The ABC poll now puts Yes at an average of 41.2 percent, well behind No at 58.8 percent.
For Yes, that’s well below the level of support recorded in the 1999 Republican referendum, which failed with just over 45 percent of the national vote.
“I’m old enough to remember that at the time it didn’t seem like a week before the case for change was particularly good,” says Professor Simon Jackman of the University of Sydney.
“This is also the case with this referendum.”
The model, developed by Jackman, is an effort to understand published polls and is not a prediction of the outcome of the vote.
The decline has been steady for months, but some recent polls have slowed the descent and left us a little less sure of the true position of the average.
In particular, Essential saw a slight shift, albeit within the margin of error, in the direction of Yes.
This is mainly due to a rise in Queensland and may well prove to be an outlier.
And although the trend is easing, it has certainly not reversed.
Jackman says this is moving in a clear direction: “At this point, all the polls are pointing to a No victory,” he says.
“It may be that there are three points of error (or) five points of error. But I really doubt that the average of the polls does not reflect the true situation of the eight points that we should see (the Yes wins ).”
“It seems incredibly, incredibly unlikely that this is the case, and incredibly, incredibly unlikely that there will be an instant turnaround.”
Greater diffusion of results than elections
Recent polls have shown a relatively wide spread in the numbers. For example, a RedBridge poll released earlier this month put the Yes vote at 38 percent, while this week’s Essential put it at 47 percent, after excluding undecided voters.
The spread of results is broader than we usually see in the run-up to general elections, and Jackman says it’s broader than one might expect due to the sampling of voters participating in the polls .
This will undoubtedly be of interest to pollsters and researchers once the referendum result is known, but that does not mean it will give much hope to the Yes camp.
“The polls are a little scattered, but they’re scattered in an area that suggests the no vote is on its way to a victory.”
Emeritus Professor Murray Goot, a political scientist at Macquarie University, agrees.
“Once it got down to about 45 or 46 Yes and it kept falling, in my opinion, no amount of campaigning on the ground or on social media could save the Yes vote,” he says.
The power of the vote is poorly understood
Goot claims that early polls suggesting that up to two-thirds of people supported the Voice were misinterpreted by many who thought they were showing strong support, rather than showing widespread but limited support for the Voice.
“The strength of the vote is something completely different,” he says.
“Once you start asking people about the strength of their vote… you get a completely different answer.
“What it showed was that most people were either undecided or not very committed to the Yes or No side.”
Over time, as the debate evolved and the referendum approached, people hardened their positions.
For the Yes vote to prevail from there, we must change the most firmly established mentalities.
To be fair, some polls suggest that a significant number of people are undecided.
But with nearly 2.5 million people voting Friday, demand is growing.
Goot says he’s not surprised to see this trend and that “it’s not unusual for polls to fail in a referendum.”
“I have looked at all the polls from the 44th referendum to the 77th referendum… generally, but not invariably, there has been a decline in support over the campaign period and, in some cases, a very substantial decline. “
He says the most dramatic fall in support was seen in the 1951 referendum, in which the Menzies government sought the power to legislate against communism and allow it to ban the Australian Communist Party.
“There was only one polling organization, the Morgan Research Centre, (and) it fell somewhere in the 80s to just under 50 votes.”
Cities and regions differ
We only track the national voting intention average, primarily because states and regions are surveyed less often and with smaller samples.
The general and consistent polling trend indicates that support for Yes is strongest in city centers and declining in outer suburbs, regional towns and remote areas.
“I suspect that on Saturday night, Sunday and in the days following the referendum, we will remember one of the most enduring aspects of Australian politics and public opinion,” Professor Jackman said.
“In city centers, income and access to higher education are accompanied by a form of cosmopolitanism (and) social progressivism.
“It was true for the plebiscite on same-sex marriage, it was true for the question of the republic.”
Of course, this is a general trend, and it won’t always be true. Cooktown is likely to vote differently to Cape York, results in Walgett will not reflect those in Whyalla and East Hills may not look anything like Elizabeth.
We can safely assume that campaigns have surveyed smaller areas to understand who is persuadable and where to direct resources.
But this type of information is kept secret and most of it never comes into the public domain.
To see these trends for ourselves, we’ll just have to wait another week to see how the votes are cast.
If you are unable to load the form, you can access it here.