I’ve made two New Year resolutions – the first is to start providing answers to your requests, and the second is to get through the huge backlog of email. So today, we’ll start on both (fingers crossed on the email).
In the previous Requestathon, CountArach asked:
Would there be any way to look at Australia’s views on Immigration vs Economic growth? This would show us if the theory that people generally blame immigrants during times of economic uncertainty is in fact true.
If we troddle on over to the Australian Election Study 1987-2007, there are three questions in the survey that provide us with some good data for this. Firstly, the AES asks the question “Do you think the number of immigrants allowed in Australia nowadays should be reduced or increased?” This gives us the data for Australia’s views on immigration levels.
Secondly, the AES asks the question “How does the financial situation of your household now compare with what it was 12 months ago?” as well as the same question but where “Country” substitutes for the word “Household”. We can combine the percentage of respondents that answered either “a little worse” or “a lot worse” to those questions and track whether negative self-perceived economic conditions walk hand in hand with changes in public opinion on immigration intake levels.
There’s a strong correlation there and the uptick in all three between 2004 and 2007 is interesting. If Rudd’s popularity holds up over the next year and the economic downturn starts to bite – will the Coalition be tempted to start banging the immigration drum? I guess we’ll see during the next 12 months.
Next up, Yaz asked:
I was fascinated by a comment in a recent post that Greens voters were quite likely to put down no religion on the census.
Any chance of some fancy graph plotting major religions vs voting habits, with some thoughts on possible causes/correlations.
To start off, let’s have a look at a scatterplot of the Greens Primary Vote vs. the proportion of the population that put “No Religion” on their Census form – and do it by 2007 Commonwealth electoral division boundaries.
So there’s obviously something to it, but the one thing that hasn’t got much to say about explaining it is median income. We’ve already seen that median income and the Greens vote level walks hand in hand by electoral division, but “No Religion” and median income has a much, much weaker relationship:
There’s not much of a correlation there at all. However there are two things that could be underlying a good part of the Greens/No Religion relationship – Education and Lone Person Households.
If we take the proportion of the population in each seat that has a Year 10 education or less and compare it the “No Religion” answer, we get the same sort of pattern we see when we compare the Greens vote to that education stat.
Similarly, if we compare the proportion of Lone Person Households in each electorate with the “No Religion” Census answer and the Greens primary vote – the two produce fairly compatible results.
If we make a really brave assumption that there is no substantial ecological fallacy polluting the numbers, Lone Person Households had a tendency to vote Green and they also had a tendency to put “No Religion” as an answer on the Census at higher levels, on average, than the rest of the population. Similarly, the lower the numbers of people in an electorate with a Year 10 Education or less, the higher both the Greens vote and the No Religion answer was, on average, across electorates.
But income doesn’t seem to have a great deal to do with it. This isn’t to say that highly educated Greens voters whom live by themselves have an overwhelming tendency shun religion – it’s extremely likely to be a hell of a lot more complicated than that, but it’s food for thought when it comes to exploring some of the cultural issues that drive certain groups of people with certain worldviews to live in certain suburbs and geographical locations – not to mention they way they vote.
You can also treat this thread as a Requestathon thread so please feel free to throw in any requests. There should be a few more answers to previous questions coming shortly(ish).
I hope you all had a merry festive season and best wishes to all for the New Year.







5 Comments
Thanks for the answer Poss! Happy new year to you too.
Thanks, Poss, I appreciate all that. Not being a stats type, I need some clarification about the graphs. Is the steepness of the line of best fit on the graph any measure of correlation, or is it simply an artifact of the choice of scale on the x and y lines? I am presuming the latter. If so, how do I gauge the level of correlation?
As to the substance of the graphs… Perhaps folks who choose to live by themselves are consciously choosing a life other than the partner/2.1 kids/dog/suburbia life, that might otherwise correlate with either of the major parties. Or perhaps Green voters are naturally more cranky and find themselves living alone because they have no choice!
On the Lone Person Household/No Religion correlation, you could hazard that it’s only people with partners and/or children who feel compelled (for some reason) to perpetuate religion!
OH NO!
Look What they’re doing to your cousins at FDacebook
http://www.new.facebook.com/tos.php?api_key=09c481eb4a47c45cc6dc190c4088a299&next=http%3A%2F%2F72.32.123.74%2Fv1%2F%3F_fb_fromhash%3D2a7863bae07b9776df3980d08e266406%26fb_sig_in_iframe%3D1%26fb_sig_locale%3Den_GB%26fb_sig_in_new_facebook%3D1%26fb_sig_time%3D1230873591.3968%26fb_sig_added%3D0%26fb_sig_canvas_user%3D663648790%26fb_sig_api_key%3D09c481eb4a47c45cc6dc190c4088a299%26fb_sig%3D40e5089345267e6976c0e1eeef58d639&v=1.0&canvas
Possum, can you factor the age distribution of the different electorates into your analysis? I wonder how much the relationship between Greens vote and religiosity reflects an age cohort effect. Religion and education are certainly strongly age cohort-related. For example, the rough national figures show that:
* 24% of 25-34 year olds stated ‘no religion’ vs 15% of those aged 35+
* 20% of 25-34 year olds had completed Year 10 or less vs 45% of those aged 35+
The 2 electorates with a Green vote above 20% also had the highest proportion of resident adults aged less than 35, but I don’t know how far this plays out. What would your correlations look like if you exclude age as a variable?