The final preference datasets for the NSW state election were published yesterday. These datasets include the full preference data for every single ballot paper cast in both houses. New South Wales is the only state to publish this data for the lower house, although the upper house data is also published for some other jurisdictions.
There’s so much to analyse in these datasets and I don’t have time to do it all for Friday morning, let alone space to fit it in a blog post. So today I’m going to focus on the two-party-preferred metric in the Legislative Assembly, and how much people are marking the ballots.
The plan for now (although I haven’t written them yet) is to publish a post tomorrow looking at the Legislative Council, then follow up with a post looking at some key metrics in non-classic/crossbench seats and look at different 2CP metrics in seats where more than one are of interest. Then I’m also going to use this data to compare swings amongst different types of voters.
Labor’s statewide two-party-preferred figure is 54.27%, which is slightly higher than the estimates from the ABC.
Of course there are questions about how much 2PP actually represents voters genuine intentions, particularly in non-classic races where lots of voters might just vote ‘1’ for an independent. But it does give a sense of how this was a clear Labor victory, even if it wasn’t with a majority.
First up, this chart shows how preferences flowed from each of the other parties to Labor and Coalition on a 2PP basis:
You can compare this chart to the 2019 equivalent in this post.
It is fairly consistent that the proportion of votes exhausting went down for pretty much every party. Even One Nation's exhaust rate dropped from 71.1% to 62.2%.
I want to particularly draw attention to the proportion of Greens votes flowing to Labor. This proportion has increased from 47.9% in 2015, to 52.5% in 2019, to 59.5% in 2023. Over the same period, the Greens preference flows from the Coalition dropped from 8% to 7.3%, so this isn't just a general increase in preference marking.
Unfortunately we only have this data for the last three elections, so we can't compare to when Labor was last winning elections in the 1999-2007 period. I suspect Greens preferences flow more strongly to Labor when they are on track to win, rather than this being a new high in Greens support for Labor.
This next chart uses the same data as above, but groups the data by its destination, not its origin.
While Labor won 54.3% of the 2PP, that number was only 46.9% if you include exhausted votes. The Coalition polled 39.5% and 13.7% exhausted.
More than half of the votes that flowed to Labor as preferences came from the Greens. One Nation votes made up the largest share of the exhaust pile.
This next chart shows how that 2PP compares to the last 36 years, when including exhaust rates. Labor's 2PP was not quite as high as they achieved in 1999 and 2003.
There has been a lot of chatter about whether there's something wrong with Labor not winning a majority off such a large two-party-preferred vote.
The flaw in this argument is that so many seats were won by independents and minor party members. 12 seats went to the crossbench, with 21 seats being non-classic races between a major party and a non-major party candidate. So if you want to assess whether it was unusually difficult for Labor to win a majority, you also need to include those seats.
There were six seats won by crossbenchers where Labor won a majority. There was no doubt about Lake Macquaire, Balmain, Ballina, Newtown and Sydney. But they also won a huge 2PP majority of 69.7% in Kiama, where they narrowly lost to ex-Liberal independent Gareth Ward. It appears that many Ward voters didn't bother to preference the Liberal candidate, so in the artificial 2PP count, Labor won easily. I'll return to the topic of weird 2PP trends in non-classic races in another post.
So when you add those six seats to Labor's 45 seats, that means Labor won the 2PP in 51 out of 93 seats. While this is not a particularly large majority (Labor lost a bunch of marginal seats by slim margins), the Coalition would still need a substantial swing to pick up the five seats that would give them a 2PP majority in a majority of seats. Labor does not have that many super-marginal seats now, so would need a uniform swing of 3.7% to pick up Penrith, East Hills, Monaro, Camden and Riverstone. Labor won the state 2PP by a 4.3% margin, so that's just a 0.6% bias. Not that much really.
We don't have historical data about how preferences from a particular party flowed prior to 2015, but we do have data that merges together all votes for non-major candidates and looks at where those votes flowed as far back as 1988 for classic races, and since 2007 for all seats.
Labor's flow of preferences is now their second-highest, just behind the 1995 rate, but it's by far their best preference flow since the jump in exhausted preferences in 1999. There also appears to be a general downward trend in exhaustion since a peak in 2011. Exhausted preferences are now at their lowest level since 1999.
Finally, I wanted to look at how many boxes voters numbered on their ballot papers, separate from the question of who makes the two-party-preferred count.
I had wondered whether there really had been an increase in preference marking, and it turns out there has been!
The rate of voters marking just one box dropped from substantially from 64% to 57%. There's still a long way to go, but I think this does show evidence that drawing attention to the value of preferences can make a difference. Optional preferential voting doesn't have to default to first-past-the-post.
My final chart for today looks at preference marking rates by party.
You can compare this to the 2019 chart in this post. Rates of voting 'just 1' have gone down for most parties: Labor, Liberal, Greens, independents, Shooters, Sustainable Australia, One Nation and Animal Justice voters all reduced their rates of just marking a single box. Interestingly Nationals voters did not.
The Greens are still way out in fron of the other parties for their voters marking preferences. More generally, minor parties of the left tend to have more preferencing amongst their voters than on the right.
Tomorrow we'll look at some similar data for the upper house.
Thanks Ben – a detailed, comprehensive and interesting report.
I am sure there is enough tea leaves in there to keep the major party numbers men and women occupied for months, trying to work out what it does or doesn’t mean. I said it before and I say it again – if I was in any of the two major parties, I would be very worried about their long term positioning and success. Living with a primary vote of 37 and 35% means living in heart attack territory, because you will be forever relying on doing deals with minor parties to get preferences and legislation through the parliament.
I have just finished reading a book about diplomacy and it is a messy game, almost like a game of chess where the ultimate goal is always 10 steps in advance, but never announced. It seem to me that both major parties will have to start to employ their own envoys to hold permanent position and talks in “the courts” of the minor parties if they are to succeed in the future.
Blimey Charlie!
Ben: how did you process that huge amount of data?
I do it in Excel and it takes 10 minutes to do the re-calculation for a single electorate
My main aim is to find out not only how many JV1s there were, but how many patterns there were [it’s huge, but not as high as nPm] and how many voters followed EXACTLY the ticket issued by the Party that got their “1” preference mark. That’s always a small proportion – typically 30% for GRN, lower for ALP, but I expect that to be reversed this year. Both parties sometimes get a bit worried by the “non-compliance”. Honchos in the ALP have been heard to say “Green voters “think”, ALP voters don’t”
Geoff
Thanks for this analysis, Ben.
While it’s good to see the allocation of preferences going up, the “drawing attention to the value of preferences” didn’t occur as a result of a fair and transparent civil education campaign. There wasn’t one.
It came about, at least in part, due to the Liberals continuing to disenfranchise voters by mimicking official signage and by spending money saturating booths with a “just vote 1” message (noting that this message wasn’t aimed at their own voters, but at suppressing preferences of their opponents’ voters). This forced others to market the “number every box” message.
It was an awful situation. People on all sides were spending their limited time and effort debating competing voting strategies, when the democratic debate should have been focused on debating competing visions and policies.
“Optional preferential voting doesn’t have to default to first-past-the-post.” Perhaps this is true, but only as technicality. In practice, the messaging becomes very much like first-past-the-post whenever it suits one of the parties to frame it that way.
While I realise that some people like optional preferential voting, it’s democratically inferior to the mandatory preferential voting system used in federal elections.
Peter
You can’t just say that optional preferential voting is democratically inferior to the mandatory preferential voting system – you need to provide your reasonings for others to judge the merits or otherwise of what you are proposing. This was the subject of much back and forth from a more recent post, so I won’t reiterate my views, which is pro OPV if that is not already clear.
I will however say, it ALL depends on the outcome that you are seeking to achieve through applying one or the other (remember – firstly define the problem). I am not a black or white person. I recognise that we live in a grey world and there will ALWAYS be competing and conflicts goals / outcomes that people want to achieve in any of these issues. At some stage, decision makers will need to make a call on which goal / outcome they preference over the other. For me I will always put the most democratic choice over and above some concept of the purity of the voting system, or what might or might not be in the voters mind when they number their ballot (an impossibility in my mind).
Thanks Ben. The high just vote 1 for Libs (73%) & Nats (75%) might reflect their how to votes. Labor often allocates preferences to Greens in the lower house, which will not be distributed, as virtue signalling & it pays with Green preferences in return? It would be much harder for the liberals to adapt the same approach with ON.
Geoff, I use R. A dataset that large is too much for Excel.
Hi Neil, we did have a good debate in that other thread (https://www.tallyroom.com.au/51352).
I did provide my reasoning there about why full preferences are more democratic, which is in summary:
Optional preferences is focused on the negative – the right of a single voter who is determined not to number their least preferred candidates (but only in NSW!)
Full preferences is focused on the positive – the right of all voters to choose their most collectively preferred candidate.
While we didn’t convince one another, we did find an outcome that would meet both our requirements: preferential voting with a savings mechanism (to allow for incomplete votes with a clear intent to be counted).
Higher preferencing rates in 2023 may be because of the increase in non-classic contests, or at least voters knew that their favourite major party may not make the final two, thanks to an increase in independent candidates. It may also be because the federal election was less than a year earlier and so numbering all squares was still fresh in our minds. The big jump in preferencing amongst SFF voters is perhaps because of the loss of competitiveness of SFF in 2023.
The high exhaustion rate of LNP voters may be a reflection of the certainty or near-certainty of the LNP coming in the top two in most races. LNP didn’t come 1st or 2nd in various inner-city seats, Lake Macquarie and Kiama.
It’s encouraging to see NSW voters are numbering more preferences at state elections. It seems that independent challengers in Northern Sydney and Climate 200’s intensive campaign to encourage voters to number every box, and some efforts to encourage voters to preference by Labor, worked and influenced voters. News such as Jacqui Scruby and Joeline Hackman’s joint statement attacking OPV and vowed to “abolish the undemocratic optional preferential voting system” if elected and the Liberal Party launching an 11th-hour challenge in the NSWEC against signs in Manly that told voters not to risk a wasted vote by numbering every box also drew voters’ attention to the importance of preferencing. Labor also put signs in North Shore and Willoughby asking voters to number every box. NSW Labor MLC Rose Jackson also posted a video on social media saying “Don’t waste your vote! If you’re voting for a minor party but you want to change the government then make sure you preference Labor. Make your vote count!”.
It seems that independent challengers in Northern Sydney and Wollondilly’s intensive campaign to encourage voters to number every box worked very well. The rate of non-major party voters not preferencing a major party dropped dramatically in Lane Cove (from 40.69% in 2019 to 29.56% in 2023), North Shore (from 55.47% in 2019 to 35.42% in 2023), Pittwater (from 51.17% in 2019 to 32.00% in 2023), Willoughby (from 49.89% in 2019 to 31.28% in 2023) and Wollondilly (from 63.04% in 2019 to 51.21% in 2023). Smaller declines also occurred in Wakehurst (from 54.38% in 2019 to 46.80% in 2023) and Manly (from 43.55% in 2019 to 38.42% in 2023). A 2PP exhaustion rate of 29.56% among non-major party voters was lower than the rate for any electorate in the 2019 election and probably a record low since all lower house ballot papers started to be data entered. 2PP exhaustion rates among non-major party voters of 31.28% in Willoughby and 32.00% in Pittwater were lower than the rate for any electorate in 2019 except Summer Hill (30.4%). Credit to these community independents for running effective campaigns mobilising voters to make their votes more powerful. Even though most of them didn’t win, making more voters number more preferences was already a big achievement. Notably, independents and Labor have made the formerly safe Liberal seats of Lane Cove, Willoughby and Wollondilly marginal Liberal vs Labor seats, sending a powerful message to the Liberal Party not to take these seats for granted any more.
Comments are closed.