become clear that in many respects the process will, from the public’s point of view, be more similar to voting at an election than to participation in a survey.  My use of the term “plebisurvey” reflects this.

I aim in this post to explore the following issues to which this approach gives rise:

  • the challenge of ensuring the legal viability and political legitimacy of the plebisurvey;
  • implications of the legal Direction by which the process has been initiated; and
  • the extent to which the process meets certain well-established criteria for freedom and fairness of electoral processes, specifically:
    • impartial administration, and transparent handling of complaints;
    • a guarantee of a vote to those qualified to do so, and exclusion of those not qualified;
    • prevention of the buying and selling of votes;
    • a guarantee of a secret and safe ballot; and
    • prevention of multiple voting.

It needs to be emphasised, however, that as at the date of writing (15 August 2017), there is still comparatively little known about how the process will unfold beyond basic information just released by the ABS, and it is virtually beyond doubt that some of the observations which follow will be overtaken by events, and that a host of new issues will arise as the details of the process come to be worked out.

Factors influencing the viability and legitimacy of the plebisurvey

Before I proceed to a more detailed discussion of what is known (and not known) to date, a preliminary but vital point must be made. If the plebisurvey is to be of any value as an input to the policy-making process in Australia, it needs to be both legally viable and politically legitimate.

Legal viability is a matter for the courts, and two challenges to the overall legality of the exercise have already been launched in the High Court, with a Full Court hearing scheduled for 5 and 6 September 2017.  It is already clear from documentation associated with one of the challenges that a key question for the Court’s consideration will be whether the opinions being sought are “statistical information” within the meaning of the Australian Bureau of Statistics Act 1975 or the Census and Statistics Act 1905.  This being the case, it will be necessary for the government to hold the line that the plebisurvey is a statistical exercise.

Political legitimacy, on the other hand, is determined by public opinion and expectations, especially of the way in which societal decisions are made; and the government clearly believes that a process involving “voting” by all will have much greater legitimacy in the public’s eyes than a statistical process such as a sample survey of the type which the ABS conducts regularly. This, therefore, points to a need for the plebisurvey to incorporate many of the features of an electoral process.

The fundamental problem that the government faces is that these requirements for legal validity and political legitimacy are pulling in opposite directions: the more the plebisurvey looks like an electoral process, the harder it will be to justify it in court as an essentially statistical exercise; but the more it looks like a mere statistics gathering process, the harder it will be to give it public credibility.

The Census and Statistics (Statistical Information) Direction 2017 

 This “Direction”, issued by the Treasurer on 9 August 2017 under paragraph 9(1)(b) of the Census and Statistics Act 1905, would appear at present to be the only instrument which defines how the process will unfold.  Its substantive provisions read as follows:

3. Collection of statistics

(1) The Statistician is to collect the following statistical information in relation to matters prescribed for the purposes of section 9 of the Statistics Act (in particular, one or more of items 5, 30 and 38 in the table in regulation 13 of the Census and Statistics Regulation 2016):

(a) statistical information about the proportion of electors who wish to express a view about whether the law should be changed to allow same-sex couples to marry (participating electors);

(b) statistical information about the proportion of participating electors who are in favour of the law being changed to allow same-sex couples to marry;

(c) statistical information about the proportion of participating electors who are against the law being changed to allow same-sex couples to marry.

(2) The statistical information is to include information about participating electors at the national level, at the level of each State and Territory, and at the level of each electoral division.

(3) The statistical information identified is to be published on or before 15 November 2017. In this section:

(4) In this section:

elector means a person:

(a) enrolled on the Commonwealth electoral roll at the end of 24 August 2017; or

(b)  who has made a valid application for enrolment on the Commonwealth electoral roll before the end of 24 August 2017.

Several points are immediately worthy of note.

First, although the plebisurvey has widely been described as a “postal” operation (including by the ABS itself), there is nothing in the Direction which compels the ABS to adopt that particular mechanism for gathering information.

Secondly, there is nothing in the Direction which obliges the ABS to obtain the specified statistics by approaching all electors; it would therefore be open to the ABS to obtain the statistics in question through a well-designed survey based on a large random sample, at a cost much less than $122 million.  Some prominent Australian pollsters have argued that such a survey would be more likely to provide an accurate snapshot of the public’s view than the self-selected, non-random sample of the electorate on which the plebisurvey will be based.  That having been said, there is clearly an understanding between the government, the AEC and the ABS that a randomly selected sample survey will not be used.

Thirdly, the requirement for results to be compiled not just at the national level but also by State, Territory and electoral division highlights the likelihood that these subnational results may be used by some MPs and senators to inform the “conscience votes” they will be expected to cast on legislation which will be introduced if the plebisurvey result is in favour of same-sex marriage.  The significance of this lies in the fact that just as election results are often clear-cut in some divisions and very close in others, it is likely that there will be individual electoral divisions in which the plebisurvey result is very close, even if the national and state outcomes are more decisive.  And in those close divisions, every detail and every weakness of the arrangements for the plebisurvey could have the potential to influence the result.

Fourthly, the definition of “elector” in the Direction has already given rise to a number of controversies, especially regarding participation by “silent” electors and 16 and 17-year-olds.  These are further examined below.

Specific sources of legitimacy of electoral processes

For the balance of this post I will proceed on the assumption that the plebisurvey, however dressed up, is fundamentally an electoral rather than statistical operation.  Considered in that light, it is clear that for the process to be legitimate, more will have to be achieved than simply arranging voting.  There is a whole literature on the definition of “free and fair” elections, which I discussed in a Senate Occasional Lecture in December 2013, identifying eight key elements of the concept, highly relevant in this context.  Five of those which look likely to be especially problematical in relation to the plebisurvey are discussed below.

“The election is administered impartially, and opportunities exist for complaints about the process to be lodged and dealt with in an even-handed and transparent way”

The first point to highlight here is that the very concept of “administering” a process requires that it be defined.  Similarly, complaints about a process can only be meaningfully made against a background of predefined rules or expectations of how it will be undertaken, including rules defining rights of participants.  Here, there is currently an almost complete vacuum: little more is known than what appears in the Direction, and in a handful of brief media statements.  For an exercise of the scale contemplated – quite possibly the largest postal ballot ever undertaken anywhere in the world – this lack of detail about how the process will unfold when a de facto “roll close” is less than two weeks away is without precedent in a modern Australian electoral operation.  In particular, it is presently unclear:

  • what activities will be undertaken by the ABS, what will be done by the AEC, and what other agencies (for example, the Department of Foreign Affairs and Trade or the Australian Defence Force) may need to be brought into the process;
  • whether the elements of the regulatory framework will have the status of laws, guidelines, or administrative instructions;
  • who will be (a) drafting and (b) approving the regulatory framework for the operation, and, in particular, whether Ministers or their offices will be involved (as will be necessary if laws are to be enacted or regulations are to be made);
  • whether “voters” or other stakeholders will have any rights in relation to the process that can be legally enforced;
  • when the regulatory framework will be finalised;
  • how it will be publicised; and
  • who will have the right and/or responsibility to enforce it.

There has been media speculation that legislation might be expedited through the Parliament in an attempt to implement for the plebisurvey some of the integrity and campaign regulation measures applicable at elections.  Dr Kevin Bonham on his blog has explored this possibility in detail.

Beyond the basic need for a known regulatory framework, for administration to be impartial it must at the minimum be effective: a sub-optimally administered process runs a high risk of being of benefit to one side or the other, even if there is no malice at work.  On this, it need only be noted that the ABS has no experience in running electoral processes; that both the ABS and the AEC are used to knowing the rules of a process and planning for its implementation well before it is upon them; that processes become harder to administer effectively when time is short, multiple agencies are involved, and their respective responsibilities have not previously been defined; and that both agencies have therefore been presented by the government with an ugly problem.  Several commentators, notably Paul Kildea and Jack Waterford, have already explored issues arising from the conduct of the plebisurvey by the ABS.

On the broader question of the impartial administration of the process, while the Minister for Finance has stated that “[a]ll arrangements for fulfilling the Treasurer’s direction will be subject to final determination by the independent Australian Statistician”, the word “independent” does not appear in either the Australian Bureau of Statistics Act 1975 or the Census and Statistics Act 1905.  It is also notable that at an election, there is a clear distinction between design of the process (which is primarily done by the Parliament and reflected in the detailed provisions of the Electoral Act) and the implementation of the process, for which the AEC is responsible.  When, as proposed for the plebisurvey, the same body will be simultaneously designing and implementing the process, the impartiality of its actions will no longer be able to be assessed (almost) purely by the extent to which they comply with the law; value judgements about the design of the process itself will also come into play.

In relation to all of these issues, the absence of clarity regarding the role of the AEC is especially significant.  On 8 August 2017 the Finance Minister stated that “The ABS will make arrangements for the secondment of officers from the Australian Electoral Commission to assist the Australian Statistician with this process as required”, while on 9 August 2017 he went somewhat further, saying that “the ABS, supported by AEC officers as appropriate, will make relevant announcements about timetables and practical arrangements …”. The ABS for its part has said that it “has formed a Taskforce to deliver the survey and will publish the results by 15 November. The survey Taskforce includes staff from the Australian Electoral Commission seconded under ABS legislation.”

Two points arise from this. The first is that by themselves AEC officers cannot do all that much to help any more: the postal voting process has long been automated, so what the ABS will need most will be access to the AEC’s offices, procedures, IT systems, management structures and contracts.

The second is that it is arguable that any such support from the AEC – indeed, any AEC support other than secondment of staff in circumstances which had them strictly on the ABS’s books and off the AEC’s – would exceed the AEC’s powers and functions under section 7 of the Electoral Act, and would therefore need to be the subject of an arrangement under section 7A of the Act.

Finally, at present there is simply not enough information about any complaints handling process to make it possible to judge how even-handedly and transparently it might operate.  What can be said, however, is that the ability of stakeholders to make complaints depends critically on their having timely access to detailed information about how the process is supposed to proceed, not least so that they can make arrangements for appropriate monitoring and scrutineering.  If the regulatory framework for the process is prescribed too late, any rights to make complaints will be of theoretical rather than practical significance.  It could also be observed that the way in which the ABS communicated with the public in the aftermath of the problems encountered on the night of the 2016 census was hardly a model of transparency.

“People qualified to vote, and only people so qualified, are able to do so.”

Several specific problems identified with the plebisurvey plans are relevant to this criterion.

First, serious questions have been raised about the extent to which the use of a postal process may tend to encourage voting by older electors who are familiar with the mail, while discouraging voting by younger electors who are thought by some to be disinclined to make use of paper-based mechanisms, and whose more frequent address changes (when compared with their elders) may make it less likely that plebisurvey forms will reach them.  Certainly all voters are being encouraged by the AEC and ABS to ensure that their postal address details on the roll are up-to-date, but this message is not being advanced through the sort of intensive official advertising and public relations campaign typically seen at election time, not does the “roll close” scheduled for 24 August 2017 appear to have the salience of the roll close for an election.

In addition, questions have been asked about how postal votes will be able to be delivered to electors overseas who are in countries with a limited or non-existent postal service.  On that point, it is notable that the vast majority of people who vote overseas at federal elections do so in person at Australian diplomatic missions: at the 2016 election, 70,232 so voted in person, and only 1,174 postal votes were received by missions.  It would therefore seem highly likely that the basic model of postal voting will have to be modified if overseas voters are to be effectively serviced.

More generally, it would be fair to observe that the universal postal voting model is essentially untested in modern Australia, having been used only once before, at the Constitutional Convention election of 1997.  At a federal election or referendum, numerous voting modalities are made available, including ordinary voting at polling places; pre-poll voting; postal voting; absent voting; overseas voting; mobile polling for the taking of votes in hospitals, nursing homes and prisons; mobile polling in remote communities; and Antarctic voting.  The use of this mix is administratively challenging, but is pursued precisely because experience has shown that it constitutes the best way of ensuring that all who are entitled to vote are able to do so with relative ease.  Any scheme based on just one modality of voting can be expected to be sub-optimal from that perspective.  Again, it is possible but by no means certain that some aspects of those different modalities may somehow be used for the plebisurvey.

Secondly, it has been noted that the AEC is prohibited by subsection 90B(6) of the Electoral Act from providing to the ABS details of “silent electors”: people whose addresses are, for their own safety, not shown on the roll.  This suggests that any arrangements for their participation in the plebisurvey will have to be made by the AEC.

Finally, the question of whether 16 and 17 year olds will be able to vote has become a matter of controversy, the argument being that those of them who enrol under section 100 of the Electoral Act are entitled to be placed on the roll and therefore, according to the wording of the Direction, should be able to participate in the plebisurvey.  On this, the AEC issued a statement on 11 August 2017 asserting that 16 and 17 year olds who claim enrolment are only “provisionally enrolled until they turn 18”, and “are not on the roll”.

There are several major problems with this statement, not the least being that the concept of provisional enrolment of 16 and 17 year olds is unknown to the Electoral Act.  In fact, a careful reading of the Act suggests that the AEC’s position on this is misconceived.  This was rapidly identified by a number of commentators (myself included), and the arguments against the AEC view have been well summarised in a blog post by Mr Stephen Murray.  It remains to be seen whether any corrective action on this will be taken.

“Votes are not bought and sold”

At present there is no legal provision specifically dealing with the possibility of the buying or selling of plebisurvey votes. The ABS for its part has stated that there are “potential penalties for providing false or misleading information to the Australian Bureau of Statistics”; and this could conceivably encompass information falsely suggesting that a particular person has completed the plebisurvey form.  But even if that applies, or even if an appropriate ban were otherwise legislated, it could be virtually impossible to enforce in practice, as the act of voting will be taking place in locations determined by the voters, in the absence of the scrutineers who at a normal election can monitor activities at polling places.  A vote buying website set up overseas would for all practical purposes be beyond the reach of Australian law.

“Voters can cast a secret ballot, without fear of any adverse consequences”

In the absence of special legislation, the provisions of the Electoral Act which require postal votes to be marked in such a way that authorised witnesses cannot see them (paragraph 194(1)(d)), and prohibit  interference with political liberty (section 327), will not apply. Section 28 of the Crimes Act 1914, which provides that:

“Any person who, by violence or by threats or intimidation of any kind, hinders or interferes with the free exercise or performance, by any other person, of any political right or duty, commits an offence”

could conceivably be invoked, though there is no law at present which would appear to give participation in the plebisurvey the status of either a right or a duty.

As things stand, people could be invited to mark their votes ceremonially in churches in front of the whole congregation.  More generally, it is often overlooked that a primary purpose of having voting at state-organised polling places is to guarantee the ability of voters to cast their ballots in an environment free of coercion.  This guarantee is lost with postal voting.  On this, Birch and Watt note that:

“The dominant currency of the home is typically that of emotion, rather than that of money or brute force.  Within the emotional economy of the home, psychological pressure may well have the same if not greater power to alter behaviour as that of money or force in the public arena.  So, though more subtle and difficult to identify, emotional coercion may pose just as serious a threat to democracy – indeed, these very qualities may make it even more worrisome.  The common distinction between the ‘public’ and ‘private’ spheres breaks down when we consider the complex overlapping spheres of ‘privacy’ within the home.  Acts that are ‘private’ within the family may not be sufficiently private for individual family members.”[1]

“Everyone votes only once”

In the absence of special legislation, attempting to vote more than once will not be a specific offence, though again, a general prohibition on the provision of false or misleading information to the ABS could conceivably be invoked.

A broader practical problem is that the plebisurvey ballot papers will be sent to everyone on the electoral roll, including people who have no interest in making use of them.  This contrasts greatly with the situation at election time, when postal ballot packs are only sent to those electors who have applied for them, or who have registered as general postal voters.  While normal legal provisions relating to interference with the mail will provide some protection against the theft of ballots from letter boxes, there will be no specific offence of taking them from rubbish bins, or generously offering to collect them for “recycling”.  There are likely to be literally millions of such ballots floating around in a relatively uncontrolled way: at the 1997 Constitutional Convention postal ballot, the turnout was 46.92%, and a similar turnout for the plebisurvey would see more than 8 million ballots delivered nationwide to people who ultimately did not use them.  Many of these would potentially be ripe for diversion by persons inclined to try to manipulate the process.

The ABS’s somewhat forlorn statement that those who choose to not complete the ballot “are encouraged to destroy it by tearing it into two or more parts” seems unlikely to have a major beneficial effect on the integrity of the exercise.

In addition, a person who applies for a postal vote at a federal election is required, in his or her application, to nominate an answer to a “security question”, and that answer is also required to be stated on the voter’s completed postal vote certificate envelope.  This mechanism provides a strong protection against completion of postal votes at elections by persons other than those who have applied for them; but it will not operate at the plebisurvey, as discrete applications for postal votes will not be required.

Overall, the ballot paper integrity mechanisms for the plebisurvey look likely to be quite startlingly weak.


It can be seen from the foregoing discussion that the plebisurvey, considered as an electoral exercise, is at risk of being defective in a number of different ways; and the likely defects are substantial, of the type which would often lead international observers to deem an election not free and fair.  Some may be resolved through the adoption of as yet unknown procedures.  Others, such as the rudimentary nature of the legal framework for the process and the weak ballot paper controls, are inherent in its basic design and are unlikely to be curable.  However things unfold, the model is clearly far from ideal, and the extent to which its outcome(s) will be taken to be a legitimate expression of the views of the Australian people is difficult if not impossible to gauge right now.

[1] Sarah Birch and Bob Watt, ‘Remote Electronic Voting: Free, Fair and Secret?’ (2004) 75:1 The Political Quarterly 60-72.


Michael Maley is one of the ACT Co-Convenors of the Electoral Regulation Research Network, and a member of the Editorial Board of the Election Law Journal.

Suggested citation:  Michael Maley  ‘The 2017 Australian Marriage Law Postal Plebisurvey: Issues and Controversies’ on AUSPUBLAW  (16 August 2017) <>