Sunday, July 3, 2016

Postcount: 2016 Tasmanian Senate

Seats Won: 4 Labor 4 Liberal 1 Green 1 Lambie
Seats In Doubt: 2 (Labor vs Green vs Liberal vs various micro-parties)
It appears likely Labor will win five seats
Lisa Singh has very good chance of re-election. 
Richard Colbeck's position is not yet clear.
It appears difficult for micro-parties to win (One Nation has a remote chance, it is hard to find a realistic chance for others)

Expected to be returned easily: Urquhart (ALP), Polley (ALP), Brown (ALP), Abetz (Lib), Parry (Lib), Duniam (Lib), Whish-Wilson (Green), Lambie (JLN)

Bilyk (ALP), Singh (ALP), McKim (Green), Bushby (Lib), Colbeck (Lib) are contesting final four seats, together with various micro-party lead candidates.

Current assessment 5-4-2-1 to Labor or 5-5-1-1 are most likely

=======================================================================

Note: I have decided to put some new modelling of this contest in a new post.  Go to Tasmania Senate: A Model Of What Might Occur for my latest comments, though those below are also often still relevant.




This post follows the post-count for the Tasmanian Senate race.  I warn that it is complex and around Wonk Factor 4/5 at this stage.  The Tasmanian Senate race is complicated by the state's high below-the-line voting rates and especially by the below-the-line voter rebellions against the preselection demotions of Lisa Singh to sixth on the Labor ticket and Richard Colbeck to fifth on the Coalition ticket.

For this reason we cannot read the Tasmanian Senate race solely off the party totals - the race is about candidates and is again complex (though not as complex as last time).  Although the Liberals seem well behind the Greens for a seat, chances arise for them to win five if their remaining vote is evenly enough split between Senators Bushby and Colbeck.

I expect to do more scrutineering in the next week to try to get a clearer read on a possible race between McKim, a Liberal (probably Colbeck at this stage) and micro-parties for the final seat.

Below the line voting rates in the Tasmanian Senate are very high (I currently expect the rate to finish at about 28%) because of the combination of the new system and the Singh/Colbeck situations.  Most ATL voters are stopping at six and many BTL voters are stopping at 12.  Some voters are voting both above and below the line (the BTL vote only counts then) but making their above and below the line votes completely different.

The following figures were current as of Saturday 9 July and will be updated from time to time:

--------------------------------------------------------------------------------------------------------
79.4% of enrolled voters have had their votes counted.  In 2013 final turnout was 95.1%, so about 83% of votes have been counted.

The current party quota totals are (a quota is about 7.69%):

Labor 4.435 quotas
Liberal 4.190
Green 1.436
Lambie 1.100
One Nation 0.325
Family First 0.252
NXT 0.188
Shooters, Fishers and Farmers 0.169
Sex/HEMP 0.168
Christian Democrats 0.111
PUP 0.093
Rec Fishers 0.091
Animal Justice 0.089
Lib Dems 0.061
Derryn Hinch Justice Party 0.054
Science Party 0.049
Renewable Energy 0.049
Aus Liberty 0.041
Flux 0.040
Arts Party 0.028
Ungrouped 0.019
Citizens Electoral Council 0.007

----------------------------------------------------------------------------------------------------------

In the post-counting the votes will move around a bit, especially based on which votes are being counted, but not normally by more than a tenth of a quota.  A wild card is that some votes that are being counted as informal will probably be found to be formal, and it isn't clear yet who this will advantage.

If all voters voted above the line it would be fairly likely that Labor and the Greens would take the last two seats and perhaps this is what will happen.  However the situation is greatly complicated by below-the-line voting for Lisa Singh and Richard Colbeck (and possibly others).  Especially, should votes between Colbeck and the remaining Liberals split in the right way, this could create a situation in which the Liberals win five at the expense of the fifth Labor candidate or Nick McKim, despite having a poorer quota total.

Thus as well as the seat balance there is a question of which major party candidates will win seats, and at present Catryna Bilyk is under threat from Singh in the event that Labor only win four, while David Bushby is under threat from Colbeck if the Liberals only win four.

Current progress

Votes for each party are counted as "unapportioned" on the night then gradually sorted into above and below the line and counted to parties or candidates during data entry.  It will take a while for the breakdown of each party's vote into above and below the line votes, and who the BTL votes are for, to become apparent.  In the early stages of this process the BTL vote often appears to be very small compared to the ATL vote or totals, but this is misleading.  A further complication is that, for whatever reason, many booths are sometimes having their ATL but not BTL votes included in the totals, and this means that making sense of what percentages candidates are actually receiving below the line has become extremely time-consuming.  Often it seems that there are a lower proportion of BTLs than there are.

The number of ballot papers that have been processed already is large, but a booth is not uploaded until all votes from that booth are processed.  Because the batches in which booths are processed don't necessarily correspond to exact booths (a booth may be spread across multiple batches) it isn't predictable when a given booth will be completed and upload - even the staff don't know.

Lisa Singh


The Singh below-the-line vote is significant for Labor's chances, as well as her own.  There appears little doubt that Lisa Singh has more than 3.5% of the vote in her own right, which means that the fifth Labor candidate John Short will not receive any ALP ticket votes, and will be knocked out of the preference cutup quickly, making Singh the effective number five. The current lineup of votes for different parties is actually about the best thing Singh could have possibly wished for, because it is so easy for her to knock out Short, and yet she can also have enough votes to be competitive.

If Singh has between 3.5% and about 7.5% Labor's chances are improved because it means that Catryna Bilyk and Singh both stay in the count as candidates short of a quota.  The ideal situation for Labor would be Singh to have about 6%, and then Bilyk and Singh could sit in the count with most of a quota each and not be caught.  A less ideal situation for Labor is that Singh herself is elected quickly, leaving Bilyk on about half a quota and at risk of losing.

If Singh gets much more than a quota in her own right this will be a problem for her party as her preferences leak with substantial numbers going to the Greens, Liberals (especially Colbeck), Lambie etc.   This is because some of the vote for Singh and Colbeck was a cross-party protest vote against perceived major-party preselection injustice and hackery.  However it does not appear that Singh will be much, if at all, over a quota.

In early counting of actual votes by my estimates Singh is on 11.8% of 13714 votes in Denison, 9.4% of 9682 in Franklin, 2.75% of 4079 in Braddon, 2.9% of 3982 votes in Lyons and 2.9% of 13982 in Bass, which would give her 5.9% statewide (0.77 quotas, leaving 3.67 quotas for the rest of the Labor ticket).  However tracking the figures has been exceptionally difficult because of issues with the AEC website displays of information so please treat these as estimates only.

The above figures are consistent with on-the-night scrutineering of three Denison and three Franklin booths that found the Singh BTL rate between 8% and 17%.

On the above projected votes Bilyk may get ahead of Singh, since although some below-the-line votes for other Labor candidates will go to Singh first, aside from that Singh can only receive below-the-lines, and all the above-the-line preferences that flow to Labor will go to Bilyk.  However there is a high chance both would be elected since for either to lose it would be necessary for three other candidates (eg both Liberals and McKim, or replace one of these with a micro-party candidate) to pass one of the Labor candidate.  I expect Singh's projected vote to improve as more BTLs from Lyons are added.  The chances for five Labor seats currently appear good.

Singh also gets quite good flows of preferences from minor party BTL votes.

As an aside, Catryna Bilyk apparently received zero of the first 1171 below-the-line votes counted in Braddon.  The only other party candidates to receive zero in this batch were the two Citizens Electoral Council candidates and ten micro-party support candidates.

Richard Colbeck

Richard Colbeck is the fifth Liberal candidate and will be elected should the party win five seats, but they are actually only about 1% over their fourth quota.  This means the only way the Liberals can win five is if both Colbeck and David Bushby stay high enough in the count to beat either Nick McKim or one of the Labor Senators, as well as all micro-parties. If this doesn't happen, Colbeck and Bushby are competing for a seat.  Colbeck will benefit from leakage within the ticket against the ticket order (for instance there are a fair few 1 Abetz 2 Colbeck votes) and after that will rely on below-the-line preferences while Bushby continues to collect ticket votes.

In early counting of actual votes by my estimates Colbeck has 3.5% of 13714 votes in Denison, 4.0%  of 9682  in Franklin, 9.1% of 4079 in Braddon, 3.9% of 3982 in Lyons, 3.9% of 13982 in Bass, which would give him 4.88% statewide (0.63 quotas, leaving 3.56 for the rest of the Liberal ticket).  However this estimate of Colbeck's vote may not be completely reliable depending on the range of booths included in the count.  Tracking the figures has been exceptionally difficult because of issues with the AEC website displays of information so please treat these as estimates only.

The above figures are consistent with on-the-night scrutineering that found Colbeck's vote in six Denison and Franklin booths to be mostly between 0 and 3% with one booth at 12%.

If Colbeck's below the line vote does not improve on the projected estimate then he will probably trail Bushby based on above-the-line votes flowing to the Liberal ticket, and it also appears difficult for him to catch Singh.  In that case he will need to stay ahead of McKim and the micro-parties, which may be difficult.

Nick McKim

On raw quota totals Nick McKim appears to be travelling OK.  The Greens are on 1.45 quotas and this will probably improve.  However there are two dangers to his position.  The first is that a favourable split in Liberal preferences could put both Colbeck and Bushby ahead of him as candidates, and if the same split happens in the Labor ticket with Singh and Bilyk, he might not catch up.  On current primaries this scenario is very much on the cards.  The second is that he might be caught by a right-wing micro-party or the Nick Xenophon Team on preferences.

Early signs are that McKim's below the line vote is not high enough to make any difference to his position other than reducing the loss of votes by leakage when Whish-Wilson is elected.

Micro-parties

The changes to Senate voting have removed the problem of more-or-less-100% preference flows between micro-parties and ensured that the preferences of micro-parties are determined by the voter.   Also, micro-parties tended to lack presence at booths to hand out cards (at my booth only Sex Party and Renewable Energy were there). The question is the level of anti-major-party sentiment and the extent to which micro-party voters at this election have chosen to put numerous other micro-parties ahead of the "big three".  A bit of a pointer here is that the Recreational Fishers Party polled 6% in the three Tas Reps seats it contested but less than one percent in the Senate (but perhaps their voters mainly voted for Lambie and not other micros).  So Reps votes for some parties were basically just "none of the above" votes to the Greens and majors.

Voters could often find how-to-vote cards for their parties online (list for Tasmania here), but micro-party voters tend to buck the cards anyway, and it's not clear how many voters would have looked.  If micro-party voters all ignore their party cards then their preferences will splatter with many going to big parties ahead of other micros.  This would stop any micro from making really massive gains.

Right-wing micros (ON, FF, SFF, CDP, LDP, ALA) have 0.986 quotas between them.  Left micros (Sex/HEMP, AJP, REP, Science, Arts) have 0.39.  Remaining micros (NXT, PUP, RECF, Hinch, Flux, CEC) have 0.47.

It seems more or less impossible for any of the left-micros to catch Labor or the Greens, and on that basis I expect that all will be excluded, releasing some preferences to the Greens and to a lesser degree Labor.  However the Sex/HEMP how-to-vote card preferences apparently exhaust (unless they changed their card), which I must say wasn't very bright of them.  Long-term Sex Party voters (if any) would be well aware their party has contentious form on preferencing, and hence likely to make their own decisions.  (Indeed in my scrutineering so far Sex Party preferences go to Labor and the Greens.)

On the right side, the Christian Democrat card preferences go to the Liberals and then to One Nation, and the Shooters, Fishers and Farmers card preferences go to One Nation.  If Family First can get over One Nation, then all the right micro HTV card preferences except the Liberal Democrats will pool with them, and they also get the Liberal preferences if Colbeck is excluded at the right time.  However it is difficult for Family First to pass One Nation whether HTV preference card flows are assumed to be weak or strong.  The evidence so far is that they are weak.

Assuming One Nation is the leading micro then they get neither Family First nor Liberal HTVs, and the donkey votes included in the Family First tally will flow to Labor.  So it seems One Nation will struggle to get up to where Labor and the Greens are, unless the Shooters, Fishers and Farmers voters follow the card (or decide to preference them anyway) extremely strongly.  It being a new system, this is quite possible, though I suspect S,F+F votes will also go to the majors.

There is some thought that preferences might pool with the Shooters, Fishers and Farmers in sufficient quantities to get them over many other micro-parties.  I think it is hard for them to get enough to overtake One Nation.

NXT would be likely to pull preferences from everywhere if not for the fact that few Liberal and no Labor or Green preferences will be released til very late in the count.  They appear to be getting some benefit from Lambie Network preferences but  I think they're just too low down.

In scrutineering I've found that micro-party preferences go all over the place, with few following how to vote cards and many preferencing the big parties.  This also applies to Liberal ticket votes.  Less than 20% of those I watched were following the ticket recommendation to preference the Christian Democrats; more even preferenced Labor.  (These were from Denison and Franklin however).

My feeling overall is that it is difficult for any micro-party to win because of the preference spraying and candidate issues with the others, but that One Nation or SF+F are the best placed micro-parties if one does.  I should note that simulations at the geekLections site are repeatedly suggesting the Shooters, Fishers and Farmers will win the final seat, albeit without considering the complexity of the below the line count.

More comments will be added through the count.

27 comments:

  1. What are Nick McKim's chances?

    ReplyDelete
  2. Fairly good I think, but by no means certain. The main scenario in which he loses is that Colbeck's BTL vote is high and McKim gets stuck alone behind both Liberals. This doesn't seem terribly likely to me at this stage but I really need more data on Colbeck.

    ReplyDelete
  3. If there are a lot of BTL votes, or if most people ignore the HTV cards, then I'm not sure how you can make a prediction for the final two seats at all. What can you base your predictions on?

    Do you expect that the postals will vary significantly from the existing first preference proportions? The media say that the Liberals do better in the postals, but is it expected to be a significant difference? And who can predict how many votes the micros will get from postals?

    ReplyDelete
    Replies
    1. I expect that there will only be strongish preference flows if lots of people follow the cards. If many voters do not follow the cards (whether voting ATL or BTL) then preference flows when parties are excluded will be weaker and generally parties will just be excluded from the bottom up, making it impossible for micro-parties (for instance) to come up through the count, and just leaving a Labor-Liberal-Green contest for the last seat. We can rely on very ideologically similar parties having some preference flow though, eg a lot of Animal Justice preferences will go to the Greens and Labor whether their voters have seen the cards or not.

      I thought I'd mentioned the postcount shift for the Liberals in the article but see I hadn't. In 2013 they improved their vote total by 0.15% (0.02 of a quota) from the postcount to the final primaries - not a very large difference, plus the ordinary votes still to come are unlikely to help them. Micro-parties generally perform slightly more poorly on declaration votes, but it varied by micro-party with not much difference in most cases.

      Delete
  4. I voted at one Hobart booth in the morning, and handed out htv cards (green ) at another and didn't see a single micro party person at either. Lots of get up people though

    ReplyDelete
  5. Peter Madden is talking up his chances on Twitter. What do you think is the likelihood of this truly ghastly outcome eventuating?

    ReplyDelete
    Replies
    1. As far as I can tell, quite remote, though anything above 0.0000000 is much too high for my liking. There is just going to be so much individual voting in Tasmania, vast numbers of BTLs, and it is going to be a lot like a Hare-Clark election - you get votes on name recognition, not because party X preferenced party Y. The best chance for him might be if Colbeck gets quota on BTLs causing Bushby to be cut out early and releasing the Liberal HTV preferences. But even the non-Colbeck Lib voters are voting BTL in significant numbers and those who vote ATL will not all follow the card.

      Delete
  6. On the AEC senate Tally Room http://vtr.aec.gov.au/SenateStateFirstPrefs-20499-TAS.htm, there is ticket votes and unapportioned, what is the difference?

    ReplyDelete
    Replies
    1. A ticket vote is a vote that is 1 above the line for a party. An unapportioned vote is a vote that was counted as a vote for a party on election night but has not yet been separated as above or below the line. It looks like what they are doing is separating votes into above or below the line, and counting the ATLs by party as they go, and then the BTLs will be counted by candidate as they are data-entered.

      Delete
    2. thank you, I thought ticket votes must have been ATLs but the numbers didnt stack up.

      Delete
  7. This comment has been removed by the author.

    ReplyDelete
  8. It appears if they have just started seperating ATL and BTL votes, still a long way to go and they still haven't counted about 85,000 - 90,000 senate votes yet at all.

    ReplyDelete
  9. More senate votes starting to trickle through, too early to make an accurate prediction but looks like strong BTL voting for Colebeck and Signh. Could end up with Lisa Signh having ~ 0.7 of a quota BTL, Colebeck ~ 0.6 of a quota BTL and Mckim picking up the Greens 0.45 of a quota remainder. Could be interesting if the BTL votes mean Signh or Colebeck actually beat out the fourth member on their respective tickets (seems much more likely to happen for Colebeck given the liberals being just over 4 quotas).

    ReplyDelete
  10. "some votes that are being counted as informal will probably be found to be formal". Yes, I am amazed that the informal rate is allegedly up from 2.2% to 5.5% in Queensland. My best guess is that the same *poorly-trained* officials who were telling voters that they "must" number 6 boxes ATL or 12 BTL (and apparently saying, or implying, no less and no more) were doing the preliminary count and were rejecting anything that didn't have exactly 1-6 ATL or 1-12 BTL. Shame, AEC, shame! I know they have a difficult job but the general level of ignorance reported from polling officials suggests the training missed a few VERY important points.

    (The silly-looking nom-de-web is because I occasionally comment on another Atom-hosted site about apartment developments in Brisbane, where it actually looks quite appropriate. But I think you know who I am, KB.)

    ReplyDelete
    Replies
    1. If there are votes being counted as informal if they have more than 1-6 ATL or 1-12 BTL then that's pathetic by both the officials and the AEC as optional preferential voting is nothing new and it makes not logical sense to exclude people for preferencing more than the minimum.

      If votes are counted as informal if they are below 1-6 ATL or 1-12 BTL in the preliminary count then that is understandable. The rule around a vote being formal even if preferenced less than the minimum is a rule only known to people who have kept a keen eye on the senate reforms and the AEC quite rightly did not want to promote this too much.

      Delete
    2. I am not aware of any evidence of normally valid votes with more than 6 (or 12) numbers being counted even as provisionally informal on the night. I do believe this happened provisionally with just-vote-1s, but they will all be sorted out on data entry anyway.

      There are a lot of strange votes out there under a new system. Some people are voting both above and below the line. I saw two votes that had a one ATL above the line then numbered 2-dozens below. (Such votes will be saved by the savings provision, but only for one party). There are a lot of votes that are extremely hard to read. There are some people who have voted 1 only below the line for a candidate (which really is informal). Some people have voted 1-6 below the line (which is formal under the savings provision.)

      I am aware that voters in many places weren't told that they could number at least so many boxes, and that in some unfortunate cases voters were incorrectly told they could only number that many. That said, the ballot papers had the correct advice on them, but this is clearly an area that needs improvement next time and that I will stress in the next round of JSCEM hearings.

      Delete
  11. With about 80% count it seems clear that Singh will be elected late in the piece with a near quota depending on the run of BTL votes with the fifth Labor candidate knocked out fairly early and the fourth Labor candidate below a quota but a big enough share to get a seat at the end.

    Five Labor, 4 Liberals apart from Colbeck, 1 Greens and Jacquie Lambie.

    The last spot between:
    Colbeck probably with about 0.5 plus of a quota by then,

    Second Greens with a fair number of votes with Recreational Fishers, Sex/Hemp, Arts, Science, Renewable Energy, Animal Justice which will presumably preference Greens to a significant extent above centre and right groups but also with a fair share going to Singh and fourth Labor and plus a share of the more centrist NXT and PUP (and maybe Lambie ticket).

    A right minor - there are enough FF, Shooters, Hanson, ALA, LDP votes (a full quota in all) to get ahead of Colbeck and the second Greens if their preferences ran well between all groups - hard to see it happening as Liberals, Labor, Greens and Exhaust will gradually whittle down the total.

    NXT also an outised chance if the high recognition X logo drags in preferences above the line.

    Looks like about 20% overall voting below the line. A triumph for the new voting system and perhaps "genius" by the Liberals to snare a fifth spot by creating opportunity for fourth Liberal and Colbeck to stay in the race which would never have existed otherwise.

    ReplyDelete
  12. With count now up to 83.4% and near 25% of votes allocated to Group or individual candidates the % of votes for both Singh and Colbeck seems to have dropped to about 0.55 quota for Singh and 0.35 quota for Colbeck. Seems to be bouncing around so there could be some non-random processes occurring.

    Given both Singh and Colbeck only collect preferences from the c20% BTL voters (but no doubt get a good proportion) then it is looking less likely for Colbeck and more likely for second Green unless some stronger than expected preference flows occur with the smaller groups. Having said that if Colbeck is eliminated then those preferences could help FF or other centre/right group if they are ahead of Colbeck at that stage??

    ReplyDelete
    Replies
    1. I am checking on this but a common problem in the count has been that some booths are added to the totals with a split into Ticket Votes and Unapportioned and without the breakdown of candidate votes, causing the apparent candidate votes to be unreliably low. I will have accurate figures up soon.

      Delete
    2. Yes again we have over 6500 votes in Franklin that are partially allocated in this way rendering the overall allocation grossly misleading.

      Delete
  13. Kevin,
    My experience of polling officials advising voters to number 1-6 above the line or 1-12 below the line, meant that the "only" was implicit rather than ordered. I'm speaking essentially of mobile booths - nursing homes etc.
    I didn't scrtuineer, so I have no first hand information about how voters treated the Senate. My guess is that the numbers doing more than the minimum either atl or btl would be miniscule. Fortified by your advice, I went 1-116 in Victoria, with the usual intense contest for the feted #116.

    ReplyDelete
  14. You must have influence KB - Tas count has reversed to previous position?

    ReplyDelete
  15. Something like that - it's extremely hard to follow.

    ReplyDelete
  16. Now count gone back similar to the 83.4% situation with lower % for Singh and Colbeck as per 2.54 pm comment.

    ReplyDelete
  17. Hey Kevin, do you know if they did any counting today (Tuesday) for the Senate in Tas?

    ReplyDelete
  18. Over 28,000 ballots were scanned and a substantial proportion of those data-checked but to date there has been no update posted on the candidate BTL totals as a result.

    ReplyDelete
  19. Thanks Kevin.. Look forward to seeing results tomorrow (hopefully)

    ReplyDelete