Friday, December 16, 2022

Victorian Upper House 2022: It's Still Not Real Democracy But It Is Funny

Legislative Council 2022: ALP 15 L-NP 14 GRN 4 LCV 2 AJP 1 SFF 1 PHON 1 LDP 1 LDLP 1

The buttons have been pressed on the Victorian Legislative Council election for 2022.  This election yet again employed the long-discredited Group Ticket Voting system even though the same system wrecked the 2018 election . Nearly all members of the 2018-22 parliament, and most of the parties contesting the 2022 election, did nothing about the fact that at least eight of the 40 MLCs in the parliament were not there on electoral merit.  

And so, this garbage system in which parties beat other parties with several times their vote and win regional seats they would not have won in any other system on the earth continued.  That said, the 2022 results aren't nearly so bad as last time.  Yet again, there have been many individually incorrect results (including one which focuses attention on yet another undemocratic junk feature of Australia's worst electoral system) but all the same,  the overall balance of this parliament is remarkably representative.  There have also been hilarious doses of karma.  Parties that supported preference harvesting or did nothing about it have in many cases been hoist on their own petard while several parties that opposed Group Ticket Voting have had success.  Glenn Druery's "alliance" has not been nearly as successful this election after a campaign in which Druery got stung not just once but twice. While its members did get three seats between them, they may well have won one of those anyway.  Nearly all the randoms who got in at the last election lost, and both major parties lost a seat they should have won as a price of their inaction on the problem.


In Northern Victoria, the Animal Justice Party (the least deserving winners in a region by vote share) ratted on a preference deal organised by Druery and snowballed their way past three Druery parties and Labor among others (to victory) while giving nothing back - not a fair reflection of voter intention but still something that several of the victims had coming.  

I add once again that this post is not intended as an attack on the calibre of those newly elected to represent parties with smallish vote shares in their region at the expense of more popular parties.  They may turn out to all be excellent MPs, and at least this time several of them support electoral reform.  Rather the point is that they were not elected by a proper electoral system and do not have a proper mandate from their regions.  We have seen this at this election - most of the crossbench elected in 2018 lost.  

Before I move onto general issues I want to comment about one result that is attracting the most attention and the most misconceptions.

Der Fall Somyurek

The case of Adem Somyurek (LDLP, Northern Metro) has been far too complex for many people's preconceptions, especially because he beat Fiona Patten (Reason) whose own situation vis-à-vis Group Ticket Voting also doesn't sit neatly with the folk-hero narrative.  The TL;DR is that Group Ticket Voting secured the bad guy's victory, but he may have won anyway.  Indeed he has clearer claims to being there on merit than the MLC he replaced has had before or would have had this time.

Patten was elected in 2014 through the Druery alliance, beating Labor for the final seat by GTV despite them having over twice as many leftover primary votes.  She famously split with Druery in the leadup to the 2018 election, but from this two myths sprang up: firstly that Patten was opposed to GTV and secondly that Patten's win in 2018 was not connected with GTV.  In fact, Patten and her party (Sex Party, later Reason) supported retaining Group Ticket Voting but wanted a ban placed on payments to external preference cartel consultants, a ban obviously aimed at Druery.  Furthermore, Patten's win in 2018 was again reliant on Group Ticket Voting flows - just not those determined by Druery.  Patten was able to get above-the-line preferences from other parties to beat Labor again despite them having nearly three times as many votes after their second quota. 

Probably because so many punters on the left admire Patten and despise Somyurek, a very common narrative has been that Patten only lost and Somyurek only won because of Group Ticket Voting.  A connected narrative has been that Somyurek could only have won off such a modest vote share because of GTVs.  But there is not a lot of truth in this.

The primary votes in Northern Metropolitan were Labor 2.002 quotas, Liberal 1.131, Greens 1.116, Labour DLP (Somyurek) 0.287 (4.79%), Victorian Socialists 0.283 (4.71%), Reason (Patten) 0.217 (3.61%).  Labor takes two seats, the Liberals one and the Greens one, and after that nobody has anything much.  This is very similar to the federal situation in which Ralph Babet (UAP) won a Victorian Senate seat - Babet had the highest vote left over in the race for the final seat and nobody even made inroads on catching him.

In a Senate-style contest this would simply be a case of cutting from the bottom up until either Somyurek or one of Victorian Socialists or Patten won.  Patten would be coming from 1.1% behind Victorian Socialists with not that many (about 6.5%) of friendly preferences to draw on; perhaps she could catch them on that and make the final contest, but it wouldn't be trivial.  In any case, almost as much of the preference pool is right-wing parties as left-wing parties and Somyurek would at least be competitive for the seat.  Perhaps he would win anyway, perhaps not, but what the GTV system has really done here is deprived us of a great scrap on the way to finding out.  The problem is that above-the-line voters for relatively middle of the road/politically confused microparties like Sustainable Australia, Hinch Justice, Transport Matters etc have had their votes sent to Somyurek en masse when if they actually troubled themselves to learn to vote properly they might have sent them left, right or to exhaust.  

Something I have been exploring is whether the huge ballots at this election may have amplified the ballot paper confusion in which Labour DLP and the Liberal Democrats poll much better on average when they draw to the left of Labor and the Liberal ticket respectively than when they draw on the right.  I have not found clear evidence of this.  The metric I have looked at is average swings for both these parties from 2014 to 2018 and from 2018 to 2022 with position changes (or lack thereof) classified as left to left, left to right, right to left and right to right in terms of position relative to the parent party.  This is what I got (not guaranteed error free):

2014 to 2018 Left to left 0.26% (2 cases), left to right -2.24% (5), right to left 1.28 (4), right to right -0.17% (5)

2018 to 2022 Left to left 1.06% (3 cases), left to right -1.13% (3), right to left 1.58% (6), right to right 0.02% (4)

As expected, moving from left to right of a parent party is a disadvantage while moving from right to left is a benefit.  Both the left to left and right to left cases saw a larger positive swing in 2022 than 2018, which is consistent with the idea that drawing to the left was more of a benefit in 2022 because of the larger ballot paper.  But the difficulty of taking this seriously as evidence on such a trivial sample size is further highlighted by the average left-to-right disbenefit being far lower in 2022 than 2018 for no reason.  Also it just happened that two of the left-to-left cases in 2022 were Finn and Somyurek, unusually prominent Labour DLP candidates who were trying very hard to get elected.  

Voter confusion resulting from drawing to the left of Labor is very likely to have had something to do with the size of Somyurek's primary vote, but that's not specific to Group Ticket Voting.  We've seen in the Senate system that even with GTV abolished, too many pointless parties still flood the ballot and confusion has still often occurred.  Most of Somyurek's high-ish vote (in a region with some historically strong areas for his new party) came because he campaigned actively and targeted areas where Labor were a bit on the nose but voters were not receptive to his opponents.  

Results and Vote Share

The 2018 election delivered a result that was severely disproportionate, with the Greens getting only one seat off a vote share deserving of four, while two parties lacking the statewide support to merit any seats won one each, and two others got more than they deserved.

At this election, the result is surprisingly proportionate.  The table below shows the current statewide vote shares (I added averages by region last time but they were little different), the number and percentage of seats won, and in the final column the number that I would expect to be won based on the votes cast if Victoria had a single 40-seat (Droop quota) division.  


How has a system that produced a disproportionate dumpster fire in 2018 come up with a 2022 result that looks like a real election? Firstly the presence of left and right GTV spiral groupings distinct from the Druery group has made it harder for candidates to win off very small vote shares.  In 2018 there were Druery group preference spiral wins off 0.62%, 0.84% and 1.32%.  In 2022 the lowest winning share was 1.56% for Animal Justice in Northern Victoria and everyone else who won did so with at least 3%.  

Secondly the tendency of Group Ticket Voting to randomise results for parties that are leading in the race for the final seat has, in this case, cancelled out one of the negative impacts of splitting 40 seats into 8 regions of 5: that the Greens can end up over-represented.  Adrian Beaumont has published analysis in the Conversation finding that the use of the Senate system would have seen the Greens win in almost every region.  The exception is Northern Victoria where I believe his comments are based on analysis I tweeted to the effect that Senate BTL preference flows to the Shooters are so strong in this area that the Shooters would probably overtake the Greens. Aside from Labor, the Coalition and Greens, it seems only two minor parties would have won: the Shooters in Northern Victoria and either Labour DLP or Victorian Socialists in Northern Metropolitan.  Ben Raue has a slightly different results set with the Greens winning six instead of seven with Labour DLP beating the Greens in Eastern Victoria; I think if any minor party could have bridged the (very large) gap it would much more likely have been One Nation.  (Whether they could do it requires detailed modelling of Senate preference flows and some assumptions about Family First and Freedom Party preferences.)

While Group Ticket Voting needs to be scrapped to make sure that wins from tiny vote shares cannot happen again (and for the other reasons outlined in the 2018 article) Adrian's analysis highlights why reform should include breaking up the current 8x5 regional structure.  There may need to be a test case to determine whether the asymmetrical double entrenchment of the 8x5 structure in the Victorian constitution was truly valid, and if it was then there needs to be a referendum to fix it, but all this should be done.  (Some interesting discussion in comments at PB re this.  There is a previous case, Trethowan, that says asymmetrical entrenchment can work, but it is controversial.)  Or the government may prefer to go direct to the referendum to avoid dismantling the rest of the Bracks' governments entrenchments.  

Trailing Winners

In 2018 ten non-Greens crossbench MLCs won by overtaking other candidates using Group Ticket preference flows; perhaps two of these would have won by voter choice.  In 2022 the number of non-Greens crossbenchers has fallen to seven (the sidebar Not-A-Poll here predicted 7.8 on average, with 5-6 having a plurality of votes over 7-8).  Of these, Somyurek was a primary vote leader.  The remainder would have all either definitely or probably not won in their regions:

* The Shooters (3.00%) would not have beaten the Greens (7.97%) in Eastern Victoria.

* One Nation (3.73%) and Animal Justice (1.56%) would not have beaten Labor (12.16% notional excess) in Northern Victoria, and it's unlikely One Nation would have beaten the Shooters (5.10%).

* Legalise Cannabis (5.26%) and Liberal Democrats (3.62%) would not have caught the Liberals (10.09% notional excess) or Greens (6.73%) in South-East Metro.

* Legalise Cannabis (4.29%) would not have overtaken the Greens (8.08%) in Western Metro.

These results don't have the same sting of unfairness that they had in 2018, because the parties involved have all ended up fairly represented at state level.  But there's no guarantee that will continue to be the case, and accountability to the electorate is lessened in cases like Animal Justice where it's not the same person getting elected as before.

GTV Incumbents Mostly Defeated

One of the arguments advanced for Group Ticket Voting is that it increases the diversity of backgrounds in parliament by electing more ordinary people.  But if Group Ticket Voting in the current vote share environment was a good system for electing good politicians then those who got elected under it from behind on primary votes would get recognised by voters as unusually good MLCs and re-elected.  This hasn't happened; only two of the ten non-Green crossbenchers elected in 2018 have been returned, and that isn't unusual either - historically less than 50% of such winners under Group Ticket Voting around the country had kept their seats before this election.  The incumbents in this term of parliament in general didn't build their vote significantly, and even if incumbents under GTV do build their vote by a lot they can still lose to GTV randomness.  (Somyurek was also re-elected but is irrelevant to this section as he was a Labor MLC at the previous election.)

Impact of BTLs

I was hoping to see a larger increase in below the line voting at this election than occurred - there was only a marginal increase from 8.9% to 9.5%.  There were two cases where the fact that some votes were below the line rather than above the line resulted in a different winner, but there were also two others with interesting below-the-line impacts.

The first of these was North-Eastern Metropolitan, where had all votes been above the line Rod Barton of Transport Matters would have been re-elected despite his party running second-last with 0.21% of the vote.  (A staggeringly low vote for an incumbent who was at times high profile and was well regarded by the commentariat, but who probably would have got more votes if he spent the whole term making random burping noises.)  Barton needed virtually every New Democrats vote to flow to him above the line to get his preference spiral going, but 42% of the New Democrats votes in North-Eastern Metro were below the line, and even without that being the case, there were two other points at which below the line votes would have caused the spiral to collapse.

The other was Western Victoria where Legalise Cannabis would have beaten the Greens by 0.29% had all votes been above the line, but in fact the Greens had a 0.08% advantage on above the line votes, and also did better than Legalise Cannabis on below the lines from other parties.

There are two more results that are interesting in terms of the impact of BTLs.  David Limbrick was the calculator winner in South Eastern Metro and won by almost as much as the calculator projected (with a big hand from Inclusive Gregory, see below).  Considering just above-the-line votes, Limbrick was actually behind, but he did well enough on below the lines to win (partly because the Liberals lost votes to leakage within their ticket, but also because he picked up many below the line preferences from anti-lockdown/antivax parties, including those whose ticket votes flowed to the Liberals).  Secondly Bernie Finn was by my estimate very slightly ahead on ATLs in West Metro, but lost vary narrowly because of BTLs.  In both these cases the calculator winner won, but given that there were BTLs, they needed help from those BTLs to do it.

Inclusive Gregory Surplus Shambles

Inclusive Gregory is an abomination that exists in the Victorian upper house as well as, eg, in federal Senate counts.  Any politician who does not support replacing it with Weighted Inclusive Gregory should resign. 

Inclusive Gregory is a system for distributing surplus votes when a candidate is elected with a surplus not on primary votes but partway through the count.  A relic from the paper count days, Inclusive Gregory means that even if a candidate who is elected during the count carries some votes at full value and others that have used virtually all of their value electing someone else, both those votes leave the surplus at the same value.  It means that the total contribution of a given vote to the count can become significantly more than or less than one vote.   Its impacts are especially severe when combined with Group Ticket Voting, because GTV often results in large surpluses involving groups of votes with very different preferencing patterns.

Inclusive Gregory has long been a problem, but the only previous clearcut case of it changing a result was a rather weird one in 2014 where it only foiled a preference harvester off 2.49% of the vote who didn't deserve to win anyway.  This election the overweighting of Labor votes in surpluses has caused the Liberal Democrats to beat the Liberals in South-Eastern Metro but also caused the Liberals to beat Labour DLP in Western Metro.  

David Limbrick retaining his seat was with the assistance of party name confusion (he had a  great draw to benefit from that), Group Ticket Voting and Inclusive Gregory - without all of these he would have lost.  But as noted above he also would have lost without superior BTL performance to the Liberals, and a further critical ingredient was building a personal vote.  (As a sign of this, Liberal Democrat incumbents Limbrick and Tim Quilty had much higher BTL percentages than other Liberal Democrat candidates.)

Finn on the contrary only got as close as he did by preference harvesting - his party's 5.2% vote was very good, but he didn't have any business beating the Liberals on 7.8%.  

It's a bit disappointing to see the ABC claiming what as far as I can tell is undeserved credit for predicting Limbrick's re-election.  Antony Green has said "Last week I published a blog post pointing out that the formula used for distributing surplus to quota preferences would result in Labor votes being hugely over-represented in the Legalise Cannabis surplus in South-Eastern Metroplitan Region. I argued this formula would result in Liberal Democrat David Limbrick defeating Liberal Manju Hanumantharayappa in the race for the final seat.

And that's exactly what has happened."

That is not true, at least not in that blog post.  The blog post said only "This over-representation brings Liberal Democrat David Limbrick close to winning the final seat, and the only reason Limbrick is even close to election is the distortion caused by the IG method."  That can fairly be taken as a statement that Limbrick could win but not that he would.  In fact the ABC provisional results page continued to list Limbrick's Liberal opponent as a "Likely" winner and Limbrick as only "Possible" right up to the button press, whereas I ultimately said that I couldn't pick who would win it before the button press.  I also noted why the ABC's mention of 67,000 vs 60,000 ATLs was not telling the whole story:

"Sunday: Limbrick made a further small gain in calculator position and finished with a calculator lead of 0.67%.  Seems to me he is seriously in this.  The ABC's commentary refers to 67000 vs 60000 locked in votes but I believe this is the situation before the Legalise Cannabis surplus, some of which is locked in and on which Limbrick clearly gains." 

Modelling these counts is very difficult and I only worked out that Limbrick was a more than outside chance on the second Thursday, having pretty much ignored him until he took the calculator lead during rechecking.  All the same, no one should claim credit for predictions they didn't make.  

The State Of The Chamber

Overall, Labor has a relatively friendly upper house, as it should have given that it won the lower house convincingly.  A 22-18 left-right split is a good reflection given there was a 55-45 2PP result downstairs.

Labor's easiest path to passing motions will be with the Greens and Legalise Cannabis.  If Legalise Cannabis are not on side they need the Greens, Animal Justice and one of the four right crossbenchers.  Another path that may be useful sometime is Legalise Cannabis plus four of the five lone MLCs.  

The Greens, Legalise Cannabis, One Nation and Animal Justice have all expressed some level of support for abolishing Group Ticket Voting, as has Limbrick in a personal capacity.  However some of these (especially AJP and I'd expect LDP) are conditional on scrapping the current regional system.  Labor will certainly have the numbers to get some kind of reform or other through and the blame will be entirely on it in this term if it does not make some kind of effort.

It will be interesting to see how Legalise Cannabis' pursuit of their major political aim affects the dynamics of the chamber.  This is their second success after picking up two seats in WA, and they also polled very significantly in places in the Senate this year.  Although I generally describe them as a left party, their appeal is not purely left and in rural and outer suburban areas there's crossover between them and the Shooters and One Nation vote.  This is the first case where they will have real power (and lots of it).  For any major parties that are stuck on how to deal with this lot, here's the big secret: if you legalise cannabis, they'll go away.

(NB I have also published a Lower House review.)

3 comments:

  1. Hi Kevin,
    One of the reasons why you're having difficulty quantifying the advantage LDP/DLP drawing to the "left" of the major party might be that this year the actual Vic upper house ballot papers had double party rows and, in my experience, were much better arranged so that you were naturally drawn to the center of the page more than with something like the federal Senate ballot's single row of parties. So just measuring "left/right" position might not be taking into account whether a party was on the top row/bottom row or actually any easier to find on the ballot this year?

    ReplyDelete
    Replies
    1. Yes I've seen Ben Raue's post here where he lists the relative ballot positions in terms of above/below etc https://www.tallyroom.com.au/50466 and I was trying to add in assessment of above/below but with so little data it was going to be useless; there's a big risk that you end up testing too many different hypotheses. Some of the positions like left of but below are ambiguous as to whether they would be beneficial or not (I just used alphabetical order, so left but below is classified as right). In Somyurek's case he was listed as just to the left of Labor (not way left) and neither above nor below.

      Delete
  2. I personally favour abolishing the regions entirely and using party lists to reduce the ballot paper size.

    You could perhaps argue that in regional Victoria the regions make some sense, but in Melbourne itself they are absurd.

    However, I don't think the Coalition will agree to it (how would they determine the Lib list vs the Nat list) and nor will the Greens for whom a single electorate in the upper house would guarantee 4-5 members each term vs 7-8 if regions are retained.

    That would leave Labor trying to argue with the micro party turkeys that they should vote for Christmas.

    ReplyDelete

The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.