Let’s make a deal

Today’s Herald-Sun reports that Labor and the Greens are on the brink of closing a deal in which Labor will get Greens preferences in sensitive lower house seats, and the Greens will get Labor preferences in the upper house. In reality, the former part of the bargain is of little consequence: it is well established that the Greens lack the power to influence the preference decisions of more than a handful of their supporters. Furthermore, Rick Wallace of The Australian notes that "how-to-vote cards for the upper house have to be submitted several days before those for the lower house", making it possible for minor parties to "strike a deal to get what they want from Labor and Liberal in the upper house, then rat on them when it comes to the lower house". That possibility aside, the deal seems like sound tactics on Labor’s part. Upper house seats decided on Labor preferences are now certain to go to the Greens rather than Family First, preventing a repeat of Steve Fielding’s Senate win in 2004. The cost to Labor is that Family First will surely not be putting them ahead of the Liberals like they did in 2004: Labor’s brains trust presumably had reason to think this was not going to happen in any case. It was earlier reported that Labor was considering a separate preference swap with Family First in eastern suburbs marginals, which could potentially have decided very close outcomes in Labor’s favour at no meaningful cost. However, this idea is said to have been scuttled due to "rank-and-file unrest".

The other ball in play is the possibility of a Labor-Liberal deal at the expense of the Greens and the Nationals. Unlike minor party supporters, Labor and Liberal voters by and large follow the how-to-vote card, so the significance of their preference allocations is not limited to the upper house. This respectively gives them power to swing the result in Liberal-versus-Nationals and Labor-versus-Greens contests, leading to talk that Labor might put the Nationals last in exchange for Liberal preferences in lower house seats vulnerable to the Greens. On the one hand, such a deal would put the Nationals’ seven lower house seats at risk, particularly Rodney and Shepparton; on the other, it would terminate the Greens threat to Labor in Melbourne, Richmond, Northcote and Brunswick. This prospect has generated much excitement in the media, with Paul Austin of The Age reporting that "senior Nationals (have) privately vowed to destroy any chance the Liberals had of victory" if it goes ahead. Specifically, they have threatened to direct preferences to Labor in former Liberal leader Denis Napthine’s seat of South-West Coast, won by less than 1 per cent in 2002. However, Ted Baillieu says he has "made it very clear" that "we are giving our preferences to the National Party and they are giving their preferences to us".

UPDATE: Some comments worth relating from Brian Costar of the Swinburne University of Technology in this week’s issue of the Weekly Times, concerning the possibility of a Labor-Liberal preference deal: "I don’t believe it will happen for one minute. Last time, (prior to the 2002 election) in the pre-poll period, Labor gave its preferences to the Liberals … but when it got serious they just gave their preferences to the Nats". Costar is quoted as saying the source of the story was likely to have been a "strategic leak" from the Labor camp, "designed to send a message to the Greens ‘not to get too smart’ as well as to spread discord between the Nationals and Liberals".

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.

22 comments on “Let’s make a deal”

  1. Hi William – Love your guide. But I can’t see how a Lib-ALP preference deal would put Benalla at risk: Labor had 41.3% primaries last time, so it’d take a very big swing for their preferences to come into play.

  2. William. The Greens can not direct their members in the lower house. Green voters will act independently from any recommendation. I would also think that the deal struck between Labor and the greens the upper-house will ensure that Family first will not climb out of the pile as they did in 2004. Any deal between Labor and the Liberals on lower-house votes will also be received favourably. Again I do not think the Greens will receive favourable preference flows as all the smaller parties want to have a chance and they will trip over each other. My prediction is Greens have zero to no chance in the lower-house. The Greens can direct preferences in the upper-house only by defalut and the fact that many will vote above-the-line. I don’t think any party will rat on a preference deal. It is bad politics and will rebound in teh long run to back off from any frim undertaking. Should teh unthinkable happen it is possible for the parties to run an underground campaign urgeiung voetrs to “vote 5 below the line”.

    Most voters do not trust the Greens they are an unknown factor and little pulic exposer or scrutiny has taken place on their policy and organisational structure.

    As the campiagn progresses, and yes its still early, I still am of the view that the Green vote will not be as strong as the 2002 results and that the Northern Metro region is their best chance The next possible seat is teh Southern Metro but that will depend on preferences as would any other seat depending ion the Liberal parties showing. (Upper-house ALP 18-19 Liberals 17-21 Greens 2-3)

    As Steve bracks was quoted as saying the ALP sees the Nationals and the Liberal Party as one and the same. They will not lose votes as a result of playing the National party last. The Nats threats of armageddon are like going to war with a feather duster.

    re Liberals/Green votes. Many Liberal supporters see the Greens Asa bigger threat to business and pressure again the Liberals will not suffer Asa result. As to the possibility of buying lower-house preferences again I state the Greens can not direct their vote.

    What is interesting is that the DLP has not yet to my knowledge nominated lower-house candidates. The DLP polled reasonably well in the senate election and it was the DLP and Democrat vote that saw family First survive the count. People Power will suck up most of the dissipated democrat vote and the disillusioned. information I am getting from the workers on the ground is the Greens are not winning hearts and minds.

    The Other big factor that is still to be played is the issue of the Howard Liberal Government have absolute power in Canberra. One can expect that the ALP will run hard in the last week on the issue of re-elect Labor to keep Howard under check. This may persuade 2-3% of the vote who feel that some checks and balances are necessary. What is interesting from news coming out of the US mid term exit polls is that national issues and not local issues are dominating the US mid term campaign. I think there is a swag of voters out there that see benefits from having a Labor state Government at a time when the Liberal Party control both the House of reps and the senate. With GWB suffering in the polls voters Howard’s policies and support will also suffer. It only takes a few percentage points to effect the outcome and every issue helps.

  3. It still wouldn’t surprise me if Family First got ALP preferences in one Upperhouse seat. It’s the pragmatic decision vs idealistic.

    In the federal election they will definitely do a deal with FF. In the victorian election its different because it’s alot easier to be idealistic when you have no chance of losing.

  4. The thing is if the Libs direct preferences to the Greens then because most major parties` voters follow the how to vote cards, the question will be whether the strays are enough to effect the outcome. If Labor get in the low 40`s then the strays will probably not be enough.

    If the Nats went against the Libs and their vote and preferences held and any anti Labor swing was not big then Polwarth could be won by Labor (margin 9.?, Nats 15%)
    and if the Nats could if the they were standing in all non-Metro seats undermine the libs in the suburban fringe seats.

  5. Charles, I was actually recycling a media report with Benalla, which I haven’t really looked at myself. I’ve removed it from the list. Geoff, this article in The Australian says the Nationals “threatened to rip up their traditional preference swap with the Liberals” in SW Coast and Polwarth. Unless I’m reading it wrong – perhaps they’re threatening to direct preferences to someone else who might conceivably win the seat? I can’t imagine who though.

  6. Hi I just received feedback from teh VEC re: publciation of data-file feeds. Extract below.

    What concerns me is that there is no plans to provide polling place above-the-line voptes for teh upper-house. One would think it would be advantageous and helpful in teh scrutiy of the ballot if there was a preliminary tabulation of above-the-line votes on the night. It could be that as a result of teh above the line tabulation tehre would be no need to distribute below the line votes. (In theory at least). Given that Above the line votes will not be data-entered a manual tabulation is by far quicker and that this, as in teh federal Elections, is best done one the noight while polling staff are available.

    I ahve request access to this information and will with Willioams permission republish here when available

    — Extract from reply received —

    The VEC are providing election night results data in a similar format to the AEC for the State election…

    These files will be XML formatted and closely follow the work the AEC has undertaken in formulating an XML standard for election data. The VEC will be making some deviations within the data from the AEC standard, let me know if you want further information about this as I am able to provide you with the AEC specification documents?

    There are a number of options as to which data you might want to use, if you select downloading either the Lite or LitePP files you would need to set up your system with the load files first. If you want to use the verbose file each time this will contain all the info you need. The order of items and a quick description is:

    Item 1 – first load file (State election)
    This file contains the voting centres to be used on election day 25 November relative to their Districts and needs to be loaded into your own application. This is available now if you want it?

    Item 3 – second load file
    The next load file we will provide will be the candidate file. This will be available on the evening of Friday 10th November after close of nominations and once ballot draws have been completed. We will provide a link on our website where you will be able to download it from. I will email you all once the link and data are available.

    Item 4 – third load file
    On Sunday the 12th November I will provide you with the final load file. This will contain the group voting tickets information (it will include up to 3 ticket elements for each group and will have full below the line preferences against each candidate for each ticket). The file will contain each group, candidates within each group plus each ticket registered by the group. We anticipate this will be late afternoon due to the fact that ticket registration does not close until 12 noon.

    Each load file is being provided so that you can preset your systems to accept the lite files that will be generated on election night and beyond. Because of this we will provide you with a generation of the lite files from the 12th, obviously there will be no results in the data file but it will let you test your systems to make sure they can accept the files.

    Item 5 – Election Night
    On election night we will generate 3 results files:
    File 1 – Verbose file. This file contains everything about election results, candidates, tickets, voting centres and the results. If you choose to use the verbose file you don’t need the load files. Each time the file is generated it will contain all the information as mentioned above.

    File 2 – Lite File. This file contains identifiers that link back to each part of the load files. There is no detail in this file i.e. candidates names or results by voting centre etc. This is the smallest version of file we provide.

    File 3 – Lite PP file. This file is much the same as the lite file but it has separate fields for each voting centre for each candidate. It also has total votes counted in each voting centre for each candidate. There is more information on the LitePP file because of the breakdown of information by voting centre.

    Note: There will be no results by voting centre for the Upper House.

    I hope you understand the above, please feel free to get back to me if you require any further clarification.

  7. For those who’ve been wondering why I’ve not been writing much on the election, read the last post by MelbCity. I’ve been too busy re-designing the ABC’s data receive programs so we can run an election night computer system. Writing on the election as well is too difficult.

    I disagree with you MelbCity. We are having more than enough trouble negotiating the way we receive lower house votes to start changing the upper house format as well. We are having enough dispute arguing over what is (or is not) counted on the night, and what is (or is not) an election night counting centre.

    No electoral office has ever done a count of ticket votes on the night. No electoral office has ever provided booth by booth counts for the upper house in the media feed. Precious few provide booth counts for the lower house, and the need for them has been the basis for a dispute between the television networks for several years.

    The AEC only provides Senate totals by group on the night. Victoria is the first state ever to do candidate counts on the night for the upper house. They of course have paper work covering the counts in each booth. What they are not doing is going to the effort of capturing all that paperwork into their election night reporting service.

  8. Antony I do not think we are in disagreement. The AEC most certainly did provide a preliminary senate distribution of the night. A manual count of above-the-line is the most efficient means possible and this is best done utilizing Voting Center Electoral office staff engaged in the conduct of the election. Victoria has seen Millions of dollars wasted in the development of duplicated voting system. It would have been far preferable and less cost to the public purse of the VEC and AEC co=ordinated their system development efforts. but then governments are not that efficient.

    Agreements are in place for the distribution of election night data. I am still to see the outcome of this effort. Although I question aspects of the model being put in place (Never could understand why postal votes are not considered as just another voting centre) What is clear is that there is a strong need for open and transparency in the voting/tabulation system. It is interesting to watch issues surrounding the introduction of electronic voting systems in the USA.

    I understand that the VEC wants to use its new multi-millions dollar vote tabulation system (As used in Local Government elections) but the truth remains that in many cases it is quicker and more transparent to undertake a manual count in the lower-house seats.

  9. The AEC provided a preliminary distribution of Senate preferences? Garbage. They hadn’t even entered the group ticket votes into their counting system. It was several days before they had differentiated between below and above the line votes. They did nothing of the sort you suggested on election night. If they somehow did that for you in Victoria, it must have been a personal deal given the number of party people who rang me on the Monday after my preference distributions.

    I agree with you entirely on the postal votes. It is currently responsible for the biggest problem we are having with the disatribution of result and a tripling of the size of the XML results file.

  10. I have just received the following email from the VEC which is slightly different from the advice given earlier.

    This indicates that it should be possible in theory to know the outcome of the Upper-house on the night. Which is good news. It may be that the outcome of the manual count will result in not having to undertake a below-the-line distribution of preferences.

    — Extract —

    In regards to counting on election day we will be counting above the line and below the line 1st prefs at each voting centre. You are also correct regarding exhaustion of ballot papers, as long as there is at least 1 – 5 the prefs can stop anywhere after and will exhaust. Also if there are duplicates after 5 they will exhaust as well.

  11. Antony. Sorry but extracted the senate votes from the AEC site on the night and the days following as they were published. I also scrutineered in one booth. I can not comment on the distribution of preferences but they certainly did a manual distribution based on above-the-line an first preferences at the polling place. Again I scrutinised the count.

  12. Maybe it was only done in Vicctoria. No hold it I collected results on the night from the AEC web site. Unless they made up the results I think you are mitaken. yes prefernce distribution I would agree. preliminary above-the-line and first prefernecs were available on the night as the results were phoned in. I can provide you with dated copies of my emails showing the results if you still are unsure.

  13. I rthink tregh mis-understanding here is that I am refering to above-the-line and first preference votes which were presorted and tabulated at the polling place on the night. This is what I refer to as a preliminary distribution. Detailed prefernec distributiion did not occure until mach later but as moast people voted above the line the results were pretty much detwermined on the night. It was interesting to rewad back on my emails though. I found one email I sent to Crikey expressing concern that teh Greens would not get over the line and that Family First could win a set. This was also mentioned in corresspondence by others ut Crikey dism,issed it at the time. We tried telling them. the Grens needed over 8.5% and the ALP over 39% fall below and they missed out. (As it turned out thats what happend)

  14. So the AEC didn’t issue a distribution of preferences? You are completely confusing me by your use of the word ‘distribution’. You said the AEC issued a Senate distribution, which I now think I understand you mean as being a break-down between above and below the line preferences.

    I can tell you exactly what the AEC did in 2004. On the night in polling places, they reported a simple tally of votes for each group. They may have differentiated between above and below the line votes in each booth, but the election night report was a total only.

    The AEC then went into post-election mode on the Sunday. Returning officers broke all Senate ballots into above and below the line votes. ATL’s were tallied and entered into their computer system as a total, the BTL’s batched and sent to the data entry centre.

    To avoid the problem at earlier elections where the Senate count went back to zero at this point, or to just a tally of ticket votes, the election night tallies were retained as group totals. In the post election period, the AEC results system had a total of ticket votes, a total for each candidate, and a total called ‘Unapportioned below the line’ which represented the difference between the election night count and the tally of data entered votes, effectively being an estimate of the number of below the line votes still to be data entered.

    On election night you will find the AEC just gave a total by group, and that was certainly the manner in which totals were provided in the electronic feed to the media. The post election broke all the Senate totals into ATL, candidate and ‘unapportioned BTL’ totals.

    Now I understand that what you are referring to is the VEC not providing separate ATL totals. That can matter. For instance, in Tasmania in 2004, it was obvious once you knew the number of ATL’s that the ticket vote result which elected Family First was not going to occur. The exhausted rate below the line may matter in Victoria. But I’m not sure it’s worth going on the war path about. Leaves a bit of room for punditry you know.

  15. Antony. we are in agreement on this one. The issue of transparency will be important as Victoria moves towards a computerised count. There are also isues related to the formula and segmentation. Depending on the fold up of the count some votes will increase in value as the count progresses.


    The pinciple of one vote one value does not exist under the system of proportional representation adopted. It will be made worst if people vote 5 only below the line. The new 5 preference rule is an issue of concern.

    If you want your vote to count preference all candidates in order of your choice.

    Too late to change for this election but it MUST be reviewed along with the segmentaion in the counting of excluded candidates.

  16. I support optional preferential voting. It has worked just fine in Tasmania for a century. It assists in undermining the engineering of election results created by group ticket voting. Full preferential voting discourages people from voting below the line. However, I think Victoria might consider adopting the surplus calculation method used in the ACT and NSW which excludes the exhausted preferences before calculating the transfer value.

  17. There is no justification of having five as a minimum if you implement optional preferential voting then why not have one as the minimum why stop at five or the number of vacancies.

    The other issue of concern in relation to optional voting is if it is allowed below the line should parties be required to express preferences beyond the minimum permitted? If the answer is no then you end up with a defacto first-past-the-post system. Yes to optional below the line but registered parties above-the-line-vote should be required to nominate all preferences…

    Issue 2.

    Problem with the ACT and NSW surplus is they both use a formula based on the number of ballot papers as opposed to the value of a ballot paper. Under this formula some ballot papers increase in value at the expense of others. ((Ctv-Q)=S) / P
    When it should be ((Ctv-Q)=S) / Ctv * Vv

    Ctv =Candidates Total value
    Q = Quota
    Vv = Value of the vote (Ballot paper)
    S= Surplus
    P= Number of ballot Papers

    With a computer aided counting there is no justification to use a number of ballot paper based formula which distorts the principle of one-vote-one-value.

    One transaction per candidate no segmentation in distributing excluded candidates preferences also is another issue. If segmentation is to be adopted then it should be on a FIFO basis and not the current system of aggregation of vote values. This is very technical but as I have tried to explain it can effect`the result under certain circumstances. So now we have computer aided technology we might as well get the principle right

    More information http://melbournecitycouncil.blogspot.com/2006/10/one-vote-one-value-calculating-transfer.html

    The correct formula should be

  18. Your wrong about NSW and the ACT. Both use the Gregory or ‘last parcel’ method of calculating the surplus, as does Tasmania. All votes transferred to a candidate arrive at the same value and cannot increase in value if distributed as a surplus. In the case of the ACT and NSW, votes that exhaust with a candidate stay with the elected candidate.

    I have to say I am totally confused by your explanation, in part because I am at a loss to know what you mean when you keep talking about segmentation, a term I’m not familiar with.

    What you are referring to is the problem of the Senate system, used in all state upper houses, which is an ‘Inclusive Gregory’ method. In calculating the surplus and transfer value, all votes are looked at, not just the last bundle. The problem with it is well described by McAllister and Farrell in the Australian Journal of Political Science two years ago. The Inclusive Gregory method has the possibility of a vote increasing in value during the count. McAllister and Farrell well outlined an example of where such a change in weight completely distorts the count.

    As this theoretical problem actually occured in WA in 2001, a proposal to fix it was included in electoral amendements in 2003. The WAEC published a substantial paper on the subject, and the proposal to change the system is still in the pipeline. (It did not pass at the time for other reasons to do with the bill) Once one electoral commission does it, the rest will fall into line.

    Choosing one or five preferences is entirely a matter of trading off principles. The ACT allows 1, the Tasmanians use the number of vacancies. The only advantage of the number of vacancies is in the case of a very popular independent, it avoids the proportionality of the system being distorted by a high rate of exhausted preferences.

  19. Anthony & Antony

    The above is very interesting, where does one get the WAEC paper?

    For the record, the XML file format for Vic was finalised yesterday (15th), there were some minorr changes:
    – Names of candidates in the TwoCandidatePreferred section for
    the lower house in the PollingPlaces section.

    – Breakdown of TwoCandidatePreferred votes in the lower house
    PollingPlaces section into the 5 categories: Ordinary etc.

    – Names of ungrouped upper house candidates.

    And definitely no PP counts for the LC

    On what is counted when, it was always my impression that only the ATL was reported from the booths on election night. Sometimes the #1 BTL votes were counted, but they were seldom included in the total. Certainly TENIS and the NTR Board guides said only ATLs appeared live. The NSW SEO uses the terms RATL and RBTL for Local Council elections, the R standing for “Random” (only NSW can have RATL?). My friendly DRO confirmed she was counting only ATLs at both NSW Leg Council, and Local Govt. elections.

    On Gregory, etc., it has sometimes seemed that the people driving the count, particularly since it became computerised, are not always sure. This was true for Vic Senate in 2001, when whether or not it was a Gregory method made the difference between the Greens getting up and not getting up, based on the way DLP preferences had their Transfer Value calculated. When we were trying to simulate this, we got the message “it IS Gregory, err… no it’s not…. err yes it is….)

    Geoff Lambert

Leave a Reply

Your email address will not be published. Required fields are marked *