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

Twitter: @AU_Truth_Seeker


Friday 4 October 2013

WA Polling booths discrepancies

Here at Truth Seeker, we seek the truth. Simple.

Click to see multiple examples of vote discrepancy...

(I may update this post from time to time. I'll go top to bottom, so any updates will be subsequently time-stamped down the bottom of this post).

We firmly believe that representative democracy is best served by having a robust system of counting votes and electing the candidates most preferred by the voters.

Truth Seeker is NOT a member of any political party and has no close relative who is. Rather, it is vital that the truth gets out and we get the representatives we voted for.

The AEC has been set to announce ALP's Louise Pratt and PUP's Dio Wang as being elected to the final two spots in the 2013 WA Senate election. Critically, the difference comes down to two micro-parties who between them shared just 2.7% of the vote - the Shooters and Fishers and the Christians. At an exclusion point, the Shooters led the Christians by just 14 votes. If this exclusion was flipped, the result would change such that the Sports Party's Wayne Dropulich would be elected, and Green's Scott Ludlam would probably be elected.

This post is not saying that Pratt & Wang should not have been elected and Ludlam and Dropulich should be. Honestly, I'm not sure how legislative outcomes would be better/worse served under each outcome. But we need to be sending people to the Red Benches who we actually voted for in accordance with the system that existed on polling day. My blog does not barrack or provide support to any party.

The most obvious way to check for vote discrepancies is to compare the number of ballots cast for the House and Senate. In theory, these should be equal. In practice, the numbers should be incredibly close and only vary by the odd vote or two where people may have stolen/shredded/burnt/eaten either (but not both) ballot papers. Also, there may be the very odd difference (usually pre-poll) where all senate ballot papers go in one box and it cannot from that point be distinguished which Senate ballot was cast in which Division (the same problem does not exist in the House, where it can be easily distinguished which ballot paper is for which division).

So, here is a list of polling places in WA where a discrepancy of 10+ votes exists:

Division Polling Place Vote
discrepancy
Brand Port Kennedy -14
Canning Oakford +14
Canning Strawberry Fields -11
Durack Derby -50
Durack Roebourne +10
Moore Connolly +10
Moore Heathridge +13
O'Connor Albany Central +18
O'Connor Esperance +18
O'Connor Harrismith +15
O'Connor Kambalda West -50
O'Connor Narrogin East +10
Pearce Yanchep -12
Perth Inglewood North +12
Stirling Balcatta North -10
Stirling Osborne Park +10
Swan Belmont +11
Swan Langford +14
Tangney Bull Creek East +20
Tangney Canning Vale Central +14
Tangney Canning Vale East +15

Please check this for yourself:
1. Go to the vtr.aec.gov.au
2. Click on "Western Australia" in the left hand sidebar under division results
3. Click on the electorate of your choice (eg, Tangney)
4. Click on "Polling Places" towards the top of the page
5. Click on the polling place of your choice (eg, Canning Vale East)
6. Note the total (informal and formal) number of House votes for the polling place (eg, 3,335 votes)
7. Now, to check the Senate results, click on Election Results "Senate"
8. Click on Division & Polling Place Results "Western Australia"
9. Click on the same electorate as above (eg, Tangney)
10. Click on "Polling Places" towards the top of the page
11. Click on the same polling place as above (eg, Canning Vale East)
12. Scroll down and to all the way to the bottom, noting the total number of Senate votes (eg, 3,350 votes)

There are additional discrepancies that exist in some pre-poll voting centres, but I have not included them in the above list as some of these relate to a booth which is common to two or more electorates. For example, Greenfields Brand PPVC had 592 more Senate votes than House votes, but Greenfields Canning PPVC had 590 less Senate votes than house votes.

Also, some of the above relate to Divisions that have not been declared - Durack and Brand, so arguably these house figures are subject to change. The fact that an uncanny number are multiples of 10 implies there is at least one "bundle error" that has not been resolved.

Overall, 600 out of 906 polling booths have recorded different vote numbers for the House and Senate, and 118 polling booths display a discrepancy of 5 or more votes.

And this is not even getting into the discrepancies that exist in postals, pre-polls, provisionals or absentee ballots.

What is wrong with a recount? Elections cost the government tens of millions of dollars. A recount would cost a drop in a proverbial ocean. Further, as long as we have an answer by 30 June 2014 we'll have the right senators in place by 1 July 2014. This Truth Seeker cannot see how the courts would blindly accept any result so close where the AEC had not gone to every possible length to ensure that the correct outcome was officially declared.

Updated - Geraldton Waggrakine
When analysing the rate of BTL voting, one polling booth stands out - Durack's Geraldton Waggrakine. To be honest, I know nothing about this area apart from what Wikipedia tells me.

But given a previous comment to my blog that a polling place manager made an expedient decision to quickly count the ballots on the night, pending a recount, I wonder if Waggrakine has slipped through the cracks?


I'm not saying anything is wrong about the way this booth was counted - it may be right. It just looks highly suspicious. The record indicates 1929 votes were cast, with just one ballot recorded as Below The Line. No similarly sized booth had anywhere near this low rate of BTL voting. Also, this booth had a 5.7% informal rate, compared to 2.9% for Durack generally. Interestingly, its House Informal rate was very similar to the Durack average.

Should half of these informal ballots actually be Below The Line votes? Would this add an additional 60 ballots to the count? Will this change the result? Don't know...

Would conducting a full recount give us more confidence in our democratic processes? Yes.

29 comments:

  1. I have looked at this issue before across the country and noted that there is always a discrepancy between total house and senate votes...its just the way it is.....the only thing I can put it down to is that either one of the other of the ballot papers is messed up and/or the voter does not want to complete at least one of the ballot papers. They either throw it in the bin or take it with them...either way there is a discrepancy........the reason could be many and varied......but this issue does exist across the country. Go and look at the total house versus senate votes in WA for the 2010 election. There is a difference of about 10k votes!!.

    Having said all of that I suspect there will end up being a recount purely on the basis of the closeness of the vote.....having said that it is quite common to get a close vote early on in a senate exclusion contest. are we going to recount all of those when they happen or just the ones which we deem to have an impact?? perhaps that is why the AEC refused the initial request.

    as for the recount I cant see how anything other than a full recount would work...it leaves open too many questions if you just do a partial recount.....I still think this one ends in court with arguments about informal votes...as we saw in Bush V Gore, when it really matters to the balance of power people can argue any outcome with a pretty reasonable chance of success...I cant see Clive backing down if a recount reverses the outcome.

    ReplyDelete
    Replies
    1. The 11,049 difference between total Senate votes and total Reps votes is not a number to focus on because most of the difference (all but 687 votes) is in declaration votes (absent, provisional, pre-poll, postal) and likely legitimate.

      Basically, if the DRO is satisfied on the basis of the declaration information on the envelope that the person was entitled to vote in WA, but not satisfied that they were entitled to vote in the division in which they have tried to vote, their Senate vote will be counted but their Reps vote will not. It happens surprisingly often, especially with provisional votes and to a slightly lesser extent with absent votes.

      The difference of 687 in ordinary votes is the net of the numbers Truth Seeker is working with. It's still a significant number and some of the individual booth discrepancies are alarming - especially the ones that were out by an even 50.

      Delete
    2. thanks for that. So are you saying that the DRO just throws away the HOR ticket? Doesn't even count it in informal? So it is not added to the total votes at all?

      Delete
    3. They are not quite thrown away, the papers are bundled up and the package signed by the DRO, in the same way as unopened envelopes that do not meet the declaration standards at all (i.e. even any contained Senate vote can not be counted).

      The papers are not counted as informal votes - and neither should they be. Informal votes are votes from valid voters. These are votes from people that it could not be established were entitled to vote in the division at all.

      The procedures are in Schedule 3 of the Commonwealth Electoral Act with sections 10(b), 19 and 20 being the relevant provisions.

      http://www.austlii.edu.au/au/legis/cth/consol_act/cea1918233/sch3.html

      Delete
  2. I'll be so bold as to say without a shadow of a doubt the Waggrakine booth was counted / recorded wrong. The only explanations in my mind do not apply - it seems to be a typical outer suburban booth. If it had been a huge nursing home or a remote indigenous community with overwhelming illiteracy a very weak argument may have been sustained. Please attempt to have the AEC provide a credible explanation. I really don't think this sort of thing should remain unchallenged. It's a little depressing really that it requires people like you who are not being paid to provide oversight and quality control to discover these anomalies. But it's refreshing that at least someone is. Btw, your blog is always fascinating and educational. Much of it I'm perceiving as black humour, possibly as a subconscious defence mechanism to stop myself bursting into tears!

    ReplyDelete
    Replies
    1. Thanks - black humour is 100% intentional :-)

      I'll keep blogging weekly or so even after the senate is finally decided by our courts.

      Delete
    2. I've just tweeted AEC the Waggrakine question... will post response, if any

      Delete
  3. That's the kind of "dog's balls" data point I've been looking for... Thanks!

    ReplyDelete
  4. From my experience counting votes the most likely reason for a discrepancy in house/senate votes is that it is *really* hard to count 50 pieces of paper. Votes are bundled together, then checked. Most bundles are out by 1-2 votes. The check will sometimes fix, and sometimes break the count.

    If you calculate it statistically: 2000 bundles of 50 with a std.dev of 0.5 equates to a sampling mean standard error of 0.012 or a standard deviation of discrepancies of (2000*0.012) = 22 votes. Judging by the numbers above the std.dev is probably more like 0.1, but it can add up.

    TS, if you have a list of every discrepancy and use that to calculate a standard error of "bundling error" per 50 votes, you should be able to calculate the probability that 14 votes is significant, amongst the ~1000 bundles of SFP/Christian votes. My estimate from the above numbers is that it won't be in 99% of cases.

    ReplyDelete
    Replies
    1. And Senate counts are much harder to get exact that HoR.

      Mind you, all of those packets should have been through fresh scrutiny since, so there really shouldn't be any excuse for complete screwups. On the night, 1 in a 100 is reasonable to expect; after fresh scrutiny you would hope for maybe 1 in 500.

      Of course, if your ballot guards aren't watching properly, you can lose ballots out the door or in the bin. Still, there are some pretty substantial errors there - I expect that the 50 noted will be just a missing bundle.

      Also, all of the votes should still be in their booth bundles, except for maybe BTL votes - I'm not sure on the process for those once they go to the state counting centre.

      Delete
  5. All of the senate votes at Waggrakine would have been counted atleast twice, once at the polling booth on the night and once again at the electorates count centre sometime in the following two weeks.

    If there was a RO that hurried the count and didn't count BTL's properly, that would have been picked up by the recount which was done at a separate location with new counters and a different RO.

    ReplyDelete
  6. What confidence is there that a second count would be better than the first?

    ReplyDelete
    Replies
    1. Huge difference.

      Fresh scrutiny is done in good working conditions with no time pressure and an emphasis on getting it right. It is also done with teams dominated by experienced hands, who discuss any contentious outcomes amongst themselves.

      It's a pretty big fillip to have your booth's count go through fresh scrutiny without changes.

      Delete
    2. More rigorous & closer checking by staff & scrutineers, knowing result is close, and hopefully (depending on how ballots from the first count are packaged and labelled) the incremental results of the first count to check the recount against. It would require errors in both counts of a given batch of votes for them to go undetected.

      truth seeker, the bottom axis does not look like a log scale as labelled to me.

      Delete
  7. WA Senate Declaration postponed till further notice (was to be at 11.30am WST) pending review of further appeal by Greens candidate

    ReplyDelete
  8. Truthseeker. I notice there are a number of booths with no BTL count not just the one you highlight. Are they randomly distributed? Is it possible that BTL votes are not always entered at the polling booth but at some central location and then allocated to another booth?

    ReplyDelete
    Replies
    1. Some very small booths will sometimes have no BTL votes. I once did a booth in a nursing home, and the results showed a zero BTL count. No doubt, the very elderly couldn't/didn't want to, number that many boxes. The same could be true in places with very low literacy and numeracy levels.

      Delete
  9. Geraldton-Waggrakine recorded 50 BTLs in 2010.

    ReplyDelete
  10. important to note that Waggrakine is in Durack which is probably the strongest area for the shooters party.........if there are votes missing I suspect the Greens wouldn't want them found!!

    ReplyDelete
    Replies
    1. No, the Greens always want the right result, nt the one that suits them.

      Pete

      Delete
  11. did anybody else notice that the Katter party only beat Rise-UP by 25 votes at the elimination point.... probably some of you but did it matter? probably not........I think it just goes to show how often the votes can be very close at the crucial earlier elimination points in the senate counts and probably explains why there is no automatic vote recount mechanism as it could create some major problems.........

    ReplyDelete
  12. Votes go missing into the bins. The punters either accidentally or otherwise put their ballots into the large AEC-provided bins with the used HTV cards.

    Some time ago (maybe 10 years or so) many polling booths would have the bins just outside the exit door and party campaign workers could sift through for undamaged HTVs to keep as a reserve stock should they run out. Or just to relieve boredom at small booths.

    Every now and then you'd find a blank ballot paper in with the discarded HTVs - and then tell the officer in charge who could work out what to do with it, I'm guessing they'd be smart to call it a spoil.

    The AEC /SEC no longer allows such HTV scavenging, probably as much for public liability risk in having people rummage through a bin that might contain sharps, as a means to ensure stray ballots dont make it into the wrong hands.

    Regards
    David B

    ReplyDelete
    Replies
    1. If a voter fails to place a ballot paper in the ballot box and that paper is found somewhere in the booth by a staff member, it is placed in a sealed envelope and marked as 'discarded'. As such, it is not included in the vote as an informal vote. It is however noted at the end of the day when the OIC is balancing the number of ballot papers given with those accounted for at the end of the day

      Delete
    2. I've been trying to find some example figures for missing and discarded ballot numbers from past statistics but so far without success. Assuming they are indeed sufficient to regularly produce small discrepancies, another test that suggests itself is to examine discrepancy size by votes cast at a booth and check for major outliers.

      Delete
    3. On the multiples of 10 thing - the two 50s have my attention as probable bundle errors but apart from that I'm doubtful it is meaningful. Given that the cutoff for the list is 10 I would expect 10 to be the most common discrepancy in the list, and it is.

      Derby, Durack - For the House the informal vote is 4.32 points higher than in 2010. The district as a whole was 1.8 points higher than in 2010. For the Senate the informal vote is 1.3 points lower than in 2010. The district as a whole was 0.33 points lower than in 2010. I wouldn't be surprised if most or all of the error was in the House informal pile, but of course it should be checked.

      Had a good look at Kambalda West, but nothing jumping out at me on that one so far. Which is a bit of a worry. Especially as the Senate vote in this booth is down on 2010.

      Delete
    4. I looked at these numbers as well and you are right. Kambalda West does look odd. I suspect if you look at anything long enough you can always come up with a different answer and vote counting would be no different. You would expect to find some discrepancies. Whether that warrants a full recount of 1.3m votes is the question. I suspect each time there was a recount you would come up with a different number, especially when it gets into the subjective area of informal votes.

      In terms of the big picture, if the biggest discrepancies we can come up with is 2 lots of 50 votes in heavily Shooters party territory that may be missing then this is not going to change the result. Indeed it may actually improve the result the other way to what the Greens are looking for......

      Delete
  13. Can someone experiened tell me, do all ordinary votes get counted and entered in the computer from the polling booth at which they were made or do some get sent somewhere else and entered? does the same apply to provisional votes? how to absents/postals and prepolls get allocated against a particular booth?

    Is it possible that any larger than normal discrepancies get explained by these procedures?

    ReplyDelete
    Replies
    1. Both. They are all entered at the polling booth on the night and submitted, then sent elsewhere and recounted and reentered at a central count center.

      All BTL votes are sent to a central BTL count center and entered (twice) somewhere else again.

      Postals/Absents/Prepolls are collated centrally then sent out to the count center for each electorate to be counted and entered.

      Delete
  14. WA BTL Votes worthy of Interest.

    http://democracy-at-work.blogspot.com.au/2013/10/wa-senate-recount-ballot-papers-worthy.html

    Unfortunately I do not have a source location list for each batch allocation as this would assist in narrowing down the list even more

    The odds are that the gap between them will widen not swing.

    It is estimated that a recount of the WA Senate will costs taxpayers around $100,000

    Had the AEC provided Scrutineers access to the preference data-file then these votes would have been subjected to more detailed scrutiny during the initial count and possibly avoid the need for a recount

    ReplyDelete