That’s it for 2020


Now that I’ve finished posting about all of the redistributions happening at the moment, I’ll be bringing down the shutters on this blog for the next month or so.

Be assured a lot will be going on in the background as I prepare my guide to the Western Australian state election (due in March 2021) and work on some new datasets based on the elections of 2020. I’m also making plans to cover the September 2021 local government elections in New South Wales, and getting ready for the next federal election.

I will also be planning to bring back the podcast in early 2021. Unfortunately I ran out of time to make one more episode to wrap up 2020 as the New South Wales and South Australian redistributions took up a lot more of my time than I had planned.

If you find this website useful, please consider signing up as a donor via Patreon. I can only do this website thanks to the regular and predictable support of my Patreon donors, and more donors will give me more time to do more here.

Victorian state redistribution kicks off


Just as the South Australian redistribution finishes, and alongside a state redistribution in New South Wales and two federal redistributions, the starters pistol has been fired in the Victorian state redistribution.

The redistribution will be conducted based on the enrolment figures as of November 30, and each electorate must be drawn to be within 10% of the average based on that data. That average is 48,625 voters.

28 of the 88 electorates currently sit outside of that quota, which reflects that the boundaries were last drawn in 2013 and thus have had plenty of time to be pushed out of balance while Victoria has boomed in population. Some seats are way over quota, with Cranbourne and Bass more than 40% over the quota, while a bunch of other seats have experienced very little population growth and have fallen behind. Rowville, Forest Hill and Mount Waverley are all at least 18% below quota.

There is a clear geographic trend that can be clearly seen on the following map:

The seats in the eastern and southern suburbs of Melbourne are all well below quota, while the south-eastern fringe and most seats in the north and west of the city are above quota. Meanwhile there are a bunch of seats in regional Victoria which are relatively close to the quota, while others have fallen behind.

Another way to look at the trends is to look at the 8 Legislative Council regions, each of which includes 11 electorates.

SA redistribution – final map and margins


The final boundaries for the South Australian state electoral redistribution were released in November, and I have just finished producing my new map of the electoral boundaries, as well as my estimates of the margins.

The map is shown below:

The boundary changes were slightly less dramatic than the draft boundaries, with boundaries reverting to the 2018 map in a number of places. Overall 18.1% of voters have been moved to a new seat, down from 19.7% on the draft boundaries.

I have produced estimates of the margin in each seat. No seats changed hands in the redistribution, but there were some dramatic changes.

NSW redistribution – draft map and margins


It’s taken some time, but I have now finalised my map of the New South Wales draft electoral boundaries for the 2023 and 2027 state elections.

The boundaries were released in early November and I blogged about them at the time, but it’s taken me until now to finish my Google Earth map, which has allowed me to produce the interactive map below.

Green lines represent the 2023 boundaries, while red lines represent the 2015-2019 boundaries. You can toggle each layer on and off.

I’ve published the full list of seat margin estimates below the fold, and you can download the Google Earth map from my maps page.

Most of the changes took place in the Sydney region, with seats outside that area changed less dramatically. The seat of Bankstown was effectively abolished, with the name transferred to the neighbouring seat of Lakemba, so the name “Lakemba” has been retired. This seat was replaced with Leppington in the south-west of Sydney.

About 561,000 enrolled voters, out of a total of 5.3 million, have been shifted into a new seat. That’s 10.55% of the total electorate. 32 out of 93 electorates either lost no voters or a tiny number of voters. 27 electorates did not gain voters. 14 seats had no significant change.

I estimate that seven seats changed their name. It’s worth mentioning Castle Hill: while the seat has retained its name I believe that Kellyville is a successor to the old seat of Castle Hill and the name Castle Hill is now applied to a seat which has primarily replaced Baulkham Hills. Both seats were dramatically redrawn.

The seats that have changed the most by margin are Auburn and Parramatta. Auburn has become 4.3% better for Labor, thanks to the removal of the suburbs around Sydney Olympic Park and the addition of South Granville. Parramatta then becomes 4.3% worse for its Liberal incumbent Geoff Lee, with the inclusion of those areas around Olympic Park and the loss of the northern end of his seat to Epping. Lee’s margin is cut from a comfortable 10.6% to a much tighter 6.3%.

Conveniently the NSW Electoral Commission publishes two-candidate-preferred counts at the booth level for all match-ups in each seat, which means I have to do a lot less guesswork in seats where the Greens made the top two. But there was still four seats which required some estimates.

In Cabramatta and Dubbo, where independents came second in 2019, I simply ignored the extra areas added to each seat. In Sydney, I chose to add in the Greens vs Coalition counts for the areas added from Heffron, Newtown and Port Macquarie as a proxy for Alex Greenwich, which increased his margin by 0.2% compared to what it would look like if I ignored those areas. A substantial part of Newtown around Surry Hills was moved into Sydney and I suspect those voters will be quite supportive of Greenwich, so it was worth trying to factor in that change.

The seat of Murray, won in 2019 by the Shooters, Fishers and Farmers, has gained the former Jerilderie council area from Albury, where the Shooters did not run. I have ignored those voters and thus the margin is unchanged.

NZ 2020 – voting data shows how voters split their votes


New Zealand voters cast two votes, but (unlike in Australian federal elections) they are cast on a single piece of paper. This means it is possible to determine exactly how many voters who voted for each party on the list voted for a particular candidate in the electorate. This data was published for the 2020 election yesterday.

68% of all voters cast both votes for the same party. This ratio varied tremendously between the major parties and smaller parties. 87.6% of National party voters voted for a National electorate candidate, and 78% of Labour voters did the same. In contrast, only 29% of Green party voters and 21% of ACT party voters cast their electorate vote for the same party. Overall 81% of major party voters and 29% of other voters cast both their votes for the same party.

65% of ACT party voters cast their electorate vote for National, and 57% of Green party voters cast their electorate vote for Labour.

There are two seats I’m particularly interested in: the neighbouring inner-Auckland seats of Auckland Central and Epsom. Green MP Chloe Swarbrick won Auckland Central for the first time in 2020, while Epsom has been won by ACT MPs since 2005.

The National Party has traditionally given the go-ahead for their local voters in Epsom to defect and vote for ACT to ensure a partner for National in parliament. Jacinda Ardern was less than full-throated in her endorsement of the local Labour candidate in Auckland Central in 2020, but Labour didn’t sit out the race like National tries to do in Epsom.

I looked at the split vote statistics in Epsom in the lead-up to the 2014 election.

Back in 2011, a large number of Green and Labour voters cast their electorate vote for National, a tactic which could have potentially elected the National candidate, but they were thwarted thanks to the large numbers of National voters choosing to vote for the ACT candidate instead.

Labour and Green voters appear to have given up on pursuing this strategy. Just 16% of Green voters and 7% of Labour voters cast their electorate vote for National list MP Paul Goldsmith.

Almost 70% of National party voters in Epsom still cast their vote for ACT MP David Seymour, and they make up over half of his total vote. Almost 11,000 National party voters cast a ballot for Seymour, compared to just over 4,000 ACT party voters and 3469 Labour party voters.

Swarbrick also relied heavily on Labour party voters to win Auckland Central, although Labour’s voters were less emphatic in supporting her, which reflects Labour running a serious campaign to win this National-held seat.

34% of Labour’s party voters cast their electorate vote for Swarbrick, compared to 58% for Labour’s Helen White. Green party voters split 81-16 for Swarbrick over White.

Despite a majority of Labour voters sticking with White, Labour party voters still contributed more to Swarbrick’s total vote than Green party voters. She polled about 5754 votes from Labour party voters, compared to 5637 from Green party voters.

Meanwhile National candidate Emma Mellow-Sandford’s vote followed a similar pattern to National candidates across the country. She gained 88% of the National party vote, along with 72% of the ACT party vote, along with a handful of Labour voters.

9 votes in Bundaberg wraps up Queensland


The final results of the Queensland state election were decided yesterday, with Labor winning a recount in the seat of Bundaberg by nine votes, and won a recount in Nicklin by 85 votes.

The original preference count in Bundaberg gave Labor victory by eleven votes, and gave them victory in Nicklin in 79 votes.

These two results give Labor 52 seats, four more than they held in the last parliament. Labor gained five seats from the LNP (Bundaberg, Caloundra, Hervey Bay, Nicklin, Pumicestone) and lost South Brisbane to the Greens.

That’s about it for now for the Queensland state election, and I wanted to just flag that the blog (and the podcast) will probably be quiet until early in December.

I am currently working on my map of the New South Wales redistribution, and I’m also busy with some other work projects that will occupy me for the rest of November. The NSW redistribution’s timing has meant some other things have been bumped into the next month.

I do still want to do at least one more podcast this year, but it won’t be until early December. If you’ve got suggestions about what we should cover, let me know.

I’m also planning to complete my datasets for the ACT and Queensland elections, and will return with some more analysis of the results when they are finished.

Then once those are all finished, I’ll go dark until January or February next year when I’ll return with a guide to the WA state election.

NSW state redistribution – draft maps released


The New South Wales state redistribution took another step forward today, when the draft boundaries were released. The maps and report can be found here.

It will be some time before I finish my own boundaries file, which I will then use to calculate my own margins. I also haven’t seen any other estimates, so for now I will keep this post broad, and readers can discuss what you find in the maps in the comments.

I will return with margin estimates and an interactive map once they are ready.

The biggest news is the abolition of the seat of Lakemba and the creation of a new seat of Leppington in the fast-growing south-western corner of Sydney.

While the name “Lakemba” has been abolished, it looks to me that the seat of Bankstown has actually been divided up between its neighbours, with Lakemba being renamed as Bankstown while maintaining a majority of its existing population.

ACT 2020 – how people voted


It’s now been over three weeks since the ACT election, and I have pulled together the statistics on how the Covid-19 pandemic affected how people choose to vote.

There was a massive surge in pre-poll voting, necessitating a collapse in the election day vote. There was also a more modest increase in postal voting, on top of a record high postal vote in 2016.

This chart shows the share of the vote cast as an ordinary vote on election day, as a pre-poll vote, a postal vote, or other methods. I’ve calculated these figures as far back as the 2001 election – it was not possible to calculate these same statistics for 1995 and 1998 based on the data I could find on the Elections ACT website (let me know if I’ve missed it).

The trend was already heading in this direction before this year. Ordinary voting had declined from 83.8% in 2001 to 59.7% in 2016. Pre-poll voting had climbed from 12.4% to 33.4% over the same period. Postal voting was around 3% in the early 2000s, increasing to over 4% in 2008 and 2012, and then hitting a record level of 5.2% in 2016.

These figures were completely flipped on its head in 2020. 69.3% of votes were cast in person at pre-poll, with another 6.3% cast as a postal vote. That’s at least 75.6% of all votes cast before election day.

I’ve also pulled together statistics on turnout and informal rates, dating back to the first Hare-Clark election in 1995.

The informal rate has steadily been declining since the first Hare-Clark election. That first election saw an informal rate above 6%, but it dropped right down to 4.2% in 1998 and, apart from a much lower rate in 2004, it has steadily dropped since then.

Yet this year’s informal rate is remarkably low, falling to just 1.4%.

The likely explanation for this low rate is due to people voting electronically at pre-poll. Electronic voting has been used in the ACT since 2001, and it has been the default method of casting a pre-poll vote since at least 2008. The big increase in electronic voting has likely led to a big drop in informal voting.

I haven’t yet compiled the ACT 2020 data for my data repository. When I do so that will allow me to calculate the informal rates for each method of voting (and also calculate the partisan voting trends by method of voting), but I have looked at the breakdown in 2016. The informal rate for pre-poll voters was 1.9%, compared to 2.9% for all other votes.

Finally, I have also taken the opportunity to calculate the turnout rate since 1995.

The turnout rate has moved within a very narrow range, but 2020 was one of the lowest-turnout elections, in line with the last two elections. The 2016 election had a turnout of 88.5%, which was the lowest ever. 2012 and 2020 both had a turnout of 89.3%. This compares to turnouts of 93.9% in 1998 and 92.8% in 2004.

It is worth noting, however, that turnout has been calculated as a share of enrolment, not as a share of eligible residents. We know that enrolment rates have been increasing as a share of the eligible population at the federal level, and this presumably has flowed through to ACT elections. If enrolment rates have improved, you’d expect a decline in official turnout rates.

That’s it about the ACT for now. I will try to finish my ACT 2020 dataset before the end of the year, and I will return with an analysis of how voting patterns between the parties vary across the different voting methods. In the meantime the ACT 2016 dataset is available for free in my data repository and ACT 2012 is available for Patreon donors who give $5 USD or more per month.

NZ cannabis referendum finalised


The results of the New Zealand election were finalised yesterday, with the reporting of the special votes.

The final result saw the National Party lose two seats, with Labour gaining a 65th seat, and the Māori Party gaining a second seat.

But today I’m more interested in the results of the referendums held alongside the election.

The first referendum, on voluntary euthanasia, was very decisive, but the second referendum on legalising cannabis, was very close.

The ordinary vote count gave 46.4% of the vote to the “yes” case, but the special votes broke strongly in favour of “yes”, winning 60.3%. This led to a final figure of 48.8% for “yes” and 51.2% for “no”.

This sort of split leads to an interesting map when show the results by electorate, which I’ve done below.

There’s a weird quirk in the vote counts. Ordinary votes are listed based on which general electorate the vote was cast in, including votes cast in Māori electorates. Votes cast as special votes are instead grouped by the electorate the person is enrolled in, with Māori electorates listed separately. I have combined these totals ignoring the Māori electorate special votes, which is not perfect but gives you a sense of the geographic trends.

Most rural electorates voted “no”, although there were “yes” wins in Northland and Whangarei in the north, West Coast-Tasman in the south, and East Coast in the east.

The “yes” case did much better in the cities. Yes won in five out of six seats in Wellington, including 67% in Rongotai and over 70% in Wellington Central. Three of five seats in Christchurch voted yes, although the margins did not get to the levels seen in Wellington. There was also a solid yes win in Dunedin and a narrow yes win in Nelson.

The “yes” case did not do so well in Auckland. It did very well in the inner city seats of Auckland Central and Jacinda Ardern’s electorate of Mount Albert, but “no” won a majority of seats, including some reasonably big margins in the south-east of the city.

Podcast #47: Queensland election, the next day


Ben is joined by 4ZZZ’s Alexis Pink for a quick podcast to talk about last night’s Queensland election results.

This podcast is supported by the Tally Room’s supporters on Patreon. If you find this podcast worthwhile please consider giving your support.

You can subscribe to this podcast using this RSS feed in your podcast app of choice, but should also be able to find this podcast by searching for “the Tally Room”. If you like the show please considering rating and reviewing us on iTunes.