The history of referendums in Australia is riddled with failure. Albanese has much at risk – and much to gain

The Australian summer has not been kind to the proposal for a First Nations Voice to Parliament. A recent Newspoll shows a majority in favour – at 56% – but with half of those only partly in favour and 37% opposed. There is plenty of material for opponents to work with.

That they have been assiduously doing. Some calls for “more detail” are probably a genuine desire for just that – more detail. Others are very likely a cover for outright opposition, or for the holding operation currently being conducted by Opposition Leader Peter Dutton.

Read more: Albanese's Newspoll ratings drop but Labor maintains big lead

He knows there is much at stake for him and the opposition in how it responds to this issue. A Liberal Party already seen as out of tune with the increasingly progressive social opinions found among voters in metropolitan seats might yet suffer more damage.

First Nations opinion is said to be overwhelmingly in favour. However, the minority of Aboriginal voices already heard against the proposal – some from the left, others from the right – have ensured the proposal will likely be unable to carry the same amount of moral force as the 1967 referendum.

The “yes” case in that year, which attracted almost 91% of the vote, not only had bipartisan support; it was overwhelmingly understood as the right way to vote if you didn’t want Australia to become another South Africa. You had to vote “yes” if you wanted a decent future for the Aboriginal children whose images featured prominently in the publicity. There was a moral clarity, as well as a sense of mission among supporters, on a question that offered white voters a kind of national redemption.

Supporters of the Voice would like a similar kind of clarity, but it is probably now already out of reach. That means that the “yes” case will need to counter the sowing of fear and doubt, the suspicion of a hidden agenda, that has been the enemy of referendum proposals since the earliest years of the Federation.

The numbers are fairly well known in a society that is usually a bit hazy about the details of its political history. Eight out of 44 – that’s how many referendum proposals have succeeded since 1901. Bipartisanship has come to be seen as essential in getting a proposal accepted, but some bipartisan proposals have also failed.

It is often said the constitutional requirement of a double majority – more than 50% of voters as well as at least four out of six states – makes changing the constitution exceptionally difficult. There is evidence to support this contention.

On five occasions, a proposal achieved a majority of voters but went down because it lacked the requisite majority of states. In 1977, simultaneous elections of the House and Senate achieved over 62% support but fell over because voters in only three states backed it. All the same, proposals have more commonly failed because they gained the support of less than half the national electorate – indeed, sometimes less than a third.

Read more: Changing the Australian Constitution is not easy. But we need to stop thinking it's impossible

Both sides of politics have benefited from negativity. In 1951, a time of great tension in the Cold War, Labor leader Herbert Vere Evatt managed between July 9 and September 22 to help turn an 80% majority in favour of banning the Communist Party into a narrow “no” majority. There was also a 3-3 split between the states. He presented the proposal as the thin edge of the wedge, arguing the Menzies government would use the power arbitrarily against people and institutions who were not communists.

In 1988, Hawke government proposals for mostly modest changes achieved embarrassingly poor support. They included a provision for freedom of religion and extension to the states of the right to trial by jury and compensation for compulsory acquisition of property. The Coalition’s “no” campaign, led by Peter Reith, mobilised opposition with great success.

Only a provision for fair and democratic elections received more than about a third of the vote – and that was just 37.6%. The proposal for simultaneous four-year terms for the two houses of parliament – perhaps the most substantial of the proposals in terms of systemic change – managed less than a third of the vote. Not a single state voted in favour of any proposal. (The Australian Capital Territory voted narrowly in favour of fair elections.)

“Anyone who got up in the next 20 years at a Labor Party conference and suggested another referendum would have a very short future,” concluded Bob Hogg, the ALP national secretary. That said, the party under Hawke’s successor, Paul Keating, did champion a republic in the 1990s. A 1999 referendum held after it lost office produced another disappointing result for constitutional reformers.

A “no” vote would be a great blow to the Albanese government. And in light of this history, it is in some respects odd that it has chosen to hitch its wagon so tightly to the cause of constitutional change. It is true they do not usually destroy governments – Menzies survived the 1951 defeat to rule for another decade and a half, and after the 1988 debacle, Labor would win elections in 1990 and 1993. But they do have the potential to undermine leaders and governments. Some commentary, for instance, suggested after the 1951 result that Menzies might leave politics.

The Voice referendum is not lost, but there is sufficient confusion and contention around the proposal to worry supporters. To suggest a good deal of that confusion has been cynically manufactured for political advantage does not much alter the level of danger.

Anthony Albanese’s recent speech at the National Press Club indicates he is seeking to strike a delicate balance. He wants to facilitate debate and consultation – especially with First Nations peoples – without leaving a vacuum to be filled by opponents with misinformation.

That task is likely to test fully the political skills the prime minister has developed over decades in the business.