Sunday 26/10. Final result: Labor 7, Liberal 6, Greens 4.
The count has now been finalised and the Greens have indeed won a fourth seat in Molonglo, their candidate leading 9457 to the third Liberal’s 8536 at the key point in the count. As I should have noted in the previous entry, there was also a close race between the second and third Greens candidates which has in fact been won by Caroline Le Couteur, who overtook Elena Kirschbaum late in the count. Kirschbaum had 4203 votes at the point where she was excluded to Le Couteur’s 4285.
Saturday 25/10
In Molonglo, we now have a preference count for 62,577 out of 88,291, and Antony Green’s assessment is that “the Greens are starting to be favourite for the final spot”. On the present projection, second Greens candidate Elena Kirshbaum leads third Liberal candidate Giulia Jones 6660 to 6166 at the relevant count. The Liberals are likely to close the gap in what remains of the count – the primary votes that have been admitted to the preference count have gone 31.3 per cent Liberal and 18.5 per cent Greens compared with 31.4 per cent and 18.2 per cent from the total – but my back-of-envelope calculation tells me they will only be able to close the gap by perhaps 200 votes.
Tuesday 21/10
The count in Molonglo is getting progressively more interesting, with second Greens candidate Caroline Le Couteur just 49 votes behind third Liberal Jeremy Hanson at the crucial point in the count. Le Couteur herself leads the third Greens candidate, Elena Kirschbaum, by 49 votes at the relevant earlier point of the count. So the result could yet be Labor 7, Liberal 6, Greens 4, rather than 7-7-3.
Sunday 19/10
This post will be updated progressively with details of late counting in the ACT election. Two results remain in play: in Molonglo, which could either go Labor 3, Liberal 3, Greens 1 or Labor 3, Liberal 2, Greens 2, and in Ginninderra, which could either go Labor 2, Liberal 2, Greens 1 or Labor 3, Liberal 1, Greens 1. The most likely results will produce an outcome of Labor 7, Liberal 7, Greens 3, but other possibilities are for the Liberals to win as few as five, Labor to win eight or the Greens to win four.
In Molonglo, the Liberals are on 2.48 quotas on the primary vote and the Greens are on 1.48, so whoever does better on preferences will win the final seat. The problem for the Greens is the 2.7 per cent recorded by Liberal-turned-independent Richard Mulcahy, which based on pre-poll votes looks likely to go about 35 per cent to the Liberals and maybe 10 per cent to the Greens. Against that is that the Greens can hope for a strong rate of preference leakage from Labor. There is also an outside chance that independent Frank Pangallo could sneak through and take the seat if he receives enough preferences from minor candidates, but it would have to be rated a long shot.
In Ginninderra, Labor are on 2.41 quotas and the Liberals are on 1.64, the risk for the Liberals being that Greens preferences after the election of their candidate will push them ahead. However, the gap is probably wide enough to get endangered Liberal incumbent Vicki Dunne home.
I’m mystified by your 600 votes comment Canberra boy. The Green vote on that distribution is 1.47 quotas as opposed to 1.46 quotas on the full primary count. The Labor ticket is 2.80 quotas compared to 2.90 quotas, the Liberal vote 2.62 as opposed to 2.52, Pangallo 0.36 compared to final 0.39. The battle is still between Hanson, Jones and a second Green. Hanson will have the higher primary vote but Jones is consistently favoured in preferences from other Liberals. Together with the operation of Robson rotation, this is keeping Hanson and Jones’s totals roughly equal, splitting the Liberal ticket and keeping them just ahead of the third Green. I think the point is that both Jones and Hanson may be 100-200 votes higher than they will be at the end of the count, which makes the final race closer. And the 2nd and 3rd Green have 0.65 quotas togather where by the end of the count they will have 0.59 and need more preferences to get to their current tally.
And the current count has Labor 0.10 short of its final primary. Corbell will have an extra 400 votes on the current count which puts him out of any difficulty.
Jones is still overperforming compared to her final total and Hanson is still underperforming on first preferences. But will other voters in booths that had high Hanson first preferences be more likely to preference him over Jones or will she continue to grab a larger slice of the later prefs?
I got as far as Fadden in the list of 2007 Canberra FE booths before the case for the opposite view occurred to me. That is, if the Liberals poll well in a particular booth that could mean that they are successfully competing for primary votes among those who would have otherwise put them second, which could mean one would expect a degree of corellation in the *opposite* direction (ie strong primary performance in a booth = bad performance on preferences in that booth). Nevertheless I shall press on with my hypothesis from #98 and see what happens to it (aint science grand?)
Good luck, Kevin.
If I didn’t have half a dozen papers to write I’d join you.
Of course that’s with the 3 dissimilar parties. When it’s within the Liberal Party the the trend could be different. The question is really where Zed’s excess and the votes from the other excluded liberals will go.
Some analysis on the latest distribution – please note caveats. Antony, I make the Lib preference split between Hanson and Jones as pretty much even, maybe Jones marginally ahead.
Now a big caveat here is that the interim distribution of preferences is still too early for my projections to be reliable, but for fun here they are:
1) On the current preference flows the greens will end up with 1.743 quotas and the Libs 2.788 – indicating that the final seat is down to 0.045 quotas. This also suggests that the winner of the final seat will not actually get a quota.
2) As I’ve said before the fate of Caroline Le Couteur is determined by whether or not Elena stays ahead of Pangallo. If she doesn’t – most likely – then Caroline ends up on 0.653 quotas. If she does stay ahead of Pangallo – less likely – then Caroline ends up on 0.772 quotas – which is probably enough to outlast Jones
3) I can’t do a remotely reliable projection of where Jones ends up on this distribution as Hanson is neither elected nor excluded. A very rough figure is 0.424 quotas- which is obviously wrong but perhaps indicates she’s behind
On preference flows within the Libs, on the current distribution:
Seselja: 16% to Hanson, 20% to Jones
White: 17% to Hanson, 16% to Jones
Kent: 30% to Hanson, 26% to Jones
Barnier: 21% Hanson, 15% to Jones
Burke: 36% Hanson, 42% to Jones
Which if anything slightly favours Hanson, but it could change
To analyse the final counts, when Mike Hettinger is excluded, the totals are as follows
Pangallo 0.44 quotas
Labor 1.74 across 2 candidates
Green 1.83 across 3 candidates, lowest Green 0.42
Liberal 1.71 across 3 candidates
The Pangallo going out theory suffers because his vote will rise as the rest of the primary count is excluded while the lowest Green candidates primary vote will fall. The gap between a third Green and Pangallo will widen for the rest of the count.
The third Green goes out and elects Rattenbury and then his surplus is distributed. Totals are now
Pangallo 0.46 quotas
Labor 1.77 across 2 candidates
Green 0.75 with one candidate
Liberal 1.73 across 3 candidates
Pangallo goes out and also elects Andrew Barr. After his preferences and the tiny Labor surplus, totals are
Labor 0.88 quotas with one candidate
Green 0.83 with one candidate
Liberal 1.85 with three candidates
Jacqui Burke now goes out
Labor 0.89 with one candidate
Greens 0.85 with one candidate
Liberal 1.75 across 2 candidates, totals 0.88 and 0.87 quotas.
Compared to the final primary totals, Labor’s vote will be higher, Liberal vote lower, Pangallo higher, Green second candidate lower, and minor parties whose preferences are leaking to Liberal slightly higher.
That’s one very close final count. The Greens need one of the two final Liberal candidates to open a lead over the other. At every stage of the count so far, the final two Liberals have been neck and neck. It’s very similar to the Brindabella count in 2004, when the Green just couldn’t get ahead of one of the two remaining Liberals.
I scrubbed the booths Oakes Estate, Tharwa and three Special Hospitals from my 2007 Canberra FE sample for being too small. For the remaining 37 the net result is that there is no remotely significant rank corellation in either direction between the Liberal primary vote and the proportion of preferences received by the Liberal candidate (r=.096, p=.56). Of course if you throw a very large amount of data at it from lots of different booths rather than just 37 booths from one electorate you might get somewhere but for a small sample either the factors pulling in opposite directions cancel out or else any corellation is totally drowned out by slop. Might do this on a larger scale sometime if it has not already been done.
And areaman is completely correct – things might very well be different when talking about booth-by-booth preference flows within a party. Maybe I should have spent the above time doing something more useful like a full notional distribution of preferences using the updated primary figures as a base.
Something worth bearing in mind when looking at these kinds of situations (and I’ve seen a few of them) – when people are talking about the Greens being .xy of a quota behind on a distribution based on incorrect primaries, and actually being .zw of a quota better than that in the real primaries, the .zw doesn’t translate fully to their final position. An increase in the Green primary at the expense of the Libs increases the Green exposure to leakage while decreasing the Libs’, meaning that only most of the vote gain is effectively reflected in the final position and not all of it.
The scenario of a candidate losing because they are the sole remaining candidate opposed to two fairly evenly split ones from another party is one that I often talk about in Hare-Clark but practical examples of it happening are rare. Will be interesting to see if this turns out to be one or not.
Kevin, it happened in Brindabella in 2004. At the end of count 45, one Green had 0.58 quotas, two Liberals had 1.52. But the Liberals had 0.71 and 0.81 which meant the final Green was distributed at this point. I think it is something more likely to occur in the ACT than Tasmania, because the vote between candidates is much more randomised in the ACT by the extra rotations, and because the individual candidates tend to be less well known.
The fact that the greens only ran 3 candidates and the Libs 7 means that they’ll have lower leakage, but yeah more votes means more leakage (in votes, not percentages).
So The Greens would have had a better chance if they only ran 2 candidates?
Jimbo Cool, hate to disagree on the preferences, but I think they favour Jones when you take account of the size of the bundles, and when you pull out Seselja’s preferences that can be identified at every stage of the count.
Of Seselja’s surplus, 827 finish with Jones,777 with Hanson, 41% to 38%
White to Jones 68 (18%) Hanson 65 (17%)
Barnier to jones 153 (24%) Hanson 132 (21%)
Kent to Jones 274 (26%) Hanson 316 (30%)
Burke 660 to Jones (42%) and Hanson 559 (36%)
However, Jones’s primary still has to slip anoth 0.08 of a quota to the final primary. That opens a real opportuinity for the Greens to get ahead.
It depends on how many votes the candidates got for the greens ticket that the party wouldn’t have if they hadn’t been there.
Anthony, the question which I asked at the top of the page that no one seems to know is if if Jones’ share of liberal preferences will fall inline with her declining primary vote.
That’s because no one knows.
That’s what I figured :).
Areaman you could make an educated guess though – using either of Antony’s or my estimates of preference flow ratios and apply them to the uncounted booths – ignoring the Zed vote. Add the Zed preferences back in at the end as a ratio of his overall primary overquota. Laborious perhaps but you’ll have an idea. To do it this way you should use my ratios I think because Antony’s done some fancy juju to the Zed overquota on his Turing machine. I’m using pen and paper and the power of procrastination…
But my argument is that her flows (as a percentage) will fall along with her primary. All doing that maths would show is how many pereferences she’d get if the ratios stayed the same.
I think the preference flows will be exactly the same. There does seem to be a slightly higher flow of preferences to her from both Seselja and Burke, but the rest are essentially random, as occurs under full Robson rotation.
Well, Kirschbaum gets to second on the Green ticket, but only on a count back, and then wins the last spot after getting 58 votes ahead of Jones. this count will go on all week.
Bummer – no update on the Elections ACT website…
I disagree that the preferences are “essentially random, as occurs under full Robson rotation” – if a voter marks out 1 to 7(which is the bulk of the major party votes) in a ballot group they do it in an order that isn’t random, and the mere presence of Robson rotation doesn’t make it so. If it were that simple surely we could just rake a random sample of ballot papers and build a confidence interval from that – no need for Antony’s juju and Turing machine, no need for a full count. Which reminds me, where’s Possum? Now that we’re talking stats surely he has a regression that will help us?
To what extent foes donkey voting occur under Robson rotation- does anybody know, is it possible to detect it? There must surely be a tendency, even for a “thinking voter” for people to say- “this is too hard” or “it’ll all come out in the wash anyway”, so they just vote down the ticket for the Party they choose and for the ballot-paper permutation they’ve been handed. The system was initially designed to stop intra-party squabbles over who was going to be top of the pecking order- a rather strange tool to intercede within a party.
These permutations are limited by the technology of the printing process (in Tasmania anyway) and generally the candidate order 1234 is just rotated in each batch i.e. to 2341, with each column rotating in lock-step. It’s just like an Enigma machine really and it might be possible to do a Turing and “break” it? You could get horizontal donkey voting as well, in which there would be a correlation between parties for a given candidate.
In the case of electronic voting it ought to be possible to present each voter with a more truly random ballot “paper” without such intra-party and inter-party correlations. Is this done? The Tasmania legislation, at least, doesn’t trouble itself with how the aim (each candidate appears at the top of the ballot in equal numbers) is achieved, so there would be no particular trouble in making it happen.
Double bummer- there is an update just as I’m trying to be clever!
Well on Antony’s figures it will take all week as we have the Greens best booth coming up next, Lyneham, followed by a good one for the Libs, Ngunnawal, followed by a good one for the Greens, O’Connor etc.
Hey all,
I missed this thread for a bit so missed some of the discussion, so I’ll respond to some old comments.
1) Oz wondered about why the Greens haven’t decided on a leader. I don’t think it’s that there’s two people competing for the position (who people would guess would be Meredith and Shane), I think it’s that both of them want the other one to do it. Remember that none of them have any parliamentary experience, and the ACT Greens have never had a leader before. It’s made more complicated by the fact that they don’t know if there’ll be a 4th MLA or who it will be.
2) I think the Greens will have to go with Labor in most circumstances. The Greens actually did suffer in the polls after supporting Kate Carnell’s government. But I think it is possible to support the Liberals if the ALP are seen as being complete bastards. So if Labor just refuses to negotiate and dares the Greens to support the Liberals, I reckon they would, and they’d get away with it. After all, that’s what happened in Tasmania in 1996, and I think that government worked quite well.
PS. Even though I live in Canberra now I haven’t been that involved in the ACT Greens, so these are just my hunches, I really don’t know what’s going on in the meetings any more than a Greens person in Sydney or Melbourne.
And I’ve also written some commentary on the progressive results at my blog.
I’d assume they didn’t have a whole lot of choice in supporting Carnell’s Government though right?
I understand you being mystified by my 600 votes, Antony. I foolishly made my comparison of the primary vote quotas in last night’s count with the Electoral Commission’s rounded-off figures of 2.5 quotas for the Liberals and 1.5 quotas for the Greens in the full primary vote count. The advantage the Liberals had in last night’s distribution vis-a-vis the Greens was actually 0.08 of a quota, or more than 300 votes more than they would have had if it were in proportion to the overall vote.
Tonight’s count, where Elena Kirschbaum takes the last seat, involves the Liberals primary vote in the distribution being advantaged relative to the Greens by 0.07 quotas or more than 400 votes when compared to their relative position in the overall primary tallies.
Hanson & Jones each have a share of the primary votes counted tonight significantly higher than their individual share of the overall primary vote. Kirschbaum similarly has a disproportionately high share – but less than the two Libs. Not surprisingly, Le Couteur’s share of the primaries in this distribution is lower than her proportion in the total count.
Antony can you explain the countback – when the two Greens are both on 2331 votes at step 175, does the countback involve going back to the last occasion that one of them was ahead, or is it a look back at their primary vote?
Just on timing of the count – I understood the Electoral Commissioner was saying yesterday that the count would take until next Wednesday. Today he seems to have said it would take until the end of next week.
There were 387 million votes cast at the last Indian general election, and the ECI had every last one of them counted, by hand, and up on their website in a week. And look at this: 200,000 votes to count and the end nowhere in sight.
The ECI probably had 100,000 people working on the count. And if memory serves me correctly, India uses first-past-the post. Hare-Clark combined with the anti-democratic, elitist Robson rotation must be a nightmare when it comes to data entry of all preferences.
How is Robson rotation anti-democratic and elitist?
Because it denies groupings of candidates the ability to present themselves in an order of their choosing; denies voters who wish to support a party (and other election systems which offer this opportunity as an alternative to full preferencing – eg Senate – show that the overwhelming majority of voters prefer it) the opportunity to do so with greatest effect for the party; and assumes that voters should be educated, informed and interested enough, and endowed with enough time, to be able to choose between individual candidates.
Hear hear.
Geoff (#123), there are some stats on the incidence of what the TEO calls “linear voting” (which is essentially donkey voting but for a particular party) in both ACT and Tasmanian Hare-Clark elections in the TEO’s discussion paper at http://www.electoral.tas.gov.au/pages/Media/PDF/Robson_Rotation_Paper.pdf
Robson Rotation takes power from elites in political parties and gives it to voters which is not elitist but in fact anti-elitist (and why Adam & co don`t like it).
With Robson Rotation voters who cannot be bothered choosing between candidates can and do do an internal party donkey vote “linear voting”.
Surely the “elites” are the people who run the political parties, not educated voters?
What Tom said.
Hardly surprised the party hacks don’t love Robson rotation – a system where the most popular candidate wins, as opposed to the one the party bureaucracy would most like to install, is never going to be their fave.
It was an elitist decision to impose Hare-Clark on the ACT in the first place. There was a poll at the time which showed that the people wanted single-member seats. But someone said “Oh no, we can’t have that because the demographics of Canberra mean that Labor would win all the seats, so they have to have PR.” That was elitist in the sense that there was a deliberate decision to ignore what the people of the ACT wanted, because it was felt that giving them what they wanted would be bad for them. In contrast, all Canberra Boy said was that parties should be allowed to list their candidates on the ballot paper in the order they choose, and that voters should be allowed to vote for a party list of they want. I don’t see anything elitist about that. I suspect Robson rotation is just a device to reduce the Labor vote, since it is mainly Labor voters who want to vote for a party list, and who are more likely to be confused if they don’t have a how-to-vote that matches the ballot paper.
Actually Adam there was a referendum when it was decided to move from modified D’Hont. The two choices were Hare-Clark or single member electorates. Hare-Clark won easily:
http://www.elections.act.gov.au/elections/referendum_92.html
Then the next election the electorate was asked if it wanted to keep it:
http://www.elections.act.gov.au/elections/referendum_95.html
which it did.
OK so they changed their minds. That doesn’t negate my point about the original decision.
Do you have anything to back up your memory of that poll?
Surely two referenda involving the entire ACT are more accurate than one poll?
No, it’s just my memory, which is usually pretty good for this sort of thing.
It’s not a question of accuracy. I’m not disputing the later votes, I’m making a point about the original decision to impose PR on the ACT, since people were complaining about the “elitism” of wanting voters to have the option of party-list voting. In fact imposing self-government at all on the ACT was “elitist”, since it was pretty clear at the time that they didn’t want it.
I don’t think “elitist” means what you think it does.
My definition of elitism (in this context) is the behaviour of people who by virtue of their position or education assume the right to impose decisions on other people either against their wishes or without consulting them, in the belief that they know best what is good for people.
Deputy Liberal leader Brendan Smyth maintains Canberrans rejected Labor on Saturday
“You’ve just got to look at the big number, that 64 per cent of people did not vote for a Labor government, they did not vote for Jon Stanhope,” Mr Smyth told ABC Radio.
Spot the flaw in this argument.
So anything that’s not populism then?
How about the fact that more did NOT vote for a Liberal government? About 68%.
Well Robson Rotation forces the candidates to get out there and campaign. If candidates are unable to inspire people to vote for them then that’s their problem. Further, there’s nothing stopping the parties from circulating material which numbers candidates in an order if they want to (which is actually what the Greens did). I still don’t understand how it’s ‘undemocratic’.
All decisions are ‘imposed’ to some extent. The decision to choose Robson rotation is no more or less of an imposition than choosing to allow the parties to determine the order their candidates are placed on ballot papers.