Full house? (part three)

With another Victorian upper house thread having extended beyond 200 comments, a further update would appear in order. At the time of the last post, Labor’s hopes for an upper house majority appeared to hinge entirely on Southern Metropolitan. Since then, the chances of a DLP win at Labor’s expense in Western Victoria – always possible, but somehow too bizarre to contemplate – have increased considerably. The DLP scored 2.6 per cent of the primary vote in this region, which is subsequently engorged by preferences from the Country Alliance and People Power, putting their candidate Peter Kavanagh fractionally ahead of Family First. Family First preferences in turn get Kavanagh ahead of the Nationals, unlocking enough Coalition preferences to get him ahead of both the Greens and Labor. At this point of the count, the Greens hold the narrowest of leads over Labor (130 votes, according to Antony Green); if they stay ahead, Labor will go out and their preferences will push Kavanagh over a quota. Otherwise, the Greens will go out instead and their preferences will deliver the seat to Labor incumbent Elaine Carbines. Another alternative scenario that might thwart the DLP is if what Antony describes as "a big whack of postals from a seat contested by the Nationals" put them ahead of the DLP at the earlier point of the count.

Southern Metropolitan remains on a knife edge, except that the Greens’ position has firmed in late counting – so it now looks like a contest between Labor (Evan Thornley) and Liberal (David Southwick) for the final place, rather than the three-way contest for the last two places that was in play earlier. According to Antony Green, the ticket votes alone leave Thornley 3844 short of a quota after the addition of preferences from Democrats, People Power and independent Rita Bentley, which the distribution of the Greens surplus should cut to around 2500; while David Southwick is 2934 votes short after receiving preferences from Family First and the DLP. However, there are more below-the-line primary votes for Liberal than Labor (5602 versus 4961), and in particular, there are more for Southwick than Thornley (1683 versus 633). Then there are the 5,275 below-the-line votes for parties other than Labor, Liberal and the Greens. The destination of these preferences look set to decide the issue, bearing in mind that many will exhaust given that voters are required only to number five boxes.

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.

171 comments on “Full house? (part three)”

Comments Page 4 of 4
1 3 4
  1. So Evan Thornley is going to be elected?
    Will the DLP win a seat in the Upper House?
    Surely there aren’t too many votes left to count?

    Today’s Newspoll: The ALP leads the Coalition by 10 points on a 2PP basis.
    How long will the Kevin Rudd Honeymoon effect last?

  2. Lyle Allan said: I disagree with random distribution of surpluses, like Antony, but I don’t feel they make much difference.

    Analyses have been done of this for Irish elections, based on the statistics of random samples. They come up with with an estimate that, over the years 2.2 to 2.5 “wrong” candidates may have been elected, when compared with Gregory-type methods. They can’t identify where (or even if) these occurred of course. As for fractional “numbers” of politicians like 2.5, this is not unusual in Ireland- nor in Tasmania- where many pollies can wear two hats without even trying (I can say that, I come from there).

  3. Antony said: Geoff, that’s not right. The surplus goes out from the highest to the lowest, as in the Senate.

    Urk!…. well, it WAS true, as version 14 of the Act from 2005 says:

    the total number, if any, of other votes obtained by the excluded candidate on transfers are to be transferred from the excluded candidate beginning with the ballot papers received at the lowest transfer value, as follows—….

    There are 20 versions of the Act on http://www.dms.dpc.vic.gov.au/Domino/Web_Notes/LDMS/PubLawToday.nsf Has anyone combed through them to see when the order was flipped? And why? Was the original a drafting error?

  4. I don’t think the above was from Version 14. It may have been from Version 19, which now cannot be found on the site. I probably downloaded the version I have been working from from another site. Version 20, which is the current real McCoy, says it incorporates amendedments made up to 31-Oct-2006. Was it really amended for the LC as recently as that?

  5. Geoff Lambert pointed out a statistical analysis of Irish election results on random sampling. Thanks Geoff for pointing this out. I think a similar mathematical exercise was done in Adelaide in the early 1980s. The late Henry Mayer referred to it in Politics once and it came to the conclusion one Senate election per century might be decided differently under random sampling.

    The trouble with random statistical analysis is that they generally don’t use actual data. The only Senate elections which might have been different since 1949 have I think one in NSW where a candidate won by a very small number of votes, but that was after the adoption of the Weighted Gregory method in 1984. Anthony Green might know of other close results. I think the 1964 McManus result was one the DLP thought that random distribution might produce a different result, but that election was, I think, decided correctly and the chances of random selection producing a result that would be different from the Hare Clark (not weighted Gregory) method were in fact reduced by the method described in my previous correspondence on this issue.

    I think off the top of my head that if a random statistical analysis were done of the 1964 Victorian Senate result it might show the chances of another random sampling producing a different result (that is, the election of Hannan rather than McManus) would be something like 998 or 999 in 1000 or even 9999 in 10,000.

  6. Sean … I can’t believe I wrote that. I think it must have been somebody masquerading as Ray.
    Believe me I do understand how Hare-Clark works and my concerns about OPV are real. However, as David points out the solutions to fix these may be worse than the problem. What others seem to find objectionable is the GVT system more so than OPV.

  7. Guys I asked this question before but didn’t get an answer. Maybe someone could help me.

    RE: SMET

    When the Democrat votes (all .1of a quota assuming BTL follow the ticket) go over to the Greens and the Greens subsequently get elected, the excess votes go over to the ALP. Here is where I get confused. People talk about the vote being diluted when it goes over to the Greens and then the ALP. What does it mean by diluted. If one votes FF and then Liberal then obviously the Liberal Party get that whole vote once FF is knocked out. If someone votes Dem, Green and then ALP, what happens when the vote goes over to the Greens but the Greens have reached their quota.

    Again for the sake of the exercise assume all BTL votes follow the ticket, does Thornley get .04 of the quota once the Greens take their .06 of the quota or is it a smaller number than .04 because of this term that I can’t get my head around “dilution of votes?” How is this dilution worked out.
    Laymans terms please in your answer.

  8. If a Democrat vote is excluded and the second effective preference is to Southwick, that goes out at full value. Effective second preference means that preferences to candidates already elected or previously excluded at the time of the Democrat exclusion are not considered at that point, that is the point of the Democrat exclusion.

    If a Democrat vote is excluded and the effective second preference is to Pennicuik, the Green candidate, and the next effective preference is to the Liberal Southwick, the preference to Southwick is devalued. That is because it is included with all the other Pennicuik votes at the time of Pennicuik’s election. If, off the top of my head Pennicuik had 90,000 atl Greens ticket votes, 6,000 btl first preference votes 2000 votes from the Democrats and 2,000 votes from others, and a surplus above the quota of 40,000, all 100,000 votes would go out at a transfer value calculated by dividing the surplus by the Pennicuik vote at the time of her election. This means the 100,000 votes would go out at a value of 0.4.

    Democrat votes that preference Southwick directly have a full value. Democat votes that preference Pennicuik and then Southwick are worth 0.4 when received by Southwick.

    That’s why Thornley supporters would prefer Democrats and others btl who are not going to vote for Thornley as an effective second preference to preference Pennicuik first, for their vote will then go Southwick at a reduced value. That’s why the Democrat Southwick vote is diluted. Of course the Democrats who number btl Pennicuik and then Thornley also go out at 0.4, but the important thing for the ALP is that the votes from Pennicuik to Southwick will be a very small proportion of the total Pennicuik surplus. The Pennicuik surplus will be largely from atl votes, which go directly to Thornley.

    Hope I’ve explained in simple terms John.

  9. John said. Guys I asked this question before but didn’t get an answer. Maybe someone could help me. When the Democrat votes (all .1of a quota assuming BTL follow the ticket) go over to the Greens and the Greens subsequently get elected, the excess votes go over to the ALP. Here is where I get confused. People talk about the vote being diluted What does it mean by diluted.

    Well what I think people mean is that only the excess over and above the quota flows on to Labor and it has to be diluted among all the other votes.

    The DEMS give the GRNS about 6,200 PAPERS, worth 6,200 VOTES.

    This contributes to the GRN an excess above the quota of about 4,800 votes.

    Every single PAPER held by GRN then flow foward to the ALP (mostly) and the LIB (a little).

    The GRN excess VOTE is only 0.079 of a quota, so the number of PAPERS going foward is mutiplied by 0.079 to work out how many VOTES they are worth. In the case of the DEMS, this pans out to about only 150 VOTES.

    Thus 6,200 papers, worth 6,200 votes, come out of the GRN cut-up as 6,200 papers, worth only 150 votes.

    That’s dilution.


  10. Geoff

    Your example using approximate results is better than mine. The Greens on posted figures last night are about 1000 votes over a quota on ticket votes from the Democrats and the Greens plus first preference votes for Pennicuik. If the below the line exclusions go to the Greens before Pennicuik reaches a quota that is more votes that will be diluted. As you correctly point out a Democrat btl effective preference to the Greens will have very little value when it goes to either Southwick or Thornley, but what Thornley will want is for as many of these btl papers to be locked up with the Greens before the surplus is reached. Thornley will be the beneficiary of the Greens surplus because all the atl ticket votes from the Greens and the Democrat tickets go to him. I suspect a majority of btl first preference votes to Pennicuik will go to Thornley too, but I noticed when scrutineering of Friday a lot of these exhausted, but these exhausted votes will be a very small part of the surplus.

    My example used made up figures for the purpose of explanation. I should have used real figures, of course, but I don’t believe my method is wrong.

    I hope John that you can make better sense of it now you have a further explanation from Geoff as well as from myself.

  11. Quite so, Ray. Props to Mr S – for my part, I was saying that these freak outcomes tend to come unstuck in late counting, which was pretty much based on the precedent of the Fremantle Hospital Support Group’s near miss in the WA election.

  12. Well, I’m *sure* this will be the last I ever hear about “the Greens representation in parliament being out of proportion with their support base”.

    Greens 10.6% of the primary vote, 5% of the seats
    Nationals 4.5% of the primary vote, 5% of the seats
    DLP 2.0% of the primary vote, 5% of the seats

  13. StephenL is right that floating quota itself doesn’t delay counts. Where I referred to NUS counts I observed being slow, the slowness was caused by the other two quirks of the system, especially when their effects were combined – the throwing of all a candidates’ votes for a surplus rather than just the last parcel, and the retention of several decimal places of detail (the latter stopped very small parcels from exhausting, which greatly increased the number of parcels prior to ATV).

    (I’m personally a fan of throwing just the last parcel. I think it’s fair that those whose support for a candidate is probably softer, as indicated by their preference reaching that candidate later or at lesser value depending upon system, should have more say in the outcome for the remaining candidates. However I may just be saying that because just throwing the last parcel makes it easier for me to rort my vote to maintain its value through the count, beware!)

    I’m not that surprised by Geoff Lambert’s case of the decisive #15 preference especially given the large number of candidates and positions and few voters in that example. Many years ago in a Hobart City Council election with roughly 16000 voters, 17 candidates and six or seven positions, some votes at full value were active with (for instance) 9 for one candidate and 10 for another at a point at which one of these two candidates lost the final position to the other by a fraction over 3 votes.

  14. Isabella (December 8th, 2006 at 5:33 pm ) posted, “What will Peter Kavanaugh talk about in his maiden speech? The threat from Communist Vietnam? The Domino Theory? Give us all a break.”

    I replied (December 8th, 2006 at 6:08 pm ), ‘I do not know Peter Kavanagh or what he will talk about if he gets to give a maiden speech, but I can make an informed guess. It won’t be the non-existent threat from communist Vietnam, though he may make a reference to the suffering of the people there. I think he will talk about life itself, about his grandfather, about persistence in a cause you believe in, among other things, but I am content to wait and see.’

    Yesterday, Peter Kavanagh gave his maiden speech. He did not talk about the non-existent threat from communist Vietnam, though he did talk about the fight of democracy against communism. He also talked about life, about his grandparents, and about persistence and about much else besides. (Jo Chandler, “I see DLP people: Kavanagh raises ghosts in the upper house”, The Age, 14/2/2007). See for report:

    Or go to the Legislative Council Hansard for a more rounded account – the speech itself.

Comments are closed.

Comments Page 4 of 4
1 3 4