Comments on: How Checkout Chicks Smashed Howard.

By: Possum

Possum — Fri, 12 Dec 2008 00:40:40 +0000

CCD level data would be great to use, but the big problem is linking it to booth level data. I’ve been playing around a little using a couple of seats I know well to develop a way to cluster the CCD data to booths (using some off the shelf spatial data aggregation algorithms in GIS), but nothing has really worked to my satisfaction yet. The smaller the booths the greater the difficulty.

If we run CCD level data against the swing by seat, a lot of really good info get’s washed out and some info will aggregate in misleading ways – but there’ll also be some richer things floating around. The problem is trying to sort the wheat from the chaff!

I’m using 2007 election boundaries.

If you could keep me up to date with any interesting things you find in the data it would be much appreciated – better still, start a blog El Nino!

By: David Richards

David Richards — Thu, 11 Dec 2008 14:29:47 +0000

Maybe you should send this to Julia Gillard and suggest that the ALP has a huge mandate to go even further with dismantling WorkChoices and making it highly unlikely such a system will ever again see the light of day. The current downturn and impending layoffs shows that no matter how good you think your AWA was in good times, the rights you signed away bite you in the bum bigtime in a downturn.

By: Pica

Pica — Thu, 11 Dec 2008 11:06:15 +0000

Poss, El Nino, fascinating to see (and only vaguely apprehend) the depth to which youse people dig down into this info. Interesting to see that work choices was poison, and then some, and that the libs have issues with forming a core demo, these are smarty pants insights that’ll get a run at my xmas parties (with acknowledgement of course!). Cheers.

By: mad cow

mad cow — Thu, 11 Dec 2008 10:52:29 +0000

I’ve always felt that in order to properly model elections you need to model the spread of ideas via social networks. Its no surprise to me that a relatively small number of people has a relatively large effect

By: El Nino

El Nino — Thu, 11 Dec 2008 10:25:05 +0000

Freaky! Thanks to recent redundancy and a bit of time on my hands, I have been keeping myself busy by running brute force correlations against the 7000+ census variables and 2007 first preference votes by electorate for ALP, LIB, Green, FFP and Nats. My laptop could fry eggs at times!

Firstly, I am interested to see that you are running electorate level census data. Wouldn’t the much smaller census districts (CD) give you more richness (effectively using the electorate swing as an averaged result for each CD)?

The results I have been seeing are fairly predictable (although nevertheless enlightening) for each party with one notable exception. Except for the Liberals, each of the parties are turning up 100+/200+ strongly correlated variables. The findings are pretty much as you would expect: Nats are rural blue collar, Greens are young, wealthy and well educated, ALP heartland is in the non-Anglo ethnic communities and transport workers, FFP are religious and dropped out of high school (actually, dropped out in Year 11, which may be an indicator of aspirantionalism).

The most interesting is the Liberal vote which turns up only about 30 significant variables and the correlations in these are generally much weaker. This is telling me that there is no real demographic that the Libs appealed to at the last election. The strongest Liberal demographic are immigrants from the UK which may explain why they had Blair and we had Howard (ha ha).

Also, did you take into account changes in electorate boundaries (and new at least one new electorate) between 2004 and 2007 as this may affect swing values? This is something that you can account for more accurately at CD level (something I am about look at).

By: girtbysea

girtbysea — Thu, 11 Dec 2008 04:50:30 +0000

Possum: If you’re getting excited by those lines through those dots, and believing they mean something, you really do need to get out more, or get some counselling. Take a break over Chrsitmas at least, this obsession of yours is just not healthy.

By: US Election On Best Political Blogs » Blog Archive » How Checkout Chicks Smashed Howard. - Pollytics

Thu, 11 Dec 2008 04:08:11 +0000

...] How Checkout Chicks Smashed Howard. – Pollytics I pulled out of the census the percentage of the total population of each electorate that marked “retail” as an answer and compared it to the ALP swing by electorate at the last election. The results are incredible. … [...