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

Twitter: @AU_Truth_Seeker


Saturday 5 April 2014

Live Blog - WA Senate election (times in AEST)

2:30am Signoff
So, what have we seen? Early ABC figures showed a catastrophic rout for ALP and a sensational showing for the GRNs, however TruthSeeker figures never quite picked this up and instead showed a consistent record of massive swings from LIB/NAT/ALP to PUP/GRN.

For the record, my methodology for basing my swings was
1. List 2013 vote by party for each booth in an electorate
2. Calculate benchmark 2013 "base" expected performance for each electorate that depends on %counted. While we can't do a matching booth comparison, we did instead do a matching by vote counted, where the order of counting each booth within an electorate is assumed to be directly propotional to the number of votes (bigger booths take longer)
3. Lookup current vote counted for each party, compare to the expected vote, calculate swing
4. Add swing to 2013 vote
The proof is in the pudding - I'm happy with the stability of my estimates as the night has progressed.

My prediction is for a 3-1-G-P outcome at this stage, with a 2-2-G-P an unlikely but plausible. Despite HEMP vote exceeding my expectations, votes of the majors has meant that a HEMP harvesting operation has not materialised. Also notable is the NAT reduction - in a previous post I referred to the Wirrpanda effect whereby the NAT vote in 2013 was inflated by nominating a footballer. Despite NAT having a much better ballot draw this time, it still shed two-fifths of its votes! In fact, had NAT held its vote and indeed increased it by 0.2% due to a better ballot position, as would have been expected with Wirrpanda, it would have overtaken LIB3 and have got elected.

But the big winner is the big Clive Palmer. To increase his vote from pterodactayl size to brontosaurus size was a great outcome for him and his fledgling party.

My next post(s) will depend on the complexity of the count. If it gets a bit closer, you'll see a lot more of me!

Let's hope that no further reelection is required for this poll.

Here are my final predictions for the night:
LIB.. .. .. .. 34.12%
ALP.. .. .. .. 21.34%
GRN.. .. .. .. 15.82%
PUP.. .. .. .. 12.21%
NAT.. .. .. .. 3.13%
LDP.. .. .. .. 1.72%
GOD.. .. .. .. 1.54%
HEMP.. .. .. .. 1.08%
SFP.. .. .. .. 1.04%
SEX.. .. .. .. 0.96%

Outcomes:
Party     Elected
--------- ---------
LIB    :     2.638
NAT    :     0.000
ALP    :     1.362
GRN    :     1.000
PUP    :     1.000

Easy.



1:38am
58% counted
LIB.. .. .. .. 34.11%
ALP.. .. .. .. 21.23%
GRN.. .. .. .. 15.90%
PUP.. .. .. .. 12.22%
NAT.. .. .. .. 3.18%
LDP.. .. .. .. 1.72%
GOD.. .. .. .. 1.53%
HEMP.. .. .. .. 1.09%
SFP.. .. .. .. 1.04%
SEX.. .. .. .. 0.96%

Party     Elected
--------- ---------
LIB    :     2.514
NAT    :     0.000
ALP    :     1.442
GRN    :     1.000
PUP    :     0.996
HEMP    :     0.040
WIKI    :     0.005
VEP    :     0.003

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.510
20-2-110    :     0.438
20-1-111    :     0.048
30-2-100    :     0.004

1:30am
And it just gets closer and closer
Here it's worth considering the relevant flow of BTL votes
But again, both ALP+GRN and LIB+NAT are starting from a similar base. This means, that there are probably the same number of BTL's sitting within each party's current estimate.

1:20am
55% counted
LIB.. .. .. .. 33.98%
ALP.. .. .. .. 21.27%
GRN.. .. .. .. 15.93%
PUP.. .. .. .. 12.22%
NAT.. .. .. .. 3.20%
LDP.. .. .. .. 1.72%
GOD.. .. .. .. 1.53%
HEMP.. .. .. .. 1.10%
SFP.. .. .. .. 1.05%
SEX.. .. .. .. 0.97%

Results are:
Party     Elected
--------- ---------
LIB    :     2.524
NAT    :     0.000
ALP    :     1.434
GRN    :     1.000
PUP    :     0.997
HEMP    :     0.034
WIKI    :     0.007
VEP    :     0.004

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.521
20-2-110    :     0.431
20-1-111    :     0.045
30-2-100    :     0.003

1:00am
48%
LIB.. .. .. .. 34.22%
ALP.. .. .. .. 21.10%
GRN.. .. .. .. 15.91%
PUP.. .. .. .. 12.18%
NAT.. .. .. .. 3.25%
LDP.. .. .. .. 1.71%
GOD.. .. .. .. 1.50%
HEMP.. .. .. .. 1.11%
SFP.. .. .. .. 1.06%
SEX.. .. .. .. 0.96%

Party     Elected
--------- ---------
LIB    :     2.570
NAT    :     0.000
ALP    :     1.385
GRN    :     1.000
PUP    :     0.995
HEMP    :     0.041
VEP    :     0.005
WIKI    :     0.004

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.565
20-2-110    :     0.380
20-1-111    :     0.050
30-2-100    :     0.005

12:51am
"Median scenario" TruthSeeker numbers, with zero variation, shows 2-2-G-P outcome. But variation either way makes it 3-1-G-P

12:50am
So, ALP now up to 39% of getting 6th seat. VEP still refusing to die. HEMP getting a bit further away.

12:45am
40% counted
LIB.. .. .. .. 34.40%
ALP.. .. .. .. 20.95%
GRN.. .. .. .. 15.99%
PUP.. .. .. .. 12.09%
NAT.. .. .. .. 3.34%
LDP.. .. .. .. 1.68%
GOD.. .. .. .. 1.49%
HEMP.. .. .. .. 1.12%
SFP.. .. .. .. 1.06%
SEX.. .. .. .. 0.94%

Party     Elected
--------- ---------
LIB    :     2.564
NAT    :     0.000
ALP    :     1.385
GRN    :     1.000
PUP    :     0.989
HEMP    :     0.056
WIKI    :     0.005
VEP    :     0.001

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.553
20-2-110    :     0.380
20-1-111    :     0.056
30-1-101    :     0.006
30-2-100    :     0.005


12:35am
And yet again, VEP get a bit of an airing. But this is incredibly unlikely.


12:30am
35% counted
LIB.. .. .. .. 34.60%
ALP.. .. .. .. 20.75%
GRN.. .. .. .. 15.84%
PUP.. .. .. .. 12.14%
NAT.. .. .. .. 3.41%
LDP.. .. .. .. 1.72%
GOD.. .. .. .. 1.48%
HEMP.. .. .. .. 1.10%
SFP.. .. .. .. 1.07%
SEX.. .. .. .. 0.95%

Results are:
Party     Elected
--------- ---------
LIB    :     2.661
NAT    :     0.000
ALP    :     1.258
GRN    :     1.000
PUP    :     0.990
HEMP    :     0.084
WIKI    :     0.006
VEP    :     0.001

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.651
20-2-110    :     0.256
20-1-111    :     0.083
30-1-101    :     0.008
30-2-100    :     0.002


12:20am
29% counted
LIB.. .. .. .. 36.45%
ALP.. .. .. .. 21.42%
GRN.. .. .. .. 15.94%
PUP.. .. .. .. 12.32%
NAT.. .. .. .. 3.51%
LDP.. .. .. .. 1.86%
GOD.. .. .. .. 1.51%
HEMP.. .. .. .. 1.11%
SFP.. .. .. .. 1.09%
SEX.. .. .. .. 0.98%

Results are:
Party     Elected
--------- ---------
LIB    :     2.766
NAT    :     0.000
ALP    :     1.179
GRN    :     1.000
PUP    :     0.955
HEMP    :     0.087
WIKI    :     0.011
VEP    :     0.002


LN-A-GPO     Prob
---------- --------
30-1-110    :     0.721
20-2-110    :     0.163
20-1-111    :     0.071
30-1-101    :     0.029
30-2-100    :     0.016


12:05am
23.2% counted
LIB.. .. .. .. 36.21%
ALP.. .. .. .. 21.32%
GRN.. .. .. .. 16.07%
PUP.. .. .. .. 12.40%
NAT.. .. .. .. 3.49%
LDP.. .. .. .. 1.84%
GOD.. .. .. .. 1.58%
HEMP.. .. .. .. 1.17%
SFP.. .. .. .. 1.08%
SEX.. .. .. .. 0.99%

Party     Elected
--------- ---------
LIB    :     2.753
NAT    :     0.000
ALP    :     1.172
GRN    :     1.000
PUP    :     0.974
HEMP    :     0.095
WIKI    :     0.006

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.727
20-2-110    :     0.163
20-1-111    :     0.084
30-1-101    :     0.017
30-2-100    :     0.009



11:55pm
LIB.. .. .. .. 36.00%
ALP.. .. .. .. 21.50%
GRN.. .. .. .. 15.87%
PUP.. .. .. .. 12.52%
NAT.. .. .. .. 3.58%
LDP.. .. .. .. 1.85%
GOD.. .. .. .. 1.56%
HEMP.. .. .. .. 1.16%
SFP.. .. .. .. 1.09%
SEX.. .. .. .. 1.00%

Outcomes again:
Party     Elected
--------- ---------
LIB    :     2.760
NAT    :     0.000
ALP    :     1.155
GRN    :     1.000
PUP    :     0.983
HEMP    :     0.092
WIKI    :     0.007
VEP    :     0.003

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.743
20-2-110    :     0.149
20-1-111    :     0.091
30-1-101    :     0.011
30-2-100    :     0.006


11:45pm
LIB.. .. .. .. 36.44%
ALP.. .. .. .. 21.23%
GRN.. .. .. .. 15.76%
PUP.. .. .. .. 12.44%
NAT.. .. .. .. 3.68%
LDP.. .. .. .. 1.85%
GOD.. .. .. .. 1.57%
HEMP.. .. .. .. 1.17%
SFP.. .. .. .. 1.03%
SEX.. .. .. .. 1.03%

And the outcomes are:

Party     Elected
--------- ---------
LIB    :     2.802
NAT    :     0.000
ALP    :     1.120
GRN    :     1.000
PUP    :     0.956
HEMP    :     0.117
WIKI    :     0.003
VEP    :     0.002

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.758
20-2-110    :     0.106
20-1-111    :     0.092
30-1-101    :     0.030
30-2-100    :     0.014


11:33pm to confirm, this assumes that all "group" votes are indeed ticket votes. This is valid until we actually start knowing the breakdown of ATL/BTL votes.

11:30pm the previous update has HEMP being elected at the midpoint scenario, but with LIB getting elected at most of the outliers.

11:23pm Update at 4.2%
LIB.. .. .. .. 35.17%
ALP.. .. .. .. 20.43%
GRN.. .. .. .. 14.13%
PUP.. .. .. .. 13.70%
NAT.. .. .. .. 3.54%
LDP.. .. .. .. 1.69%
GOD.. .. .. .. 1.40%
HEMP.. .. .. .. 1.21%
SFP.. .. .. .. 1.15%
SEX.. .. .. .. 1.02%

Party     Elected
--------- ---------
LIB    :     2.711
NAT    :     0.000
ALP    :     1.012
GRN    :     0.999
PUP    :     0.998
HEMP    :     0.230
VEP    :     0.032
WIKI    :     0.018

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.708
20-1-111    :     0.277
20-2-110    :     0.012
30-1-101    :     0.002
30-1-011    :     0.001


11:13pm
That is, if HEMP elected, then it will be likely at the expense of LIB3

11:10pm
Latest top 10 parties
LIB.. .. .. .. 34.88%
ALP.. .. .. .. 20.98%
GRN.. .. .. .. 14.08%
PUP.. .. .. .. 13.35%
NAT.. .. .. .. 4.13%
LDP.. .. .. .. 1.71%
GOD.. .. .. .. 1.35%
HEMP.. .. .. .. 1.16%
SEX.. .. .. .. 1.05%
SFP.. .. .. .. 0.97%

...leads to the following results:
Party     Elected
--------- ---------
LIB    :     2.857
NAT    :     0.001
ALP    :     1.025
GRN    :     1.000
PUP    :     0.993
HEMP    :     0.105
WIKI    :     0.011

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.850
20-1-111    :     0.117
20-2-110    :     0.025
30-1-101    :     0.007
21-1-110    :     0.001


11:08pm
HEMP polling more than 1% - putting them in the frame?

11:01pm
Recent data in much better for greens as vote counted crosses 2%.
Using the following top 10 estimated primaries
LIB.. .. .. .. 34.23%
ALP.. .. .. .. 21.60%
GRN.. .. .. .. 14.66%
PUP.. .. .. .. 13.03%
NAT.. .. .. .. 3.47%
LDP.. .. .. .. 1.72%
GOD.. .. .. .. 1.60%
HEMP.. .. .. .. 1.15%
SEX.. .. .. .. 1.14%
SFP.. .. .. .. 0.92%

Outcomes with 10% variation are:
Party     Elected
--------- ---------
LIB    :     2.717
NAT    :     0.000
ALP    :     1.222
GRN    :     1.000
PUP    :     1.000
HEMP    :     0.052
WIKI    :     0.007

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.717
20-2-110    :     0.222
20-1-111    :     0.061

10:51pm: First M-C outcomes - +/-10% variation for everyone, so pretty small:
3 LIB, 1 ALP, 1 PUP looking likely.
Party     Elected
--------- ---------
LIB    :     2.997
NAT    :     0.002
ALP    :     1.333
GRN    :     0.608
PUP    :     1.000
HEMP    :     0.001

LN-A-GPO     Prob
---------- --------
30-1-110    :     0.606
30-2-010    :     0.332
30-1-011    :     0.059
21-2-010    :     0.001
21-1-110    :     0.001
20-1-111    :     0.001



10:47pm
Estimated primaries for top 10 parties
LIB.. .. .. .. 35.70%
ALP.. .. .. .. 21.66%
PUP.. .. .. .. 13.54%
GRN.. .. .. .. 12.16%
NAT.. .. .. .. 4.39%
GOD.. .. .. .. 1.64%
LDP.. .. .. .. 1.51%
HEMP.. .. .. .. 1.20%
SEX.. .. .. .. 1.12%
SFP.. .. .. .. 0.94%


10:30pm First simulation coming up very soon....

10:30pm Disagree with ABC swings at this stage. I'm getting:
WIKI -0.1%
NAT -0.8%
IND 0.6%
DEM 0.0%
ALP -5.7%
LDP -2.0%
PUP 6.6%
LIB 0.5%
SFP 0.4%
HEMP -0.1%
GOD 0.0%
GRN 0.8%
SEX -0.4%

at this point in time. PUP stronger swings in regional areas than ABC seems to be giving credit for


9:50pm Is low voter turnout going to cost Labor? TruthSeeker analysis from the Griffith byelection will result in a net swing from Labor to LIB of ~0.15%. Not massive. In fact, borderline nothing!

9:41pm: Has the AEC had a technical difficulty?

9:40pm - Hmmm - I didn't round up. Quota is 2, and only 4 have quota. So result would be 2 LIB, 2 NAT, 1 PUP (and drawing of lots for the 6th electee, from anyone!)

9:30pm - 8 votes in. Just that. If this was replicated state wide, we'd see 3 LIB and 3 NAT. 4 LIB 2 NAT. Ha!


9:20pm Welcome folks - here we go again. Stay tuned for updates as the night progresses. This page will be updated as time rolls on.

27 comments:

  1. re 9:30pm post: You mean 3 LIB, 2 NAT and 1 PUP? There are only 2 National candidates.

    ReplyDelete
  2. Gosh. 2 NAT. But their prefs go LIB so it would be a 4-2 split.

    ReplyDelete
  3. I'm calling 3 Lib 1 ALP 1 GRN 1 PUP.

    ReplyDelete
  4. 2 Lib, 1 ALP, 1 GRN, 1 PUP look locked in, the question is who will get the last seat?

    ReplyDelete
  5. Agreed - it's between ALP and LIB at the moment with the ALP ahead.

    ReplyDelete
  6. The ABC is now predicting the Green vote at 20.5%. I just can't see that being possible?!

    ReplyDelete
  7. I think the HEMP result may fade if the Labor outcome is a little higher? I'm running sims with Labor at a base of 23% and get HEMP or VEP often getting to the last three but then getting tipped out by Labor who then lose to whichever of LIB3 and PUP was not already elected.

    ReplyDelete
    Replies
    1. I'm getting the HEMP-ALP gap as ~0.1 quota or ~1.4%, so it's quite large

      Delete
    2. I actually have Labor beating them even with a lower vote. I can get HEMP to around 0.75 of a quota but after getting preferences from SEX and a GRN surplus (Pratt being selectively preferenced so she is ahead of HMP) I always have the ALP ahead of them.

      Delete
    3. Correction to the above - I get HEMP to a little over 0.5 of a quota, not 0.75. But the rest of it holds - I have them behind Labor at that point.

      Delete
  8. BTW I'm claiming dibs for picking the result at 10:30PM :-)

    ReplyDelete
    Replies
    1. Yep! Great pickup. But stlil not 100% clear yet.

      Delete
    2. ABC is showing 2-2-1-1 now, ALP by 800 votes.

      Delete
  9. What would be the effect of BTL preferencing Pratt over Bullock? I know some tickets (most importantly the Greens) have already done that, is that a good enough estimate do you think?

    ReplyDelete
    Replies
    1. BTL prefing Pratt > Bullock would make no difference. Bullock elected immediately on primaries, so is thence elected and excluded.

      Delete
  10. Well, looks like the Lib/PUP arb paid off. I put the risk-free profit on GRN, and am now up 7.5% - not too shabby. Thanks Centrebet!

    ReplyDelete
  11. One consideration is that pre-poll and postal voters would have missed some of the PUP advertising blitz and the ALP Bullock controversy. We might see some decrease for PUP and increase for ALP as those votes are included, increasing the odds of 20-2-110 relative to 30-1-110.

    How do the outcomes change if you reduce the PUP vote by say 0.5% or 1.0%, while increasing the ALP vote by the same amount?

    ReplyDelete
  12. "2. Calculate benchmark 2013 "base" expected performance for each electorate that depends on %counted. While we can't do a matching booth comparison, we did instead do a matching by vote counted, where the order of counting each booth within an electorate is assumed to be directly propotional to the number of votes (bigger booths take longer)"

    This worked much better than the ABC's model which at times projected the Greens to close to 20 (and in very early counting above it) and quite a long way into the count projected Labor to below 20. I didn't agree with their projection for Labor at the time and said so and why.

    Nice work on that.

    ReplyDelete
  13. http://kevinbonham.blogspot.com.au/2014/04/wa-senate-take-two-preview-live.html

    Link to my comments thread. There appears to be a tipping point determined by the question of whether PUP cross on the HEMP preferences, as currently looks likely, or on the earlier LibDem preferences instead. Trivial change in the PUP vote alters final outcome by thousands. Worth keeping an eye on but my tip is vote distribution to move away from this tipping point not towards it.

    ReplyDelete
  14. And further on the tipping point - on current primaries BTLs could put PUP over it, meaning that Reynolds probably isn't actually winning yet (she also has the issue of leakage from the Nats BTLs before she gets on level terms with Pratt). Still think there'll more likely than not be enough drift to the Libs in post-count for the tipping point and BTL scenarios to disappear but this is less straightforward than it at first appeared.

    ReplyDelete
  15. Rather large primary vote changes tonight, possibly on formality checking. On current votes PUP would cross before the tipping point but Labor would still lose by a thousand or so. It probably gets worse from here.

    ReplyDelete
  16. Some of the damage tonight was just data re-entry but Labor are still down several hundred for the day.

    ReplyDelete
  17. Libs gaining heaps on early postals. Looking over on this evidence.

    ReplyDelete