How to design a second referendum ballot

If a second EU referendum becomes a serious possibility for the UK, the design of the ballot paper will become a battleground in its own right. This article is not in itself a plea for or against a second referendum. It merely seeks to evaluate some available options for the form the ballot might take.

Many countries, most notably Switzerland, have considerable practice of phrasing referendum questions. Expertise is also found in the field of public international law, which deals with cases where the secession of territories has been made conditional on a popular vote. Drawing on this experience, good practices can be identified. The main objective is to avoid manipulative bias.

The decisive weight is usually assumed by a simple majority: 50 per cent of those who voted. Supermajority requirements give a veto to a minority of voters. In the absence of good reasons why a minority should be privileged, a simple majority is both necessary and sufficient.

The question is best put in a binary manner, to which you can either say Yes or No. There is an established practice that a Yes vote is support for change. A No vote helps to maintain the status quo. The 2014 Scotland referendum phrased the question in this way: “Should Scotland be an independent country?”

Neither of the UK’s European referendums have strictly adhered to these practices. The question in 1975 was “Do you think that the United Kingdom should stay in the European Community (the Common Market)?” with Yes being a vote for the status quo (albeit only for 17 months). The 2016 question was “Should the United Kingdom remain a member of the European Union or leave the European Union?” with the answers “Remain a member of the European Union” or “Leave the European Union”; the Electoral Commission deemed that this wording was the “most neutral” of those it tested.

Nevertheless, it is best to consider the wording of another question in the light of best and established practice. If another Brexit referendum were to be initiated today, the status quo would be determined by the fact that the membership of the UK in the European Union ends on 29th March 2019. But since the voters are evaluating change, it is undesirable for the status quo to fluctuate substantially between the time the referendum question is decided and the date of the poll.

The question could, for example, be: “Should the notice to leave the EU be revoked?” But at the time of any referendum it may make a significant difference whether a withdrawal agreement has been properly ratified, with all necessary legislation passed by parliament. In times of volatile change, a referendum is highly impractical, if not impossible. At this stage in the process, any second referendum would require extending the two-year Article 50 notice period, at least for the duration of the referendum process.

Up to a point, a referendum can include multiple questions. But it is better to be economical with the number of options. If multiple options are included, the first-past-the-post system, which lets a relative majority or plurality decide the outcome, is untenable. It would lead to arbitrary results, which might depend purely on, for example, the number of options on the ballot paper. The country should not be bound to follow a course only supported by 20-30 per cent of voters.

It is for this reason that a multi-option ballot with a first-past-the post-mechanism should not include an option to Remain in the EU, since this would allow for the reversal of the 2016 referendum by a minority of voters. The only option, then, would be to ask voters what kind of Brexit they prefer.

For a multi-option ballot to be feasible, the design must ensure that the end result is supported by the simple majority. This could be done by employing a form of the alternative vote. The ballot paper would require all voters to rank their first, second, third (and so on) options. The least popular option would then be eliminated and the second choices of its supporters added to the remaining options, and so forth, until one option reaches the majority.

However, it is doubtful whether a decision thus produced would come across as a clear and straightforward representation of the voters’ preferences. Moreover, do we really expect every voter to have a clearly ranked preference between Norway, Customs-Union, Common Market plus Customs-Union, Canada+ and whatever other options were to go onto such a ballot sheet?

A more appropriate mechanism would be to ask two questions. Each question has as its alternative today’s status quo. The questions can go in different directions. Question 1 could ask: “Do you want the UK to be bound by the draft Withdrawal Agreement?” Question 2 could ask: “Do you want the UK to revoke the notice under Article 50?” The No to each question supports the status quo right now.

If there is a majority for one of the questions but not the other, this settles the matter. The matter is also settled if neither of the questions finds a positive majority: the status quo prevails. However, in the event of a majority voting Yes for both questions, a third, decisive question needs to be on the ballot paper, namely: “Assuming there is a majority of Yes for both questions, which option do you prefer?”

Broadening the range of options to more than two is feasible, but then additional decisive questions would be needed to establish the voters’ preference for all possible outcomes in which more than one question commands a majority of Yes.

All of which goes to show that selecting and phrasing appropriate questions can be highly controversial issues, diverging due to political positions and in expectations of voters’ preferences. There may indeed be growing support for a second referendum. It may be much more difficult to agree upon the specific design of a ballot paper.