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

Twitter: @AU_Truth_Seeker


Wednesday 18 September 2013

Senate update - 18-9-13 - 10pm

A lot more counting in the Senate today, as we move closer towards having some more results

NSW:
My monte carlo simulations are showing the following probabilities of election for the final seat:
LNP: 82%
DEM: 9%
SFP: 9%

At the heart of this, as explained earlier, is the gap between AMEP and FF at an early stage of elimination. Today, it has reduced from 480 to 440.

WA:
Dropulich (Sports Party) continues to poll well in late counting, particularly absentee votes. Given WA has a high percentage of absent votes compared to other states, I wonder if this is due to youngish men on FIFO mining jobs voting outside their electorates? Hence, would it make sense to consider these voters to be more of people feeling disenfranchised, and more likely to throw a vote to "Sports Party?" The vote for Sports in WA Absent votes is significantly higher (0.33% vs 0.22%) so unless there is a very high "Voice Party" BTL%, WA is looking increasingly safe for Dropulich.

TAS:
Appears extremely close, between PUP and LNP for the final spot.

QLD:
Looking quite safe - 3 LNP, 2 ALP, 1 PUP

VIC:
Looking safe. Monte Carlo model showing 100%, indicating near impossibility of any other result.

SA: (updated)
Monte Carlo analysis of this state shows a 16% chance of election by No Carbon Tax. However, this probably overstates NCT's chance of election. At the end of count 29, Family First's Bob Day (4.00%) exceeds Lib Dems of 3.75%. Unusually, these are mainly all ticket votes so there's not too much BTL action to consider. If anything, it is unlikely that other parties' BTLs will flow to Lib Dems over FF. On the other hand, Lib Dems seem to be doing better on Declaration votes as is the case in NSW. On balance, it is unlikely for there to be a switch here, so Bob Day looks relatively safe.


ACT/NT
As outlined previously, very safe 1-1 in each case. This is despite ALP's Nova Peris in NT and the LIB's Zed Seselja in ACT will receive enough preferences to win.


8 comments:

  1. Why do you hate SA :(

    ReplyDelete
    Replies
    1. SA is important - it's got almost the same number of voters as Western Sydney! :-)

      Delete
  2. Howdy... In your MC simulations, are you reducing the margins of uncertainty for each party at all as the count progresses?

    ReplyDelete
    Replies
    1. I intended on reducing variation %s progressively. I even trialled a formula that was based on the % of the count remaining. In the end, I have not reduced the %'s. This is for my own consistency so I can track how things are changing, and keep an eye on outside outcomes.

      But, I am trying to avoid putting too much attention on the raw percentages, as you will know I have stopped routinely publishing full tables of percentages. In the interests of full disclosure, i am using +/-3% for majors and +/-15% for micros (That is, 0.05% +/- 0.0075%). So, as you can see it is more indicative and allows me to scenario test, which was the main benefit of applying variation in the first place. Additionally, the main focus now is in applying variation in votes.

      I have also introduced a scenario tool which allows me to "flip" each elimination, and selectively choose the 2nd lowest or 3rd lowest candidate to see what would happen in an alternative scenario.

      Delete
  3. PUP seem to have picked up a bit in Tasmania today. I have them at -1249 vs LDP "by the calculator" which keeps them competitive (though I expect it to get worse again). The Bass EPPVs counted so far were harmful for them as expected so there must have been some EPPVs or adjustements from somewhere else.

    This could be going to the button as a modelling tossup. Perhaps by then Sex Party will be more clearly out of it; they are not building enough of a lead at the moment. FF do appear to be gone as they are making no progress.

    ReplyDelete
  4. Antony's blog points out a new critical point that I hadn't been previously considering, and it's now as close or closer than Sports vs RiseUp or Sports vs NoCarbonTax.

    The issue is between Australian Christians and Shooters & Fishers. If those flip, then [PUP + ALP2] are in and [SPORTS + GRN] are out.

    Being later in the count this may be more dependent on BTL votes but those could go either way.

    This count is fascinating and terrifying.

    ReplyDelete