Sunday, April 13
This is probably my final update, since the result is well and truly beyond doubt. On the raw votes, the ABC calculator produces a result at the final count of 194,282 (14.86%) to Linda Reynolds and 179,150 (13.71%) to Louise Pratt, and my own projection is hardly different (14.91% to 13.66%). As Antony Green points out on Twitter, Labor below-the-line votes are producing the very unusual result of the second candidate, Pratt, outpolling the first, Joe Bullock, the current numbers being 1285 to 1039 with a great many more still to be apportioned, although it seems unlikely Pratt’s lead will be overturned. A precedent for this noted by GhostWhoVotes is that Barnaby Joyce outpolled George Brandis as the respective second and first candidates of the Liberal National Party Senate ticket in Queensland in 2010, the circumstance here being that Nationals loyalists who opposed to the LNP merger expressed their displeasure below the line.
Friday morning
Antony Green and Kevin Bonham are both calling it for Linda Reynolds, and I’m not going to argue. Yesterday saw the addition of another 13,530 postals and 2034 absent votes from Brand (on top of the 1653 that had been counted there already, these being the only absent, pre-polls or provisional votes entered into the count so far), together with more rechecking. My projection now has Reynolds’ lead over Louise Pratt at the final count at 190,430 (14.57%) to 183,002 (14.00%), or 7428 votes, which is lower than yesterday because of some tinkering I’ve done with the model. Putting the raw vote into the ABC calculator, Reynolds now leads 189,988 (14.54%) to Pratt’s 183,443 (14.04%), increasing the margin to 6545 from 3407 yesterday. The postal results have been consistent with the contention that the Joe Bullock story breaking the day before the election caused a shift in support from Labor to the Greens, Labor’s postal vote (24.64%) being higher than its ordinary vote (21.83%), while the Greens are much, much lower (6.98% compared with 15.78%).
Thursday morning
The addition of 11,138 out of what should be at least 90,000 postal votes has blown a hole in Labor’s hope that votes cast earlier in the piece will be relatively favourable for them, making a Louise Pratt victory look increasingly unlikely. With numbers reported from Brand, Curtin, Durack, Hasluck and Perth, the results respectively show the Liberal vote 11.1%, 11.1%, 10.3%, 13.4% and 9.6% higher than the ordinary vote, equalling or exceeding the similarly large differentials in September. Putting the raw votes into the ABC calculator previously showed Pratt in the lead, but now Linda Reynolds holds a lead of 3407 votes (0.26), or 188,421 (14.42%) to 185,014 (14.16%).
On the model I’m using to fill the gaps in the count, Reynolds finishes 8499 (0.65%) clear with a lead of 190,963 (14.61%) to 182,474 (13.96%). For pre-polls, postals and provisionals, the model assumes parties’ vote shares will differ from ordinary votes to the same extent that they did in September, producing percentage figures which are applied to estimated totals based on declaration vote data published by the AEC (1653 absent votes were added today from Brand, but as absent votes tends to bounce around depending on where they were cast, I will continue using the aforesaid method until a large number of votes are in). For postals, the party vote shares recorded so far for each of the five electorates for which votes have been counted are extrapolated to an estimated total, likewise based on the AEC data. For electorates where results have not yet been reported, the method is the same as for pre-polls, postals and provisionals.
The Liberal margin will come down by perhaps around 3000 if Palmer United’s position improves to the extent that it doesn’t need HEMP preferences to get elected, in which case HEMP votes will be passed on to Labor at their full value rather than a much-reduced transfer value. However, the improvement in PUP’s position needed for that to happen is a not insubstantial 0.3% going on the modelled figures.
Wednesday morning
I’m not going to do serious number crunching until we start seeing pre-polls, absents and postals, but the Liberals gained at least 1500 votes on yesterday’s re-checking and the addition of special hospital results as such, Kevin Bonham putting their lead at 2504 based on the current numbers. Kevin also observes that Labor’s position might improve by “thousands of votes” depending on the arbitrary fact of whether Palmer United reaches a quota after Liberal Democrats preferences are distributed, or whether the job still needs to be finished with the subsequent exclusion of Help End Marijuana Prohibition. In the latter case, HEMP will go into the mix of votes to be distributed as the Palmer United surplus at a fraction of their value. Otherwise, their preferences will transfer at full value to their next party of preference, namely Labor. However, the odds are in favour of the Liberals on either scenario.
Tuesday morning
Rechecking and perhaps a few delayed booth results yesterday added 2161 votes in Durack, 1076 in Forrest and 152 in Hasluck, to the extremely slight advantage of Labor. The West Australian reports counting of postal votes “may get under way today”.
Monday morning
Nothing new on the counting front yesterday, which the AEC presumably devoted to very carefully transporting votes to the divisional offices where the primary vote totals will be rechecked over the coming days. Ben Raue at The Tally Room observes that “the numbers of absent, provisional and pre-poll votes have dropped to 20-33% of the 2013 levels, while the number of postal votes has increased” – which would seem to bode ill for Labor, given how heavily postal votes traditionally favour the Liberals (47.6% in September compared with 38.8% on ordinary votes).
Sunday morning
For those of you who have just joined us, the WA Senate election result looks to be two seats for the Liberals, one each for Labor, the Greens and Palmer United, and with the last seat a tussle between the third Liberal, Linda Reynolds, and number two on the Labor ticket, Senator Louise Pratt. Both major parties were well down on the primary vote to make way for a surge to the Greens and Palmer United. Scott Ludlam was handsomely re-elected off a quota in his own right, while Palmer United’s Zhenya Wang will get there with preferences from a range of sources, the most handy of which are HEMP, Shooters & Fishers and Family First. The following quick summary of the results shows the raw percentages, and how I’m projecting them to look after pre-polls, absents and postals are added. There follows projections of the race for last place as derived by plugging both raw and projected results into Antony Green’s Senate election calculator.
As I write, 38 booths out of 814 are still to report results. The only electorate where all booths have reported is Moore, where 69,323 ordinary vote have been cast compared with 72,507. This makes turnout difficult to calculate, but it seems to me to have not been as bad as some were suggesting. The number of ordinary vote cast in Moore amounts to 70.14% of enrolled voters, compared with 74.59% at the election last September. In Brand and Fremantle, which in each case have had all booths report but one, the totals are 70.6% and 69.8%, compared with 77.7% and 75.1% at the election.
Saturday
11.39pm. Back from my ice cream break to find the count at 937,396 (63.3%), with 62 out of 814 booths still to report. The latest projection puts the Liberals on 33.8% and the Nationals on 3.2%, Labor on 21.3%, the Greens on 16.0% and PUP on 12.2%. On the ABC computer, third Liberal Linda Reynolds’ lead over second Labor Louse Pratt at the final count has narrowed to 14.84% to 13.73%.
10.05pm. Count up to 661,954 (44.7%). My statewide projections are the same as Antony Green’s, so I’ll drop the metropolitan model and work off those instead from now on. I’m projecting 39.2% for Liberal, 3.4% for the Nationals, 21.1% for Labor, 16.1% for the Greens and 12.2% for Palmer United. Plugging that into the ABC calculator has third Liberal Linda Reynolds beating second Labor Louise Pratt at the last exclusion 15.1% (1.0553 quotas) to 13.49% (0.9446 quotas). Kevin Bonham and Truth Seeker think Labor are doing a little better than that: I’ve no idea about their methods, but I suspect it’s because they’re going off the raw vote totals, whereas I’m going off swings.
9.36pm. Count up to 526,235 (35.6%), Liberal projection down a shade to 2.93, Labor up to 1.57. But Labor’s position in the race for the final seat hasn’t improved since my 9.00pm update, because the Greens vote has come down slightly and reduced the size of the surplus available to Labor.
9.24pm. To explain all that in vote terms, the Greens vote is variously projected at 17% or 18%; Labor’s at a bit below 21%; Liberal at 34.5% plus Nationals at 3-4%; PUP at around 12%.
9.15pm. I have two models on the go here: the one I’ve been quoting, which extrapolates metropolitan swings across the rest of the state, and one which looks at the swings of all electorates, the problem with which is that non-metropolitan electorates should improve for Labor later in the night as bigger booths from regional cities report. But with the count now up to 367,945 (24.9%), the difference between the two seats of figures is diminishing – apart from the Greens, who are on 1.24 quota in the statewide model and 1.34 in the metropolitan-only model, and PUP are a bit higher in the former (1.18) than the latter (1.12). But both pretty much have the Coalition about 0.03 short of a third quota, and Labor on about 1.55.
9.00pm. With the same caveats applied in my 8.43pm comment, I’m now having Labor narrow the gap a little: Liberal 2.94 quotas, Labor 1.55, Greens 1.36, PUP 1.13. With the Greens surplus pretty much all going to Labor and PUP pretty much all going to Labor, the score at the final count would have Liberal winning 1.07 to 0.91, but with the numbers still certain to keep shifting around as the count progresses, and perhaps still the outside chance of both losing out to a micro-party boilover.
8.55pm. Antony observes current numbers in fact find that final vote going to Voluntary Euthanasia, but the statistical chance of that sticking would be low. Nonetheless, it should be emphasised that the final seat which I’ve been representing as a race between third Liberal and second Labor could be less predictable than that.
8.43pm. The picture isn’t getting any better for Labor as the count moves up to 121,082 (8.2%). My present projection based on metropolitan area swings has the Liberals on 2.96 quotas, Labor on 1.51, Greens on 1.36 and PUP on 1.14. That would easily get the Liberals to a third seat when the PUP surplus was distributed. Still plenty of room for caution though: the swing may be quite different outside Perth, and the swings I am calculating are derived not from booth-matching, but by extrapolating from the current electorate totals from metropolitan seats with their results from last September.
8.33pm. “Most of my modelling is based on the Perth vote”, suggets Antony, indicating my belated idea to run off the metropolitan swings gels with what he’s doing. With over 5% counted, very big transfer from Labor to Greens looking sticky.
8.23pm. Count up to 47,611, or 3.2%. Metropolitan swing projections: Coalition down 7.1%, Labor down 5.7%, Greens up 8.6%, Palmer United up 6.3%. Applying metropolitan swings to 2013 statewide results is the best rough guide I can come up with, because metropolitan booths do not have the issue with regional ones that a relationship exists between their size and their partisan tendency (i.e. these booths that are reporting early from O’Connor, Durack and Forrest and very conservative rural booths). Doing so confirms the picture noted previously, with a very close race between third Liberal and second Labor for the last seat.
8.11pm. Sam Dastyari concurring with my assessment that it’s likely Liberal 2, Labor 1, Greens 1, PUP 1, with the last seat a battle between a third Liberal and a second Labor.
8.08pm. Antony Green projecting a perilously low Labor vote, but the data available to him isn’t as good as usual and there’s still on 2.2% counted. My crude early projections for the metropolitan area are a 5.4% swing against Labor, 7.0% swing against Liberal, 6.2% towards Palmer, 9.3% towards Greens.
7.57pm. My early indications are of a 7.0% Palmer United swing in the metropolitan area, and all on the ABC News 24 are talking of a Scott Ludlam win as an accomplished fact. So you might start punting on a 2-2-1-1 result, unless Labor ends up doing badly enough that it comes in at Liberal 3, Labor 1, Greens 1, PUP 1.
7.34pm. With the count up to 5718, my PUP swing projection is now at 6.7%, which is a winning score for them. I’ll be interested to see what Antony’s next projection for them says. The lower micro-party vote is making a HEMP win look unlikely.
7.26pm. Antony Green’s data-matching off the earliest fraction of the vote – which is still a lot cruder than what he’s usually able to do – concurs with a drop in the micro-party vote.
7.18pm. Count now up to 2459. We’re at least getting evidence of a lower micro-party vote: I’m crudely projecting solid drops for parties such as the Liberal Democrats, Australian Christians and Fishing & Lifestyle.
7.11pm. To illustrate that point, an increase in the vote count to 1586 has been enough to push my PUP swing projection up to 4.0%.
7.07pm. Vote count up to 1216. The least useless of my projection figures based on the available data is the Palmer United swing, which I have at a less-than-expected 2.9%. Still pretty useless though.
7.03pm. Keep in mind that big unwieldy Senate ballot papers are slower to count than than lower house papers, so it’s to be expected progress will be slower than we’re used to.
6.56pm. Five small booths in from O’Connor, which would not even be representative of that electorate never mind the rest of the state, since they offer no insight on the larger towns. Also a booth from Pearce, for a grand total of 355 votes counted. Much talk from political operatives about a drop in turnout of about 15%, putting it in the high seventies.
6pm. Polls have closed in Western Australia’s Senate election. Absent any media commitments, I’ll be closely following the results as they come in on this post. I’m still unclear as to whether the AEC will be publishing booth results, but at the very least will be able to analyse the figures based on crude matching of reported results at the division level to the 2013 figures. Antony Green will be covering the results on ABC News 24, but I’m not exactly clear what format that will take.
Antony GREEN@999
I wondered if it might be something like that. Thanks.
Assuming the re-entries are done now, Labor’s position has still markedly worsened tonight.
an insiders tale from Ludlam’s campaign
[It’s four o’clock on election day, and Scott Ludlam is dead to the world. He’s due on the ABC at 6pm and the results will start taking shape an hour after that, but until then he’s asleep on the office couch, which is about thirty centimetres shorter than it needs to be. Jess McColl, the adviser on loan from Rachel Siewert’s office for the campaign, is holding the fort — fielding calls from the occasional nutbag, calming panicky booth captains, trying to put out any fires before they start. She’s the one who has set up and overseen the massive operation that’s driven The WA Greens’ re-election campaign, at times doing much of it virtually by herself. Ludlam loves her — he drops her name constantly in radio interviews.]
http://junkee.com/headliners/how-scott-ludlam-won-the-west/
William, the impact of Palmer United being elected before HEMP is distributed is much smaller than keeps being suggested. It only effects about 2,000 ballot papers, around 1300 Australian Democrat votes and 700 Katter’s Australian Party votes. If Palmer United is elected with HEMP preferences, these ballot papers go to Labor worth about 200 votes. If Palmer is elected before HEMP preference are distributed, the ballot papers are worth around 2,000 votes instead. The effect was wiped out by yesterday’s counting.
Antony GREEN@1004
These votes don’t just go onto the Labor total though; they also go off the Liberal total because the Liberals miss out on the benefit of having the full value of those votes largely replaced by PUP-Liberal votes.
So the impact on the bottom line margin is actually double the number of votes Labor gets. If Labor gains 1800 then their bottom-line margin improves by 3600.
On the other hand the AMEP votes cut to a small degree the other way; if PUP is elected on LDP without significant surplus they are worth nothing to Labor, but if PUP crosses on HEMP then Labor gets c. 240 votes out of them.
So an impact on the ultimate margin of say 3000 votes (based on 1500 more votes going to Labor) is quite plausible. Hence the c. 3300 figure I found initially. The confusion is caused by some people talking about how many votes Labor gains and others talking about the impact on the Lib/ALP margin at the end.
On that basis even with the c.1600 deficit in the mainline calculator version after last night, Labor should still be 1500-ish in front if PUP crosses on LDP. But in my simulations last night where I ramped the PUP vote up I wasn’t getting it as that much and thus only had Labor with a slim and arguable lead. I think rounding issues in the ABC manual entry calculator are contributing to that, though there may be other factors caused by the flow on impacts of increasing PUP. (I’ll have to set up my own!)
At the 69.51% mark (1734 Brand absents added) I have Labor ahead by 790 if they cross tipping point, but this is not indicative as the only non-ordinaries added so far are absents from a friendly electorate. So while Labor are notionally just in front (and may become more so if more absents go in) I still expect them to probably be blown away later on. My assumption for the tipping point was to give PUP enough extra primary votes to reach quota, hold all their feeders constant, and reduce everyone else proportionally.
This compares to a margin of 1295 the other way on the calculator.
Whether the impact is 1800 or 3600, the point I’m making is it’s nothing like the 75,000 votes distributed at that point.
This count is slooooooooow.
Fixed a small issue with my calculations arising from the two ungrouped independents. I currently have Labor up by 492 if the tipping point helps them; I doubt this will last through the postcount.
First postal vote counting is good news for the Liberal Party.
Lib postals up 10-13 points on Lib booth votes in five different and disparate electorates. I now have Libs up by over 1000 even if the tipping point is triggered – which the way this is going it won’t be, BTLs notwithstanding.
The Liberals must be in an extremely strong position based on these postals and the likelihood that that sort of pattern will continue. I can’t see what hope there is for Labor after seeing this.
That’ll be the Joe Bullock factor – apparently.
Last time the postal and prepoll went strongly to the right whilst provisional and absentee was more even.
Absentee is the smallest of these and probably won’t change much.
Provisional is when you aren’t on the roll so I would expect these to be down if many of the problems from 6 months ago are resolved.
Whilst I don’t think it is going to happen, there would be a funny side if the Greens got 17% of the ordinary vote (say 75% of the total) and only got 8% of the remaining 25% and they came up with 14% and fell just short of a quota.
If
Never understood why an appraisal of the ‘Joe Bullock effect’ was seen to depend on whether postal votes were more favourable to him or not. It didn’t take the publicity over ‘that speech’ for voters to realise he was a currant – they already knew that.
When postals are just as bad just watch all the dinosaurs grab that as proof he was really a great candidate, it was the (mining tax) (carbon tax) (Shorten) (blah blah) anything but that nice Ol’ Joe…..
At the moment the Green vote is running much worse in postals compared to ordinary votes than in the same electorates in 2013. If that lasts, that might be the Bullock factor – it might show voters (a few percent) moved from Labor to Green late in the campaign. After all if someone is concerned about Bullock they’re most likely left-wing and they are hardly going to shift to the Coalition over it.
However maybe this pattern will change and there could well be other explanations in terms of differences in postal effort level.
I’ve called this now and so has Antony.
hairy (should that be Harry?) @1014
Disagree.
Unlike the people who visit this blog – the average punter (even most of the rusted on party voters) pay liitle or no attention to elections, let alone the WA Senate rerun until the last couple of days and without that last birst of sunlight on Bullock they would have been none the wiser of who he is, why he is #1 for the ALP and what is views are – most of them don’t know who the SDA/Shoppies are. A number of regular posters here from the left of politics declared their lack of understanding of how the unions divvie up the preselections through labrynth, long-running deals.
KB @1016
HUZZAH!
Freemantle hasnt been updated since Monday night, all other divisions last updated today.
Is it unusual for a division to be 4 days behind in the counting ?
(imagine if they had lost some ballots)
What an absolute shocker of a result for the ALP. Largely self-inflicted and might I say RICHLY deserved.
There’s no way, NO WAY, a supposedly progressive, modernist, moderate party can put a homophobic dick like Joe Bullock at the top of any ticket and not pay the price. I’m not surprised at all Ludlum got in, I would have voted for him in a heartbeat over whatever detritus the SDA sees fit for pasture. And boy, have I held my nose for Labor sometimes – this is not a Green or a Trotskyist saying this.
A missed opportunity, a profound disappointment, a real kick in the balls. Not to mention a gigantic gift to the reactionaries in every sense of the phrase. Ugh. They have good reason to celebrate this result and I don’t begrudge them that one bit.
There’s nothing sweeter to behold than an opponent’s own goal.
[absolutetwaddle
Posted Sunday, April 13, 2014 at 6:23 am | PERMALINK
What an absolute shocker of a result for the ALP. Largely self-inflicted and might I say RICHLY deserved.]
Can I suggest that the reason Bullock put himself there was he thought he would get away with it.
The exact reason why is in your very post:
[ And boy, have I held my nose for Labor sometimes – this is not a Green or a Trotskyist saying this.]
Get it?
Any expectations for when the button will be pressed?
[There’s been plenty of talk on the system of preferential voting in the wake of the Western Australian election rerun. But mathematics masters student Casey Briggs says we should focus on the method of counting itself.]
http://www.crikey.com.au/2014/04/14/counting-votes-the-wright-way-what-the-aec-should-be-looking-at/
What can I say… I highly agree, we need to review the counting rules and the way the vote is counted.
T^he system that is in place was designed to facilitate a manual count. It is a trade off of bad practice short cuts designed to reduce the time taken to count the vote not to reflect the voters intentions.
With computerized counting we need to strip the system back to its basics and apply a reiterative count where the transfer surplus is weighted to the value of the vote and any surplus distributed as a single transfer without segmentation.
Likewise we need to do the same with exclusion. But so as to ensure that a full value vote originally for a minor candidate is always distributed to the next continuing candidate the best way to achieve this and to reflect the voters intention is to reset and restart the count afresh as if the excluded candidate had not stood.
A vote for an excluded minor candidate is redistributed so that the next available preference becomes their first preference vote and if need be forms part of a surplus of an elected candidate without skipping any candidate that had not been excluded. Only surpluses are distributed in any given iteration.
In the past the time taken to apply a reiterative count was prohibitive. with the use of a computerised counting system it is desirable and should be implemented, scraping the various tweaks, loops skips hops and jumps that the have been introduced to facilitate manual counting. Methods that distort the proportionality and simplicity of a Proportional Representation STV count.
No segmentation, a single transfer of surplus votes n any one iteration.
In addition to a reiterative count we should also consider scraping the Droop Quota (X/(y+1)) and revert to a pure proportional system (x/y) with the recommended Wright system of counting the vote
I did have the opportunity to day to outline to teh JSCEM committee this issue
If I can suggest that Crikey publish the counting flow chart at outlines the Wright counting process along side the above article as it may help its readers understand the process better.
Thanks for the review