Senate voting reform – the bill drops


So we now have the government’s legislation for Senate voting reform. You can read it here.

The key points are as follows.

Abolition of group voting tickets

From now on there won’t be any distribution of preferences beyond a single party group unless the voter marks it themselves on the ballot.

Introduction of optional preferential voting above the line

From now on you will be allowed to number as many boxes as you want above the line, and your vote will flow through each party group in ticket order.

The ballot paper will carry instructions saying the voter must number “at least 6” boxes above the line, although that would revert to being full compulsory preferential voting if six or less groups nominate.

Having said that, votes just containing a ‘1’ above the line will be formal, and there is no “Langer clause” which would prevent parties or other groups advocating for a person to number less than six boxes above the line.

No major changes to below-the-line voting

Despite JSCEM recommending optional preferential voting below the line, this bill only slightly loosens the requirements for below-the-line voting. You’ll still need to number most boxes for your vote to count, but you’ll be allowed up to five sequencing errors, up from the current three.

Party logos on the ballot paper

This one wasn’t expected! Presumably this is motivated by Liberal concern about confusion with the Liberal Democrats. It’s not unheard-of: New Zealand has party logos on the ballot.

Prohibition on being Registered Officer of multiple parties

This is to address the concern about David Leyonhjelm being the registered officer (who is the official who liaises with the AEC and nominates candidates) of multiple parties.

I will have some more commentary this evening about the political impact of the reforms, but feel free to use this post to discuss the reforms as they unfold today.

Update: No counting of Senate ballots

Unfortunately I missed one major change. The legislation proposes that Senate ballots are no longer counted and recorded in group totals on election night. Antony Green has looked into this.

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


  1. I’ve been trying to think of reasons for not making below-the-line voting easier. The main one I can think of is that it may be to reduce the workload for the AEC by encouraging above-the-line votes. It’s not clear yet whether the AEC intends to go to full data entry of all ballots, but even if they do, above-the-line votes will be much quicker to enter.

    I’m also wondering how long it will be until someone raises a constitutional challenge to the whole concept of groups as offending the requirement that Senators be “directly chosen”. One consequence of these changes is that in addition to the existing situation of ungrouped candidates being unable to receive above-the-live votes, they will now also not receive any preferences from above-the-line votes.

  2. The reason for not making below-the-line voting easier is that parties want you to vote for their candidates in the preference order they’ve grouped them in. Making below-the-line voting easier ‘risks’ more voters deciding to number a party’s candidates in a preference order different to the order they appear on the ballot paper. It’s just about party control, nothing more.

    I don’t think parties should be so concerned though. In NSW state elections there’s never been any significant below-the-line voting patterns departing from the preference order of groups. Even under the same model in NSW local government elections it’s extremely rare for below-the-line preferences to get a candidate lower down on a group elected over someone above them. The only example I can think of where this has happened in local government was where a high profile councillor ran as an ungrouped below-the-line candidate and issued HTVs recommending a 2nd preference to the 2nd candidate in another group, who then defeated that group’s lead candidate on the ungrouped candidate’s surplus, but it basically never happens and I can’t believe it would happen in a Senate election.

  3. It is disappointing for voters below the line. Why can`t an individual vote vote below the line with the numbers 1 to whatever i.e. number of senators to be elected. This system works very well for Legislative Council Elections in Victoria.

  4. The changes will mean that we will have a lot fewer smaller political parties. It will also mean that the Liberals and Labor will become more similar. This is because they do not need to take notice of non-mainstream issues.
    John Flanagan

  5. The proposal to retain a prima facie requirement for full preferential voting below the line, with just a slight relaxation of the current savings provision, is risible. It means that the problem of discrimination against those who want to come up with their own preferences – for example to vote for the women candidates before the men – will continue on.

    And it will introduce a whole new absurdity: if I mark the required 6 preferences above the line, that will translate into a certain number of preferences for candidates – say 25. If I marked the exact same 25 preferences below the line, my vote would be informal.

    It’s completely devoid of logic: you might just as well have different formality rules for ordinary votes and postal votes.

  6. I agree 100% with Michael Maley. This reform would be a travesty unless it frees us to vote as we choose below the line.

  7. I’m actually quite sad about the registered officer changes. It made the LDP’s front parties so easy to track.

  8. Thanks for the post, Ben. If voters can vote above the line for 1,2,3,4,or 5 groups what is the significance of “at least 6” boxes being required?

  9. I too agree with Michael Maley. I can see that perhaps it was intended to reduce the time spent checking formality and data-entering ballots, but the idea that a BTL ballot may be informal whereas an ATL ballot expressing exactly the same preferences is formal is risible. It makes it look like the only reason the BTL option is being kept at all is as a figleaf to cover the constitutional requirement for “directly chosen” Senators.

    Surely this means that they’ll have to allow ungrouped candidates their own column?

  10. Does the absence of a “Langer clause” mean that parties can encourage a single 1 vote above the line in their HTV material?

  11. Here is a long comment – apologies in advance – on the amendments to the Senate electoral system proposed in the Commonwealth Electoral Amendment Bill 2016.

    The main defect of the Bill is its rejection of the unanimous recommendations of the Joint Standing Committee on Electoral Matters relating to Senate votes marked below the line. The Committee had recommended that in conjunction with above the line preferential voting, there be provision for:

    “…‘partial’ optional preferential voting below the line with a minimum sequential number of preferences to be completed equal to the number of vacancies:

     six for a half-Senate election;
     twelve for a double dissolution; or
     two for any territory Senate election.”.

    The Committee further recommended that “appropriate formality and savings provisions continue in order to support voter intent within the new system”.

    In his “Additional Comments” on the Committee’s Report, Senator Xenophon noted that the provisions of his own Commonwealth Electoral Amendment (Above the Line Voting) Bill 2013 were consistent with the Committee’s recommendation on that point. It follows that the approach now being taken by the Government also represents a rejection of that proposed by Senator Xenophon.

    The following points are worth emphasising.

    • The Explanatory Memorandum to the latest Bill provides no substantive explanation at all of why the Government has rejected the Committee’s unanimous view on this issue.

    • The Bill proposes instead that section 270 of the Commonwealth Electoral Act 1918 be amended to permit up to five numbering errors on a Senate ballot paper marked below the line, rather than the three which have been permitted from 1984 until now.

    o No justification is provided for the choice of five, rather than four, or six, or any other number one might care to specify.

    o There has been no evidence published that I have seen concerning how many informal ballot papers at the last Senate election would have been rendered formal by this change, though as the preferences were all captured electronically by the AEC this could presumably be determined. (Some programming changes would be needed to get that information, but some such changes will be required anyway if the system proposed in the Bill is to be computerised.) The Committee would be well advised to request this information.

    o Nor has there been any evidence provided that I have seen on how many Senate votes would be informal under the scheme proposed in the Bill, compared with the scheme recommended by the Committee. Again, this should be able to be determined by the AEC through the analysis of 2013 data, and the Committee would be well advised to request this information.

    The scheme proposed in the Bill will create an anomaly never previously seen at Senate elections: identical preferences for candidates may produce a formal vote if expressed using the above the line mechanism, but an informal vote if expressed using the below the line mechanism. Consider a NSW voter in 2013 who voted 1, 2, … 6 above the line for six of the two-candidate groups. This vote would be counted as expressing twelve preferences in total, for the candidates of those six groups. If, however, the voter had instead written those exact 12 preferences below the line for the candidates in question, the vote would, under the scheme in the Bill, have been informal. This is absurd on the face of it: why should a particular set of preferences constitute a formal vote if expressed using the above the line modality, but not if expressed using the below the line modality? As I noted earlier, one might just as well have different formal numbering rules for ordinary and postal ballots.

    In a detailed article on Senate electoral reform which I published last September on the Australian Public Law blog (, I noted that the current system has three sources of illegitimacy, which can be summarised as follows.

    • True and false preferences – “…ballot papers these days are largely clogged up with obscure candidates, sometimes running for parties whose beliefs, as professed in their names, cannot necessarily be taken at face value. Many of these candidates scarcely campaign, leaving voters with no basis for assessing their relative merits. Faced with such ballot papers, voters who wish to specify their own preferences rather than adopting a ticket as their vote have no option but to lie, writing random or otherwise meaningless numbers against candidates once they have run out of genuine preferences to express. For voters who wish to use the ticket voting option, the proposition that they have consciously adopted as their own personal preferences for each candidate those contained in a ticket is little more than a legal fiction.”

    • Discrimination against some voters – “…the current system … discriminates against some voters, by making the act of voting comparatively difficult for them. This is so, at least, for any voter who wishes to cast a first preference vote for a candidate who does not appear in the first position on a ticket; such voters are required to number every square on the ballot paper. … One can give as an example here the case of a voter who wishes to vote for all the female candidates before all the male ones.”

    • Unmanageable ballot papers – “At the 2013 election, the Senate ballot paper in New South Wales met a particular benchmark of absurdity: magnifying sheets had to be provided at polling places to enable voters to read the print, which had been reduced in size to enable all the candidates and groups to be shown.”

    I noted in my article that “The outstanding feature of the JSCEM’s main proposal for reform is that it addresses (to a greater or lesser extent) every one of these fundamental sources of illegitimacy.”.

    The same cannot be said of the proposal in the latest Bill. By doing away with preference harvesting it should go some way towards dealing with the problem of unmanageable ballots, but by continuing to require below the line voters prima facie to number every square, and by offering them nothing by way of relief from that other than a slight relaxing of the current savings provisions (the benefits of which do not appear to have been modelled), the problems of false preferences and of discrimination have been left untouched.

    These points can be drawn together into one overriding point: the scheme proposed in the Bill will be an incoherent one, with no clear underlying principles apparent. The current system at least makes sense, in that it prima facie requires full preferential voting both above the line and below, with allowance only for mistakes. The Committee’s proposal of optional preferential both above and below the line also makes sense. The Bill’s proposed, for optional preferential voting above the line but full preferential voting below the line (again with some allowance for mistakes), makes no sense.

    Finally, there is a major practical issue to be considered. Under the proposals in the Bill – and this hasn’t been widely remarked – there will be no counting of Senate votes to candidates or groups on election night, nor after they have been sent back to AEC divisional offices: they will only be counted at the central Senate Scrutiny Centres for each State and Territory. Now assume for the purposes of argument that a manual scrutiny may still be on the cards. The formality checking of below the line votes was looking difficult enough even under the scenario of optional preferential voting below the line with at least 12 squares to be marked in a double dissolution election, with more perhaps marked by some voters. It would be beyond horrendous if:

    • the bulk of below the line voters had followed the instructions and numbered all the squares;

    • five mistakes were permitted rather than three; and

    • all of the counting was being done centrally.

    There is not a single person in this country who has ever done a manual formality check on a Senate ballot paper with 110 + candidates on it using the current 90% rule, let alone with 5 squares.

    Under the proposal in the Bill that all Senate counting be done centrally, it is anyone’s guess how long it will be after the election that anyone gets even the vaguest indication of what the Senate result might look like – and that’s going to be true whether or not the count is computerised. For better or for worse, the AEC has of the order of 70,000 staff on hand to do an initial count on election night. There is no way known that a comparable number of people could be deployed thereafter and supervised effectively under just eight Central Senate Scrutiny locations.

  12. Have they left the party registration requirements unchanged – ie deposit required and minimum number of members? If so, that’s rather disappointing. It would mean that new parties would have to show significant community support to be established rather than someone just ringing around a few mates. I doubt the restriction on being a registered officer will stop the Leyonhelms of the world from establishing front parties. They will just get a few different people to be the registered officers but still have the same microscopic membership.

  13. I have some possibly useful data on the number of errors in BTL ballots. I have a spreadsheet with all the BTL ballots from the recount of the 2013 Senate Election in WA. The data does not include anything from informal ballots, but I was able to run some analysis on the number of errors in the formal ballots. There were 62 candidates in that election, so a maximum of nine numbering errors were permissible – three corrections would get you back to 90% correct.

    Of the 50,131 formal BTL ballots, 6,620 or 13.2% contain some sort of numbering error. The number of errors was distributed as follows:

    1 error – 4,076

    2 errors – 1,947

    3 errors – 384

    4 errors – 121

    5 errors – 45

    6 errors – 33

    7 errors – 11

    8 errors – 2

    9 errors – 1

    Obviously the number of errors is trailing off. It is conceivable that the new rules (allowing a maximum of eleven errors) may have saved a further ballot or two, or maybe none.

    The outcome may be slightly different for the states with bigger or smaller ballots, but I’d be surprised if the changes save more than dozen or so votes nationwide.

  14. Has anyone done any analysis on the proportion of BTLs that cross party lines vs those that could be reduced to preferences for parties?

  15. In response to Kevin Bonham’s question about cross-party preferences, I’ve dived back into my BTL dataset from the 2013 WA Senate count. In addition to cross-party preferences, I also looked at the number of ballots that rank candidates within one or more groups in an order different to that in which they were listed, and ballots cast for ungrouped candidates (see also final note below), as these also cannot be replicated by preferencing at the party level.

    I limited the analysis to the 43,511 ballots that contained a full 1 to 62 sequence. Of these:

    – 350 ballots (0.8%) cast for the sole ungrouped One Nation candidate

    – 21,592 ballots (49.6%) had candidates of one or more groups ranked in something other than the listed order

    – 13,530 ballots (31.1%) had cross-party preferences

    – 9,764 ballots (22.4%) are in two of the categories above, and 115 (0.2%) are in all three and therefore double or triple-counted

    This leaves 18.033 ballots (41.4%) that did not fit into any of the categories.

    Some other notes and qualifiers include:

    – The 21,592 ballots with candidates within parties ranked in a different order included 6,909 first preferences for candidates that were not at the top of their party ticket. Re-including the ballots that did not contain a full 1-to-62 sequence there were a total of 7,916 ballots (15.8% of all BTL votes) cast for down-the-ticket candidates. This was boosted in WA by a controversial #1 candidate for the ALP, leading to 2,198 votes for the second candidate.

    – The 13,530 ballots that included some cross-party preferences included 4,599 where the first and second preferences were for different parties (this excludes first preferences for the sole One Nation candidate). Among these were 1,393 such votes for a popular former footballer who was the #1 candidate for the National Party.

    Notwithstanding minor contributions from the above unique factors, it appears likely that over half of BTL voters like to vote in a way that is not supported by ATL preferences.

    However it is also worth noting that ATL preference do not allow a preference to be allocated to an ungrouped candidate. As such, the only BTL votes that can be 100% replicated by ATL preferences are those in elections with no ungrouped candidates, or when there is a single such candidate that the voter wished to rank last.

  16. As a change this is pretty good probably a B or B-

    They relatively well cut the gordian knot that is moving from just 1 above the line to 1,2 etc
    Probably they have done this better than NSW at this.

    The below the line is disappointing, but as someone that always votes below the line I learnt the hard way you probably need to go way past 15 (in NSW) to ensure that you have a valid vote that cant further be of use to someone that you approve. The number of people voting below the line will be low regardless

    The no senate count is dumb, I hope this eventually goes.

  17. I think I goofed on my earlier analysis of the number of errors in BTL ballots. The data was correct, but my interpretation of the legislation may have been wrong.

    On re-reading, I think the three permitted error-corrections need to get you to a correct 1 to 56 sequence (for a 62-candidate ballot) rather than to just any 56 of 62. So while it is still possible to have nine “errors” (three in the 1-56 range and then miss the last six numbers), it is the three that is the important number.

    Re-running the analysis of the 6,620 mis-numbered ballots, the number of errors in the 1-56 range is:

    0 errors – 2,559

    1 error – 3,102

    2 errors – 809

    3 errors – 150.

    The zero error cases are where the only errors or missing numbers are int he 57-62 range.

    The outcome suggests that the savings provision may save more votes than the previous analysis suggested. While the number of errors is still trailing off, it’s conceivable that there could be tens of votes with 4-5 relevant errors in WA and maybe even a hundred or two nationally, especially with the bigger ballots in some states.

  18. What will be shown on how-to-vote cards?

    Will they recommend party preferences. If so, it is little different to the current system, just the the three major parties will have all the clout.

    Observers might recall that the greens started as 5 independents, steadfastly opposed to Parties!

    The proposed system strengthens parties over individuals.

  19. I would hope parties recommend preferences. That is completely different to the current system because:
    -The voter sees the preferences.
    -The voter can choose to opt out and mark their own preferences if they disagree.
    -It doesn’t create the bizarre situation where every single preference from a group flows to the same person, so early exclusions can make a massive difference later on.

    You think the current system doesn’t favour parties to an extreme degree??

  20. But unless you vote below the line, you can only give preferences to parties, and its the party bosses who dtermine the order on the party ticket. In Tasmania Lisa Singh, dropped to number four won’t have a chance.

    It is essentially a party list system.

  21. How does that differ from the current system? You say it *strengthens* which suggests something has changed.

    In practice the power of primary votes above the line is so strong that no candidate can overcome their party’s ticket.

  22. >>It is essentially a party list system.

    Well that works well in a lot of countries including NZ and Germany.

    If you want to have a say in the order of candidates on a list, join a party and campaign for open pre-selection ballots. The GRN offer this to members most of the time.

  23. The GRN offer this to members most of the time.

    Untill and unless it becomes inconvenient, or a rogue gets through. They are no different to the rest.

    In any event you still can’t pick and choose across parties, unless you go below the line, as current.

    Why not a straight preferential vote for 6 or 12 preferences. Almost everyone can count to twelve. No list voting; just voting for the candidates.

    Tasmania’s Hare Clark is the model.

  24. I have endeavoured to encourage Andrew Wilkie to propose an amendment to do away with ‘the line’ as it won’t be needed anymore and will only serve to reinforce ticket voting. Because of the numbers it can only be a token effort, but I have argued that it should be made. He will need a seconder.

Comments are closed.