The urban geography of attitudes to marriage equality (or, did car drivers vote 'no'?)

Yesterday, the Australian Bureau of Statistics announced the results of the postal survey it was directed to conduct by the Turnbull Government about the Marriage Act. Across Australia, 61.6% supported changing the Marriage Act to allow same-sex couples to marry. Yay!

But the uneven geography of this vote has also attracted attention. The 'Yes' vote was slightly lower in NSW - largely because 12 of the 17 Commonwealth electorates that had majority 'No' votes were in NSW, and specifically, in Western Sydney. This map of survey results has been doing the rounds:

Map of Survey Results in Sydney. Source: https://marriagesurvey.abs.gov.au/results/response-map.html

There are some particularly sharp divides here. Grayndler (not named in the map above, it is the purple electorate that sits in between Sydney and Watson) had a high 'Yes' vote of almost 80%. Watson, right next door, had a high 'No' vote of around 70%.

So, what should we make of this difference? How should we try to explain it? A few articles are popping up on the theme....

In this piece in the Guardian, Sam Dastayari - a Labor Senator from the region - argues that people in the region tend to be socially conservative. As for why this is so, he is quoted as saying: "The only demographic indicator that matters is ethnicity", and that any explanation based on a single religion would be incorrect. "[It’s across the board – Muslim, the Coptics, Christians, communities from migrant backgrounds contributed to voting no."

Likewise, Andrew Jakubowicz writes in The Conversation that "social conservatism among many ethnic communities loomed large as a factor" in the 'No' vote in Western Sydney". But he tells us not to essentialise this conservatism, and suggests a range of factors that might explain it:

The opposition to same-sex marriage ... was particularly resonant in communities where people have fairly poor educational backgrounds, somewhat limited English language skills and their information is mediated primarily through religious institutions.

So, in localities where there are strong communities built around Eastern Orthodoxy, Islam, Eastern Catholicism, African Christianities, Asian Christianities (ranging from Catholic to Evangelical), and even in other areas with pockets of Orthodox Judaism, there were singular funnels of information presented in cultural and moral terms.

There’s little information available to those people from any other source that they would trust, or to which they have easy access.

He goes on to argue that "It is important not to allow stereotypes to overwhelm analysis", and that "This is not simply about religion; it’s about culture in a more complex sense".

My colleague Dallas Rogers has written a piece for the ABC which seeks to head-off the stereotypes of those in 'the west' that are ever-present in discussions of Sydney's urban geography, and which are being hastily mobilised in some of the initial reactions to the poll. He worries that:

The discussion about the No vote will almost certainly follow a familiar narrative: the west is full of variously conservative new migrants, old people and ethnic and religious minorities. Religious and migrant media will come under fire. The familiar troupe of derogatory Westie, migrant and religious discussion will follow.

Dallas doesn't seek to explain away the poll results. Indeed, he identifies himself as a Westie who voted Yes, and as one who is disappointed by the results in this part of Sydney. But he makes a really important point when noting that:

The temptation might be to think about the diverse peoples and communities of western Sydney as the sole custodians of their views.

While the vote was clearly uneven, there were lots of 'No' votes from people outside Western Sydney, and lots of 'Yes' votes from people in Western Sydney, so we should not fall into the trap of focusing only on the colours of that map above. He also goes on to point out what we might forget now that the results are in -- that this postal poll was an incredibly divisive political tactic deployed by the Turnbull Government, which many have criticised for the fear-mongering and hate that it unleashed.


So, while it's tempting to try to explain the unevenness of the vote with reference to some of the rather obvious correlations between the poll results and the socio-demographic characteristics of an electorate derived from the 2016 Census, it's best not to jump to easy and unsupported conclusions.

Sure, there are significant differences in factors like religion, income, and education between majority 'Yes' electorates in inner urban areas like Sydney and Grayndler, and majority 'No' electorates in Western Sydney like Watson and Blaxland:

	Sydney	Grayndler	Watson	Blaxland
No Religion (%)	43.7	40.7	15.6	13.4
Avg Weekly Income ($)	933	997	490	462
Bach Degree or Higher (%)	43.8	42.6	22.9	17.3
English only Spoken at Home (%)	51.0	66.5	28.5	25.5

There are also some data that are surprising. For instance, of these four electorates, Sydney has the lowest percentage of people born in Australia:

	Sydney	Grayndler	Watson	Blaxland
Born in Australia	39.2	59.1	44.4	43.9

But before we start reading explanations from these correlations, note that there are also stark differences in other characteristics of the majority 'yes' and 'no' electorates, like public transport usage:
 

	Sydney	Grayndler	Watson	Blaxland
Public Transport to Work	35.7	38	25.2	21.3

Were people in Western Sydney more likely to vote 'no' because of their education levels, or because they are more likely to drive a car to work? There's a good correlation for both, but I haven't seen anyone try to explain the results in Western Sydney with reference to public transport usage! That's because, as they say, correlation does not equal causation...

More sophisticated analyses are going to be required to better understand which of any correlations might have some explantory relevance. And of course, such an analysis would need to take account of a range of other factors beyond simple demographic differences.

For instance, in his piece in Junkee, Osman Faruqi argues that:

The huge variation in the result across electorates has a number of factors, but is partly driven by the fact the No campaign appeared to focus its resources on migrant and non-English speaking communities while the Yes campaign spent its time and energy turning out the vote rather than convincing people to change their position on the issue.

I don't whether this is right, but it certainly does stand to reason that we'd need to look at the geography of the campaigns and information if we're interested in the geography of the result.

Sounds like something for some human geography students to get getting on with...!