The downward trend in support for the voice to parliament has slowed, but a majority of people are still intending on voting no, according to the latest update from Guardian Australia’s poll tracker.
Despite the addition of several recent polls to the Guardian’s model, there is no sign of a trend back towards a yes vote for Saturday’s referendum. The yes side has an estimated 41.6% support nationally and is trailing in every state except Tasmania, which has been excluded from our model due to low sample sizes.
The need for a double majority to pass a referendum – a majority of voters nationally as well as majorities in four of the six states – means that the national yes vote likely needs to be closer to 53% for the yes campaign to have a good chance.
Support is under 40% in Queensland and Western Australia, and is highest at 45% in Victoria. There is a lot of uncertainty around the Tasmania results, where there have been relatively few polls throughout the campaign, and many which had small sample sizes. Dr Kevin Bonham, an election analyst, estimates support is about 45.9% in Tasmania, based on a longer-term average than the measure used by Guardian Australia’s tracker.
The share of undecided voters nationally is between 7% and 14%, depending on pollster methodology, according to Bonham. This is roughly in line with the share of undecided voters at recent elections.
Support for the voice to parliament has declined more than 20 points since August last year. A lot of “soft” yes and no voters – people who weren’t certain of their voting intentions, but were leaning one way – have drifted to the no camp. Some 40% of respondents in recent Essential polls have been “hard no”, about double what it was in February. About 30% are “hard yes”, which has remained roughly constant.
Guardian Australia collated more than 60 polls from 12 different pollsters since August last year. Our tracker does not predict the referendum outcome, but gives us a rough average and trend given differing polling methodologies, sample sizes and questions across pollsters and throughout the campaign.
You can read more about polling and our methodology here.
The downward trend in support for the referendum has slowed in the past couple of months, with a few recent polls as high as 47%, once undecideds are removed. “As the downwards trend has trailed off, you expect to get more polls that show an insignificant upswing just by random chance,” says Bonham.
“In the republic referendum, there was a late upswing [in the polls] that proved to not be real.”
As it stands, if the referendum were to pass it would mean a polling error about three times larger than the 2019 election, which experts say is very unlikely.
The ACT and the Northern Territory are not counted towards the state majority required for a referendum to pass. Tasmania contains only about 2% of the national population, making it hard to get decent sample sizes. All three have been excluded from our model.
The number of undecided voters has declined by about 11 points over the past year, according to the Guardian Australia tracker. But this can be a noisy measure and some pollsters have stopped reporting undecideds/unknowns as the campaign has progressed.
Bonham says the average in polls that ask two questions before marking someone as undecided is actually around 7.5%, and in polls that ask only once it is about 14%.
“In the recent federal election in polls that asked twice, [the undecided vote] was down around 6%” says Bonham. “There are some people who will always say that they are undecided even if you ask them on the day.”
A large share of voters were “soft” yes or no in early polls. Many would have been unfamiliar with the concept, and some details like the exact wording of the question were not revealed until later in the campaign.
But the proportion of undecided voters has come down as the voice has become more prominent in the news and people have seen ads or had conversations with friends, say pollsters.
