This thread is finished now - go to button press day thread
Current live count leaders
Summary of contest:
Live count suggests a contest between Animal Justice and Coalition-7 for final seat with Coalition well ahead on primary votes. However, Coalition lead fell rapidly in very late counting. While the Coalition's lead in the initial count appeared sufficient, the initial count slightly overestimated it and it is not clear the Coalition's lead will survive the preference flow.
This page is classified Wonk Factor 5/5. Beware!
----------------------------------------------------------------------------------------------------------------------
Donations welcome!
If you find my coverage useful please consider donating to support the large amount of time I spend working on this site. Donations can be made by the Paypal button in the sidebar or email me via the address in my profile for my account details. Please only donate if you are sure you can afford to do so.
----------------------------------------------------------------------------------------------------------------------
Tuesday 9:35 The live count is finished and the Coalition continued to underperform the initial count even adjusted for BTLs. The Coalition on 6.551 quotas leads Animal Justice on .482, a lead of .069. At present left preference sources have .794 quotas after accounting for the Greens being just short of two, and right preference sources have .752 quotas. Around half the left preferences will exhaust. For the right this may be about two-thirds, but Shelton and Bosi group preferencing rates are unknown. Also, the Shelton group may include some ATL donkey votes which would flow to Animal Justice.
The advantage for Animal Justice is that these preferences are splitting between five parties, three right (Coalition, Shooters and Liberal Democrats) and two left (Legalise Cannabis and Animal Justice). As against that with the exception of One Nation votes being likely to be good for Shooters, it could be the preferences generally are better for Coalition than the other right parties.
The left preference sources are EFI .291 quotas, Sustainable Australia (not really a left party but for preference purposes behaves like one) .204 quotas, Public Education .165, Socialist Alliance .081 and ALP .055. The right sources are One Nation .303 Shelton .278 and Bosi .171. I have treated IMOP (.102) as unclassifiable though it might be AJP and LC do better off it. There is also a scattering of below the lines.
It is still not certain that Animal Justice can close the gap but they are definitely close enough if they are good enough, and they were good enough the last two elections.
6:45 Coalition has dropped back again slightly. now .085 Q ahead in the check count with about 96% counted. Also the seats to come are not great for them so it may well be they pull up significantly short of the projected lead from the initial count. PS One of the LDP's people has noticed that the LDP's ATL tally in the check count has already passed its ATL tally in the initial count.
6pm: 93.6% counted and the Coalition have almost erased the discrepancy, the gap now at .115 Q compared to .126 Q in my model. Labor continues to run below the model at this stage.
Monday 17th 5 pm: Zooming through the check count, with 89.3% of the initial count done. Now in the check count we have (model first, check count second):
Coalition 6.615/6.554
Green 2.003/2.004
LC 0.796/0.822
SFF 0.682/0.699
LDP 0.749/0.780
What is interesting here is that the LDP is now the party that is running most over the initial count. The Greens are no longer overperforming in the check count and LC and SFF far less than yesterday. The question is whether this is something about the remaining votes or whether it is possible that name confusion in counting could yet knock 0.03 or so of a quota off the Coalition which would give AJP more chance but still no certainty of winning. Another possible issue that could be affecting the majors is votes that are numbered both above and below the line but where the BTL part is informal. These would be counted as others (ie not as ATLs in the initial count) but would not appear as BTLs in the check count. They would be most likely to be cast for minor parties.
Sunday 16th: The below the line count continues to cause slight fluctuations in my model based off the initial count (it now has the Coalition lead at .124 again), but this is not going to change. The question is whether the initial count is completely accurate. The check count currently includes 76.5% of the ballot papers that are in the initial count and there are some parties with significant differences (initial-based model vs check count):
Coalition 6.600 vs 6.487
Shooters 0.681 vs 0.733
Legalise Cannabis 0.796 vs 0.852
Greens 2.002 vs 2.069
(and for the record AJP 0.475 vs 0.490 and LDP 0.750 vs 0.773)
The question is whether the check count, in which Animal Justice are now ahead, will as it should converge on the initial count as the final 23.5% is entered. It is possible to use the list of percentages counted by district to do a rough projection of whether parties are over or under counted, but I have not done this yet. (A catch in this process is that certain kinds of votes can be over or under counted too, especially post-count votes, and this can throw off such projections.)
Friday 8:20 It appears the initial count is finished as it hasn't moved for over a day. My model now sees only minor fluctuations in the Coalition's lead based on slight changes in the below the line voting rates (it has just crept up to .125, for what it's worth.) In the check count, the lead is currently still only .044 but the check count remains very incomplete.
7:45 88.1% counted on ABC site and even that small addition has knocked the Coalition's lead down to .124 in my model (6.599 to 0.475)
Wednesday 12:00 87.9% counted on ABC site and the Coalition's lead is now .133 (6.606 to .473).
Overnight Tuesday/Wednesday 87.2% counted on ABC site and the Coalition lead has fallen rapidly in my model and is now down to .140 (6.612 to .472). That's big progress for Animal Justice though it still looks like the lead will end up slightly larger than I think they would want. Note that a significant proportion of what is left to add will be BTLs that my model already caters for. Shooters on .679 Q are maybe not completely out of the woods here either though that should be enough (especially with a lot of One Nation preferences.) Also the available left vote pool has come up a bit, now to about .83 Q.
6:30 86.4% counted on ABC site, Coalition lead in model now down to .155 (6.626 to .471). This suggests the lead could finish somewhere around .13 to .14 which is probably too much but it's hard to be sure. Another issue we need to keep an eye on as the check count approaches completion is the possibility of mis-sorts. In the check count the Coalition is currently on 6.506 to ALP .479 - presumably differences to the model are caused by some divisions being much more counted than others, but if votes for any party were being tossed in the wrong (adjacent) pile during initial sorting, that will eventually show up as a difference.
10:30 84.9% counted on ABC site, Coalition lead in model now down to .169. I should also note my model does not account for below the line leakage, which might cost the Coalition as much as .001 quotas. I think the target for AJP is to get the lead below .1 Q but it's not easy to say exactly what the target should be when so many preferences come from the Farrelly ticket.
Tuesday 11th: Now there is finally some movement. The AJP is up to .472 Q in my model and the Coalition is at 6.648. The gap is now down to 0.176 quotas which is the closest it has been for a long time. The count is at 84.4% on the ABC site (compared to 88.0% for the lower house count which will be very close to finished). However the Labor primary has been dropping, with the Greens rising towards a second quota. At present the total of spare left preferences, minus the Greens' shortfall, is 0.804 quotas. I would expect slightly less than half of that to flow as preferences. At this stage AJP don't seem to be quite gaining fast enough.
Sunday: The AJP is starting to improve in my model, now at .467 Q but the Coalition is at 6.663 Q (and has at times been slightly higher). In the lower house count there were gains for the left in the last several percent of the count and in theory if this repeats in the upper house the Coalition-AJP gap could close. However so far there has been no signficant closing with the initial count now at 81.6% according to the ABC.
Wednesday: My model now has the Coalition strengthening further to 6.657 Q (AJP .463), mainly because the vote for BTL-only groups in the check count has been falling. I just want to describe at this point why it seems so hard for AJP to win without major shifts in the primary vote count. Left-friendly preferences would at this stage come mainly from Elizabeth Farrelly Independents, Sustainable Australia, Public Education, Labor and Socialist Alliance, but most of the Labor votes especially won't preference anyone and overall AJP will be lucky if there is much over .4 of a quota in left preferences. Some of these will exhaust but the others will be splitting to AJP, Legalise Cannabis and currently a small amount to get the Greens over quota. It is very hard to see where AJP can gain .2 of a quota from, especially when there are also some right-wing preferences (maybe .2 of a quota worth).
Saturday: Nothing to see in the last few days, model currently has Coalition 6.645 to AJP 0.460. From here on I will post updates rarely and probably only if there is real change.
Thursday midday: I am now using the BTL percentages from the check count for all tickets - notably that for the Greens is significantly down so far on last time. My model now has Coalition up to 6.646 quotas in the initial count which is now at 62.1%. AJP haven't moved.
Wednesday night: With a more realistic sample of BTLs I have started to use the BTL percentages for the tickets that did not run last time; these are high so far in the case of Group A (Shelton), Elizabeth Farrelly Independents and Group U (Bosi). Currently in my model I have the Coalition on 6.618 quotas and Animal Justice 0.462 compared to the ABC's 6.70 vs 0.46. At some stage I may try to estimate what Animal Justice could reasonably expect to gain on Coalition off available preferences, but at this stage it is nothing much because the Labor and Green surpluses have all but disappeared.
Wednesday: The check count commences! From the first 3378 formal votes the BTL-only candidates have a combined nine votes between them (0.27%), which would damage the Coalition by about .015 quotas compared to estimates that ignore these candidates. The Coalition is still strengthening and is on 6.70 quotas, effectively about 6.64 quotas after what I will now be calling double BTL adjustment. As the check count becomes more advanced I will refine previous BTL adjustments for the parties, but too early for that just yet.
Tuesday late night: The Coalition continues to improve, now to 6.67 quotas on the ABC, which I make as 6.59 after BTL adjustment (the Animal Justice figure hardly changes). Hopefully at some stage we will get an insight into the BTL vote for candidates without ATL boxes.
Tuesday 2 pm: The sharp swing in the postcount for the lower house that happened yesterday has had a predictable ripple upstairs. The Coalition ticket has shot up to 6.60 quotas on the ABC tally, which I make as 6.51 after below-the-line adjustment, leading Animal Justice on 0.47. I put One Nation on 1.30 and still well off the pace; other movements are minor. If the button was pressed right away Animal Justice might well still beat the Coalition on preferences off these numbers, but they would not want the Coalition to keep improving. 45.9% counted.
Intro (Sunday afternoon)
Welcome to this year's thread about the NSW Legislative Council postcount; I hope it won't be as long as 2019's. We are still at an extremely early stage and while it may look like there is only a fight between Animal Justice and the Coalition for the final seat, there is a lot that can still change.
This year things are slightly simpler because above the line votes for all 15 groups with above the line boxes are being counted, removing the need to compare between counted and uncounted groups, which was a massive complication in 2019. What we have at the moment is a count of above the line votes for these groups, plus counts for blank informal ballots and "others". Others includes non-blank informals, but also includes below-the-line votes.
This is the first trick to be aware of because past experience is that some parties have higher below the line rates than others, and hence will be underrepresented in the current count. I deal with this by applying a multiplier to each party's current total. The current multipliers I have (from 2019) are:
Greens 1.040
Animal Justice 1.036
Labor 1.019
Coalition 1.013
Liberal Democrats 1.034
One Nation 1.065
Socialist Alliance 1.097
Sustainable Australia 1.044
The rest are new. For now I've put Legalise Cannabis on 1.030 and everyone else on 1.040; it is unlikely to matter.
These multipliers are important because the major parties have probably not received as many votes as the ABC count implies, while One Nation has probably received more. So while the ABC has the Coalition on 6.43 quotas and One Nation on 1.23, after adjusting for BTLs that is more likely to be something like 6.36 and 1.28 respectively, ie the Coalition is not that far ahead of One Nation.
I currently have the following quota estimates for the live count:
Labor 8.10 quotas, Coalition 6.36, Greens 2.24, One Nation 1.28, Legalise Cannabis 0.87, Liberal Democrats 0.74, Shooters 0.68, Animal Justice 0.47, various others 1.26 (A quota is 1/22nd, ie 4.545%)
I am ignoring here a scatter of below the line votes that will be for the six BTL-only tickets and also for ungrouped candidates, though I may adjust for these if evidence appears that they are significant. At the moment I suspect they total less than 1% between them.
Currently Labor would win eight seats on raw quotas, Coalition six, Greens 2 and One Nation 1, leaving four to be filled by remainders and preferences. These would presently go to Legalise Cannabis, Liberal Democrats, Shooters and most likely Animal Justice, with the Coalition closest to beating AJP. Labor would dearly like Animal Justice to win the seat as this would create a left majority of Labor, Greens, AJP and Legalise Cannabis and mean they only had to work across the aisle with Shooters and so on when one of the others got picky. If the left wins only 11 seats then that's a deadlocked upper house if one regards the Shooters as right. The Shooters and Labor are often chummy but getting the Shooters to see eye to eye with their tribal adversary Animal Justice in a case where you need both can be difficult. However, the deadlock could also be resolved by making someone from the "right" side President.
The live count is not very advanced though. At this stage only 36.5% of enrolment has been counted. The current count is heavily composed of booth votes, and when more postcount votes are added it is likely to get better for the Coalition, and all else being equal would make them much more competitive for a seventh seat (if their vote goes up by 1%, that's .22 of a quota). The count is also far more advanced in some divisions than others. At this stage I don't have a time-efficient way of processing the individual count pages into a single spreadsheet to model off that. [EDIT: I have just been sent one so more insights coming soon.]
In 2019 as of Tuesday the Liberal Democrats were on 0.65 quotas in projection but this soon crashed to below 0.6 and they ended the count with only 0.48 and lost. This may have been connected to the early inclusion of iVote and Sydney Town Hall votes (iVote is not being used in this election) but there were also general patterns involving slow counting in seats outside the inner city and some specific complications involving the way votes were thrown in 2019. So a question here is whether there is any risk of the Liberal Democrats coming down from 0.74 quotas (at a significantly earlier stage of the count) to something where Animal Justice and the Coalition (or even someone else) both overtake them. There's reason to suspect the crunch effect might not be so bad this time so it will be interesting to see if they remain competitive. (The Lib Dems had a less than ideal leadup with a massive faction fight including a failed attempt to remove lead candidate John Ruddick.)
During this week I may have time to look at this count in more detail, especially later this week. I also welcome comments from other modellers and links to data scraping of the seat by seat results. Once again, it disappoints me that the NSWEC does not make presenting data in a fashion that can be readily used without IT expertise a priority.
UPDATE: Thanks to a very helpful colleague (come for the spreadsheet, stay for the beautiful colour maps further down) I have the votes processed thus far by division and I have run a simple average on them to compare with the ABC's tallies (this is not accounting for different numbers of formal votes likely to arrive in different seats, but that is extremely unlikely to matter). I find the following average percentages per electorate out of all the ATL votes that have been counted so far, with the ABC's total percentages in brackets:
Labor 37.4 (37.1), Coalition 28.8 (29.2), Greens 9.7 (10.0). One Nation 5.7 (5.6), Legalise Cannabis 4.0 (4.0), Liberal Democrats 3.3 (3.3), Shooters 3.2 (3.1), Animal Justice 2.2 (2.1).
This suggests that ignoring the issue of skew by vote type, the count at the moment is a balanced sample on a geographic basis, so a very large amount has to change based on vote type to unsettle the large leads of the first 20 apparent seats over 22nd position.
If there is a contest between Animal Justice and Coalition for the final seat then AJP currently look like having a juicy supply of Green preferences while Coalition may not have too many friends in this count. So I think the Coalition ticket would need to be well ahead.
LNP now up to 6.7 quotas at 55% counted. Do we know whether the 55% is representative or the remaining? Unless you know the remaining will favour AJP then the LNP should be favourite for the 21st seat now. Any chance that SFF is at risk and AJP and LNP overtake them for the 20th and 21st seats on preferences?
ReplyDeleteDon't know if it is representative but suspect it is given the trends in prepoll in the lower house. I don't think AJP are catching anyone else from c. 1% behind on preferences but this may be worth trying some simulations on later. Shooters will benefit to some degree from what One Nation preferences there are.
Delete