Friday, September 7, 2018

Tasmanian Senate Major Party Preselections

Tasmanian major party (Labor and Liberal) Senate preselections are expected to be announced on Saturday 8 September.  I will be posting the results when I know them, and discussion of such voting details as emerge, but I will be out on a field trip for some of Saturday, and I don't yet know at what time the news will be revealed.  For starters then, this post looks at what different Senate placements for some of the candidates would mean.  It will be expanded to cover all relevant aspects of the preselections that emerge.

Before I get any further, I would like to say this: media should not denote any spot on the Tasmanian Senate ballot as "unwinnable".  Unwinnable, obviously, means it is impossible to win.  Lisa Singh was dumped to a so-called "unwinnable" spot in 2016, and won, as a result of below the line votes.  Clearly then her spot was not unwinnable!   It's embarrasing to hear intelligent journalists say stuff like "Lisa Singh was elected from an unwinnable position".   An acceptable alternative to "unwinnable position" is "position winnable only through below the line voting".  This description applies to all positions below 3 for either major party.



For the 2016 Senate election Labor dumped incumbent Lisa Singh to fourth as a result of a member/union/delegate ballot conducted according to a misconfigured implementation of the Hare-Clark system (see more here). This became sixth when a double dissolution was called.  Singh famously polled most of a quota on primary votes (6.12%), fifth-on-the-ticket John Short was eliminated, and Singh went on to reach her quota on below the line votes only (just) and even crossed the line just ahead of the fourth ALP candidate Catryna Bilyk.

On the Liberal side, ballots are conducted by sequential individual ballots of preselectors for each position.  In this process, moderate Richard Colbeck was demoted to fifth, partly as a result of new candidate Jonathan Duniam managing to win the third spot.  Colbeck also attracted a below-the-line campaign and managed just over half a quota below the line (3.97%).  After the initial eight surpluses, Colbeck was running tenth in the distribution of preferences.  However, he (like Singh) could only receive below-the-line preferences while his ticketmate David Bushby was still in the count.  As a result he moved slowly on preferences and was overtaken by Bushby, Labor's Catryna Bilyk, the Greens' Nick McKim and One Nation's Kate McCulloch.  He therefore finished 14th.  Colbeck got his seat back when Liberal #2 Stephen Parry was disqualified from the Senate, and is now again in the outer ministry.

Colbeck has been supported by the new Prime Minister Scott Morrison, as well as by Peter Dutton and Mathias Cormann, but we will see what sway that carries with local preselectors.  Earlier on I was frequently hearing rumours that Colbeck did not have anything near the numbers; it will be interesting to see if this has changed.  There has been intense pressure on the Liberals to nominate a woman to one of the top two positions, as well there should be given that no living woman (but twenty living men) has served the Liberal Party as a federal MHR or Senator for Tasmania.

Half-Senate Election Based on 2016 Figures

The following chart shows the primary vote figures for Tasmania 2016 and the quotas if 2016 had been a half-Senate election:


The quota is about 14.286%.

If the 2016 vote shares are repeated, then Labor and the Liberals win two seats each, the Greens win one and Jacqui Lambie wins one.

The most likely way that this changes is that either Lambie Network or, much less likely, the Greens do so badly that one of the majors manages to beat them by getting a swing to it.  There are reasons Lambie might do badly, but it's also possible that she'll do as well or better.  The reasons she might do not as well as in 2016 could include that she has been less in the spotlight (no longer being a Senator), that the JLN state election campaign was a trainwreck ending in a disappointing vote, and that there could be increased competition for the populist vote from One Nation, right-wing independent Steve Mav and also Lambie's estranged ticketmate Steve Martin (defending his seat as a National).

The Greens' brand is in a sorry condition in the state at present with their vote being continually gouged by the ALP left and independents but it is still hard to see them doing so badly here that they'd actually drop from a notional fifth to seventh or worse and lose.  As for anyone else, it's quite hard to see other parties doing well enough to get over either the Greens or Lambie, but just maybe if Lambie does extremely badly there could be an opening for someone else.  I'm doubtful Martin has had enough publicity in a short term to get more than a few percent, if even that.

Let's say, however, that the final seat comes down to a contest between one or more major parties and either Lambie or the Greens.  How then might the ticket order impact on that?

Here I'll call Singh and Colbeck the "below the line" (BTL) Senators, because that is effectively what they were last time.

If Singh or Colbeck are selected first or second, then those are pretty much automatic-win positions.  It's hard to see either major party failing to get two seats in the state - one might fall below two quotas, but will be saved by preferences and exhaust if so.  In these cases their BTL appeal might drop considerably, but it wouldn't make any difference anyway as they would not need those votes to win.

If the BTL senators are selected third, then after the first two Senators are elected, they will effectively carry the can for their party.  In this case, BTLs for them may as well be party #1s for most purposes, so the scope for a BTL campaign may be reduced, but they may try it anyway to try to get voters in who would never vote for their party. In particular, if Singh is third, it would be in Labor's interests to encourage lots of voters to vote BTL for her with the hope of spreading the vote between the #2 and #3 candidates and getting both across the line with less than a quota.  To pull this off would require Singh to get an even higher all-BTL vote than last time, maybe about 10-11% (In Hare-Clark I call this the Ginninderra effect - warning, extremely wonky link.)

If the BTL senators are selected fourth or worse, then both their BTL vote and the vote of the rest of their party become important.  The reason for this is that this triggers a fight to see who is really the third candidate.  In the 2016 scenario, each major party has not quite two quotas in the rest of its ticket.  As a result, the #3 candidate can't get any above-the-line preferences until the #2 candidate has been elected.  Probably, #3 gets eliminated early in the race, and a Singh or Colbeck at #4 takes over as the effective #3 and starts getting ATL preferences once #2 is out of business.

We can get a fair idea of how "fourth or worse" works from the Section 282 recount, which recounts the 2016 results between the 12 winning candidates as if the election was for six seats.  Singh is fifth on Labor's ticket, but has a huge effective primary vote.  Urquhart and Polley (ALP #1 and #2) are elected with quotas (because of the extra votes from other parties) but Bilyk and Brown are chopped making Singh the effective Labor 3.  From 5th on the ticket she fights with Lambie and Whish-Wilson for the final seat in the recount (but loses by almost 7,000 votes in that instance.)  This shows that even with the sort of BTL vote she got in 2016, Lisa Singh could be vaguely competitive from fourth or worse on the ballot paper - with her fate then determined by her party's performance.

However if Labor do really well on the party vote, this can be a threat to Singh as well as a boost to their chances of winning three.  The reason is that if Labor #3 stays in the race with, say, 0.2 of a quota, they might be able to start snowballing preferences from micro-parties and eventually overtake Singh and eliminate her.  Her preferences would then help Labor win three seats.  So there's a delicate balance there - Singh needs her party to do well on party votes as well as to do well on BTLs, but too well could be a problem.

OK, that's enough number crunching for what will be a total fizzer if the BTL Senators are both preselected first or second - updates to follow on Saturday afternoon or night, once I am home and know something.  I'm also considering doing live coverage of the interesting Wagga Wagga by-election, but this will depend on how busy I am with the Senate story, and anything else going on.

Groundhog Day for Singh

Singh has been demoted to 4th on the ticket behind Brown, Bilyk and Short in a repeat of what happened in 2015.  The ABC has reported Singh had the second highest number of rank and file votes but was swamped by the delegate votes.  I am refusing to link to their report because it describes Singh's position as "considered to be unwinnable", without stating who considers it thus or why the opinions of geese are relevant to electoral analysis.  Singh's position is

********W*I*N*N*A*B*L*E********

(that large enough for you all?) but, as explained above, rather challenging, and harder than her position last time.

There has been some confusion regarding the importance of different vote types.  As I understand it the total value of rank and file votes is 50% and the total value of delegate votes is 50%.  However because there are fewer delegate votes, each delegate vote carries a few times the weighting of each rank and file vote.

...But The Liberals Have Actually Listened

Liberal preselectors, however, have actually listened to voter complaints about their all-male 2016 ticket and the dumping of Richard Colbeck down the ticket.  They have endorsed 1. Richard Colbeck, 2. Claire Chandler 3. Tanya Denison.  This means their chances will ride mainly with the party vote and they will have to hope enthusiasm for their modernised ticket will lift them to the level required to get three seats.  Currently this looks difficult given how the party is travelling federally.

What If Singh Quit Labor?

I've already had one question about this and I am sure there will be further speculation about it, so let's have a look at how the race might play out if Lisa Singh were to quit Labor and run by herself.  I should make it clear that Senator Singh has already been reported as ruling out or saying she had no intention of running as an independent, which is why I've used "quit" rather than "quits" in the heading above. Ellen Coulter (ABC) has tweeted "Can confirm Senator Singh was very quick and very firm to say she would not run as an Independent."

The logistics of running as an "independent" are that Singh would either have to form her own party (which she can do without needing 500 members as she is a sitting Senator) or else run as a non-party group with its own above-the-line square. A non-party group doesn't get a logo, so she would presumably form a party.  In either case Singh would need at least one support candidate.  The reason Singh would have to take one of these options is that otherwise she would only be able to receive preferences below the line, and winning entirely on BTL preferences (as she did in 2016) is unlikely to be possible at a half-Senate election, even in Tasmania.

The advantages of running outside Labor would be:

1. Singh would be able to receive above-the-line preferences from other parties without voters from those other parties having to preference Labor.
2. Singh would be able to receive above-the-line votes from voters who would have liked to vote for her but were put off by having to number 12 boxes (I suspect the number of such voters is very small given that ATL voters are asked to number 6 boxes).
3. Singh would be able to get votes from voters who would not vote for her while she was Labor.
4. Singh would be able to offer to vote independent of Labor's stance on refugees.

While this might sound like a nightmare for the Greens, I think these advantages are easy to overstate.  Voters who vote for Singh do so because she is a person they respect, rather than because they expect her to vote a certain way on the floor of Parliament.  Tasmanian voters are used to voting below the line based on 29% doing so at the last election.  Left-wing voters sympathetic to Singh are particularly likely to be voting below the line already.

The disadvantages of running outside Labor would include:

1. Singh would be unable to get Labor above-the-line votes as preferences in the event that Labor polled around two quotas and Short was quickly eliminated.
2. Singh would be unable to get votes from voters who would only vote for her as a Labor candidate.
3. Singh might be perceived as a sore loser of an internal party ballot.
4. Singh might lose access to within-party logistic support.

In this scenario Labor would be doomed to only win two seats (unless they polled astonishingly well) and Singh would be competing with Lambie and the Greens in a straight fight for two seats.  As noted above, in 2016 Singh got 6.1% on primaries, compared with 8.3% for Lambie and 11.2% for the Greens, so she would need to lift, or have one of them crash, to have a chance.  But if Lambie's vote falls, Singh could well win from on the Labor ticket anyway.

Singh would have a realistic chance running outside Labor, but whether it is a better chance is not that easy to say.

Singh #4 Instead Of Singh #3 Could Be A Blunder

Above I commented on the scenario in which Labor polls so well that John Short has a significant surplus from the top two candidates.  He then accumulated above the line preferences and overtakes Lisa Singh.  Singh is eliminated and Short remains the #3 candidate.

In comments commenter Jamie raises the issue of Singh preferences leaking in this scenario meaning Labor would not be guaranteed the seat.  My initial thought was that if Singh is eliminated, Labor will probably be doing so well that it doesn't matter, assuming her below-the-line vote is as high as last time.  But I have looked at it in detail and it is potentially a massive problem.

This is where the #1 votes for Singh went next in 2016:

57.8% Other Labor candidate
19.8% Greens
9.4% Lambie
6.4% Colbeck (Lib)
1.8% Other Liberal
1.3% Hoult (NXT)
3.5% Somewhere else

Only 43.6% of all 1 Singh votes flowed 2-6 Labor.

A possible three-candidate race mentioned by Jamie is Short vs McKim vs Lambie.  I agree this is the most likely scenario if Singh is eliminated, though if Lambie polls really terribly she could have already been eliminated thus ensuring Short's victory.

In this case this was the 2016 three-candidate breakup of Singh's #1 votes:

Short (ALP) 50.8%
McKim (Green) 26.7%
Lambie 20.7%
Exhaust 1.8%

(In fact exhaust could be slightly higher).

Based on the 2016 figures, in this scenario on Singh's exclusion Labor would lose over 10,000 votes with McKim gaining over 5500 and Lambie nearly 4300.  This would weaken Labor's position relative to its challengers by around 14,000 votes compared to if Singh was number 3.

But it's not even that good, because Singh also attracts substantial below the line flows and these would also leak at even greater rates than the above.  So Labor could at some stage be sitting seemingly pretty on 3.0 quotas with Singh on nearly half a quota, only to drop something like 0.2 quotas on Singh's exclusion and lose.

In comparison if Singh is #3 then the below-the-line votes for Singh never go anywhere.  Of course with Singh at #3 it is possible the number of not-normally-Labor voters flocking to voting Singh BTL to save her from the nasty factions would be a lot less - but over 10,000 votes less when her place is in danger either way?  I doubt it.

Of course widespread awareness of this issue might change the dynamic considerably.  If putting Singh #4 is indeed a blunder because of the risk that Short passes her and her vote leaks, then this might encourage more Labor voters aware of the problem to vote 1 Singh as the best way of ensuring three seats for their party, whatever they may think of Singh versus other ALP candidates.

Alex Jago has identified another scenario: if Jacqui Lambie bombs out so badly that Tanya Denison (Lib #3) outlasts her, Labor's preselection order might conceivably cause the third seat to be lost to the Liberals.  The Liberals are in disarray federally and may suffer major damage from this whenever the election is, but they may also be rewarded for their positive preselections.

Such are the hazards of preselection by member ballot for a complex system!

6 comments:

  1. I have heard that Labor shafted Singh on the ticket for tactical reasons. The premise being that they can run a "progressive within Labor" directly against Greens in Green voting areas, while Labor itself runs a completely different campaign statewide. Then, they can use the Ginninderra effect to boost their chances of getting 3 senators.

    Do the numbers give any credibility to that theory?

    ReplyDelete
    Replies
    1. I think the dominance of the factions over the non-rank-and-file section of the vote probably explains what happens without needing to assume actual strategic intelligence. However, if those voting were trying to be strategic about it it would have been very tempting to dump Singh to 4 in order to pick up "save Lisa" votes again and to try to make the Ginniniderra effect trick fly.

      Delete
  2. I think it’s not safe to assume that BTL voters that vote for Lisa Singh will direct preferences to other Labor candidates. I suspect a lot of voters will direct their number 2 to Nick McKim, as he has a good personal brand even if his party doesn’t. If Lisa Singh gets eliminated I expect Greens, Labor 3 and Lambie to go head to head for two seats, which could favour Greens this time around. Tanya Denison as #3 is a good move too, as she at least has name recognition in the south.

    I know as a normally safe labor voter, the top 3 of the Labor ticket will be well down on the list, I may put them after the Libs for first time ever.

    ReplyDelete
    Replies
    1. Yes the leakage from Singh's personal vote, especially to McKim, will be high if she is eliminated with Short still in the contest. However if that is the case and if her personal vote is as high as last time, Labor may well have done well enough to win anyway. Someone could check the 2016 votes to find the exact rate at which her votes would have leaked in 2016 - something I don't think I sampled much in scrutineering because it was reasonably clear that Singh was winning - but something like 25-30% leaking would not surprise me.

      If Singh's BTL vote is a bit lower than last time and she is excluded then it becomes more likely that Labor costs themselves the seat by putting her 4 instead of 3.

      Delete
  3. Singh would not need the 500 members if she wanted to start her own party since she is a sitting senator - her party would be classed as parliamentary. As long as she did not choose an obviously challengeable name a la Clive Palmer, she would be fine.

    ReplyDelete