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

Twitter: @AU_Truth_Seeker

Tuesday, 10 September 2013

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

Welcome to my latest daily update of Monte Carlo simulated Senate forecast.

The forecast summary is as follows:

Unsurprisingly, no change to the "most favoured" outcome.

However there are some interesting signs of life in the Monte Carlo simulations:

I will have a look on a state by state basis the change in votes in the last 24 hours. This will hopefully become a regular daily feature.

As a usual caveat, these numbers are representative of only above the line (ATL) votes, and assume that all below the line (BTL) votes follow the same pattern as ATL votes. For some states, this may be misleading... See Tasmania down the bottom of this post.

Yesterday: 69.63%
Today: 70.11%

Change in primary votes:
LNP -0.01%
ALP -0.03%
LDP: +0.02%
All others: +/-0.01% or zero

Change in election likelihoods:
Reduction in likelihood of LNP3 (Arthur Sinodinos) from 85% to 80%.
Increase in likelihood of election of DEM (8% to 12%) and Shooters (4% to 6%)
This reduction in LNP3 is outside the MOE and the second day in a row that this has happened. Also, it is worth noting it has happened off an increase in counted vote of a miniscule 0.48%. I wouldn't be encouraging DEM supporters to get excited at this point, but it is worth noting that the DEM primary vote is actually below that of the now infamous Sport Party.

(Note that my model is effectively taking Monte Carlo samples around a mean - expect an MOE of approx 3% @ the 95% confidence level)

Yesterday: 65.68%
Today: 67.41%

Change in votes:
LNP +0.36%
ALP -0.24%
GRN: -0.17%
AMEP: no change
Others: minor and inconsequential

Change in election likelihoods:
No change - my model is still spitting out 100% to all six senators. This is despite there being a decent new 2% of the vote counted today and the composition of which is, by backsolving, 53% LNP and just 24% ALP.

Yesterday: 66.38%
Today: 66.75%

Change in votes:
LNP + 0.05%
Others inconsequential

Change in election:
No change, apart from a bit of statistical noise

Yesterday: 61.75%
Today: 62.19%

Change in votes:
LNP +0.02%
SPORT, RUA, NCT: no change.
All else +/-0.01% or zero

Change in election likelihoods:
SPORT up from 68% to 71% likelihood, but this change is minor compared to the MOE. Importantly, no change (to 0.01% level) to SPORT or the two parties it needs to be ahead of on a critical election path - NCT and RUA. Correspondingly, PUP likelihood goes from 28% to 25% and ALP from 12% to 10% while GRN from 88% to 90%.

Yesterday: 71.46%
Today: 74.48%

Change in votes:
LNP: +0.40% (biggest daily increase of anyone in my series to date)
ALP: -0.16%
XEN: -0.05%
GRN: -0.03%
PUP: -0.04%
FF: -0.02%
All else -0.01% or 0%

Change in election likelihoods:
GRN and FF firming slightly towards 100%, NCT increasing from 5.7% to 6.5%, but this is not overly significant.

Yesterday: 79.46%
Today: 81.97%

Change in votes:
LNP +0.23%
ALP +0.02% (surprisingly against the national decreasing ALP trend for the second straight day)
PUP -0.11% (another reduction - this implies that the cumulative average vote for PUP as counted in the last two days is just 2.9%, consistent with polls showing the Palmer surge was very late)
SXP -0.04%
All others movements rather insignificant

Change in election likelihood:
The likelihood of PUP winning is definitely decreasing - at the final stage of elimination they lead the LDP by 0.07%. If they fail, then a liberal senator is elected.
Also, there is another scenario. If, at count 21, SXP sneaks ahead of ALP3, then ALP and GRN surplus flow back to SXP, catapulting into victory. This gap is presently 0.24%.
So, overall, the likelihood of PUP winning the final Tasmanian senate seat is waning, likely to be replaced by LNP. But don't rule out the SXP, especially if ALP vote declines from here on.

The other factor is the high BTL vote in Tasmania, compared to other states. Quantitatively, this is 11% according to Kevin Bonham. In my opinion, this high is driven by voters who in State elections are used to numbering many boxes in the order of their preference and wanting to choose their own destiny.

In summary, my prediction now changes from one of "PUP win" to "too close to call".


  1. Hi TS:
    If you read Kevin Bonham's analysis of Tasmania you'll see that PUP winning is very likely as the LDP and SXP rely too much on preferences.

    1. I have a lot of respect for Kevin - he's one of Tasmania's leading psephologists.
      Yes, I've read Kevin's article, and commend him for his thoroughness. I encourage all readers to read Kevin's article.

      In tonight's update, I will delve more into the vagaries of BTL votes, and try to use data from 2010 to come up with some estimates of BTL leakage. I still believe, at the moment, that this is "too close to call".

    2. PUP did take a nasty whack on ordinary votes this morning, which raises the, in my view still rather slim, chance of them still missing the boat at the key tipping point.

    3. The key concepts we agree on. Haven't analysed the numbers yet though to quantify it. Will try to do so tonight.

  2. Do you know why the counts slow so much at this point. I would not have expected over 30% of the votes to be BTL or postal/pre poll.

    1. Btl will be low, but counting hor votes will take preference. Big percentage perhaps 20-30 will be postal or pre poll imo.

      A colleague of mine worked in a pre-poll counting centre on saturday night. Apparently she spent about 7 hours just flattening Senate ballot papers.