Facebook Google Menu Linkedin lock Pinterest Search Twitter


Voting behaviour

Apr 15, 2010

Class, voting and broad left demography

One of the most historically accurate predictors of the way people vote in Australian elections has always been the notion of “class”. At its most reductionist, working class voters


One of the most historically accurate predictors of the way people vote in Australian elections has always been the notion of “class”. At its most reductionist, working class voters generally support the ALP, owners of capital and employers generally support the Liberal Party and in the post-war years, rural Australia has generally supported anti-Labor candidates of various hues.

To a large extent that basic pattern still holds, but class structures are evolving to the point where many of the relationships we used to see are no longer as predictive as they once were.

For instance, manual labourers used to be the backbone of the Labor Party voting stock (that and the public service). The more manual labourers that lived in an electorate, generally, the larger was the ALP vote. Yet if we look at the electorate level results of the last election and compare the ALP primary vote with the proportion of people in the last Census that described their occupation as “Labourer”, the relationship has pretty much broken down:


Historically, we would have expected to see that scatter plot producing a pattern that started from the bottom left and ended in the top right of the chart – signifying a strong positive correlation between the two. Now, however, it’s almost a random cloud showing no correlation between manual labourers and the ALP primary vote at the electorate level.

There’s all sorts of reasons for that, from the changing socio-economic mix of our urban electorates through to the changing nature of our economy itself.

Yet, despite historical relationships like this no longer turning up with electorate level data, there are some proxies for class that still dominate the electorate level data.

The biggest predictor of the Labor Party primary vote at the last election wasn’t the usual suspects. It wasn’t the proportion of the electorate that came from a non-English speaking background, it wasn’t the proportion of the electorate that received family tax benefits, it wasn’t the proportion of the electorate that worked for the public service.

The biggest predictor was an inverse relationship, and one that explained 55% of the variation in the Labor Party vote – it was the proportion of the electorate that classified themselves in the Census as working as a “Manager”. The higher the proportion of managers in an electorate, the lower was the ALP primary vote.

If we run a scatter and regression line of the two for all 150 electorates, this is what we get:


For every 1% increase in the proportion of the electorate that worked as “Managers”, the ALP vote decreased, on average, by 2.3%.

As you would expect, the Coalition experienced the opposite effect:


Although the strength of the effect was slightly weaker, with the proportion of the electorate working as “Managers” only explaining 37% of the variation of the Coalition primary vote, and where a 1% increase on the proportion of managers working in an electorate, the Coalition primary vote increased, on average, by 1.8%

The other interesting thing that pops up is where the Greens come into all this.

Why are the Greens strongest in the inner cities? What is the cause of Greens voters living in the inner cities?

One thing that may help explain it is occupation. If we use the census data to sum together the people in each electorate that work in the three categories of Arts & Recreation Services, Information Media & Telecommunications and Education – and then scatter that against the Greens primary vote across 150 electorates, this is what we get:


The proportion of people working in these three industry sectors explains 51% of the variation in the Green’s primary vote. What really strikes me here, and makes me think it’s more a causative result than a proxie for “something else”, is the way that the linear relationship holds strongly across the entire spectrum of the Greens vote.

If we do the same thing again, but this time run a locally weighted polynomial regression (Loess curve) through the scatter as well, it will give us a non-linear trend line of the correlation between these industry employment types and the Greens vote as we move across the Greens vote spectrum from 2% through to nearly 23%.


The black line is our linear regression line, the red line is our loess curve – and I’ve adjusted the left axis scale here to visually maximise the difference between our two trend lines.

The loess curve sits very close to the linear regression line across the entire spectrum of the Green vote, telling us that the linear relationship isn’t being skewed or distorted by some block of either Greens voters or industry workers. If the red line started diverging massively from the black line, it would tell us that the correlation isn’t consistent across the whole spectrum as well as letting us know through what values the inconsistency exists.

But here there is only a very small divergence at around the 10% Greens vote mark, but then only maxing out at around 1% of the proportion of these industry workers before heading back to match the linear trend.

This is why I think there’s some strong causative aspect to this relationship – the consistency across all electorates and across all values is so strong, from the lowest Greens voting electorates, through the bulk of the middle Greens voting electorates and into the top Greens voting electorates.

If there wasn’t a causative aspect to this, but just, say, some generic correlation between the inner cities and the people that work in these particular industries, then we would expect to see the linear nature of the relationship break down as we moved away from the high Green voting seats and into the middle and low voting Green seats – we would expect to see the linear relationship breakdown as we move away from the inner city clusters of high Greens voting electorates.

But we don’t – the linearity of the relationship remains consistent.

So what does this mean in practice?

For starters, a large part of the “cultural” aspect of Greens voters living together appears to come as a result of hard cultural issues such as their employment, rather than because of soft, vacuous stereotypes like the cultural pursuit of the perfect latte – the sort of nonsense promulgated by pop-demographers like Bernard Salt (a story of which I should tell you about one day – a question I asked him at the recent Qld population summit resulting in some horseshit about “Metropolitan Chauvinism” and latte drinkers living in West End. Ugh!).

People working in the arts, education, media and technology industries are more likely to vote Green, and as a result of the distribution of workplaces for these industries having a higher density in the inner suburbs, the people living within close proximity to their workplaces naturally leads to the inner cities having higher levels of Greens voters.

As the density of these industries reduces, so to does the Greens vote.

In one respect, it’s just a modern evolution of class based electoral analysis – whereas the Labor vote used to correlate strongly with manufacturing and low skill, labour intensive industries – today, the Green vote correlates with new skilled services industries like arts, education, media and technology.

So it’s not something particularly new, it’s just a relatively new type of something that we’ve long seen at play in the Australian electorate.


We recommend

From around the web

Powered by Taboola


Leave a comment

44 thoughts on “Class, voting and broad left demography

  1. lorenzo aka erudito

    Great post, particularly since it fits in with an analysis of mine of environmentalism and PC as being driven by folk who work in “symbol manipulation” jobs — I wish I had known about this post when I did mine.

  2. Andrew Norton » Blog Archive » Why do graduates lean left?

    […] short time in most people’s lives. People spend far more time at work than at university. Scott Steel has shown that electorates with high proportions of people working in Arts & Recreation Services, […]

  3. Just who are these Greens characters anyway? | Who says?

    […] and I won’t go into it too deeply. If you want more information, you can go here, here and especially here. All of those analyses try to sell Greens voters as really ‘tops dudes’. This is […]

  4. Some Guy

    Possum, have you compared data from older censuses for occupation against votes in electorates to see if there has ever been a relationship between the two? For example, census data might show that there has never been much of a correlation between manual labourers and the ALP primary vote at the electorate level?I’d also point out it’s a bit down to the analyst’s intepretation in the census data of who is a labourer, manager, etc.

    There’s also the issue nowadays of persons who you’d think of as manual labourers (e.g. brickies, painters, etc.) earning the kind of money that only managers and other white collar professionals could earn. On the other hand there are a lot more managers around these days in retail, hospitality, etc. many of whom would be earning less than labourers … How do people vote when they’ve got a working class occupation but on a middle-class income and vice-versa?

  5. Unrepresented

    Your hypothesis is that “class” is still important, but you weaken this with your claim that a complete hodge podge of a group determines who votes Green. Can you come up with any definition of the class that the Greens are? (apart from latte sipping inner city singles with no children?)

  6. J-D

    Have you looked for a correlation between ALP or Coalition vote and proportion of people who are employers and/or proportion of people who are employees? If you’re looking for a line between ‘classes’, then the line between employers and employees should be considered.

  7. Liberal Slayer

    Hey Poss, Sorry Off Topic:

    Do you know where I can obtain a Mean Income by Suburb list? or how can I calculate it from the census data???

  8. Venise Alstergren

    MD: Are you absolutely sure? Hang on. I don’t think it was the Greens themselves. It might have been the commentator on Channel Two.

    I don’t usually make this sort of mistake. However, if you say so….

  9. chinda63

    preview FAIL! That should have read “southern and eastern urban-fringe seats”

  10. chinda63

    Possum – could the movement of electoral boundaries have any effect on any of these graphs?

    I’m thinking specifically of how urban boundaries tend to be the ones moved around more in order to meet fairness and population fluctuations (particularly in a state like SA where we do it every 4 years). This process has seen sections of safe Liberal seats being moved into more blue collar Labor seats and vice versa, particularly on the urban fringes. We seem to have some strange boundaries drawn where, for example, some seats in the north are absolutely boganville in the northern areas of the electorate and toffy-nosed in the southern areas – one end is solid Labor-voting and the other is solid Liberal-voting. The same things happens in the southern and urban-fringe seats, and in the west the nearer you go to the sea.

    Is there anything similar going on in the other states and, if so, can this have an impact when looking at things in these terms?

    In other words, is it possible that you would get a more expected outcome if you go right down to CCD/polling booth level, rather than looking at the overall electorate?

  11. mfs

    Jaeger: What about Greens voters who own VERY expensive coffee machines? I know a few 😉

  12. mfs

    Thanks for another insightful piece, Poss.
    Admittedly I’m in an electorate with a high Green vote (Denison), but if my experience is anything to go by, people (colleagues) employed in Science and Research vote overwhelmingly Green. However, many of these would be at least loosely affiliated with a University (education).
    Hobart seems paradoxically Green, though I suppose there are historical reasons for this as well. But science (and science policy) is a major employer down here, with CSIRO Marine and Atmospheric Research, the University of Tasmania, the National Oceans Office, the Marine Division and the Australian Antarctic Division all located in our area. I presume people employed in science are perhaps more acutely aware of our negative impacts on the environment and intelligent enough to do something with their vote…

  13. Jaeger

    I’m somewhat bemused by the Leftist latte stereotype. Surely the affluent right are the coffee snobs? A bespoke $1000+ Italian coffee machine whose name can be dropped casually into conversation with other coffee snobs to elicit nodding approval, lots of free time to grind your custom-roasted beans by hand – that kind of thing.

    I guess Nespresso drinkers must be Leftist – loathed by coffee snobs and Greens alike.

  14. don


    [Poss which electorate is the one with only 10% ALP vote? It sits way off the ALP vs Manager trend. ]

    As Poss says, New England. But there is a good reason for that.

    We have an independent federal member there, Tony Windsor, who can have the seat for as long as he likes.

    He is seen by the electorate as fair, hardworking, and canny. And he seems to be doing a great job for the electorate. Although I would normally vote labor, there is now way I would want anyone else but Tony. It seems that any number of nominally ‘labor voters’ in this area thinks the same.

    But we are doubly blessed, because we’ve got Richard Torbay as our state member, (Northern Tablelands) another independent. I’ve never seen anyone like him, he is phenomenal. Incredibly hard working, and gets things done for the area in a way we have never seen before.

    Finding a credible labor candidate who would be willing to go up against either of these two knowing they haven’t a snowball’s chance in hell of winning, is virtually impossible.

    Thus the low labor vote. We have had a very good (state) labor member here some time ago, Bill McCarthy, who died in office. Thus the seat is winnable for labor, but not while Richard Torbay and Tony Windsor keep throwing their hat in the ring.

  15. Possum Comitatus

    Toypoodle, when I can remove the rural electorates it doesn’t make any real difference.

    Gazo said:

    [What happens when a “Green” person working in IT/Arts/Media becomes a manager?]

    They consult this page:


    Dekay, I’ll have a play around with the income and see.

  16. gazo

    Interesting question: What happens when a “Green” person working in IT/Arts/Media becomes a manager? Do they switch their vote to Liberal? Can the stats shed any insight on that proposition?

  17. Toypoodle

    On manual labourers: is there a way to disaggregate rural and urban? It may show some differences.

  18. gef05

    If you don’t know the story behind the “toughs and toffs” photo then please don’t use it. It does not show what you think it shows.

  19. dekay23

    I think that difference between education level and household income might be more revealing than education level alone. High income relative to education level voters are more likely to vote conservative? Would probably need to use education and income ranks to determine…

  20. MD

    I don’t think the Tasmanian Greens ever claimed to be winning 10 seats, in fact Nick McKim was very quick to play down the likelihood of even winning 6 seats in the aftermath of the EMRS poll in late February. And certainly by the close of counting on election night I don’t think anyone was expecting anything more than 5 seats to the Greens.

  21. Venise Alstergren

    Possum: I haven’t finished reading the Post but in case you don’t make the psychological possibility an issue….

    Couldn’t the Green voters of urban Oz have it at the back of their minds that ‘One day, I’ll be in a position to enjoy the rural scene. Meanwhile, I’ll do everything in my power to make sure a lot of the original land and animals are left intact.’

    Couldn’t it be that simple?

  22. Venise Alstergren

    Dear Poss,

    I have a question to which I’d very much like to know the answer.

    Why is it that during an election the Greens always claim to be winning a hundred percent more seats than they actually win?

    Eg: Tasmania. On the night of Tasmania’s last election I went to bed in a fairly happy frame of mind. Safe, in the knowledge that they had won ten seats. Came the final reading of the tea leaves, they had won five.

    This is far from being the first time this has happened.

    Dear Uncle Poss, please explain.

  23. TomBrisbane

    Interesting stuff, but possibly misleading…

    It’d be interesting to know what sort of jobs these Green voters actually do. They may not all be ‘new industry’ types. Part of the shift to a high inner-city Green vote amongst workers in these sectors may have come from a large scale defection of teachers from Labor to the Greens. That’s speculation, but if true, they’d dwarf the number of workers in IT, arts and media, I would have thought.

    Also, there’s a statistical thing with Labor and the labourers. This section of the workforce (defined by the ABS) has consistently shrunk over the years. Perhaps in the 1930s or 40s, labourers accounted for 50% of an electorate, generating a high Labor vote. Now, however, the labourers share of the electorate is so small, between 3% and 20% on Possum’s figures, it’d be surprising if they had much impact on the overall Labor vote which varies from 20% to 65% in any electorate. It rather depends on who the other 80%-97% of the electorate are!

    However, possibly confounding my argument, there *is* a correlation between Labor’s vote and % managers, and they are about the same proportion of the electorate as labourers, so perhaps they tend to live in more socially homogeneous (i.e. uniformly middle class or wealthy) suburbs than labourers. More speculation.

    More useful, but difficult to get, would be vote by occupation.

  24. Possum Comitatus


    No electorate has even 18% of their population working in those areas combined.Look at the chart!

  25. green-orange

    IT, media, arts, recreation, and education ?

    That’s about 40% of workers.
    I hardly think you can get anything out of such a broad base.

  26. Possum Comitatus

    Rationalist, no significant (or any at all!) correlation between mining and party votes. Engineering isn’t categorised as a stand alone occupation in the census, but seems to be rolled into “technicians and trades workers”.

    If we use that, there’s no relationship between it and the major party votes, and theres negative correlation between that occupation and the Greens vote – but mostly as a result of the geography of seats with low proportions of technicians and trades workers are inner city. Removing the inner metro and running the regression again, it pretty much makes the correlation disappear completely.

  27. rationalist

    Hi Possum,

    Is there any significant electoral correlation (for say Labor, Liberal or Greens) for those who work in mining, minerals, engineering or industry?

  28. Down and Out of Sài Gòn

    Brilliant analysis. But there is one tidbit that you haven’t explored enough.

    …the sort of nonsense promulgated by pop-demographers like Bernard Salt (a story of which I should tell you about one day – a question I asked him at the recent Qld population summit resulting in some horseshit about “Metropolitan Chauvinism” and latte drinkers living in West End. Ugh!)

    Tell us more, Possum. 😉 Seriously, I’m curious of those who despise West Enders like myself. It’s a rare pathology. Personally, I prefer to plunge my own coffee these days, but there’s nothing wrong with a latte. I guess Salt is too proud to go that route, and prefers two spoonfuls of bitter International Rust in his cuppa.

  29. Robert Beswick

    Excellent work Poss.

    This along with Anthony Green’s great piece on the prospects of the double disolution show us all once again the sad gulf between the truth that is obtainable through expertise and ability to do the math as compared the complete BS that usually passes for news and “analysis”.

  30. Wobbly

    Mr Denmore, I am a Green voting Media Manager with a career’s experience in the industry, but I voted Green before I was promoted to Management. A managerial colleague who reached Managerial level in other industries before joining my organisation/industry are somewhat less progressive.

    Great work Poss!

  31. EnergyPedant

    When talking about Education I meant where those within education worked, not the level of education they had. My hypothesis is that those employed by Universities are likely to be more green than those who work at Secondary schools. Those university lecturers also tend to be concerntrated in the inner-urban green areas like Brunswick, Fitzroy, Carlton.

    Very Bad Simplistic Hypothesis: University staff => Green, Government School Staff => Labor, Private School staff => Liberal.

  32. Alex H

    Just realised that you wouldn’t have data for individuals voting intention and education level and that you need the electorate location to bring the data together. Oops

  33. Alex H

    On a similar line to Energy Pendant, I think that changes in education have had a major impact on voting intentions as well as a big shift in the type of work being done in the workforce. Comments below are completely opinion and conjecture, I can’t reference any facts to support 😉

    The traditional bastion of the left wing parties was in large unionised workforces which initially didn’t have a lot of power or adequate rights and conditions, ensuring strong loyalty to the labour party. In the post war years there has been a massive reduction in that sort of work due to mechanisation, automation and outsourcing of manufacturing to other countries. At the same time conditions and rights have improved considerably lessening the need to vote labour.

    Over the same period there has been a big increase in people going through tertiary education and entering white collar jobs. So the demographics within the electorate have changed significantly, and the old drivers for voting intention aren’t as significant, as the analysis above shows.

    My gut feel is that people vote moderate right (read Liberal) from self interest (Managers, people with capital) or because of conservative values (older generation, WASPs). Far right parties such as Family First and One Nation get the rednecks of one form or another, although a fair few of these might vote Liberal.

    I would guess that you would get an interesting correlation based on education level. This would capture the impact of the older generation given the recent increase in tertiary level education and would certainly capture dyed in the wool rednecks. The correlation would be weakened a bit by well educated managers and people with money voting liberal. In previous posts you have looked at the impact of age on voting intentions which was very illustrative and would probably tie in with the correlation between age and education level achieved.

    Rather than % of tertiary graduates vs primary vote it would be interesting to graph directly level of education achieved against voting intention for each party, this would go further to pinning down a correlation. This is because by looking at the % in the electorate you don’t take into account what the voting intentions of the balance of the electorate are.

  34. DodgyKnees

    Thanks Possum for putting flesh onto the bones of my vague suspicions.

    Is there any way of building profiles of Left, Centre, Right voters and comparing them with those of Left, Centre, Right politicians ?

    I often wonder how or if they match.

  35. Mr Denmore

    So if you’re an arts/media/communications professional who is a manager, who do you vote for? What takes precedence – your identification with your profession or your supposed solidarity with the forces of capital? A test of false consciousness.

  36. Possum Comitatus


    New England, where “Country Labor” scores 9.83% of the vote:

    All the data suggests is that 50% of the variation in the Greens vote is explained by the variation in the number of people in those occupations across all 150 electorates.

    So it’s certainly not “all” Greens votes coming from those industries, but a substantial proportion would.

    On the different levels of education – that data is in the census. I might have a play with it for another post some time if something interesting pops up with it.

  37. EnergyPedant

    Poss which electorate is the one with only 10% ALP vote? It sits way off the ALP vs Manager trend.

    Is there a way to separate out different levels of education? I suspect the greens vote is more strongly tied to tertiary educators than those in the school system.

    Also does you Greens chart imply that just about every greens vote comes from those few industries?

  38. Possum Comitatus


    There’s probably a disparity in the voting intentions of caravan parks.

    On the one hand, retirees and grey nomad types would probably be mostly Coalition voters, while low income, non retired voters living there would probably be more mixed – maybe with a lean to Labor.

  39. Possum Comitatus

    Bogdan, I didn’t actually! It was probably just my mangled twittering.

    But since you asked, here it is:

    Not much to see really 😛

    Unfortunately I don’t have that data at my fingertips at the moment, but I can have a sniff around.

  40. Bogdanovist

    This seems a more plausible explanation for Greens voting patterns than the ‘Doctors Wives’ pop-sociology hypothesis to explain the high Greens vote in safe inner city Liberal seats.

    I see from your twitter feed that you made a managers vs Greens graph (which would be good to see included above as well!). I’m not sure if it is possible given the data you have, but a “managers in Arts/Rec, IT and Education industries” vs Greens graph would be interesting.

  41. BK


    Great work – as usual.

    It would appear that Abbott’s “Listening Tour” was largely confined to people in caravan parks. Can you opine on whether bias would be present in that population and what the bias may be?

  42. vp


    Great data collection and analysis. Any conclusion drawn is dependent on the individual definition of “manager”. There is high correlation (and inverse) between primary vote and those who call themselves “managers”. It looks to me that the self-styled movers and shakers vote for pollies they think will let them do what they want and the other lot, the ne’er do wells, will vote for them what don’t colur themselves blue.

  43. don

    Poss, I’ve said it before and I’ll say it again.

    You think of questions I wish I’d been capable of thinking of – and answer them!


  44. calyptorhynchus

    Excellent, I don’t like Liberals and I don’t like managers!

    Now I can economise on my dislikes!

    Seriously, this is brilliant stuff Possum. Thanks.

Leave a comment


https://www.crikey.com.au/2010/04/15/class-voting-and-broad-left-demography/ == https://www.crikey.com.au/free-trial/==https://www.crikey.com.au/subscribe/

Show popup

Telling you what the others don't. FREE for 21 days.