2019 Western Australian State Redistribution

The landscape for the next Western Australian election has been finalised this morning with the Electoral Boundaries Commission releasing the new boundaries that will apply at the next election.

With the state’s population growth having slowed since the height of the mining boom, the scale of the changes wrought by the redistribution are much smaller than those produced by the last re-draw in 2015.

Despite population growth being concentrated in Perth and the south-west, the Commission has not repeated its 2015 decision to abolish a rural seat and create a new district in Perth. This means that 38 of the 43 seats in Perth have an above average enrolment.

On paper the boundaries increase the McGowan government’s hold on office, increasing the uniform swing needed for a change of government.

This post was updated, 28 November, with more information and adjusted margins for Hillarys and Joondalup.
My publication on the redistribution for the WA Parliamentary Library is now available at this link.

(Details on all the new boundaries can be found at the WA Electoral Redistribution Commission’s website.)

The Labor Party won 41 seats at the 2017 election, reduced to 40 following the loss of the Darling Range by-election in June 2018. The swing for the McGowan government to lose its majority on the old boundaries had been 5.8%.

Labor’s most marginal seat, Joondalup, looks to have slipped on to the other side of the pendulum on the new boundaries. The new boundaries have boosted Labor’s position in a number of marginal seats the government needs to retain to hold office. (See notes on calculations later in this post to explain the change to the estimated margin for Joondalup.)

On the new boundaries, the uniform swing needed for a change of government increases from 5.8% to 7.9%.

However, there is always variability in swing from seat to seat. Labor achieved a 12.8% swing at the 2017 election, but nine of the seats Labor gained were achieved by swings of more than 16%, the biggest winning swing being 23% in Bunbury.

Some seats won by Labor in 2017 will be easier to defend in 2021 because the formerly well-known Liberal MPs were defeated.

Some Labor MPs defending marginal seats will need to develop a personal vote if they hope to withstand any swing back from Labor’s high point in 2017. Some Labor MPs who won normally Liberal seats in 2017 will find winning the second time round tougher.

Some of the key seat changes are as follows.

• Balcatta – Safe Labor voting territory in Westminster replaces a less favourable Labor area in Hamersley, boosting Labor MP David Michael’s margin from 5.8% to an estimated 8.0%. Balcatta is a traditional Labor seat that only fell to the Liberal Party for one term on the Barnett government’s landslide re-election in 2013.
• Bicton – A 10% Liberal seat when created by the 2015 redistribution, Bicton was won narrowly by Labor’s Lisa O’Malley at the 2017 election.  The addition of the last parts of  East Fremantle boost her margin from 2.9% to 3.6%.
• Burns Beach – a narrow Labor gain in 2017, the new boundaries for Burns Beach include strongly Labor voting Quinns Rocks transferred from Butler, while more evenly divided parts of Joondalup and Iluka have been moved into Joondalup. Labor MP Mark Folkard’s margin increases from 2.5% to an estimated 5.4%.
• Hillarys – loses strong Liberal voting territory in Sorrento to Carine, replaced by slightly Labor leaning parts of Beldon and Mullaloo. This cuts the margin of Liberal MP Peter Katsambanis from 4.1% to an estimated 0.4%.
• Jandakot – Parts of Liberal voting Leeming have been transferred to Riverton, boosting the margin of Labor MP Yaz Mubarakai from 1.0% to an estimated 1.8%. Jandakot was one of the seats that fell from the clouds to be won by Labor in 2017 after a swing of 19.3%.
• Joondalup – Collects Liberal voting territory from Burns Beach while losing Labor voting areas in Beldon and Mullaloo to Carine. These changes virtually wipe out the margin of Labor MP Emily Hamilton. Those changes are enough to turn the 0.6% margin of Labor MP Emily Hamilton into an even closer notional Liberal margin of 0.4%.
• Kingsley – The addition of Labor voting Hamersley from Balcatta increases the margin of Labor MP Jessica Stojkovski from 0.7% to 1.2%.
• Landsdale – previously known as Girrawheen, the Labor margin in this seat has been slashed from 16.7% to 9.6%, though it remains a safe Labor seat.
• Morley – A small boost for Labor’s Amber-Jade Sanderson from 11.4% to 12.3% as the electorate’s eastern boundary moves from the Tonkin Highway to Beechboro Road. A seat that Labor should not have lost to the Liberal Party in 2008.
• Swan Hills – A reduction in margin from 14.5% to 12.1% for Labor MP Jessica Shaw doesn’t look troublesome, but Swan Hills was won by Labor with a swing on 18% in 2017.
• Wanneroo – Labor MP Sabine Winton has her margin increased from 7.3% to 8.6%.

Notes on Hillarys and Jondalup

Update added 28 Nov 2019
William Bowe published estimated margins for all seats in July this year when the draft boundaries were released. The analysis can be found on his Pollbludger website. There were only the tiniest differences between the draft boundaries and the final boundaries released yesterday.

Bowe had slightly different margins for Hillarys and Joondalup, estimating that Hillarys slipped narrowly on to the Labor side of the electoral pendulum (my estimates left it as marginal Liberal), and Joondalup remained as a very marginal Labor seat (my estimate was that it became a marginal Liberal seat).

The differences are caused by how polling place catchment areas divided by new electoral boundaries are split between new electorates. I’ll explain how the differences came about, and why I’ve changed my estimate for both seats.

Hillarys – the new boundaries transfer the suburb of Sorrento to Carine. This area contains two polling places, Sorrento Primary School which recorded a 59.2% Liberal two-party preferred vote, and Sacred Heart College which recorded 71.7%. If you transfer both polling places and the related proportion of declaration votes, then the new Hillarys comes out with a tiny Labor majority. However, if you do the full transfer of both polling places, votes transferred comes out at 94% of the electors transferred, which is too high compared to 89% for Hillarys at the last election. For that reason I split the Sacred Heart booth as being closest to the new boundary, which is why my estimates had Hillarys as a Liberal seat. I’ve adjusted the way I divided the polling place and my estimated margin is reduced from 0.6% to 0.4%. Given the seat has a Liberal MP, I am entirely comfortable with this estimate. Small changes to the way you deal with a single polling place can have a significant impact on estimated margins.

Joondalup – The complexity with Joondalup is its new northern boundary. A part of Joondalup was transferred from Burns Beach to Joondalup, as well as the suburb of Iluka. However, there are no polling places in Iluka. There are two polling places on the Burns Beach side of the new boundary, at the Francis Jordan Catholic School in Currambine, and the Bramston Park Community Sporting Facility in Burns Beach. On my initial part transfer of those polling places, Joondalup was calculated as a marginal Liberal seat. Having reviewed my partial transfers, Joondalup ends up with a tiny Labor majority of 0.03%, putting the seat back on the Labor side of the pendulum, where William Bowe had originally estimated it to lie. A result of this change also cuts the Labor margin for Burns Beach from 6.1% to 5.4%.

This new calculation has what I think is a better split of the Burns Beach booths. It also leaves Jondalup on the Labor side of the pendulum, which has an advantage in trying to explain the next election. Seats that are classified by past votes rather than by the party status of the sitting MP are a major source of complaints concerning the ABC’s election results site. If having Joondalup classified as a Labor seat minimises the number of outraged emails about idiots at the ABC who don’t know how to classify seats and analyse elections, then it gets a tick of approval from me.

The above calculations show that all redistributed margins are estimates based on assumptions and calculations. I would say there is always around half-a-percent of fudge factor in redistribution estimates, and with secret ballots, there is always a limit to how accurate you can make estimates.

A Note on Classifying Seats

In the pendulum below, two seats need explanation.

Darling Range was won by the Labor Party at the 2017 election, but won by the Liberal Party at a June 2018 by-election. The Liberal margin from the by-election has been used.

In Geraldton, sitting MP Ian Blayney resigned from the Liberal Party and now sits as a National MP. His seat is shown as National held with the Liberal two-party preferred margin from 2017.

On my calculations, the new boundaries push on to the Liberal side of the political ledger. It is shown on the pendulum with its estimated Liberal margin but highlighted as having a Labor MP. Others have estimated Joondalup as a very marginal Labor seat based on different assumptions about splitting polling places and allocating declaration votes. See notes above on new calculations for Joondalup.

Estimating new margins after a redistribution requires assumptions to be made about splitting polling places and re-allocating postal, pre-poll and absent votes. Such calculations are an inexact science.

The calculations cannot take account of the popularity of sitting members. This is the first redistribution in some time that hasn’t pushed Albany on to the Liberal side of the electoral pendulum, and as all followers of Western Australian politics will know, current Labor MP Peter Watson has defied new margin calculations after every redistribution since he was first elected in 2001.

Electoral Pendulum for the 2021 Western Australian Election

Summary of Changes for all Districts