In a post last month about the results from Australia’s 21 May federal election I mentioned that the electoral commission would eventually produce a full account of the preference flows, and that we might take another look then. It has now done so, in this table.
In 1983, the first election for which we have a full official two-party-preferred count, Labor won 53.2% of the two-party-preferred vote off the back of 49.5% of the primary vote. The Coalition’s 46.8% two-party-preferred came from 43.6% of the primary vote. In other words, preferences made very little net difference, increasing Labor’s lead by just 0.5 of a percentage point.
Fast forward almost forty years, and the most obvious difference is the steep decline in the major parties’ share of the primary vote: both have dropped from the high 40s to the low or mid 30s. Obviously, that means that preferences from the rest have become much more important. But those preferences have also shifted. In 1983 Labor won 53.6% of them*; this year it won 61.5%.
The corresponding figure in 2019 was 59.6%, so Labor has improved its performance on preferences after going backwards last time. That’s not surprising, since Greens and pro-climate independents represent a large share of the non-major vote, and their issues were particularly salient this year. Labor’s share of Green preferences went from 82.2% to a record 85.7%.
What’s more surprising is that Labor did better out of the far-right parties as well. Its share of One Nation preferences grew from 34.8% to 35.7%, UAP preferences from 34.9% to 38.1%, Liberal Democrat preferences from 22.8% to 28.2%, Katter’s Australian Party preferences from 33.0% to 38.7%. As Bonham comments, “The change in flow has happened across the board rather than a sharp shift from any one source.”
So while the Coalition still, naturally enough, gets a much bigger boost from far-right voters than Labor does, it’s much less than Labor gets from the Greens. And despite the Morrison government’s avid pursuit of those voters, it’s not getting any bigger.
The other thing worth looking at is the independents. Independent preferences in 2019 favored Labor to the tune of 59.4%; this year that figure rose to 63.8%, and because a lot more people voted independent, the net benefit to Labor in terms of votes more than doubled – from about 90,000 to more than 200,000.
Unfortunately the AEC hasn’t done a seat-by-seat breakdown of those flows, so in seats where an independent was in the top two we can’t say exactly where their preferences went. But we can calculate an approximate figure by looking at the difference between the three-candidate-preferred result (that is, the result immediately before the final elimination) and the final notional two-party-preferred result, and Bonham has compiled a table showing all those numbers.
From it, we see that votes from the seven successful “teal” independents all favored Labor, by margins ranging from 68.8% in Mackellar to 80.1% in Kooyong. The two Victorian seats showed the highest flow, consistent with the hypothesis (discussed last month) that there was more tactical voting there by Labor supporters. Independents in rural seats, who shared some of the teal objectives but were not branded the same way, typically favored Labor less strongly: 58.9% in Calare, 63.2% in Cowper, 66.2% in Indi, 65.5% in Wannon.
Keeping in mind that the real figures are all probably a bit less, since the three-candidate vote includes a substantial Green component, this is less one-sided than the flows from Labor to teals, which ranged from 76.2% in Curtin to 84.8% in Wentworth. But for candidates that were not officially directing preferences and had no tradition of party discipline it’s still pretty strong.
If the Coalition harbors an illusion that teal voters are mostly Liberal supporters at heart and will punish their MPs if they do deals with Labor, these figures suggest it will probably be disappointed.
* Technical note: that figure is a slight overestimate, because Labor would also have picked up some benefit from leakage of Liberal and National preferences in three-cornered contests, but the effect is very small.