Sunday, March 16, 2014

State Election Post-Count Thread: Braddon

This is the last of my state election post-counting threads.  There is now one for each electorate and they will be updated frequently.  During the cutup (which starts in the middle of the week after next) I will be on remote fieldwork and will update nightly, but hope to take a day off at the peak counting time.

SEAT OUTCOME:  4-1-0 (Count finished) 
WINNERS: Brooks (Lib), Rockliff (Lib), Green (ALP), Jaensch (Lib), Rylah (Lib)

Braddon is the most fascinating of the five preference distributions unfolding before us, with a real chance something quite unusual could occur.  The Liberals have racked up a whopping 3.53 quotas (nearly 59%), Labor is on 1.40, PUP have 0.43, the Greens have 0.40, the Nationals 0.13 (headed up by their rogue candidate Ken Dorsey!) and the rest a trickle.  

Adam Brooks and Jeremy Rockliff are both way over quota, with 0.91 quotas surplus between them.  Bryan Green is far enough ahead of the rest of the Labor ticket that he will be elected, though it will take him a very long time to do so.  

The Liberals will obviously get at least one of Roger Jaensch and Joan Rylah over the line and the question is whether they can get two.  Their rivals for this feat are Brenton Best (ALP), Kevin Morgan (PUP) and Paul O'Halloran (Green).



The Liberals are ahead on raw quota totals, but that will change when the two surpluses are thrown and votes leak out of the Liberal ticket.  However, this election isn't about party quota totals so much as it is about the positions of various candidates in the count.  The reason for this is that Bryan Green will be way short of quota, but so probably will be the third and fourth Liberals.  So if Labor is on 1.45 quotas at a given time, Bryan Green and Brenton Best might be on (say) 0.55 and 0.9.  Likewise if the Liberals are on 3.4 quotas, their last two candidates might have 0.6 and 0.8.

The Greens and PUP do not have access to this resource: they will just be left as single candidates with about 0.4 of a quota each.  PUP have even greater problems in that their vote is spread across five candidates with only half of it with Kevin Morgan.  Leakage from the rest of their ticket will be such that Morgan may even be cut out before O'Halloran.  Especially, Steve Green is a well known local alderman and a lot of his votes could well leave the ticket.  Also, the preference landscape for the Greens and PUP is very difficult - nobody in this count is likely to preference the Greens ahead of the other parties in it, while the Greens aren't likely to help PUP much even if PUP stay ahead of them.  So both these parties are struggling.

My expectation is that O'Halloran and Morgan should be eliminated, with a lot of exhaust from both the Green and PUP tickets.  This will leave a situation in which one of Best, Jaensch and Rylah is last and loses, and the other two are elected.

Making allowances for leakage and assuming the Labor preferences split evenly between Green and Best, the Liberal preferences split evenly between Jaensch and Rylah, and all other preferences split evenly between these four candidates, then I have Best effectively 1300 votes behind Rylah.  But given Best's form during the campaign I wonder if he might find himself isolated on preferences by his ticketmates, especially those voting for Dr Shane Broad.  So on that basis things could be worse for Best than they appear.  On the other hand, there is a view that Jaensch will get way over half the Brooks and Rockliff surpluses placing Rylah in danger of being behind Best.  There is also the possibility that PUP preferences (those that don't exhaust) will favour Best over all other candidates.

All up though, I think the Liberals have a slightly better than even reasonable chance on current primaries of pulling off an incredible four-seat haul in Braddon (if their vote goes down in late counting this assessment may change). It will be Hare-Clark history if it happens, since it hasn't been done since the five-member system came in in 1998.  That said, under the old seven-member system there were twice five-seat hauls in Braddon, and one of those (the Liberals in 1992) would probably have been a 4/5 result in the current system.

Update Sunday evening: Braddon now has 90.58% counted and today's added votes have not affected the balance between the parties to even the second decimal place of a quota.

Simulation Sunday evening: I have run some simulations for this using some very rough estimated preference flows.  The surprising result of this was that Morgan wasn't completely out of it - if he stayed ahead of O'Halloran (dubious) then O'Halloran would put Green over quota allowing Morgan to compete with Best on level terms.  If Morgan then outlasted Best then Best's preferences might allow him to catch the fourth Liberal.  I'm a bit wary of some of the assumptions here because Morgan is not an incumbent and might not perform so well on preferences as my estimates (especially my estimates that PUP might get more Liberal leakage than Labor).  And he has to kick a lot of winners along the way.  So I still think it is most likely between the major parties.

As for Rylah if the split between Jaensch and her on Liberal preferences is reasonably close to even she stays far enough ahead of Best and Morgan to win.  If it is dramatically in Jaensch's favour though, so that Jaensch is elected on the throw of whichever of Best and Morgan is eliminated, then it is back to a straight race on party quota totals, and one which she might lose as Best and Morgan could well help each other on preferences, and because of the surplus leakage issue.  Of course I assume Jaensch will do at least as well as Rylah (and I think he will) but we don't know that.

In my simulation Rylah beat Best by about a thousand votes, but that was assuming a split of Liberal preferences that was between even and proportional (thus slightly favouring Jaensch over Rylah).  Even if Jaensch does hit quota, it's still possible Rylah beats Best on raw quota total, but in cases like that the incumbent does tend to find some way to win.  In discussions of this count online I've a few times encountered the viewpoint that the split between Jaensch and Rylah will be nowhere near even - that Jaensch will dominate it - and if this is so then Best should win.

This is really not clear just yet - unless we get some quality scrutineering data - and we will need to see how Jaensch and Rylah are travelling after the surpluses are thrown.

More Monday 17 March:  I now think there's a good chance that Brenton Best can win this, unless his antics have caused ALP voters to desert him on preferences entirely.  As discussed in the Bass thread, when both incumbents for a party have surpluses, meaning that all the incumbents remaining in the count are from other parties, we should expect massive leakage from the surpluses.  I was using an 11% leakage assumption in the model I discussed above, but it's likely to actually be in the mid-teens at least and probably the high teens if not even 20%.  That takes maybe 400 votes out of Rylah's position on the model I discussed above (with its rather generous assumptions on the Liberal vote split), and it also means that the Liberals very probably can't beat Labor on quota total and can only win it by keeping both their candidates ahead of Best while having both short of quota.  Experience suggests that when these things get close incumbents are like cockroaches in cross-party battles with non-incumbents - if they can crawl out of a bad position they generally will.  (Update Monday 24th: turns out the original estimate of 11% was actually a very good one, so scrub the reasoning in this paragraph.)

A "part rechecked" total shows the Greens' position slightly improved to 0.42 quotas, which (if it holds) is bad news for PUP but largely irrelevant to everything else.

Wednesday 19th:  Both major parties are upbeat about their chances (Close go for fifth Braddon seat).  The Liberals are claiming that "Our vote seems to have stayed quite disciplined and stayed within the column" but no figures on this are provided, and as has been shown above the raw leakage rate is not the Liberals' only problem.  Still it would help them enormously if they could somehow keep it down to 10-12% or so.  There are no significant changes to the primary total, and this is not a race where fifty votes here or there is really going to help much in knowing what might happen.

I have become aware of very small scrutineering samples of Brooks and Rockliff preferences that had a lean to Jaensch over Rylah (but not a huge one) and leakage in the case of Brooks to Kevin Morgan.  If Liberal votes leak substantially to him rather than the ALP then that can put him more in the running.

Thursday 20th: Labor's rounded quota total has slipped from 1.40 to 1.39, aside from that not a lot new to see.

Friday 21st: And back up to 1.40 again. As noted above little changes to the primary total don't tell us much here.  Word on the street is that the TEC may start early on distributing the monster surpluses next week.

Monday 24th: I'll be having a very close look at this count in coming days but the provisional distribution of the Liberal surpluses has Jaensch just 832 up on Rylah, probably enough to give him the third Liberal seat but still fairly close.  We haven't seen really massive leakage here either (as in Bass the hold has been better than might have been the case with two incumbents elected). Leakage was only 10.6% and not the high teens figure considered possible (Labor got 47% of this leak with nearly all of that going to Green and Best).  That plus the closeness of Jaensch and Rylah has greatly strengthened the Liberals' chances of getting both up.

The Liberals are now last on spare quota total, but this is about candidates and they have a quite even split, so that's irrelevant.   Rylah is on .556 of a quota already and will get more when Howell is removed.  On the Labor side, an even split of preferences between Green and Best would put Best up to only about where Rylah is now; there is doubt whether he will get that and even if he does he would need a large preference boost off other parties to overcome the contribution of Howell to the Liberal tally.

Best will almost certainly at least stay in the count until one of O'Halloran and Morgan is gone.  As for which of those goes first, PUP are currently 239 votes ahead of the Greens.  But there are only 1248 Green votes sitting with minor candidates compared to 2420 sitting with PUP.  So PUP are more vulnerable to leakage and could drop back over 100 votes compared to the Greens.  If leakage from the PUP ticket is really bad they could go out before the Greens.  Even if they hold on they would have to be favoured on Greens preferences over Labor to go any further.

The Greens were excluded in this seat at the federal election and their preferences split ALP 69% PUP 22% Lib 9%.  So the idea that Greens preferences will help Morgan get over Best is far-fetched and it doesn't seem Morgan has any realistic chance.

Brenton Best's best hope from here - of all the ironies - is that he might be saved by Greens preferences.  If, say, 60% of the Greens' vote stays in the count then a similar flow to the federal election could be worth enough to see both Labor candidates home.  I'm sure the flow will be very much more muted than at the federal poll (in which preferencing was compulsory) so it will be interesting to see if it is even enough to erase Rylah's lead.  Obviously, Best's position against the party makes him the least likely Labor preference recipient from the Greens.

Tuesday 25th: Today we've seen a number of minor exclusions including three Greens and two PUP.  PUP are nearly 300 up on the Greens but still much more leakage-exposed; all the same O'Halloran should still be excluded before Morgan.  (It seems it matters little).  We haven't really learned a lot else today except that Labor seem to be picking up some decent bundles from the minor players and leaking Greens votes go to Bryan Green rather than Brenton Best.  Antony Green thinks the evidence that the Liberals will get four is rather strong; I'd like to see where Best is when only Bryan Green and Best are left on the ALP ticket first, and probably the Greens preference flow before too strongly discounting Best's chances of hauling himself up the mountain.

Wednesday 26th 1:25 pm: This count is moving rather slowly but we've just seen the exclusion of the National Party's final candidate, with his preferences helping Labor.  The overall picture hasn't changed greatly and it could be well into this afternoon or evening before we know much more about where this is going.

5:40 pm: We now have the exclusion of the fifth Liberal, Kyron Howell, to come. Howell's votes may well put the identity of the third Liberal winner beyond any doubt (Jaensch is currently 806 ahead of Rylah.) Best is currently 1737 behind Rylah but has access to 900 more votes from his own party than she does, so he should be able to close that gap by at least a few hundred, by the time both Howell and Broad are excluded. There will then be over 9500 Greens and PUP preferences thrown, but many of these will exhaust.  I think this is still quite unclear given what we have seen of the behaviour of Greens and PUP preferences so far - yes these votes go to Bryan Green more to Brenton Best, but do they also go to Best once Green is elected with quota?

6:40 pm: Following the exclusion of Howell, the Jaensch-Rylah gap is down to 719, meaning we still can't be totally sure which Liberal will win if they win only three.  However, it is a large gap to close on Greens and PUP preferences so it should still be Jaensch.  Again the Liberal ticket is holding a very even split between its two candidates which is very much what Best doesn't want.  The good news for Best is that once Green crosses quota on preferences from O'Halloran, Best will be the sole Labor preference target left, while preferences going to the Liberals will still have to split two ways.

Note that at the moment Labor has 1.54 quotas and the Liberals 3.51.  So it seems that Labor are closer to their second quota than the Liberals are to their fourth.  But assuming Shane Broad's preferences split evenly between Best and Bryan Green, then if the count was ended at that point Best would lose.  Before the election I speculated Labor might hold two seats even with a sickly primary total if they could get an even split, thanks to the Ginninderra Effect (warning: ultra-wonky link).  Instead at the moment it is the Liberals who have a chance of beating them by the same method.

9:20 pm: Paul O'Halloran has rather surprisingly managed to catch Kevin Morgan and outlast him by nine votes, off superior leakage from Shane Broad (ALP).  Prior to Broad's exclusion, O'Halloran was 90 behind. Coming into the throw of the PUP preferences then the Greens preferences, Joan Rylah leads Brenton Best by 1539, and trails Roger Jaensch by 735.  It could hardly have been scripted better - the preferences of the Greens will be the final decider in whether the anti-Green Labor rebel keeps his seat.

The closeness of this exclusion may lead to some degree of rechecking before Morgan is thrown.  Anyway, it looks like the expensive, noisy Palmer United bid for a seat in the Tasmanian parliament ends here in absolute failure.

Thursday March 27 11am: Almost all the PUP preferences have been thrown and Best has made negligible progress - he still trails Rylah by 1504.  Worse still for him, Bryan Green has not yet crossed and has 592 to go to quota.   There are still 196 votes of Morgan to come.  Green will very likely hit quota and have a surplus on O'Halloran's preferences, but soak up votes likely to otherwise go to Best in the process. Most likely Green and Best combined will need to beat Rylah by more than 2000 (currently 2096) out of about 5300 Greens preferences, of which a substantial percentage (maybe 30-40%) will exhaust.  It's very tough, but it isn't impossible if the remaining core of Greens support is really anti-Liberal enough to preference even Brenton Best. 

11:22 am: The Morgan votes now being distributed are few in number, but they are all from the Liberal surpluses.  The first batch of 143 has given Brenton a nasty little kicking with more going back to the Liberals; he trails by 1523 with 53 to go.

11:46 am: The PUP preferences are finished and Best's net gain on Rylah off them was a whole seven votes.  He trails by 1532 and given his feeble effort on the PUP votes I really don't like his chances of getting enough flow out of the Greens either.

12:12 pm: I calculate roughly that a 72% share for Labor of the non-exhausting Greens preferences will be required to get Best over the line.  Given that it is Braddon where Greens preferences tend to split weakly, and that Brenton Best has given Greens voters so many reasons not to preference him, this seems extremely difficult.

2:19 pm: Hare-Clark history is made as Brenton Best has made significant progress on Greens preferences but in the end failed to catch up.  He is 1053 votes behind with only 1039 to go.  Joan Rylah will win and become the first woman elected to Braddon for quite a while (Di Hollister in 1996 was the last).  The Liberals will secure the first two-seat gain in an electorate under the 25-seat system, and the first four-seat result under it as well, something that is very hard to achieve.  And Brenton Best will join the list of victims of the dreaded Ginninderra - Labor has 1.79 quotas to the Liberals' 3.68, so it looks like Labor are closer to a second seat, but because Best is behind both individual Liberal candidates, he loses.

And there we are then - 15-7-3.

2:54 pm: Best 1064 behind with 938 to go.

3:25pm: A gallant late charge by Best but it was too late.  Rylah defeats Best by 451 votes.




13 comments:

  1. With my spread, I get Green elected on the elimination of O'Halloran, and Jaensh within 100 votes of being prior to the elimination of Morgan (after exhausted votes are removed).

    As you say, Best is a long way behind. I think that this will be four Liberals as well.

    ReplyDelete
    Replies
    1. I will be trying a detailed model of this too. What assumption are you making about the split of the Lib surpluses between Jaensch and Rylah?

      Delete
    2. I think it was an even split, but allowing for the donkey cascade.

      Delete
    3. Actually, not quite so. The original break up from the excess with Brooks and Rockliff was split in ratio with the votes received. There after, even split apart from the donkey cascade.

      Delete
    4. I also got Green in off O'Halloran but I didn't even quite get Jaensch in after the elimination of Morgan. (Had Best, Rylah, Jaensch all without quota with Rylah about 1000 up on Best). That said I was using more muted preference flow assumptions re Jaensch/Rylah, which may turn out to be completely wrong.

      Delete
  2. After the counting today, I've found I had been under a misapprehension as to the nature of the Robson rotation. It not only rolls but reverses, which disperses the donkey vote much better through the latter preferences. Also, the level of donkey voting is much lower than I'd anticipated; I'd been assuming about 80% of the votes of the lowest candidate in each party; today's counting suggests this should be at most 50%. I'll rerun my cuts when I get a chance, but I've stuff to do tonight and more counting tomorrow.


    ReplyDelete
  3. It just goes to show how sexist Braddonites are, imo. Joan Rylah has been campaigning strongly for far longer than all except Brooks. She banged nails into the forestry coffin all over the state, whinged about there being no breeding stock if too many young women leave the state and it still appears she might miss out.

    Annie (for some reason blogger won't let me post in personal name)

    ReplyDelete
  4. Looks like this will end up 4:1:0 now. Given the full tickets of both the Greens and Palmer, it would be a surprise to see enough votes not exhaust to change the standings of the remaining candidates. Bryan Green may or may not reach quota from the exclusion of Morgan and O'Halloran; he needs around 11% of what is excluded in those two counts to reach quota prior to the elimination of Best. The challenge for Best is even greater; he needs to pick up around 1500 more votes than Rylah from those two exclusions; can't see it happening.

    ReplyDelete
  5. In both Denison and Bass the final PUP exclusion slightly favoured Labor with 48% and 30% respectively exhausting. I don't have the exact figures for Franklin and in Lyons there were no Liberal candidates remaining at that point.

    It looks like the exhaust rate from the Greens' final exclusion in Franklin was very low with leakage plus preferences from the party strongly favouring Labor there. Shame scrutiny sheets for completed counts are not up yet.

    ReplyDelete
  6. The scrutiny sheets are up in the offices, just not online.

    I've had a couple thoughts about the process: one is that the system would work significantly better with a requirement to vote to 10 rather than to 5; two, that elected candidates really should continue to receive preferences and have those passed through. On the second, I can see why they don't in terms of time costs, but at the theoretical point when the calculation goes electronic, there are a few touches to the method that should be added (elected candidates not treated as excluded, exhausted votes reduce the quota, last candidate elected)

    ReplyDelete
  7. From the 5267 votes from the greens, Best got 471.

    Karma, baby.

    ReplyDelete
    Replies
    1. Actually that 471 was his share of Kevin Morgan's 4892. From O'Halloran's 5267 Best got 822 directly and another 659 via Bryan Green's surplus - not bad given that 2283 ended up in exhaust.

      Delete
  8. As a long time Green voter, I have found Brenton Best's denigration of the Labor-Green alliance bewildering and mainly the sounds of a man proving the old adage that empty vessels make the most noise. I directed my preferences to the Liberals in the hope of getting rid of this person who contributes negativity as his main form of discourse. We need a diverse range of open-minded and positive-minded individuals representing our hopes for a better future. he is not one of these and now can become a jobs detective for himself.

    ReplyDelete