One Nation claimed victory yesterday in the final doubtful seat, Narungga, so the South Australian election is basically over. (See official results here.) The sitting Labor government finishes with a huge majority in the House of Assembly, 34 of the 47 seats, up seven from last time (two of which it had won mid-term in by-elections). The Liberals have been reduced to five (down 11), One Nation four (all new) and four independents (unchanged).
In terms of votes it looks rather different. Labor has 37.6% of the vote, down 2.4%; One Nation is second with 22.5% (up 19.9%). The Liberals are back on 19.4% (down 16.3%), followed by the Greens on 10.0% (up 0.9%) and Family First 1.7% (down 2.0%). Independents in aggregate had 4.8% (down 2.5%). If seats were allocated proportionally, Labor would have won 18 seats to One Nation’s 11, with ten Liberals, five Greens, a Family First and two independents.* A quick calculation gives me a Gallagher score of 28.4, which is dreadful (higher is more unfair; most democracies routinely manage to get under ten).
So all the opposition parties have cause to complain about the electoral system. But their interests are different. One Nation, for example, will not support proportional representation because that would give more power to its arch enemies, the Greens. Instead it argues for optional preferential voting, removing the need for voters to number every square (as already applies in New South Wales).
Optional preferences are a fine thing: it would be a significant reform to our system because it would prevent what are otherwise perfectly good votes being discarded due to mistakes in numbering. But pretty much everything else One Nation says about voting systems is wrong. So it’s worth having a look at the actual impact that preferences made in South Australia.
The first thing to note is that their overall effect was not great. In only four seats did they change the result: the ALP won from second place in Morphett, One Nation did the same in Hammond, and independents won from second place in Kavel and, remarkably enough, from fourth place in Finniss. (It’s believed that has never happened before in a state or federal election.)
So if the election had been conducted under first-past-the-post – as an increasingly vocal segment of opinion on the right seems to want – the result would have been (assuming no corresponding change in voting behavior) almost unchanged. The Liberals would have picked up two extra seats and Labor one, at the expense of two independents and one One Nation candidate.
In my preview I said that “if three or more One Nation MPs are elected on Liberal preferences it will set off a firestorm in the Liberal Party.” That did not happen. One Nation MPs in Hammond and Ngadjuri can certainly thank the Liberals; if Liberal preferences had flowed even fairly modestly to Labor instead it would have won both seats. But in a third, Light, where a decent flow of Liberal preferences would have got One Nation up, it didn’t get them. Liberal preferences went about 50-50 and Labor held the seat.
The other two One Nation victories, in MacKillop and Narungga, were against the Liberals, so Liberal preferences were never distributed. In both One Nation started with a big lead on primaries (11.3 points in MacKillop, 15.1 in Narungga). Labor preferences flowed strongly to the Liberal in each, but were not quite enough for them to overtake the far right – although in Narungga they got within 77 votes.
There were also a few seats in which One Nation preferences mattered. Morphett was the mirror image of Light: One Nation was the one that came third and its preferences split about 50-50. If they’d favored the Liberal he would have won. In Finniss One Nation preferences helped the independent to beat the Liberal, who could have held it if they’d run in his favor. And in Kavel One Nation preferences could have put the Liberals into the final count against Labor, but they appear to have favored the independent instead (details of the distribution there have not yet been published).
There were also four seats in which Labor preferences helped other candidates to beat One Nation: the Liberals in Chaffey and Flinders, and independents in Mount Gambier and Stuart. But in each case the One Nation candidate was behind on primaries to start with (although only by forty votes in Mount Gambier), and in Stuart at least the Labor vote was so low that it wouldn’t have made much difference.
In summary, there is no warrant for the claim that preferences worked against One Nation, or more generally that it is disadvantaged by the electoral system – except in the sense that all parties except Labor were harmed by the system of single-member districts. But One Nation and its supporters have no interest in questions of truth or falsity, so no doubt the conspiracy theories will keep coming.
What South Australia does show, however, is that the flow of preferences between Liberals and One Nation can potentially affect a lot of results. Each party has to choose what moral it takes from that: whether to pursue a preference deal, or whether to use its preferences to try to inflict maximum damage on the other. And numbers alone won’t decide that question.
(For more on the South Australian result, especially the question of why Labor’s relatively modest vote converted into such a big swag of seats, Ben Raue’s analysis is particularly interesting.)
.
* Technical note: That’s a Sainte-Laguë calculation; D’Hondt gives the same result except that Labor wins the final seat instead of Family First. I’m assuming that the independents would have sufficient coherence to run a single ticket; without them, Labor and One Nation would each win an extra seat.
The World is Not Enough 