We are now starting to get a substantial sample of votes for the Legislative Council, and it’s giving us some sense of who is likely to win. But the sample is far from complete, and the incompleteness of the sample could make a real difference to where parties stand.
When all the votes are counted, we will have total primary vote counts. These don’t necessarily dictate who is elected, but they are the ‘starting point’ for preference distributions, which usually only change one seat. So understanding where that starting point might be is useful for understanding how close the race is.
As of writing on Sunday evening, about 2 million Legislative Council votes have been recorded. About 80,000 of these votes were blank. Another 103,000 either have below the line markings or are informal – the NSWEC hasn’t yet examined them properly. That leaves about 1.8 million cast above the line for one of the 15 groups with an above the line box. That compares to about 2.6 million formal votes recorded for the Legislative Assembly. There are about 5.5 million voters enrolled in NSW, and 4.7 million votes were cast in 2019. So there’s a lot to be counted.
Let’s start with the current totals for each group, expressed as the number of quotas.
Group
Quotas
Labor (D)
8.15
Liberal/National (I)
6.43
Greens (R)
2.21
One Nation (Q)
1.23
Legalise Cannabis (O)
0.87
Liberal Democrats (J)
0.73
Shooters, Fishers and Farmers (N)
0.67
Animal Justice (C)
0.47
Elizabeth Farrelly (H)
0.27
Lyle Shelton (A)
0.24
Sustainable Australia (S)
0.20
Riccardo Bosi (U)
0.18
Public Education Party (L)
0.17
Informed Medical Options (M)
0.10
Socialist Alliance (F)
0.08
On these numbers, seventeen seats are decided on a full quota. Legalise Cannabis, Liberal Democrats and the Shooters are clearly in front, with Animal Justice in a close contest with the Coalition for the final seat. From now on, I will only focus on those top eight groups.
But we have a bit of data that might help us understand how the current sample may be biased. There are two things to be aware of:
- Only certain votes have been entered into the system so far. No postal votes have been counted, very few pre-poll votes have been counted, while most ordinary votes have been counted.
- Below the line votes have not yet been counted. Some parties do better out of below-the-line votes. If there is a big below-the-line vote for Mark Latham, as there was in 2019, that frees up some of the One Nation above-the-line vote to flow to their next candidate.
First up, let’s compare what votes have come in for 2023 compared to 2019.
Type
2019
2023
Ordinary
2,902,781
1,962,664
Pre-poll
1,023,986
52,946
Others, including postal
826,060
0
Now we know that the pre-poll and postal vote has increased this year, so the proportions should not end up the same, but this shows clearly that nearly all votes counted so far are ordinary. About 83% of all ordinary booths have reported, compared to 3.9% of pre-poll booths.
Next, let’s compare the quotas for the top eight groups for the total vote so far, and just for the ordinary votes. That way we can compare that vote to the ordinary vote from 2019.
Group
Quotas (ordinary)
Quotas
Labor (D)
8.17
8.15
Liberal/National (I)
6.40
6.43
Greens (R)
2.21
2.21
One Nation (Q)
1.23
1.23
Legalise Cannabis (O)
0.88
0.87
Liberal Democrats (J)
0.73
0.73
Shooters, Fishers and Farmers (N)
0.67
0.67
Animal Justice (C)
0.47
0.47
The non-ordinary votes so far are a very small share, so they haven’t shifted the votes too much, but Labor is already doing worse on the pre-poll votes and the Coalition is doing better.
At this point, I want to try and isolate the most equivalent share of the 2019 vote, and compare it to the final total. So what I’ve done is compared the quotas just in the ordinary above-the-line vote, since that is most of what has been counted so far.
Type
2019
2023
Ordinary
2,902,781
1,962,664
Pre-poll
1,023,986
52,946
Others, including postal
826,060
0
Now we know that the pre-poll and postal vote has increased this year, so the proportions should not end up the same, but this shows clearly that nearly all votes counted so far are ordinary. About 83% of all ordinary booths have reported, compared to 3.9% of pre-poll booths.
Next, let’s compare the quotas for the top eight groups for the total vote so far, and just for the ordinary votes. That way we can compare that vote to the ordinary vote from 2019.
Group
Quotas (ATL ordinary)
Quotas
Difference
Liberal/National (K)
7.84
7.66
-0.18
Labor (J)
6.61
6.53
-0.08
Greens (D)
2.18
2.14
-0.04
One Nation (T)
1.42
1.52
+0.10
Shooters, Fishers, Farmers (A)
1.14
1.22
+0.07
Christian Democratic Party (Q)
0.50
0.50
+0.00
Liberal Democrats (O)
0.46
0.48
+0.02
Animal Justice (E)
0.42
0.43
+0.01
Keep Sydney Open (N)
0.38
0.40
+0.02
One Nation and the Shooters both benefit, being boosted by roughly 0.1 of a quota. The three biggest parties all went backwards. I was a bit surprised about the Greens, but I suspect this is about the additional of special above-the-line votes contributing a lot more than below-the-line votes.
If you compare this to the table at the top, it doesn’t look like the Greens or One Nation have any hope of increasing their vote enough to challenge for a third or second seat respectively. But it could make a difference in the gap between the seventh Coalition member (who dropped 0.08 quotas in 2019) and Animal Justice (who lifted 0.01). AJP are currently leading the Coalition by 0.07 quotas on the ordinary vote counted so far. If the same trend took place in 2023, you’d expect that gap to widen to 0.16 quotas, which is quite substantial.
I also want to note that there may be other biases we haven’t detected in terms of which ordinary votes are yet to be counted. I had originally planned to also download Legislative Assembly data to get a sense of what booths are outstanding and how they lean (to the left or right etc) but it was too much of a task to get this done on a reasonable schedule. So I’m going to try something a bit simpler. I have simply calculated a correlation between each party’s vote by seat in the Legislative Council and the turnout so far in each seat, to see whether there might be any major geographic biases in which votes have reported. It’s a bit of a clumsy measure but I’m curious. Of course it may be imperfect if some seats have higher turnout than others (which definitely happens).
Correlations range from -1 (as turnout drops, the party’s vote goes up) to 1 (as turnout increases, the party’s vote goes up) with 0 representing no correlation.
Group
Quotas
Labor (D)
-0.10
Liberal/National (I)
0.18
Greens (R)
0.14
One Nation (Q)
-0.17
Legalise Cannabis (O)
-0.21
Liberal Democrats (J)
0.04
Shooters, Fishers and Farmers (N)
-0.18
Animal Justice (C)
-0.12
It looks like seats with a higher Coalition vote have had more votes counted so far, with a higher Labor vote being associated with a lower count. This is also true for One Nation, Legalise Cannabis and the Shooters. It is consistent with the AJP’s vote going up and the Coalition vote going down, but it’s a very weak correlation.
So what does this mean? It suggests that the current figures are reasonably accurate. I don’t think we’ll see parties with 0.2 quotas of surplus catching up and winning another seat. It also suggests that the AJP may have an advantage over the Coalition in the remaining counting.