Wednesday, July 6, 2016

2016 Senate Postcount: A Very Long Way Still To Go

This thread contains some general comments about the overall state of the Senate race and the campaign.  There is one thing I want to make clear about the Senate race right now before I get any further:

We still have an extremely long way to go!

What we have at the moment are primary vote counts that are often little more than halfway complete and may still move around a lot based on votes still to be added.  We have a new Senate voting system and no past-data experience of how voters use it.  People are trying to pick from this which micro-parties might win seats when we don't know what the final votes are let alone what they mean.

I understand the media's desire to communicate information to readers, but this stuff is hugely complex and I just want people to accept that it is going to take a long time to know who will be winning particular seats, and in some cases we won't know what is going to happen until the pressing of The Button when all the votes have been entered.

The information required from scrutineering to adequately project the final counts would be vast even with full primaries, and we are nowhere near full primaries in most cases.  Please expect me to be highly impatient with people seeking confident projections for the final few seats in most states or wanting to know if such-and-such-second-Green-candidate will win because all such projections are guesswork at best.  Be very wary of models that purport to say which party on 2.3% will beat which party on 1.8% with a high level of confidence because all such models will be useless.  Next time around we'll have a better idea of how this is going to work.


Over the last few years there was a lengthy debate about Senate reform which led to the Senate voting system being changed.  The previous system of group ticket preferencing was abolished, giving voters back control of their own preferences.  Defects of the old system had included:

* parties being elected off trivial primary vote shares based on complicated preference-gaming
* voters being unable to practically exercise control over their own preferences without laboriously numbering many dozen squares below the line
* parties having a strategic reason to deal with parties that were ideologically opposed to them, hence often causing themselves brand damage by sending preferences somewhere that they didn't want to go
* seats being decided by irrelevant tipping-point contests between parties with no chance of winning, creating a great risk of elections being voided due to errors

The new system has allowed voters to control their own preferences either above or below the line, but this also means that it is very much more difficult to project the outcome from the primary vote count.  This is especially so as this is the first time we have had this system so we have no past preference flows to base modelling on.

During the Senate reform debate opponents of Senate reform, especially on the left, tried to argue that the model agreed to by the Coalition, Greens and Nick Xenophon would exterminate all micro-parties and leave only those groups and Labor standing.  I repeatedly pointed out that this was rubbish, for instance showing in simulations that a double-dissolution held under the new system with the 2013 election votes would have returned nine micro-party and three NXT Senators, and that even at a half-Senate election under the new Senate system micros would have won some seats.  It's possible even that those who played up these obviously silly claims have created a false belief that the new system was rigged against micros, and hence fuelled the micro-party vote and thus helped defeat their own prophecies.

Many who have sought a more representative Senate will now complain because the wrong people were elected.  Whatever micro-parties win at this election will be winners on merit - because, in a race for twelve seats in their state, either they polled a substantial portion of a quota or else voters chose strongly to preference them.  Moreover they will know that to get re-elected next time they need to keep faith with the voters who supported them and keep their primary votes at a decent level.  Under the old system there was little accountability since getting back in at the end of the term had nothing to do with what vote you got but everything to do with the quality of your preference deals.  Not surprisingly, the existing crossbench has struggled, with Xenophon and Lambie the only known winners, Leyonhjelm fighting for a seat, Lazarus and Day apparently struggling, Muir apparently defeated and Wang and Madigan polling ludicrous vote shares.

The surging micro-party vote: where did it come from?

At the time of the Senate reform debate most observers including me thought the large micro-party vote in the 2013 Senate was an aberration.  A campaign pitting a disunified Labor government that had no idea how to campaign against an Opposition led by an unpopular and stridently right-wing Opposition Leader turned off voters in droves and caused them to flock to the populist gimmick party Palmer United, while in NSW voter confusion caused the Liberals to shed several percent to the Liberal Democrats.  Palmer United crashed and burned and the use of logos fixed the LDP problem.

However, as it turns out, the micro-party vote (excluding NXT in South Australia) has increased in some states.  In Victoria the total micro vote has jumped from 16.6 points to 24.9, in Queensland from 24.1 points to a crazy 31.5 (even with the loss of ten points of Palmer United), and in Tasmania from 18 points to 22.7.  It is down in NSW from 26.5 to 24.5 (because of the LDP confusion issue last time), in South Australia from 17.9 to 12.1 and in Western Australia from 24.7 to 22.1 (LDP confusion also a factor in those two).  It is also down in the NT (where Rise Up Australia have polled 7%, securing them about $10K of public funding) and roughly unchanged in the ACT.

Nationally, the overall "Others" vote is up only slightly in the Reps (12.88% from 12.42%) but it is up from 23.54% to 25.70% in the Senate.  The demise of Palmer United has been cancelled out by the success of parties that are running more actively in the Senate than the Reps (especially One Nation and Derryn Hinch Justice Party).  There is also an increase in the NXT vote because they are running in multiple states not just SA, and beyond this the level of support for other micros is nationally little changed.

When the possibility of Hanson winning was first mooted I thought it was overrated given One Nation's very poor results in 2013.  Over time I reversed that ferret because of the nature of the campaign and a single strong seat poll for her party. The modern, rationalised campaign run by both major parties had absolutely no appeal to the sorts of people who might vote for right-wing populist outfits.   There was a widespread view among analysts (which I shared to some degree, but didn't buy as strongly as some) that the Others vote was not going to stand up to its levels in Reps polling because there would be no-one on the Reps ballots worth voting for.  It didn't matter - there was a substantial mood to vote for anyone at all except the big three.  A good example was Tasmania, where the Recreational Fishers Party has recovered its deposit in three Reps seats but polled hardly anything in the Senate.

There is, predictably, a lot of stupid stuff about the new Senate system having caused the election of Pauline Hanson.  With the vote she has achieved at this election she would have won under a double dissolution at the old system as well as the new system and she would very probably have won at a half-Senate election under either system.

There will be much focus now on the wisdom (or lack thereof) of going to a double dissolution.  Senate voting had to be reformed before this election (or preference-gaming would have been permanently entrenched), but a half-Senate election could have had most of the old crossbench blocking the government's bills in revenge til 2020.  The problem is really that rather than listening to why people were voting for micro-parties, the government's campaign tried to bully them with claims that their vote would be a vote for chaos and they should "stick to the plan" (the translation being that they were idiots for considering other parties).  With the last two majority governments being wracked by infighting and instability, many voters were happy to vote for their preferred major party in the House but something else in the Senate.

And so whoever forms government will just have to get used to the difficult art of multi-party negotiation for now, until the next two half-Senate elections reduce the size of the crossbench.

How many crossbenchers will there be?

I've frequently pointed out that micro-party voters don't just all preference any other micro-party no matter what it is, they also preference bigger parties, and so as micro-parties are cut out under the new system, majors will get preferences as well.  However, big parties can only get preferences while they still have candidates in the count.  For this reason, on current primary votes it appears that at least eleven non-Greens crossbenchers will win, and possibly a few more.  However if primary vote patterns change in late counting this might reduced.  Given that a quarter of voters voted for micro-parties, but that voters for micro-parties don't necessarily like all other micro-parties, something in this range seems about fair representation.

I have a lengthy article about Tasmania, where the apparent trend towards 5 Labor 4 Liberal 2 Green 1 Lambie is complicated by apparently huge below the line voting rates and by voter rebellions against bad preselections by both major parties.  This creates the possibility that the Liberals might snag a fifth seat at the expense of the Greens or Labor, and there is also the possibility of a right-wing micro-party sneaking into the mix.

In Queensland the primary count is only 50% (of enrolment) complete.  On current primaries the LNP has four seats with a realistic chance of five, Labor has three seats with a realistic chance of four, and Hanson and the Greens have a seat.  Numerous micro-parties (with the LDP currently leading) are fighting for at least one seat and any chance that might exist of winning a second off one of the majors. There's no point considering preferences for these until we have a more complete primary count.

In New South Wales the primary count is 57.7% complete by enrolment.  On current primaries the Coalition has four seats and should win five, Labor has four and needs to lift to be in contention for five, the Greens have one and micro-parties appear to be fighting for two, with One Nation leading the Liberal Democrats and Christian Democrats and various others further back.  Again we need a more complete primary count, especially as One Nation started very strongly in primary counting and might fall back.

In Victoria the primary count is 51% complete.  The big parties have four seats each and would need to seriously lift to compete for a fifth.  The Greens have one and are competitive for a second, and Derryn Hinch is short of a quota but would win easily.  It's still possible the Coalition will take the final spot but if it doesn't there's a race on between a seething mass of micros and again, we need to see more primaries.

In South Australia the count is 60.7% done.  On current primaries the Liberals have four, Labor three and most of a fourth, NXT would get three, and the Greens should hold their one.  But depending on the strength of preferencing it's possible Family First can come up on Liberal preferences and challenge either the Greens or Labor.

In Western Australia the primary count is 57.2% done.  On current primaries the Liberals have five and Labor appear comfortable for four.  The Greens have one and are in the mix for a second, while at least one micro-party would win.  One Nation has a handy lead on primaries at the moment in that race.

As usual there's nothing to see in the territories, with one seat for each major party.

The early indications are that this Senate could be similar to the previous one in balance.  It's not that easy for Labor to get more than 27 seats or the Greens to get above nine (four of which are at some level of risk).  The Coalition on a good day might manage 30 (or might not) so it's likely to need the support of most of a large number of micros to pass anything.  However negotiation might be easier should the number of "moving pieces" be smaller, as would be the case if One Nation won multiple seats.  On the other hand, the joint Coalition-Green majority looks rather shaky at this time.  A Labor-Green blocking majority does not seem credible but if a left micro wins in Victoria then we might be close to that should things go well for the Greens.

Allocation to terms

Reports have surfaced that the major parties are already discussing how they will deal with the allocation of Senators to six-year and three-year terms.  No doubt the tactical geniuses who convinced the ALP that Senate reform would cause a Coalition blocking majority are involved again.  An article by Antony Green is compulsory reading on the subject of how this works, but beware: it is premised on much higher votes for the "big three" than have actually been recorded, and the conclusion that a micro-party that got a quota would only get a three-year term does not necessarily follow at all.

Here I will refer to two methods that have been used or canvassed in the past: the old method (in which Senators receive terms based on order of election) and the Section 282 method (in which a recount of the election for six places is held among the 12 candidates originally elected.)  The AEC is required to conduct a Section 282 recount but the Senate is under no compulsion to adopt it.

It's been patently clear that either those having the discussions or the journalists they're talking to don't actually understand the Section 282 method, since it would not lead to a freeze-out of minor party seats (eg Labor would probably only get two six-year terms in Queensland, and on current primaries Hanson might well get one).

Derryn Hinch is complaining about the Section 282 method, but he would certainly get a three-year term under the old method, whereas he would have some chance at least of a six-year term under the Section 282 method.

It is too early to say what the Section 282 method would produce (especially in Tasmania where we don't even know which major party candidates will win or in what order).  We don't even know which Senators are going to be re-elected, and modelling the outcome of the Section 282 method (which was ignored the only time its result was previously determined) would be difficult enough even with that information.  Beyond some kind of tentative agreement that they might work together, it would make more sense for the major parties to just wait until the result of the Section 282 method is known.

The other thing to bear in mind is that the Senate can allocate the terms however it likes.  The big parties could in theory conspire and give all the Greens and micro-parties three year terms if they wanted to, but this would not be well received.

I am not sure how much more coverage of the non-Tasmanian Senate races I'll find time for, but those are some general comments that I hope will do until the primary counts are more complete, at least.

Vote Mixup In WA

In Western Australia (of course) there has been a potentially significant problem in the division of Pearce with 105 voters being given Victorian Senate ballot papers.  As a result their votes have been deemed informal.  Fortunately under the new Senate system the risk of a micro-close result in which such an error could alter the outcome is very much lower, but it still could happen.  Should a Senate seat be decided by fewer than this number of votes, the losing candidate could petition to have the election cancelled and rerun on the grounds that had these voters been given the correct ballot papers they might (in theory) have voted for the losing candidate.

The difference with the 2013 vote-loss situation is that the close tipping point that decided the election was between two irrelevant parties that could not possibly win a seat.  Had one of them been excluded at a certain point instead of the other, the consequential group ticket flows would have changed the order of two seats.  In this system, the order of exclusion of irrelevant parties generally won't alter the final outcome, but another form of tipping point is still possible. The following are cases under which the loss of this number of votes could affect the outcome:

* the margin for the final seat is within 105 votes.
* a candidate is excluded at some point by 105 votes or less and would have won had they not been then excluded.
* a tipping point is created because an elected candidate either crosses quota at a given point by 105 votes or less, or fails to cross quota at a given point by 105 votes or less.  In the first case the preferences of whoever put them over the line become diluted in the candidate's surplus, while in the second they continue at their existing value.  This can have an impact on the final outcome that is much larger than 105 votes.

We will probably have to wait until the press of "the button" to see if there is any ground for an appeal based on this situation, though a margin that close might well be recounted anyway.


  1. Hi Kevin,

    Thanks for this and all your work on the election.

    I've read Antony Green's article on the allocation of terms, and there's one claim he makes that doesn't seem obviously correct to me:

    "The difference between the two methods will be greatest where a party achieves more than the double dissolution quota of 7.69%, but less than the 14.29% quota at a half-Senate election. Parties that fall between these two percentage votes would be allocated long term seats under the order elected method, but short term seats under the re-count method."

    The One Nation Party vote in QLD is 1.19 DD quotas and so fits this description. But it seems possible or even likely that Hanson would get a seat on the recount method, and so get a long term regardless of which method is used.

    Can you comment?

    1. I am thinking that Antony was just not anticipating how extremely low the major party vote shares might be in, say, Queensland at this election. His example has an "Others" vote of 5.6% and Queensland is several times that. When the major parties are not capable of securing third quotas by the recount method, and when the Greens are supplanted as the leading minor party, that conclusion falls over.

  2. Let's see how many below the line votes exhaust after one or two parties. My mother said that she liked the new system because she could now vote below the line for just the 2 parties she wanted to see in the Senate and then not send preferences to anyone else. And friends on social media have said they voted above the line and it was hard to find 6 parties they wanted to vote for.

    There could be many voters who want to vote for just 1 or 2 parties and then stop. If they're following the ballot paper instructions and don't know about the savings provisions, they'll do this by voting below the line and stopping after 12 candidates.

    I know it's useful to avoid exhausting your ballot, and my own vote had preferences for 12 parties, with the last few preferences choosing between parties I don't like but which might be contending for the last few seats. However I only found out it was worthwhile from psephological blogs - that level of detail isn't in the AEC material.

    1. My quick eyeball of the papers is that below the line voters go all over the place, there just isn't many of them.

      Going all over the place makes no difference if your 2 through 54 have already been excluded anyway!

      I assume in most contests they will make very little difference (Tasmania is an obvious likely exception).

  3. Under the current situation does the "short" term for the newly elected senators expire in June 2019 or June 2020?

    The latest possible date for the next federal election is 29 June 2019?

    1. The new Senate expires in mid-2019. Malcolm Farnsworth has covered this comprehensively here:

  4. Kevin, I totally agree that we just don't know how the senate is going to end up at this stage, especially given the uncertainty of how the preferences are going to flow with the new system and how many votes will be exhausted.

    What I am wondering though, is once things are all known and the dust has settled in a month or so, will it be possible to ‘re-run’ the results as if it was a normal half-senate election and see exactly how the seats would have fallen? Or does that require access to the full results and/or electoral commission computing system (‘The Button’)? I just think that might be interesting a) in terms of seeing how the system might work in the future, and b) for indicating whether Turnbull’s DD decision did truly backfire or not.

  5. Based on past cases it should be possible to rerun the results as if it was a half-Senate election if someone has the time and computing power to do it. In past cases all the voting data have been made available to the public. I am not sure how this will go this time given the much larger volume of data entry.

    The AEC is required to conduct a form of half-Senate rerun for the purposes of informing the Senate regarding options for the allocation of three and six year terms. This rerun however bulk-excludes candidates who did not win at the start (it is just between the twelve elected Senators, considering all the votes count.) In many cases the results will be the same but probably not all.

  6. Where do the micro parties come from?

    A campaign pitting a disunified LNP government that had no idea how to campaign against an Opposition led by an unpopular and ruthless Opposition Leader turned off voters in droves and caused them to flock to the populist gimmick party (insert party you least like here).

  7. Hi Kevin, thanks for all your work keeping us all up to date. Wondering your thoughts on whether in the future parties could conceivably 'game' the system by putting a very popular/well known candidate lower on their party ticket not out of factional deals but as a way to get an extra candidate elected.

  8. It's possible, although I think the attraction in the Singh/Colbeck case is not just the perceived quality of the candidates but the outrage at the parties for putting them in danger. If a party was obviously doing it strategically rather than out of stupidity voters would be likely to smell a rat.

  9. What happens to those that voted both above and below? Did I hear mention that the below vote will be counted if 12 or more are numbered?

    I was told 1-6 atl or 1-12 btl, and now wonder if my mind added the "or" because that's what I expected. I did hear the lack of "at least" or "a minimum of". It seems in the haste inadequate training was given to new AEC booth staff.

    You're right, people were outraged about Singh and even more so about Colbeck, in the NW. Colbeck is one of the few reasonable Liberal Senators from Tas. He turned up to forums that Whiteley refused to and spoke with respect to people, unlike Whiteley, who accused people of being either Labor stooges or of being used by Labor.

    And, as expressed above, Thank you.


  10. If a voter votes both above and below the line then if their vote below the line is formal, their below the line vote counts and the above the line vote is ignored. If the below the line vote is not formal then the above the line vote counts, if it is formal. A below the line vote must have the numbers 1-6 once and once only to be formal (any mistakes with numbers higher than 6 are irrelevant). An above the line vote must have the number 1 once and once only to be formal. Although they say vote 1-6, some people just voted 1 because that was the old system. These votes still count.

    One case I saw twice was someone numbering 1 Liberal above the line then 2 to 59 for candidates below the line. The 2 to 59 below the line is informal as there is no number 1, so the 1 above the line counts. Their vote runs down the Liberal ticket then exhausts if all Liberals are elected or excluded.

    1. From what I could see senate ballot papers just marked '1' were going into the informal pile on the night. I didn't see any '1,2,3,4,5,6' below the line, but I suspect they might have to.

      I assume the recount is done by people more familiar with the Act and savings provisions, but it will move the numbers around a bit.

      I did have a question for you Kevin and don't see where else it would fit in.

      On the night, how do the counters assume the candidate that will come second in the electorate? especially by booth to booth? In Grayndler it looks like they assumed Jim Casey will come second, but he is at present in third place (quite possibly he will move to second) and at individual booths he is in second or third place (Haven't seen fourth, but that's a possibility for some candidates too). I would have thought they would just do Liberal or Liberal as second place on the night unless the 'third' candidate is polling 30% plus, just because the TPP Liberal v Labor is the classic breakdown.

    2. There is a pre-decided pair of candidates to count between for the whole seat on the night, which is typically based on what happened in the same seat last time.

  11. Kevin, I notice votes by Senate candidates starting to appear on the AEC site. What's the difference between 'ticket votes' and 'unapportioned'? I presume those votes alongside each candidate are the below the line votes?
    John Lawrence

    1. On the night votes are counted by party, but all together, whether above the line or below the line. They are then later sorted out into above the line (ticket votes) or below the line (votes for given candidates). Until a vote has thus been sorted it is recorded as "unapportioned" (meaning it could be either above or below the line). As the count continues the number of unapportioned votes will gradually fall away to zero.

    2. When I looked this morning there were 13,284 ticket votes for ALP in Tas. Now there are only 9,740. The unapportioned numbers have risen from 79,555 to 82,105. The group total has fallen. What are those guys doing?

    3. There is some funny stuff going on in which some booths will sometimes appear in the totals sorted as Unapportioned and Ticket Votes, with the Unapportioned presumably being BTLs, and then those booths will be taken out again. This happened yesterday with Lower Sandy Bay, Glen Huon and Geeveston booths, and there are several similar booths that are likely to go the same way today.

      I don't know why this happens; it is quite annoying. In the 2013 Senate count booths were often taken out of the total then put back in, I think while they were being rechecked.

      By the way I have a Tasmania Senate specific thread at

  12. Re Unachimba's question about the choice of the two main contenders for the 2PP on the election night count. My understanding is that the AEC makes a judgment in advance; it would be hopelessly messy if it was decentralised. Obviously I'll defer to Kevin's more precise knowledgeon the subject.

  13. Thanks for this. Good to know I wasn't wrong to question Antony Green's conclusions on the 3/6 terms. It's probably fair to say that the majors may unite to vote for whichever method benefits them the most.

  14. The potential for the 105 votes in WA to cause a problem is possibly increased by the interpretation the High Court placed on section 365 of the Commonwealth Electoral Act in the challenge to the 2013 election. In dealing with the possibility of using the initial count for the missing ballots, Hayne J appeared to take a wide view of the section 365 prohibition on evidence of "the way in which an elector intended to vote".

    Such a wide reading gives the outcome you suggest: if there are 105 votes missing and an 80-vote margin at a critical tipping point between two micros each with 1% of the vote, no evidence can be taken as to the very low likelihood that the tipping point was affected.

    If this interpretation of the section holds (noting that it is currently a decision by a single judge), I think it might be wise to consider amending it even if it doesn't cause a problem this time. While it makes sense that you don't want voters getting on the stand to say how they intended to vote, I think the door could be opened to expert statistical evidence on the likelihood that a group of disenfranchised voters would have affected the result, even if their numbers exceed the critical margin.

  15. I posted this on PB but here it is more directly for you, KB: I notice that the number of ‘unapportioned’ Senate votes is now only about 10-15% of the total, even in NSW. At what stage do they get ‘apportioned’ between the ATLs and BTLs – is it based on some further eyeballing of them before they go into the input pile for the scanning/OCRing process, or does it only happen as the scanned votes are checked by the manual operators and scrut’eers (in which case it would seem the data input is about 85-90% done)? And if not the latter answer, is there a page somewhere on which the AEC reports the progress of the data input process? If not, there should be – but, failing that, if you know please update us all.

  16. I understand a vote can't go up as BTL for a specific candidate before all votes from that booth have been completely through the manual checking process by data operators on the screens. In the case of Tasmania this was happening with some booths from a very early stage - I was watching particular booths go through the process and by the end of the day or the next day those booths had appeared in the VTR in full.

    What I am not at all clear on is the process by which a given booth appears in the system, either temporarily or permanently, with a breakdown into ATL and Unapportioned, but without the Unapportioned being sorted into BTLs for specific candidates. Initially several booths were appearing in this form and then being removed, but at the end all the outstanding booths returned permanently.

  17. Kevin, from your Tasmania article I understand that the major parties are picking up a high flow of preferences.

    Do you give the 5th Labor senator (.0732) any chance in the NSW election?

    1. This is a really good question because if the flows seen in Tasmania were repeated in other states then it would seem major parties with very small remnants might overtake all the micros and I've been thinking about this a bit. In the Tasmanian case the Labor ticket seems to outperform most of the prominent micros at a rate of about .13 votes/vote on microparty preferences, and presumably outperforms the obscure ones by even more. So if that was repeated in NSW you could actually see the 5th Labor candidate going all the way up the ladder, though some of the exclusions would be very marginal.

      My suspicion is that because of the larger NSW ballot paper and the proliferation of micro-parties, plus NSW being a weaker 2PP result for Labor than Tasmania, that the gain rate won't be quite as high, and that even if it is slightly lower the scenario will fall over somewhere. But with this being a new system and all I cannot confidently say that such things are impossible!

      Those who have been sampling NSW may have more informed comments.

    2. Yes David, we tried to. Of course, although I'd like to think that a lot of Greens voters are as intelligent as my tribe, it's not necessarily much of an answer to your question about preference flows once the Greens are eliminated. Time and the virtual tally room will tell. Some Green voters, of course, are pro-environment but scared of the Unions "Liberals For Forests" types and will preference the Libs - though probably no more than 20% of them.

  18. Thanks very much Kevin.

    I'm having a lot of fun playing around with various preference flows in a spreadsheet.

    Do you have any notion what would happen to the preference flows to Labor later in the count once all LNP & Green candidates are elected or eliminated? Should we expect a high exhaust rate at this point?

    1. Well, Kevin may have seen a wider sample (yes, Kevin?) but I can tell you David that in my family of 4 (in Qld)we all voted for a few minors, then Greens, then ALP and Lazarus (or reverse) then finished off with Lib Dems and LNP in case they were in a battle for last place with someone worse. As any intelligent leftie would do.

      (If it had come to ALP v Lazarus for last place (it won't) we would have cancelled each other out, but otherwise we were consistent.)

    2. Thanks Southbank stroller. Sounds like you covered all bases :)

  19. In my sampling for Tasmania roughly 23% of micro-party preferences exhaust without hitting any of Labor, Liberal or Green. But I have not looked at whether those preferences landing or starting with Liberal or Green go on to Labor, as opposed to a micro or exhaust, because that is probably not a live question in the Tasmanian count (unless One Nation gets over either the Greens or Colbeck, which is possible.)

    1. Thanks for that Kevin. Still following with interest.

  20. NT preference data for valid votes has just been released following announcement of the result earlier today. The data is in a new format which I think I've figured out... glad to get this data now so that I can figure out how to parse it into my database before the more interesting results come out. On first glance I can see a surprising number of ballots with votes both above and below the line.

    1. The cool thing about the new format is you get to see votes that were both ATL and BTL that were formal in one and informal in the other. We will also get to see how many people vote both ATL and BTL but inconsistently. Really there needs to be "VOTE EITHER ABOVE OR BELOW THE LINE, NOT BOTH" across the top of the ballot paper in big bold font.

  21. Thanks again for your coverage of Tasmania.

    Can we take anything from the Tasmania result to help predict some of the other races?

    Am I right in thinking that McCulloch got a slightly higher share of micro-party preferences than McKim? Do you think we can expect a similar showing across the nation or did McCulloch have a particular advantage in Tas?

    1. Yes, McCulloch had closed the gap to McKim from a notional 0.1169 quotas to .046 quotas by the exclusion of Colbeck. A small part of the closing was the loss of .014 quotas from the Green ticket because of leakage.

      Parties One Nation did well from were Shooters, JLN, ALA, DHJP, PUP and not too badly off Family First. All these exist in other states and JLN had only a small surplus in Tasmania. Tasmania did however have a weaker lineup of left micros than Victoria, for instance. In Victoria I don't see ON as a threat to Janet Rice because the ALP will exit the count quickly there and she will get a stronger flow off left micros for that reason.

      I have been considering whether ON are a threat to the Liberals in Victoria but in Tasmania they did not really gain against the Liberal ticket. I do think anyone expecting ON to be caught on preferences in either NSW or WA can forget it, but I was thinking that anyway.