2025 TASMANIA SENATE
CALLED ELECTED Carol Brown (ALP #1), Richard Dowling (ALP #2), Claire Chandler (LIB #1), Nick McKim (Green #1)
CONTEST Richard Colbeck (Liberal #2, incumbent) vs Bailey Falls (ALP #3) vs Jacqui Lambie (JLN #1) for two seats.
Lee Hanson (One Nation) has polled well but does not appear to be in contention.
Projection of current live count after preferences has Lambie ahead of Colbeck ahead of Falls - however margins in projection are fairly close and this assumes preferences flow the same way as in 2022 (they may not!)
One Nation preference flow is likely to be crucial to result
WARNING: Projecting the Tasmanian Senate count is very complex. This article is rated Wonk Factor 4/5
------------------------------------------------------------------------------------------------------
It's a great pleasure to have to write an article on an actual inter-party contest for the Tasmanian Senate race! This year isn't quite as messy as 2016, when the complexities of high below the line votes for Lisa Singh and Richard Colbeck in a double dissolution caused me to do 15 hours of scrutineering and write 8440 words across two main articles about modelling the outcome. But nonetheless it's interesting, for shock value if nothing else.
The Liberal Party has had a disastrous result in the Tasmanian House of Reps contest, losing Braddon with a 15% 2PP swing (the "safest" seat ever won by a first-term government from an opposition), Bass with a near 10% swing and also copping a swing just over 10% in Lyons. Why this happened is outside the scope of this article and the detectives are already out there sifting the remains. The damage has flowed through to the Senate where the party's live result is now around 12% below its final 2022 result in both Bass and Braddon, nearly 8% below in Clark and 9% below in Franklin and Lyons. The scale of the wipeout is such that, completely unexpectedly to me, Liberal Senator Richard Colbeck is in danger of losing his seat again. Jacqui Lambie is also in some danger, and the danger to them both is Labor. It may take a long time to be all that confident who has missed out in this three way contest and I may not know with confidence until the mythical Button is pressed in about four weeks time.
As I start this article the count is at 59.82% of enrolment; this figure will eventually reach about 90%. Labor has 2.4752 quotas (a quota is c. 14.29%) in the live count, Liberals 1.5558, Greens 1.2196, Jacqui Lambie Network .4759, One Nation .3667, Legalise Cannabis .2589, Trumpet of Patriots .2272, Shooters Fishers and Farmers .1633, Animal Justice 0.0938, Sustainable Australia 0.0848, Libertarians 0.0343, Aus Citizens 0.0258 and Fenella Edwards 0.0190.
Lee Hanson (One Nation) has come reasonably close to catching Jacqui Lambie but with a deficit of over 1.3% I am confident that she will not overtake Lambie. On current primaries we will see exclusions from the bottom up (significantly including the Greens' excess, which is going to favour Labor) until Hanson is excluded in eighth place. On the distribution of her preferences, the music stops and whoever is seventh at this point loses. (It is possible that one of the three could cross quota and create an extra throw to settle the final two here, but it won't be a big one if that happens.)
The Liberals start in the lead but in 2022 the parties that are being excluded were unhelpful to them. As mentioned the Greens preferences overwhelmingly favour Labor, One Nation, United Australia and Shooters, Fishers + Farmers preferences in 2022 helped Lambie and the only party the Liberals actually did well off was the Liberal Democrats, who under their new name of Libertarians have lost three-quarters of their vote.
I have built a model of the preference distribution based off the 2022 preference flows and in that model on the current live count if the button were pressed with no more votes added, Lambie ends up winning with .983 quotas with Falls winning on .920 and Colbeck out on 0.838. That's a projected margin between the majors for the final spot of 1.17%.
However there are a few reasons this might not hold up. The first is that the current count is unrepresentative. It is slightly heavy in votes from Franklin and Lyons, which have over 52000 votes each counted, compared to Bass and Braddon that are still below 44000. Bass and Braddon are slightly stronger for the Liberals (if you can call what they have polled there strong) and when corrected that improves the Liberals' position by about 0.3%.
The second is that the ordinary votes counted generally include the day booth votes but are light on prepolls, which in the past have been relatively strong for the Liberals compared to day votes. So the Liberals' position may improve slightly on that as well. The remaining changes to the count will include absents, out of division prepolls, remaining postals and provisionals.
The other hope for the Liberals is preference shifting - primarily, that voters for One Nation this election will be far more sympathetic to them and less so for Lambie. The Coalition did try to cozy up to One Nation more this election, and One Nation's Lee Hanson teed off at Lambie a few times during the campaign. However a very large shift (c. 20%) in the One Nation preference flows is needed to affect the result by itself. One Nation did include the Liberals on their how to vote card this year (in 2022 neither Liberals nor Lambie were included) but in 2022 only 3.7% of above the line voters for One Nation in the state copied their how to vote card, so the difference made by this is probably not much more than 0.1%.
My model is in some respects simplistic still. Preferences for JLN are assumed to be the same as 2022 although Lambie herself was not a candidate then - on the other hand the JLN campaign that year was bigger. Also I am not yet accounting for a division between Nick McKim's votes and votes for other Greens (the latter flow on at full value while the former are largely used to elect McKim).
Overall at the moment Colbeck's position looks the most difficult of the three, even if it is likely to be closer than my live count projection.
This article will be updated frequently through the next few weeks.
Wednesday 7th 8:15 am: The count is now at 72% of enrolment. The current count is about 95% ordinary 5% postals. The electorate composition has changed in that it is now very heavy on Braddon and Franklin and light on Bass and Lyons, which roughly cancels out. However the postals count is heavy on Clark and Franklin and light on the other three, which suggests that the postals count specifically is underselling the Liberals.
The latest raw quota totals are Labor 2.4858 Liberal 1.6149 Green 1.1839 JLN 0.4983 One Nation 0.3585 LCP 0.2365 TOP 0.2212 AJP 0.0916 SAP 0.0822 SFF 0.1528 LTN 0.0321 Citizens 0.0243 Edwards 0.018
In my model after preferences the live count has improved significantly for Colbeck. Lambie wins with 0.977 quotas, Falls wins with 0.893 quotas and Colbeck misses out with 0.883. It is easily within the error margin of my assumptions about preferences that Colbeck could be surving at Labor's expense on the live count, or less easily so that Lambie could be losing instead. In any case, because the current count is overly heavy on ordinaries and the postal count is weighted towards Clark and Franklin, I would currently expect Colbeck to improve and displace Labor from my model. A part of this is that for the time being the Greens vote has dropped back slightly. So at this stage Colbeck looks a fair bit more competitive than my first look at the race but there is still a very very long way to go.
Thursday 8th / Friday 9th overnight: The count is now at 81.94%. The leaders are now ALP 2.488 Lib 1.653 Green 1.144 JLN 0.506 One Nation 0.357 LCP 0.232 TOP 0.223. In my model (now with Citizens Party preferences added from the 2016 election! Joy!) Lambie now wins with 0.980, Colbeck wins with 0.921 and Falls misses out with 0.871. But that's close enough that it could be unsettled if there were across the board preference shifts against the Liberals and to Labor. The postal count remains skewed towards Clark and Franklin so there is potential for Colbeck to improve further.
By vote type the count is now 92.1% ordinaries, 7.6% postals, 0.3% absents. By electorate the count is relatively light on Lyons and high on Braddon but the distortions are now minor.
Something I should have mentioned earlier is the irrelevance of minor party how to vote cards. Tasmanian voters especially just don't follow them. The One Nation card last time was followed by 3.7% of One Nation voters. Not much times not much is nearly nothing.
Saturday 10th: Count now at 84.06% and the changes since the last update are too trivial to justify running my model again but for what it's worth there is a small further move towards the Liberals.
Wednesday 14th: I've only been updating this every few days as the numbers are changing very little as the last votes trickle in to the primary count while votes are being scanned and verified for the final count. The count is at 87.19% and the leaders are ALP 2.487 Lib 1.656 Green 1.130 JLN 0.503 One Nation 0.360 LCP 0.233 TOP 0.228. My model's output now is Lambie 0.976 Colbeck 0.927 Falls 0.870 - but bear in mind that the preference flow to either the Liberals or Lambie (or both) could weaken in line with their performance, and the flow to Labor could strengthen - a projected gap of 0.82% for sixth place is not very large in the context of what we saw in the Tasmanian Reps seats.
Sunday 18th: Count now at 89.3%. The Greens have lost 0.003 Q and One Nation 0.005 Q since last update while Lambie has gained 0.006 Q. Other changes are negligible.
I wonder how much more of a swing the Greens would have needed for Vanessa Bleyer to be in the mix? (Less than I would have thought before this election played out the way it has!)
ReplyDeleteBy my calculations, if Labor can get 3 in Tas (as well as likely in Vic/SA/NSW/WA) and keep Libs to 1 then they are on 31 with a non-Green ‘other’ of 8. This gives them an extremely narrow and complex path to getting something passed if Greens and LNP against. Otherwise if it’s Labor or Lambie dropping out then Greens have sole balance of power.
ReplyDeleteAm I right in thinking that all this complicated stuff -- models, graphs, historical comparisons, 2PP/3PP choices, etc. -- are only required to predict the outcome in the face of incomplete ballot data? I.e. once all the ballots are in, tabulating each outcome becomes a simple exercise, correct?
ReplyDeleteEven with full primary vote data, projecting the Senate can be very complex because we don't know the preference flows. We get the data from every ballot counted in the Senate but not til after the count is finished.
DeleteAssume 13 days is actually 16 days as we don’t work weekends? Still find in strange re: the lack of information out there in respect to when we will hear the result.
Delete13 days is the deadline for postals to be received but there is actually still a lot of data entry to occur before the exact number of primary votes is known and then the distribution can be done, I expect it will be in the last week of May. The data entry is about 70% complete.
DeleteMorning Kevin
ReplyDeleteI see the Greens' nrw bit of cope is that this is "best Senate result the Greens have ever had." I see this may be the case on current numbers, led by a 0.5 rise in NSW, but how likely is this to remain the case as more votes come in, including what i'm assuming is all the postals?
I'm not in a position to project that right yet but will say that at least some postals are in the count already, there are plenty in the count in Tasmania. Their all time high was 13.11% in 2010, they can't be far above that now.
DeleteAre you doing one for WA soon? I think I’m in a three-way race with 3rd Labor & PHON - even if the ABC won’t admit it! Would love your take on it
DeleteI may do a national Senate thread late tonight.
DeleteMy guess would be that Jacqui is not everybody's number one choice, but she will be their number two preference.
ReplyDeleteIf Lambi loses she can blame the blunder of not running reps candidates.
ReplyDeleteHi Kevin, when they are counting the senate, why doesn’t the AEC record and publish preferences? Don’t understand why they say that quotas can’t be struck and preferences distributed until 13 days after election day. It feels like they could share more information. I know they are waiting on all the votes to come in but they could at least start counting and providing detail on preferences for candidates at the bottom that are being excluded. They could show indicative results based on previous distributions.
ReplyDeleteIt wouldn't be a good idea for the AEC to publish indicative results based specifically on previous election flows, firstly because Senate preference flows change from election to election - a lot (a good example is the great strengthening in flows between minor right parties in 2022). Secondly there are always new parties that can't be projected and other unusual issues like whether there are candidates with significant below the line votes. It's not the job of the AEC to issue rough models involving subjective judgement calls as people like me can do that. Also because of the large number of candidates with low and similar vote shares the AEC can't be mathematically sure of the exclusion order at any point of the count.
DeleteWhat the AEC could instead do is what is done in the ACT and in Tasmanian council elections where at certain points of the computer data entry process, an interim distribution based on the votes received so far is produced - what would happen if those votes were all the votes. This would provide very useful information as to how preferences were flowing and make it much easier to project the results, but it would still take a while to get up to critical mass. For instance for Tasmania at the moment the manual primary vote count has reached 84% of enrolment (it will probably get to about 90) but the computer data entry count where preferences are keyed in is only at about 11%.
I haven't actually explored why the AEC doesn't do these interim counts for information before. Therefore I am not sure if the obstacle is just that the legislation doesn't let them do it or whether there are practical issues with it (cost or otherwise), or perhaps the AEC doesn't want to go down that path. I will see if I can find out sometime!