Wednesday, December 12, 2018

Group Ticket Voting Wrecks 2018 Victorian Upper House Election

The buttons have been pressed on the Victorian upper house election.  In the end, none of the results were all that close, and all regions have been declared.  If anyone can find a legal basis for challenging the results, they will now have to do so in court.  On that, it would be nice if professional preference harvesting could be deemed to be a bribery offence under Section 151 (3) (d) but I  suspect that it doesn't work like that, and that that section is aimed at bribery connected with how-to-vote cards.  I can only assume what has happened is all legal, but history should record it as another upper house election that was trashed by Group Ticket voting. 

I should add that this post is not intended as an attack on the calibre of those elected to represent parties with small vote shares.  They may turn out to be excellent MPs.  Rather the point is that they were not elected by a proper electoral system and those elected on very small vote shares do not have a proper mandate.



The total seats won are as follows:

Labor 18, Liberal 10, National 1, Greens 1, Derryn Hinch Justice Party 3, Liberal Democrats 2, Transport Matters 1, Sustainable Australia 1, Shooters Fishers + Farmers 1, Animal Justice Party 1, Fiona Patten's Reason Party 1.

Labor should have a cushy time of it, being able to pass anything supported by all the DHJP MPs (assuming that they stay together as a party!) and having various routes for other legislation: left-wing through AJP, FPRP, Greens and maybe others; while also less green paths if needed on environmental matters through the Liberal Democrats and Shooters.  This is to be expected given the magnitude of their Lower House triumph, and it would be foolish to object to a party that won the Lower House so decisively having close to a working majority upstairs.  Nearly all Labor's Upper House wins came on raw quota, with just a couple where they needed a little hand up after polling most of what they needed to win.

Where the results of this election, as expected, go pearshape is in the results for the minor parties.  This election has doubled the representation for parties other than Labor, the Coalition and Greens from five seats to ten.  For the first time in Victoria we have seen parties able to preference-harvest their way to victory even from below 1% of the vote, "beating" parties with as much as fourteen and a half times their primary vote off the back of luck and backroom preference deals.

Below the line voting increased from 6.08% to 8.87%, but it had only minor impacts on the outcomes.  Compared to if all votes had been cast above the line for the same parties, there were only three changes.  The Liberals won in Northern Metro in a situation where the ABC calculator modelled them as losing by 1021 votes, a margin that even a trivial rate of below the line voting would have overturned.  In Eastern Victoria, a preference spiral involving the Aussie Battler Party (1.22%) fell over and the somewhat deserving Shooters, Fishers and Farmers (5.01%) retained the seat instead.  Finally in South-Eastern Metro, below the line votes altered which party spiralled to victory almost entirely relying on above-the-line votes, but not the fact of it happening.  Indeed,  Transport Matters (1.27%) were replaced by the even lowlier Liberal Democrats (0.84%).

The well-oiled LDP troll machine - well one of them at least - has been claiming that the LDP won the seat on below-the-line votes.  That's not true; had all votes been below the line the LDP would have been excluded in South-Eastern Metro at a very early stage. It's more accurate to say just that BTLs caused a rival party to lose it.  What is notable about David Limbrick's win is it is the first time the Liberal Democrats have won a seat at a GTV election without being assisted by the name confusion that continues to result (even with logos printed on ballot papers) when the party draws to the left of the Liberals on the ballot.

Results Compared To Vote Share

In the table below I compare the seats won by each party to the vote share each party achieved.  I use two indicators for this:

* share of the total state vote
* average share of the vote per region

This follows the same approach I used for analysing Senate voting reform.  The reason for it is the number of voters per region varies a fair bit (though nowhere near as much as in the Senate!) In this case using both figures doesn't make much difference, but I do it anyway.

The new Senate system provided a stunningly proportional result when weighted by state in 2016 (bearing in mind the Senate's state-based malapportionment.)  The 2018 Victorian Upper House election, however, produced severely disproportionate results:


(See a nicer table visualisation of these numbers thanks to @Gergyl.)

Proportionally by primary vote, the major parties got pretty close to what they deserved.  The Greens, however, got only one seat whereas four would have been proportional.  Only one smaller party that would have deserved a seat under a statewide proportional system by primary vote (Labour DLP) got nothing, while two that would not have done so (Sustainable Australia and Transport Matters) won seats, and two others (Liberal Democrats and DHJP) got a seat more than they deserved.

In analysing the Senate system I have pointed out that primary votes aren't the be-all-and-end-all of proportionality in Australia and that if a party (like One Nation in 2016) punches above its weight because the voters for likeminded micro-parties gave it lots of preferences, that's fine.  But in this case the Greens' underperformance doesn't have anything to do with what voters from other parties thought of them, and has everything to do with the deals done by other parties to exchange above the line preferences for pragmatic advantage.

The Fate Of The Greens

The Greens were not involved in the Glenn Druery preference deals to my knowledge, but still produced some rather odd preference orderings that may in cases have had something to do with trying to get Lower House preferences.  They also nearly cost Fiona Patten her seat by preferencing DHJP above her for reasons that remain unclear, though few Greens supporters would have chosen this order based on Senate distributions (in 2016, nearly six times more Victorian Green voters preferenced the Sex Party second as preferenced Hinch second).  To what extent the Greens were trying to play the preference game themselves is not too clear, but however hard they tried their attempts would have been doomed to failure.

The reason for this is that the Greens were an unpromising partner for smaller parties expecting to poll a low percent of the vote and seeking to swap preferences.  A smaller party would only get Greens preferences if it happened to snowball its way ahead of the Greens in the count.  But the Greens were excluded either late in the counts or not at all, meaning that the chance of a given party ever getting ahead of them to get their preferences would be slim.  A micro-party would be better off swapping its leading preferences with other micro-parties that it had higher chances of outlasting.  This became even more the case once there was already a group of micro-parties networking (through Druery). 

As a result, the Greens were disadvantaged by their expected vote being too high to make them a good partner for preference swapping, but generally not high enough to put them over the line on raw quotas.  In 2014 this hadn't counted against them in most seats - the smaller party mix in 2014 was more ideological and less tightly networked on preferences, and the combined vote for such parties was slightly lower (19.7%).  It's also worth noting that while the Greens got 12.5% of seats off 10.75% of primary votes in 2014, they did poorly relative to their primary vote in 2006 (7.5% off 10.58%) and 2010 (7.5% off 12.01%).

Some Labor partisans have been very keen to obfuscate about what has actually happened here.  Many have pointed to the Green vote dropping from what the Greens polled in 2014, as if that's the main cause of the result.  It's a part of the cause, but not the main part.  Firstly under a Senate-style system, the Greens would easily have won seats in North Metro, South Metro and East Metro.  In most of the remaining divisions they would have led various centre-to-right minor parties and they would have been likely to hold on to at least one of those leads (perhaps more but it's difficult to tell).  Secondly, had the Greens polled their 2014 vote again at this election, they would have still only won two seats (North and South Metro).  So a fair apportioning of the blame for the Greens' drop from five to one is one deserved seat loss based on actual drop in vote share, and three undeserved ones caused by Group Voting Tickets.

Greens sympathisers in turn have been keen to blame Labor for preferencing "Druery parties".  That doesn't seem to be a significant cause either, though not for want of trying on Labor's part.  Even in the case of the relatively close Sustainable Australia result, the Labor above-the-line vote was too reduced in value to change the outcome.  In many districts Labor preferences either weren't distributed while the Greens were in the count, or would not have saved them from their fate.

The major cause of the Greens' defeat was simply that the smaller parties had too many primary votes between them (22.1% statewide) and they more effectively exploited the Group Ticket Voting system to create artificial near-100% preference flows using #1 above-the-line votes.  These flows are fake because, as the Senate election showed, when voters are asked to choose preferences, their preferences tend to (i) scatter (ii) flow along ideological lines to a degree and (iii) favour well known parties over obscure ones.  For this reason, the argument that a quarter of the seats is a fair return for nearly a quarter of the votes does not hold up - it would if voters for all smaller parties preferred the other smaller parties to Labor, the Coalition and the Greens, but they do not.

Apologists for the indefensible argue that the Victorian system is OK because it allows the voter to escape from GTVs by voting 1-5 below the line.  It's true this is an improvement on the pre-2016 Senate system and that it better escapes the charge of discriminating against voters who want to order preferences their own ways.   But it creates a way in which a party can be systematically disadvantaged by getting several percent of the vote, in a way that a party getting less might not be, and this is unfair even if everybody voting 1 above the line for any other party knows exactly what they're doing and where their preference goes.  Which of course they don't.

Arbitrary Winners

Also, if the Victorian system was delivering seats to the smaller parties on merit, one would expect the highest-polling examples to win more often those further down the chain, and one would expect parties to win in districts where they actually did well.  But the first isn't the case at all, and the second isn't always either.  Ignoring all candidates elected on quota and also all major party candidates who later won, the winning smaller party (excludes Greens and Nats) candidates were variously 2nd (twice), 3rd, 4th (twice), 5th (twice), 7th, 10th and 12th out of all the parties (typically 18 in total) on primaries.  In the only case where one of these parties had the largest remainder by this method, it was out-snowballed by two others!  Thus the best performance outside the majors and the Greens (an impressive 7.85% for the Shooters in Northern Vic) came unfairly to nothing while we were treated to such complete absurdities as:

* Transport Matters (0.62%) beats Greens (9.01%) in Eastern Metro.

* Lib Dem (0.84% - the party's worst result in any region!) beats Liberal (12.3% over quota) in South-East Metro.

* Sustainable Australia (1.32%) beats Greens (13.46%) in South Metro.

Most of the smaller party wins were not so patently outrageous, but nonetheless wouldn't have happened had all voters chosen their own preferences.  Only a couple (Catherine Cumming (DHJP) in Western Metro and Jeff Bourman (Shooters) in Eastern Vic) were arguably wins on electoral merit.  Even those came unreasonably close to losing to parties on much lower vote shares.

A Fairer System?

I've previously pointed out that the parties in the previous parliament - yes even the Greens - are all to blame for this nonsense, since even those who supported reform failed to do enough to raise the issue and put pressure on the waverers.  It is understandable that Victoria's Electoral Matters committee chose to adopt a wait and see approach pending Senate reform, but there was ample time to refer the matter back and pass reform after concerns about Senate reform proved groundless in 2016.  By not doing so, the Victorian parliament has ignored the warnings of Antony Green and others that this day was coming, and has created a result that is not only incredibly silly but also should be seen as disquietingly unfair.  The system can not only systematically disadvantage a party that gets several percent in an environment in which the total micro-party vote is high, but is also self-entrenching because it keeps flooding the parliament with MPs who don't deserve to be there, who can't get re-elected without it, and who therefore mostly won't support reform.  As with systems with entrenched self-policing gerrymanders, the prospect for change appears bleak unless a need for bipartisan reform is accepted by both major parties.

Had this election been held under a Senate-style system, my rough estimate of the seats won would have been Labor 19, Coalition 14, Greens 4, Shooters, Fishers and Farmers 2, Hinch Justice 1.  The Shooters would probably benefit from the regional concentration of their vote, although one of their two seats would be uncertain anyway.  It is not that easy to eyeball results for seats where the Greens had a modest lead over smaller parties on primaries, and at some stage if it is useful and time permits I may run a simulation using Senate preference flows.  There's a case that such a result, while much fairer than the one that has occurred, would be a bit harsh on some of the stronger-polling smaller parties, and for those who agree, the ideal solution would be Senate-style reform augmented by having fewer regions with more members in each.  On the other hand, under such a system it's likely not as many micros would have run, and that this could have improved chances for the more substantially-supported smaller parties.  [EDIT: @sorceror43 has helpfully pointed out that 8 regions with 5 members is constitutionally entrenched in Victoria, so it would seem we are stuck with it.]

Overall though to prevent this from happening again, the overwhelming imperative must be to get rid of Group Ticket Voting, as has been done federally and in NSW and SA.  Whether the solution is exactly like the Senate system, more like NSW's, or something that encourages preferencing to a stronger degree than either, isn't such a major issue.  There is a small-scale Twitter campaign by a person unknown to me for "5 above or 5 below" in which the voter directs their preferences either above the line or below it with a minimum of 5 boxes in either case.  While a 1-5 below the line is likely to be less effective than 1-5 above, the slogan is an elegantly simple one for voters to follow, and I think the idea deserves thought.  At least subject to Senate-style ATL savings provisions.

Australia has a proud record of getting elections right.  We are the home of preferential voting, the first full federal election under which occurred 99 years ago tomorrow (with the first by-election 100 years ago on Friday).  In a world where some supposedly great but in fact badly flawed democracies are still in the nappies of first-past-the-post, we provide a shining light to places like Maine, which recently led the US by adopting preferential voting.  We should take our electoral leadership role seriously and get rid of the farce that is group ticket voting.

See Also

Tally Room - includes breakdowns based on primary vote under various methods for statewide proportional voting.

Inside Story Victorian election summary by Tim Colebatch (covers both houses)

William Bowe at Crikey

Adrian Beaumont

19 comments:

  1. Thanks Kevin - very useful (and I agree 100% with your conclusions). I'm not sure whether you can answer this question straight off or whether you'd need to do more analysis, but I'm wondering - in the cases where the BTL votes made a difference was it because they had a positive effect (actual preferences tending to favour the eventual winner) or a negative effect (exhausting, so preference simply not going to the beneficiary of the ATL deals)? Or a bit of both?

    ReplyDelete
    Replies
    1. In the cases of Hughes (Aussie Battler) and Khan (Transport Matters) losing, the candidates needed to get past cut-off points involving other parties that weren't actually in the hunt. Both of those candidates failed because the below-the-lines for their feeder parties sprayed all over the place instead of going in the same direction as the ATL preferences, which is basically your negative effect. It means that what the calculator modelled as a 100% preference flow might be only 90%. The eventual winners of those seats didn't need any positive effect themselves, they just needed the losers to get knocked out at an early stage.

      In the case of the Liberals in Northern Metro, below the lines had a net positive effect for them as they did eventually reach quota. But had this not been the case they would have been saved by negative effects against their rivals as well.

      It's also worth mentioning South Metro where negative effects against Sustainable Australia and positive effects in favour of the Greens made the race very much closer than the calculator had it. Sustainable Australia finished up over 15000 votes worse than the calculator forecast and the Greens finished up almost 6000 better, but their position was so good even this couldn't stop SA from winning.

      Delete
  2. Thanks Kevin. I infer from that that a goodly fraction of BTL voters went beyond the minimum of 5 preferences and gave further contingent preferences to candidates whom they didn't mind too much (as an intelligent voter would). Correct?

    ReplyDelete
    Replies
    1. I don't have any stats for 2018 in that regard. This is from the VEC's report for 2014:

      "Of those who voted below-the-line, more than half (52.67%) simply voted 1 to 5.
      One seventh (14.65%) numbered all the squares on the ballot paper.
      The remaining below-the-line voters finished at various points, with
      fairly small numbers for each finishing point (though several
      thousand voted 1 to 10, possibly for two groups on the ballot paper).
      There was some variation among the regions; broadly, voters in
      regions with a higher below-the-line voting rate were less inclined to
      stop at 5, and showed a slight tendency to go to the end. For
      instance, in Northern Metropolitan Region, 45.42% voted 1 to 5, and
      15.15% numbered all the squares on the largest ballot paper in the
      election."

      Delete
  3. Hey Kevin, should "a party that won the Upper House so decisively" (first para after the seat tally) actually read "Lower House"?

    Massive thank you to the time and effort you put into covering the ongoing count this election. Your work really was second to none!

    ReplyDelete
  4. Hi Kevin,

    Adrian Beaumont wrote a piece at the conversation on this as well. If you could compile a list of phesologists who've pointed out this result is a distortion of democracy, that would be a very useful thing next time a reporter talks to Tanya Plibersek, or a reader engages with someone in a comments section.

    ReplyDelete
  5. to change to a system similar to the senate will invite onp in.... I note not one onp mp in the Vic upper house

    ReplyDelete
    Replies
    1. There were no One Nation candidates.

      Delete
    2. Pauline is our (Qld's) poison. I'd be surprised if the rightwing whirlwind coalesces around her candidates a second time. It's not the voting system, but the voters that elected Pauline and her fair weather friends. Given the completely random outcome of this election I'm not sure that GVTs would be any protection from One Nation except perhaps in Queensland where they'd be just as cursed as the Greens.

      Delete
    3. A number of points here:

      1. One Nation are just another little party nowadays. The big parties might put up the cordon sanitaire to try to stop them winning under GVTs, but the micros don't, so the argument that using GVTs shuts them out is now completely obsolete. Had NSW 2013 been a double dissolution, Hanson would have won by preference-harvesting off 1.2%. We have no idea how many they would have won in 2016 with GVTs but certainly at least 1 and perhaps as many as 7.

      2. One Nation don't poll well in Victoria anyway. It was their worst state in 2016.

      3. The seats One Nation won in the Senate in 2016 were won on merit, either on primary vote or on superior preferencing performance compared to other parties in the hunt. We should be supporting systems that will make it possible for parties to get elected if their support level deserves it, no matter how odious we might find those parties to be.

      Delete
  6. Kevin is is worthwhile for you to do a post detailing the difference between upper house election systems in the varying states and the Senate? Perhaps when politicians have a sense of the strengths and weaknesses of each we may get some momentum for change.

    ReplyDelete
    Replies
    1. The politicians who sit on Electoral Matters type committees are usually well aware of the different systems so I'll be making those points to them provided I am aware of when the relevant inquiry is open. Just in summary, both SA and NSW have got rid of Group Ticket voting successfully (though NSW's system is a bit limited because of the high rate of just voting 1), Tasmania's upper house has single seats so it's not an issue there, Qld, ACT and NT don't have upper houses so it's only Vic and WA that still have group ticket voting. WA has an even worse system than Vic because it is malapportioned and anyone wanting to vote below the line needs to number every box. (It does have a better method for surplus transfers.)

      Delete
  7. I’ve been thinking about the results and the impact of GTVs and it strikes me that there is another aspect to the voting system which is just as unfair: the left over votes that don’t count towards getting anyone elected. The left over votes could in theory be as high as a quota minus 1 of vote (just over 17% of voters) which is an extremely high number of voters that effectively don’t have their votes counted.

    So for example at this election the 14% of people who voted Green in Southern Metro never had their second choice considered but people who voted for minor party candidates didn’t just have their second choice distributed, but their 3rd, 4th, 5th, 10th, 20th 30th, etc, preferences count. So in theory someone’s 50th preference (or higher) can count towards electing someone but the 2nd preference of the left over candidate never gets considered. Therefore the fifth seat is not necessarily won by the most preferred candidate and receiving a high primary vote can actually be a liability.

    I’m thinking back to the Victorian Senate at the 2004 federal election. The 6th seat was a contest between Labor, the Greens and Family First. Before the last exclusion Labor was on 0.56 of a quota, the Greens were on 0.71 of a quota and Family First were on 0.72 of a quota. Labor was excluded and since they directed preferences to Family First, the Family First candidate was elected with 1.26 of a quota and the Greens were left with 0.73. However, clearly the Greens would have preferred the Labor candidate to win and you could argue that Labor (even though they had the lowest % of a quota) was the preferred choice if you had of distributed Green preferences (although probably only by a very small margin). The ridiculous thing about this example is that if more Green voters had actually voted for Family First then the Green candidate would have been eliminated and therefore the Labor candidate would have been elected (likely pleasing Green voters). Therefore we have a voting system where voting for your absolute last choice candidate could actually help your preferred candidate win (eg, if more Green voters voted for Family First the Labor candidate would have won). This strikes me as ridiculously unfair. Please tell me if I am missing something here.
    Without analysing all the results from the Victorian election, I assume this scenario above has played out in at least one of the regions?

    Kevin, has there been much analysis on left over votes and do you have any ideas on how we could address the anomalies I've described above to make the system fairer (assuming you agree with my analysis?

    ReplyDelete
    Replies
    1. So-called "leftover votes" are not remotely as unfair as GVTs, and are arguably just a fact of electoral life, although they do introduce some unpleasant strategic voting issues (which are made much more severe by GVTs). Put simply, in elections there are winners and losers, and not everyone can vote for a winner all of the time. A vote not contributing to someone getting elected is not at all the same as a vote not counting.

      For a simple version of this consider a single-seat election. A single seat election is much the same as a multi-seat election but with a quota of 50%. In a very close seat, like Ripon, 49.98% of the voters' preferences did not end up with the winner and did not contribute to getting anyone elected. But that doesn't mean those votes didn't count or were wasted votes. They counted alright and indeed they were counted many times; it's just there weren't quite enough of them.

      As the number of seats increases, the percentage of people whose votes do not determine any winners decreases, but it is not possible to eliminate the issue altogether. I pointed out the many fatal problems and contradictions with one attempt to do so (using the "Hare quota") in a very technical article at http://kevinbonham.blogspot.com/2017/04/wonk-central-why-we-dont-use-hare-quota.html

      It is true that a voter can reduce the effectiveness of their vote in deciding the results between candidates further down on their list if they vote 1 for a party that is going to poll a high primary vote, is not going to win, but is never going to be excluded. However it is only really GVTs that make this situation likely to happen in a multi-seat preferential contest. In the new Senate system a party that polls most of a quota will generally win, so not only will its voters help get it elected, but part of their vote will probably flow on as a surplus to others.

      Regarding the Family First case, the first thing to mention here is that FF only won because of Group Ticket Voting. Had voters made up their own minds, firstly Family First wouldn't be that high in the count anyway, and even if they were the Labor preferences would flow more to the Greens and the Greens would win.

      Odd cases where voters would be better off voting for an adverse party rather than their own (not only to help a more likeminded party win, but even to help their own party win!) happen in any of these systems where candidates are excluded from the bottom up. However Group Ticket Voting makes these cases much more likely and there were many of them in the 2013 Senate election. As for this one, at one stage there was a risk that Fiona Patten could lose because of votes for minor Green candidates that preferenced her instead of preferencing the other Greens. In Western Metro, I believe the Shooters would have won had more of their voters voted 1 Liberal instead of 1 Shooters. There are bound to be other such cases; I haven't looked closely.

      Delete
  8. Hi Doc,

    Insightful analysis on upper house results in Vic and especially commentary regarding participants in the Druery scam, potentially engaging provisions of the Electoral Act; you qualify your comments by distinguishing between HTVs and GVTs. My thinking is that if both require to be registered with the VEC they are intrinsically linked by virtue of the fact that both direct preferences. I would welcome your further comment on this.

    Further it is still unknown as to who performed the task of submitting the nominations for candidates which may engage 151 3b.

    There is clearly property and benefit which would engage 1 a, b, or c and the participants election conduct has undoubtedly been “influenced” by the preference whisperer as can be clearly seen by the tight preference flows between micro parties at each exclusion point.

    S3c is likewise in play!

    ReplyDelete
    Replies
    1. I'm not a lawyer, despite my extensive experience of application of electoral law. Regarding 3b I think it goes to the fact of whether a candidate nominates, not the mechanism by which they nominate. It would have to be proven that someone had been induced by payment to nominate who would not have otherwise done so, or vice versa. In the context of supposed payments to a consultant to arrange preference flows, there's nothing of that sort.

      Regarding 151 3c, this could depend on how "support" was read. After all I don't think there's any question of someone being induced to make public statements in support of a candidate, which is usually how this is interpreted. Whether arranging a preference deal to the benefit of a party for payment could be interpreted as "support" - that would be more interesting to see discussion of.

      Re 151 3d, both do direct preferences, but the question is whether the GTV preference flow can be interpreted as "set out in the vote of the elector". I'd expect the defence lawyer would just say, no, the elector didn't set out those preferences at all, they just put a 1 in the box and the preference was implied. It is true that a 1 above the line is taken as if those preferences were set out by the voter, but I don't think that's the same thing.

      I'd be interested to see a dispute of return by a candidate or elector to clear up these matters. Of course a difficulty is having evidence of the alleged conduct in question in any of the contests. While it's being widely reported and not denied that Druery was involved in coordinated deals and payments, I don't know if anyone challenging a result would be able to prove it.

      Delete
  9. That didn't take long. Media reports that Catherine Cumming has quit or been pushed out of DHJP after losing their leadership ballot.

    ReplyDelete
    Replies
    1. As anticipated by our Kevin - see the "(assuming that they stay together as a party!)" aside just under the listing of the results early in the main post. She seems to see her resignation more as an anti-Druery move than an anti-Hinch one - interesting...

      Delete