In Queensland, Labor was ahead in 11 seats after election night, and the Coalition led in 18, representing a projected five seat gain for Labor vs the 2013 election. However, all projected Labor gains had ordinary vote splits of less than 51.5% Two Party Preferred (2PP) to Labor, and the Coalition currently has big postal vote margins in all these five seats, with many more postals still to be counted.
As a result of postal counting and favourable booth rechecks, the Coalition now leads in 74 seats, with 71 for Labor and five crossbenchers. 73 seats are solid for the Coalition, so even if Labor wins the remaining undecided seats, they would still trail 73-72.
On election night, it had seemed possible that the Liberals could lose Grey to the Nick Xenophon Team (NXT). However, the NXT has failed to get the required preference flow, and the Liberals will retain it.
While Hindmarsh in SA is currently the closest seat where Labor is leading, most postals have been counted, and absentee votes should move it back to Labor. I believe that both Hindmarsh and Cowan in WA will be Labor gains.
In Queensland, Forde has moved into the Coalition column on booth rechecking and a 56-44 Coalition split on postals. If postals continue to flow strongly to the Coalition in Queensland, they could also win Herbert, Capricornia and Flynn - all seats where Labor had election night leads. That would give the Coalition 77 of the 150 House seats, a four seat majority.
Labor appeared to have a comfortable 51.5% 2PP in Flynn on ordinary votes, but postals are currently breaking 64-36 to the Coalition, halving Labor’s lead. This is not unusual in Flynn: in 2013, postals went to the Coalition by a 69-31 margin.
Labor’s hope is that the postals counted so far are too favourable to the Coalition, and that later postals will be better for Labor. This could happen if the batches counted so far are from regular postal voters, and later batches are more reflective of the population at large. Labor will do better on absent votes, which are still to be counted, than on ordinary votes, but there are far more postals, particularly in Queensland and Victoria.
An issue with seats where the incorrect final two candidates were selected on election night is that distributing preferences between the correct candidates involves going through the seat booth by booth after election day. Because of variance between booths, the first booths to have the correct final two candidates may be unrepresentative. For example, the earliest results showed Rob Oakeshott winning Cowper, which is incorrect.
The Senate vote count is lagging well behind the House count, and it is too early to project which candidates will win the final seats.
Authors: Adrian Beaumont, PhD Student, School of Mathematics and Statistics, University of Melbourne