New numbers change Victorian federal redistribution equation


Federal redistributions in Australia rely on two sets of numbers – the actual enrolment numbers at the time that the process commences, and an estimate of how many voters will be enrolled at a future point, about three and a half years after the conclusion of the process. While electorates need to be drawn within 10% of the average enrolment at the start of the process, the projected numbers are far more important, since seats need to be drawn within 3.5% of the average on those figures.

These projected numbers ensure that electorates are drawn so that faster-growing seats have smaller populations, and slower-growing seats have larger populations, and thus population change reduces malapportionment, rather than increasing it.

Unfortunately, if those projected numbers are no good, the whole thing is undermined.

The projected figures for the Victorian federal redistribution were released in October, and I posted about them here.

There have been a number of people raising concerns about those figures in the comments, and earlier this week the AEC acknowledged the issue, saying that “The AEC has been informed by the ABS that there was an error in the initial enrolment projections supplied for use in the redistribution of Victoria.” They have now released this corrected data, and it does change the distribution of population in a way that will favour outer suburban growth areas, which will be drawn with smaller electorates than if the original projections had been used.

I haven’t personally investigated the problem with the projected numbers, but as an example Zac Gross posted this graph, showing that almost all Victorian SA1s were assumed to have growth of almost exactly 10%, whereas in other redistributions the growth rates vary (as you’d expect).

First up, I’ve modified the following table that I posted in October which breaks down electorates in Melbourne into different parts of the city north and south of the Yarra River, and breaks rural Victoria into east and west.

The original projections had the 26 seats of Melbourne collectively about three-quarters of a seat under quota, but that deficit was spread out across the city. The 16 seats south of the Yarra were about half a seat under quota, while the 10 seats north of the Yarra were about a quarter of a seat under quota. It was particularly surprising that the six seats of western Melbourne, an outer suburban area where you’d expect fast growth, was projected to barely gain any population relative to other parts of the state. This looked very different to the trends in the NSW redistribution, where Western Sydney is set to gain a seat while the eastern half of the city loses two.

But this picture looks different with the new figures. The ten seats north of the Yarra are pretty much spot on quota, although the central city seats are under quota and will probably need to expand to take in surplus growth in the western suburbs. The 16 seats south of the Yarra are now 84% of a seat under quota, rather than 50%. Indeed the eight seats I defined as “eastern Melbourne” are almost half a quota under themselves.

This makes a huge difference to the implications for the redistribution. It’s now clear that the seat to be abolished will be in the eastern suburbs of Melbourne.

This doesn’t mean that other areas won’t be affected – overall the seats north of the Yarra (urban and rural) fall short of their quota by 17% of a seat. It’s possible this deficit could be spread out amongst the 19 seats on the north side and thus doesn’t require a seat to cross the Yarra, but there will definitely be a need to distribute population differently within that area.

Another way to look at the figures is via this map. It has two layers. The first shows the revised projected quotas for each seat, while the second shows how much each seat’s projected quota was changed by the revisions.

The first map now makes a lot more sense. Most seats in Victoria are under quota (as you would expect when a seat has been abolished), but the outer suburban fringe on the north-west and the south-east both tend to be over-quota. This was not the case on the original figures.

When you toggle to the second tab, it’s very clear that the new projections have favoured outer suburban areas.

La Trobe, Lalor and Calwell were all projected to be under quota but are now projected to be well over quota. The change in La Trobe was 13.2% of a seat’s population!

It’s quite unfortunate that this mistake was made after the first two rounds of submissions, which would have been made based on those projections being correct. The mapmakers will be able to use the correct figures, but will be relying on public submissions based on entirely different numbers. But I’m not sure the alternative of allowing further rounds of public submissions would have been viable. At least the problem has been identified and fixed. It would have been far worse to continue with incorrect numbers, that would have likely led to fast-growing outer suburban areas being under-represented.

Western Australia was also affected by this issue, and I’ve got a blog post coming up covering WA this afternoon.

Liked it? Take a second to support the Tally Room on Patreon!
Become a patron at Patreon!


  1. I haven’t looked at the new numbers but my original proposal removed all the rural parts of both La Trobe and Holt, making them entirely focused only on the growth suburbs (Holt being the Cranbourne Line and La Trobe being the Pakenham Line).

    Then I had Bruce move west and focus almost entirely on City of Dandenong. With Aston being considerably more under-quota now, I would move Endeavour Hills into that seat. Being north of the freeway and a bit of a PT black hole that isn’t along the Dandenong train lines, it doesn’t seem out of place with suburbs like Rowville & Lysterfield.

    I don’t know if that still works on the revised numbers though. But basically that’s how I had it looking.

  2. Macnamara – Current enrollment 112881 – Projected enrollment 122119
    After redistribution – Enrollment 117142 – Projected enrollment 125205

    [edited for length]

  3. @ Trent, I would support Endeavour Hills going into Aston it is a middle class area in contrast to the suburbs such as Dandenong. Also Lysterfield South is in Casey LGA and right next to Endeavour Hills so it will move together into Aston.
    I am still hoping the rural component of Holt and La Trobe can be removed.

  4. baed on what ive done liberals should retake aston, olidifit their hold in deakin, possibly win mcewen. the new menzies/jagajaga would be lost and chisholm would strengthen for labor.

  5. @John renaming Melbourne to “Batman” wouldn’t be very well liked by Adam Bandt I wouldn’t think given that John Batman was a very controversial figure.

    I’m curious as to why you named Melbourne “Batman” though. The old seat of Batman (named after John Batman, the founder of Melbourne, not Bruce Wayne, the superhero from DC Comics) was actually what is now the seat of Cooper. As unique as it is to have a seat that coincidentally shares its name with a superhero, I think it should stay as Melbourne given that there’s a seat called Sydney. Furthermore, I believe Melbourne is a Federation seat, meaning it existed in 1901 and hasn’t been abolished since, so the AEC won’t rename it.

  6. @nether portal. im allowed to make suggestions. i was thinking of renaming Sydney too but didnt put that one in.

  7. Come on John, you can’t post a giant list of enrolment numbers for every seat in the state. Make a document somewhere else and link to it with a summary.

  8. John

    Hate to tell you but I think your Holt is under the minimum quota for 2023. It works fine at the other end.

    It is is really hard to get the fast growing seats to work at both the near year and the further year. If I recall my 2020 submission for Holt was something like 100 electors in from the minimum at the near year and 100 electors in from the maximum at the far year. If I recall, I might have come unstuck in Calwell because Craigieburn was growing so fast.

  9. redistributed is right but there is a pretty easy fix. Just put the remainder of Narre Warren South – West into Holt.

    I’ve have some pretty close calls

    La Trobe:
    Current: 90.27
    Projected: 103.42 (I’ve pretty much just been playing around with Yarra Ranges small towns putting one in and one out to get both numbers to fit)

    Current: 90.78
    Projected: 102.95

    Current: 91.32
    Projecred: 103.26

    I’ve kind of wondered what would happen if an electorate was basically unable to be drawn without going over one of the limits. Would they just have to draw the weirdest boundaries to try and get it to work

  10. @Trent and @Nimalan, I would disagree that Endeavour Hills should be in Aston as it has strong cultural and community connections to Dandenong given it is multicultural with many going to Dandenong to shop.
    Rowville in comparison is more Anglo and Upper Middle Class and are more likely to lean their leisure to Knox City rather than Dandenong. This is even evident from the voting patterns of Endeavour Hills and Rowville.

  11. @ Marh
    I actually would say Rowville is increasingly multicultural although still slightly less diverse than the Greater Melbourne. Rowville has a growing Chinese community almost 13% and much higher than the Greater Melbourne average. Both Endeavour Hills and Rowville have some of the largest Sri Lankan community. Endeavour Hills in terms of SES is somewhat of a demographic midpoint between the Dandenong and Rowville it is a transition suburb like Mulgrave. I would also describe it as entry level middle class. I do concede many residents would likely shop in Dandenong or Fountain Gate rather than Knox City and that is less affluent.

  12. Casey needs to gain voters not lose them. I have added bayswater north to Casey and Vermont and Vermont south to aston

  13. @redistributed i stand corrected thanks for pointing that out il suggest the aec bum it up to the correct numbers by scrounging up some more voters from la trobe or bruce.
    @drake wont work the projected nmbers are at 129k so adding another 10k will push it over
    @marh that works out then 🙂 i did it because i wanted t remove whitehorse from deakin. now instead of spanning 4 divisions (kooyong, menzies, deakin and chisholm) is now only spread across 3 (kooyong, chisholm and aston)

  14. While I haven’t yet looked at how the new numbers impact my own maps, I feel like they would actually improve the one part I really wasn’t happy with which was the McEwan / Casey border.

    I had to stretch McEwan a lot further east than I wanted to, in order to get it within the quota range. Now that the projected electors have gone from -2.4% to +6.4%, I feel like I’d be able to move that boundary back towards the west where it makes more sense and is a bit more compact.

  15. Has anyone found a neat way to redraw Lalor?

    Transferring Williams Landing, the rest of Truganina and the rest of Point Cook to Gellibrand brings Lalor down to +3.90%, but transferring Werribee South would put Lalor just below -10% on current figures.

    I’d be nice to keep Skeleton Creek as a strong boundary, which would suggest that Werribee South needs to be split, but is there a better way? Lalor could take the rest of Little River from Corio to balance this out perhaps?

  16. @Angas I had the exact same problem, and Lalor was the first seat I tried to draw. I initially tried the opposite of what you suggested, having Corio take Little River and Cocoroc (no one lives there). Little River fits fine in either district. I ended up just putting 3 SA1’s in the NE corner to Tarneit Nth into Gellibrand. It’s not ideal but it’s better than the other options.

  17. @Angas
    Spliting Werribee South isn’t that bad of an idea. The Wyndham Harbour development (to the east of Duncans Road/Beach Road) is all new and probably has more in common with Point Cook than Werribee (along with the rascals of Campbells Cove). Old Werribee South of beach shacks and market gardens belongs more with Werribee proper. It means you have to split the SA1 21305136839 along Duncans Road, but it doesn’t have many people and is all farm land anyway. So I would leave 812, 813, 827 and most of 839 in Lalor, and shift 834, 835, 842 & 843 into Gellibrand, along with the more obvious moves you mentioned.

  18. @Witness
    Thank you for the breakdown. I wasn’t familiar with the different parts of Werribee South, but looking at the SA1s, I was already thinking that Duncans Road would be a ‘good enough’ spot to draw a boundary. So it’s good to hear that this would be a reasonable option.

    Yeah I think it would be fine to unite all of Little River in either Corio or Lalor, but I suppose the committee would probably prefer to keep the hard LGA boundary as they traditionally have.

    @Drake @John
    So therefore a small slice of Tarneit would seem to be the only other alternative to splitting Werribee South. Either the part north of Leakes Road and east of Derrimut Road, or the part north of Sayers Road and east of Morris Road could work well.

  19. Anthony Green raised Casey as a possible seat to abolish which could see the Deakin moving to the upper Yarra Valley and Dandenongs going into Aston and La Trobe which in turn Monash and divisions in south-east Melbourne might get moved into the abolished parts of La Trobe. It does seem to have merit although it might over quota Deakin unless if Mcewen makes up for this.

  20. @John, I don’t think Menzies will be abolished as it is well associated with Manningnam like Aston is to Knox plus Aec always tries not to remove seats that is named after a Prime Minister

  21. I suppose the committee will probably leave Gippsland and Monash unchanged as they both have relatively strong borders, but I wonder if they will be tempted to try to top up Gippsland since it is projected to be at -2.79%.

    Moe-Newborough is too much for now and the alternatives of Mirboo North or Welshpool aren’t really that good.

    Previously it looked like Monash would gain Lang Lang and maybe Koo Wee Rup from La Trobe, but that is looking less likely now that Monash is basically on target at -0.16%. That would probably only come back into play if Gippsland expands.

  22. Here is my rough initial proposal:

    – Abolish Deakin but transfer the name to Chisholm. The name of Hotham can be changed too
    – Regional areas mostly stay the same but Woodend into Bendigo and Kilmore into Nicholls, with Corangamite’s excess going to fix up Wannon and Corio. Mallee, Indi, Ballarat, Monash and Gippsland all stay the same.
    – Inner city areas largely reverting to previous boundaries (Melbourne gains Kensington, Wills rest of Brunswick, Cooper gets Clifton Hill)
    – Jagajaga gets most of Bulleen and a bit of Templestowe Lower so that it can lose most of Nillumbik to McEwen, which is needed so McEwen can top up Nicholls, Scullin and Bendigo.
    – Maribrynong is still a bit of a bits-and-pieces seat but someone has to gain from Calwell
    – Isaacs gains most of Dandenong council except for Dandenong itself. Unless you make a seat entirely based on Dandenong council, it’s pretty difficult to make a seat that doesn’t combine middle class areas with it. Hotham’s section of Dandenong council just did not work with any seat.
    – The high growth seats of Calwell, La Trobe and McEwen in particular didn’t really have a lot of options that worked both with current and projected enrolment. The Clyde North SA2’s and Mickleham SA2 in particular were hell to get to work with any seat.

    I might try later on doing a Melbourne/Macnamara transfer or a Casey/McEwen transfer to see if they work any better. I feel I was pretty limited based on how strict the Calwell, La Trobe and McEwen numbers needed to be.

  23. @Drake
    Nice proposal!

    I really like what you’ve done with Casey and La Trobe, although that merger of Menzies and Deakin feels a bit awkward.

    Greater Dandenong and Kingston are difficult to handle at the moment, so I think it’s fine to make the kind of compromise you’ve made by putting parts of Greater Dandenong into Isaacs.

    Your Maribyrnong is good, but if there’s a way to readjust the outer-North so that McEwen can drop either the Calder or Nillumbik, I think that would be beneficial. Overall I think it’s pretty difficult on these numbers.

    Everything else looks really good though. Lots of sensible changes. Nice and minimal.

  24. @Angas

    Yes I’ve wanted to have McEwen lose Macedon council but it’s really hard to get the numbers to work. You can’t have it gain from the top of Calwell (where there actually is a lot of similarities between the two seats) because doing so leaves McEwen over projected enrolment. McEwen is kind of always going to be wonky because you need to pair it with low growth areas, and it’s surrounded by a bunch of high growth areas.

    Yeah I might play around with Chisholm, Menzies, Casey. They were the last ones I did and at that stage I was just trying to get the numbers to work. The Menzies border looks weird but I feel it at least kind of makes sense on paper. Parts of Croydon/Ringwood have been in Menzies before anyway.

  25. Il finish my Vic maps after the last 2 wa ones (Burt & canning) just need to confirm the boundary of canning/burt the other maps are available for viewing. The new seat will either be the swan based one ofpr darling scarp based. Just need to check the numbers of the swan seat to see where the majority of hasluck voters go.
    @angas yea but the way I’ve drawn them works so why fix it when they work. Most of my eastern divisions are at the higher end and western lower-par. I’ve abolished Menzies and the renamed jagajaga to preserve the name. ,y Hotham still contains 2/3 voters of the old Hotham so I kept the name

  26. i still think something has to be done about mcewen unfortunately given the current configuration i was able to do any major changes to it. i moved kilmore into nicholls and mernda north into scullin and then added some parts of jagajaga. hopefully can be fied next time

  27. Am also having trouble with this. I have Calwell/Lalor/McEwen all finely balanced, but can’t get the other numbers to work subsequently. I think I’ll have to force a swap of high-growth and low-growth areas with one of the other seats, but I can’t find a neat way to do this (unless I leave McEwen looking dreadful, which I really want to fix)

    SE Melbourne I got to work by mangling Dunkley further east, but I ended up with really nice boundaries elsewhere so accepted it.

  28. The best I’ve managed to do in regards to McEwen is to remove the parts along the Calder.

    Something like:
    – Top up Hawke with Gisborne (after it has lost Bacchus Marsh Surrounds SA2 to Ballarat and Hillside to Gorton)
    – Top up Bendigo with Woodend and Macedon
    – Top up Nicholls with Kilmore
    – Top up Scullin with Wollert in exchange for the rest of Mernda
    – Reduce Calwell by taking everything north of Donnybrook Road (which does split Mickleham, but means that Maribyrnong and Wills don’t need to move further north)
    – Finally, top up McEwen with the remainder of Nillumbik Shire except for the majority of Eltham

    It’s not perfect, but I think this would move a few of those northern divisions in the right direction. The main issue would be having to split Macedon Ranges Shire between 3 divisions. In particular, I don’t know where to put Romsey. Does it fit better with Bendigo, Nicholls, McEwen or Hawke?

  29. Personally I think Macedon ranges should go to hawke. I think the deficit in hawke Ballarat etc will push hawke east and the Gordon should be based on Melton. Personally the old numbers worked better as mcewen could have been abolished and broken up

  30. In regards to Bendigo I’ve kept my original suggestio. Of adding lockington gunpowder sa2 as I believe Echuca would be better off in Bendigo in a future redistribution given communities of interest and transport link with bendigo


Please enter your comment!
Please enter your name here