Blogging Senate forecasts and results in the WA Senate re-election until officially declared.

Twitter: @AU_Truth_Seeker

Tuesday, 24 September 2013

Senate update - 23-9-13 - 10pm update

Well, what a big day it's been in Senate counting - the AEC has updated Senate results from 146 of the 150 electorates.

While I am busy at work, I will be tweeting any odd or revised Senate outcomes so follow me @AU_Truth_Seeker

Here is my estimation for the overall Senate makeup:

Firstly, this relies on an assessment of WA and TAS, which remain very uncertain and volatile. But if we can move past this, recent counts point to the fact that ALP/GRN/PUP may well have together 38 Senators - enough to block legislation. While for some legislation, this may not matter (can't see Palmer voting against to block a Carbon Tax repeal, for example), it may matter if Palmer decides to play an obstructionist card to leverage alternative policies.

WA (90.19% counted)
Reports of the change in elected Senators appears correct, if there were to be no further votes counted, as the Shooters currently lead the Christians by 79 votes (calculator says 346, but is overstating actuals). This means that PUP's Zhenya Wang and ALP's Louise Pratt have overtaken the Greens' Scott Ludlam and Sports' Wayne Dropulich.

An earlier elimination than "Will the Shooters catch the Christians?" is the Animal Justice vs HEMP. If HEMP falls behind Animal Justice, this strangely locks in the election of Ludlum/Dropulich. But currently HEMP is leading by 183 according to the calculator, but a large 486 if we correctly account for BTLs. This margin is actually higher than yesterday (by about 150 votes), reversing a narrowing trend of the last few days.

A later elimination is PUP vs LDP, with the winning party to leapfrog the other onto the Red couches. Currently, PUP is winning by 700 votes (according to the ABC Calc) but by 1600 votes once BTLs are considered.

Assessment: With the AJP v HEMP gap widening, this appears to increase the likelihood of ALP/PUP.

NSW (91.17% counted)

The NSW Senate race is perhaps the most complicated. See my previous post for information on the critical counts.

Basically, although the critical counts vary in "Calculator" magnitude of 227 to 1138 votes, the actual margins, including BTLs, all vary within the narrow range of 555 to 1011 votes.

I have set up a spreadsheet with all the data auto-feeding from online and will tweet and post if it gets interesting tomorrow.

Assessment: Looking increasingly likely for 3 LNP, 1 LDP, 2 ALP. The outside chance is now more likely to be Shooters and Fishers, not Democrat, as LNP vote has reduced significantly recently enabling the exclusion of Sinodinos to elect SFP. This remains, however, unlikely.

Instead of recreating the wheel, I'll point you in the direction of Kevin Bonham's specialised Tasmanian Senate analysis.

In summary, this one is way too close to call, with too many BTL votes throwing a degree of randomness that simply cannot be modelled accurately. If I had to take a punt, I'd suggest PUP's Jacqui Lambie is slightly more likely to win the final spot.

Other states/territories
To be honest,  I cannot see any real likelihood of any of these states producing different outcomes. I am keeping a watching brief on SA, in particular, but believe the margins to be insurmountable.


  1. I also now think PUP best placed in Tas but there is so little in it. Enthusiasm for the Sex Party's 795 vote calculator lead faded on close analysis.

  2. WA: The prevailing thinking seems to be along pretty clear lines, it's either Dropulich+Ludlam or Wang+Pratt. I thing Dropulich+Pratt is out of the question but I wonder if Wang+Ludlam is still a possiblility.

    If all votes were ticket votes, obviously not. However there are currently (91.05% counted) 28,333 BTL votes and 50,563 unapportioned votes (most of which will likely be ticket votes, but not all). The margin between Pratt and Ludlam at the very final count (using the ABC calculator) is now 6,830.

    Of those 28,333 BTL votes, ignore the votes going directly to Ludlam (6,976) and Pratt (1,145) - those are "known". That still leaves 20,212 "unknown BTL" in play. On the assumption that the race comes down eventually to 2 candidates, the relative positions of Ludlam and Pratt on each of those tickets could be important. Certainly a large fraction of those could be diminished by transfer values but I think there's still enough uncertainty at the moment not to call it either way.

  3. You need to take into account the value of the BTL's which will be used when 5 out of 6 senators are elected. For example taking the first 3 Liberals BTL votes (3146) + Wang (1376) + Bullock (3040). These vote will be transfered at a almost meaningless value.

    Which leaves 12650 BTL votes floating around at full value. Ludlam would need to win around 9000-1000 of these to overtake Pratt. Highly unlikely.

    1. Sorry. 9000-10,000.

    2. Hmmm... I'll try to add this to my spreadsheet on Thursday night's update... It's not the most unlikely situation I've ever modelled....

  4. I have to disagree... if someone voted for Bullock but then NOT for Pratt (let's say the religious right of the ALP could have some particular reasons) then those 3040 votes currently have a TV of 0.4633, making that bloc worth 1408 alone. That's nearly half the gap closed when you consider that each vote that flips reduces the margin by 2. A TV of 0.4633 is not almost meaningless.

    Perhaps less plausible but mathematically valid, anyone who voted for Johnston and then leaked out of the Liberal ticket would find their vote transferred at 0.6384 - so there's another 2008 votes there. I don't expect that many of those would have leaked out of the ticket, but I'm going to argue the maths - that's more than meaningless.

  5. Say 10% of Bullocks BTL votes are religious right wingers who wouldn't put a lesbian before a far left green (and I think 10% is very generous) and assuming none are distributed to the 3rd liberal candidate or the Palmer candidate before the Greens, that is a total of 139 votes.

    If over 50% do vote Bullock and then skip Louise for far right religious reasons, it’s almost certain that they will elect the Palmer Candidate and/or the 3rd Lib before the final distribution.

    On the Liberal ticket, the vast bulk of those BTL votes will be used to elect three liberal candidates and then maybe a Palmer candidate. The numbers of people who would vote 1 Johnston and then Ludlam before the other Libs or Palmer would be insignificant in the scheme of the count.

    It’s fair to assume that (and im happy if anyone has evidence on this to prove me wrong) that over 90% of people voting BTL for a candidate, will continue to number the other candidates in that party 2, 3, 4.

    I think it’s likely that these excess votes could be enough to sway some of the minor party contests, which could win Ludlam the seat but not enough to change the final distribution.

    1. BTL:You see most 1,2,3,4 down the sheet. Second most is 4,3,2,1. IF the one is in the second or third spot, those often squirrel away to other parties for at least one of the other candidates in the list. In Tassie for instance, there are a lot of Lib voters who use BTL to exclude Colbeck for some reason.

    2. Once we can get the full list of BTL votes, I'll be doing this sort of analysis. Not tasmanian so don't know about the relative popularity of their senators. Also remember that Tasmanians are used to expressing their opinions by voting in multi-member-electorates the way THEY want to vote - explains high BTL percentage in this state.

  6. I've started a new thread for posting about the outcome and the preference distribution tomorrow:

  7. I definitely enjoying every little bit of it and I have you bookmarked to check out new stuff you post."