Photo finishes: Fairfax

This post will be progressively updated to follow the late counting in the undecided seat of Fairfax.

Friday 3pm. Absents have now been wrapped up: 298 were disallowed, and the last 105 counted have broken 59-46 to Palmer, increasing his lead from 98 to 111. Still to be processed are 415 pre-polls and 155 postals, of which Ted O’Brien would need nearly 60%. However, the comment from Julian T below suggests that a number of the pre-polls will likewise be disallowed, and have been left out of the count to this point pending a final determination on the matter. If that’s so, Palmer looks to have it wrapped up, pending what might show up on a recount.

Thursday 5pm. The race that’s stopping the nation has today seen the addition only of 201 provisionals, which were the biggest imponderable of the remaining count. It turns out that they have broken very handily to Clive Palmer, 148-53, boosting his lead from three to 98. Still outstanding: about 650 absents and as many pre-polls, and about 150 postals. If they break as such votes have until now, the LNP will only be able to carve about 50 votes out of Palmer’s lead. UPDATE: Julian T in comments observes that a number of the supposedly outstanding pre-polls and absents will not in fact make it into account by virtue of being “disallowed”, as they involve voters submitting votes for Fairfax when they actually turned out to be enrolled elsewhere in Queensland. Ballots from such voters can be admitted for the Senate, but not the House. That likely makes the hill very difficult for Ted O’Brien to climb.

Wednesday 4pm. Clive Palmer lost the lead earlier today but at the time or writing has recovered it – by three votes. Troublingly for Palmer (assuming he does actually want to win), a batch of 395 absents were less good for him than those admitted previously, going 52.6-47.3 his way against a grand total of 56.6-43.4. By contrast, 474 pre-polls have maintained their trend of going 58-42 against him. Still to come: about 650 absents, on which he would hope to gain maybe 80; 550 pre-polls, on which he should lose them again; a trickle of postals (100, perhaps) that would then be likely to put him behind; and about 200 provisionals which will do who-knows-what.

Tuesday 6pm. From 502 on the weekend to 362 yesterday, Clive Palmer’s lead has now worn down to 65 after today’s counting of pre-polls went 830-588 against him and postals went 436-378. However, the postals were somewhat less bad for him than previous batches, and there’s now very few of them left. The bulk of the outstanding vote consists of about 1000 absents and pre-polls each, which appear likely to cancel each other out with the former going 57.0-43.0 for Palmer and the latter going 57.5-42.5 against. There are also likely to be about 200 provisional votes, which are an unknown quantity. Assuming the latter go 50-50, I’m currently projecting Palmer to win by 34 votes.

Monday 7:30pm. The AEC explains the Coolum Beach anomaly. Clive Palmer sought a Federal Court injunction today to have counting suspended, though to what end I’m not quite sure. As the AEC notes, the normal practice would be to petition the Court of Disputed Returns. The court has so far reserved its decision.

Monday 7pm. It appears the Coolum Beach PPVC mystery has finally been resolved. The votes for that booth had incorrectly been entered for the Buderim PPVC and vice-versa, and only now has the error been corrected. Since there are still no Senate votes recorded for Buderim PPVC, my earlier scatterplot did not show up a corresponding error for that booth. While it doesn’t look like there will be salvation for Clive Palmer in the form an uncovered vote count anomaly, he has gained ground with the addition of 1291 absent votes which continue to favour him quite strongly, in this case breaking 722-569 his way and giving his faltering lead a badly needed boost from 209 to 362.

Monday 4pm. Looks to be going right down to the wire, with another 1223 postal votes slashing Palmer’s lead by 293 to 209. There remain to come 2400 absents and a similar number of pre-polls, which have respectively been heavily favouring Palmer and O’Brien, along with maybe 500 postals which have been favouring O’Brien 62-38. My projection of where this is headed leaves nothing in it. In other news, Clive Palmer has today been raising the issue of the Coolum Beach PPVC discrepancy, though not in a way that might inspire the casual observer with confidence in his claims.

Saturday 8pm. Another 1306 pre-polls have gone badly for Clive Palmer, favouring Ted O’Brien 762-544 and cutting Palmer’s lead from 718 to 502. My projection of the share of the outstanding 7200 or so votes needed by O’Brien is down to 53.5% (assuming once again that nothing comes of this Coolum Beach PPVC anomaly).

Friday 3pm. Another 1792 postals have been just as favourable to Ted O’Brien as earlier batches, cutting 414 into a lead for Palmer that now stands at 718. There should be a bit over 2000 to come, which should further cut into Palmer’s lead by about 550. That leaves the result well and truly down to absents and pre-polls, the likely behaviour of which is a bit of a mystery at this stage. As to the apparent Coolum Beach anomaly, the commenter who raised the matter has received what to my mind is an unsatisfactory response from an AEC worker who appears not to have properly grasped the issue.

Thursday evening. Leaving aside the Coolum Beach PPVC issue, which is yet to acquire a life independent of this website, postals are flowing heavily enough to LNP to suggest an extremely close result. The addition of 2363 have cut 611 votes from Palmer’s lead, and could potentially take out a further 1000 of the remaining 1132 if there are indeed 4000 of them outstanding and they continue to behave like the previous batch. However, Palmer has done much better out of 855 absents, on which he has gained 79 votes and of which there are about 2000 more to come. There are also around 2000 pre-polls, which look likely to favour the LNP.

Wednesday evening. Michal Klaus in comments notes that the Coolum Beach pre-poll voting centre booth, the addition of which did so much to swing the count in Ted O’Brien’s favour, “could be one of the most extreme outliers anywhere in Australia”. He’s not wrong:

As the above chart clearly shows, the 55.87% primary vote recorded for Ted O’Brien at the booth is entirely out of whack with its 36.27% Senate vote for the Liberal National Party. There are also 575 more votes recorded there for the House than the Senate. Clearly there is an Indi-style error waiting to be uncovered here, either in relation to the House or the Senate vote. Given there are ballpark similarities between the Coolum booths for the Senate and a strong discrepancy with Coolum Beach PPVC on the House numbers, it seems safe to assume that the error is with the House results, and that Ted O’Brien’s tally is consequently higher than it should be. To shift the Coolum Beach PPVC data point to where it should be in the above chart, about 1000 votes need to be deducted from O’Brien’s total.

Wednesday 5pm. The count continues to trend away from Clive Palmer overall, who perhaps faces a further headache from the yet-to-report “BLV Fairfax” pre-poll voting centre in Maroochydore. If it’s anything like the other PPVCs in Maroochydore, it could cost Palmer as much as 1000 votes of a lead which currently sits at 1664 (UPDATE: LTEP notes in comments that the BLV booths in fact do very limited business). Other late counting has been a mixed bag, with absent votes favouring Palmer 827-565, but pre-polls going 381-328 the other way. The total number of such votes should perhaps be around 4500 and 2500, and there should further be about 5000 postals, none of which have been counted yet. So a fair bit of life in this one yet.

Election night. I don’t believe Antony’s projection of 50.9% to Clive Palmer is based on anything other than a guesstimate as to preferences. It’s still clearly the best thing available, but the 0.9% figure would be well within its margin of error. I’m guessing the AEC will conduct a preference throw in fairly short order that will put the issue beyond doubt one way or the other.

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics.

193 comments on “Photo finishes: Fairfax”

  1. Stephen Spencer on Twitter says that Palmer will win by 60, if current trends hold.
    We are headed for a recount whoever wins, I guess – either one initiated by the LNP, or one initiated by Palmer.

  2. @Gdfollow on Twitter says they have alowed 450 Provisionals … The first 201 went PUP’s way 147 to 54

    It might be the LNP who find themselves running off to court.

  3. If it’s really 337 votes left to count, it’d have to be an extraordinarily good batch of votes for O’Brien for him to win by the skin of his teeth.

  4. Local radio might be confusing the remaining Absent votes with the total. Unless the current dec vote totals remaining are wrong. There seem to still be 911 votes to throw according to the AEC.

    Oh and most of the remaining provisionals got rejected so whoever said they allowed 450 provisionals was misreading it too.

    My projection now has Clive by 46 but still no sure thing. That amazing split on provisionals was huge for him, though it’s a bit less amazing when you look at the LNP’s dud performance on provisionals statewide in 2010.

  5. In regard to the provisionals, can some tell me the difference between:

    Ballot papers disallowed 233

    and

    Envelopes rejected at preliminary scrutiny 445

  6. Ballot papers disallowed relate to electors who claimed to be enrolled at an address in Fairfax, but were found to be enrolled in another Queensland division. Their Senate votes are counted, but their Reps votes are “disallowed” (because they were cast for the incorrect division).

    Disallowed votes will also be a significant discount from the remaining pre-poll and absent votes, as there are none recorded yet. This possibly means that Mr O’Brien has already reached a bridge too far.

  7. Julian T@158

    Ballot papers disallowed relate to electors who claimed to be enrolled at an address in Fairfax, but were found to be enrolled in another Queensland division. Their Senate votes are counted, but their Reps votes are “disallowed” (because they were cast for the incorrect division).

    Disallowed votes will also be a significant discount from the remaining pre-poll and absent votes, as there are none recorded yet. This possibly means that Mr O’Brien has already reached a bridge too far.

    I looked at those seats that have been declared and all of them have no disallowed EVPPs in the final declaration vote screen. However some have substantial numbers of disallowed absents while others have zero. Odd.

  8. Dr Bonham

    The threshold for the declaration of a poll is that the outstanding votes could not alter the outcome. No division yet has final figures, as there is still a day to go for postal votes to be received. However, I’m quietly confident that seats that have been declared would comfortably comply with the test in the first sentence, as the AEC are nothing if not prudent.

    For most divisions, the recording of disallowed votes is simply a final accounting measure.

    Every division will end up with disallowed absent, provisional and pre-poll votes. (I think they will have disallowed postal votes as well, but my knowledge of issuing procedures is slightly outdated, so I am not sure of this.) (Most absent, pre-poll and provisional votes are issued on the basis of an elector’s claim…the limited introduction of netbook computers will certainly have ameliorated this problem, and the scheme has the potential to virtually eliminate it.)

  9. OK + thanks very much Julian! I did notice the declared seats generally still had small numbers of unthrown votes but what I didn’t notice is that every time there is a zero given for rejected votes, there are still envelopes listed as awaiting processing in that category.

    So that would explain why there are reports that there are really only a few hundred left to throw in which case it is indeed extremely likely to be over, but for the recount.

    I have added a detailed note on Palmer’s performance on provisional votes today to my coverage at http://kevinbonham.blogspot.com.au/2013/09/2013-federal-election-late-counting.html . It is off the scale. He outdid his ordinary 2PP vote by over five points greater than the highest gap recorded by any candidate in any seat in 2010. Probably had I had a detailed look at 2010 provisionals I would have been expecting something like 60:40 or slightly better but nothing like 73:27.

  10. I presume that the way a person’s anonymity would be breached would be by those opening envelopes and any scrutineers present, who could see the sender’s name and address. I guess this also means there are strict procedures for removing ballot papers from envelopes so they are not viewed by handlers. But what if a ballot paper is incorrectly folded so the numbers are in full view when removed?

  11. My view on what will happen

    Palmer will win

    He will object to himself winning

    The court will order fresh election

    The electorate will be upset that they have to vote again

    Some donkey from the LNP will win the seat

  12. Palmer up by 111 after all absents thrown. 235 of them were binned.

    Now O’Brien is still in contention if the rejection rate of EVPPs is much lower than for absents, but I don’t know what that rate will be. Might see if I can dig some figures up from 2010. Even if all votes in theory on the table are thrown I would still project him losing by 18, but Palmer would want to win by more than that to feel safe on the recount.

  13. Clive out to 111 in front based on what seems to 109 more Absents counted going 61-48 to him, which is a vote or 2 below the trend for Absents.

    It seems that another 11 postals have come in, but not yet been counted. As per the link from Dr Kevin B, perhaps they’re waiting to do all the remaining declaration votes together to maintain anonymity – which means possibly waiting until tomorrow.

    Clive still the favourite, but should be by under 100 either way so a recount would occur regardless.

  14. And it has reappeared, whatever the reason.

    Clive Palmer has won.

    It is now up to the recount and the courts to decide matters. Lots of fun if Palmer keeps going off like he has been.

  15. What a thriller this is! O’Brien must have got something like 70:30 off those absents so far. I believe there are not really 415 to count as some of those absents are rejected. My latest projection of what is left to count has O’Brien “winning” by 10, but whatever it is, you can’t project a recount!

  16. @GdFollow on Twitter says 301 votes to count, which would mean 114 ballot papers disallowed. If those were all EVPP’s, then that would be about 2% of the EVPP total, which seems consistent with Banks (2.7%) and Bass (0.9%).

    I’ve got O’Brien winning this by 13 votes, but a poll-style “margin of error” is around 17 votes.

  17. Oh dear. Clive did pretty well on postals to only drop another 23. I don’t know if the seven remaining postals are real or going to be rejected. I project +1 or -1 depending on which way that goes (+/- 10 or so, +/- recount, +/- court case).

  18. Clive now only 22 in front with everything gone except 253 pre-polls (& only 7 postals, which seems a bit odd), many of which could be disallowed.

    On trend the Libs would get back in front if around 150 of the pre-polls were accepted.

    What’s the odds on a tie? (it’s all a bit academic seeing we’re going to have a recount anyway, but would add to the fun.)

  19. Conflicting reports of Palmer final margin of 8 and final margin of 36. It seems that someone has the split of the last absent bundle the wrong way around.

  20. 36 is final pending recount according to AEC on Twitter.

    Don’t think I’ve ever seen such an “unprojectable” contest. Every time it seemed clear which way the wind is blowing, or even that it was blowing towards a tie, one of them or other has done something statistically remarkable.

    36 must be likely to survive the recount. Recount plus court could knock it over though. Worth 37 votes combined in McEwen 2007. 🙂

  21. It’s certainly been a count for the purists. The sheer randomness of the vote flows at the end! A finish for psepho purists, no doubt enhanced by the personalities involved. Are we sure the counting is all over?

  22. Well, that’s a surprise. ABC TV news Melbourne reported that there will be a “full distribution of preferences” early next week, and if the margin is fewer than 100 votes there will be a re-count. So we don’t already have a full distribution of preferences?

  23. triton@186

    Well, that’s a surprise. ABC TV news Melbourne reported that there will be a “full distribution of preferences” early next week, and if the margin is fewer than 100 votes there will be a re-count. So we don’t already have a full distribution of preferences?

    No; what we actually have is a 2PP between Palmer and O’Brien on the assumption that they are the final two candidates. But mathematically it is possible, though unlikely in the extreme in practice, that the Labor candidate could beat Palmer into second and exclude him. The Labor candidate trails Palmer by 8.24 points but candidates below them combined have 13.95%. So they need to do the full distribution to confirm for the sake of certainty that this is not the case. In the process they are quite likely to pick up some of the minor errors that have been made along the way.

  24. Dear Australian Electoral Commission,

     I request that the Electoral Commission as part of its obligations
    and commitment to maintaining an open and transparent electoral
    process publish progressively during the data-entry process copies
    of the electronic data entry preference data files used to tabulate
    the election results.

    This information should be readily available and certified copies
    of the data files MUST be published prior to the declaration of the
    polls.

    There is no legislative or overriding reason why this information
    is not published on line as is the case with other electoral
    results. An electronic record of the data is the same as any other
    document or record.

    It is fundamental that our electoral system is open and transparent
    and subject to full independent scrutiny

    By refusing to make this information available for scrutiny
    undermines public confidence and brings the Electoral Commission
    and the election itself into disrepute

  25. I presume they’ve started on the recount, but eariler today the AEC site had Palmer’s winning margin changed from 36 to 37, and now it has moved from 37 to 38.

    all being recounted anyway I guess, but seems a bit odd.

  26. My guess is that they’ve discovered a few errors and corrected them. I hope they continue to provide updates of the progression of the re-count.

    I would think that they have organized the counted votes in such a way that the reason for any differences from the first count can be identified. Say, if votes for a given candidate are placed in bundles not too large (say, 10 or 20) they can check that each bundle has the right number and that every vote in it is valid and for that candidate. Then a not-too-large number of those bundles are themselves bundled, and so on. It shouldn’t be like counting a jar of jelly beans where you do it twice and get a different number but you don’t know why.

  27. Palmer’s lead now at 39, he told us it was rigged!

    It’d be nice if there was an indication of the progress of the recount (although I’m not sure exactly how this could be done).

  28. ltep
    [Palmer’s lead now at 39, he told us it was rigged! ]

    Creeping up gradually, but at the last moment the AEC will find an error that rips it all away. (Cannot let Palmer get away with these outrageous rigging claims!)

  29. If the Member (apparent) for Fairfax claims to possess compelling evidence, then he should present it via the appropriate channels.

    If not, then the Member (apparent) should resume his seat.

Leave a Reply

Your email address will not be published. Required fields are marked *