Victorian election endgame

A week and a half on, the finishing touches are being applied to Victoria’s rather extraordinary election result.

Saturday morning

Late excitement in Ripon, where contested ballots during a recount appear to have worn away to nothing. The VEC’s communications on the recount have been a bit confusing, but the buzz on social media suggests the Liberals have achieved the tiniest of leads, with the result perhaps to be decided by a single digit margin. The recount is to resume this morning, and is proceeding slowly as every dubious ballot is scrutinised in minute detail.

Thursday morning

My results platform is now fully updated with the latest results, and hopefully more or less works. Many more preference counts have been conducted, but not in the one contest that remains of potential interest, namely Melton.

UPDATE: By latest results, I mean the latest VEC feed. Unfortunately, this hasn’t been reconciled with the preference counts yet.

Tuesday evening

The Victorian Electoral Commission is now a fair way through the preference distributions, and it seems the numbers in the new parliament will be Labor 56, Coalition 26 (Liberal 20 and Nationals six), Greens three and independents three. The one remaining chance for a boilover is Melton, which will presumably be retained by Labor, but they have only 34.9% of the primary vote with the remainder scattered among the field of eleven other candidates. Then is the upper house, which I’m afraid I haven’t been able to give its due over the last week and a bit, but do stay tuned.

Notable results from the resolution of the count:

• Sam Hibbins retained Prahran for the Greens by a 7.4% margin over the Liberals, which he was able to do because he again squeaked ahead of Labor at the last exclusion. At that point in the count, Liberal candidate Katie Allen was on 14,824 (36.7%), Hibbins was on 12,911 (32.0%) and Labor’s Neil Pharaoh was on 12,647 (31.3%). The 264 votes separating Hibbins and Pharaoh compares with 31 votes when the exact same candidates faced the exact same situation in 2014. The difference on that occasion was that Pharaoh landed only 277 votes clear of Liberal incumbent Clem Newton-Brown on the final count – here as in so many other places, the Liberals were not a feature this time.

• Labor won the western Victorian seat of Ripon by just 31 votes, Sarah De Santis finishing with 20,030 (50.04%) over Liberal incumbent Louise Staley on 19,999 (49.96%).

• The vague prospect of an independent win in Benambra did not eventuate, with Liberal incumbent Bill Tilley emerging with a winning margin of 2.4%. Independent Jacqui Hawkins reduced Labor to third place, at which point Tilley was on 19,517 (47.1%), Hawkins was on 11,778 (28.4%) and the Labor candidate was on 10,110 (24.4%). That left Hawkins needing 88% of preferences, and she managed 78%.

In the seats that were being followed closely on the earlier post, Liberal member David Southwick made it home in Caulfield by 205 votes (0.3%); Labor’s Jackson Taylor prevailed over Liberal incumbent Heidi Victoria in Bayswater by 296 votes (0.4%); Labor’s John Ormond Kennedy was a 329 vote (0.4%) winner over Liberal member John Pesutto in Hawthorn; Tim Read of the Greens won Brunswick from Labor by 504 votes (0.6%); and Labor’s Chris Brayne finished 767 votes clear (0.9%) of the Liberals in Nepean.

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics.

171 comments on “Victorian election endgame”

  1. Sorry Kevin but I don’t get how the Greens political party would “easily” win eastern Metro. They have barely half a quota. It certainly seems far clear that they have been hard done by (where you could make a much stronger case in southern metro)

    Perhaps they are more likely to slip over the line for the fifth place but hardly seems an inevitability. Like you’ve alluded in country seats, a smaller set of micro parties taking the “not the majors or the greens” vote, would pretty much all have at least one of the major party surpluses “directed” to them ahead of the Greens

    The biggest factor impeding the Greens LC outcome in this election is the big Labor vote. Votes getting second or third Labor members over the line would not have got back to the Greens even if voters were able to direct their own votes above the line

  2. @Roger-

    There are two key elements that are of importance, based on experiences with the Senate voting system in both the 2016 election and in the states (NSW and SA) that use the same system:

    1) A lot of votes will ultimately exhaust at some point, meaning that quotas at the start are larger in importance than just their raw numbers may suggest;

    2) Many of the voters will, at fairly early points, direct their preferences to the parties that did best in terms of obtaining first preferences.

    To demonstrate, some notes involving likely outcomes in each district using a Senate-style system:

    Eastern Metropolitan: ALP and Liberals each win two seats on quotas. After that, it would be

    GRE 0.53
    LD 0.25
    ALP 0.22
    LIB 0.19
    DHJP 0.15
    AJP 0.14
    DLP 0.10

    In this circumstance, it is hard to see how the Greens would lose the race for the fifth seat- it would require a preference flow to the ALP that seems unlikely in a circumstance where the voters are directing their preferences.

    Eastern Victoria: The Coalition and ALP win two on quotas, but have, respectively, only 0.05 and 0.01 of a quota left. For the last seat, there is:

    GRE 0.40
    SFF 0.30
    DHJP 0.27
    LB 0.24
    AJP 0.19

    Here, I see it as likely that the Greens would ultimately lose to a micro-party- but, in this case, it depends heavily on what choices the voters who backed the smaller parties not in this list made, as either the Shooters, Hinch, or the Liberal Democrats could have a shot depending on that.

    Northern Metropolitan: ALP wins two and the Greens one on quotas, and the Liberals one fairly rapidly once preferences start to flow. From there:

    ALP 0.58
    SOC 0.25
    DLP 0.25
    REA 0.20
    DHJP 0.12
    AJP 0.12

    I don’t see anyone catching the ALP for the fifth seat, especially since I strongly doubt supporters of the DLP or Hinch (or, for that matter, quite a few of the Reason voters) would back the Socialists or vice versa.

    Northern Victoria: The ALP and Coalition win one each on quota and will get a second each on preference. For the fifth seat:

    SFF 0.47
    GRE 0.39
    DHJP 0.29
    LD 0.23
    ALP 0.14
    VEP 0.12
    ACP 0.10

    I don’t see it as likely that anyone would stop the Shooters- there aren’t enough left voters for the Greens to be likely (especially given that a good hunk of the smaller ones will be expended getting the ALP its second seat), nor do I see Hinch or the Liberal Democrats passing the Shooters (especially since, again, a lot of the smaller right vote will be expended getting the Coalition its second seat).

    South-Eastern Metropolitan: The easiest story: the ALP wins three either on quota or close to it, the Liberals one on quota, and there’s no real way the Greens (the only other party with more than a fifth of a quota) will pass the Liberals for the fifth seat.

    South Metropolitan: The Liberals and ALP win two on quota- after that:

    GRE 0.80
    LIB 0.30
    AJP 0.13
    REA 0.12

    It would require a statistically implausible preference flow for the Greens to lose.

    Western Metropolitan: The ALP wins two on quota and the Liberals one- after that:

    ALP 0.76
    GRE 0.52
    DHJP 0.42
    LIB 0.30
    DLP 0.21
    AJP 0.16
    SFF 0.12
    LD 0.10

    The ALP will win the fourth seat (and their third). As for the fifth: it depends heavily on how preferences do or do not flow, as the Greens, Hinch, and even the Liberals all have a shot.

    Western Victoria: The ALP wins two on quota and the Coalition one- after that:

    COL 0.80
    GRE 0.45
    ALP 0.29
    DHJP 0.27
    SFF 0.27
    AJP 0.17
    LD 0.16
    VEP 0.11

    The Coalition would win the fourth seat (and their second). For the fifth seat, it depends on who the last two parties standing are- I don’t think the micro-parties of the right could win (especially since a lot of their support is likely to flow to the Coalition), but who does depends on, firstly, Hinch or the Shooters passing the ALP, and, secondly, the ALP passing the Greens. Of the three most likely options (in order):

    Greens vs. Hinch/Shooters- Greens wins on ALP preferences
    Greens vs. ALP- ALP wins on Hinch/Shooters preferences
    ALP versus Hinch/Shooters- ALP wins on Greens preferences

    As these calculations demonstrate, the Greens would have at least three seats and possibly (if all breaks went their way) as many as five. In comparison, there seem to be only two seats where micro-parties are guaranteed a win, and only one more where it is a reasonable possibility- and, in these cases, they are seats where the micro-party in question either polled a large first-preference vote (Northern Victoria, Western Metropolitan) or where the two largest parties leave the count at an early stage (Eastern Victoria).

    Meanwhile, the ALP is, by my math, guaranteed nineteen seats and the Coalition fourteen- this demonstrates that the micro-party bloat affects all the parties with high first-preference votes.

  3. @G. Rochon Loll

    I appreciate the analysis and don’t disagree with anything you’ve said.

    However, I think if the problem is that both voter and party behaviour would change if the Fed Senate system was introduced in Victoria.

    I don’t know of anywhere where a regionally based upper house count on the new Fed rules has been tried. 2016 doesn’t count because of the DD drastically reducing the quota.

    I honestly believe that in East, South and West Metro the seats have flipped from the Greens to Labor rather than the micros, due to the massively increased Labor vote. The micros in those regions have won at the end expense of the Liberals. The micro voters are much more likely to be right wing and I doubt they’d give the Greens any preferences at all.

  4. It is interesting to note that the Lib/ Nats in the last 2 elections have gone from 21 Legco seats to 10.

    That has to be some record

  5. @3Z But in some cases it is Labor and Reason preferences that are resulting in a RW micro getting over the top of the Green. Look at Southern Metropolitan: https://www.abc.net.au/news/elections/vic-election-2018/results/smet/ Specifically I direct your attention to count 20, where at the exclusion of the Reason candidate votes originally for Labor and Reason are going to Sustainable Australia! There’s 4.5% of the vote wrapped up in that, and the Greens are on 13.9% at that stage. You can’t honestly believe voters for the ALP or Reason would be flocking to Sustainable Australia over the Greens if it wasn’t for the distortion of GTV.

  6. @DVC

    The last time I did the numbers was yesterday and the only time then (by the calculators) that Labor preferences were material was in West VIC where they get the AJP over the Green.

    In the metro areas the labor overhang wouldn’t have had an effect on the outcome anywhere.

    I haven’t looked at Reason but it would be rich to question their preferencing decisions. The Greens preferenced Hinch above Reason I’m North Metro.

  7. Try again now, GG. Wait a few seconds for loading and do a bit of hard refreshing if it doesn’t work at first — and if it still doesn’t work, let me know. I’ve also just discovered that the whole thing doesn’t work on MSIE 13, but I’m not going to let that worry me.

  8. G. Rochon Loll has made my point for me well. The new Senate system has shown that when voters are given the choice of what to do with their preferences, you do not get strong flows from right wing micros to other right wing micros, or even from micros to micros in general. Firstly preferences have a tendency to flow more to the well known parties than the lesser known ones, and secondly preferences scatter. A lead for the Greens of 4.71% over another party would not be caught when voters chose their own preferences. That is especially when the lead is over the Liberal Democrats, who are often very poor performers on preferences because (i) their primary votes are obtained mainly as a result of confusion (ii) their candidates are often obscure. There are also some micro-parties whose preferences would flow reasonably strongly to the Greens – AJP, VEP, SA, VS, FPRP.

    Exhaust also plays a role in making it hard for parties to catch up but even if it wasn’t a feature, the spray of preferences that results when voters make their own decisions would ensure that such a lead for the Greens held.

    In a fair system, the fact that Labor’s vote has gone up would not stop the Greens still winning seats where they have a big lead over any other party.

  9. Seen reports from Twitter folk presumably watching media that suggest margin in Ripon has come down to 3 or 4 votes. There will be a “further full recount preference distribution” (whatever that is, is it a full recount or isn’t it?) tomorrow.

  10. @3Z: This is what I’m referring to:

    Count 20: Jill Mellon-Robertson (Fiona Patten’s Reason Party) excluded
    8,425 (2.00%) votes originally from Fiona Patten’s Reason Party distributed to Sustainable Australia (Clifford Hayes) via preference 7.
    4,252 (1.01%) votes originally from Voluntary Euthanasia Party distributed to Sustainable Australia (Clifford Hayes) via preference 7.
    2,980 (0.71%) votes (73,174 ballot papers at 0.0407 transfer value) originally from Australian Labor Party (Ticket 1 of 2) distributed to Sustainable Australia (Clifford Hayes) via preference 12.
    2,980 (0.71%) votes (73,174 ballot papers at 0.0407 transfer value) originally from Australian Labor Party (Ticket 2 of 2) distributed to Sustainable Australia (Clifford Hayes) via preference 12.

    I’m not questioning Reason’s preferencing decisions as much as I am the distortion created by the system as a whole compared to what the *voters* for those parties would choose. Even if only a third of the preferences above flowed to the Greens instead of Sustainable Australia then the Greens get over the line.

  11. William Bowe @ #115 Thursday, December 6th, 2018 – 8:24 pm

    Try again now, GG. Wait a few seconds for loading and do a bit of hard refreshing if it doesn’t work at first — and if it still doesn’t work, let me know. I’ve also just discovered that the whole thing doesn’t work on MSIE 13, but I’m not going to let that worry me.

    Donuts, mate!

  12. Aaaand – at 9.30 am Friday the VEC’s Ripon page at https://www.vec.vic.gov.au/Results/State2018/distributionRiponDistrict.html shows 91.39% counted (ie, as near as dammit to complete) and preferences distributed step by step, and Labor ahead by 31 votes!

    And re the LC – can I suggest that running hypothetical counts based on the ATL votes (as the ABC does and maybe some of you are doing using your own software) is pretty pointless. The time for “predictions” is when the VEC has entered all the BTL votes and run the proper count. Oh, that’s right, they won’t be predictions, they’ll be the official result! As the Ripon count shows (and many others show), when people distribute their own prefs, they go all over the shop.

  13. Sad, but not surprising, that people think Labor is a substantially different party from the Libs.
    No wonder they have been widely branded ‘Liberal-lite’
    Oh, the dichotomous thinking of the typical voter comprising the majority of the populace
    Baaa-baaaaaa-baaaaa

  14. Vic, look at their workplace relations policies – and the laws they pass when in power! Yes, they’re too close on refugees and coal, but if you want workers to have rights there’s a yuuuge difference between them.

  15. Kevin (or anyone) – as votes exhaust (obviously fewer in Vic upper house, but more in Federal Senate), is the quota progressively reduced mathematically or does the same original quota number keep applying until no-one can get that number anymore. I would imagine there must be some races where the last two end up on say 0.9 and 0.8 of a quota with the 0.9 one being declared the last winner.

    But it is conceivable that using a ‘running’ quota throughout the elimination process could deliver a slightly different result as a candidate could be declared elected and their surplus distributed earlier in the counts, rather than a few more minors being knocked off the bottom of the list before this same candidate got to an original ‘full’ quota.

  16. RR, no – the quota is not reduced, so there is a theoretical chance that the last person elected will not have a full quota. But from memory I don’t think that happened in the last Senate election. We will see if it happens in any of the LC regions.

    There are systems that reduce the quota as the number of live votes decrease – eg, the Meek system https://blog.opavote.com/2017/04/meek-stv-explained.html which is used in NZ for local government elections. But it’s quite complicated, so even the Proportional Representation Society recommends its use only where “the expertise of those conducting the count, the communication of the method to the electors and candidates, and the reliance on computerised counting are not problematic.” (See http://www.prsa.org.au/meek_stv.htm ) So no Australian jurisdiction uses it.

  17. There is a massive difference in between the ALP and the Liberals. One only has to look at their social policies and how they fund them.

    The ALP funded the Safe injecting facility as a government funded entity that has to report to the health department.

    The Liberals had promised more cash for “not for profit organisations” “The Babes Project” and “Olivia’s Place.” These Pro-Life facilities in all but name were going to get close to Million bucks worth of Tax-payers cash thrown at them – with no reporting direct back to the government or no management of their activities.

  18. The original quota keeps applying. In the context of the Senate I recommended that a progressively reducing quota be implemented. JSCEM has made the following recommendation:

    “2.26 The Committee recommends that the Australian Government commission a technical report on the most appropriate count and surplus transfer methodology for Senate elections.
    2.27 This technical report should consider the need for a progressively reducing quota.”

    I would be surprised if this was complete in time to be implemented for the 2019 federal election, even assuming the recommendation is accepted.

    A progressively reducing quota should be used for the reason you suggest – otherwise a candidate may continue receiving preferences even after their election is mathematically assured, which is basically a waste of votes. This was an issue in the Tasmania Senate count where the last two did indeed finish below quota and there were various preference bundles that could have carried higher weights with a progressively reducing quota.

  19. Just a quick calculation: 56.3% of the votes and 88 seats implies 49.5 seats proportionally – does ALP Victoria have a disproportionate number of seats? are the electoral boundaries well drawn?

  20. Single-member-per-seat systems aren’t designed to create proportional representation and no amount of electoral redrawing will make them do so. Also even if they did, proportionality would be measured off the primary vote and not any estimate of 2PP (which is what I assume the 56.3 is above.)

  21. B. Bach – It’s well known that single-member districts, even when drawn fairly, tend to exaggerate majorities. Some people even see this as a good thing, at least when their party is on the winning side of the equation. It’s been suggested that, on average, a “cube rule” applies – https://en.wikipedia.org/wiki/Cube_rule . Though as the wiki article says it doesn’t always work that way; sometimes a party with a majority of the overall vote ends up with a minority of seats.

  22. Also that assumes all seats have the same number of electors

    I think the ALP prior to the election had more over enrolled seats and less under enrolled seats (compared to the average) than the Coalition. This has an impact on the meaning of the 2PP

  23. The VEC preference distribution page for Ripon – https://www.vec.vic.gov.au/Results/State2018/distributionRiponDistrict.html – keeps saying it’s been updated to a new time, currently 15:30:01 Friday 7 December, but keeps showing the same 20030 to 19999 result. So do they just show the same “Last updated” time across all districts when they’ve updated anything? Or do they change the “updated” time across the Ripon pages when they’ve only updated the primary count?

  24. Ahh yes Kevin. I see from your Friday 1:30 post on your own site that you’re puzzling over the same thing. I think they’re a bit careless with their “Last updated” times!

  25. That ratio of cubes of TPP rule is complicated by the presence of third parties.

    56.3:43.7 gives 2.138 cubes ration which gives 60:28 seats ratio in 88 member house.

    If Labor win Ripon, and you group the Greens seats on the Labor side (as even Prahran would not have been won by the Liberals this time), and the independents on Coalition side (as they were seats Labor would not have won), you end up with 59:29 which is pretty close.

    Jack Aranda and Kevin – thanks for that – like all these very small changes, they could make a difference at occasional ‘pinch points’ in these counts.

  26. That will be it and Staley will be seated. If Labor challenges in court and succeeds in either being declared winners themselves or else forcing a by-election, then only upon that ruling being made would the seat become vacant.

  27. I would say that’s the end of the recount in Ripon. With such a close result, a petition to the court of disputed returns would be likely if there was either (1) a view by Labor scrutineers that valid votes had been improperly excluded or (2) evidence of voting or campaign irregularities. Once the count is finished, it will become clear fairly quickly if there is any dispute. Chances are this is the end of the matter.

  28. Very hard to see why Labor would challenge the final outcome in Rippon in the courts. With a huge majority there’s very little for the ALP to gain from an extra seat, while any court action will be costly, have marginal chance of success given the scrutiny the ballots have already been subjected to, and would be widely seen as poor sportsmanship. Much easier to completely avoid the potential for bad PR and get on with governing.

  29. The ALP may well decide a new election is not worth it and hope the redistribution adds enough Ballarat or other similarly ALP friendly area to Ripon and subtracting some less ALP fierndly areas in the east of the seat. If the ALP think they have a chance of winning in court without a new vote, it may be worth it for them to take it to court.

Leave a Reply

Your email address will not be published. Required fields are marked *