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

Twitter: @AU_Truth_Seeker


Sunday 29 September 2013

Updated WA Senate BTL flows

As I continue to refine my model for estimating BTL flows, I now add additional information derived from analysis of WA's 2010 Senate election.

Combining this information with the TAS 2013 Senate estimated still results in a lead to the Shooters in BTL votes, but only of 76 votes instead of 136 votes. Hence, ALP's Louise Pratt and PUP's Zhenya Wang retain a narrow advantage in the 2013 WA Senate election.




In my previous post, I published a table that listed estimated BTL vote percentage flows to the Shooters and the Christians.

In adding extra analysis of the 2010 Senate election, the following table best represents current point estimates of preference flows to the Shooters and the Christians.

BTL votes
Shooters
Christians
Gap
Smokers Rights
599
15%
5%
+60
Socialist Equality Party
158
5%
3%
+3
Australian Voice Party
55
6%
14%
-4
Sex Party
1660
8%
3%
+81
Secular Party of Australia
482
5%
2%
+13
Australian Independents
351
7%
17%
-33
The Wikileaks Party
1624
6%
4%
+29
Katter's Australian Party
396
17%
11%
+24
Family First Party
472
3%
32%
-140
No Carbon Tax
90
14%
6%
+7
Stable Population Party
311
1%
2%
-3
Stop The Greens
139
18%
6%
+18
Australian Democrats
573
4%
4%
-3
Animal Justice Party
710
6%
4%
+13
Australian Fishing & Lifestyle
211
43%
5%
+80
Australian Motoring Enthusiasts
410
19%
3%
+64
Rise Up Australia Party
382
3%
39%
-140
One Nation
415
13%
12%
+5
9038
+76

In summary, the addition of WA data where applicable, currently weighted 50-50 with the 2013 TAS data has sliced 108 votes off my expectations of the Shooters BTL vote gain. (Interestingly, applying the 2010 WA flows without giving the 2013 TAS votes a weighting gives a narrow preference loss to the Shooters)

Key changes involve FF and RUA vote - these went more strongly to the Christians in 2010 than in 2013 in Tasmania. However, I have had to assume that flows in WA in 2010 to the CDP is equivalent to forecast flows in 2013 to the Australian Christians. Given the CDP represented a strong(er) name with greater name recognition, there is a risk that I am overestimating the flows to the Christians. My previous estimate of FF--> Christians vote from TAS was very low as almost all of the FF vote went to the Liberals instead.

However, I have applied an arbitrary -25% variation to the Shooters vote in all WA 2010 estimates - this is to allow for some Shooters party leakage to the Sports Party.

Recall that this "additional" BTL impact is in addition to "locked in" or guaranteed votes. There has been no additional votes counted today in WA, so this means the point estimate is an implausible +28 votes. This is 0.002%. Given my estimates, I would have to apply a variance of +/-100 votes. Accordingly, this could go either way.

Assessment: (92.89%) As per yesterday, the Shooters maintain a slim lead over the Christians which if continued would enable a victory to ALP's Louise Pratt and PUP's Zhenya Wang. But the additional votes to be counted will be few but important. Any margin of error in BTL calculations would exceed the current forecast margin.

6 comments:

  1. Would the Shooter and Fisher party having a MLC in the state parliament have any impact on BTL preference flow? I guess it would be too hard to model.

    ReplyDelete
    Replies
    1. Yes, too hard to model. When doing a 10 party preferred analysis, it is not the whole "least worst" party that you often find with 2PP - the average preference number that defines the preferred party is 4.1. So, it is usually the 2nd preferred party, or often the 3rd preferred party.
      If SFP have a member of parliament who is even half popular, then it would help people think "I'm putting SFP 2nd".

      Still - impossible to quantitatively model, so I can't include it in my analysis. But it may explain an element of variation in the post-count analysis, if such variance exists.

      Delete
  2. Thanks Truthseeker. I guess it is really hard to model. One thing I notice is that you have the Christians getting much more of the BTL votes for the Aust independants than the Shooters. However as we know the Independants card preferences the Shooters. This seems to make a big difference in your model...your thoughts? I understand that people voting below the line are implicitly rejecting the card but even still.....

    ReplyDelete
    Replies
    1. The point of estimating preference flows is to deliberately ignore the submitted group tickets. In the case of AI I'm estimating percentages based on known preference flows in Tasmania. The margins of error are high for some of these estimated flows, but these point estimates certainly represent my best guess based on actual data.

      Delete
  3. I've seen articles from a few days ago giving the button press time as 11 am (WA time) Wednesday.

    ReplyDelete
  4. Have you factored in the segmentation distribution and the method of calculating the surplus transfer value? This can add over 10,00 votes to a candidates total. IN Victoria 2007 the distortion as a result of the Surplus Transfer value was eqiv. to 70,000 votes

    In QLD 2007 segmentation resulted in the wrong candidate being elected. Try recounting the QLD vote excluding all by the last seven candidates remaining (3 ALP, 3 LNP and 1 GRN). This simulates a reiterative final count.

    Votes for excluded candidates redistributed as if the excluded candidate(s) had not stood

    Does not matter which system of calculating the surplus transfer value is used. The Greens should have been elected to the QL:D Senate in 2007

    IN 2010 the method of segmentation could give the Sports/Greens parties the edge. We will know when the AEC finally published the BTL preference data. In 2010 it took them months., This time i think it will be quicker


    ReplyDelete