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

Twitter: @AU_Truth_Seeker

Sunday, 3 November 2013

Final WA Senate results... until overturned by the courts

Well, what a busy few weeks its been! Apologies for not prew in the PUP/ALP vs GRN/SPORT tag team match we've just witnessed). It is also pleasing that this means that informal BTLs were reconsidered, as per my request to AEC.

The revised data dump available online shows the Shooters vs Christian margin which was 14 votes in favour of the Shooters (and hence PUP/ALP) is now actually 12 votes in favour of the Christians (and hence GRN/SPORT).

What has happened is that the number of Christian votes is 25 higher than previously, while the number of Shooters votes is 1 lower. I could make a reference to this being a Christian miracle, but I will let the reader decide if He would intervene to elect an additional Greens MPs!

Notable changes in votes as a result of the recount:
All parties gained votes BTL, half gained and half lost votes ATL
Christians: +26 (incl +5 ATL, +21 BTL)
Shooters: -6 (-23, +17)
Aus Ind, AFLP ATL: -7 (these flowed through to the Shooters via Group Voting Ticket)
No Carbon Tax: -15 (This flowed through to the Christians via Group Voting Ticket)
This means the Christians fared 14 better on other BTLs, and the Shooters fared 12 better on other BTLs. This is slightly high, but not crazily so, given the new number of BTL votes recorded is 701 higher.

(post modified): It is now widely known that there are 1375 missing votes. I think it almost inevitable that a new election will be called. If so, it becomes a gamble to everyone, as we move into unprecedented territory. While a more pleasing option would be to include the lost votes in the manner they were at the counted in the original count, I doubt the courts will do this. I advise readers to view the comments thread from my previous post.

As a slight aside, I also talked about the Kambalda West vote (O'Connor), and stated that the Senate vote for Labor looked 50 votes too low. I note the recount has "found" these votes, with an additional 50 coming Labor's way and an additional 2 (probably BTL) as well. If I was more enthusiastic, I'd rerun my whole discrepancies spreadsheet that I created... (unlikely :-)

I have put some thought into possible Senate voting reform options, but I could not write anything better than the excellent summary that Kevin Bonham has come up with. The only addition to Kevin's commentary I would like to add is that the randomness of voting columns has distorted outcomes. While we acknowledge randomness is fair in theory, I think that listing parties by their performance in the last Senate election would be fairer. Sure, the Liberal Party would be Group A, Labor Group B, Greens in Group C in just about all states. But when 85%+ of people want to vote for these parties, what is wrong with this?

For the future... I doubt I will have the time to post weekly as I previously stated. I think it is highly likely I will resume blogging for the new WA Senate election, the Vic upper house election, and possibly some number crunching for the SA upper house election.

If I don't post again soon, I owe a big thanks to the tens of thousands of you who have visited my blog over the last two months. Also, thanks to Tim Colebatch, Kevin Bonham, Andrew Crook and the prolific Poll Bludger Will Bowe for trusting my work and reporting my conclusions. And thanks to those who have provided me informal information of what's happening behind the party lines, and the mysterious Maxine for sharpening my modelling. My initial aims were to draw attention to a range of unrepresentative election scenarios, and to test whether financial analysis techniques such as Monte Carlo Analysis could be applied to elections. I'm satisfied that I more or less achieved my aims.


  1. The missing 375 votes have not been counted at all. They are not included in this count which is to be declared on Monday.

    That is why they are significant.

    1. I have to agree - the lost votes have been excluded from the count and there is sufficient evidence to show that according to the booth totals from the on the night count 13-15 more shooters votes got excluded then Christian votes, thus completely altering the outcome.

  2. And their disappearance has determined the result.

    It is known which booths they are from and even who they were counted for the first time around. See the comments thread to your previous post for the running commentary from your readers... short version is that the missing votes include 18 that would have been with the Shooters and 5 that would have been with the Christians. That 13-vote swing would make it a 1 vote margin for the Shooters (and therefore Wang and Pratt) if re-included. (Note that Ben Raue at The Tally Room puts the impact at 15 votes - he is including two changes that I believe are genuine recount changes rather than lost ballots).

    The AEC has declared the result on the votes they now have. I don't think they have any choice in this. The Court of Disputed Returns WILL have a choice and will almost certainly try to re-include the missing votes based on the original tally. However if that results in a 1-vote margin they may be almost forced to declare a new election given that there is a not insignificant chance of there being a relevant error in the missing 1375 votes.

  3. "But, the most curious aspect, albeit not result changing, of the recount was the loss of 1428 Liberal ATL votes. Where did these go??'

    My guess - and that's all it is - is that part of the answer involves the infamous missing votes being specifically Liberal bundles. If you look at the polling booth figures on the AEC site, you'll see that at two of the four booths from which votes went missing - Wundowie and Mount Helena - the number of Liberal ticket votes recorded is a grand total of zero. In another of the booths, Henley Brook, the Liberal group total is 30.79%, which isn't plausible given their House of Reps vote in that booth was 58.65%. It's also clear that there are far too few informal votes recorded for the Bunbury East (four out of 1529) and Wundowie (zero out of 232) booths. If it is actually the case that these missing votes are Liberal and informal bundles, they won't have had any influence on the result.

    1. 886 Liberal votes lost in the 4 booths as originally counted.
      For the record, the others are: 164 Labor, 112 Greens, 21 Animal Justice, 5 Democrats, 3 Stop the Greens, 2 Climate Sceptics, 11 Family 1st, 3 Katter, 11 Wikileaks, 4 Aus Independents, 1 Secular, 14 Shooters, 7 HEMP, 3 Christians and 8 Smokers. Total 1255 plus 120 informal

    2. Which booth were the Democrats votes missing from?

      My tally of the missing is short of the public 1255 by 5 votes and this would solve it, but I'm not seeing any movement in the Democrats figures at any of the booths.

      I'm also short by either 6 or 10 on Informals depending on how you treat the 4 they still have from Bunbury East. I suspect they lost all of the original 81 but found 4 in other stacks, in which case I'm at 110 (81 plus 29 from Wundowie - I'm also assuming the movements at the other booths were just normal re-count shifts).

      Did they maybe find some Democrat votes and Informals after releasing publicly the number that they were missing?

    3. 5 Democrats votes missing from Bunbury East.
      Informals: Bunbury East (80), Wundowie (29), Pearce Provisional 1 (11). I believe you are correct that the 4 informal from Bunbury east were new from the recount.

      I don't know the answer to your last question but I would have thought the commission would have had the final number of 1375 missing for a few days

    4. Thanks for the clarification on the Informals. I'd trawled through the data looking for a "fifth booth" but never found a convincing candidate - but the data set was only ordinary votes. That said, Corrigin (O'Connor) has a strong odour of fish about it - Informals down 16, from 19 to 3, leaving it with an unlikely 0.54% informal rate.

      I'm still struggling to see the movement in the Democrats votes at Bunbury East - I have 5 in both counts and the current data on the AEC site still says 5. I think I'm going to assume they found them.

    5. Anonymous 1:56 am, November 04, 2013 indicates that Informals: Bunbury East (80), Wundowie (29), Pearce Provisional 1 (11) Can you please explain what Pearce Provisional 1 is, is this a fifth booth?

  4. From my perspective, its not who has won, its the failure of the process that bother's me.

    Looking at some of the comments on social media, I have an inkling Scott Ludlam might not put up a big fight at the CDR. If he did, he would look rather silly given his comments over the last few days.

    Be interested to see how much the votes might change if there was a new election

  5. My comment of 1:27am was made before I read the most informative comment on the previous thread concerning the anomalies in the vote totals from the four affected booths. So please disregard!

  6. Thanks for all your work truthseeker. Please come back for the State Elections!

    (Except NT/QLD)

  7. Even if the courts could determine that the margin is some small number of votes I think we will get the usual arguments over double voting, over particular voters being wrongly allowed or disallowed to vote and so on. I just don't see a margin of one or three or whatever surviving.

    Note that the CDR is entitled to amend the apparent margin based on its rulings on the votes challenged by scrutineers, though there may not be much difference from those. (There are nearly 1000 such votes, very few of them being for authenticity.)

  8. Thanks to Anoymous who has the data to calculate the impact of the "missing booths". As I have pointed out most of the swing in the vote has come from the electorate of Pearce...not just in the missing votes from the 3 booths but also the fact that 10 extra ticket votes were found for the Christians in the recount. This also had an impact. I wonder if it is possible to work out from which booth those extra votes were found? I don't have the data to do it. I thought it might be interesting to know............

    1. My polling booth data only covers ordinary votes and so doesn't square to final divisional movements. Also there seems to have been some tiny changes after I downloaded. However with those disclaimers out of the way...

      The big gain for the Christians in Pearce appears to be Yanchep, where they gained 11 votes. It appears this may have been a case of votes being place in the wrong pile because the Australian Democrats lost 11 on the re-count. However the issue may have been more complex than that as the Nationals also gained 19 at that booth.

      The Christians also had some big movements in other booths in other divisions, but largely netting out. The biggest ones were +20 at Padbury (Moore) where it appear the votes had been erroneously credited to the Liberal Democrats and -16 at Alexander Heights (Cowan) where it's possible some votes were double counted as the overall booth total is down 18 - which appears correct as Senate and House votes now align.

      The only big booth movement for the Shooters is the -14 at Mount Helena, which is one of the booths with the missing votes.

      The fact that even without the lost ballots there are double-digit changes arising from the recount (and the above are not the biggest - the ALP picked up 52 votes at Kambalda West - likely a whole 50-vote bundle not counted the first time, plus some singles misplaced in other stacks) and the fact that there were 701 valid BTL votes previously erroneously recorded as Informal suggests to me that a re-count due to a 14-vote margin at a critical elimination was more than justified and a case could be made even when the gap is significantly more than that.

      Note that this doesn't mean ANY close elimination would justify a re-count. There are plenty of eliminations early in the count that come down to a single vote or even a tie, but the order doesn't end up mattering because the other candidate is soon eliminated anyway.

      The reason the Shooters-Christians elimination created a perfect storm was because one of the candidates preferenced the other (the Christians preferenced the Shooters) but not vice-versa, meaning that if the Shooters beat the Christians they got a boost past the Sports Party (who then go out) while if the Christians win the elimination, they are next out anyway with the Sports Party getting both lots of preferences.

      I'm really hoping that ticket votes are eliminated. That way you don't have most of the preferences on an elimination all going in a single direction determined by cynical preference deals. Without that mechanism the chance that the order of elimination this early in the count then determines the entire direction of the rest of the count is much reduced.