WA redistribution – follow-up analysis

By
Ben Raue
-
21

I wrote a quick blog post yesterday after the release of the draft electoral map for the next Western Australian state election. Now I’m back with a more detailed analysis of what has changed. I’ve included a map comparing the old and new boundaries, as well as my estimates of the primary vote for the larger parties.

I made a few minor tweaks to my redistribution code overnight. This caused some small changes to the margins, but I’ve updated the table in yesterday’s post.

By my estimate, 86.9% of voters remain in the same seat that they were in prior to the redistribution.

No changes were made to Bateman, Bunbury, Kingsley, Rockingham, South Perth and Southern River. While there was rather a lot of new area added to Kimberley, I estimate the new population makes up just 0.4% of the total.

The new seat of Mid West is primarily a successor to Moore. 76.5% of the enrolment in the new seat comes from Moore, with just 23.5% coming from North West Central. I think it’s fair to say that North West Central was abolished.

There are some dramatic changes in the outer suburbs of Perth which become obvious when you look at a list of what proportion of each new seat’s electors have come from another seat.

The renamed seat of Padbury is a successor to Carine, but only 51.7% of electors came from Carine. The neighbouring seat of Hillarys was also severely redrawn, only containing 53.2% of its prior population. This is easy to see on the map: Carine and Hillarys were previously seats that were roughly square-shaped, with Hillarys sitting to the north of Carine on the north coast of Perth, but they have been redrawn into two thin rectangles covering the same area. Judging by the comments on yesterday’s post, I suspect that change will be subject to challenges.

At the other end of the metropolitan region, the southern seat of Kwinana shifted south, and only 54.8% of its new enrolment was already in the seat.

The equivalent metric was just 69.6% in Jandakot, 71.7% in Bibra Lake (replacing Willagee) and 72.7% in Baldivis. All other seats had 75% or more.

Now I wanted to turn my focus to the political impact. Yesterday I pointed out that the newly created seat of Oakford is a Labor seat with a 28% margin, while the Nationals seat of North West Central was abolished. The super-marginal Labor seat of Churchlands flipped from Labor to Liberal, so the net number of Labor seats is steady at 53, while the Liberals have gained one seat net from the Nationals, so they each hold three.

But this is not particularly interesting, since the last election was so unusual. It’s more interesting to consider what would happen if there was a swing back towards a more competitive outcome.

Of course, a swing would not be uniform, but it’s the best we have.

For the Liberals and Nationals to gain the 24 seats necessary for a one-seat majority, they would have needed a uniform swing of 23.1%. Now that swing would be 23.4%. This suggests that the redistribution has slightly favoured Labor – this makes sense, considering the abolition of a Nationals seat and the creation of a safe Labor seat.

For what it’s worth, a swing of 23.4% would leave the Liberals and Nationals with 53.5% of the two-party-preferred vote, but an inconsistent swing could well give them a majority with quite a bit less.

Forrestfield was previously one of Labor’s top 29 seats, but has dropped out of the list with the redistribution cutting the Labor margin from 25.5% to 22.6%. Forrestfield has been replaced in that list by Oakford.

The biggest changes in margin were in Padbury and Hillarys. Labor won Carine by just 2.5%, but the margin in Padbury is now 13.6%. Meanwhile Labor won Hillarys by 19%, but the margin is now just 8.9%.

Labor also improved their position in Kwinana by 3.3% and Kalamunda by 2.5%.

Overall the redistribution is not particularly dramatic. It is simply the next chapter in the long story of urban population growth outstripping rural growth and forcing the abolition of rural seats to ensure fair representation of city voters.

Up next I’ve got a map showing the old and new boundaries. You can toggle each set of boundaries on and off, as well as the names of each seat. And after the map I’ve posted my estimates of vote for Labor, Liberal and Nationals (combined), Greens and others on primary votes per seat.

I should also note that I am not currently planning to make my own KML Google Earth boundaries file. I have been making my own boundary map files since the late noughties and they are all posted here. I started making these because at the time I did a lot of my mapping in Google Earth, and it was hard to use proper shapefiles in GE. I also would use the methodology to make maps that were not available as digital boundary files.

But electoral commissions have gotten better at publishing shapefiles promptly and my technological practice has changed, so the need has gone away. They also take a long time and add a risk of error that isn’t there when using the proper data.

But using the official shapefile means I need to shrink the file a bit to make it easily readable. I think it’s been done very efficiently but in a few places you might notice little gaps between electoral boundaries on the map.

I might still keep making my own maps for obscure boundary files, such as new ward boundaries, but that time has mostly passed.

Seat Labor Lib/Nat Greens Others
Albany 49.3 31.0 5.3 14.4
Armadale 73.0 8.9 4.8 13.3
Balcatta 65.4 20.3 7.3 7.0
Baldivis 79.1 9.1 3.7 8.1
Bassendean 71.0 12.3 7.9 8.8
Bateman 45.8 37.0 6.9 10.3
Belmont 70.4 16.1 7 6.5
Bibra Lake 65.9 17.4 10.2 6.5
Bicton 56.3 30.3 9 4.4
Bunbury 60.5 23.2 4.8 11.5
Butler 77.2 13.6 4 5.2
Cannington 71.9 13.0 6.4 8.8
Central Wheatbelt 35.0 54.1 2.2 8.7
Churchlands 38.5 45.0 9.9 6.6
Cockburn 69.8 17.6 6.2 6.4
Collie-Preston 61.4 23.5 4.1 11.1
Cottesloe 29.0 45.6 12.9 12.6
Darling Range 56.5 28.0 4.4 11.2
Dawesville 57.0 33.7 3.2 6.0
Forrestfield 65.4 22.0 5 7.6
Fremantle 55.4 16.4 20.1 8.1
Geraldton 52.7 39.0 2.6 5.7
Girraween 72.4 10.3 5.6 11.7
Hillarys 51.6 37.1 6.1 5.2
Jandakot 60.9 26.3 5.7 7.0
Joondalup 66.5 19.2 4.6 9.7
Kalamunda 53.7 29.4 8.6 8.4
Kalgoorlie 52.1 36.5 2.2 9.1
Kimberley 53.9 26.7 15 4.4
Kingsley 60.5 30.2 5.6 3.7
Kwinana 79.2 7.6 5.4 7.8
Landsdale 69.0 20.1 4.2 6.7
Mandurah 67.6 21.5 3 7.8
Maylands 62.0 15.6 16.1 6.2
Mid West 33.9 55.0 2.8 8.3
Midland 66.8 18.3 6.7 8.3
Mindarie 71.1 17.6 5.6 5.7
Morley 69.3 16.6 6.4 7.7
Mount Lawley 59.0 23.4 11.1 6.5
Murray-Wellington 57.8 27.3 2.8 12.0
Nedlands 35.8 35.0 13.7 15.5
Oakford 70.7 16.8 4.3 8.2
Padbury 55.4 33.0 7.4 4.2
Perth 63.5 17.8 16.1 2.7
Pilbara 58.9 28.3 3.8 9.1
Riverton 52.2 32.6 7.4 7.8
Rockingham 82.8 9.8 3.2 4.3
Roe 28.8 57.4 3.7 10.1
Scarborough 49.2 35.1 10.2 5.5
Secret Harbour 75.3 15.2 3.6 5.9
South Perth 49.9 35.4 10.4 4.2
Southern River 76.0 11.7 3.6 8.8
Thornlie 74.1 13.9 4.9 7.1
Vasse 34.3 49.1 9.6 7.1
Victoria Park 63.7 16.2 12.6 7.5
Walyunga 70.9 18.3 4 6.8
Wanneroo 71.8 18.3 4.5 5.4
Warren-Blackwood 32.2 41.5 14.9 11.4
West Swan 78.7 11.3 3.8 6.3
Liked it? Take a second to support the Tally Room on Patreon!
Become a patron at Patreon!

21 COMMENTS

  1. In my objections I’m gonna suggest they do a little bit of housekeeping as I think it’s a good idea where practice to line up district boundaries with lga boundaries. And will definitely be objecting to the elongation of carine and Hillary’s. I also think they should keep with the tradition elsewhere of naming districts after the most populous lga

  2. @Ben Raue I like the kmz files you make better though because you can change the colours of the seats so you can represent parties holding each seat on an SVG map (I do this on Mapshaper). This particular shapefile however does not have any fill colours for the seats (so I can’t even change the colour because the colour value is not in the SVG).

  4. They’ll be hearing from me about the changes to Hillarys and Carine! I’ve lived in either the seat of Hillarys of Kingsley for 30 years, and know the area very well.

  5. I disagree with your margin on Churchlands, what are you basing it on? Pollbludger has it at 0.1% and after looking at the booths overlapping, I am not seeing how you have it 0.6% more liberal than bludgers margin. would love to be proven wrong.

  6. @daniel what we can agree on is that margin should be more at the election. That’s just a notional margin based on the previous election

  7. Well firstly my estimated Liberal margin in Churchlands is 0.5%, so if William’s is 0.1%, that’s only a difference of 0.4%. I wouldn’t worry too much about such small margin differences, particularly when a small area has been moved. We’re all just producing estimates. Effectively Churchlands is a dead heat, but a bit better for Liberal than it had been previously.

    I’ve moved Doubleview PS into Scarborough, City Beach PS into Churchlands from Cottesloe, and Osborne Park Voting Centre from Scarborough into Churchlands.

    Doubleview had 946 formal votes, and was 60.5% Labor, while City Beach was 1035 formal votes, and 65.2% Liberal. And Osborne Park was 60.7% Labor but was much smaller (173 formal votes).

    In my methodology I then transfer the appropriate number of special votes so the overall transfers are proportional to the population transfers between the seats. So in the case of Scarborough more special votes were transferred than Cottesloe because the border booth was smaller but that wasn’t matched by the actual transfer of population. I also weight the special votes transfer to match the bias in the ordinary votes. So if the booths moved to the other seat are much more pro-Labor than the rest of the original seat, I take more Labor votes in the special votes transferred.

    Overall there were three batches of votes moved:
    -Cottesloe to Churchlands – 1681 votes at 31.4% Labor 2PP. Which boosted Liberal by 627 votes.
    -Scarborough to Churchlands – 1872 votes at 60.4% Labor 2PP. Which boosted Labor by 389 votes.
    -Churchlands to Scarborough – 2242 votes at 59.5% Labor 2PP. Which boosted Liberal by 426 votes.

    Overall that’s a net boost to the Libs of 664 votes.

    The 2021 result was 12821-12413. A net result of 408 votes for Labor. Instead my estimate is 13,400-13,144, or a Liberal margin of 256 votes. 408 + 256 = 664. A margin of 50.48%.

  8. Good to see the proposed Secret Harbour straddling the Perth metropolitan boundary at the point where its at its most artificial. Despite the abolition of upper house regions, a lot of submissions were oddly intent on strictly adhering to the boundary, resulting in some strange suggestions for Dawesville.

  9. @david Walsh it was entirely possible to keep the boundary between the metropolitan area and the regions without odd suggestions as you can see in mine

  10. I’m with you Anton, I live here and I can’t see this sticking for the life of me, regardless of whether they’ve been named Marmion, Whitfords, Hillarys or Carine, the seats that cover this area have always maintained a similar shape, why abandon it now? I anticipate a long list of objections from everybody except WA Labor (who’ll almost certainly retain Padbury whilst lose what was the old seat of Carine).

    The new boundaries elsewhere are pretty funky, Oakford and Belmont are now ‘spare parts’ districts that seem to group suburbs with little in common (South Guildford with Kewdale, Causarina with Brookdale) because they don’t fit anywhere else, which makes me wonder, why must the electoral commission stick to the magic number of 59? Surely a slight expansion (or contraction) of one or two might be a just way of leveling this out.

  11. The electoral commission doesn’t have the power to increase the number of Legislative Assembly seats above 59 (or reduce them) as Parliament alone determines this.
    I don’t agree that the boundaries of Belmont lack community of interest; the seat is based squarely on the medium sized City of Belmont and wraps around Perth Airport on its northern and southern sides.

  13. Belmont is clearly not a spare parts district it covers the entirety of the City of Belmont + South Guildford and Hazelmere. The latter suburbs in an ideal world would be better suited being in the seat of Midland but 90%+ of the electorate is within the Belmont LGA.

  16. It looks like most of the original proposal still stands, with some minor differences (main one being the adjustment of Hillarys and Carine back to their usual configuration)

  21. John, I’ve considered it and I won’t be doing posts on the VIC and WA submissions. Other projects instead. You can keep discussing those redistributions on the previous posts.

Comments are closed.