Could the wording of the question win the Voice to Parliament referendum? Or is there a chance its length may lead to defeat?

The Albanese Government has finally released the wording of the referendum question on the proposed Aboriginal and Torres Strait Islander Voice to Parliament.

The wording is: “A proposed law: To alter the Constitution to recognise the First Peoples of Australia by establishing an Aboriginal and Torres Strait Islander Voice. Do you approve this proposed alteration?”

So what’s in a word, or 29 or them to be exact? Well, quite a lot when it comes to referendums.

Research undertaken by yours truly and colleagues has found that the wording of referendum questions can influence the outcome of the vote. We will get to that shortly.

But first, let’s look at some other ballot questions from votes across the world.

The question posed to voters in the United Kingdom in 2016 was 16 words long. Photo: Claudio Divizia/

Referendums come in many shapes and sizes, from the bizarrely bland to the blatantly biased.

For example, in Northern Ireland in 1998, voters were asked to approve (or otherwise) the rather neutral question, “Do you support the agreement reached in multi-party talks on Northern Ireland and set out in Command Paper 3883?” In all, 71.2 per cent of voters did.

Command Paper 3883 was a coded reference to the official document containing the so-called Belfast Agreement on a peace deal in the then terrorist-ridden British province. Despite the lack of references to the policy contained in the Command Paper, voters were seemingly aware of the issues and knew perfectly well what they were voting for.

A government referendum poster in Belfast in 1998. Photo: CAIN

In other referendums, voters have been asked more leading questions. In a bizarre example from Chile in 1978, the junta asked voters whether they supported the following statement: “In the face of international aggression against the government of our fatherland, I support President Pinochet in his defence of Chile’s dignity, and I once again confirm the legitimacy of the government of the republic in its leadership of the institutional proceedings in this country.”

Not surprisingly, almost 100 per cent of voters endorsed the proposition. But then again, Chile under Pinochet was a dictatorship.

So, let’s look at the wording of questions in constitutional referendums. What does the research show? Several things, in fact.

When the French-speaking province of Quebec held a referendum on independence in 1995, 62 per cent of the support for sovereignty was down to the wording of the question.

But there’s also research that shows this relationship holds more generally across democratic countries.

When looking at constitutional referendums, the probability that a referendum will pass increases if it has emotive words like ‘agree’ or ‘approve’ in the question.

A poster titled 'How to Vote Yes' which displays the ballot papers for the 1999 referendum with the word 'Yes' written into the voting box.

Ballot papers at a polling booth for the 1999 referendum. Photo: Loui Seselja/National Library of Australia, nla.obj-145819844

Colleagues and I have looked at all 150 constitutional referendums held since 1970, and our preliminary findings suggest that inclusion of such words can improve the chances of success by up to four per cent.

It seems that emotive words make voters feel more positive and less threatened.

If, as some polls indicate, the Voice referendum will be a tight race, a two to four per cent lead could make all the difference.

So far, so good, for those cheering for the Voice.

But there is a snag to this research; a reason for supporters to be cautious.

My earlier research about referendums on ethnic and national issues, such as the Voice, shows the longer a question on the ballot, the greater the risk (or chance) of a ‘no’ vote.  

It might be that a longer question makes voters sceptical. If in doubt, vote no, runs an old adage. One might speculate that this is due to confusion caused by longer, and hence less intelligible, questions.

I discovered that for every extra word in the question there is a tenth of a per cent risk (or chance) of a no vote. So, if we do the maths, 0.1 x the number of words in the proposed question means we get 2.9 per cent.

Effectively, the long question published by Prime Minister Anthony Albanese cancels out most of the benefit of its emotive words.

So, there is all to play for.

Words make a difference, but neither side can be confident of winning the Voice referendum based on the linguistic equation of this question alone. And that’s rather assuring.

You may also like

Article Card Image

Democracy Sausage: the world of the pre-citizen

Legal and youth justice expert Faith Gordon joins Mark Kenny to discuss young people, social media and democracy.

Article Card Image

Democracy Sausage: the state and democracy

This week we are returning to the building blocks of politics and democracy, as philosopher Philip Pettit joins Mark Kenny to question assumptions.

Article Card Image

Prime Minister Anthony Albanese is determined to advance reconciliation

The Prime Minister reflected on the outcome of the Voice referendum on an episode of ANU podcast Democracy Sausage.

Subscribe to ANU Reporter