The last constitutional referendums in this country were held in 1999, on adding a preamble to the constitution and changing Australia to be a republic. The latter referendum may have some insights about what we could expect on Saturday, so I thought I would delve a bit into the trends in those results.
I will mostly be ignoring the preamble referendum, and looking at the republic.
While every referendum is different, I think there’s a lot of reasons to think that the coalition supporting Yes will have similar trends to the Yes coalition in 1999 – inner urban voters, leaning to the left but also including seats that had traditionally favoured the Liberal Party.
At the end of this post, I’ve done a rough calculation of how each 2022 electorate would have voted in the 1999 referendum (made much easier because most votes were cast on referendum day in 1999).
42 out of 148 seats voted Yes in 1999. Of those, 25 were Labor seats and 17 were Coalition seats – all Liberal seats.
Of those seventeen Liberal seats, less than a third of them are still held by the Liberal Party. Six are now held by teals – of the seven urban teal seats, only Mackellar voted No in 1999. Five are now held by Labor, and one is held by the Greens (Ryan).
Those five Liberal-held seats that voted Yes in 1999 are Berowra, Bradfield, Deakin, Menzies and Sturt.
If you use my rough estimates of 1999 vote by 2022 seats, there are now 46 Yes seats out of 151. Those 46 include 35 that gave Labor a 2PP majority in 2022, and just eleven that produced a Coalition 2PP majority. Those eleven seats include those same six teal seats and the same five Liberal-voting seats. This may partly reflect that Labor had a better election in 2022 than 1998, but later on we’ll discover that it’s more complex than that.
This next chart compares the Labor 2PP in 1998 to the Yes vote in 1999.
The horizontal and vertical lines delineate 50% in each of those metrics, so they make four quadrants.
There is definitely a correlation between Yes vote and Labor 2PP, but it’s not a very strong one. But in this next chart, you can see that the 1999 Republic Yes vote lines up much more strongly with the Labor 2PP in 2022.
You can boil this down to a simple metric. The correlation between the Yes vote and the 1998 Labor 2PP was just 0.396, but the 2022 correlation was 0.613. Bear in mind that a correlation of 0 means the two numbers have no relationship, while a correlation of 1 means they are perfectly correlated.
In simple terms, this means that the way that people vote in 2022 lines up much closer to the 1999 Republic referendum than how people voted in 1998. And I believe we will see this flow through on Saturday: Labor, teal and Greens seats will be much more likely to vote Yes. Although with polling suggesting a baseline support for Yes of around 40%, plenty of Labor seats will likely vote no.
You can also look at the divide by regional status. The AEC classifies electorates into four geographic categories: inner metro, outer metro, provincial and rural.
The voting trends were gradual, with Yes winning a majority in inner metropolitan seats, and losing by more as you head further out of the city. Yes actually won a slim 50.6% majority in metropolitan Australia, while No won 63% in rural and provincial Australia.
The seat results were more decisive, with about two thirds of inner metro seats voting Yes. There were another 14 outer metro seats that voted yes, and just one outside the big cities (Newcastle).
I thought I would also examine how much the vote related to the size of the booth.
Back in 1999, nearly all votes were cast on the day. Antony Green’s recent post points out that almost 91% of the vote was cast on the day.
Smaller booths are usually counted first, so the bias towards the No vote in the smaller booths would produce an early trend that would favour No more than the final result.
Small booths with less than 500 votes strongly favoured No. Those with 500-1500 votes were also slightly favoured, but beyond that the trend was very subtle.
If the voting trends are similar in 2023 as in 1999, you’d expect the first-counted booths to be biased in favour of No. That’s before we consider the impact of so many more votes being cast early, but let’s put that aside for now.
Finally, this map shows the 2022 electorates by how each area voted in 1999. This is not as accurate as a typical redistribution calculation, with over two decades having passed. But mostly it seems to hold up.
Generally the Yes vote was highest in seats where the Greens are now strongest – the three Greens-held seats in Brisbane, Sydney, Grayndler, Canberra, Melbourne and Macnamara. It seems quite plausible we’ll see a similar trend on Saturday.
Ben,
great pick up to see the correlation with 2022 – that’s a fascination graph.
You concludes with a “general” statement, but don’t mention the exceptions – where the Green’s aren’t now strong yet the 1999 Yes vote was high. i.e. Bradfield and North Sydney. They have though dramatically moved on a 2PP basis towards Labor since 1998.
Great Analysis Ben,
One demographic trend that was observed at the 1999 referendum which was observed was much stronger stronger support for the Republic in CALD areas even if they were lower socio-economically than other more Anglo-Celtic seats for example Fowler voted Yes while Mackellar, Cook & Hughes voted No. Same with Greater Melbourne with Yes votes in Western Melbourne but no votes in Casey, Dunkley and Flinders. There is been reports with pre-polling exit polling around Redlands LGA that is very dire for the YES side but in Cragieburn (Outer Metro growth area) which younger and more ethnically diverse area it seems promising for YES. So i will be analyzing if there is correlation between ethnicity and Yes for the Voice referendum not just Education & Income. One demographic quirk i dont believe we will see but was observed in 1999 was the Catholic/Protestant divide.
@Nimalan, definitely that is a factor but what I would say is that nowadays most electorates are a lot more ethnically diverse than they were in 1999. That being said, I don’t think the Voice would be as popular among ethnic minorities as the republic since the No campaign has been putting out a pretty significant campaign on why Indigenous have a voice but not other ethnic minorities. It is interesting to see seats like Menzies or Fowler have a very strong republic vote while I don’t think said seats would have a strong Yes vote for the voice.
@Dan M
Yeah i agree with you i am not confident enough to predict that Menzies, Fowler will vote Yes. Maybe we need to compare diverse areas and Anglo areas that are similar socio-economically to see if there is a difference such the Hills District (more diverse) with the Shire (more Anglo) or compare Doncaster with Mentone/Aspendale etc. This is the opposite of SSM vote where Anglo areas were generally more supportive. In terms of changes in ethnic diversity since 1999 i would say some of the dramatic changes have actually occurred in growth corridors such as Casey/Cardinia LGAs in Vic Outer North West Sydney (parts of Greenway/Mitchell) , much of Brisbane South of the River while a lot of Bayside Melbourne, Northern Beaches, Redlands LGA probably havnt changed much since the 1999
I have seen polling (https://docs.google.com/spreadsheets/d/11ODTJjPtOL58EkRWLsjwNYHrqrtd0GXk/edit#gid=2112880336) that indicates only 22 seats (of 151) are projected to return a yes result.
All teal seats except North Sydney seems to be projected to vote no, but North Sydney still has a YES margin of 6.6% or something. I knew Warringah and Mackellar were going to give NOs but I am surprised to learn that even Wentworth is a NO.
Given Bennelong, Reid and Chisholm are also holding up, and Kooyong is only just a marginal loss, I suspect this is the Chinese immigrant vote.
(While North Sydney did vote 60% yes on the republic, this was before the influx of chinese people into Chatswood)
On a separate note, Greenway is still projected to say yes though. Dunno why given neighbouring Parramatta (Samaras singled this one out as a perfect demographic for YES votes) and Chifley (reasonably high First Nations population) are both NOs on this.
@ Leon
This is interesting, Mackellar which voted No the Republic i suspected will have the strongest No vote among all the Teal seats even more than Bradfield it has fewer professional more wealthy tradies etc. Interestingly Calwell which had the biggest Anti-Labor swing is projected to have a similar voice result to Mitchell/Menzies both much wealthier seats at 57% No and is even better than Mackellar at 62% No. With respect to Wentworth i am interested to know if the Jewish vote will be more Pro-Yes for example would Dover Heights, Rose Bay & Bellevue Hill vote more Yes than Mosman, Clontarf/Seaforth etc despite being socio-economically similar.
@Leon I would be interested to see how the methodology of the poll was conducted and how sampling was obtained. I read somewhere it was similar to how a seat by seat poll was obtained in the federal election which I believe significantly underestimated the swing towards Labor in Victoria in the seats with large Chinese populations such as Menzies, Aston and Chisholm, underestimating the swing to the Libs in Tasmania, while also underestimating the swing towards the teals in NSW.
Ben
What seems striking from the latest Resolve poll (I think) is the division by age – under 34 are strongly yes, 35-55 evenly divided and over 55 – strongly no
There is some correlation between Labor 2PP and support for a republic/voice. Labor holds more seats in metropolitan areas and this is where there is a greater intention to vote Yes. It’s clear that there further you go out from a capital city CBD, the more likely you’ll find No voters.
A better correlation or predictor of Yes at the Voice referendum would be Green/teal vote for each electorate. On demographics, university degree attainment and median age would be better predictors.
@Nimalan, The Republic referendum is different. I think non-Anglo-Celtic voters and younger voters had less affinity to monarchical traditions and Mother England and therefore were more pro-Republic.
Since 1999, there has been increasing urbanisation and the growth sea/tree changers as well gentrification of blue-collar cities like Newcastle. I wouldn’t be surprised if Eden-Monaro, Richmond and Newcastle are more Yes-voting than they were in 1999 (or at least more Yes than the national average).
Here in Melbourne it will be MUCH closer than the polls. I don’t believe Yes is winning 60-40 here!! No has a shot here in the CBD, believe me. I haven’t seen really much Yes material since moving down here and I have seen many No posters around here. I think migrants are actually more likely to vote ”No” because they see this as referendum as racial division, and they are opposed to that.
I think Yes will be lucky to win in 5 seats nationwide. Maybe Clark, Sydney and a couple of others. It will be lineball here in Melbourne, I see it 50-50 right now. Lets see how right or wrong I am on the night.
@ Votante
Agree with you i think Richmond, Eden Monaro, Newcastle and even Corangamite are much more progressive than they were in 1999. That being said i expect there will be a wide variation within Richmond with a very strong Yes within Byron Shire while Tweed/Ballina Shire will vote No so the Delta will be much greater than in 1999. Same with Macquarie with the Blue Mountains much more progressive than in 1999 but no real demographic change in Hawkesbury Region so i expect to see polarization within the electorate. One other change, i believe in Outer Suburban growth areas which today often have a better Educated demographic For example, there are reports that Greenway is looking good for Yes. I also expect that Blair will be better for Yes than it was in 1999 and will be closer to the Queensland Average and not among the Top 5 no voting electorates in the Nation like it was 1999 due to demographic change around Spring Mountain, Springfield Lakes with a better educated and younger demographic moving into the new developments.
Having spoken to a few people who have been handing out both yes and no materials across pre-polling booths in Melbourne, Sydney & Hobart the consistent feedback is that no is doing better than expected.
I know that this is anecdotal evidence but when seats such as Clarke, Warringah, Higgins, Toorak, North Sydney are doing better for no than expected it is difficult to see how yes will prevail.
In my mind the question is whether or not there will there be a yes vote in any State. My feeling is that there will not be a yes vote in any State.
Another observations:
I was at a corporate lunch with about 20 people a couple of weeks ago. There was a presentation on why the yes case was important and should be supported. After the presentation there was general discussion about the vote. The organizer and I later had a chat and she was confidant that everyone in the room was voting yes. I suggested that my read on the room was 12-14 people would vote no on the basis that about 8 people had said nothing and a number of others used the phrases along the lines of “my friends are all voting no what should I say to them……”. Which I took to mean “I am voting no unless you give me a good reason to vote yes and I still haven’t heard it”. And the 8 who said nothing weren’t going to piss off their boss by saying that they were voting no. I was told that I was absolutely wrong and how could I think that anyone at the lunch would vote no…. we shall see.
Best
Pollster
I do agree that the pre poll results will be strongly No – even in a place like North Sydney the pre poll vote is quiet conservative. Young people and service professionals aren’t pre polling. The feedback being attained so far is only from pre poll. I predict there might be a glimmer of hope for Yes when some big urban election day booths come in, but then the pre poll will darken the skies.
Even in North Sydney on the pollling booths i expect to see variation for example a much stronger Yes Vote in the Higher Density areas along the train-line with a younger demographic while Longueville, Northwood and especially Hunters Hill will be better for the No camp.
Pollster
I can certainly concur about a sort of…well, I don’t want to use the word ‘arrogance’, but that is pretty much what it is, within elements of the Yes campaign. Some members of my family are quite fervent Yes supporters, but get quite irritated when I point out the signs and historical precedents which indicate the Voice will fail. They usually say very scientific things like “you just don’t know” and “it could be different this time”. I don’t wish the impending loss on them though. This will be a war lost not by the soldiery, but by the generals.
Daniel T, Melbourne CBD although is largely migrant heavy, most don’t have citizenships as they tend to be international students, newer immigrants, and expatriates this is why the federal electorate of Melbourne has an 83% Yes Support for Same-Sex Marriage in 2017.
So the three federal seats in Brisbane that are currently held by the Greens were the only seats in Queensland that voted Yes to the Republic in 1999? Interesting.
The yes vote was overwhelming in the federal seat of Melbourne today on pre poll. I would go 90-10 yes from when I was down there. Not much support for no and a bit hostile to them today
The number of volunteers or signs is not a true reflection of what the actual result will be. I drove past a pre-poll booth in a Labor electorate and it was plastered all over with Yes signs but I doubt it will be a landslide Yes vote in that electorate. I still think inner-city and CBD seats like Melbourne will have the highest Yes votes.
Voted today in Melbourne North, I don’t see the enthusiasm for ”Yes” at all. Although this polling booth caters several electorates and I was given the ”Melbourne” card as I live in this electorate.
There’s no way 90% voted Yes today. I saw several people in line taking no pamphlets.
Which booth were you at Tno? I was at the one on Victoria street
@Nimalan
I am in North Sydney.
I would agree all the high density areas incl Chatswood would vote yes.
I reckon Wollstonecraft and Waverton, whilst low-density and harbourside would vote yes as well given the demographics and past 2pp results – in fact all of North Sydney LGA probably will.
What would your read be on Lane Cove Proper and Willoughby booths? The latter is my home booth and I only see yes signs on homes (and I would consider it one of the older/whiter/conservative areas of the electorate, though by no means extremely so)
Pollster your comment that
“I was told that I was absolutely wrong and how could I think that anyone at the lunch would vote no…. ”
is fascinating.In my view it sums up why the yes case will lose-the proponents simply don’t understand that there are perfectly good arguments for voting no.
@ Leon
Agree with you i feel that all of North Sydney LGA will vote yes including Kirribilli etc. I also think Lane Cover Proper and Willoughby (suburb) will vote Yes but more narrowly. i think the Strongest Yes vote will come out of Chatswood, St Leonards and Artamon which are more Labor friendly and St Leonards has a lot of unionized healthcare workers due to proximity to the RNS hospital. On the eastern fringe Castle Cove, Castlecrag and Northbridge will be leaning towards No while in the seat as a whole it will likely be St Ignatius College, Riverview or Hunters Hill that will be competing for the stongest no voting booth.
With respect to Melbourne, while i dont think it is 90% Yes i still think it will be in the Top 3 in the nation to Vote Yes and with the possible exception of the East Melbourne booth i expect all to vote Yes strongly.
Agree Melbourne won’t probably be 90% Yes given SSM was 83% although the boundaries were slightly different back then. I believe it would be somewhere between 70% – 80% (I found the 64% in the poll a bit too low)
@ Marh
Good point about the demographics of the CBD i have noticed Docklands is now increasingly resembling that with a large number of international students previously it was a lot of Double Income No Kids. There seems to have a big demographic shift in the last 15 years there.
@Sebena – I thought it was interesting which is why I posted. If I am correct in my read of the room there are a lot of the “shy tories” floating around. The question is then does this also flow into the opinion polls? If so, 40% national vote for yes would be pretty optimistic.
Best
Pollster
To my knowledge, Voice referendum seems very similar to California Prop 16 (2020) where it seems even voters in progressive places vote more conservative on referendums such as race relations
https://en.m.wikipedia.org/wiki/2020_California_Proposition_16
Electorate with the highest Yes vote in 1999 was Melbourne. Newtown North (in Grayndler) had the highest Yes vote of any booth with 83.25%. Next highest booths were Carlton Central, Fairfield North, Fitzroy North and Carlton North – all in Melbourne except for Fairfield North (Batman/Cooper). Fairfield North, while not surprising to see it voted strongly Yes, it’s surprising to see it was the third highest in the country. I will be voting at that booth tomorrow.
I suspect Melbourne will again have the highest Yes vote in this referendum.
@ Adam
Agree with your analysis. i suspect Fairfield north will not be in the Top 20 booths in the nation for Yes vote today. I this Fitzroy (Sacred Heart PS) will be the highest Yes and there has been further gentrification in Collingwood, Fitzroy, North Melbourne etc since 1999 with more density. In Batman/Cooper i suspect the Westgarth booth will be highest for Yes. Fairfield-Alphington while quite affluent has not seen as much densification, it remains a quiet village atmosphere so i suspect even the Bell PS booth in Preston maybe higher now for Yes. By the way do you have the link for 1999 results by booth i would like to do some comparisons on a booth levels to look at demographic changes since 1999.
@Nimalan Ben has posted a spreadsheet with the results by booth in the Data section of this website.
Thanks Adam,
Interestingly, the Carlton Central booth was the Highest Yes vote at 93% so exceeded the Republican vote in 1999. However, Fairfield north was not among the highest in the Nation the booths around Northcote came close to 90% Yes while Bell PS which i mentioned prior to the results was 82% Yes higher than Fairfield North. This shows how much more progressive the Inner City has become compared to the country as whole.
Comments are closed.