I’ve started to see a few questions about how preferences flowed at this election. So while we don’t have the full picture, I thought I would do what I can to get a sense of how preferences flowed, how that compares to preference flows in the past and how it played out in some key seats.
For now it is clear that preference flows to Labor have increased, and exhaustion rates in two-party-preferred counts has gone down to the lowest level since 1999.
We will eventually get full preference data for each ballot which will allow us to calculate how preferences flow from every party and candidate, both on a two-party-preferred basis and in the two-candidate-preferred contest in that seat. For now, all we have are the primary votes for each candidate, and a two-candidate-preferred count between the top two candidates. From this we can calculate how many votes from the pool of minor parties and independents flowed to each of the two major parties, but we can’t distinguish preferences from Greens voters from those from, say, One Nation voters.
For this analysis I’ve narrowed my focus to the 71 classic races where the 2CP matches the two-party-preferred or 2PP (those where it is Labor vs Coalition). I’ve also had to excluded a few piles of votes where the primary vote has been counted but not the 2CP. Absent and enrolment/provisional votes from many seats are in this category, as well as pre-poll booths in East Maitland and Wyong. There are also four seats (only one of them classic) where some but not all postal votes have had their preferences distributed. This is all as of Monday evening.
First of all, let’s look at the statewide picture for those 71 seats. This next chart shows the make-up of the total preference pool in 2019 and 2023, and where those votes flowed on a 2PP basis. I have just looked at the 71 seats. For 2019 results, I have used my post-redistribution calculations for those 71 seats.
The Greens made up about 40% of the preference pool in both 2019 and 2023. Independents made up a slightly larger share, while the Shooters went down. Sustainable Australia ran in over 80 seats, and made up over 10% of the preference pool. One Nation preferences increased, but were still less than Sustainable Australia.
The exhaustion rate has dropped from 52.5% to 47.5%. I did a blog post before the election looking at these flows since 1988. Just looking at classic races, the exhaustion rate was 51.2% in 2019, so it looks like it has fallen below 50% for the first time since the exhaustion rate shot up in 1999.
This is also the highest proportion of preferences that have flowed to Labor since 1995.
If you calculate preference flows as a proportion of the total formal vote, Labor gained 7.7% of the vote in preferences, up from the previous record of 6.9% in 2019. The Coalition gained 3.4%, which was their highest preference gain since 2011.
If you exclude exhausted preferences, the proportion flowing to Labor was a tiny bit lower than in 2019, when Labor gained 70% of preferences that flowed, the highest level on record.
I have also looked at these statistics specifically for 18 classic seats which at the time of writing have a margin of under 6%. I will list show the 2019 and 2023 preference flows for these seats at the end of this post.
Labor's share of preferences increased in 15 of these 18 seats. The only exceptions were Riverstone, South Coast and Tweed. It's worth noting that Riverstone and South Coast were seats where only three candidates ran in 2019, so the entirety of the preference pool came from Greens voters.
It will be interesting to examine these seats when we can break down preferences by seat, but obviously there are substantial differences in who makes up the preference pool in each seat. Greens voters made up just 16.8% of the preference pool in Upper Hunter, and 73.7% in Drummoyne. One Nation only ran in three of these seats: Penrith, Holsworthy and Penrith. Those seats all saw an increase in preference flows to Labor - perhaps not the biggest increases, but still an increase. I suspect we will find that One Nation preferences did not flow particularly strongly to any side when that data is available.
We know that Greens voters are much more likely to mark preferences, and much more likely to preference Labor. This next chart shows that this trend is likely still true in 2023. There is a strong relationship between how much the Labor share of preferences changed, and how much the Greens share of the preference pool changed.
While this trend is clear, there are also a cluster of nine seats where the Greens share of the preference pool did not change much, yet they all saw Labor preference flows increasing. I suspect we will find that, even once you adjust for the different parties contributing to the preference pool, we will find an increased preference flow from Greens to Labor.
That's it for now. I will be working on a piece for tomorrow trying to estimate how many seats might have changed hands under the compulsory preferential voting (CPV) system used at federal elections. But for now, here's the change in preference flows for the 18 closest seats.
Hi Ben, You might be interested in this analysis of the Willoughby By-Election in 2022:
this methodology shows that the independent Larissa Penn would have probably got in.
Would love to hear your critique. Also here is a link to Hansard when OPV and need to have a referendum to change it back to compulsory preferential was legislated.
https://api.parliament.nsw.gov.au/api/hansard/search/daily/searchablepdf/HANSARD-290296563-829
From page 4183 onwards OPV is introduced on 4186 , Liberal objections start at 4187
By the way, how have you determined the 2023 exhaustion rate?- I had not thought this data was available yet.
It’s not hard. We know the exhausted vote count by subtracting the 2CP votes for the two leading candidates from the formal vote. If you subtract the primary vote from the 2CP vote for each of those candidates, the two remainders plus the exhaust is the total preference pool, so the exhaust rate is a proportion of that.
The result in Tweed cost Labor the seat? Confirms that Labor ran the wrong candidate? Personally I am still getting my head around the 800 Green’s voters who preferenced Ward in Kiama. He won on their preferences. what were they thinking?
In early to make up some BS take about Greens voters without any evidence, normal auspol commentary eh Roger. Seems far more likely the 13% of voters who exhausted were wanting nothing to with any of them was a bigger factor, on the little actual evidence there is
I am convinced that Labor won key seats following big PV swings to Labor. A small uptick in the Greens PV in key marginal seats plus an increase of preferences from minor parties, other than the Greens, being the icing on the cake. Labor’s PV swing may be attributable to tactical voting because of OPV and the Liberals having “you can just vote 1” signs, which probably persuaded anti-Liberal votes not to waste their vote (if they were only going to put 1 anyway).
Roger, you couldn’t possibly say how many Greens voters preferenced Ward. To get a like-for-like comparison you’ve gotta subtract the absent and enrolment/provisional since we don’t have a 2CP for those yet. Amongst the rest, there is 11,203 votes for the Greens, Liberals and Sustainable Australia. That includes about 5400 Greens votes and about 5900 Liberal votes. Of those 11,203, 55% exhausted, 28% preferenced Labor and 17% preferenced Ward. That’s a pretty low rate for Labor, presumably because almost half the preference pool are Liberal voters.
If about one third of all Liberal voters preferenced Ward that would be enough to explain every preference he received.
I’d also note that just 12% of Greens voters preferenced Ward in 2019. That was 699 votes. I’m sure some Greens voters will preference Ward but nothing like 800.
Comments are closed.