Sunday, November 30, 2014

Victorian Election Postcount: Prahran

 
(Correct prediction posted to tallyroom.com.au . Notice date.)

=======================================================================

UNDECIDED SEAT: Prahran (Lib, 4.7%) 
SUMMARY: Contest between Clem Newton-Brown (Lib) and either Neil Pharaoh (ALP) or Sam Hibbins (Green)
RESULT:  Hibbins (Green) wins seat - awaiting formal declaration

This article followed the post-count in the undecided seat of Prahran, won by the Greens (subject to official confirmation) on preferences from third on primaries after eleven days of counting, including a recount.

The original (after reworking) article appears at the bottom of the post with updates scrolling to the top.  However, there was a surprise in the prepoll count - Labor did much better than projected on within electorate prepolls, which I'm told they targeted heavily - so much of the original modelling for questions 1 and 2 soon became irrelevant.  The modelling for question 3 turned out to be slightly pessimistic for the Greens compared to the reality.

The results of the three key questions, after recounting, are being reported unofficially as:

1. Does Pharaoh stay ahead of Hibbins?  Hibbins knocked out Pharaoh by 31 votes.

2. If the final count is Newton-Brown vs Pharaoh, who wins? Newton-Brown led by 25 based on the quick throw.  This margin remains to be confirmed (and is presently irrelevant.)

3. If the final count is Newton-Brown vs Hibbins, who wins?  Hibbins has won by 277 votes thus apparently winning the seat.

Assuming these results are now officially confirmed, Hibbins has won the seat for now.  It is possible that there will be a court challenge, but even if there is one Hibbins will be able to sit in parliament until such time as a court might decide otherwise.

The result is a major success for the Greens, who have unseated a Liberal in a single-seat electorate for the first time in any Australian state or federally, who have won both lower house seats they targeted, and who will hold more than one seat in an Australian parliament elected under a single-seat system for the first time.  Indeed, they have only previously won four seats in single-seat elections (one of them twice) and two of those wins were in by-elections.



Full preference distribution from VEC
There's even an animated GIF of it for those who like such things!

Some quick trivia:

1. The Greens started the post-count needing 84.2% of preferences to win based on primaries at that time.  This came down to 83.0% at the end of the post-count.  They actually got 84.16%.

2. There was some speculation that Family First preferences would hurt the Greens.  They in fact got 40 more of them than Labor ... and appear to have won by 41!

3. The expression "non-monotonic" is used for election results in which a candidate was harmed by gaining particular votes, compared to if those voters had put their chosen candidate lower.  Famous non-monotonic election results include Frome 2009 (where Liberals could have won the seat by instead voting Labor) and Denison 2010 (where Labor voters could have won the seat by voting Liberal).  However, at the moment, Prahran doesn't have that property.  Sufficiently many Liberal voters could have caused Labor to make second by voting 1 Labor, but in doing so they would have caused Labor to win!

4. Cases of candidates winning from third in single-seat elections are rare.  This Tally Room thread has a list of some in comments.   It could be a big week for this theme with a threat of it also happening in the SA Fisher by-election.

======================================================================

Recount Updates

11:10 pm Wednesday: Yep, it's confirmed: the recount distribution is posted. 
That's just about the end now so thanks all for the interest!  Over 13,000 page views from over 5,000 readers. 

10:30 Wednesday: A final recount margin of 277 over Newton-Brown has been tweeted by a Green scrutineer on Twitter.

9:10 Wednesday: Well that was quick - Dinesh Mathew has tweeted that the Greens have won the exclusion point in the recount by 31. Awaiting official confirmation but it looks like the result will be declared and that Hibbins will take his seat pending any possible court challenge.

8:40 Wednesday: Recounted primaries have been posted. I'm not totally sure they are the final recount primaries, but suspect they are. Changes from the primaries for yesterday's distribution are: Newton-Brown (Liberal) -22, Pharaoh (Labor) +7, Hibbins (Green) -11, Gullone (AJP) +2, Walker (FF) 0, Goldsmith (Ind) -6, Stefanopoulos (Ind) -3, Menadue (Ind) -1.  In all, a net loss of 34, presumably to informal.

These changes won't prevent Hibbins from beating Newton-Brown if he makes the final two.  As for the margin between Pharaoh and Hibbins, there is a projected tightening of 18 votes to 23 based on these changes to the two candidates alone.  In the worst-case scenario for the Greens (that the votes gained all preference Labor and the votes lost all preferenced them) this could come down by another 12 to 11, but that's highly unlikely; more likely, they might lose just a few more.  So at the moment any risk for the Greens (if that is the final primary for the recount) comes if any votes are found to have been mis-allocated as preferences as well as primaries.  There's also an increasing risk if the margin gets closer that the seat will go to court, but that would not prevent them holding the seat until such time as the court decided otherwise.

I also note that if those are the final recount primaries then only the votes of the minor candidates (presently 1776 in all) are capable of overturning the margin, and the Greens would have to be extremely unlucky for at least six of those to be in the wrong pile. Indeed it is hard to see how such a thing would have escaped scrutineering.

6:10 Wednesday: Reports circulating that we may have a result tonight.  Between 10 pm to 1 am according to VEC via Alison Savage (ABC).  This is a very fast recount! 

3:00 Wednesday: The recount is underway.  The Greens, whose modelling and sampling has stood up rather well throughout this post-count, have modelling suggesting that Labor will still lose even if they get over the Greens in this recount.  If that is true it may be that the 2PP quick throw between Labor and Liberal had errors in Labor's favour.

9:15 pm Tuesday: Justin Barbour tweets that a recount has been ordered to commence tomorrow morning.

I am expecting a recount in this seat based on the closeness of at least the margin between Pharaoh and Hibbins, not to mention the other margins.  Both major parties would be likely to request recounts and the VEC can grant them apparently without any constraint at its discretion.  Updates on the expected recount will now be posted in this section, scrolling up.

========================================================================

Postcount Updates

9:00 Tuesday: Now being reported the margin is officially 262! "Reconciliation error".  It remains to be seen if this affects the margin at the earlier exclusion.  9:15: Nope, the earlier margin is now 41 votes.

7:27 Tuesday: James Campbell has reported that Hibbins has unofficially won by 128 votes.

7:15 Tuesday: Greens scrutineer Dinesh Mathew (who I add has been very helpful through this process) is claiming victory on Twitter.  Liberal scrutineers are reported by James Campbell as expecting to probably lose by around 120 votes.

7:00 Tuesday: Unverified reports that Liberal scrutineers are looking like a bunch of Appalled Cats at the moment.  I've also heard (a few days back) that the Liberals were rather slow to cotton on to the risk of the Greens getting up from third and were rather distressed when they realised.

6:30 Tuesday: They really might now!  A scrutineering sample has been reported with a flow of 407/459 (88.7%) to Hibbins.  I don't know if this includes minor candidate preferences via Pharaoh but in any case there are very few of those, and if the sample holds off Pharaoh's primaries only that would be enough.  Indeed if it holds for all votes the Greens could win the final exclusion by over 500.  A note of caution though: we don't know if these votes are still stratified in any way.

5:30 Tuesday: The throw of Pharaoh's preferences will take hours so don't expect a result immediately.  We should keep in mind that these preferences include not only the ALP's primaries, but also a few hundred votes that Pharaoh received from the minor candidates - votes that will not flow as strongly since none of them follow the card.  I'm somewhat sceptical that the Greens will quite get there, but they might.

5:00 Tuesday: Unconfirmed multiple reports that Hibbins has knocked out Pharaoh by 38 or 39 votes in the distribution of preferences, a figure likely to go to a full recount and possibly to court if confirmed.  Pharaoh also now trails by 25 votes on the 2PP.  Hibbins needs 85.7% of Labor preferences to win.

4:00 Tuesday: No, I haven't fallen asleep; there is sadly no news to report.

1:20 Tuesday: This really is extremely slow.  Still no clear indication of final primaries.  When the distribution starts I'll likely be relying on whatever scrutineers might post (if anything) as the VEC site won't update after each exclusion (at least if Frankston is any guide).  Meanwhile I should note that one major bookmaker has paid out bets on all three candidates.  No real information is conveyed by that move; it is just a publicity/customer-loyalty stunt.

12:20 Tuesday: A small primary update though the primary still lags the 2PP by 110.  The ALP-Green gap comes down by another four to 397.

11:07 Tuesday: More figures are up and if they are correct Pharaoh has closed Newton-Brown's 2PP lead down to a ridiculously close 11!  (Even if he loses I'm amazed he's got that close from nearly 300 behind).  The primary count is still lagging slightly but the last update had the ALP-Greens gap down to 401 votes with the Greens now needing 82.9% of preferences in the event that they are second.  Remember, if the Greens can get second on preferences then the currently showing 2PP gap is irrelevant.

10:15 Tuesday: The last dribbles of primaries have moved Newton-Brown ahead in the 2PP quick throw by 61 votes (up from 41) but as mentioned below this may not be reliable (it was in Frankston where the margin moved by only two).  Perhaps these votes have not been rechecked yet as the 2PP count now runs 152 ahead of the primary meaning that we do not know the final primaries yet.

11:00 pm Monday: I'm told that we should not expect wonders from the outstanding votes mentioned below.

3:00 pm Monday: Word that there are still about 200 provisionals and postals to throw.

2:00 pm Monday: Well it now seems today isn't the day - News Corp journo Liam Quinn has tweeted that kickoff for the preference distribution is tomorrow morning.  That said they should be able to get rid of the tiddlers reasonably quickly and we should at least know who is provisionally third before too long.  We might however see some crumbs later today in terms of the handful of remaining uncounted votes.

1:30 pm Monday: I'm back online but there has not been any sign at all of white smoke from the Vatican yet.

10:15 am Monday: Today's the day apparently, but it's quite likely just the day when we find out we are headed for a recount.  I'll be out til lunchtime or so but in the meantime here's Ben Raue's take on this complete with maps.  I've also heard the Greens are on for an extra 16 votes though if so the correction has yet to appear.

6:00 Saturday:A further complication that can still save Pharaoh aside from a recount is the possibility of significant error in the quick 2PP distribution (which currently shows Newton-Brown ahead by such a wafer-thin margin.)  Adrian Beaumont advised me of a case of a rather large such discrepancy (Brunswick in 2010, 193 votes).  Looking at the 27 (generally close) electorates for which such a discrepancy can be established (it seems that non-close seats are rechecked less thoroughly), the vote change in fact exceeded the current Prahran margin in 17 of those, and exceeded 100 votes in ten.

In seven of the 17 cases with a change exceeding 41, the change from the initial 2PP count to the final one favoured Labor, and in ten it favoured the non-Labor party.  The seat in 2010 in which Labor made the largest gain from the initial 2PP count to the count at the end of preferences was, amusingly, Prahran (344 votes).  

One might infer that there is less scope for errors in this seat given that it is bound to be one of the most scrutineised seats of the election, which would suggest errors could well have been caught earlier on.  But this is yet another uncertainty concerning the current state of play. 

8:00 Friday: And yet another closing of the 2PP count between Newton-Brown and Pharaoh, which would be astonishing enough now by itself even without the question of who is second and whether the Greens win from second.  The gap is now 41 votes with the addition of some more provisionals.  The Labor-Green primary gap has also closed, by six to 426.  (NB A poster on pollbludger, billie, says there are still 100 postals being held back for Monday.  This is a common method of preserving the anonymity of the final trickle of postal votes.)

6:30 Friday: William Bowe advises that the preference distribution for this seat will be done on Monday after the addition of any remaining straggling postals.  That should at least tell us how many candidates will be begging for a recount.  He also advises that there is nothing bar outstanding postals left for the primary count that day, meaning that Newton-Brown will provisionally win it unless errors in existing counting are found.

6:14 Friday: Another 17 postals added (or adjustments to existing counting) and Newton-Brown wins them 10-7 extending his lead to 60!  The few remaining primaries have been added as well and Labor now leads the Greens by 432, with the required rate for the Greens against Newton-Brown now 83.2%
 

5:36 Friday: Almost all the primaries showing in the 2PP count are in, with only 150 more absents showing in the 2PP count and, for some unknown reason, the prepoll primary count 48 over.  The Newton-Brown vs Pharaoh count has closed to a ridiculously tight 57 votes.  This could move as the 2PP allocation of votes are checked, and if they are the final 2PP candidates then they are likely headed for a recount.  Also still up for grabs is the Labor to Green gap, currently at 443 primaries.  There are 822 Animal Justice votes (which are believed to be worth about 1 vote to the Greens over Labor in every two) making this one impossibly close to call.  The required rate for the Greens against Newton-Brown, if it comes down to that, is now 83.4%.  (Pharaoh currently needs 82.7% and apparently isn't quite getting it.)

3:10 Friday: The great Pharaoh comeback continues; he is now a mere 83 votes behind on the 2CP.   178 postals, 175 prepolls and 133 absents seem to have been added.  However I'm not sure there is anything left but late postals.  If  Newton-Brown vs Pharaoh is the real contest then we could be headed for a recount.

Meanwhile more primaries have come through and the Greens have failed to make gains on these - they're still 553 behind Labor on primaries with about 882 postals, 885 absents, 402 prepolls and 399 provisionals (plus any votes not yet in the 2PP count) to go.  They may do very well on remaining absents.  The asking rate for the Greens if they are second is back up to 84%.

3:00 Friday: Updates this afternoon may be slow, have to go into town for a little while.

1:45 Friday: If the patterns on the already-counted non-booth votes continue when the full booth count is in, the Green percentage required against Newton-Brown will drop to around 79%.  That might be a trifle generous given remaining postals (which are not factored in) but if it does drop that far then it is seriously gettable - assuming, of course, that they are second!  We don't know how fair a sample these absents are as the early-2PP-counted absents were more Liberal than the later ones, which may mean the Greens' position could be better than this projection.

1:13 pm Friday: We now finally have meaningful primary figures!  With 3233 of 3937 known formal postals, 7843 of 8072 known formal prepolls, 1469 of 4661 known formal absents and 0 of 399 known formal provisionals included, Labor's lead over the Greens is 522 and the Green 2CP required is down to 83.5% (and will probably keep falling).  If these patterns hold up then I project the Greens closing to about 350 votes behind Labor and probably just passing them on AJP preferences but it's looking very tight.

1:06 pm Friday: The ABC site shows another 399 2PP votes added with Labor making still further gains and cutting the margin to 133.  The VEC site has yet to update so it is not clear whether these are the provisionals (edit, yes they are). Very close but I understand there's virtually nothing left to be added.

12:00 pm Friday: Movement at the station with some more absent votes counted and a big gain by Labor to gain 89 votes off 745 (a 56% flow).  I understand that's just about the end of the absents though, if so leaving Labor with 184 to catch off 500-odd provisionals and a very small trickle of postals.  In 2010 Newton-Brown won the 2PP for provisionals getting over 57% of them; Labor would be needing something like 2:1 in their favour this time around.

11:20 am Friday: The checked primary vote count continues to trickle up to where it was when it was taken offline for rechecking, but is still about 5000 votes short of that point.  No new 2PP votes have been added.  (My understanding is that they are actually thrown on a two-candidate basis quickly without primaries actually being counted, so it is not as if primaries for the late additions exist that are being suppressed.)

10:50 Thurs: Another few hundred absents added and Labor actually got a good rate (57%-ish 2PP) on this parcel to close to 275 behind, but don't expect this to happen consistently.  Some scrutineers are saying the Pharaoh-Hibbins flow is good.  I also understand that what is left includes still significant numbers of absents.  We don't yet know what the primary split on these is but the more of them, the better the Greens' chances.  Hence I have downgraded it from a likely Liberal retain back to "unclear".

5:10 pm Thurs: Most postals are now rechecked in Prahran but there is still more checking to go to get us to the point where we learn anything much new about the Labor vs Green primary race and the Green vs Liberal asking rate.

4:20 pm Thurs: A further gain in the 2PP for Newton-Brown vs Pharaoh with his lead now out to 322 votes, based on more absent votes of which 3798 are now counted.  As Antony notes the flow of out-of-electorate prepoll votes can vary a lot in this sort of electorate (depending on where they are from) but it would take spectacular variation on the relatively small number of votes left for Pharaoh to get over this gap.  Most likely they're a trickle from a wide range of different electorates.

Based on the 2010 turnout there might be about 2700 votes left, leaving Labor requiring a 56:44 split of the remainder - highly unlikely wherever they are from, though Labor did beat this rate in five booths within the electorate.  I also infer that overall turnout is actually slightly down and that the number left might be lower.  We still don't know anything more about primaries except that rechecking of the ordinary booth votes had very little net impact. 

12:30 pm Thurs: Further counting of absents has seen Newton-Brown perform extremely well on them with a remarkably good 54.2% 2PP share over Labor off 2700 votes (which is most of them).  There is not much left to throw now and Pharaoh's chances of pulling back 268 votes are negligible; Labor will just about certainly not win the seat.  We are still waiting on primaries to see how the Greens are travelling and whether they have any hope.  As often happens with a count of this kind the different variations in different kinds of votes appear to be cancelling out and coming back to where we started: that Newton-Brown is probably winning.

10:50 Thurs: While waiting for something to happen I should note that there are probably still about 1400 formal prepolls to go according to the calculations published by Poll Bludger, suggesting that the number of absents is only 3500-4000 (some of which are already counted, but we do not know the primary breakdown until they have been rechecked).  The smaller number of absents makes it touch and go as to whether a repeat of 2010 patterns would be enough to get Hibbins into second with AJP preferences - he would probably just make it, but the big doubt is whether those patterns will repeat.

5:20 Wed: Newton-Brown back in the 2PP lead over Pharaoh by 72 votes after another c. 2000 are added, including 1550 absents of which the 1471 formal votes broke 50.9% to Newton-Brown (compared to my 2010-based projection of 49.1%).  However, we don't know where these absent votes came from - because Prahran is so diverse, absents from different surrounding electorates may vary a lot in their behaviour. 

4:00 Wed: Counting of 1068 prepolls has increased Labor's 2PP lead by 38 votes to 52; these are out-of-electorate early votes which based on 2010 patterns should have favoured the Coalition slightly.  The all-important primary breakdown of these votes doesn't seem to have appeared anywhere yet. 

12:50 Wed: Clem Newton-Brown will nominate for the deputy leadership of the party despite his seat being unresolved.  The article says "The Liberals say around 8000 uncounted absentee and postal votes are likely to favour the incumbent." The postals most certainly will, strongly, but there are very few of them left, and the absents will slightly favour Labor (and also the Greens) against Newton-Brown if 2010 patterns hold up.  At the moment I think the incumbent's chances are about 50-50.

12:30 Wed: The rechecking process is causing a lot of confusion in this seat and others.  At various times during rechecking the primary vote totals shown on the VEC and ABC sites are only partial totals from those booths that have been rechecked.  This has caused a number of people to believe the Greens are now ahead but this is not true; Labor are still 645 (give or take a few) ahead; it is just that not all the primaries are currently showing.  Also it was believed 100 votes had been found for Hibbins during rechecking but this was a false alarm.

5:20 pm Tues: Still hasn't been any absent vote counting today, looks like that might not come til tomorrow afternoon or Thursday with more rechecking tomorrow.  Meanwhile scrutineering figures on the AJP preferences are suggesting Labor's real current lead on the Greens may be below 300 prior to absents. 

3:10 pm Tues: I think it might be useful to keep tabs on the target 2PP percentage for Hibbins if he makes the final two.  That rate is the rate of votes he would need to reach him ahead of Newton-Brown to win the seat.  It can be found using both candidates' primary vote percentages by the formula:

(50 - Hibbins)/(100 - Hibbins - Newton-Brown)

When I first calculated the get rate it was 84.2%.  It is still 84.2%.  (Pharaoh needs 83% on current primaries and we know from the 2PP that he is getting very slightly above that.)

11:10 am Tues: I have had some comments from a scrutineer suggesting that the Greens' flow from AJP is a bit better than my model.  The suggestion is that although they are currently 645 behind Pharoah on primaries, this is really only something in the low 300s after preferences.  The remaining prepolls are from out of the electorate, meaning that if absent voting favours the Greens relative to the booth vote as strongly as last time they are very capable of finishing second.

8:30 Mon: More postals have been added, cutting Pharaoh's lead to 14. As postals tend to favour Newton-Brown anyway, this doesn't tell us anything we don't already expect.

8:00 Mon: Given that my original projection had Pharaoh losing the 2PP battle with Newton-Brown by about 790, and given that his performance on the prepolls so far has beaten the projection by about 800, it would seem that if he can continue to perform as well on the remaining prepolls, of which there are about 2900.  On that basis Pharaoh currently seems more likely than not to tick both boxes, and although both seem currently not much better than line-ball, if he ticks one that makes it much more likely he ticks the others.  It is quite a mystery why the Greens' performance on the prepolls counted so far has been so bad given the huge effort put into the seat.  I thought Pharaoh had very little chance last night but the early votes have placed him in a fairly good position.

5:00 Mon: Pharoah has indeed hit the lead off the early votes, but his lead is 113 and not the 1000 being claimed on Twitter! A long way still to go with absent votes (which should help Labor) and remaining postals (which have been ultra-strong for Newton-Brown, though there won't be nearly as many of them as absents) still coming.  There are some tensions over the Greens' reluctance to concede; at this stage, I wouldn't either in their position.

4:22 Mon: Sands are shifting rapidly here with a claim on Twitter that the 2PP for Newton-Brown from these early votes is very unfavourable and that Labor will win!  Again waiting for the official update but the last claim from the same source was on the money.

4:15 Mon: The mystery "figures" posted here yesterday were actually quite close.  Over 7000 early votes have been added and they have broken 42.3% Liberal, 30.5% Labor, 22.8% Green.  On this basis my model moves Pharaoh to a projected 326 vote primary lead over the Greens (from a deficit of 520) so at this stage he is marginally favoured to be second (very little in it though.)  The 2PP count (Liberal/Labor) has not yet been updated.

2:45 Mon: Based on updates being posted by scrutineers it appears that the strange truth is that the scrutineering figures posted by an apparent imposter yesterday are on the money (for all I know they might even have been real figures!).  There is a report (@melbcity) that Pharaoh has increased his lead on Hibbins by 517 on prepolls. I'll wait until they are official figures before reconfiguring my model.

========================================================================
Original Article

This post follows (at about Wonk Factor 3/5 level) the post-count in the undecided seat of Prahran, the only seat where votes were counted today (Sunday).  The only seriously unclear lower house seat from the election (though there are half a dozen others not quite nailed down yet), Prahran is a fascinating three-cornered mess that truly deserves a detailed thread of its own.

I may write a postcount thread to follow other seats later but for the time being I point readers to Poll Bludger for the Legislative Assembly, Poll Bludger, Tally Room and geekLections for the Legislative Council and Michael McCarthy for why Brunswick is still interesting. (Any more?)

Prahran, a politically diverse seat with a large LGBTI population, is a three-way fight between the Liberal one-term incumbent Clem Newton-Brown, Labor's Neil Pharoah and the Greens' Sam Hibbins.  The seat was one of two heavily targeted by the Greens, the other being Melbourne, in which their margin dropped during checking, placing the seat in some doubt. Based on 2010 patterns (in which the Greens did much better on absent votes than votes cast on the day) I believe their margin will increase and they will still win it.

Newton-Brown is way ahead on primaries (currently with 45.83%) but Pharaoh (25.1%) and Hibbins (24.69%) are very close to each other for second.  Of the remaining candidates the Animal Justice Party has 2.29%, Family First have 0.65% and three others have 1.44% between them.  In vote terms, Pharaoh leads Hibbins by 88, with the AJP with 496 and the rest combined with 453.

Here is a table, various parts of which I will explain as I go along.  Anything with a question mark is my projection from 2010 data and available evidence re 2014 vote numbers only.  Note that my figures include projected formal votes only. (The VEC site has some very annoying habits when it comes to including informal votes in percentage tallies.)


(Note: A scrutineering sample claiming the real Early vote split to be Lib 41 ALP 33 Green 25 - on which figures Labor would come second - was posted here by an imposter claiming to be a specific scrutineer.  The real scrutineer states that no early votes have been opened yet.)

There are three questions that must be answered:

1. Who Will Make The Final Two, Pharaoh (ALP) or Hibbins (Green)?

The Animal Justice Party has preferenced the Greens on its how-to-vote card, and while few of its supporters will follow it to the letter, it would be expected that many of them will preference the Greens anyway.  So Pharaoh is likely to need a larger lead than he has now.  In the two federal Victorian seats the AJP contested in 2013 their preferences split 55:45 Green to Labor in Batman (but with at least 7% and probably somewhat more going to the Liberals first) and 73:27 (at least 9% through Libs) in Melbourne.  Without having personally seen sampling I'm thinking that the three way split of AJP preferences could be something like 60% Green, 30% Labor, 10% Liberal, meaning Pharaoh's primary lead needs to increase to at least 150, and possibly more like 200, for him to outlast the Greens.  (Update: in scrutineering at Prahran booth, where AJP cards were handed out, the AJP flow to the Greens was 80%.)

Based on my estimates in the table as published above,the Absent votes look like the key factor. A repeat of 2010 differences between the Labor and Greens primaries will improve the Greens' position by around 491 votes.  The remaining votes as projected on my table would then have stood to improve the Greens' position by another 111, with the Greens projected to lead by 520.  So at the moment it looks like the answer to this question is probably Hibbins, unless anyone has any real scrutineering figures otherwise.  However, it's possible that the patterns will be different enough this time to change that.

2. If Hibbins (Green) Is Excluded First, Who Wins?

At this stage the 2PP is 50.92% to Newton-Brown, a lead of 397.  Based on the above table I infer that at a similar stage (with booth votes and around the same number of postals) counted in 2010, Newton-Brown was on a 2PP of about 54.7, perhaps a shade higher accounting for early postals being more conservative than late ones.  This is, extremely fortunately, almost exactly the same 2PP as he finished with, meaning that I can assume it is the same and simply apply the differences from average for each vote type last time to the votes expected to be remaining.

The chart shows that I expect Absent votes to slightly favour Pharaoh compared to Newton-Brown.  But there are not that many of them and they are only projected to favour him by the sort of margin he is currently behind by, off four times as many votes.  The other categories all project to favour Newton-Brown.  Off my back-of-envelope estimates of the numbers of votes left in different categories, I actually project Newton-Brown expanding his margin over Pharaoh by a further c. 390 votes.  That's a fair amount and should be treated with caution - what's important is the evidence doesn't currently suggest Labor closes the gap here, let alone closes it by enough. So I think the answer to question 2 is very likely to be Newton-Brown (Lib).  I am not prepared to absolutely call that yet.

3. If Pharoah (Labor) Is Excluded First, Who Wins?

Unfortunately this is the bit we know the least about, and it may be crucial. We do not have a Liberal-vs-Green 2PP yet, only a Liberal-vs-Labor one which may not be directly useful.

If we look at the current 2PP count with Newton-Brown beating Pharaoh with 50.92% 2PP off primaries of 45.83% to 25.10%, it turns out that in that count Pharaoh gets an 82.5% preference flow from all other candidates combined (most of them Hibbins).

If we assume Hibbins beats Pharoah then Hibbins needs a very slightly better flow from Pharaoh than Pharaoh is currently getting from him.  On my current estimate of primaries, it comes out to 84.2%.  However, that can easily change - depending on how remaining primary votes go it might be that he needs, say, 90%, or 75%.  That's a point we have to keep track of as we find out more, if it looks like the Greens do come second.

I have seen some reports from Greens scrutineers who claim that Hibbins was getting virtually every ALP preference at their booths.  But the samples they report are too small to be reliable and apparently come from strong Green booths in a very diverse electorate.  I've also now seen one sample reported from a strongish Green booth (Prahran) that was not so unanimous. The Liberal primary vote in Prahran ranges hugely from 23% in Balaclava to 68% in Toorak West.

We know that the ALP how-to-vote card follow-rate in Victorian elections in city seats is about 44.5% (average of Melbourne, Brunswick and Richmond in 2010) so this percentage of the Labor flow can be allocated to Hibbins automatically.  It's harder to predict the 55.5% who make up their own minds.

Here we have some precedents from other state elections.  The one that comes from a compulsory-preferencing regime is the seat of Heysen at the SA election in 2014, at which all Labor preferences (including those following the how-to-vote card) went 80.25% to the Greens ahead of the Liberals.  In SA, how-to-vote cards are displayed in polling booths.

The optional-preferencing states provide examples of how voters distribute preferences if they decide not to follow the card when it says just-vote-1.  Examples include Balmain, NSW, 2011 (59% of Labor votes exhausted, but those that did not went 77% to the Greens) and Noosa, Qld, 2012 (63% exhausted and those that did not went 80% to the Greens).  There were 12 NSW 2011 cases where the Coalition won an outright majority with the Greens second, and there are not complete preference distributions online, but there is enough information to find the total Greens share of all preferences.  As these were seats won heavily by the Coalition in an election where Labor was on the nose, I haven't extracted them yet.

(My thanks to @sorceror43 on Twitter and Blake Andrew on Poll Bludger for info on the above.)

If Hibbins can get above 75% of the Labor votes that do not follow the card then on current primaries that would leave him with a gettable portion of the remainder.  But I'd infer (especially based on the SA example) that the preference splits from non-card-followers in a compulsory preferencing election are somewhat weaker than the splits from those who choose to give preferences under OPV.  I'd infer this both from the Heysen outcome, but also because those who strongly want to preference the Greens would follow the card, but those who strongly want to preference the Liberals must do so in their own order.  I am doubtful therefore that the flow of Labor preferences to Greens will be much above the 80% in Heysen.

On that basis I suspect that if Hibbins does make the final two he is likely to end up a few hundred votes behind, based on current primaries.  But there are a lot of uncertainties involved in the modelling estimates so the seat still appears very much open.

We may have a much better idea where this is going after counting on Monday and I will update the thread then.  It could turn into a classic or it might fizz out so quickly that I call it tomorrow.

Updates to follow in coming days.

16 comments:

  1. This comment has been removed by a blog administrator.

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. Comment attributed to "Luke Corcoran" was fake by an imposter and has been deleted.

      Delete
  2. Kevin surely Melbourne is still very much in play as well with only 300 vote difference and a very low count (52%)


    Peter N

    ReplyDelete
    Replies
    1. Oooh, I didn't know that! Looks like 111 votes have changed hands in post-checking cutting the margin by 222. That does make it interesting. Should still win if postcount trends hold from 2010 but maybe they won't.

      Delete
  3. Kevin in Melbourne the VEC website has a 50.74 /49.26 split at the close of counting but with only 52% counted seems way early for the Greens to be claiming definite victory today with the Libs primary vote likely to increase in pre polls and postals and mainly flow through to ALP

    ReplyDelete
    Replies
    1. Green vote will increase in the votes to come, so if they're winning from this early position then Labor+Liberal are very unlikely to regain enough ground to take back the seat.

      Delete
    2. This is also what I expect based on 2010 patterns. Again it's mainly the Absent votes that are the issue - they very strongly favoured the Greens compared to booth votes last time, at Labor's expense.

      Delete
  4. Based on Luke's sample I get Pharaoh needing something north of 75% of Green preferences, not necessarily over 80%.

    ReplyDelete
  5. Sadly (though perhaps not for the Greens) the data was fake and posted by an imposter.

    ReplyDelete
  6. Based on the difference between VEC and ABC, the second batch of postals seems to have broken wildly (62.7%) to CNB. Could that be?

    ReplyDelete
    Replies
    1. Yes. Discrepancy in postal campaign effort between CNB camp and ALP apparently.

      Delete
  7. Thanks very much for the detailed post Kevin, it's much appreciated

    ReplyDelete
  8. Kevin - how much of the Greens preferences would've had to leak away from Labor to prevent Labor winning?

    ReplyDelete
    Replies
    1. Around 14.3%. It would have been almost exactly the same requirement for Labor off the Greens as the other way round. The provisional 2PP throw suggested that just over that amount would have leaked, but it may be a while before we know the ALP/Liberal 2PP for sure now.

      Delete