Photo finish: Victorian Senate

Saturday evening. This post will be used to follow the final stages of the count for the Senate in Victoria, where at the close of election night counting the third Liberal candidate Scott Ryan has a 1.3 per cent lead over Greens candidate Richard di Natale. Of the remaining five seats, three have been won by Labor and two by Liberal.

Sunday evening. I don’t believe any votes were counted today. What we have so far is all polling booth votes counted for above-the-lines and first preferences from below-the-lines. Hopefully someone will correct me in comments if I’m wrong on either count.

Tuesday 5pm. Still nothing doing. Recent activity as consisted of breaking booth votes down into above-the-line and below-the-line voters. Counting of postal, pre-poll and absent votes will begin shortly. Strong performance by the Liberals here in lower house voting does not bode well for the Greens.

Wednesday 5/12 8pm. Counting of nearly 200,000 pre-polls and 100,000 absent votes have dashed any hope of a late-count upset here, boosting the Coalition vote from 38.9 per cent to 39.2 per cent and cutting the Greens from 8.4 per cent to 8.2 per cent, increasing the Greens’ deficit to an insurmountable 2.1 per cent.

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics.

129 comments on “Photo finish: Victorian Senate”

Comments Page 3 of 3
1 2 3
  1. Karl, yes I see — didn’t think of that. How many of the 106,000 or so “unapportioned” do you expect will go to Jacinta BTL?

  2. Not Telling!

    You’re missing the point of the unapportioned. Votes that are ALP 1 will stay within the ALP group. There will be a very small proportion of votes that stray outside the ALP group after being used to elect at least 1 ALP candidate.

    For example, in the last federal election, the election of Kim Carr resulted in a surplus of 650887 votes, 650172 of which went to the 2nd ALP candidate, 304 went to Labor candidates 3 & 4, and 376 went to all other candidates combined. (The balance of 35 were exhausted (10) or fractional votes lost in transfer (25)).

    It is worth noting that 376 out of 3000000 votes is 0.01% or 0.0009 quotas.

    Even if you take the denominator as the total number of ALP BTL votes instead of total ALP votes, you still get 376/14598=0.026. So only 1 vote in 40 ALP BTLs will start ALP but go to another party before leaving the ALP group.

    As ALP is 7000 votes over quota, this 300 odd votes is NOT going to make a difference. The assumption that AG’s calculator makes of having all votes stay within a group is still 99.99% valid.

  3. On another note, there are two ways of looking at the remaining votes to be counted in the Senate.

    The difference in the votes between the (almost) complete House scrutiny and the incomplete Senate scrutiny is as follows:

    Ordinary votes: 2320 extra senate votes remain to be counted
    Absent votes: 29489 extra votes remain
    Provisional: 1220 more votes have been counted than the House. (WTF?!?)
    Pre-poll: 39155 extra votes remain
    Postal: 25865 extra votes remain

    Since Lib votes go to the Greens, any extra Lib or Greens votes will end up with the Libs and any extra Labor votes will stay with Labor. The noise generated by the other minor parties is insignificant and can be safely ignored.

    So, what is the CURRENT margin (ALP – Green vote), according to the ABC calculator? 40750 votes. (Pre-liberal surplus distribution)

    Well, the Libs are 25867 over quota and about 91% of this will go Green, and the balance will go Labor. This halves the margin to 19800.

    Extra votes:
    Ordinary: 0.7% weaker (than total) for libs, 0.55% stronger for Labor and 0.14% stronger for the Greens. Net extra labor votes: 0.7%+0.55%-0.14%=1.11%*2320=25 votes. Labor margin now 19825

    Provisional: There’s more than there should be, so let’s leave this well alone.

    PrePoll: 2.24% stronger for libs, 1.29% weaker for labor, 1.07% weaker for Greens. Net extra labor votes: -2.24-1.29+1.07=2.46%*39155=-963 votes. Labor margin now 18862 votes

    Postal: 8.64% stronger for libs, 4.53% weaker for Labor, 3.27% weaker for greens. Net extra labor votes: -8.64-4.53+3.27=9.90%*25865=-2561 votes.

    Margin now 16301 votes. That’s equivalent to 0.5%.

    What chance is there of a swing of 16301 votes when there’s only 95000 votes remaining? Pretty low. Not zero, but approximately zero.

    Sure you can assume weird stuff with the BTLs going 1 Collins, 2 Green, 3 Feeney, but there’s not going to be more than 300 of those. If anything, you could assume more Lib BTLs going to Labor before the greens.

    SO, as said above, 3-3. Not certainly but very likely. Libs prefs going to Greens not Labor as previously assumed makes a difference, but not significant enough to switch the result.

    Numbers correct as of now.

    Future research would look at the individual electorates in which these votes remain to be counted and repeat the above 4-step process, but, frankly, I can’t be bothered!

  4. Sorry, I think I meant to ask something like “how many of the unapportioned will leave Labor after Jacinta or after Jacinta and Gavin?”… but I’m a bit slow on the numbers game.

    And your point about the tiny proportion of votes that stray from ALP1 or ALP2 isn’t lost on me, though the Kim Carr example you cite is difficult for me to believe at first glance — so only 376 out of the total number of ALP BTL voters strayed from the first 2 candidates in their 3rd preferences? That’s… um… not many.

    Are you telling me that my hope-in-the-face-of-all-odds is better invested in something else? Say, a freak earthquake that swallows entire Divisional offices whole?


  5. Hang on, I’ve got a better one; Antony Green’s senate calculator is adding weight to the ALP vote due to an inbuilt glitch on ABC computers (as is apparently compulsory in all ABC departments) and the reeeeeeeeeal margin is actually in the Greens’ favour. (What’s that? A straw? CLUTCH IT! CLUTCH IT!)

    Revised prediction:
    Richard by 6. (That’s even more than Fran Bailey!)

    You heard it here first. Again.


  6. NT:

    Spot on with the bias comment.

    As for the 376, relooking at the numbers, I suppose you could increase it, as that’s 376 stray for Carr and an additional stray of 123 from Conroy. Also note that Ristrom only got 191 of this 499, and that some of this stray balance (308) could have easily come back to Labor via a minor party.

    Note that this is approx 600 ballots each time but that gets reduced to 376 and 123 respectively due to the transfer values of the ballots following the election of the top 2 candidates (TVCarr=0.60, TVConroy=0.21)

    I suppose these are an underestimate in that voters could have gone ALP1, ALP4, Green, although I reckon there’d be less than a dozen of these that go from an elected ALP candidate to an excluded ALP candidate, and then to elsewhere. Multiply these dozen by a 0.6 or a 0.2 TV and you’ve got a not-very-clutchable straw! (an additional 2 votes – not even enough to switch McEwen!)

    The download for the 2004 Distribution of preferences is at:

  7. Can someone please check my logic? Ta.

    (As at 4.30pm.)

    Below: Party, quotas on primaries + quota collection from micro parties assuming assuming 100% ATL votes = total, respectively:

    Labor, 2.9162 + 0.0452 = 2.9614
    Coalition 2.7747 + 0.3376 = 3.1123
    Green 0.6990 + 0.2222 = 0.9212

    i.e. Senators elected so far = 3 Lib, 2 ALP.

    So, 0.1123 Lib (and Lib-contributing micro party) excess to be distributed, of which Greens need 0.0788, or 70%. My estimate is that 90% of the excess will go to the Greens.

    Ergo, Richard is not out of the race yet.

    Obviously this doesn’t account for BTL votes, but there is still room to move, non?

    Be nice to me. I’m new to this number-crunching thing.


  8. Antony Green’s Senate Calculator says, with the 4.30pm votes entered, that Labor reaches their 3rd quota after Family First’s preference distributions (which see One Nation’s votes which originally went to FF go to Labor), with 0.0107 quotas extra.

    0.0107 quotas is 4,735 votes. That’s bugger all — and it is most likely less than this, surely — my thinking:

    – the senate calculator assumes 100% ATL voting, but some BTL Labor voters may have gone [1] Collins [2] DiNatale or [1] Collins [2] Marshall [3] DiNatale. There is no way of knowing how many people did this, but let’s assume that it’s ever so slightly higher than last time due to the media scrutiny of Labor preference deals that the Family First 2004 deal brought on. Does anyone want to hazard a numerical guess?

    – other BTL voters from minor parties are more likely to go to the Greens before 2 major parties than go the way the group ticket went.

    If, for any reason, Labor DOESN’T score these 4,735 votes, when the Liberal excess is distributed, the Greens will receive about 90% of these, overtaking Labor. Both Labor and Greens will finish over a quota after that distribution, but the Greens will score more, securing them the last seat.

    I just don’t know how to get a detailed fix on the likelihood of those 4,735 votes not actually being there in reality.



  9. Re 108:

    There IS a way to determine (or at least, get a feel for) how Collins btl voters have tended to vote with their 2nd and subsequent preferences.

    In fact, there are two.

    The first way is to scrutineer the recount at the AEC divisional offices, and note where Collins btl voters are sending their preferences.

    The second way is to head out to Hawthorn and watch the data entry.

    I have done the former, but not the latter. I think a higher proportion of Collins btl voters have preferenced the Greens in a 2nd or 3rd spot, compared to Carr btl voters in 2004. OldTimeHack points out that about 1 in 40 btl strayed from Labor in 2004. As a guesstimate, I would put the figure around about 1 in 25 this time; maybe 1 in 20. Still, there just aren’t the VOLUME of Collins btl votes necessary to influence the outcome.

  10. Sorry Jen, I’ve got no idea! On another matter while we wait, are you in a position to explain why the Greens put up 6 candidates for election in Victoria? The only other parties that did so are the DLP and Family First. Doesn’t it make more sense to just have 3 candidates (or even two)?

  11. I’m with Jen, when will we know? Just put us out of our misery so we can get on with x-mas. Then again, if the Greens get up, it will be a very, merry x-mas indeed!

  12. Karl,
    there was much discussion re: putting up 6 candidtes and not all were in agreement ( i thought it was unnecessary to have more than 3).
    However the rationale was that each candidate attracts a personal vote which preferences the no.1 candidate, and every vote counts as we know!
    In the post-election review all these matters will be discussed and all opinions welcome, so go for it.

  13. Surely the reason for running 6 is to gain extra exposure?

    As it’s so hard to find any particular party the “stand out” effect probably makes it worth it.

    Now… Let’s look at the ABC calc….. I’ll be back!

  14. NT 107:

    Your numbers seem a bit wrong when compared to the ABC senate calculator (as at now) with the split being:

    Note that these numbers almost but not quite add to 3.000.

    So after the distribution of the Lib surplus, it is impossible for both Labor and the Greens to be over 3, since the libs will retain their 1 and distribute their surplus and the total live surpluses will be a tad less than 2.

    (On the other hand it is technically possible, but extremely unlikely for both Labor and the Greens to both end with less than 1 quota (0.9999 to 0.9998) and in that case, the Act says the leading party is elected.)

    In terms of the BTL impact, IMHO this effect will favour Labor rather than disadvantaging Labor. I reckon there’d be more people voting BTL Lib but then giving their preference to Labor for ideological reasons rather than to the Greens who share almost nothing with Libs policy-wise.

    Assume a worst case scenario (for Labor) where there is an increased Labor drift and a decreased Liberal drift. The extra impact this would have is in no way going to exceed 500 votes, or 0.001 quotae.

    As for minor party votes going Green before the majors, you’re right there. But don’t discount extra Lib votes heading Labor’s way before going green.

    Karl, I don’t doubt your observations, but unless they’re backed up numerically (eg 40 of 1000 votes I saw drifted) I am always hesitant to accept as truth. I have an extensive scrutineering background and always tried to get samples to back my observations. Remember, that just like the polls, there is always a margin of error, and if you observe 4/100, then there will be an associated MOE too.

  15. A healthy skepticism makes for a good watchdog! OldTimeHack: I accept, and concur, that my observations should be backed up numerically. Once I’ve completed my report on my scrutineering to the Greens, I’ll upload some numbers, but that won’t be until next week.

  16. OldTimeHack and Jen:
    Thanks for your comments re: numbers of Senate candidates. I can’t say I have an opinion one way or the other at this stage. Just curious.

  17. Late counting in the Vic Senate is flowing to the ALP away from the Libs. Small changes only. But with maybe 1.7-2% of the vote to go, the trend could see the 5th and 6th places swap over again between Lib and ALP. That could be interesting.

  18. I would be surprised if 5 and 6 swapped. The gap is 20000 votes, and 2% to go is 60000 votes. That’s a 30% swing!

    Karl, I suppose you could argue that those whose fans that vote BTL who are more likely to stuff it up (ALP?) should have a responsibility to make it as easy as possible to vote BTL by running less candidates, whereas those who are committed, have spare time and are passionate (greens?) can get away with making it harder to vote BTL! But I think that any such effect would be outweighed by the extra exposure and increased noticibility having 6 candidates would get you….

    Let’s just hope Ruddy calls a double dissolution. That would be so much fun for amateur psephs like me! I might even run as a senate candidate with 11 mates and hope to get a harvest 🙂 Quota <8%

    Karl, I eagerly await your numbers (if they let you publish)

  19. OldTimeHack:
    I believe they increased the candidate fee this election to $1000 for senate candidates … so if you and your 11 mates want to have a go you’re looking at a cool $12k! Ah, the price of democracy…

  20. On election night I was scrutineering for another party in an inner Melbourne seat and whilst my colleague and I were focused on the House of Reps, being new to scrutineering I watched some of the senate votes being counted mainly out of interest.
    I was surprised by the number of Greens votes that were informal due mainly to the fact that many Green voters voted below the line and didn’t complete their vote correctly. Whilst this is only speculation I can’t help wonder if the result might have been different if more Greens had taken the above the line option. I have always voted below the line out of principle, but next time I think I’ll go above the line unless I’m not happy with the preference deal.

  21. AEC Hidden Secret – You can do both. If both above and BTL votes are formal, the BTL counts. Also, standard rule of thumb is that a vote is valid (at least in Victoria) if at least 90% of the boxes are filled and formal and a clear intention can be made. Handy little trick I forgot to use last month.

  22. 125

    Kruscica . . . the 90% rule holds true nationally. And it’s also possible to exhaust your vote by repeating the same preference more than once (provide you have a single 1st preference indicated). The rule is that as long as by replacing no more than three wrongly-stated (gaps in sequence, repeated numbers, etc.) preferences a valid BTL vote could be registered, the paper is formal. But all preferences expressed after the last correctly indicated sequential preference (i.e. from and following the break, repeat, etc.) exhaust and are not counted.

    I am certainly not advocating this practice, for two reasons – one practical, one legal – a) it’s all too easy in the fast flow of the examination/count for the vote to be wrongly set aside as informal; and b) the act of advocating such a vote is illegal (it’s a sort of mini-Langer vote)

  23. Hey Dembo, long time no see.

    $12k, eh? Could I claim it as a tax deduction if I were a staffer? (Not that OTH has ever held any full-time or part-time job for any MP union or consultancy)

    Dyspnoeia is spot on. In elections with at least 10 (or is it 20?) candidates, if you would have completed at least 90% of the boxes had no more than 3 numbers be changed, then your vote is counted. There’s only a small chance it would not get counted because it would almost certainly be manually entered rather than being set aside as informal.

    Just make sure you vote for a weird candidate at a small polling booth (as did OTH) and then you can see that there has been “1” vote for this BTL candidate, proof that your vote has been counted.

    And as kruscica says, just vote a 1 above the line as well, so if you do happen to stuff anything up, then it will revert to an ATL vote.

    OTH approves of this method, but would never encourage anyone to vote 1-2-2-2-3-4-5-….-58-59, which actually would have been an accepted vote in Victoria with 68 candidates.

  24. dyspnoeia – are you sure the act of advocating such a vote is illegal? Until recently, I thought it was illegal to advocate an informal vote, but such a vote is not technically informal anyway, and on checking I found that S.329A (regarding which Langer was prosecuted for contempt following an injunction not to breach) was abolished in 1998.

    I would quite like to know what the actual legal situation is, as lack of knowledge of it was about the only thing that stopped me publicly endorsing Fat Cat for Denison this time round. 🙂

  25. Thanks Dyspnoeia

    I don’t advocate leaving ballot spaces blank as a rule, just pointing it out that it can still be valid as a corollary to the validity of Senate papers, as another little-realized thing in the counting rules, and no, I most certainly wouldn’t advocate doubling up on numbers 1-2-2 or 1 … 57-57. It is disturbing though to recall the Langer case…

    On another note, as a longtime lurker with now two posts to my name, thanks to all who’ve participated over the long period, offered morsels for edification, and to William for setting up the site, husbanding it, and marshaling with a light touch.

Comments are closed.

Comments Page 3 of 3
1 2 3