Preference counts shed light on Labor-One Nation contests

OnThe Electoral Commission of South Australia has been spending the last two days recounting votes in seats where they have had to do a new two-candidate-preferred (2CP) count after incorrectly choosing candidates to count on election night. They still have some way to go but are making progress. We now have some 2CP data in 37 out of 47 seats, and we are particularly gaining data for the many Labor vs One Nation races.

I’m going to look at what data we have, and then see what it tells us about the remaining undecided races.

Just as a reminder, we have a bunch of seats where we required a correct 2CP count to know who wins, or if they remain very close. There are also a handful of seats where a full distribution of preferences is necessary to determine the winner.

I had originally written that ECSA wasn’t planning to conduct 3CP counts, but it sounds like they’ve done some exploratory counting to see if independent Lou Nicholson will stay ahead of Labor in Finniss, and that seems to be happening, so I’d expect an Independent vs Liberal preference count there soon.

When analysing preference flows, I prefer to either analyse preferences based on the party of first preference, or by comparing 3CP to 2CP, as these isolate any local seat oddities and ensure you’re comparing like-with-like. Unfortunately none of that data is currently available, so for this post I am looking at the difference between the primary vote and 2CP vote in booths where both have reported, and calculating the percentage of preferences gained for each of the top two candidates.

This first chart looks at the overall flow in all seats with a given 2CP combination. This covers 19 Labor vs One Nation races, 13 Labor vs Liberal races, 3 Liberal vs One Nation races and 1 One Nation vs Independent race. This includes two seats where I have doubts as to whether the current 2CP is correct (Heysen and Finniss) but the data is still of interest. There is no 2CP data for Kavel, Ngadjuri or MacKillop, where there is also doubt about the order of elimination.

Labor is in the 2CP in almost all races, and they do slightly better against One Nation than against the Liberal Party. I suspect that when we eventually have more data we will find that generally One Nation gets a slightly worse preference position for a given primary vote, and in some seats the Labor-Liberal margin is closer than the Labor-One Nation margin.

It’s worth bearing in mind that we don’t know how many of those preferences in Labor-One Nation races come from the Liberal Party. The Liberal vote was severely depleted in some of these seats so many of these votes come from the Greens or other independents or minor parties.

It is a small sample size, but Labor preferences favour Liberal 2:1 when Labor is knocked out. Because of this, the Liberal Party looks solid in Chaffey and Flinders despite strong One Nation challenges, and the Liberal Party looks competitive in Narungga despite One Nation starting with a lead (but more to come on that).

Next up, this chart shows the same data, but broken down by seat. It also shows (in brackets) what proportion of primary votes counted so far have a matching 2CP count.

You can see that Labor-Liberal races have close to complete 2CP counts in most places, which is much less true for Labor-One Nation races.

There is also lots of variety within each category. Preferences strongly favouring Ashton Hurn in her race in Schubert also skew the overall data. If you exclude Schubert, then Labor preferences in Labor-Liberal races improves to 58.3%, much closer to the rate in Labor-One Nation races. Honestly it doesn’t look like Labor’s preference performance differs that much depending on whether they are opposed by Liberal or One Nation.

So what does this mean for the seats where we’ve been waiting for 2CP counts?

Elizabeth and Light

Preference count has caught up in Elizabeth, and 29% of the count is finished in Light. Labor is on 55.7% in Elizabeth and is projected to end up around 56.5% in Light. Both seats called for Labor.

Hammond

No preferences counted here. Labor and One Nation are basically in a primary vote tie, both on 57%. You’d typically expect this to mean Labor is likely to win, but One Nation has received preferences from Airlie Keen and the Liberal Party. Both ABC and Poll Bludger have One Nation in front, so we’ll wait for real preference data here.

Mawson

We have 7% of preferences counted here, which isn’t much. Labor is gaining slightly more preferences than One Nation, and has a big lead, so this is probably a Labor win.

Mount Gambier

31% of preferences have been counted, and independent Travis Fatchen is gaining a very large majority of those preferences. Independent win.

Narungga

Only 13% of preferences have been counted. So far the Liberal is leading on the 2CP count, but those preferences come from Liberal-friendly booths. One Nation is looking to still be the favourite but it is close.

Stuart

Preferences from two tiny booths have reported since I started writing this post. They are slightly favourable to Geoff Brock, who had a large lead. Brock has won.

Current state of play

While the ABC updates more often, here is my seat count:

• Labor – 32 + Mawson likely (awaiting 2CP count)
• Liberal – 4 + leading in Morphett and Heysen (both close, but also possible Greens could overtake Labor on 3CP in Heysen)
• One Nation – leading in Hammond, Narungga (need 2CP count), Ngadjuri and MacKillop (need 3CP count)
• Independents – 2 + leading in Finniss (need 3CP count) and Kavel (need full distribution)

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