Avid media consumers will have noticed the following two articles which were based, in part, on my analysis:
1. The Age – Double Dissolution forecasts
2. Crikey – prospects and rationale for a WA Senate recount
In response to the comments on this blog in my electronic absence:
The Waggrakine anomaly
This Truth Seeker does not usually like to take credit for things, but it is pleasing to note the potentially miscounted Waggrakine result has been picked up by a local news website and a similarly focused Facebook page.
To recap, my site pointed out late last week that the polling place of Waggrakine, in the WA electorate of Durack, recorded in the Senate an unreasonably low rate of BTL votes (just 1 out of 1929) and an unreasonably high rate of informal votes. Given a recount request is being considered, and a potential discrepancy of 50 BTL votes have been incorrectly recorded as informal, this is a significant discrepancy.
It is interesting to note that at least three people insist they voted below the line at Waggrakine, and a staff member at the polling booth also seems to imply that there were more than one ballot during the counting process. To clarify, I have no objections to the staff on election day or their role – it is not the responsibility of election day polling place staff to data-enter votes or test for formality. My understanding is that all they need to do is assign votes to tickets.
I have not raised this issue because I want the result changed, or think it should be changed, or think it may change if this booth was included. Rather, there are any myriad of issues, undetectable by standard data mining techniques, that may compromise the counting of the ballots.
It would be interesting to know if anyone from Waggrakine knew exactly who they voted for and in which order. This may be via a specialised online Below The Line vote counting site, including www.belowtheline.org.au or senate.io. If you did, I would love it if you could send me an email which is to a gmail.com account, theoriginaltruthseeker at … you get the picture (Apols, trying to avoid printing my email address to avoid spam).
For the record, the only BTL vote currently registered in Waggrakine was for the ALP’s Louise Pratt.
My assessment of likelihood of recount
I agree with the assessment of an anonymous commentator:
“For the AEC I think there is a 60-70% chance of a recount purely based on the logic of a close result AND the political/public pressure.”
I agree with the commentator who also implies that the likelihood of a full recount being required will increase if higher appeals and/or legal action is required. Kevin Bonham raises the prospect of the recount process being overrun by facetiousness (my paraphrasing) similar to the high rate of individual vote disputes seen in the seat of Fairfax. Perhaps, scrutineers could be allowed only two unsuccessful challenges per batch of votes, like the way Hawkeye challenges work in Tennis? (Please – no-one take this suggestion seriously!!)
I find it unlikely a court will refuse to accept a full recount, given the extensive time period we have between now and when the new Senators are expected to show up and be seated on the red couches at Parliament house. I think a partial recount should be ruled out, due to the closeness, that any vote that is for any other candidate may influence this outcome.
I agree with the suggestion by regular psephological commentator IntuitiveReason that a recount with closer scrutiny cannot but help improve the reliability of the count process – all minds will be 100% focussed on the outcome and allow the result to be more representative of the collective will and wisdom of our voters. Thus, a second count will necessarily be more reliable than the first count. If there’s still doubt, why not count it a third time? We’ve got nine months to get this right.
Analysis of the stated results shows the Shooters and Fishers outpolled the Christians by a 3:1 ratio – that is 54 to 18 votes. Hence, it is likely that BTLs for these parties would be slightly more favourable to the ALP and PUP – by approximately 2 or 3 votes. But the point is that the count can almost be proven to be incorrect, so let’s redo it to get the right outcome.
Upon such a recount being conducted, it is almost certain that enough votes will change one way or the other that will revert this Shooters v Fishers battle to a 50-50 prospect. So, toss a coin to predict the winners.
Miscellaneous other comments
The x-axis on my graph was not a log scale. I was experimenting with several display options, changed the axis title, then changed the axis-type, then published. Oops.
Other exclusions, apart from the ones I had previously identified over the last few weeks are irrelevant and would not have changed the outcome. This is because ~97% of votes were ATL and hence future preference flows were known.
The Kambalda West polling booth has been commented on by Kevin Bonham and others, as there were precisely 50 votes less recorded in the Senate than the House. I note that the %Labor vote in the Senate and the House is very similar across the O’Connor electorate, including in neighbouring Kambalda. However the gap between the Kambalda West Labor vote in the Senate and the house is... precisely 50. Coincidence?
Next steps
I am back in an area with great internet now but I do have a particularly busy week at work. I will mainly update things via Twitter (@AU_Truth_Seeker) and will post here every day or two. I have one exciting numerical idea up my sleeve and intend on getting the time to do this analysis and post here towards the end of the week. I will also keep readers updated with my take on the near unavoidable court action relating to the WA recount.
Beyond that, I will update this blog approximately weekly, with analytical comments on the election we’ve had and predictions of future polls.
The problem with a re-count is that ultimately on close counts it will come down to subjective opinion on what is and what is not determined to be an informal vote. Even if challenges were limited to 2 per batch, would this amount to more than 14 votes? Of course it would.
ReplyDeleteMy personal view is that whilst Scott Ludlam has a legitimate right to request a recount based upon his 124000 first preference votes, it is a joke that the ASP with under 3000 first preference votes has even sought a recount.
Whilst the count may very well be wrong, the fact is that it will be more accurate then any potential recount could possibly be as the subjectivity will no doubt now be skewed by all 4 parties concerned, GRNS, PUP, ALP and ASP.
But most of the discrepancy is not going to be coming from questions of formality - it will be questions of addition. Do the number of ballots add up? Has someone miscounted a 50-bundle?
DeleteThis is where us, the voting public, is most owed a recount. The fact that party X or party Y's low or high vote has influenced the election of party Z is irrelevant.
With respect, that's not what S279B and S280 of the Electoral Act provides for.
DeleteFormality disputes in the Senate (barring abuse of process) should be relatively rare. Very few ATL votes are at all contentious. Even with BTLs there will now and then be a dispute about whether a vote exhausts at a particular point or not (eg if a voter's 5 and 8, or 11 and 17, are very hard to tell apart) but apart from voters whose 1s and 7s are very similar there should not be too many outright formality margin calls. Voters with bad handwriting caused by infirmity, or who are prone to confusion and not good with numbers, may well just vote 1.
DeleteI'd also expect that most Senate votes subject to formality disputes would have very little bearing on the outcome. The outcome is determined by two parties that between them accumulate less than 5% of the state vote at the point where one is excluded.
The big unknown to me - and the one that justifies a full recheck in principle (the problem being how to prevent abuses of process) is what proportion of the ballots that are for other parties might actually belong to S+F or AC.
In the event of a re-count, I bet there will be more than 14 disputed ballots heading for the Court of Disputed Returns.
ReplyDeleteThis is going to turn into a circus just like the US Presidential Election back in 2000 where no matter how accurate any recount is, the losing parties and their supporters are going to cry foul no matter what.
This is not an excuse for us to accept an incorrect election. With more scrutiny we will get a more accurate solution.
DeleteThe circuit breaker in 2000 was a partial judiciary who made a seemingly "courageous" call and not allowed a complete recount to proceed.
All very good comments. I suspect in my gut that this will degenerate into a court contest and that it will centre very much around the issue of informal votes, as these things usually do. Yes there may be some votes that have not been counted as truthseeker points out but even he has only come up with 2 worthwhile anomalies and even then these are unlikely to have much impact on the result if they are in fact anomalies and have not been counted. So I tend to agree with the comment that informals will be the major focus in the end. This is a lawyers picnic potentially. I think comments on Bush v Gore is appropriate in these sort of situations. But I do respect the comments of Kevin and Truthseeker.
ReplyDeleteAs for the 50 votes in Kambalda West, yes Labor is down 50 votes in the senate but as we know at this election the minor parties in the senate are attracting a much bigger proportion of the vote so I wouldn't expect the major parties to keep the same vote in the senate as in the HOR. For example the combined Coalition vote is down about 60 votes on the same basis. So it would seem to me if the thesis is correct, that votes haven't been counted then it is likely to be across both Coalition and Labor.
But once again this discrepancy issue assumes that all data for a booth has been correctly entered against the booth and not against some other booth or included (accidentally) in declaration votes. The votes may well have been counted but just misallocated. The same goes for the BTL votes in Derby. If as people have pointed out that the votes are counted twice and at 2 separate locations then this type of error could occur...you might even expect it to occur....having said that, yes it is possible that the votes haven't been counted at all..
The LIB Senate vote in O'Connor was heaps less than the primary, the ALP Senate vote in O'Connor was very similar... Except in Kambalda West.
DeleteThe 50 vote discrepancy here is not one which will change the result. But it's an indicator that the count is not perfect and adds to the weight of evidence that a fresh complete recount should occur.
I have looked at the number closely and don't entirely agree with that assessment, but we could go back and forth on this. The combined coalition vote is well down in that booth on the O'Connor average. At any rate I cant make the numbers neatly add up to 50 votes simply being missed from Labor as you suggest. it looks much more likely to be across both of the major parties on my numbers.
DeleteWhat needs to be pointed out is that we are working with incomplete information. Purely what is publicly released. We have no way of verifying the information so you could be right. But to justify a full count based on one piece of arguable evidence does not satisfy the recount test. I think you would need to show something more systemic than one or 2 discrepancies to satisfy that test. A recount will always be better than the original count but you could say that for every count.
On reflection and looking a bit closer at the numbers I am coming around to your viewpoint. I take your point and on the basis of the numbers you are talking about it seems like the number will be closer to 50 from Labor. I just had a problem that it could account for the entire difference. But it is possible on the flow of numbers although not entirely certain as my gut feel would be that a senate vote would translate to slightly less than the HOR vote for a Labor candidate purely because of the greater choice of left parties in the senate. It is possible that all Labor HOR voters gave 100% of their senate votes to Labor, I just thought it unlikely. So I bow to your greater knowledge! Keep up the good work on the numbers.
DeleteI still stand by my comment regarding discrepancies. I doubt these are numerous and big enough to make the case for a full recount. Nonetheless I think you will get your wish on the basis of the AEC being seen to do the right thing.....
In a recent article in the Fairfax press the Electoral Commissioner said there were about 10 cases of (wholly unintentional) multiple voting per electorate in 2010. There are 15 seats in WA so that’s about 150 cases of multiple voting, the vast majority (or all) of which would just be double votes. The typical profile of a double voter is an elderly person or a migrant from a non-English speaking background.
ReplyDeleteIt strikes me that elderly people may be more likely to vote for an explicitly Christian party and migrants living in Perth may be less likely to vote for an outdoor recreational party. I don’t know what other preferences the two parties are carrying by the relevant stage, but maybe there could be a couple of votes in it – I haven’t tried to do any calculations and obviously the vast majority of the double votes would be irrelevant to the key contest.
Of course, a recount doesn’t help with double votes. I think if the margin of victory is smaller than the number of multiple votes in the House, a new election is called. I’m not saying this should happen for the Senate though – other people have pointed out that margins at early exclusions are usually very small.
Here’s the Fairfax article: http://www.canberratimes.com.au/technology/technology-news/hacking-fear-if-electronic-vote-replaces-pencil-paper-20130920-2u5au.html
Fascinating! Yes, I remember this article. THat 10 votes per seat would probably also throw open the NSW Senate election to a challenge by the Shooters... But I haven't checked these numbers in a while although it was closeish to 500 from memory.
Deletebtw has anyone looked at these Fairfax recount numbers released by the AEC. Maybe I am reading this wrong but it looks like there are a massive number of ballots being challenged. They mention 2900 ballots out of 20000 being sent to the top decision maker......how can that be? that is almost 15% of the ballots! and about 50% of the ballots were challenged initially mostly on authenticity.
ReplyDeleteAm I a complete muppet or is that astounding.....
Happy to be guided by the usual sage advice from Kevin and Truthseeker.
If right it shows the perils of a full recount in WA........
Yes I have covered the high challenge numbers in the Fairfax recount here: http://kevinbonham.blogspot.com.au/2013/09/2013-federal-election-late-counting.html
DeleteRegression by booth strongly suggests that the Palmer crew are claiming many if not most votes that oppose them are not authentic ballots, and are then challenging the district level decision on authenticity and sending the papers to the top. There is nothing in the rules to stop them doing so as a scrutineer can object to any ballot and can challenge any decision by the DRO.
(To the Anon 12:11pm; I'm not sure if this will appear as a threaded reply.)
ReplyDeleteKevin is all over those Fairfax numbers: http://kevinbonham.blogspot.com.au/2013/09/2013-federal-election-late-counting.html
Truth Seeker called it "facetious" in the original post; I think 'vexatious' is a better word, by analogy with vexatious litigants. My read of the numbers is that the PUP scrutineers are probably challenging the formality of any vote for the LNP that's in any way ambiguous (this is run-of-the-mill stuff, and the LNP scrutineers would be doing the same on votes for the PUP), and challenging the authenticity of the vote otherwise. The latter is vexatious, and this is the behaviour that Kevin worries about with a full Senate recount. It's hard to tell from the table because it's not clear what the blank cells should be, but the AEO might end up with 40,000+ ballot papers to rule on, and of course the process is taking a lot of time. Hopefully a lot of the blank cells are actually zeroes and it won't be quite that ridiculous.
(Do the journos who read this blog also read the comments? It'd be good to interview someone from PUP about what's going on. I can't really make sense of the strategy at all.)
The procedures on recounts were presumably written with the assumption that scrutineers would be acting sensibly, but they'll need re-writing to avoid this in future.
(Off-topic: there's a commenter here who goes by "David B" - just so it's clear, he's not me!)
I know that media from a number of different outlets follow my blogs, and I think my comments. At least one MP and a number of staff of MPs follow my twitter feed, but I use a "quality rather than quantity" approach to twitter.
DeleteI don't think Journos will like sharing with each other what their next story is. But perhaps PUP is not keen to divulge what they're doing anyway.
Not withstanding a miscount in the ATL group voting and a missing bundle of votes. I am yet to do a both by both analysis comparison with the lower house.
ReplyDeleteWhat also needs to be noted is that had a Weight Surplus Transfer system been adopted the Greens and PUP would have been elected to the last two positions. If we also remove the distortion that exists in the segmented distribution of excluded candidate votes and adopt a reiterative count where the vote is reset and restarted on each exclusion then PUP and the ALP are elected.
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