Senate preferences – how might they flow? Part 1


Up until now, I’ve largely ignored preferences when working out who are the leading contenders for Senate seats. While the new system will likely make it impossible for parties on a tiny vote to jump ahead of a number of contenders, preferences are still likely to matter.

I visited the NSW Senate count on Saturday in Moorebank. Apart from seeing the new scanning technology in action (a story for another day), I got to observe ballots being data entered for a few hours, and got a sense of some general trends.

The main takeaway is that nearly everyone is preferencing, but very few people have preferenced more than six parties above the line. In my time, I only saw three ballots which had a ‘1’ and no other numbers above the line, and a similar number who had numbered more than six boxes. This observation has been confirmed by Kevin Bonham at the Tasmanian count and appears likely to be a national trend.

This tells us that preferences will definitely be flowing, particularly between parties with some similar policies, but that preferences won’t flow all the way. Plenty of votes will exhaust after electing one person and the last seat could still be decided on substantially less than a quota.

In this post I’m going to run through the latest figures in the Senate count and the general political make-up of the preference available to flow. We don’t really have detailed information on how minor party preferences will flow. Even if you were to scrutineer, there are so few of them in such a large bundle of votes that you don’t get a decent sample.

In this post I cover the races in NSW, Victoria and Queensland. I’ll come back tomorrow with a similar analysis for Western Australia and South Australia. I’ll do a separate update on Tasmania, as the below-the-line voting there makes that race unique and worth following more closely.

The other thing to note is that these primary vote totals are based on an incomplete sample: as more declaration votes are counted the numbers should shift. I also ignore the effects of below-the-line voting, as we don’t yet have any data on this for the mainland states.

For each state, I group together all the votes for minor parties (and small leftover quotas for larger parties) into four groups: left, right, centre and random. “Random” covers unclassifiable parties like the Citizens Electoral Council, the Socialist Equality Party, Online Direct Democracy, etc. “Centre” covers Derryn Hinch, Nick Xenophon Team, the Veterans Party and the Motoring Enthusiasts. The rest should be clearly left or right.

New South Wales

Full quotas: 4 LNP, 4 Labor, 4 seats left to be decided

Parties in the race:

  • 0.9212 – Greens
  • 0.7559 – Nationals
  • 0.5392 – One Nation
  • 0.3568 – Liberal Democrats
  • 0.3461 – Christian Democratic Party

Votes to be distributed

  • 0.7280 – right
  • 0.6446 – left
  • 0.3684 – centre
  • 0.3401 – random

This is a race between three right-wing candidates for the last two seats. The Greens should get to a quota with Labor and other left preferences. The fifth LNP candidate (the Nationals’ John Williams) should also get home comfortably, although his ticket could well soak up some other right-wing preferences that could have gone to other candidates in the race.

Beyond that it’s very hard to say. You would expect a lot of left-wing voters, who may have numbered Greens, Labor and a plethora of other left parties in their top six, will exhaust before choosing between the right-wing choices. So it likely comes down to where preferences from the other small right-wing parties go.

The Shooters, Fishers and Farmers make up 0.26 quotas, Family First and the DLP each make up 0.15 quotas, and everyone else has less than 10% of a quota. Family First would normally be assumed to favour the Christian Democrats, but they are very close to the LDP and LNP on the ballot: any vote which puts either of these parties ahead of the CDP won’t get to the CDP.

It seems likely that the CDP and LDP will both do better out of preferences than One Nation, but One Nation is unlikely to be starved of preferences from other minor right-wing parties, and has a significant lead in the race.

Assessment: Greens and Nationals win ninth and tenth seats, with CDP, LDP and One Nation in close race for last two seats.


Full quotas: 4 LNP, 4 Labor, 1 Greens

Parties in the race:

  • 0.7802 – Derryn Hinch
  • 0.3785 – Greens
  • 0.3315 – Liberal
  • 0.2313 – One Nation
  • 0.2149 – Animal Justice
  • 0.1964 – Sex Party
  • 0.1958 – Nick Xenophon Team

Votes to be distributed

  • 0.9324 – right
  • 0.4172 – left
  • 0.2056 – random
  • 0.1161 – centre

Derryn Hinch is well and truly in the lead and should win the tenth seat. The fifth Liberal candidate and the second Green are then in the lead for the last two seats. They both have a decent lead over their rivals, and a lot will depend on how much preferences flow.

One Nation is the only minor right-wing candidate with a chance, with almost a quota of votes to distribute. It seems likely that the Liberal will do better from these votes than One Nation, pushing them further ahead. It’s also hard to see One Nation getting enough votes from this bloc to overtake the Green.

There is about 0.4 quotas sitting with left minor candidates, including Drug Law Reform and Marriage Equality. You’d expect the Greens to be sitting high in the top six of most of these voters, so the Greens should do well from this batch. On the left, there is a potential threat to the Greens from Animal Justice and Sex Party but they are a long way behind. The best chance for one of these parties would be to pick up some support from the other groups, and then get support from the other party when they are knocked out, but that doesn’t seem likely.

Animal Justice is in position C on the ballot. Their how-to-vote gives preferences to the Greens, although it’s not likely that many voters would have received it. The Sex Party didn’t give preferences to either the AJP or the Greens, but the Sex Party and the Greens are next to each other on the ballot and you’d expect preferences to flow.

The Nick Xenophon Team could also do well from preferences but would find it hard to get ahead of the Greens.

Assessment: Hinch to win 10th, Liberal and Greens likely for last two spots.


Full quotas: 4 LNP, 3 Labor, 1 One Nation

Parties in the race:

  • 0.8877 – Greens
  • 0.5940 – Liberal National
  • 0.4993 – Labor
  • 0.3542 – Liberal Democrats
  • 0.2542 – Nick Xenophon Team
  • 0.2497 – Family First
  • 0.2268 – Katter’s Australian Party
  • 0.2176 – Glenn Lazarus Team

Votes to be distributed

  • 0.7344 – left
  • 0.6783 – right
  • 0.2161 – random
  • 0.0877 – centre

The Greens should clearly win the ninth seat, with the LNP and Labor in a strong position to win the tenth and eleventh seats. The LDP has a lead for the twelfth seat but could easily be overcome.

There are a lot of left preferences but you’d expect most of these to be absorbed by Labor and Greens or to exhaust. On the right, the LNP will soak up many of these preferences, but a large proportion is made up of the One Nation surplus and Australian Liberty Alliance. These two parties polled over 0.3 quotas, and preferenced each other, but Family First was the next significant party on both of their how-to-votes. It seems likely these votes will favour Family First and KAP and are less likely to help the LDP.

It’s also worth thinking about where the Nick Xenophon Team would go. I suspect they may do reasonably well out of preferences from all corners.

Assessment: Greens to win ninth seat, Labor and LNP to almost certainly won tenth and eleventh seat, LDP leading in race against Family First, NXT and KAP for last seat.

Liked it? Take a second to support the Tally Room on Patreon!


  1. I voted below the line in NSW, but didn’t preference the LDP (I suppose being a positive alternative to CDP and PHON who of course I didn’t preference). We will see how many Greens, Labor and left preferences go to them after candidates hit quota or are excluded

  2. Were you able to tell if there was any evidence that people were more likely to preference familiar parties than the various lesser-known parties in making up their 6 preferences? Also what about donkey vote effects where people either preferenced the adjoining parties to the one they were voting for, or the first few at the end of the ballot paper to ‘make up the numbers’?

    It seems as though the instructions being given out by polling staff were ‘number 1-6 above the line or 1-12 below the line’, not including the words ‘at least’ which would’ve been more correct. I certainly got the first instruction, and everyone else I’ve heard from did too. If this was the official instruction from the AEC of what polling staff were to say, then that’s a bizarre blunder as it clearly should correctly have been ‘at least 1-6, at least 1-12’. Either way few people probably would’ve gone beyond 6 anyway, so I don’t see it as a disaster, but it’s an error that should be corrected next time. It also incidentally highlights how much voters are guided by the instructions from polling staff and don’t conscously remember voting procedures – a reason why fear of confusion from the system changing was always over-stated.

  3. My count is that if people follow HTV that in Qld KAP are likely to get in ahead of the LDP on the back of DLP, GLT and FF preferences.

    Similarly in NSW CDP will beat out the LDP on the back of SFF and DLP preferences.

    Probably a big if though, particularly in NSW. What SFF voters would follow the advice to preference CDP over the LDP?

  4. Ben,
    Thanks for the Article.
    I have been scouring the internet for a detailed analysis with realistic senate predictions but it is a desert out there.
    There are many speculations about what the outcome COULD be – e.g. One nation with 3-5 senate spots or the crazy Christians with 2 or 3. If feels like these statements have been made to create sound bytes for a sensation hungry media. They are only theoretical based on primary votes and not backed by any statistical predictions of preference flow. Your valuable analysis has shed some light on this.
    I have a gut feel that an unexpected preference flow will go to NXP or the greens and they might grab an extra 1 or 2 of the contested spots. This might just be wishful thinking. I like my independents to be centrist, not nutbags that isolate minorities or restrict freedom of choice.
    It will also be interesting to see how the 3 year terms are handed out, Could you do a prediction if it is based on reaching quotas with primaries?

  5. Hi Ben,

    I was at the counting centre at Emu Plains.

    The centre covered the counting for the seats of Greenway, Lindsay, Macquarie and Chifley.

    What stood out to me the most was how well One Nation was faring with preferences- I saw numerous 1. ALP 2. One Nation votes. Also many Labor voters preferencing One Nation 3rd, 4th, 5th and 6th. One Nation polled very strongly in the Lindsay, Greenway and Chifley seats in their own right too.

    I can easily call that more ALP voters were preferencing One Nation over the Greens (barely any Labor voters were following their HTV’s. I guess the story may be quite different in other parts of Sydney but no doubt One Nation did very in those seats.

  6. Fascinating insight into how the senate votes are going, thank you. Clearly AEC needs to provide more education somehow on voting instructions. And more info on the odds and ends parties, many of whom are seriously misnamed.

  7. I did the whole country on Sunday night (about 65% vote counted) and it didn’t change much Monday, perhaps using a slightly different method to you, and the Senate is as follows:
    NSW Coal 6, ALP 4, Greens 1, 1 to either Coal or ALP, 1 unsure (close contenders One Nation/Lyneham (Lib Democrats) but I don’t think they will necessarily get it).
    Vic Coal 5, ALP 4, Greens 1, + 1 ALP or Greens and 1 unsure (noone seems to be preferencing Hinch)
    Qld Coal 6, ALP 4, Greens 1 One Nation 1- no uncertainties there.
    WA Coal 6, ALP 4, Greens 1, and 1 going to either Greens or ALP.
    SA Coal 4 ALP 4, Xen 3, Greens 1- no uncertainties there
    Tas Coal 5, ALP 4, Greens 1, Jacqui 1 and 1 going to either Greens or ALP.
    NT Coal 1, ALP 1 – no uncertainties there
    ACT Coal 1, ALP 1- no uncertainties there.

    Coal 34
    ALP 26, Greens 6 (& 3 to either of these from WA, Tas and Vic)
    Xen 3
    Jacqui 1
    One Nation 1
    and 2 unsure (1 in NSW and 1 in Vic)

    giving the balance of power I think to Xenophon, as in a crunch where everyone sides with the Coalition except Labor and Greens, as even if the Coalition gets both extra (unsure) seats they have at most 36, whilst the ALP and Greens will have 35. So 3 Xenophon votes will make or break legislation.

  8. N.B. In the above I should have written it:
    NSW Coal 6, ALP 4, Greens 1, 1 to either Coal or ALP= 1 unsure
    Sorry gave NSW an extra Senate seat (unlucky 13?)

  9. I can’t help but wonder which demographics are contributing to the One Nation result.

    Is it the older, more “traditional” voters who tend more towards racist attitudes (that were much more common 50 years ago), or is it the younger voters who weren’t politically aware during Hanson’s first time in parliament?

    I know we aren’t going to learn it through the count, or anything, but clearly one of these two groups, or both, or some other similar group, is behind One Nation’s strong performance, particularly.

  10. the one obvious is the last 2 seats are almost a lottery to predict in these 3 states
    onp is only sure of one seat in qld……..

  11. Glen: There’s a whole cohort of younger votes who would only know of Hanson from Dancing With The Stars.

  12. My thirty-something Chinese born wife only knew about Hanson from Dancing with the Stars, having moved to Australia in 2003. I had to fill her in on the whole “Asians form ghettos and don’t assimilate” thing.

  13. The demographic is low income blue collar types and social conservatives giving a big up yours to both major parties. This is why the lower house preferences from One Nation are splitting 55/45 (apart from Longman where the HTV directed to Labor).

  14. AJP in Vic flows might be greater than you expect – they were one of the non-majors with a steady presence handing out HTVs, along with FF and Rise Up, at least in the south east of Melbourne.

  15. Simon – I also get KAP as the favourite in Queensland, with Family First as the threat.

    In fact if Family First can outlast the second One Nation candidate, I have them taking the seat on One Nation preferences. However in most scenarios that doesn’t happen because One Nation get enough of a boost from the Shooters and the Australian Liberty Alliance to be ahead of Family First, whose preferences then deliver the seat to KAP.

    I also expect One Nation to pick up a disproportionate number of non-HTV preferences due to name recognition. But hopefully not enough to actually give them a chance at the seat.

  16. There is also the question of whether or not the inter-group preferences in the Senate will be used to choose the long term Senators, by using the who would have been elected in a half-Senate election, or whether only the first preferences/intra-group preferences will be used, by using the first elected method (as there are more than enough senators elected on first and intra-group preferences to fill the first six in each state).

    There has been talk of using the first elected method. The biggest effect of that would be in three states discussed in this post. In Queensland and NSW it would mean short term Green Senators only (unless the Greens make up the difference between a quota on primaries in late counting). In Victoria it means a short term for Hinch.

    I suspect that Hinch is in with a chance at re-election in 2018/2019 but that the Greens are very unlikely to poll 2 quotas each in Qld and NSW in a half-Senate election. Thus the main effect is harming the Greens in the next Senate.

  17. The Shooters did not man booths. The ALA covered some booths in seats they ran in, but I can tell you at my two local booths they turned up at 10am and were gone by early afternoon missing the first two hours which is peak turnout time.

    I think the HTV instructions will not be followed by plenty of voters which limits the ability of One Nation to leapfrog (or Katter).

    I really think it is LDP vs Family First here. Mind you if there is enough funky below the linet that could change.

  18. My Queensland modelling only assumed 35% compliance with HTVs for the micros that didn’t man booths. Just a wild guess assuming some voters will have sourced it online or from newspapers.

    The rest I had randomly generated with a bias towards parties with similar philosophies (obviously subjective).

    Anything is possible until we see some preference data. If they’re more random than I’m assuming, then the LDP obviously comes back into the equation.

    I’d actually like to gather some proper data for Queensland if anyone knows a micro candidate that would appoint me as a scrutineer for a couple of days.

  19. Nick C, the polling staff at my local booth (Niddrie Central, Division of Maribyrnong) also left out the ‘at least’ before referring to 1-6 and 1-12. But at least ( 🙂 ) the staffer gave the right answer when I asked if I could number more.

  20. 1 to 6 above or 1 to 12 below was likely said (almost) everywhere. The text to read actually did say ‘at least’ or words to that effect, but polling booth staff are as trained as the average voter, and not experts on the difference of numbering after 6. They are also in a hurry and when you are giving someone the ballot papers you are also trying to get the voter to understand that they have number every box in the House of Reps. So at least 1 to 6 or 1 to 12 is working against you there.

    There certainly were voters explaining to polling booth staff that they could (and would) number after 6 on the Senate paper.

    Also to be honest all the HTVs had 1 to 6. We know about 75% follow HTVs when they get them right? so I imagine many people would have happily ignored instructions to keep on preferencing if they chose to do so.

    Im just happy the ‘doomsday’ scenario of just ‘1’ didn’t happen. That’s almost what happens in the NSW upper house(or 1 and 2) so I think most votes in this election got to where the voters wanted them to go.

    (I used to preference every box below the line, gave up after 50 or something with the top LDP candidate as my next least bad choice, we will see if the point something vote mattered)

  21. I could actually make a list of things that polling booth attendants are trying NOT to get the voters to do.

    Destroy their paper.
    Walk out with their paper
    Vote informally
    Vote above and below the line (though that can count so up to the voter I guess)
    Identify themselves on the paper.
    Pretend to be someone else

    A small mistake there is pretty significant.

    Sticking to just 1 to 6 will make sense for a lot of voters as they really have no idea who most of the Senate parties really are.

  22. Just looking at the Queensland count. Both Labor (now 3.46)and the Greens (0.87) quota is falling. LNP (4.62) and LDP (0.36) increasing.

  23. Just looking at the Queensland update. Preference flow from parties in the 0.1 to 0.2 range will be important. These include Hanson’s overflow, Shooters Farmers and Fishers, ALA, Sex/Hemp, Animal Justice, Marriage Equality. If any of these flow strongly to Family First, Katter or Xenophon then the LDP and possibly the fourth Labor senator could be in trouble.

    That said I expect they won’t flow strongly and will scatter. It still will be interesting to look at the flows when they are revealed.

  24. I still have the final Queensland seat as being between KAP and Family First, with Family First now a slight favorite. I think Labor are very safe for the eleventh seat.

    The scenario for KAP to win requires the second One Nation candidate to overtake Family First, giving KAP a boost from whatever proportion of Family First voters followed their published HTV.

    For One Nation to catch Family First would require some combination of Shooters and ALA voters following the HTV and One Nation getting a larger share of the “do it yourself” preferences. I think the latter is plausible – a party that can attract first preferences can attract second preferences and One Nation polled 9% compared to the 1-3% of the other contenders.

    Despite their big lead on primaries, I struggle to see the LDP holding on. I just can’t see where they will get preferences from. Not just from not being on HTVs – I also think they’ll struggle to get their share of the random preferences. And I think enough of those random preferences will remain live to allow them to be overtaken.

  25. The LDP may get some preferences from voters aiming for the LIberals and being confused. There will however be far fewer of these than if the logos had not been introduced. Maybe there was a realisation in the government about this additional risk of lost votes at later stages in the count and this is what caused the introduction of logos?

  26. Not relevant to any of the contests being discussed here, but for anyone wanting to play with some raw data, the AEC has just released the preference data for NT after announcing the result earlier today.

    They’ve used a new format compared to that used previously for BTL data. I think I’ve figured it out but it’s going to be painful to get it into a format to feed into my database. But happy to have a data set to work with ahead of the more interesting counts.

  27. Labor down to 3.43, Greens up to 0.89, LDP up to 0.37 and LNP down to 4.59.

    Katter on 0.23, Xenophon on 0.26, Family First 0.249.

  28. Based on the incredible preference splatter seen in the Tasmanian count (see Kevin Bonham’s stats, I’m withdrawing my analysis which assumed significant following of HTVs.

    That LDP lead now looks very formidable and the fact that they got 4.2% of all second preferences in Tas despite being on the right of the ballot and only having a primary vote of 0.8% suggests that they will get some preferences in Qld.

    However exhausted votes were lower than expected in Tas, and while they had a smaller ballot, it may mean there is more room to catch a candidate. If it takes even 0.7 of a quota to grab the final seat, that only requires someone like Family First to out-preference the LDP at a little over 4-to-3 in order to catch them, and even One Nation needs less than 2-to-1 if they can catch some early prefs from parties like ALA (got 18% of them in Tas).

  29. Due to the Greens being under a quota, Labor’s preference flow will struggle. I do not think the fourth seat is locked up. There are a lot right parties (one nation, ALA, Rise Up, Christian dems, country minded) which will help Katter or family first. Labor is the favourite and has a head start, but it is an interesting race.

  30. One of the other lessons from Tasmania was that micro voters don’t just preference other micros. That was a big factor in the low exhaust rate. While the majors don’t pick up preferences at anything like their primary rate, they don’t just get them at the same rate as the micros either.

    Ranking the parties by the portion of ATL second preferences they get from votes on which they didn’t get first preference, the top four were Labor (19.8%), Greens (13.2%), Lambie Network (11.3%) and Liberal (10.9%).

    Labor in particular got second preferences from everywhere – including 17.8% from the Liberals! They got 7% or more from everyone except ALA and CEC.

    Even with the bigger ballot, I think they cruise to a seat, out-preferencing the micro contenders by at least 2-to-1.

Comments are closed.