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

Twitter: @AU_Truth_Seeker

Wednesday, 2 October 2013

WA Senate - rolling observations post Distribution of preferences

Keep coming back here this evening for oddities regarding the WA Senate election.

(Updates in reverse order, so hence the "reverse donkey" numbering)

6. If all WIKILEAKS voters had trusted their party's unusual preferencing of the Sports Party and Nationals, GRN and Sports would have been elected.

5. If all BTL votes for #2 candidates for the Christians and the Shooters had gone to their #1 candidates, the Christians would have been ahead and GRN/SPORT likely winners.

4. Here's a graph of the progressive (non-ticket) margins for the Shooters v Christians battle.

Most observations are obvious:
a) Shooters held a slight lead on its ticket votes (incl 3689 Australian Independents and 5516 Fishing and Lifestyle ticket votes) compared to the Christians (incl 1404 No Carbon Tax ticket votes)
b) Christians gained strongly on its BTL votes and Shooters on its BTL votes, but...
c) Shooters "lost" 16 votes from its #2 candidate, the Christians "lost" 33 votes from its #2 candidate including 3 that went to its opponents.
d) Strong flows from Rise Up and Family First to Christians
e) Moderate flows from Fishing and Lifestyle and Motoring Enthusiasts to Shooters.
f) Surprisingly, very strong flows from the Sex Party to the Shooters. I would have expected a small flow instead in line with 2010 results.

3. Wikileaks BTL votes - 71 to Shooters, 41 to Christians - it's official. Despite the best intentions of WIKILEAKS preferencing crew to stop the greens, it was their BTL voters who inadvertently helped ensure Ludlam wasn't elected.

2. There were 19 BTL votes for "Stop the Greens" that gave their next preference to the Shooters and Fishers. I think we can say they achieved their aims.

1. If 15 Greens voters had voted for the Christians, GRN and PUP  SPORTS could have been elected


  1. The AEC stated in its media release this afternoon re count request: "As with all aspects of the count, the automated distribution of preferences undertaken this morning was open to scrutineers appointed by the candidates."

    Is this an indication from the AEC that closeness of the count is not a valid reason, there has to be proof the count is wrong?

  2. It would seem that they are positioning themselves for an excuse to say "no recount". I'm just not sure what the courts will think of this.

    And bugbears of a nameless pundit that the BTL file is not continuously available may not work in the AEC's favour.

  3. That wording appears in the press release for various states, not just WA.

  4. If I read it right, they are going to ask for a specific recount of the relevant votes to determine the Christian v shooters race...probably only the ticket votes......ludlam seemed quite specific in his wording

    But as Kevin points out how can you actually do that without counting every ticket vote again.....but maybe you can......

    I still suspect this ends up in court and that informal must be re examined. It tends to be how close votes always end up....just look at Gore v Bush for precedent.

    Are there any precedents for this?

    1. There is no Senate precedent, that I'm aware of, but I am no Senate historian or leal expert - I'm just a humble financial modeller who has no employment history in Politics.

      I think the precedent is McEwen 2007 ( where a judge personally analysed and interpreted 643 votes and "changed" the status of a few dozen votes. It would be interesting to see a judge and a bunch of lawyers drool over some 1.3 million votes! Every "Is it a 1 or a mosquito poo?" or "has the grey lead got fainter over the last month?" little debate could be significant.

  5. I've seen it noted that had Wikileaks preferenced the Animal Justice Party instead of the Sports Party then Animal Justice would have gone over HEMP causing Green and Sport to win. So there's another irony for the files - the Wikileaks Party caused the Sports Party to lose (assuming the loss is confirmed) by preferencing it.

  6. There's not a lot you can actually do, see or complain about once it gets to the "button press" stage anyway. And as for the ability to scrutinize 50,000-odd BTL preferences once the data entry is complete... What anyone thinks the scrutineers can work out from that is beyond me. We can maybe test whether the AEC's software has a bug in it or not, but not whether those 50,000 accurately reflect the ballots that were filled in.

    A far more likely error is that one bundle of ATLs could have been skipped in one count, much more mundane but more realistic.

    Remember that WA is about 15 times the size of Fairfax so a margin of 14 would be equivalent to less than one vote when presented at the same scale!

    1. Yes and no. Final result between shooters and Christians was 23,515 v 23,501. One could say it was 14 votes out of 47,016 or 14 votes out of 1,400,000

    2. Any 14 vote variation could matter - for example, and thanks to preferences, a bunch of votes that were put to Group P, instead of Group O could matter, even though this is not a contest between either Group P or Group O.

    3. Those votes would have been checked a number of times already, both at the booth and at the various electorates count centers.

      The AEC enters all BTL votes twice into the system and barcodes them. If there is any discrepancy in the data entry the ballot is reentered a 3rd time.

      I'm not saying that if they recount that some votes wont change here and there, but the AEC would have a legitimate excuse not to recount if the Greens cant point to flaws in the count.

  7. Truthseeker, when you say that if 15 Greens voters had changed their votes to Chirstians then a Greens and a PUP would have been that right? I would have thought it was the usual Sports/Greens solution....I wasn't aware there was any other combination that would work?

    1. Amateur error. Sorry. I have fixed it up.

  8. BTW all congratulations to Truthseeker for his attempt at calculation of the result and actually coming in very close on a number of counts.The methodology seemed to hold up very well and may be useful in the future. Well done.

    1. As one anonymous person to another - thank you! I wish I had more time to do this properly, but my responsibilities of a busy job and busy family have to take precedence.

      But this is not the end of my blog. I plan on posting ~weekly with additional information regarding the analytical side of elections.

  9. AEC declines recount