Comments on: Redistribution latest

By: Geoff Lambert

Geoff Lambert — Fri, 11 Aug 2006 11:26:25 +0000

Hmmm…. a Chi-Sq test on these numbers returns P

By: David Walsh

David Walsh — Thu, 10 Aug 2006 14:48:36 +0000

Interesting Geoff. But I wonder if the large % discrepancies are as significant as you make them out to be. Or whether they’re simply exaggerated by the small-ish sizes involved.

For instance, to use the Dee Why Central example:
The 481 Mackellar voters went 58.4% Liberal
The 1546 Warringah voters went 52.8% Liberal

A difference of 5.6% may seem large. But 5.6% only amounts to 27 voters for Mackellar.

Or to look at it another way, the 2027 Dee Why Central voters went 54.1% Liberal in total. If you split that between the two electorates as per the 481:1546 ratio above, you get a result that’s only 21 voters off either side.

A drop in the ocean in electorates of 80,000+ voters.

By: Sacha

Sacha — Thu, 10 Aug 2006 09:30:38 +0000

Geoff: perhaps results in booths shared by neighbouring electorates could be used to look at candidate effects on how people vote.

It might be interesting to look at this in areas that are pretty homogeneous.

By: Geoff Lambert

Geoff Lambert — Mon, 07 Aug 2006 04:18:39 +0000

Sorry about this, but I overlooked the peculiar case of the booth at Rottnest Island- you can vote in any WA Federal Electorate at Rottnest. If you take Rottnest out, the above numbers become 5.6%, 0.01% and 24%.

By: Geoff Lambert

Geoff Lambert — Mon, 07 Aug 2006 04:10:04 +0000

Just to expand on the point about shared booths…

There were about 230 of these at the 2004 election (not counting the CBD booths which serve ALL electorates). The average absolute difference in the TPP for the two (it’s nearly always only 2) electorates shared by these booths was 6.1%, the smallest difference was 0.01% and the largest was 40% (!)

By: Geoff Lambert

Geoff Lambert — Mon, 07 Aug 2006 03:45:06 +0000

Iâ€™m sure even Malcolm and Antony would be the first to admit that recalculating TPPs for redistributed electorates can be a bit iffy. Having done so myself, I would think that itâ€™s so iffy that, in most cases, itâ€™s probably not worth the effort- which is very considerable. There are all sorts of issues to do with the population of booths within an electorate, how they may change from election to election, what their â€œcatchmentâ€ is and how to split up their votes when their existing catchment partly changes from one electorate to another. These problems loom particularly large in the border areas between two urban seats which are being redistributed. There is also the problem that the people who vote at a particular booth, although they are presumably of the same demographic, vote differently according to which electorate they are in. You can see this in places like Dee Why, where the main booth services both Warringah and Mackellar (both fairly blue-ribbon-ish Liberal). The TPP at this booth can differ significantly depending on whether the voters vote in a Warringah or a Mackellar cubicle.

When we get down to splitting hairs at the less than 1% change level (as in Bennelong), we really are wasting our time starting the whole process. This doesnâ€™t mean that I donâ€™t take advantage of Malcolmâ€™s hard work in doing my own tea-leaf reading prior to an election, but I donâ€™t think I could show an increased prediction performance on the basis of it.

Geoff Lambert

By: Sacha Blumen

Sacha Blumen — Thu, 03 Aug 2006 13:35:55 +0000

From a previous comment of mine:

Here are the 2 party preferred votes in the booths in the 2004 election for the proposed Blair eletorate:

from Oxley: 7240 ALP, 4460 Lib
from Forde: 1330 ALP, 3749 Lib (including 1/2 of Flagstone)
from Blair: 19067 ALP, 27527 Lib

The part from Forde is Boonah shire with perhaps some extra bits, so you can see why the ALP wants Boonah shire to stay in Forde.

By: Paul

Paul — Thu, 03 Aug 2006 12:16:22 +0000

you gotta love the masses who cant understand why seats go away – you saw the same in SA in ‘04

By: David Walsh

David Walsh — Thu, 03 Aug 2006 01:59:22 +0000

Blair remains of interest. The commission may not be done cutting down the Liberal margin.

Both John Cherry and, unsurprisingly, Labor object to the inclusion of Boonah LGA and instead suggest expanding Blair into Ipswich East.

The political effects are straight forward. Boonah LGA contains 6000 registered voters. I haven’t fully figured out all the polling booths belonging to the shire, but it looks like they voted Liberal roughly in the order of 70-75% in 2004. Whilst Redbank & Redbank Plains in Ipswich East voted Labor roughly 60%. Curiously, Cherry adds more voters here than Labor does, 4800 vs 3600.

On the face of it, the suggestion appears to have a fair bit of merit. Why does Blair shed rural territory to the north only to gain rural territory to the south? A seat even more centred on urban Ipswich makes sense. As does retaining Boonah in the predominately rural Forde.

The real test is whether the changes are outweighed by less than ideal knock-on effects to other seats.

By: Ben Raue

Ben Raue — Wed, 02 Aug 2006 02:36:40 +0000

Check out the number of objections to the redistribution in NSW. I expect that a lot of them are form letters related to the impending demise of Gwydir.

http://www.aec.gov.au/_content/why/redistributions/2005/nsw/object.htm