Report | Doc. 14791 | 07 January 2019

Updating guidelines to ensure fair referendums in Council of Europe member States

Committee on Political Affairs and Democracy

Rapporteur : Dame Cheryl GILLAN, United Kingdom, EC

Origin - Reference to committee: Doc. 14299, Reference 4296 of 30 May 2017. 2019 - First part-session

Summary

Recalling that, in some countries, recent referendums have raised concerns about the process and/or the fairness of the outcome, the report underlines that referendums should be embedded in the process of representative democracy and should not be used by the executive to override the wishes of parliament or be intended to bypass normal checks and balances. It recommends three kinds of change:

– the Venice Commission’s Code of Good Practice on Referendums, adopted in 2007, should be updated to take account of changes arising from the growth of the internet and social media and to reflect the importance of ensuring that quality information is available to voters;

– member States’ compliance with the Code should be enhanced. Constraints on government campaigning and the independence of the body supervising the referendum and its powers to enforce the rules, including the power to impose sanctions in case of breach, should be ensured;

– finally, in areas where legal prescriptions are not appropriate, there is much scope for sharing good practice between countries, particularly in relation to methods for enhancing citizens’ participation in democratic deliberation, both before and after a referendum is called, for instance through citizens’ assemblies.

A. Draft resolution

(open)

1. Experience drawn from referendum practice since the Code of Good Practice on Referendums (hereinafter “the Code”) was adopted in 2007 by the European Commission for Democracy through Law (Venice Commission), upon a request from the Parliamentary Assembly, calls today for strengthening the rules on conducting referendums, enhancing compliance by Council of Europe member States and sharing good practice throughout the continent.

2. In particular, in some countries, recent referendums have raised concerns about the process and/or the fairness of the outcome. Other countries have explored innovations in referendum practice which could be of benefit to policy makers in all member States. Moreover, the growth of the internet, and especially social media, has fundamentally changed the nature of political campaigning and of people’s expectations of democracy.

3. The Assembly therefore welcomes the fact that a process for updating the 2007 Code has been initiated by the Venice Commission and invites the latter to take into account, in the revised Code, the following general principles:

3.1. referendums should be embedded in the process of representative democracy and should not be used by the executive to override the wishes of parliament or be intended to bypass normal checks and balances;

3.2. proposals put to a referendum should be as clear as possible and subject to detailed prior scrutiny, including by parliament, to ensure that they reflect voters’ concerns and can reasonably be expected to deliver on voters’ wishes;

3.3. the conduct of the campaign should ensure balance between the sides and allow voters to have access to balanced and quality information on the options in order to be able to make an informed choice.

4. With regard to specific aspects of the conduct of referendums, the Assembly invites the Venice Commission to consider including in the revised Code the following elements:

4.1. it should not be possible for the executive to call a referendum on a constitutional proposal, except where the decision to hold a referendum has already been endorsed by the legislature or where the proposal that is put to a referendum has been passed by the legislature;

4.2. where possible, referendums should be post-legislative; where this is not possible, a process should be set out requiring two referendums if the first referendum does not allow voters to choose between the options that are ultimately available;

4.3. to avoid the danger of low participation while maintaining the principle that referendum results should not be subject to turnout thresholds, referendums should, as far as possible, be called only on subjects that are likely to attract significant public interest;

4.4. it should not be possible to put to referendum a proposal which would run counter to Council of Europe membership conditions, such as a proposal to reintroduce the death penalty;

4.5. questions requiring replies other than “Yes” or “No”, including multi-option questions, should be allowed if they give voters a clearer choice;

4.6. an impartial body should check any proposed referendum question to ensure it is clear, accessible and unbiased. Where a fixed format for referendum questions is used, this should be reviewed periodically to ensure that it provides for a ballot paper which fulfils these criteria;

4.7. in the case of citizen-initiated referendums, the number of signatures required to trigger a referendum should be high enough to ensure that the proposal has genuinely wide support; the development of procedures whereby a citizens’ petition would not lead directly to a referendum, but rather to a citizens’ assembly which would recommend follow-up action could be encouraged;

4.8. there should be sufficient time for all sides to develop and make their points and for voters to hear the arguments and form an opinion. While a considerably longer period of preparation is desirable, particularly if the topic has not already been subject to widespread public discussion, the absolute minimum time between calling a referendum and polling day could be set at four weeks;

4.9. the prohibition for the authorities to use public funds for campaigning purposes should last throughout the campaign period;

4.10. in the case of public funding, the principle of equality between the sides should take precedence over that of proportional distribution of resources;

4.11. the principle of transparency should apply both to the sources of campaign funding and to how those funds are spent; spending and/or donation limits should be encouraged and foreign donations prohibited;

4.12. new rules on the transparency of campaign materials are needed, including clear labelling of who produces all advertising; rigorous independent press regulation and impartial fact-checking should be encouraged for the purpose of tackling misinformation;

4.13. the responsibility to provide official information should be entrusted to an independent body, rather than the authorities; information should include, as a minimum, the referendum question and details of when and how to vote and, where possible, explanations and analysis of the proposals;

4.14. sanctioning powers should cover all aspects of campaign regulation, including breaches of the campaign finance rules; fines should be commensurate with the scale of campaign funding.

5. The rise of digital media and the increasing convergence between printed, broadcast and digital media, especially social media, calls for regulation across all media sectors in relation to all electoral processes. The Assembly notes that the Venice Commission is currently working on these issues and hopes that relevant guidelines will be developed for both referendums and elections.

6. Considering that compliance is occasionally lacking in the following areas, the Assembly calls on member States to ensure that:

6.1. all fundamental aspects of referendums, as defined in the current Code, with particular emphasis on rules governing the franchise, are fixed in legislation for referendums in general (rather than on an ad hoc basis); such legislation cannot be changed less than a year before a referendum is held;

6.2. the body supervising the referendum is independent of government and has powers to enforce the rules, including the power to impose sanctions in case of breach;

6.3. public funds are not used by the authorities for campaigning purposes throughout the campaign period;

6.4. optimal solutions are developed, in co-operation with internet companies, with a view to designing repositories of online political advertising.

7. The Assembly considers that the existence of an independent body which would check any proposed referendum question, supervise the conduct of the campaign, take all necessary measures to ensure that this is properly conducted and possess the means to enforce its decisions and sanction possible breaches, would be one of the most efficient means to enhance member States’ compliance with referendum rules. It therefore calls on the Venice Commission to consider recommending in the revised Code the creation of such bodies in Council of Europe member States.

8. The Assembly underlines that promoting citizens’ participation in democratic deliberation, both before and after a referendum is called, can address voters’ lack of trust in and feeling of disconnection from decision-making processes. Thus the Assembly, recalling also its Resolution 1746 (2010) “Democracy in Europe: crisis and perspectives”; and drawing from existing practice in some member States:

8.1. encourages all member States to explore opportunities for citizen deliberation both prior to referendums and during the campaign, for instance through citizens’ assemblies;

8.2. invites the Venice Commission to highlight, in the revised Code, the role citizens’ assemblies and other similar mechanisms could play to ensure proper scrutiny of proposals before a referendum is called and improve the quality of information and debate during the referendum campaign.

B. Explanatory memorandum by Dame Cheryl Gillan, rapporteur

(open)

1. Introduction

1. This report examines ways in which rules and practice in the conduct of referendums throughout Europe could be strengthened and updated.

2. Upon the initiative of the Parliamentary Assembly,

the current Code of Good Practice on Referendums (CDL-AD(2007)008rev-cor, hereinafter “the Code”), including guidelines (CDL-AD(2006)027rev) and an explanatory memorandum, was adopted by the European Commission for Democracy through Law (Commission de Venise) in 2007. The Code was subsequently endorsed by the Parliamentary Assembly and the Committee of Ministers.

3. Much has happened in the years since then to suggest that a careful review is now overdue. Referendums have continued to take place frequently, which means we now have extensive evidence both on compliance with the existing guidelines and on whether those guidelines address all the issues they need to cover.

4. The Venice Commission has provided legal opinions on a number of referendums held in Council of Europe member States, most of which were on constitutional matters.

The Assembly, for its part, has also continued to observe the conduct of referendums in member States of the Organisation under monitoring or post-monitoring procedure.

Both the Venice Commission’s opinions and the Assembly’s reports have occasionally expressed concerns about procedural or substantive issues and have issued recommendations that have also been taken up by the Monitoring Committee in its relevant country reports.

5. At the same time, some countries have explored innovations in referendum practice from which policy makers throughout member States may draw lessons. The rise of the internet, and especially social media, has fundamentally changed the nature of political campaigning and of people’s expectations of democracy, in ways that demand urgent attention.

6. For all these reasons, interest in a review comes from a variety of sources. In April 2017, Lord George Foulkes and other members of the Assembly tabled a motion for a resolution entitled “A commitment to introduce rules to ensure fair referendums in Council of Europe member States”.

This mentioned votes on gay marriage in Ireland, the “Brexit” referendum in the United Kingdom, referendums on immigration in Switzerland and in Hungary, and the constitutional referendum organised in Turkey in April 2017. According to the motion’s signatories, in each of these cases, “questions have been raised about the process of the referendum and the fairness of the outcome”.

7. The Venice Commission has initiated a parallel review process.

Careful examinations of referendum practice are also being conducted within some member States: Ireland’s Citizens’ Assembly examined “the manner in which referenda are held” in January 2018;

in the United Kingdom, an Independent Commission on Referendums proposed wide-ranging reforms in summer 2018.

8. This report takes account of these developments. It is intended both to assist the Venice Commission’s review, which focuses on the law of referendums, and to encourage wider sharing of democratic good practice across member States. It addresses issues exclusively connected with national referendums. Local referendums are intentionally left aside since they pose different problems and deserve their own approach and a separate report.

9. As regards methodology, this report has been prepared drawing on documentation produced by the processes described above, especially the Venice Commission’s existing Code and its compilation of opinions relating to referendums. In light of the discussion held at the Committee on Political Affairs and Democracy in January 2018, I also drew up a further questionnaire, which was administered through the European Centre for Parliamentary Research and Documentation (ECPRD). This asked a series of questions relating to the referendum franchise, the administration of referendums, and the conduct of referendum campaigns. Thirty-eight member States replied to this questionnaire.

10. In carrying out this work, I was assisted by a referendums expert Dr Alan Renwick, Deputy Director of the Constitution Unit, Department of Political Science at the University College London. Dr Renwick was Research Director for the United Kingdom’s Independent Commission on Referendums. Besides his existing expertise and the evidence base described in the preceding paragraph, Dr Renwick consulted widely with referendums experts in a range of European countries.

11. In addition, as soon as I was appointed rapporteur, I held meetings with the President of the Venice Commission, Mr Gianni Buquicchio, and its Secretariat, to ensure that, as was the case back in 2007 with the drawing up of the Code, the exercise of updating and expanding it would also be the result of close co-operation between our Assembly and the Venice Commission.

12. Professor Nikos Alivizatos, in his capacity as a Professor of Constitutional Law and as a former member and current expert of the Venice Commission on referendums, gave me general and specific comments which I introduced into the report. He also participated in an exchange of views with the Committee on Political Affairs and Democracy in October 2018 and shared his expertise and his position with all members of the committee.

13. After that meeting, I asked members of the committee to suggest concrete amendments to the draft report. I thank those who have replied to my call although it is of course my ultimate responsibility, as rapporteur, to decide which of these suggestions to take on board and how.

14. Building on the reports and other documents set out above, as well as on the contributions by the Venice Commission and colleagues, chapter 2 of the report proposes principles that rules and practices for referendums should seek to advance. Chapter 3 then examines the implications of these principles in the following specific areas:

where the rules on referendums sit: whether in constitutional law, ordinary legislation that regulates referendums in general, legislation passed for a particular referendum, or executive orders;

when referendums take place: whether the circumstances in which referendums are held are laid down by rules or are a matter of political choice; who can call a referendum; what topics can be put to a referendum; and what procedures surround a decision on whether to call a referendum;

the structure of the vote: who can vote; the nature of the question or questions on the ballot paper; the processes for setting the question; and the administration of the vote;

the status of the result: whether it is binding or advisory, and whether the validity of the outcome is subject to special thresholds and other safeguards;

the conduct of the campaign (I) – Fairness between the sides: the role of government in the campaign; campaign finance; and media balance;

the conduct of the campaign (II) – Information available to voters: transparency of who is saying what to whom; measures to tackle misinformation; measures to ensure reliable information is available; and measures to promote direct citizen engagement;

enforcement of the rules: sanctions against those found to have breached the rules; and rules on when a referendum should be partially or fully rerun.

15. After examining the nature of the changes that may be needed in each of the areas of referendum conduct in chapter 3, chapter 4 concludes that change is needed across the board. Concrete proposals are listed in the draft resolution.

2. Principles guiding referendum design

16. It should be stressed from the outset that referendums should be embedded in the process of representative democracy and should not be used by the executive to override the wishes of parliament. In particular, in a world marked by continuous and unexpected changes, democracy should be able to adapt itself to new situations through reforms, without delay. To the extent, however, that decisions endorsed by referendums cannot be easily overruled or amended, recourse to referendums should be exceptional.

17. Before delving into the details of referendum regulation, this chapter considers broad principles that should underpin any set of guidelines about referendum practice.

18. The Venice Commission’s Code is built largely upon the four principles of universal, equal, free and secret suffrage. The application of two of these principles – those of universal and secret suffrage – does not raise many issues that are specific to referendums as distinct from elections. Those relating to equal and free suffrage do have particular implications. Specifically:

Equal suffrage: referendums generate particular requirements for balance between the sides during the campaign – for example, in media coverage and in access to funding. There are, however, differing views on what the appropriate balance should be: one conception is that each side should have equal access to resources irrespective of the level of support it enjoys; an alternative view is that the sides should have access to resources in proportion to their support levels. I discuss these two concepts further in the appropriate chapters below;

Free suffrage: free suffrage includes not only the freedom of voters to express their wishes, but also the freedom to form an opinion. This implies that the proposal, or proposals, put to a referendum should be as clear as possible, and that voters should have access to reliable and trustworthy information that will help them decide how they wish to vote. How this is best achieved in a referendum is not necessarily the same as for an election.

19. Beyond the four suffrage principles, the Venice Commission locates referendum practice within the principles that human rights should be protected, and the rule of law should be upheld. The latter principle implies that referendums should not be called extra-legally and should not be called with the intention of bypassing normal checks and balances. It also implies that referendums should not be designed to reverse the preconditions of Council of Europe membership, such as, inter alia, to reintroduce the death penalty.

20. Beyond the principles articulated by the Venice Commission, it should be recalled that democracy is not just about voting (vital though this clearly is). It is also about the processes of discussion and deliberation that lead to the proposals that are to be voted on. These processes matter for democracy in both their form and their substance. In terms of form, citizens should be able to contribute to decision-making not just by voting, but also by participating in prior discussions, including discussions leading to the delineation of the options on the ballot paper. In terms of substance, the proposals put forward in a referendum should be designed to uphold the preconditions of Council of Europe membership – including ratification of Protocols Nos. 6 and 13 to the European Convention on Human Rights (ETS Nos. 114 and 187), and not the contrary. They should also have been thought through carefully and subjected to considerable scrutiny to ensure, as far as possible, that they reflect voters’ concerns and can reasonably be expected to deliver on voters’ wishes. Such an approach helps the democratic community as a whole to make decisions that are inclusive.

21. These principles imply that, beyond the requirement that the actual vote in a referendum should be conducted in a manner that upholds universal, equal, free and secret suffrage, processes around calling a referendum and the conduct of the campaign should ensure: a) that the options put to voters in a referendum should respect human rights and the rule of law, should reflect voters’ concerns, should have been subject to careful prior scrutiny, and should be as clear as possible; b) balance between the sides (appropriately defined) and voters’ access to the information that they want in order to make their choice.

22. The following chapters examine the implications of these principles for each aspect of referendum conduct and explore possible answers to the question of who checks whether they have been complied with and how and when these checks are performed. As explained above, for each aspect I consider whether any changes are needed: to the Venice Commission’s existing guidelines, to compliance with those guidelines, or to broader norms and practices.

3. Implications of the principles for each aspect of referendum conduct

3.1. Where the rules on referendums sit

23. Rules on referendums can be set out in constitutional law (either in the Constitution itself or in other entrenched provisions), in ordinary legislation that applies to referendums in general, in ordinary legislation relating to a specific referendum, or in executive orders. Alternatively, it may be that there are no legal provisions for certain aspects of referendum conduct in some countries.

24. In Section II.2, the Code of Good Practice on Referendums states:

“a. Apart from rules on technical matters and detail (which may be included in regulations of the executive), rules of referendum law should have at least the rank of a statute.

b. The fundamental aspects of referendum law should not be open to amendment less than one year before a referendum, or should be written in the Constitution or at a level superior to ordinary law.

c. Fundamental rules include, in particular, those concerning:

• the composition of electoral commissions or any other body responsible for organising the referendum;

• the franchise and electoral registers;

• the procedural and substantive validity of the text put to a referendum;

• the effects of the referendum (with the exception of rules concerning matters of detail)

• the participation of the proposal’s supporters and opponents to broadcasts of public media.”

25. The essence of these guidelines does not need to change. As the Venice Commission argues, it is important that basic rules around referendums are fixed for referendums in general; otherwise there is a danger that they will be manipulated to suit the interests of those calling a specific referendum.

26. There is a question as to whether the list of “fundamental rules” should be extended to other aspects of referendums. This particularly applies to the aspects of the regulation of referendum campaigns that are examined in chapters 3.5 and 3.6 of this report.

27. In practice, most Council of Europe member States adhere to these guidelines for at least some of the fundamental rules of referendums: that is, most set out a framework for the conduct of referendums in their constitutional law and/or in ordinary standing legislation. But not all do so for all fundamental rules. In the United Kingdom, for example, standing rules for the most basic aspects of referendum conduct are set out in ordinary legislation. But this legislation does not address the referendum franchise, which is decided on an ad hoc basis for each referendum. In consequence of this, attempts have been made by both sides in recent referendums to change the franchise to their advantage.

Among countries that have held referendums in recent decades, Norway stands out for having no standing legislative framework for referendums at all. All aspects of regulation are therefore decided ad hoc.

28. It would be desirable, therefore, for all countries to ensure that their standing referendum provisions (whether those are in constitutional law or in ordinary law) address all of the fundamental rules. In addition, the principle should be maintained that fundamental rules will not be changed in the year preceding a referendum to which those rules apply.

3.2. When can referendums take place?

29. Some of the most basic aspects of referendums relate to when they can be held: on whose instigation, on what issues and through what procedures.

30. Some countries also have provisions that make referendums mandatory in certain circumstances. The Venice Commission offers no guidelines on whether there should be mandatory referendums, and it should maintain that approach: whether such referendums are appropriate will depend upon the wider constitutional structure and traditions within each country.

3.2.1. Who can call a referendum

31. Referendums are most often called either by the legislature (the national legislature for national referendums, and equivalent bodies at other levels of government) or by citizens via petition. In some countries, it is also possible for a referendum to be called by the executive (most commonly, the President), by a minority of parliamentarians (where they wish to put a measure that has been passed by the legislature to a popular vote), or (in the case of a national referendum) by a number of regional assemblies.

32. The Code of Good Practice on Referendums contains no guidelines on who should be able to call a referendum. For the most part, this is appropriate: this is again a matter for determination within individual countries.

33. Several Venice Commission reports or opinions on recent referendums do, however, highlight concerns about constitutional referendums that are called by the executive over the heads of the legislature. An executive power to put to the people a bill that has been passed by the legislature, before signing it into law, is not inappropriate: this effectively supplements the normal decision-making process with a popular vote. But if the executive can put to voters a proposal for a constitutional amendment that has not been passed by the legislature, this risks subverting normal processes of democratic discussion and scrutiny.

34. It should be stated in the Code that it should not be possible for the executive to call a referendum on a constitutional proposal, except where the decision to hold a referendum has already been endorsed by the legislature, or where the proposal that is put to a popular vote in a referendum has been passed by the legislature. There is a strong case for extending this restriction to cover referendums on all matters, not just constitutional laws.

3.2.2. What can be put to a referendum

35. The current Code makes two stipulations regarding the subject matter that can be put to a referendum. First, at a general level, it says:

“Texts submitted to a referendum must comply with all superior law (principle of the hierarchy of norms). They must not be contrary to international law or to the Council of Europe’s statutory principles (democracy, human rights and the rule of law)” (section III.3).

36. Second, much more specifically, it says:

“It is advisable for constitutional rules relating to referendums to be put to a referendum, compulsorily or at the request of a section of the electorate” (section III.5.d).

37. The first of these stipulations is clearly appropriate in light of the Council of Europe’s commitment to human rights and the rule of law. The second is justified in terms of parallelism in procedures.

38. Some countries’ Constitutions contain further provisions regarding the topics on which referendums are permissible. For example, some, including Denmark, Italy, and Portugal, do not allow referendums on matters of finance or taxation. Decisions on what can and cannot be put to a public vote are, for the most part, appropriately left to individual countries.

39. Nevertheless, two further general guidelines might be developed. The first is that the proposals put to a public vote should be as clear as possible.

40. There is a danger that a referendum about a principle or a “generally worded proposal” does not give sufficient clarity about what is proposed to permit voters to make an informed choice. In their Joint Opinion on the draft law “On introduction of changes and amendments to the Constitution” of the Kyrgyz Republic of 2015, the Office for Democratic Institutions and Human Rights of the Organization for Security and Co-operation in Europe (OSCE/ODIHR) and the Venice Commission said: “the matters that are being decided by a referendum should never be too imprecise or too vague, and the draft legislation adopted in this manner should not leave important matters to future laws … Asking citizens to engage in such a “blind vote” would dilute the very purpose of popular referenda, and should be avoided.”

41. Similarly, the United Kingdom’s Independent Commission on Referendums, established in autumn 2017, drew a distinction between pre- and post-legislative referendums: the former ask about a question of principle or generally worded proposal, the latter about a specifically worded draft that has already passed through the legislature. The Independent Commission pointed out that post-legislative referendums typically give much greater clarity. It recommended that referendums should be post-legislative wherever possible. Where a standalone post-legislative referendum is not possible – as where a referendum is deemed necessary in order to decide whether to start negotiations on the proposed change – the Independent Commission said that a process should be set out requiring two referendums if the first referendum does not allow voters to choose between the options that are ultimately available.

The Venice Commission’s Code should include the same recommendation.

42. The second general guideline that might be added is that, as far as possible, referendums should only be called on subjects that are likely to attract significant public interest. This stems from the principle stated in chapter 2 that referendums should assist in allowing the democratic community as a whole to make decisions that are inclusive and considered. If participation in a referendum is very low, it is possible that a decision may be taken that does not reflect the wishes of the community as a whole.

43. The current Code recommends that referendum results should not be subject to turnout thresholds, and I propose that that should not change (see chapter 3.4.2 below). In the absence of such thresholds, however, there is a danger that some matters could be decided on very low turnouts. The best way to avoid this (unless compulsory voting is in place and enforced) is to use referendums only where turnout is likely to be high. That clearly cannot be a legal stipulation, as turnout is unpredictable. But it would be desirable as a norm. It also implies that legal requirements to hold referendums on relatively technical matters – such as requirements that a referendum should be held on any constitutional amendment no matter how uncontroversial it is – should be avoided.

3.2.3. Processes for calling a referendum

44. As the principles set out in chapter 2 indicate, the early stages of a referendum process are vital for the democratic quality of the process as a whole. Referendums typically offer only two options. While the choice between those options is important, the choices that lead to the delineation of these options deserve equal attention. Furthermore, if the purpose of a referendum is to engage citizens directly in decision-making, it is desirable that that should include the phase of delineating options as well as the final decision.

45. The current Code recognises these points to some degree

and reflects the principle that any proposal put to a public vote should have been subject to detailed prior scrutiny. It presumes that, where parliament calls a referendum itself, such prior scrutiny will necessarily take place.

46. It would be desirable to state more explicitly the general principle that any proposal put to a public vote should be subject to detailed parliamentary scrutiny. That may indeed be automatic in the case of a post-legislative referendum – though it should be emphasised that the fact that the final decision lies in the hands of the people is never a reason for parliament to subject a proposal to any less scrutiny than it would otherwise. In the case of a pre-legislative referendum, it is not automatic, even if the referendum is called by parliament. In the case of such a referendum, the proponents of the proposal should publish a detailed prospectus of what they propose and expect to achieve, and this should be subject to detailed parliamentary scrutiny.

47. In addition, opportunities for citizen deliberation should be explored. Ireland has recently pioneered the use of “citizens’ assemblies” prior to referendums. A Constitutional Convention comprising politicians and randomly selected members of the public led to the referendum on same-sex marriage in 2015. A citizens’ assembly composed entirely of members of the public paved the way for the referendum on abortion liberalisation in 2018. In both cases, the members of the assembly met over multiple weekends to learn about the issues, deliberate in depth among themselves, and reach conclusions. Their recommendations were vital in shaping the draft changes that the Oireachtas subsequently put to voters in the referendum. They gave politicians a much deeper understanding of informed public opinion and helped frame the debate during the referendum campaigns that followed. By bringing solid evidence and reasoned arguments to the fore, and by placing ordinary citizens at the heart of the discussion, they may have reduced polarisation on these contentious topics. They ensured that voters were involved throughout the process of policy development.

48. Citizens’ assemblies have been used in the early stages of referendum processes in several countries besides Ireland – most notably in Canada.

Citizens’ assemblies and other related public deliberative processes have also been demonstrated to be successful in a variety of other contexts.

Still, they remain relatively novel as democratic instruments. The Parliamentary Assembly should encourage member States to experiment in their usage as a means of enhancing democratic debate and engagement.

The Venice Commission should also highlight them as a mechanism for ensuring that proper scrutiny takes place in the early stages of the referendum process.

49. Chapter 3.4.2 below, raises further points regarding the calling of referendums through citizen petitions.

3.3. The structure of the vote

50. Once a referendum has been called, the next set of issues concerns how the vote is structured: who can vote in a referendum, the nature of the question on the ballot paper, the process by which that question is set, and the administration of the vote.

3.3.1. Who can vote

51. The current Code states that the same principles of universal franchise should apply to referendums and to elections (section I.1.1) and that electoral registers should be maintained so as to be accurate and up to date (section I.1.2). As noted in chapter 3.1, it also says that the franchise should be set well in advance of any referendum to which it applies and thus should not be tailored to suit the interests of one or other side in a particular vote.

52. These provisions are all appropriate. Furthermore, responses to our questionnaire show that compliance with them is generally high: most countries do provide – either in their Constitutions or in general legislation – that the referendum franchise is the same as the franchise for national legislative elections. Still, there are a small number of exceptions to this and in countries where the franchise is set in ordinary law, vigilance is needed to ensure this is not manipulated with particular policy goals in mind.

3.3.2. The nature of the question

53. I discussed in chapter 3.2.2 the nature of the subject matter that can be put to a referendum. More specific than that is the nature of the question itself. I stated that the proposal put to voters in a referendum should be clear. The wording of the question should be clear too.

54. The Code also offers guidelines on the structure of the question.

55. These provisions are for the most part appropriate. The stipulation that “voters must be able to answer the questions asked solely by yes, no or a blank vote” may, however, be too restrictive in two senses.

56. First, it may be preferable, at least in some cases, to ask voters to choose between options rather than to answer a Yes/No question. In the case of the United Kingdom’s referendum on membership of the European Union, for example, the question was “Should the United Kingdom remain a member of the European Union or leave the European Union?”. Voters were asked to place a cross next to either “Remain a member of the European Union” or “Leave the European Union”. In explaining its decision to propose such wording, the United Kingdom Electoral Commission said that some of its consultation respondents felt that a Yes/No question (“Should the United Kingdom remain a member of the European Union? – Yes/No”) was biased: first, because it mentioned only the remain option in the question, and, second, because the “Yes” response was the status quo.

The Electoral Commission did not say that Yes/No questions are always problematic, and indeed it had proposed such questions for previous referendums. But there may be concerns that “acquiescence bias” – a tendency for people to favour a “Yes” response over a “No” response – means that such questions are inherently biased.

57. Evidence on these points is not strong enough to suggest that Yes/No questions should be avoided, and indeed the vast majority of referendums continue to employ questions of that form. Nevertheless, it would be desirable for the Venice Commission’s guidelines to allow that a non-Yes/No question would sometimes be preferable.

58. Second, it may sometimes be appropriate for a referendum to offer more than two options. Such referendums are rare, but they do occur: there were, for example, three-option referendums in Sweden in 1980 and Slovenia in 1998, and New Zealand allowed voters to choose among five or six options in referendums in 1992, 2011 and 2015. If more than two options have significant backing, it may be better to allow voters to choose among these than to force an artificial binary choice. Such referendums are complex: experiences in Sweden and Slovenia show that they should not be held using a first-past-the-post voting system, as they may then deliver an ambiguous result, so either a preferential ballot or a multi-question (and possibly multi-round) format is required.

But the guidelines should at least allow that they can be held.

3.3.3. Processes for setting the question

59. If the principles for referendum questions considered in the preceding subchapter are to be upheld, there must be a process designed to achieve this end. On this, however, the current Code is largely silent.

60. In some countries, referendum questions have a fixed format, asking whether voters approve a specific bill or draft constitutional amendment. A danger with such questions is that they can become extremely complex.

61. As discussed further in chapter 3.6.3, the Code does state that voters should in such cases receive the text that they are voting on, together with a detailed explanation of what changes it involves. It may be preferable, for example, to break the question up, so that a preamble gives the name of the bill and a short explanation of its content, and voters are then asked a short question on whether they approve it. What is most appropriate will vary depending on traditions in particular countries. But the Code should state that, where a fixed format for referendum questions is used, that format should be reviewed periodically to ensure that it provides for a ballot paper that is clear, accessible, and unbiased.

62. In other cases, referendum questions do not follow a fixed format, but are rather determined each time a referendum is held. Where the referendum is called by the legislature, it is in most cases the legislature that decides the question. The current Code provides no guidelines on the procedures that should be followed. That is a clear lacuna that should be filled: checks are required to ensure that the principles set out in chapter 3.3.2 are met. Otherwise, there is a danger that those calling the referendum will bias the question to their advantage. There have been accusations of manipulation in the wording of recent referendums in, for example in Greece,

Hungary

and “the former Yugoslav Republic of Macedonia”;

an independent question-setting process would help to avoid these.

63. One country that does have a procedure for impartial checking of proposed referendum questions is the United Kingdom. The legislation enabling a referendum must contain the referendum question. When a bill for a referendum is presented to parliament, the Electoral Commission conducts detailed scrutiny of the proposed question, convening focus groups and consulting with campaigners, language experts and others. It assesses whether the question is readily intelligible and (perceived as) neutral.

It also tests potential alternative wordings. It then makes a recommendation on the precise wording that should be used. To date, it has recommended changes to all of the questions that it has assessed. Parliament is not obliged to accept these recommendations, but it has done so in the great majority of cases.

64. In other countries, it may be that a different procedure – for example, a judicial procedure – would be more appropriate. The Code should therefore stipulate simply the requirement that proposed referendum questions should be subject to rigorous scrutiny by an impartial body before they are set. This process of scrutiny should assess the question, at minimum, for clarity, absence of bias, and conformity with the other principles currently set out in the Code.

65. The current Code does include provision for question assessment in the case of citizen-initiated referendums.

66. In line with the recommendations above in relation to referendums initiated by the legislature, this provision should be revised to provide that this check on question wording should be made by or in light of a rigorous assessment by an impartial body using the criteria I have discussed.

3.3.4. Administration of the vote

67. The Code sets out guidelines for voting procedures (section I.3.2.a), establishment of the result (section I.3.2.b.iii), organisation of the referendum (section II.3.1), and observation of the referendum (section II.3.2). These are all appropriate and I do not propose changes to them.

68. Compliance with these provisions is important but is not always universal. For example, some countries – including some long-standing democracies – lack an independent body responsible for organising referendums. In other cases, while the election management body is formally independent, it is in practice captured by those in power. OSCE and Assembly reports have occasionally expressed concerns about the independence of election bodies in relation to referendums in some member States. Unless there is a strong and unchallenged tradition of administrative impartiality on electoral matters, such arrangements can create serious risks of violation of the rule of law and impartiality of administration, and every effort should be made to eliminate them. The body supervising the referendum must be independent of government and should have powers to implement the rules and sanctions to enforce them.

3.4. The status of the result

69. Two important issues relate to the status of a referendum result: whether that result is binding or advisory; and whether it is subject to any special thresholds or other safeguards.

3.4.1. Binding or advisory?

70. A distinction is commonly drawn between referendums that are legally binding and those that are in law only advisory. The Code currently makes three principal stipulations on this point. First, it states that:

“The effects of legally binding or consultative referendums must be clearly specified in the Constitution or by law” (section III.8.a).

Second, it provides guidelines on what those effects should be:

“When the referendum is legally binding: For a certain period of time, a text that has been rejected in a referendum may not be adopted by a procedure without referendum. During the same period of time, a provision that has been accepted in a referendum may not be revised by another method. The above does not apply in the case of a referendum on partial revision of a text, where the previous referendum concerned a total revision. The revision of a rule of superior law that is contrary to the popular vote is not legally unacceptable but should be avoided during the above-mentioned period. In the event of rejection of a text adopted by Parliament and put to the popular vote at the request of a section of the electorate, a similar new text must not be put to the vote unless a referendum is requested” (section III.5.a).

Third, it advises on when referendums should not be binding:

“Referendums on questions of principle or other generally-worded proposals should preferably not be binding. If they are binding, the subsequent procedure should be laid down in specific rules” (section III.8b).

71. The first of these provisions is clearly appropriate: it is essential that the status of a referendum should be clear in advance.

72. The second provision is also appropriate as far as it goes. In addition, however, it should be acknowledged that there are intermediate possibilities between fully binding and fully advisory referendums. A binding referendum as conceived in the current Code binds the legislature not to legislate in a manner contrary to the referendum outcome. One step down from this, it is possible for a referendum to bind only the executive. In the United Kingdom, for example, the principle of parliamentary sovereignty makes it impossible to bind parliament. A 2011 referendum on the voting system, however, had it passed, would have bound the government to implement the Alternative Vote electoral system, which was set out in the legislation that enabled the referendum. Parliament could legally have overridden the referendum result and amended the legislation, but the executive was bound. At a further step down, a referendum may bind government or parliament to consider a matter, but not to take a particular decision. Under the provision for abrogative citizen-initiated referendums that was in force in the Netherlands between 2015 and 2018, for example, a public vote for abrogation required that parliament considered a bill to repeal or amend the legislation in question, but parliament was legally free to reach any conclusion it wished.

Such provisions may be appropriate so long as they are clearly set out in law and publicised in advance of the referendum. At the very minimum, after an advisory referendum, the executive or legislature should always consider the results and recommend a course of action. Consultative referendums have been abolished in the Netherlands as they tended to be interpreted as binding by the population.

73. The third provision is also appropriate. It should be recalled, however, that, in practice, most referendums are treated as politically binding even where, in law, they are only advisory. That is, I am aware of only a small number of cases in which elected representatives have chosen not to follow (or substantially follow) the result of an advisory referendum. This means that the mere fact that a referendum is only advisory in law is no reason to think it can be treated in a more casual manner than can a formally binding vote. As indicated above, even in the case of a non-binding pre-legislative referendum, everything should be done to ensure the proposals are as clear as they can be and have been subject to rigorous scrutiny before the options on the ballot paper are fixed and the referendum is called.

3.4.2. Thresholds and other safeguards

74. The simplest threshold for a referendum is that the proposal passes if it is backed by 50% + 1 of the valid votes cast. This threshold has commonly been applied in referendums in countries such as Austria, France, Iceland, Ireland, Sweden and the United Kingdom. In a small number of other cases – for example, in several provincial referendums in Canada – that threshold has been raised to a supermajority requirement: in the Canadian cases, 60% of those voting had to support the proposed change.

Much more common than a raised threshold is an additional threshold that exists alongside the simple majority requirement. There are three main types: a turnout threshold (or “turnout quorum”) requires that a certain proportion of the eligible electorate participates in the vote; an electorate threshold (or “approval quorum”) requires that a certain proportion of the eligible electorate supports the proposed change; a multiple majority threshold requires that a majority be attained not only across the country as a whole, but also in a specified number of regions or other sub-units within the country. Turnout and electorate thresholds are widely used in Europe: for example in Denmark, Hungary, Italy, Lithuania and Poland. Multiple majority thresholds are rarer, but are employed, notably, in Switzerland.

75. The current Code advises against both turnout and electorate thresholds. It states:

“It is advisable not to provide for: (a.) a turn-out quorum (threshold, minimum percentage), because it assimilates voters who abstain to those who vote no; (b.) an approval quorum (approval by a minimum percentage of registered voters), since it risks involving a difficult political situation if the draft is adopted by a simple majority lower than the necessary threshold” (section III.7).

76. These stipulations are appropriate and concur, for example, with the recommendations of the United Kingdom’s recent Independent Commission on Referendums. The damaging impact that turnout thresholds can have by encouraging disengagement campaigns has frequently been observed in referendums.

Perceptions of electorate thresholds may vary more, depending on political traditions in a particular country.

77. Supermajority requirements are similar in their effects to electorate thresholds, and advice against them might be added to the Code.

78. In contrast, multiple majority thresholds may be appropriate, particularly in federal systems. Such a threshold is used, for example, in Switzerland, where referendums must secure a majority in the country as a whole and in the majority of the cantons. It is for individual countries to determine whether multiple majority thresholds are suitable in the context of those countries’ wider constitutional arrangements.

79. Special thresholds such as these are designed to make it less likely that decisions will be taken that go against the considered will of the population as a whole. As the preceding discussion indicates, that is a noble goal: referendums should not allow major changes to be pushed through without careful reflection and, as far as possible, broad support. Special thresholds are not the best means for achieving this goal. It is important, therefore, to attend carefully to the question of what might be a better approach. Other parts of this report suggest several avenues:

there should be careful and widespread scrutiny of proposals before a referendum is called, in order to reduce the chances that ill-considered proposals reach the ballot paper (chapter 3.2.3);

where possible, the decision-making processes of which referendums are part should include citizens’ assemblies or other similar opportunities for public deliberation, so that people are engaged throughout the process of exploring the issues and developing proposals (chapter 3.2.3);

referendums should be embedded in the process of representative democracy. They should not be used by the executive to override the wishes of parliament. Where they are initiated by citizens via petition, the proposals should be scrutinised in depth by parliament, which may devise counter-proposals (chapters 2 and 3.2.3);

proposals should not generally be put to a referendum unless they are likely to attract high levels of public engagement and participation (chapter 3.2.2);

referendum campaigns should be conducted so that, as far as possible, voters have access to balanced, quality information on the options (chapters 3.5 and 3.6).

80. These conditions are hardest to meet in the case of citizen-initiated referendums: by definition, the process is then less subject to control by the authorities. A number of safeguards can nevertheless be considered:

the number of signatures required to trigger a referendum should be high enough to ensure that the proposal has genuinely wide support. The safeguards against the commercialisation of signature collection in the current Code (section III.4.e) should be strictly applied;

consideration should be given to the development of procedures whereby a citizens’ petition would lead not directly to a referendum, but rather to a citizens’ assembly, at which the petition sponsor’s proposals could be carefully scrutinised, alternatives could be considered, and recommendations could be reached. In alternative versions of this approach, it could be that only the assembly’s recommendation goes forward for decision in a referendum, or that the original proposal and the assembly’s counter-proposal go forward. It could also be open to the assembly to recommend or decide against a referendum. Such approaches would be innovative, but all offer ways of building on existing good practice in European democracies to strengthen the participatory and deliberative quality of decision-making.

3.5. The conduct of the campaign (I): Fairness between the sides

81. How a referendum campaign is conducted is fundamentally important for the democratic quality of the process as a whole. Our discussion of principles in chapter 2 identified two key values: the campaign should be conducted in a manner that is fair between the two sides; and it should enable voters to access the information they want. The first of these is developed in this chapter and the second in chapter 3.6.

82. One requirement for fairness is that there should be sufficient time for all sides to develop and make their points and for voters to hear the arguments and form an opinion. It should not be possible to call a “snap” referendum at very short notice, such that opponents of the proposal have insufficient time to organise. The absolute minimum time between calling a referendum and polling day could be set at four weeks. A considerably longer period of preparation is desirable, however, particularly if the topic has not already been subject to widespread public discussion.

83. Fairness in referendums also requires balance. As mentioned in chapter 2, there are two concepts of balance. One says that there should be equality between the sides in a referendum, irrespective of the level of support they have. The other says that the resources available to each side should be proportional to their support. The current Code combines elements of both of these, as do the rules in many countries. That is appropriate: an unduly pure application of either conception would create problems.

3.5.1. The role of government in the campaign

84. The Code prescribes “a neutral attitude by administrative authorities” in referendums (section I.2.2.a). It develops this further:

“Contrary to the case of elections, it is not necessary to prohibit completely intervention by the authorities in support of or against the proposal submitted to a referendum. However, the public authorities (national, regional and local) must not influence the outcome of the vote by excessive, one-sided campaigning. The use of public funds by the authorities for campaigning purposes must be prohibited” (section I.3.1.b). (See also section II.3.4.b.)

85. This strong prohibition against the use of public funds to campaign on one side is highly desirable. When a referendum is held, it is for voters to make up their own minds. While it is important that voters should know the positions of their elected representatives, if government is allowed to campaign strongly, there is a danger that the referendum mechanism may be used by governments to entrench their own authority and suppress opposition. Serious concerns of this nature have been raised by referendums in a number of member States in recent years.

86. Compliance with this prohibition is, however, relatively rare across member States at present: the evidence gathered for this report identifies only seven member States that have clear government impartiality rules (Ireland, Italy, Latvia, Lithuania, Portugal, Spain and the United Kingdom), though it is possible that there are some others. In fact, this is the area of the Code where compliance is weakest and member States should take steps to address this.

87. In addition, the Code should itself be strengthened in one respect. While it says that public funds should not be used for campaigning purposes, it does not say for how long this requirement should apply. In the United Kingdom, for example, this rule applies only during the final four weeks before polling day, whereas campaigning typically proceeds for much longer. The Code should be explicit that the prohibition should last throughout the campaign period.

3.5.2. Campaign finance

88. Campaign finance regulations may address a variety of issues. As referendum experts Theresa Reidy and Jane Suiter point out, they can include limits or bans on donations or on spending, transparency requirements, public funding provisions, and enforcement rules.

89. The current Code is most expansive in relation to public funding. The prescription of “a neutral attitude by administrative authorities” noted above is explicitly applied to “public funding of [the] campaign and its actors” (section I.2.2.a). The Code then elaborates:

“Equality must be ensured in terms of public subsidies and other forms of backing. It is advisable that equality be ensured between the proposal’s supporters and opponents. Such backing may, however, be restricted to supporters and opponents of the proposal who account for a minimum percentage of the electorate. If equality is ensured between political parties, it may be strict or proportional. If it is strict, political parties are treated on an equal footing irrespective of their current parliamentary strength or support among the electorate. If it is proportional, political parties must be treated according to the results achieved in the elections” (section I.2.2.d).

90. This guidance encompasses the alternative conceptions of fairness indicated above. There is a strong case for saying that, in the case of public funding, the principle of equality between the sides should take precedence. The purpose of public funding is to ensure that voters can hear the arguments on each side of the debate: it thus ensures that the minimum requirements for democratic choice take place. The same minimum should apply to both sides.

91. The Code currently has little to say on other aspects of campaign finance. In fact, it has only two provisions:

“Political party and referendum campaign funding must be transparent.” (section I.2.2.g) “The principle of equality of opportunity can, in certain cases, lead to a limitation of spending by political parties and other parties involved in the referendum debate, especially on advertising” (section I.2.2.h).

92. These points should be developed further. Transparency is of paramount importance in democracy. Concerns about corruption and attempts to buy undue influence are widespread, and transparency is an essential first step for addressing them.

Such transparency should apply both to the sources of campaign funds and to how those funds are spent. In particular:

at the very least, the sources of campaign funds should be revealed to an independent regulator. For donations, this should apply to donations above a minimum threshold;

there should be rules on who may contribute to campaign funds: foreign donations, for example, are often prohibited and such prohibition is desirable. Disclosure of funding sources should enable the independent regulator to police these rules;

in addition, donations above the minimum threshold should be made public: voters have a right to know if individuals or bodies provide substantial resources to political campaigns;

campaigners that spend more than a minimum threshold should be required to submit detailed spending returns. These should make it clear what money has been spent on and from whom services have been purchased.

93. There is no harm in one side being able to outspend the other if it has greater popular support and/or greater support among elected representatives: voters are perfectly entitled to let what others think influence their own views, so the strength of the campaign on each side may be useful information. Nevertheless, if one side can overwhelmingly outspend the other, that may inhibit free choice, particularly if much of that funding comes from a small number of wealthy sources. Beyond transparency, therefore, constraining spending through spending limits, donation limits, or both, is also desirable. The rules applied should, as far as possible, be consonant with other rules and traditions in the country concerned.

94. Campaign and party finance in relation to all electoral processes is a wide-ranging subject and is of such importance that it should probably be addressed by the Venice Commission in a document separate to its Code of Good Practice in Electoral Matters and its Code of Good Practice in Referendums, which could set out good and bad practice without being unduly prescriptive. As such, it should therefore be considered as a separate matter from this report, although it has to be referred to as it remains a significant element of referendums and their conduct.

3.5.3. Media balance

95. The Code offers two kinds of guidelines in relation to media coverage of referendum campaigns.

96. First, the Code says that coverage in public broadcasting should be balanced:

“In public radio and television broadcasts on the referendum campaign, it is advisable that equality be ensured between the proposal’s supporters and opponents.” (section I.2.2.b). “Balanced coverage must be guaranteed to the proposal’s supporters and opponents in other public mass media broadcasts, especially news broadcasts. Account may be taken of the number of political parties supporting each option or their election results” (section I.2.2.c).

97. Second, it says that other broadcast coverage should give at least some access to both sides:

“Financial or other conditions for radio and television advertising must be the same for the proposal’s supporters and opponents.” (section I.2.2.e). “In conformity with freedom of expression, legal provision should be made to ensure that there is a minimum access to privately owned audiovisual media, with regard to the referendum campaign and to advertising, for all participants in the referendum” (section I.2.2.f).

98. There is legitimate debate about how far the two notions of fairness should apply in the context of public and private broadcasting. Ireland has some of the strictest rules requiring perfect balance between the sides. There is concern, however, that this sometimes creates “false balance”, on issues where the great bulk of opinion falls clearly on one side.

This is particularly problematic given Ireland’s requirement for a referendum on any constitutional amendment, which means that public votes are sometimes required on very uncontroversial proposals, such as local government reform (1999) and allowing judges’ salaries to be reduced (2011). The precise requirements should therefore be tailored to the circumstances of individual countries.

99. The Code makes no stipulations regarding non-broadcast media. This is particularly significant in light of the rise of digital media, and especially social media. Indeed, the increasing convergence between printed, broadcast, and digital media means that it may become increasingly difficult to justify markedly differing approaches to regulation of these media sectors. The United Kingdom’s Independent Commission on Referendums recommended “a comprehensive inquiry into the future of political advertising across print, broadcasting and online media”.

Such careful thought is needed in other countries too. The Venice Commission is also currently working on these issues and it is to be hoped that it will develop relevant guidelines for both referendums and elections.

3.6. The conduct of the campaign (II): Information available to voters

100. The second principle identified above regarding campaigns was that voters should be able to access the information they want. It is not for the authorities or anyone else to dictate to voters what information they ought to absorb before voting. But if voters want additional information before casting their vote, they should be able to access it from sources that they trust.

101. This is an aspect of referendums where the current Code makes only limited provisions, however it has become much more salient since this Code was drafted. Recent years have seen considerable concern over the quality of political discourse and the prevalence of so-called “fake news” and other forms of unreliable information. In addition, the rise of digital media has transformed how citizens receive information about politics. Particular concern focuses on highly targeted online advertising that can be minutely tailored to suit particular audiences but is invisible to everyone else (the phenomenon of so-called “dark ads”).

It is essential that the Code should keep abreast with these developments.

102. Recent research identifies four approaches to regulating the conduct of campaigns that are intended to improve the quality of information and discourse: promoting transparency; confronting misinformation; providing quality information; and facilitating citizen deliberation. In many ways they are not specific to referendums but apply also to elections in general. I will focus, however, on their application to referendums.

3.6.1. Transparency of who is saying what to whom

103. Chapter 3.5.1 discussed the importance of transparency in relation to campaign finance. Transparency is also essential in relation to campaign messaging: voters should be able to find out what claims campaigners are making and who is making them. They should also be able to see if campaigners are putting out contradictory messages to different groups of voters or seeking to portray different images of themselves to different voters.

104. The current Code makes no mention of this. But the rise of online “dark ads” has considerably increased its prominence. New guidelines are urgently needed.

105. At a minimum, all advertising materials, irrespective of medium, should be clearly labelled, so that citizens can readily identify who has produced them.

106. Beyond this, there have been increasing calls over the past year or so for the development of publicly accessible repositories on online advertising, so that anyone can see the advertisements that campaign groups are producing. The United Kingdom’s Independent Commission on Referendums, for example, recommended “the creation of a publicly available and searchable online repository of political advertisements, which should include the advertisement itself and information on when it was posted, which groups were targeted, and how much was spent”.

107. The major internet companies have begun to make moves in this area. For example, Facebook launched its own repository of political advertising for the first time for the 2018 Irish referendum on abortion liberalisation, and it ran an improved version for the 2018 United States mid-term elections.

But it would be unwise to leave such provisions entirely to the companies themselves. First, that entails outsourcing important aspects of campaign regulation to private multinationals, whose incentives might not lead them to do what is best for the democratic community as a whole. Second, transparency would be greatest if a single repository were created covering all online platforms.

108. It may be too early to lay down precise legal guidelines on these matters. But careful work is needed by governments and internet companies co-operating internationally to urgently develop optimal solutions.

3.6.2. Tackling misinformation

109. Beyond transparency, the question arises of what should be done where inappropriate messaging is identified. The current Code is silent on this.

110. There have been some calls in recent years for measures to ban campaigners or others from making false claims, to require them to stop making claims that have been found to be false, and to punish them if they fail to do so.

I am aware of very few democratic jurisdictions in which this is done at present, however, and none among the Council of Europe member States. There are good reasons for that: this approach is fraught with dangers. First, it is susceptible to abuse by unscrupulous authorities, particularly where state neutrality is inadequately guaranteed. Second, in some circumstances it may backfire: campaigners who have been told to desist from making certain claims may successfully portray themselves as martyrs fighting against an establishment that refuses to allow certain truths to be spoken. Third, it may have little practical effect and therefore become discredited. In order not to constrain free speech, such mechanisms could be used only against manifestly false claims. This means, however, that claims that are not strictly false but clearly intended to mislead will often get through. Confidence in the system may be undermined as a result.

111. The Code is therefore right not to recommend this approach (though such an approach could be effective in some jurisdictions, where State neutrality and public confidence in such neutrality are both high). This is again in line with the recommendation of the United Kingdom’s Independent Commission on Referendums.

112. Other approaches are possible and should be encouraged. One traditional approach is rigorous independent press regulation, through which misreporting can be identified and corrected. The rise of digital media again means that the future design of such regulation requires careful and urgent consideration.

113. More novel is the process of “fact-checking”, through which claims made by campaigners and others are subjected to rigorous scrutiny by a scrupulously neutral independent body, which then offers a verdict on their accuracy. The intention here is not to ban misinformation, but to expose it, thereby helping citizens make their own minds up in an informed way. There have been concerns about “backfire effects” in relation to fact-checking,

though some studies suggest that these may have been exaggerated.

It appears clear that careful fact-checking and prominent reporting thereof have an important role to play in the democratic mix. Their development and publicity should be encouraged and facilitated.

114. In some cases, such a fact-checking function may be performed by an official body. In Ireland, for example, the referendum commissions that are established for each referendum to provide information on the options (see chapter 3.6.3) have in some cases also opted to intervene during the campaign, calling out campaigners for making false claims and urging them to desist.

Similarly, in the 2016 referendum on European Union membership in the United Kingdom, the UK Statistics Authority intervened on several occasions to upbraid campaigners for making inaccurate use of official statistics.

Where a public body is sufficiently independent and commands sufficient public trust to perform such a function, it may be appropriate for it to do so.

3.6.3. Provision of information

115. The third approach to improving information quality during referendum campaigns is for the public authorities themselves to provide neutral and reliable information. The current Code does include provisions in this regard. The Code says:

“The authorities must provide objective information. This implies that the text submitted to a referendum and an explanatory report or balanced campaign material from the proposal’s supporters and opponents should be made available to electors sufficiently in advance, as follows:

i. they must be published in the official gazette sufficiently far in advance of the vote;

ii. they must be sent directly to citizens and be received sufficiently far in advance of the vote;

iii. the explanatory report must give a balanced presentation not only of the viewpoint of the executive and legislative authorities or persons sharing their viewpoint but also of the opposing one” (section I.3.1.d).

In addition, as noted in chapter 3.2.3, it says that parliament should be able to express its opinion on referendums that it has not itself initiated (section III.6). The explanatory memorandum clarifies that this is in part because of the importance that voters should be informed (section III.6, paragraphs 46-48).

116. Practice around the provision of information in referendums varies widely across Europe. Many countries provide no substantial official information, whereas others provide quite extensive information. In Switzerland, the authorities prepare a booklet that explains the proposal and sets out the positions of the Federal Assembly and the proposal’s initiators. In Liechtenstein, a brochure is provided that must give space for both sides to express their positions – an approach that has also been adopted in some referendums in the United Kingdom. In Ireland, an independent Referendum Commission is established for each referendum, comprising four senior officials (who hold their roles ex officio) and a senior judge, who acts as chair. It must explain the proposal being put to the vote, which in practice it does through a leaflet sent to all voters, a website, extensive advertising, and media appearances by the chair. It is sometimes criticised, however, for not providing all of the information that voters want: it limits itself to the legal effects of the proposal, meaning that it can neglect further effects that voters may be particularly interested in.

Outside Europe, there are some examples of more extensive public information provision that does explore further effects, most notably in New Zealand.

117. It is not realistic to expect a government or parliament with a view on a referendum to produce balanced information materials. The Code should therefore be amended to clarify that an independent body should be responsible for provision of official information.

118. At a minimum, the information provided should set out the referendum question and the details of when and how people can vote. Where possible – in particular where trust in the independence and authority of the body providing the information is sufficient – it should also provide explanations of proposals, and it may go further, providing information allowing voters to weigh the options against their own evaluative criteria.

119. Where the official information provision does not cover all these information types, the work of media outlets and of independent experts is crucial. They should be encouraged to provide extensive information that is accurate, accessible, and unbiased, and that addresses the issues that voters care about.

3.6.4. Citizen engagement

120. The final approach to enhancing the quality of information and debate in referendum campaigns engages citizens directly in deliberation about the referendum topic. The use of citizens’ assemblies and other similar mechanisms before a referendum is called to deepen understanding of voters’ considered perspectives and thereby help frame the issues and the debate was mentioned above. Similar mechanisms can also be employed after a referendum has been called. For example, in the United States State of Oregon, since 2010, a citizens’ panel has been convened in the early stages of the campaign every two years to hear about the issues, deliberate, and prepare a statement about the issues and arguments; this statement is included in the information pack that is sent to all voters. Evidence suggests that voters value this material and trust it more than material coming from campaign organisers.

121. Such practices remain somewhat experimental, and the optimal approach will vary from country to country. But they offer the prospect of deepening citizens’ participation in democratic discussion, improving the quality of democratic deliberation, and addressing voters lack of trust in and feeling of disconnection from decision-making processes. Further trialling of different approaches in different countries should therefore be encouraged. This is in line with the recommendations of the United Kingdom’s Independent Commission on Referendums, which says that “citizens’ assemblies should be piloted during future referendum campaigns”.

3.7. Enforcement of the rules

122. The final aspect of referendum regulation to be considered is the enforcement of the rules. The current Code stipulates two kinds of enforcement mechanism. First, it envisages sanctions for breaches of the rules.

Second, it indicates that it should be possible to annul the result of a referendum – in whole or in part – if there is a danger that rule breaches affected the referendum outcome.

The Code also sets out the procedures for deciding on such matters, which should be conducted by an electoral commission and/or a court.

123. These provisions are appropriate. It might be clarified that sanctioning powers should cover aspects of campaign regulation that are not explicitly mentioned at present, such as breaches of the campaign finance rules. Consideration must also be given to the scale of sanctions. In particular, fines should be commensurate with the scale of campaign funding, such that they are not treated simply as tolerable campaign expenses.

4. Conclusions

124. In conclusion, it has become clear that three kinds of change are needed:

first, the Venice Commission’s Code of Good Practice on Referendums, agreed back in 2007, should be updated. This is needed particularly to take account of changes – since it was written over a decade ago – arising from the growth of the internet and social media and to reflect the importance of ensuring that quality information is available to voters. There are, however, other areas in which amendments would also be desirable. Concrete suggestions, including general principles and their implications for specific aspects of referendum conduct, have been made in the preliminary draft resolution;

second, member States’ compliance with the Code should be enhanced. Constraints on government campaigning are often particularly weak and there are also other issues that should be addressed; the areas where compliance is mostly lacking are also listed in the draft resolution;

third, in areas where legal prescriptions are not appropriate, there is much scope for sharing good practice between countries, particularly in relation to methods for enhancing considered public debate around referendum issues.

125. Among the suggestions to be considered by the Venice Commission when updating the 2007 Code one stands out in particular as one of the most efficient means to enhance compliance by member States with referendum rules: the creation of an independent body which would check any proposed referendum question, supervise the conduct of the campaign, take all necessary measures to ensure that this is properly held and possess the means to enforce its decisions and sanction possible breaches.