Victoria 2022: New Lower House Distributions And 2PP Pendulum

Last week saw some late excitement for those of us interested in the entrails of the 2022 Victorian election with the release of 39 previously unavailable preference distributions, albeit officially unofficial and "indicative".   These are the full preference distributions for the seats where previously there was either no distribution (because a candidate was elected on primary votes alone) or an incomplete distribution (because the winner crossed 50% against two or more non-excluded rivals).  36 of these were classic two-party preferred contests (mostly lopsided ones), the remainder being Narracan (a supplementary election that Labor didn't contest), Brunswick (Greens vs ALP) and, entertainingly, Mulgrave.

In the original postcount the VEC kept Mulgrave as a Labor vs Liberal seat although independent Ian Cook (of "Slug Gate" fame) held a narrow primary vote lead for second over the Liberals' Michael Piastrino.  This led to complaints from the Cook camp seeking something they thought was called a "recount" (in fact what they wanted was a realignment).  Cook claimed that "according to my scrutineers, it will bring Daniel Andrews down a few per cent to make the seat marginal".

It didn't (though it might have done the "down a few per cent" bit had Cook's primary lead over Piastrino stayed at 4%).  Cook did in fact hold his eventual primary lead of just under 1% over Piastrino all the way to the final exclusion, but Andrews won the 2CP against Cook by a margin of 60.83% vs 39.17%.  This was, in fact, more than Andrews won the 2PP quick throw vs Piastrino by (60.20-39.80).  

Before moving on I'd like to say that all states should require and fund their electoral commissions to throw all single-seat contests to completion (this includes you Tasmania (Legislative Council) and especially you Queensland).  We have the unusual situation that these VEC distributions which are bound to be far more accurate than the previous 2PP quick throws (in cases with differences of several hundred votes) are technically throws for information purposes only.  Officially Piastrino finished second but actually Cook was Andrews' final opponent in the eyes of voters.  

For my purposes the most useful 2PP figures are the most accurate ones because they are the most predictive for future elections.  A similar situation arose with an uncorrected error in the federal division of Isaacs in 2019, which meant Isaacs was considerably less marginal going into 2022 than it appeared.  Therefore I am posting revised 2PP figures based on the new distributions and a 2PP pendulum ditto.

At the time of writing I have the state 2PP using the revised figures at 54.95 to Labor (1,987,820 to 1,629,811) compared to the original 2PP of 55.00 but in the 2PP pendulum below I have the figures this is based on so any clerical or other errors can be spotted and corrected.  If any errors are found I will edit this article accordingly.  Note that to one decimal place the figure I have rounds to 54.9 not 55.0.

However for the purpose of modelling the next election it is necessary to estimate a 2PP if Narracan had been contested by Labor.  For this I get 54.79%.

2PP Pendulum

I produce 2PP-based post-election pendulums for elections because I have a strong dislike of pendulums that mix classic and non-classic seats.  These miss the purpose of a pendulum, which is to predict how many seats will fall for a given 2PP swing.  

The conventions I adopt for 2PP pendulums are:

* A seat held by the crossbench is listed separately with its 2PP margin shown in brackets.

* A seat won by a major party against someone else is listed in the main part of the pendulum with the 2CP margin and opponent shown in brackets.  

* I also note the following 3CP situations:

- 3CP: The 3CP swing to change the result in cases where the 3CP loser would have won had they made the top two

- S3CP (survival three candidate preferred) - the swing required for the winner to be eliminated at the three-candidate stage by another candidate who would have been competitive but lost

- L3CP (losing three candidate preferred) - the swing required to replace the 2CP loser with a different candidate who would have been competitive but lost

The definition of "competitive" is an estimated 2CP inside 56-44.

The pendulum below shows what I have as the 2PP votes for the major parties in each seat, in case any errors are spotted (click for larger version):

(S) indicates sophomore effect is expected for a first-term MP, provided they recontest - this may be worth 1% or so.  (SS) equals sophomore surge and denotes a seat where the other party's incumbent was defeated.  

The pendulum shows some marginal Liberal vs ALP seats are not a massive ALP to Green swing away from becoming Liberal vs Green (might the Greens someday make the top two in Caulfield?) In Sandringham (4.79% 3CP swing needed) and Brighton (3.93%) the Labor to Green margin (which is double the swing figure) is also not massive, but in these cases the Greens would not be competitive in the 2CP.  

The situation in Preston, where an independent who finished fourth almost made the 2CP, is too unusual to be worth codifying.

It should be noted that four of the seats where Labor has massive 2PP margins are 2CP-marginal vs the Greens, so the Greens will be chasing eight seats if all goes well for them (or badly for Labor) next time.

Pendulum Pain For Coalition!

In my original article I noted the issue of "pendulum tilt" - the current pendulum is wildly favourable to Labor in terms of converting 2PP votes into seats.  With these new distributions I can revise estimates slightly.  By uniform swing (adjusted for Narracan estimate) and assuming no crossbench changes:

* The Coalition needs a 51.86% 2PP to put Labor out of majority

* The Coalition needs a 52.83% 2PP to win a majority of its own

Swings are never actually uniform and a probability model taking personal votes into account would give slightly different results, but on a uniform swing basis and assuming no crossbench changes there is a 4.69% gap between Labor's expected minimum for a majority and the Coalition's.  That's a hefty advantage.  I will produce such a model in the leadup to the next election once I know if there will be a redistribution or not.

A further problem for the Coalition is that while increased crossbench success might make it harder for Labor to win a majority, that would most likely involve the Greens taking seats from Labor (which doesn't help the Coalition govern) or could involve teal-type independents taking seats from either side (probably ditto).  Right-wing alternatives are negligible at the moment, as was shown up at this election by the non-viability of the Victorians Party and the sinking without trace of media-hyped "purple independents".