Breaking up the Senate? Why it’s such a bad idea


TheĀ AustralianĀ reported on Sunday about an idea from Queensland LNP senator James McGrath that would see the Senate broken up so that senators represent “provinces” within each state. While details are scarce, this proposal would see the end of proportional representation in the Senate, would likely wipe out all minor parties and would see one-party rule in the Senate, largely replicating the results of the House of Representatives.

The idea is obviously self-serving and is very unlikely to go anywhere, but it’s such a bad idea that I think it’s worth explaining why it’s so bad. I also had a go at modelling what Senate elections would look like under such a system.

Exhausted votes in the Senate drop in 2019


Opponents of Senate voting reform in 2016 focused a lot of attention on the danger of votes exhausting – which happens when a voter hasn’t marked a preference for any of the remaining candidates.

The rate of exhausted votes was relatively low in 2016, but that didn’t stop exaggerated claims about exhausted votes being a problem before the 2019 election.

The voting system does make it much easier to exhaust your ballot, as a side-effect of making it much easier for voters to mark their own preferences rather than relying on the discredited system of group voting tickets. Yet this problem was significantly reduced by a policy of encouraging voters to number at least six boxes above the line or twelve boxes below the line (which was what AEC staff were meant to tell voters, was printed on ballots, and was advocated for on most how-to-votes).

So what happened in 2019? Exhausted vote rates went down nationally, although the rate did increase in two states.


The exhaustion rate depends on a number of factors: the number of candidates and groups (the more boxes on the ballot paper, the more you need to fill out to minimise the exhaustion risk), the number of seats to be elected, and the partisan balance at the conclusion of the count.

The ballot paper was much smaller in 2019, in part due to the half-Senate election. The rate of just-vote-1 ballots remained very low (although it went up slightly) with most people still marking 6 preferences on their Senate ballot.

There was a substantial increase in the exhaustion rate in Victoria and a small increase in South Australia. The rate dropped in the other four states, in particular in New South Wales and Western Australia.

While exhaust rates were consistently higher in bigger states, where a larger ballot meant that a standard six preferences would be less likely to ensure a vote that didn’t exhaust, it appears the trends may have something to do with who was standing at the final round (ie. candidates who were either elected without distributing their surplus, or were the last candidate to be excluded).

The Greens were still standing at the end of the count in NSW, Queensland and Western Australia, while Labor’s Lisa Singh was still standing in Tasmania. But there was no Labor or Greens candidate in the count at the end in Victoria and South Australia.

One Nation were still standing in five states at the end of the count. In South Australia the last three candidates were Liberal, United Australia and One Nation. A left-wing voter wouldn’t have had many options. In Victoria, the last three were Liberal, One Nation and Derryn Hinch.

All of this analysis is based on the proportion of the total vote which has exhausted by the end of the count. Many of those exhausted votes would have already helped elect someone before exhausting, so even fewer voters would have had their vote exhausted without contributing to the election of a senator.

The overall conclusion is similar to in 2016: most votes helped elect someone, or ended up with the last candidate to be eliminated. As long as voters mostly mark multiple preferences, exhaustion rates will stay low. It will also help if ballot papers continue to shrink as we move further away from the group voting tickets era. The new Senate system is working reasonably well to ensure that most votes count.

Below-the-line rates go up in 2019


This is the latest in my occasional series looking back at the final results of the 2019 federal election.

The 2019 federal election was the second election held under the new Senate voting system, which included changes to make it easier to vote below-the-line. The election saw the rate of below-the-line voting increase nationally, with particularly large increases in New South Wales and the ACT.

I previously discussed the reasons for the big surge in New South Wales in June. Namely the spike in support for Liberal candidate Jim Molan. I’ll return to the ACT in a future post.

This chart shows the rate by state, and below the fold I’ll include the same information in a table.

I’ve also included a map below the fold showing the below-the-line vote for every booth in the country.

Voting system change boosts Labor and Greens in Brisbane City


I posted back in March about the Queensland government’s proposed reforms to local government in that state. The reforms include a bunch of other changes to electoral finance and council procedures, but I focused on two proposed changes: introducing compulsory preferential voting for single-member elections, as well as introducing proportional representation with compulsory preferences for multi-member elections.

It appears that the latter point is not included in the legislation which was introduced earlier this year and was the subject of a parliamentary inquiry, with it being treated as “the subject of further consultation”. This solves the problem of voters being required to number an absurd number of boxes, but it means we’re stuck without any proportional representation in Queensland council elections for now.

The government is proceeding with the other key piece, which would switch from optional preferential voting (OPV) to compulsory preferential voting (CPV) for elections of mayors and for single-member wards (the map of which councils would be covered by this change is included in this post). This echoes the change made to the state electoral system prior to the 2017 state election.

This system change will most likely have the biggest impact in the City of Brisbane, Australia’s biggest and most important local council, where optional preferential voting has seen a lot of Labor and Greens votes exhaust rather than helping the other party, and where the LNP’s large council majority could be under threat.

NSW to raise council election costs to make private providers “competitive”


New South Wales council elections are due in September 2020, which means that local councils right now are having to decide who they will contract to run their election.

This may seem strange to people not familiar with NSW council elections. In most states, all council elections are run by the state electoral commission. Yet in New South Wales, local councils can choose to either use the NSW Electoral Commission (NSWEC) or a private contractor.

In this post I will run through some of the history of this quirk of NSW electoral administration, why I think it’s such a bad idea, and what looks set to happen soon.

The NSW Independent Pricing and Regulatory Tribunal (IPART) is due to hand down a final report, as early as today, in an inquiry into what price should be set by the NSWEC for conducting local government elections.

The IPART is considering a pricing structure which would increase the average cost by 62% compared to the last round of elections in 2016 and 2017. The IPART has justified this proposal in part because of a concern that private election providers have not been able to compete with the NSWEC under the current pricing structure.

But do we really want private election providers to be “competitive” with the NSWEC? Why would that be a good thing?

NT redistribution – changes to a handful of seats in second draft


I blogged about the draft boundaries for the Northern Territory redistribution before I took a break back in June. At the end of July, the NT Electoral Commission released a second draft (not a final map), which contained changes to the boundaries of five electorates. Because the changes between two seats were reasonably significant, this triggered another round of consultation, before a final map is released later this year.

I won’t go into depth about this map, because it’s mostly the same as the May map. There were changes in two areas.

There was a substantial exchange of population on the border of Katherine and Arnhem. The seat of Nelson also retracted slightly from the Darwin area, with this area moved into Karama and Fong Lim.

This map below shows the new draft in green, with the old boundaries in red and the first draft in blue. It’s zoomed in on Katherine by default.

This new map is now available for download as a KMZ file from the maps page.

Brisbane City – draft redistribution margins


The Electoral Commission of Queensland (ECQ) last Friday released the draft boundaries for the Brisbane City Council election, due in March 2020. I’ve now put together a map of the electoral boundaries, and I’ve also calculated margins in all 26 wards, as well as primary votes for the three main parties.

The changes have helped Labor in a couple of marginal LNP wards, but overall has not had a big impact on Labor’s prospects of gaining control of the council in 2020.

ACT redistribution finalised


Along with the Northern Territory, Western Australia and Brisbane City Council, the ACT has also been redrawing its electoral boundaries for the local Legislative Assembly, with the boundaries finalised in July.

In this post I’ll share a map showing the changes to the electoral boundaries, along with my estimates of the vote percentages for the bigger parties in each electorate before and after the redistribution.

I was close to publishing the draft electoral boundaries and finalising my vote estimates back in early June right before I went on parental leave, but while I was away from the site Elections ACT announced that the final map would not feature any changes compared to the draft map, so that work can apply to the final version.

All five electorates were modified, with two seats gaining territory, two losing territory, and the seat of Murrumbidgee gaining and losing territory.

The Belconnen-area electorate of Ginninderra took in some territory from the Gungahlin-area electorate of Yerrabi, but these two seats did not exchange any territory with the remainder of the electorates.

The other three electorates all effectively shifted slightly north. The Tuggeranong-area electorate of Brindabella expanded north to take in half of the suburb of Kambah from Murrumbidgee. Murrumbidgee then shifted north to take in Yarralumla and Deakin from the central electorate of Kurrajong. This brings Murrumbidgee right up to Capital Hill and Lake Burley Griffin.

My redistribution vote estimates are my first test of a new formula I’ve used for distributing special votes in a way which reflects the different voting trends in different parts of an electorate. I might go into more detail about how this works down the track.

Labor received a boost in three out of five electorates, doing particularly well by picking up almost 1.4% in Kurrajong.

The Greens boosted their support in both of their current electorates but experienced no change in Ginninderra, their best prospect for winning another seat.

The Liberals did particularly well in Murrumbidgee, while they were knocked back badly in Kurrajong.


You can download the Google Earth map file for the 2020 boundaries from my maps page, along with every ACT Legislative Assembly map dating back to the first electorates in 1995.

WA redistribution – draft boundaries map


The draft boundaries for the Western Australian state redistribution were released three weeks ago, but it has taken me some time to put together the map of the new boundaries, which are available for download now.

The commissioners implausibly managed to avoid moving an electorate from the country to the city despite a growing gap in enrolments.

Ten out of 59 seats were left with no changes. Most seats underwent small changes, with the most dramatic changes taking place in the north-east of Perth. The seats of Girrawheen and Mirrabooka were completely redrawn, with Girrawheen shifting south to take in much of Mirrabooka, and a new seat of Kingsway drawn in the northern half of Girrawheen. Kingsway is an unfortunate name for the new seat, being created right next door to the similarly-named Kingsley.

I haven’t yet done my own calculations about the new margins, but we can use William Bowe’s estimated margins, published at Poll Bludger.

William has one seat changing hands, with the Liberal seat of Hillarys, currently held by a 4.1% margin, turned into an effective dead heat with Labor just out in front. This seat was unusually close in 2017 thanks to the sitting Liberal MP running as an independent, and should be easily retained by the Liberal Party in 2021.

There are a number of marginal Labor seats where their position has been improved: from 5.8% to 8% in Balcatta, from 0.7% to 1.2% in Kingsley, from 2.5% to 4.9% in Burns Beach, from 1% to 2% in Jandakot, from 2.9% to 3.6% in Bicton, from 7.3% to 9.2% in Wanneroo, and from 1.4% to 2.3% in Murray Wellington. Labor’s margin in Joondalup has dropped from 0.6% to 0.1%.

You can toggle this map below to show the 2017 boundaries (red), the 2021 draft boundaries (green) or both.

You can download the Google Earth layers for both the Legislative Assembly and the Legislative Council from the maps page, along with WA electoral boundaries dating back to 2008.

The final electoral boundaries are due to be published by the end of November this year.

New dataset – upper house votes from South Australia


It may have taken about 500 days to be published, but the Electoral Commission of South Australia finally published the complete statistics report for the 2018 state election two weeks ago, on July 31. This report was the first place to include the upper house vote count broken down by seat and by polling place.

This has allowed me to finally complete my dataset for the 2018 South Australian election. The dataset includes upper house and lower house results (including two-candidate-preferred figures) at the seat and booth level, as well as candidate lists, a summary of the turnout by electorate and a list of polling places. This is all available in my data repository. ECSA has also published the upper house booth results here, but their data is much less flexible than mine, being broken into separate documents for each seat.

As a demonstration of what can be done with this data, I’ve put together the following map which shows the upper house primary vote by booth for the three biggest parties: Liberal, Labor and SA Best, as well as a colour-coded map showing which of these parties won the vote in each booth.

If you find this dataset useful, bear in mind that I also have the equivalent dataset for the 2014 election which is available to those who sign up as Patreon donors.