2021 WA Election – Legislative Council Update

6 April – all six regions have been declared and I’ve included the updated results in the table below. There is more detail on the final result in each region at the ABC’s Legislative Council results page.

WA Legislative Council – Projected results

Region	ALP	LIB	NAT	GRN	OTH
East Metropolitan	4	1	..	..	1
North Metropolitan	4	2	..	..	..
South Metropolitan	4	1	..	1	..
Agricultural	3	1	2	..	..
Mining and Pastoral	4	1	..	..	1
South West	3	1	1	..	1
Council (36 seats)	22	7	3	1	3
Change	+8	-2	-1	-3	-2

The ‘Other’ seats are two Legalise Cannabis WA and one Daylight Saving Party. The parties that no longer have representation in the Legislative Council are Pauline Hanson’s One Nation (-2), the Liberal Democrats (-1), Shooters, Fishers and Farmers (-1) and the Western Australian Party (-1).

1 April – the three non-metropolitan regions have been decided. The Daylight Saving Party won in Mining and Pastoral despite polling only 98 votes. The last two seats in South West were won by Legalise Cannabis and the Nationals. More details at the ABC’s Legislative Council results page.

Update 30 March – all above the line votes have now been counted. The scanning of BTL votes and reconciliation of ballot papers marked above and below the line is currently underway. Button press for the distribution of preferences is expected this week. I’ve switched the ABC’s Legislative Council calculators to manual updates for the final stages of the count.

Update 25 March – the small amount of additional votes added in the last three days have done little to change the comments I posted here three days ago. The WAEC is proposing to press the button to distribute preferences in each region separately as counting is completed next week.

A quick wrap of the final lower house numbers before talking about the Legislative Council.

The result in all 59 lower house seats is now clear, Labor winning 53 seats, the Nationals four and Liberals two. At this stage the results are not “official”. A formal distribution of preferences will be undertaken in all seats this week, after which official declarations will be made.

The two closest counts are Churchlands and North West Central, both decided by around 300 votes. There would need to be a major counting error found during the distribution of preferences to bring either result into doubt. It is unlikely a re-count request would be granted in either seat if the lead of around 300 votes is confirmed.

Several seats remain in doubt in the Legislative Council, but one of those that isn’t is the most remarkable victory.

In Mining and Pastoral Region, Wilson Tucker of the Daylight Saving Party looks certain of election despite polling only 0.19% of the vote. His victory may guarantee the abolition of group voting tickets for the Legislative Council.

Using output from my ABC Legislative Council calculators, combined with modelling based on 2017 above and below the line rates, my best estimate of the composition of the new Legislative Council is –

WA Legislative Council – Projected results

Region	ALP	LIB	NAT	GRN	OTH	Dbtfl
East Metropolitan	4	1	..	..	..	1
North Metropolitan	4	2	..	..	..	..
South Metropolitan	4	1	..	1	..	..
Agricultural	3	1	2	..	..	..
Mining and Pastoral	4	1	..	..	1	..
South West	3	1	..	..	..	2
Council (36 seats)	22	7	2	1	1	3

Whatever the outcome of the final seats, the Labor Party will have a majority in the Legislative Council for the first time in the state’s history. It will also have a constitutional floor majority with at least 21 seats after supplying the Council President.

The final seats in doubt are –

• The last seat in East Metropolitan which remains a close contest between the fifth Labor candidate and the Legalise Cannabis candidate. For reasons I outline below, I favour Legalise Cannabis candidate Brian Walker to defeat the fifth Labor candidate Robert Green.
• The last two seats in South West Region are difficult to pick with a close contest between Labor’s fourth candidate John Mondy, lead National James Hayward and Legalise Cannabis candidate Sophia Moermond.

At the bottom of this post I have an extensive discussion of technical issues relating to the conduct of the count and how I have done my modelling. Those interested in the nitty gritty of my methods can skip to that discussion.

NOTE: All references in the following discussion to Legislative Council Calculator totals are based on totals reported on the morning of Monday 22 March. As the site updates automatically based on a feed of votes, the values mentioned in this post may not match future calculations.

East Metropolitan Region

Labor has polled 4.7 quotas and will elect four MLCs. Labor’s three current MLCs are re-elected, namely Alanna Clohesy, Samantha Rowe and Matthew Swinbourn. They will be joined by a fourth Labor member in Lorna Harper.

The Liberal Party has polled 0.98 quotas and Donna Faragher will be re-elected. The Greens are polling 0.38 quotas and MLC Tim Clifford will be defeated. So will Charles Smith, the WA Party (ex-One Nation) MLC.

However, a race early in the count to determine which of Legalise Cannabis and the Western Australia Party are excluded first, will determines whether Green preferences elect Labor’s Robert Green to the final seat as a fifth Labor MLC, or whether Green preferences elect Legalise Cannabis candidate Brian Walker to the final seat.

If you look at the ABC calculator output for East Metropolitan, at a point labelled Count 20, Walker is excluded. At that point he has 0.2200 quotas compared to 0.2258 for Charles Smith of the Western Australian Party. (The numbers do change slightly with each update of votes.) So Walker is excluded, his preferences flow to the WA Party and leave the Greens in last place, and Green preferences go on to elect the fifth Labor candidate.

However, if the WA Party trails Legalise Cannabis at this point, WA Party preferences are distributed to Legalise Cannabis, and then the last placed Green is excluded and preferences flow to Legalise Cannabis and Brian Walker eventually wins the last seat.

At the start of the count, Legalise Cannabis has 0.1734 quotas, the WA Party well behind on only 0.0558 quotas. By the key choke point where the two parties are last and second last, Legalise Cannabis has received one ticket of preferences from Animal Justice who began the count with 0.0466 quotas. The WA Party relies on ticket preferences from eight different parties to reach the choke point, from Independents Doan and Larsen, and from the Great Australian Party, Daylight Saving Party, Sustainable Australia, Health Australia, Liberal Democrats and Liberals for Climate. The WA Party was the beneficiary in East Metropolitan Region of the micro-party alliance preference deal arranged by preference ‘whisperer’ Glenn Druery.

To reach the choke point, Legalise Cannabis risks leakage of only Animal Justice BTL preferences while the WA Party is wholly reliant on BTL votes having the same preferences as ATL ticket votes. Given a 5-10% rate of BTL voting for all the minor parties, I expect that the WA Party will fall short of Legalise Cannabis at this critical juncture, making it more likely that Brian Walker of Legalise Cannabis will win the final seat on Green preferences ahead of Labor’s Robert Green. But for such a complex count, it is best to leave the final seat in doubt.

North Metropolitan Region

The result in North Metropolitan is clear cut. On ATL votes Labor has polled 4.18 quotas and the Liberal Party 1.65 and the Greens 0.48. Once BTL votes are included in the count, the Labor vote will drop slightly but still be above four quotas, the Liberal vote will drop slightly and the Greens rise slightly. Preferences from a collection of right-of-centre parties will be enough to allow the Liberal Party to reach a second quota.

Labor’s Martin Pritchard is re-elected, Pierre Yang is elected after transferring from South Metropolitan Region, and Ayor Makur Chuot (elected as the first Australian MP with South Sudanese background) and Daniel Caddy are the newly elected MLCs. Labor MLC Alannah MacTiernan moved to and was elected for South West Region. The re-elected Liberals are Peter Collier and Tjorn Sibma, with Michael Mischin defeated. (Mischin had been demoted to the unwinnable fifth position on the Liberal ticket.) Greens MLC Alison Xamon will also be defeated.

South Metropolitan Region

Labor will elect four MLCs, the Liberals one with the final seat in my view to be won by the Greens. This differs from the current ABC calculator output for South Metroplitan Region. At what the calculator calls Count 26, the Greens trail Labor very narrowly 0.4646 quotas to 0.4706. The calculator excludes the Greens and Labor’s fifth candidate goes on to win the final seat.

However, the calculator does not take any account of BTL votes. Once they are included, Labor’s vote will fall slightly and the Green vote rise. At the 2017 election, only 2.35% of Labor’s votes were BTL compared to 11.13% for the Greens. At the start of the South Metropolitan count, Labor has 4.4706 quotas to the Green’s 0.4500. Modelling percentages for all parties based on 2017 ATL/BTL rates, I estimate that a better starting count would be Labor 4.4002 and the Greens 0.4858. That means at the equivalent of Count 26, the Greens should lead Labor and go on to win the final seat on Labor preferences. That may be reversed if Labor attracts a lot of below the line votes, but I think the Green projected lead will be enough.

The Labor MLCs re-elected are Sue Ellery and Kate Doust, joined by newcomers Klara Andric and Stephen Pratt. Sitting MLC Pierre Yang moved to and was elected to represent North Metropolitan Region. Liberal MLC Nick Goiran is re-elected while fellow MLC Simon O’Brien did not re-contest after losing Liberal pre-election.

Liberal Democrat MLC Aaron Stonehouse was defeated, his party’s vote declining from 3.9% to 0.9% after losing the advantage of being to the left of the Liberal Party on the ballot paper. He will be replaced by the Green’s Brad Pettitt, or if all my predictions prove to be wrong, by a fifth Labor MLC in Victoria Helps.

Agricultural Region

On first preferences Labor has polled 3.20 quotas, the Nationals 1.88 quotas, Liberals 0.87 quotas and the Shooters, Fishers and Farmers fourth on 0.28 quotas. Preferences do give a huge boost to the Shooters, Fishers and Farmers, and result in the election of the first Liberal ahead of the second National, but the SFF rely totally on ticket votes to catch the second National and that will be unlikely once BTL votes are included.

The 2017 election was the first time Labor had ever elected two MLCs from Agricultural region, so 2021 is the first time the party has elected three. Labor’s Darren West is re-elected, joined by newcomers Shelley Payne and Sandra Car. Labor’s Laurie Graham retired. Nationals Colin De Grussa and Martin Aldridge are re-elected. The new Liberal MLC is Steve Martin, replacing Jim Chown who was defeated after demotion to third on the Liberal ticket. The result will be the loss of a seat for the Shooters, Fishers and Farmers. Sitting MLC Rick Mazza did not re-contest after choosing to contest South West Region where he was defeated.

Mining and Pastoral Region

The initial quotas in Mining and Pastoral Region are Labor 4.03 quotas, Liberal 0.76, National 0.73, Greens 0.33 and SFF 0.25. But the boil over in Mining and Pastoral is that the Daylight Saving Party’s Wilson Tucker is certain of election despite his party polling just 0.19% of the vote, 0.0131 quotas.

The Daylight Saving Party was the anointed beneficiary in Mining and Pastoral Region of the micro-party alliance preference deals organised by preference ‘whisperer’ Glenn Druery. Knowing Druery as I do, he will be proud of pulling off victory from the lowest ever first preference vote. Given Mining and Pastoral Region has previously voted solidly against ever adopting daylight saving, the election of a party supporting that cause from such a low base will surely drive a nail into the coffin of group voting tickets in Western Australia.

The party starts with a very low count, but gets the preferences of every party and independent polling lower. It then leap frogs and collects preferences from other low polling parties, than larger parties, and reaches a quota to win the fifth seat after passing both the Liberal and National Party candidates who began the count with more than 50 times as many votes. Having looked at the count in detail, I can’t see how Tucker does not win the fifth seat.

At the end of the count, Tucker is elected on Shooters, Fishers and Farmers ‘s surplus preferences, One Nation and Australian Christian preferences then delivering the final seat to the Liberals over the Nationals.

Even if Labor’s vote falls just short of four quotas once BTL votes are added, Labor will reach a fourth quota during the count without impact on who wins the other two seats. Labor’s Stephen Dawson and Kyle McGinn are re-elected, joined by newcomers Peter Foster and Rosetta Sahanna. Wilson Tucker is elected for the Daylight Saving Party, and Liberal Neil Thomson is the new Liberal MLC. One Nation’s Robin Scott is defeated. There were three retiring MLCs in 2021, Liberal Ken Baston, National Jacqui Boydell and Green Robin Chapple.

South West Region

South West is complicated because while Labor has polled just short of four quotas, its only real chance of winning election is on the leakage of below the line votes. There are no group voting ticket preferences flowing its way unless the Legalise Cannabis Party is excluded earlier than expected, which would free Green preferences to elect a fourth Labor MLC.

On the ABC’s calculator output, Labor starts the count with 3.9641 quotas, the Liberals 1.3062, Nationals 0.4172, Greens 0.3948, SFF 0.1705 and Legalise Cannabis 0.1473. At the count labelled 23 in the Calculator output, the Greens are excluded, then the SFF, at which point Sophia Moermond of Legalise Cannabis and National James Hayward win the final seats.

Taking account of ATL/BTL rates in 2017, I think Labor is more likely to start the count with around 3.9000 quotas which will make it much harder for Labor to reach the fourth quota but it is possible. If BTL votes reduce the flow of preferences to Legalise Cannabis resulting in it being excluded before both the Greens and Shooters, Fishers and Framers, then the last two seats change totally and Labor and the Shooters, Fishers and Farmers win the final two seats.

Certain of election are re-elected Labor members Sally Talbot, Alannah MacTiernan (transferring from Northern Metropolitan) joined by newcomer Jackie Jarvis. Labor’s Adele Farina did not win Labor pre-selection. Liberal Steve Thomas is re-elected, and Green Diane Evers and One Nation’s Colin Tincknell will be defeated.

If Labor wins a fourth seat it will go to newcomer John Mondy. National Colin Holt retired, so any National seat would be won by newcomer James Hayward. The possible Legalise Cannabis winner is Sophia Moermond, and if the SFF win, the seat will go to Rick Mazza, who chose to contest South West rather than his existing Agricultural Region seat.

Details on the Counting Process and Calculator Modelling

As at the close of counting on 21 March, the count in the Legislative Assembly had reached 85.0% of enrolment, close to final figures. The Legislative Council is at 81.3%. The WAEC is currently tallying only above the line (ATL) vote by manual counting. All below the line (BTL) votes were batched up by counting centre and sent for scanning and data entry and they are not included in the current count. So 3.7% represents the number of BTL ballot papers as a percentage of enrolment, which corresponds to about 4.6% of ballot papers, the same rate of BTL voting as in 2017. The currently reported rate of LC informal voting is a very low 2.2%, but this will increase as the scanning and data entry of BTL ballot papers begins to find informal votes.

Given almost all LC ATL votes are accounted for, further counting is not going to change the current LC percentage votes by party. Differences will come about by the inclusion of BTL votes, and predictions of the final outcome will be influenced the rate of leakage away from party tickets during the distribution of preferences.

Much of the analysis that follows is based on the output of my ABC Legislative Council calculators. I always get complaints about the calculator, that it doesn’t declare certain seats as too close to call, which means its prediction on single seats can flip-flop as the count progresses. It also makes its prediction based on assuming all votes are ATL votes and this is criticised as unrealistic. However, without the calculator it would be impossible to assess who is winning the final seat in each region. Much of the analysis of LC results on various blog sites is based on arguing variations from the calculator output, not on independent re-calculation of the preferences.

The most important point to understand comparing the calculator in WA with how it has worked elsewhere is that it ONLY includes ATL votes. Elsewhere, where BTL votes are in the initial count, the debate is all about how BTL votes will deviate from the ticket. In Western Australia, as only ATL votes are counted, two errors have to be accounted for in assessing the calculator. These are variations in primary vote once BTLs are added, and then leakage of BTLs away from the ticket.

To get around this problem, I have compared % of ticket votes by region and party in 2017 with % total per group by region by party to calculate adjustment factors which I apply to the current party percentages for 2021. For the major parties this lowers the 2021 party % slightly, while for other parties it inflates the percentage.

Using this method, the only region where it appears to have a significant impact is South Metropolitan Region. Taking into account the variation in BTL voting rates, it is much more likely that the Greens will win the final seat in the region rather than Labor. It is for this reason I have suggested the Greens will win the final seat despite the LC Calculator output suggesting the seat will be won by a fifth Labor candidate.

For other regions the main issue will be the traditional one, whether BTL votes will drift away from the party’s lodged ticket. As described in my analysis of East Metropolitan Region above, deviation of BTL votes from the party ticket improve the chances of Legalise Cannabis defeating the fifth Labor candidate, though the seat remains rightly in doubt.

The last two seats in South West Region could be impacted by leakage from ticket, but the current calculator output looks the most likely outcome unless Labor attracts enough BTL preferences to reach a fourth quota.

Some Seriously Technical detail about the conduct of the WA Legsialtive Council Count

The electoral systems for the Senate and for the WA Legislative Council are subtly different variants of Proportional Representation by Single Transferrable Vote (PR-STV). There are two very technical points concerning how the count is different in WA compared to Senate elections.

To explain the difference, you have to know the difference between a ballot paper and a vote. All ballot papers have what’s called a transfer value attached to them. A Vote is equal to a ballot paper times its transfer value. Ballot papers with equal transfer value are bundled together as a total of ballot papers, and the number of votes equals the number of ballot papers multiplied by their transfer value.

All first preference ballot papers have transfer value ‘1’, as do lower house votes, so the number of votes equals the number of ballot papers. But when a candidate reaches a quota, a number of votes equal to the surplus are distributed as preferences. This is done by distributing all ballot papers at a reduced transfer value so that the correct number of votes are distributed. A simple example is that if a candidate has 1.25 quotas of ballot papers, all ballot papers are transferred at (0.25/1.25)=0.20 transfer value. So 100 ballot papers transferred to another candidate as preferences are equal to (100*0.2) = 20 votes.

This is obvious for a candidate elected with a surplus on first preferences, but is more complex in cases where a candidate reaches a quota in the middle of the count. What votes do you look at to determine the surplus and calculate the transfer value?

There are two main methods for doing this in Australia. The ‘last bundle’ or Gregory method, again with minor variants, is used for the NSW Legislative Council and for Hare-Clark in the ACT and Tasmania. As the name suggests, only the last transferred bundle that put a candidate over quota is examined for preferences and used in determining the transfer value. All votes already with the candidate before they reach a quota are left with the elected candidate and not examined for further preferences.

In 1984 the Senate system was changed to use what is called the ‘Inclusive Gregory Method’ where all ballot papers with a candidate at the point where they are declared elected are examined. The one problem with the Senate method is that the calculation of the transfer value is based on the number of ballot papers rather than the number of votes which can give a strange distortion to the further flow of preferences.

Western Australia changed its system in 2005 to adopt the ‘Weighted Inclusive Gregory Method’ where the calculation of Transfer Value is done based on votes not ballot paper. The transfer of preferences therefore reflects the votes that elected the candidate, not the ballot papers. The ABC’s calculator uses the weighted method for Western Australia.

Then there is the issue of the bundling of votes when a candidate is excluded and their votes distributed. Again WA differs from the system used for the Senate.

In the Senate, when a candidate is excluded, all votes with equal transfer value are distributed together whatever their source. The ballot papers with the highest transfer value (usually 1) are distributed first, followed by other ballot papers in order of descending transfer value.

This means all first preference ballot papers with a candidate, whether for the candidate or transferred to the candidate, are distributed as a single count. Then all ballot papers at the next lower transfer value, and so on. It means that when the second candidate on a party ticket is excluded, their first preferences are distributed before the much larger number of ballot papers transferred from the lead candidate on the ticket at reduced transfer value. Each exclusion of a candidate corresponds to several Counts in the official distribution of preferences, each count corresponding to ballot papers of the same transfer value being distributed.

This is not method used in Western Australia. In WA, ballot papers are distributed from an excluded candidate in small bundles corresponding to the order they were received. So a candidate’s first preferences are distributed first. Then individual counts are conducted for every bundle of votes received by the excluded candidate in the order each bundle was originally received, and irrespective of the transfer value. It means that every exclusion in a WA count may consist of tens or even hundreds of individual counts.

Where this matters is that under PR-STV, after every ‘count’, a test is done to see if any candidate has reached a quota. If they have, that candidate can receive no further preferences, and the exclusion of the candidate whose votes are being distributed continues.

In Senate counts with larger transfer bundles, this means that multiple preference tickets may end up in a candidate’s surplus and be mixed in with the candidate’s surplus to quota preferences.

In the WA Legislative Council system, only a single bundle is transferred so there can be only one ticket creating the surplus. Subsequent bundles with the same transfer value that at a Senate count would end up in the surplus, in WA end up outside of the surplus and skip the provisionally elected candidate and go on to the next available preference. These tickets end up distributing at their original transfer value rather than being washed through the surplus of an elected candidate. By having smaller bundles, the WA system means there are smaller surpluses when candidates pass a quota, and more votes from other bundles are distributed at full value.

All this may sound like unnecessary detail, but it mattered in 2017. As explained (with a few typos) here the calculator in 2017 got the final seat in East Metropolitan Region wrong because it used the Senate’s bundling method. When I wrote the specification for the original calculator in 2007, I included individual bundling as one of the methods. In 2017 I was not aware the method was in the system. In 2020 the software was re-written with a new interface, which is when I discovered the correct bundling method was available.

After an argument with the technical people who couldn’t understand why I wanted to use an option that hadn’t been used before, the 2021 calculator uses the correct bundling method.

I’m not sure this will make any difference in 2021, but if you look closely at the calculator output, you will see that preferences are being distributed in order of receipt, and that candidates can be declared electing in the middle of a distribution producing a change to further preferences.

The difference this can make depends on the size of each transfer. In a really close contest, the imprecision created by assuming all votes are ATL ticket votes means the calculator can get the wrong final seat winner even with the correct settings.

In the end the calculator is a good approximation to the actual count, and the best available for trying to analyse the result. But the output needs to looked at closely if you want to make sure that results are not being predicted with greater certainty than is justified.