Monday, July 1, 2019

Not-A-Poll: Best State Premier/Chief Minister Of The Last 40 Years: Final Round 1

A loooong time ago when the world was young and innocent I started a runoff series to select this site's choice as best state Premier or Chief Minister of every state and territory in the last 40 years.  The plan was to then run a final with all the state and territory winners together.  Ultimately and unsurprisingly with the left-wing skew of readers on this and other psephology sites, Labor leaders won every round convincingly, so I also ran a runoff to get a token Liberal into the final as well.  Earlier this year I got too busy with all the elections going on to run new rounds when each month started, so I have waited until the elections were over before starting the final.

Our contestants and their histories in this contest are:

NSW - Neville Wran, Premier 1976-1986.  Topped the NSW group first round with 37.8% and thrashed Bob Carr 152-50 in the runoff.

Victoria - Daniel Andrews, Premier 2014-present.  The only current Premier to win a state, Andrews polled second in the Victorian group first round with 25.3%, but with a landslide election victory under his belt, defeated Steve Bracks 158-102 in the runoff (which was postponed in an attempt to reduce contamination from the state election.)

Queensland - Wayne Goss, Premier 1989-1996.  Tied with Peter Beattie on 29.3% in the Queensland group first round then cleaned up Beattie 122-73 in the runoff

Western Australia - Geoff Gallop, Premier 2001-2006.  Gallop won the WA first round narrowly with 32.5%.  The first runoff against Carmen Lawrence was tied 97-97 and I hadn't made a rule for ties so there was a second runoff, which Gallop won 75-71.

South Australia - Don Dunstan, Premier 1967-1968, 1970-1979. Smacked it out of the park with 57.4% in the SA group first round.

Tasmania - Jim Bacon, Premier 1998-2004. Polled 37.3% in the Tasmanian group first round and defeated Lara Giddings 105-84 in the runoff.

ACT - Katy Gallagher, Chief Minister 2011-2014. Polled 40.3% in the ACT group first round and defeated Jon Stanhope 119-52 in the runoff.

NT - Clare Martin, Chief Minister 2001-2007.  The only other leader to win in the first round, with 57.6% in the NT first round (none of the others even managed double figures!)

And the Coalition wildcard is Nick Greiner, NSW Premier 1988-1992, who eventually won a long series of Coalition eliminations, defeating Kate Carnell (ACT), who I had been suspecting would win the Coalition series when I started it, 58-38 in the final.

I've decided not to add any more wildcards.

The rules for the final runoffs are:

* The last candidate in each round is eliminated.
* Any candidate failing to poll 8% in a round is eliminated.
* Ties are resolved in favour of the last leader on primaries at a previous stage at which there wasn't a tie, and failing that in favour of the candidate least recently in office.
* Any candidate who could not mathematically win or tie from their position in a preferential election is eliminated.

Spruiking is, as always, welcome in comments.  Voting for round 1 is open til the end of August.


  1. "Any candidate who could not mathematically win or tie from their position in a preferential election is eliminated." - doesn't this leave your poll vulnerable to challengable assumptions about voter flow? For example, if I assume that all the votes from those eliminated go to the next challenger in popularity of like party affiliation, we end up with a two-horse race as the polls stand after my vote - the third place candidate vaulting over the top of the second-placed candidate to take a narrow win off the current runaway leader. Thus, everyone below third would be excluded. But, if a spread the preferences more evenly, or even proportionately amongst the other candidates, including the current front-runner, he quickly reaches a position in which he is unbeatable, capturing more than 50% of the vote. Thus, the even-distribution model calls (in this case) for a single winner and for everyone else to be excluded. The first is more likely to result in a head-to-head or 3-way runoff, with a clear favorite emerging; the second is more likely to produce an instant clear winner. Or have I misunderstood your planned approach completely?

    1. What I mean is that a candidate will be eliminated if they could not win from that position if they got every preference from every other candidate below them or who they might overtake (party status is ignored). For example suppose it is down to four candidates who poll 42%, 30%, 18% and 10%. Obviously the candidate on 10% is eliminated because they are last. But if it was a preferential election and all the fourth candidate's voters preferred the third candidate, then the third candidate would still not get into second place. So the third candidate is eliminated too and only the first two continue to the next round.

      On the other hand suppose four remaining candidates have 39%, 30%, 21% and 10%. The fourth candidate is eliminated. But it would be mathematically possible for the candidate on 21% to win a preferential election from this position (though highly unlikely in practice) so they go through to the next round.

      As things currently stand nobody would be eliminated by this rule because everyone above the 8% threshhold could in theory get to 50% based on voters for other candidates switching to them. The rule is mostly useful for quickly culling uncompetitive candidates if two leading candidates have over two-thirds of the vote between them.