Embedded Files
In my free time, I'm often bored, so what's a better thing to do than occupy my time by using the 2022 Federal Election to speculate on proportional representation! Feel free to critique my speculation through email or mastodon. Or alternatively, use the creative commons license to improve upon my work.
Also worth noting, I know the limitations of simulating elections like how people might vote differently if presented a party list. I also recognise the constitutional challenges associated with some systems, but this is a product of me being bored and wanting to do something.
I've done different types of mixed member simulations, like:
- South African style elections (MMP where both the region and national lists are elected proportionally)
- New Zealand style MMP but adjusted for preferences
- Japan inspired or MMM (Parallel Voting)
- Scottish inspired or AMS (Additional Member)
General STV proportional representation at different levels for both Queensland* and Federal.
*The data I used for the Queensland parliament projections are from the 2022 Federal Election because data is easier to get as the AEC consolidates election results much better than ECQ, and Queensland has 30 divisions in the lower house, so it's easy to divide electorates and keep population proportional (except for one, my first experimentation where I use divisions resembling the Queensland Regions and malapportion regional and rural votes because I wanted live on the legacy of Bjelke-Peterson, jk)
As well as looking at where One Nation's votes where the highest.
I've consolidated a lot of my earlier work into a google doc, because they were saved in as text files, edited with notepad. There's a button below to take you do the document.
Queensland Regions (Mar 24)
Queensland Regions (Mar 24)
From google doc, to summarise: this was my first attempt, and I had essentially invented a really bad electoral system. I didn’t even use an excel spreadsheet or anything; I remember using the ABC website for the 2020 Queensland election results and because of the nature of using a double PR MMP system, the malapportionment doesn't really have an effect. The regions are just symbolic.
South African inspired MMP, federal (April 24)
South African inspired MMP, federal (April 24)
Also on google doc. Summary: creation of large regions (see google doc for more), as well as a national list vote to level out. Using the same 'multiplier system' of multiplying all results by 2.5 because I needed to calculate 250 seats.
QLD Federal Seats grouped into 3 for QLD parliament. (May 24)
QLD Federal Seats grouped into 3 for QLD parliament. (May 24)
On google doc. Wanting to improve on my first attempt of the Queensland parliament, and wanting to use the 2022 federal results and federal divisions, I decided to retry. In this case, I divided the federal seats into 3 to create 10 electoral districts (but in retrospect not happy with the divisions I mapped out), plus 20 seats from a state wide list. For some reason I was really interested in South Africa's electoral system (and the election was in the month I made this) and that's why I was probably obsessed with a system like this.
Also note that these elections are probably similar to the 2024 Queensland state election, except that Labor is under represented in place of the Greens, and the UAP would be split between Family First, One Nation & the LNP.
Visualisation not working? try: https://public.flourish.studio/visualisation/18920485/
AMS Simulation (July 2024)
AMS Simulation (July 2024)
The additional member system (AMS) of proportional representation is similar to MMP, but the party list doesn't dictate the proportion of seats in parliament, but rather adds seats to try to make the parliament more proportional. It is currently for the Scottish & London Assembly elections. The parallel voting method (MMM) which is used in Japan is included as an example as it is different again. Specifically that, the proportion of the parliament is not accounted for and the party list vote is not subject to the local vote. This theoretical election uses the 2022 Senate election and elects 200 MP's. In the visualisation below you can compare all 3 types of proportional representation: national proportional representation using STV/quota method (PR), additional member (AMS) and parallel (MMM).
The second visualisation provides detail of AMS 'topping up' states to be more proportional, e.g. under my model, Tasmania gets 7 seats, 5 electorates and 2 top up seats, with an extra seat going to the liberals and the greens picking up a seat.
Visualisations not working? try: https://public.flourish.studio/visualisation/18646274/ , https://public.flourish.studio/visualisation/18657132/
One Nation Popularity (Aug 24)
One Nation Popularity (Aug 24)
For some reason, I decided to look at the popularity of One Nation in each electorate in Australia, and made a scale, if you'd like to see the scale I used, download the excel workbook for specific details. Capricornia had the highest popular vote with 14.6% of first preferences, with 22.59% of the popular vote when eliminated. Kooyong had the lowest share of votes in an electorate with 0.72% voting for the party, and 0.8% when eliminated. New England is an outlier, being the only rural region below the average this is probably due to the fact that this is Barnaby Joyce's seat.
One Nation did run a candidate in Higgins or Kennedy.
QLD STV 15 Divisions (Sep 24)
QLD STV 15 Divisions (Sep 24)
On Google Doc: In this simulation I used the 2022 Federal Election to elect a Queensland parliament of 90 members. I did this by joining 2 federal electorates together to create a new district that would elect 6 members. Considering the 2024 Queensland election and since there is an absence of the UAP in this model, this is probably the most accurate PR model for the Queensland parliament.
Visualisations not working? try: https://public.flourish.studio/visualisation/22086768/ , https://public.flourish.studio/visualisation/19249055/
QLD STV 5 Divisions (Sep 24)
QLD STV 5 Divisions (Sep 24)
Instead of joining 2 electoral divisions, why not group them in groups of 6? This would create 5 main electoral regions, I called them: North & Outback Queensland; Agricultural Queensland; North Brisbane; Beachside; and Brisbane & Logan. Each region would elect 18 members to compose a parliament of 90 members in the Queensland parliament. Again, used the 2022 House of representatives federal election for reasons mentioned at the start. For the application of the 2024 Queensland election, UAP would likely go the LNP or ONP, depending on the region, or perhaps family first; Labor is under represented in place of the greens.
Visualisation not working? try: https://public.flourish.studio/story/2577467/
Ranked Choice MMP (Mar 25)
Ranked Choice MMP (Mar 25)
The German snap election was coming up when I made this and I was inspired by the MMP, and I found out that New Zealand Also used the system, so I decided to try an apply this the Australian way by using preferences! For this, I combined the 2022 Senate & Lower House results. For the national list, I decided that preferences would keep 'flowing' to parties until there's a point were all parties have reached above 5%.
In this simulation, I'm using New Zealand rules, so if a party wins an electorate or Aboriginal seat they are entitled to full representation. In this simulation there were 3 over hang seats 250, so 253 members were elected to the house. I also used STV to elect Aboriginal members to parliament and guessed that representation, I think this could be changed. Aboriginal representation would be a proportion of Aboriginal registered voters on the electoral role, I assumed all would vote ~500k chose to register on the roll, so they are entitled to 5 seats, but if only 300k enrolled on the Aboriginal list, there would be 3 divisions. Furthermore, everyone gets the same national list vote, there's no special 2nd vote (or party vote)
I grouped the Independents for the lower house and hypothetically imagined that they decided grouped into a 'teal' bloc to benefit them, I did not include David Pocock and I added him to list of Independents because a lot of the independent candidates that got votes that would influence the election were all 'teals,' but that ~60000 voters for Pocock would imitate the 60000 who didn't vote for a candidate supported by Climate200.
All preferences are assumed on educated guesses, with people who vote minor parties continually preference minor parties.
Visualisation not working? try: https://public.flourish.studio/visualisation/22093355/
Google Sites
Report abuse