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

Twitter: @AU_Truth_Seeker

Wednesday, 11 September 2013

Senate results - updated - 11-9-13 - 10pm

Welcome to my latest daily update of Monte Carlo simulated Senate forecast. Today, my “10pm update” comes 2 hours early, because I’m tired, so it may not represent the end-of-counting of the day.

The forecast summary is as follows:

Guess what? No change to the "most favoured" outcome, again. Note that in Tasmania, my model is giving a greater than 50% chance of success to the Palmer Party, for the moment... I believe this is due to my model auto-updating to 100%, whereas in actual fact the BTL votes for ungrouped candidates would otherwise be excluded. This ever so slightly affects quota, creating a minor discrepancy.

Major changes by state:
NIL, apart from the Tasmanian issue.
Note that Kevin Bonham's blog represents the best discussion of the impact of below the line votes.

Noteable vote changes:
VIC: LNP+0.20%, ALP -0.08%, GRN -0.07%.
WA: ALP +0.13%, GRN -0.09%
TAS: LNP+0.21%

Minor changes to chance of election by state:
VIC: AMEP now less than 100%. LNP candidate (Helen Kroger) now has a miniscule chance, according to my model.
WA: ALP now slightly more likely to defeat the GRN, SPORT slightly more likely to defeat
TAS: Discussed yesterday. In practice, although the PUP vote is in freefall, it still may have the best chance of winning the last seat. I still think, however, this is "too close to call".


  1. I just wanted to say you are awesome. I have been wanting to understand what is happening in the senate for a long time and there was absolutely nothing decent on it until I found your via that piece in The Age. Its like all the conventional media simply ignored the senate because they had no way of understanding it. Even Antony Green effectively had nothing. This has been extremely frustrating because I knew that it was comprehensible to someone abke to do the analysis as you have shown.

    Considering the senate is the body with the greatest potential for a varety of outcomes and that it is much closer to a real democratic representation of Australia it has been a real failure by the traditional media to not cover it.

    1. +1
      I've had a strong interest in these ridiculous preference flows since a vote for Labor in 2004 helped get Family First elected, and the analysis since then has been pretty limited other than your blog and a couple of others for this election. Antony has always been good at talking about how stupid the system is (see youtube clip with Virginia Trioli) but I don't think he has time to focus on Senate outcomes.

      Perhaps if there's no change to the Senate election process (god forbid), then maybe the newspapers & pollsters will think about putting a small amount of resources into Senate polling and potential outcomes.

      And just wondering if you thinking of "open sourcing" your excel sheet post the election? Antony's Senate calculator is good but just doesn't account for uncertainty.

    2. Shucks! You guys flatter me. It's 100% in my spare time, owing to a very busy job, but it's a bit of fun.

      Yes, I agree the system is silly. Democracy is not having minuscule parties elected with three umpteenths of the vote - unless you live in a third world country.

      As for my model, my extensive experience with excel modelling is that if something is never designed for external distribution, it is hard to back-solve it there. It's got so many side calcs hanging off it now that it would be next to useless for anyone else. Besides, I'm very bad at putting comments into my VBA code....

      So, I think I would rather not do it, as I have zero spare hours a day to provide support to the model, and would fear people misusing it and misquoting it to promote non-quality controlled outcomes.

      I have made my theories very open, and I suspect this is enough for any Advanced excel user to replicate, using a bit of VBA for the simulations.

      Sorry - it's not a "no" but i see personal downsides which outweigh communal upsides, at this point in time.

    3. No problems, I've had the 'joy' of trying to debug undocumented VBA macros that broke with new versions of Excel and it's not a fun thing to do.. For me it would probably be easier to run scripts against the ABC website and scrape the results, but am thinking that would probably breach their terms of use if I did 1000 simulations. Hopefully this kind of analysis isn't required with the next election anyway, fingers crossed..

    4. Agree that it would be ideal if it was web linked. I note other websites, such as this one: do dynamically link into the ABC calculator, but I am not sure how to do so as I've never done any HTML work/modelling/analysis.

    5. And yes, would rather not have these issues next time, hoping that the system is reformed by then.

  2. Sex Party closing in in Tasmania! I have redone some aspects of my model and now believe they will take the calculator lead. At this stage, not by enough to win.

    There is also some very odd stuff going on in Tas with votes disappearing off the screen, eg Denison had nearly 5000 postals processed earlier, now showing <3000.