Sunday, July 3, 2016

Election Wrap: A Total Mess!

ELECTION OUTCOME: Hung parliament or narrow Coalition majority

Apparent Coalition Wins: 69
Apparent Labor Wins: 64
Apparent Crossbench Wins: 5

Links to threads will be added here.

Undecided Seats (12):

Coalition vs ALP (9):
Flynn (see below)

Coalition vs NXT

Coalition vs ALP or Greens

ALP vs Greens

I have not called Flynn because although Labor currently has a 1.8% margin there, in 2013 the Coalition gained exactly that in the postcount in that seat.  

Over coming weeks I'll be posting more details of the amazing mess that is the 2016 federal election outcome.  It will take us well into the postcount before we know whether the Coalition has flopped over the line in enough close seats to claim a small majority, or whether it might have to go looking for one or two crossbench friends.  The Coalition needs to win eight out of 11 unclear seats to win a majority.  Labor would have to win everything it can to win majority government, but may form minority government if it wins most of the seats still in play.  I have listed 12 undecided seats above but it is possible I will add others that might move back into the frame, or should I find a credible NXT scenario in some other SA seat.  (A case for NXT having some sort of shot might be argued in a few other seats but requires heroic preferencing assumptions.)

The current two-party preferred swing in the 138 "classic divisions" (Coalition vs Labor) is 3.3%.  On 2013 experience this should come down to about 3.0%, once all the non-ordinary votes are added, so about a 50.5% 2PP (exactly as per the final Newspoll), or maybe slightly more.  The polls were right, and any shift in preferencing behaviour based on the primary votes appears to have been very small indeed (I currently have it at about 0.2 points).  

The Coalition would have been expected to just win a majority on these sorts of numbers, and might yet just do so.  But the very real chance that it will not arises from some unexpected feral swings in areas that based on seat polls and aggregated state polls did not appear to be at so much risk.  In particular, my home state has delivered three seats to Labor, including the seat of Bass with a 10.8% swing, when none of four seat polls by two different companies had that swing above 5%, and when aggregated state polling often implied a swing to the Coalition!  Another fine example is the seat of Macarthur, which has fallen 59:41 when three Newspoll/Galaxy seat polls had the Coalition at 49, 50 and 50.  It appears seat polling has avoided the disaster that befell it in 2013 (when it skewed hopelessly to the Coalition) but has found a new way to be unreliable.  

If the Coalition does come up a bit short in seats for the actual 2PP, I suspect it is the distribution of the swings that will be the cause.  Seats with small swings or swings to the Coalition were often useless seats and Labor may have just done better in the marginals.  The election also reveals weaker evidence for sophomore effect than is usually the case, with the "double sophomore" seats showing swings only 0.5 points below the national average (though this rises to 1.1 points if the outlier Bass is taken out.) In some cases this is a poor reflection on the calibre of the first-term Coalition MPs.

It scarcely matters (except for election addicts and potential MPs) whether the Coalition gets a Reps majority or not, because it is facing an even worse Senate than the last one.  Australians disenchanted by Turnbull's high-tech innovation nation and Bill Shorten's progressive-left Labor have turned in droves to protectionist populist Senate parties of the right and centre (as I suspected after reversing my initial Hanson-scepticism) as well as micro-parties of all kinds.  The Coalition looks about 9-11 seats short of a Senate majority.  It is unclear there will be enough non-Green crossbenchers for the Coalition to pass bills without the support of the Greens, and even if there are, the only benefit will be that there will be more party blocks included in them (three NXT from South Australia, and possibly multiple One Nation Senators from various states.)  Hinch and Hanson are winning, and there looks to be one as yet unknown Queensland micro and two in NSW.  

Calling a double dissolution under the new system has restored democracy to the Senate after the disastrous 2013 election, and those who win seats will do so based on the actual will of the voters, but the messy outcome of this proportional representation election with such a small quota will raise the question of whether a half-Senate election and putting up with the existing crossbench until 2019 would have been pragmatically wiser.  We will have to see exactly who is in this mess to see whether the government has any chance of passing its ABCC bills or even bothers with the joint sitting at all (that is, assuming it is returned.)  

Of the seats I list above as close there are two that require special explanation.  Melbourne Ports is a Labor win on the published two-party preferred (Labor vs Liberal) but Michael Danby is only 2.0% ahead of Steph Hodgins-May (Green).  However the Others vote in Melbourne Ports includes 1.9% for Animal Justice, 1.4% for Drug Law Reform, and 1.4% for Marriage Equality, which all sound like potential Greens feeders.  Perhaps Danby will move further ahead of the Greens in the post-count but on present numbers they could very well overtake him in the preference distribution.  If this happens then the Greens currently need 73% of all preferences over the Liberal candidate to win.  Danby is getting 80% of all preferences himself, but Danby infamously tried to direct his own preferences to the Liberals (reports vary of what cards were handed out and when).

The other is Grey.  Currently the ABC projects Grey as a Liberal retain, but this is not based on real numbers - it is just a guesstimate of preference flows.  In coming days Grey should be realigned as Liberal vs NXT and we will see whether Andrea Broadfoot can get the 72% of preferences she currently needs.  Rowan Ramsey may make that mountain steeper if his position improves in the postcount.

Tomorrow I will start rolling out postcount threads, for the following:

* Tasmanian Senate
* Melbourne Ports
* Grey
* One for all remaining two-candidate seats
* Possibly one for the remaining Senate races once the percentage counted is higher.

General analysis will continue to be added here.

A hint on the Tasmanian Senate: Lisa Singh at least is in with a serious chance; I have seen scrutineering figures with her getting 8-17% of the vote (not just the BTL vote or the Labor vote - all the vote) in four southern booths.  How well she does in the north of the state is as yet unknown.

(My usual voting declaration: Wilkie in Reps, and effectively (albeit not 1 for partly tactical reasons) Greens in Senate. I often make things difficult for the Tasmanian green movement but the major parties' policy failures on same-sex marriage (Liberal) and Senate reform (ALP) were completely unacceptable to me, and the Greens at least ticked both of those boxes.)


Explainer: Why Are The AEC totals and election-night ABC totals different?

The AEC totals currently show the ALP leading in 72 seats and have 7 seats "not yet determined".  The figure for ALP leads is a few seats higher than the ABC had on election night.  The reason for this is that the figures on election night included a projection that takes into account post-count performance at the seat at the last election.  The "not yet determined" seats are mostly not close; they are seats where the AEC has not yet formally determined who the top two parties are because they differ from last time.  The "not yet determined" seats are Grayndler (which Labor will win), Grey (unclear NXT vs Liberal) and five other seats that the Coalition should win, with very minor doubt about Barker and Cowper depending on preference flows.


  1. Hi Kevin, the AEC s currently showing 4.4493 senate quotas for the ALP. Am I right to interpret this value as a mix of ATL #1 votes for ALP, and BTL #1 votes for individual ALP candidates?

    As I understand it, the AEC will have already tabulated BTL #1 votes. Will we have to wait until Tuesday to see this breakdown?

  2. Yes this is a party total of ATL and BTL #1 votes for each party. However I do not believe we will get a quick ATL/BTL split and candidate breakdowns; rather, this emerges gradually as votes are entered for data-processing over coming weeks.

  3. Hey Kevin, thanks for the update.

    I'm just curious about the discrepancy between the national 2PP shown by the AEC via the Tallyroom and the 2PP shown by ABC/Yourself/Misc others?

    At it's maximum difference it was leaning nearly 1.5% more towards Labor on the Tallyroom than elsewhere, right now that's down to 0.4%.

    Any idea why this is happening?


    1. The Tallyroom 2PP (currently 50.11) is a live total of votes that have actually been counted on a 2PP basis. It is not adjusted as a projection for the tendency of post-count votes (postals, absents, out-of-electorate prepolls and provisionals) to favour the Liberals on a 2PP basis. Last night it would have been very dependant on which votes had been included at various stages and which not.

    2. So was Barrie Cassidy right?

    3. About what? The preferencing / internal polling thing he got from Labor, or something on the night? If the former then the reason for Labor's internal seat polling painting a better picture than the public seat polling most likely has very little to do with preference shifting (the argument as to why preferences should shift being wrong anyway) and a great deal to do with the public seat polls being rubbish in their sampling of primary votes.

  4. G'day Kevin,
    What's going on in Barker? The ABC show less than 50% of ALP and Green preferences going to NXT, is their computer just having a hissy fit?

    1. There are no real preferences in Barker and the ABC computer outfit is nonsense. Probably a bug rather than bad estimates.

  5. Another interesting SA result to watch will be the last senate seat. At the moment the ABC gives NXT, the ALP and the Greens 1 likely seat each but they only have about 2.2 quotas between them and it would appear that there are at least 1.2 quotas unaccounted for spread across all right wing parties. Could the Greens be wiped out in SA?

  6. Kevin, are pre-poll figures included in the AEC "ordinary" count to this point? Or do they constitute "declaration pre-poll"? I'm watching from the States, and this is my first Australian election. Thanks!

    1. Prepolls are included in the ordinary count if they were cast within the voter's electorate. Out-of-electorate prepolls are declaration pre-polls.

  7. Under the new system I expect that preferences won't flow strongly enough between the various micros for them to catch up but I have not looked at the HTV cards for SA closely enough to say. I have just been looking at something similar in Tasmania where the gap between the micros and ALP/Green is much lower.

  8. In regards to your voting declaration do your tactical considerations make much difference? For left people that understand how to do it I'd understand a tactical below the line vote for some popular and credible minor left candidates that runs into in the Greens, then flows on to other less left parties and ends with the least bad most centre/right wing candidate you're happy with (perhaps 1 to 35 or something like that). I feel the tail end matters more because the point something percent of a vote after (or if) the Greens make quota might at least end in say NSW a Xenophon versus One Nation elimination before it vanishes. The chance of another left wing party staying in the count is very remote.

    1. It all depends on where the totals land. I sent mine through a few minor parties before sending it where it ended up and would have sent it through more but there were just too many that I seriously disliked. Maybe I would have put those micros first anyway though even the micros closest to my own views mostly had red cards for opposing Senate reform. I think it's gone where it needs to be in terms of beating any micro that might come up through the pack (all of which are bound to be something horrible).

      The Tasmanian Senate field this year was ghastly, or maybe I'm just finding more and more reasons to dislike parties. I could hardly find anyone who I was comfortable voting for at all, but dozens I was desperate to vote against. We need to design a system where a voter can prioritise expressing who they don't want elected rather than who they do, but I have no idea how to do it. An early attempt decades ago caused a senate full of Australian Democrats in a simulation, so I scrapped it.

  9. Thanks for all your hard work, Kevin.

    Some of the prepoll booths are showing up now for Grey. Seems to be showing weaker than required preference flows to NXT so far but can we read much into it considering it's prepoll and also that NXT had stronger primaries in these booths too.

    1. I am having a look at this in detail now in writing a post about Gray. The golden rule is that anyone who gets excited about so-and-so being in front on the 2PP and tries calling the seat is a mug!

  10. Congratulations, Kevin, on an excellent election coverage. Well done!

  11. If there is another possible NXT scenario, might it be Sturt rather than Boothby (as I've heard from some quarters). Cwissy Pyne's Primary is down below 45%. The Primary gap between ALP & NXT is just a tick over 2% (22.62 to 20.55) with GRN a tick below 7%. Outside chance I will admit but Barker is essentially off the table due to higher Lib primary and larger FF vote.

  12. I'll have a look at Sturt on the Grey thread soon.

  13. Kevin, I have the Coalition on 69 here. Going off the AEC seat summary:

    - They have 5 out of the 6 "Not Yet Determined" column: Barker, Cowper, Durack, Higgins and O'Connor.
    - They have 37 out of 40 of the Liberal party seats, the other three being Chisholm, Gilmore and Dunkley which you have in doubt.
    - LNP 18 seats.
    - Nationals 9.

    Have you given Labor a seat from the Coalition where they are currently trailing?

  14. You are correct! Thanks, it's easy to lose track of one while shuffling seats in and out of doubt, especially at 3 am.

  15. Matthew G
    Just a little confused buy the AEC senate page ... i realise that it's still early and still a lot of counting to be done even before preference cut up.... but what is the difference between "ticket votes" and "unapportioned"... i would have thought the former was the ATL votes but even for the major parties these numbers are small compared to the latter... can you enlighten me?. I'm interested partly because me and the Missus tried this tactical voting thingy after reading your sage explanation on the matter and our first vote for a worthy but no-hope candidate still has no votes recorded against their actual name. Cheers

    1. On the night all the booth votes are counted by party but are not separated into above the line votes for the party or below the line votes for a candidate. An unapportioned vote is a vote that has been counted to a party in this way. A "ticket vote" on the other hand is a vote that has been fully processed through the system as a 1 above the line vote for that party. Over time all the unapportioned votes will resolve into ticket votes or votes for individual candidates. However the process is very slow because the votes from any given booth are not included until every vote from that booth has been processed, and votes from a given booth are not always processed together. In past elections it has taken a few weeks for all the unapportioned votes to be sorted to the relevant candidate or as party above-the-lines.

    2. Where you say ' A "ticket vote" on the other hand is a vote that has been fully processed through the system as a 1 above the line vote', would this refer to any votes made above the line or only those made above the line that followed the party's preference order shown on their how to vote card?

    3. Under the new system it refers to any above the line vote. The AEC keeps no official record of how-to-vote cards, and in any case some parties changed their cards through the campaign.

  16. Capricornia and Herbert seem to have postals and prepolls ... absent votes outstanding has some one worked out the % rate to each party in each seat for each?

    1. I've been out all day but will deal with these soon on the Vanilla Reps Postcount thread.

  17. With what remains, it looks like unless there is a massive change that Capricornia should go to Lib/Nat by about 900-1000 votes, Hindmarsh to Labor by 700-800 votes and Herbert too close to call (maybe slight Lib/Nat advantage). All up should be 76 to 77 seats as most people have predicted.