NSW 2023 – where preferences have flowed over time

The New South Wales Legislative Assembly has been elected by the optional preferential voting (OPV) method since 1981. Unlike compulsory preferential voting (CPV) which is used in most of Australia, voters are free to number as few or as many boxes as they wish. If a ballot paper is incomplete, the vote ‘exhausts’ once all of the candidates numbered on the ballot are no longer in the count.

This is the first part of a four-part series looking at how preferences work in the lower house. For this post, I’m using historical data to look at who benefits from preferences, and where those preferences come from. In the next post, I will look at some more detailed data we have just for the last two elections which tells us more about how voters for each party mark their ballots. Finally I am going to do a post focusing specifically on Labor and Greens reciprocal preferences in particular seats.

For this post, I’ve compiled a dataset of 685 contests between 1988 and 2019. This covers every seat which was a classic race, where at least three candidates ran, and where preferences were fully distributed. There were 44 seats between 1988 and 1995 where only two candidates ran – it hasn’t happened since then. I also excluded 136 non-classic contests where the final two-candidate-preferred count was not between Labor and Coalition. For this post, I want to analyse how preferences affect races between the major parties in seats that they are competing over.

This chart shows the make-up of primary votes in these seats cast for candidates other than those in the final count.

The first thing that jumps out is the role of Greens preferences. Since 2003, Greens preferences have made up over 40% of minor preferences in these races, peaking at 49.5% in 2019. The Greens appear to have largely replaced the Democrats. If you combine Greens and Democrats preferences, they made up over 40% in 1991 and 1995, and over 32% in 1999.

The role of independent preferences in deciding the outcome in Labor-Coalition contests has significantly declined. Over 60% of these preferences came from independents in 1988, steadily dropping to 1999. It reached a second peak of 30% in 2011, but was only worth about 10% of preferences in 2019.

Bear in mind that this analysis excludes non-classic seats, which excludes any independent who won or made it to the top two. These are specifically independents who are too weak to be a serious contender for their seat. It appears that over time the vote for independents has become more concentrated in places where they have a credible chance of winning, with less polling a smaller share of the vote, but that's a topic for another day.

The Christian Democrats made up a small share, growing to a peak in 2015 before mostly disappearing in 2019. One Nation peaked in 1999 and then quickly disappeared. The share for other parties reached a peak in 2019, when they made up 36% of preferences. This group in 2019 was dominated by the Shooters, Fishers and Farmers, Sustainable Australia, Animal Justice and Keep Sydney Open. Other prominent parties in this category include No Land Tax (2015), Unity (1999-2007), and Against Further Immigration (1995-2007).

It's also worth noting the small share of preferences flowing from a second Coalition candidate, in seats where both the Liberals and Nationals ran a candidate. I'll return to that at the end of this post, but suffice to say this phenomenon has died out since the 1999 election.

So where do these preferences flow? Unfortunately we can't identify preference flows for each specific candidate (and thus party) from this pool of preferences. In some cases you could identify an individual candidate's preference flow by examining the distribution of preferences, but it can be difficult to separate first preferences from votes that passed through a party. For example, if the Greens are eliminated last in a race with six candidates, we know how many votes then flowed from the Greens to the major parties, but we can't separate the Greens primary votes from primary votes for the three other parties that flowed to the Greens along the way.

We do actually have this data for the 2015 and 2019 elections, and I will look at them in the next post, but for historical purposes I'm just going to look at these preferences as one big pool. Where did those preferences flow, and how large was that pool?

The Coalition received the greater share of preferences in 1988, and the parties were in rough parity in 1991 and 1995.

Things changed dramatically in 1999. The total pool of preferences increased dramatically as the major party vote declined. This was the one election where One Nation had a big impact in the Assembly, polling 7.5%.

I know that Queensland Labor used the tactic of encouraging "just vote 1" at the 2001 state election - I haven't found evidence that this tactic was also used at the 1999 NSW election, but it certainly has been used in more recent elections. The 1999 election saw a dramatic increase in candidate numbers. 732 candidates nominated for the Assembly, a 60% increase compared to 1995, and still the record. That would have made it harder for voters to fill out a full ballot paper and likely resulted in fewer voters bothering to complete their ballot.

Exhausted votes made up a majority of all preferences from 1999, and that has continued to be the case ever since.

Labor received more preferences than the Coalition, but never by very much. Exhausted votes dropped in 2003 and 2007 as One Nation disappeared, but spiked again in 2011. The Greens did not recommend preferences to Labor in most seats in 2011, and Labor barely gained more preferences than the Coalition.

There has been a subtle shift over the last two elections, with preferences to the Coalition as a share of the total formal vote dropping to their lowest level since at least 1988, while Labor preferences recovered in 2015 and reached a new high in 2019, with almost 7% of all formal votes flowing to Labor as preferences.

A question for myself is how this might have looked differently with compulsory preferential voting. While Labor does receive more preferences than the Coalition, what about all of those other votes exhausting? Where would they have gone if compulsory preferences were mandated? A large share of the preference pool was vaguely left of centre, including the Greens, Animal Justice and Keep Sydney Open.

I'll return to this topic in the next blog post, analysing the richer data we have for the last two elections, and comparing some of this data to preference flows at federal elections.

On another note, I mentioned the share of preferences flowing from another Coalition candidate. This chart shows the number of seats where both the Liberal and National (Country) parties contested at each state election.

The introduction of optional preferential voting did reduce the number of "three-cornered contests", but not by much. The number of seats had peaked at 5-6 in the early 1970s, when the Coalition was in power, and was four at the last CPV election in 1978. This number dropped to just two in 1981.

The number spiked to its highest ever at the 1988 election. That was when the size of the Assembly was expanded from 99 to 109, producing quite a dramatic redistribution. Quite a few seats were drawn straddling Liberal and Nationals territory, and both parties contested seven such seats.

The number dropped to just two in 1991, when another dramatic redistribution was undertaken to revert back to 99 seats. There were none in 1995, but discipline in the Coalition broke down after they lost power.

In 1996, both Coalition parties contested the Southern Highlands by-election, triggered by the resignation of former premier John Fahey. The Liberal Party narrowly held on thanks to Labor preferences, meanwhile the Nationals lost the seat of Clarence to Labor, shoring up the Carr government's precarious majority.

There was another outbreak of three-cornered contests in 1999, when a reduction in seats from 99 to 93 caused similar issues to 1988. But the exhaustion rate surged, and lost a close race in Monaro despite the Coalition parties collectively outpolling Labor by 10%.

After that 1999 experience, we've never seen another three-cornered contest in New South Wales. While they do sometimes happen in CPV systems, under OPV it's never happened again, and as long as exhaustion rates are so high.

Liked it? Take a second to support the Tally Room on Patreon!