The Indigenous vote in rural electorates


There has been some confusion and even some misleading analysis looking at how the Yes vote performed in seats with large Indigenous populations.

In particular, the focus has been on the seats of Lingiari, Leichhardt, Parkes, Kennedy and Durack, all of which voted No. The No vote was very strong in four of these seats, but a lot closer in Lingiari, where No only polled about 55%.

When you compare election results to demographics, there is a serious danger of falling afoul of the ecological fallacy – assuming population-level trends apply to the individuals within these electorates.

This is particularly dangerous when looking at electorates based on their share of Indigenous voters. There is no federal seat in Australia where Indigenous voters make up a majority of the population. The 2021 Census found that 40.3% of people in Lingiari were Aboriginal and/or Torres Strait Islander, and 51.6% are not, with the remaining 8.1% not answering the question. That is the closest we get. We don’t have “majority minority” electorates as you find in the United States.

So in all of these seats, this analysis is not looking at how Indigenous voters have voted – you’re looking at how electorates with a lot of Indigenous voters have voted. And the non-Indigenous voters play a major role.

When you dive down to the booth level, you are able to see clearly that places with large Indigenous populations within this electorate did vote emphatically for the Voice.

I think there were two main claims in the referendum campaign about the voting intentions of Indigenous people. The first was about how the total Indigenous population around Australia was planning to vote.

Early in 2023, Ipsos and YouGov polls found 80-83% of Indigenous Australians were supportive of the Voice proposal. Later in the campaign, a Resolve poll in late September found Indigenous Australia splitting 59-41. This is a difficult question to poll, but it makes sense that, as the Voice proposal lost support generally around Australia, it also had an impact on support amongst Indigenous people. We can’t divine the truth of this question from booth results, since many Indigenous people live in big cities and towns where their votes are mixed with others.

The second question was about particular Indigenous people – those that live in small towns and remote communities. The No campaign made accusations that the Yes campaign was driven by more privileged urban people and was out of touch with the interests of Indigenous people and that remote communities didn’t support the change.

This accusation never made much sense, and the results show it to be completely untrue.

It is easiest to see the truth by looking at the seats of Lingiari and Leichhardt. A large proportion of votes in Lingiari are cast via remote polling teams, where the AEC schedules a time and date for voting in a small town, drives or flies in, collects the votes, and then moves on. At the 2022 federal election, these teams made up 22% of the total vote in Lingiari.

These remote booths are strongly dominated by Indigenous voters. Outside of these remote booths, almost all of the remaining booths are in the urban centres – Alice Springs, Katherine and the urban fringe of Darwin. While the remote booths usually vote strongly for Labor, those other areas are more marginal or lean towards the Country Liberal Party.

The seat of Leichhardt is dominated by the city of Cairns, which has made up over 80% of the ordinary votes. It also covers the rest of Cape York, including the Torres Strait. You can identify booths in the Torres Strait, as well as a number of Aboriginal communities in Cape York.

The remote booths in Lingiari voted 73% for Yes. The outback pre-poll booths, which draw voters from all over, were 49.4% Yes. Alice Springs (which has a substantial Indigenous population mixes with non-Indigenous people) was about 44%, while the Darwin surrounds were more like 22%.

In Leichhardt, those Indigenous communities are not as populous as in Lingiari, but they also voted clearly for Yes. 64% of voters in the remote booths voted Yes, as did 67% in the ordinary booths in the Torres Strait and 61.5% in the Cape York Aboriginal communities with a standard booth. In contrast, the other Cape York booths were only 34.3% Yes and Cairns was 33.7% Yes. This explains why these seats voted No overall despite a strong Yes vote in Indigenous communities.

It is harder to isolate a large vote share that is Indigenous in the other three seats but you can see it at the booth level. In Kennedy, the Yes vote was 78% on Mornington Island, 76% in Yarrabah and 67% in Doomadgee. The remote booths in Durack were 71.2% Yes, and the Kimberley region overall gave Yes a slight majority. Roebourne, Bluff Point, Cable Beach and Fitzroy Crossing all polled over 60%.

You can use the ecological fallacy to explain why it’s not useful to compare the total Indigenous population to the Yes vote, but I think it might explain something more interesting, if not a happy explanation.

There is a phenomenon in political science called “low-contact intergroup threat perception”, where groups that do not interact but live next each other are more likely to be in conflict. Thanks to Luke Mansillo for referring me to some research on this topic that I am yet to read fully.

In the 1967 referendum, there was also evidence that seats with large Indigenous populations had a lower Yes vote than in other seats, although they still recorded a Yes majority. While Yes polled 90.77% nationally, it managed just 71% in Kalgoorlie and 77% in Kennedy. I think it is logical that non-Indigenous voters who live in seats closer to large Indigenous populations were less willing to vote Yes. I’ll leave it to you to interpret what that could mean.

Liked it? Take a second to support the Tally Room on Patreon!
Become a patron at Patreon!


  1. It also looks quite high for the APY Lands in South Australia. Grey is at 21.63% YES but its “Remote Mobile Team 1” is at 77.30%.

  2. You may need to consider what JCP has said about the tactics of some AEC staffers and accompanying activists, that go out into the remote locations.

    Plus, the mentioned Ipsos poll suggesting an 80% pro Voice among aboriginal Australians is on rather thin ice. Not only was the panel size tiny (300 total or 181 weighted), but it was rather limited in that the panel was selected from persons who previously agreed to take part in polls. That’s hardly the disadvantaged mob living on the fringes of our outback towns, or the traditional way far out on the land.

    It’s more likely the self-identifying ‘aborigines’ living rather comfortably within the Tesla belt of our capital cities.

    Details here:

  3. It looks like if the NT had 3 seats, presuming the third seat was a Darwin surrounds and Palmerston seat, Lingiari would likely have voted Yes.

  4. @Peter Sherman –

    What exactly should we consider about what JCP has said? What evidence has she, or anyone else, provided for her expressed concerns (or rather, insinuations)? In the absence of such evidence what conclusions do you suggest we draw, even tentatively?

  5. JCP made statements reflecting on the professionalism of the AEC – I haven’t yet heard her offer any evidence. Unless she produces something of substance she should publicly apologise. The evidence across remote area polling and indigenous booths in the electorates of Leichhardt, Kennedy, Durack and Lingiari is consistently strongly supportive of Yes.

  6. Matt, Luke mentioned it in the context of V.O. Key in Southern Politics (1949) finding that the vote for the Dixiecrats in the south remained highly correlated with the black population. The references are pretty obscure.

  7. One idea I’ve seen is the idea that white people living in communities where they are in contact, but not regular contact, with indigenous people are more likely to vote no than those with no contact. That is voters who think of “indigenous people” as “those people over there” (with higher poverty and crime rates) rather than thinking about abstract ideas of decolonisation are more likely to vote no.

    Do statistics bear this out? Does No start low (affluent inner city communities), reach a peak at a certain % indigenous (country towns) then go down again as communities become characterised by an indigenous vote.

  8. One polling booth located in an area which, according to the 2021 census, has 91% of the local population identifying as indigenous and voted no is Woorabinda in the seat of Flynn.

    Woorabinda Yes: 116 No: 118

    It would be interesting to know what local factors were in play at Woorabinda.

    I was disappointed to hear JCP make a comment about the AEC without backing it up. Although the fact that the vote in the remote polling booths was not closer to 100% perhaps needs investigation. If I identified as part of a race that was being offered a constitutional voice to the exclusion of all others I can’t imagine anyone voting no in such circumstances…..

    Separately, extremely interesting analysis by John Black in yesterday’s AFR (page 47). His analysis of the voting patterns on a seat basis was that they very strongly correlated to income and wealth. The higher the income and wealth the higher the yes. If you get the chance to read the article it is worth the effort.



  9. This was an interesting comment; ‘There is a phenomenon in political science called “low-contact intergroup threat perception”’, in other words ‘no’ voters were motivated by a sense of feeling threatened. In reality many voters in electorates like Parkes believed that a committee was never going to solve complex social issues. It is sad that the analyists are so convinced that the Voice was a brilliant idea that they cannot believe there could be valid reasons for rejecting it!!

  10. Peter Sherman, that’s an utterly disgraceful and frankly, a racist comment. Aboriginal and TSI people are allowed to live in major cities and have decent incomes. It doesn’t make them less Aboriginal or TSI, and it’s not your place to decide how legitimate they are. I’d also suggest that plenty of Aboriginal and TSI people in major cities are also doing it tough, despite your attempt to claim they’re mostly wealthy and therefore not as Aboriginal/TSI as their rural counterparts.

  11. Some additional stats from AEC:

    – Halls Creek, with a 78 per cent Indigenous population had a narrow 52 per cent Yes vote
    – Brewarrina, in the electorate Parkes, has an Indigenous population of 56 per cent. It delivered a Yes vote of just 35 per cent.
    – Wilcannia, in the same seat, is 51 per cent Indigenous and was at 39 per cent support. It had five remote mobile teams, all of which returned results favouring No.

  12. We do have “majority minority” electorates, like how the US has “majority minority” congressional districts. Banks and Bennelong for example are majority Asian. I’ll try and compile a list actually because this seems interesting.

  13. It seems that in Lingiari remote mobile voting teams (RMVTs) have a mixture of results but actual polling places have majority No votes.

  14. Anyway, back to “majority minority” electorates: it’s harder to calculate given that there is no race question on the Australian Census. There is, however, a nationality question, so I’ll add up all the nationalities and see.

  15. I’m having trouble finding full results for federal electorates as the AEC only includes the top five, however we can assume Bennelong is a majority minority seat as 34.8% are either Chinese or Korean, so add in the others and it’d be likely over 50%.

  16. And 31.8% of Fowler are either Vietnamese or Chinese so we can also assume that Fowler is majority minority.

  17. As for state electorates, I’ll start with NSW:

    * We can assume Ryde is majority Asian as it is 35.8% Chinese or Korean alone
    * We can also assume that Epping is majority Asian as 36.9% are Chinese or Korean alone

    Note that “Asian” doesn’t include Indian or Middle Eastern people, so if we include them as “Asian” then:

    * 45.6% of Epping is either Chinese, Indian or Korean
    * 41.1% of Bankstown is either Lebanese, Vietnamese or Chinese
    * 41.0% of Ryde is either Chinese, Indian or Korean
    * 38.3% of Kogarah is either Chinese or Nepalese
    * 38.2% of Auburn is either Chinese, Lebanese or Korean
    * 37.2% of Parramatta is either Indian, Chinese or Lebanese

    And so on.

    Let’s look at that the NT:

  18. The NT has several Indigenous-majority electorates:

    * Arafura (82.6%)
    * Gwoja (75.4%)
    * Arnhem (68.9%)
    * Barkly (64.5%)

    For the record, Daly and Namatjira are slightly more non-Indigenous than Indigenous.

  19. Anyway, for the record, only 32 electorates voted Yes. Of these, 12 were in Melbourne, eight were in Sydney, three were in Brisbane, three were in the ACT, two were in Perth and two were in Hobart. The remaining two were in Newcastle (obviously in Newcastle) and Cunningham (in Wollongong). No electorates in all of SA or the NT voted Yes.

  20. No, we can’t assume Fowler and Bennelong are majority minority. Prove it, instead of making logical leaps.

  21. I would say Chisholm & Calwell are non European majority in the latter case a large Middle Eastern & growing South Asian community with only an Italian Community being a significant European community. Possibly Bruce as well including Afghan community. However, i am doubtful about Banks being majority minority the Chinese community is 20% there is Nepalese community around Hurstville (would be less than 5% of Banks as while though but that is it no real Indian community etc. Fowler has a large Iraqi community as well. I would expect Blaxland to be majority minority and much more diverse than Barton/Banks.

  22. I don’t think any of those seats count because you are lumping together a bunch of different communities. I don’t think “majority non-European” meets the criteria.

  23. I’ve seen majority-minority used to refer to both cities/areas/electorates/etc. that have a majority of one specific non-white ethnic group, and also those that, when combining all non-european/white people, are majority-minority.

    Despite the fact that the ABS does not record racial data we can assume a number of the latter exist here by making estimates with ancestry data. I interpreted Ben’s comment on the lack of majority-minority electorates here in the above post to be referring to solely Indigenous people anyway given the subject.

  24. Interesting discussion, Ben.

    I wonder if remoteness affected the propensity to vote Yes, that is it isn’t connected by a major highway, has a small population and is miles away from a town of any significance. There was a strong Yes result in Goodooga in Parkes but not in Wilcannia which is a bit over half indigenous but is connected by the Barrier Hwy and has a larger population.

    In Queensland, Palm Island and Mornington Island are not accessible via land and had strong Yes votes.

Comments are closed.