As increasing numbers of Australians have chosen to vote early, we have now reached a point where the country is split roughly into two halves – one half voting before election day, through pre-poll voting, postal voting and remote booths, and the other half casting their vote on the day. There are also a few small batches of votes that can’t be identified as being cast on election day or before, but nearly everyone falls into one of these two groups.
For today’s post I am not looking so much at how many people vote early, but instead at how those people vote, both at a national and local level.
Let’s start with something simple – the two-party-preferred vote at a national level.
The early vote has always leaned to the right, but as the size of the early vote got bigger, that had a bigger influence on the election and made the election day vote more progressive. Labor actually won a majority of the election day 2PP in 2016 and 2019.
As the pre-poll vote has expanded, it has also slowed down the counting of votes on election night – I’ll return to that topic tomorrow with some graphs showing this very clearly.
But for today I want to talk about how this complicates the projecting of results. Because pre-poll tends to come in as much larger batches later in night, a differential swing between election day and pre-poll can mean that the earlier projections are off. This wouldn’t be an issue if the pre-poll was broken up into smaller pieces that were reported simultaneously with election day, but since it isn’t, it can cause problems. The problem is not caused by pre-poll being more conservative than election day, but them changing at different rates.
This next chart uses the same data as above, but compares between elections to calculate swings.
Labor did better on the election day swing in 2010, 2013 and 2019, while they did worse in 2004, 2007, 2022 and slightly in 2019. There was little difference in 2016.
The sample size is small, but it seems like early voting tends to swing more, in either direction, than election day – this was true in 2007, 2010, 2013, 2019 and 2022. 2004 is the only clear example bucking that trend, but the early vote has grown substantially since then.
With the volume of the early vote increasing quickly, it would be expected that the make-up of those groups could change. This hasn’t changed enormously – it would have been reasonable to assume that the early vote would become less conservative as it expanded, but that hasn’t happened.
Next up, let’s dive below the national level and look at individual seats. Firstly, I’ll show a scatter plot of Labor’s 2PP in all 150 electorates (adjusted for the redistribution). The black diagonal lines indicates a result where the early vote 2PP and election day 2PP were the same.
Almost every seat has a higher Labor 2PP on election day than pre-poll. Labor won the election day vote in 93 seats and won the early vote in 79. There were 15 seats where Labor won on election day but not the early vote, and just one which was the opposite.
The one big stand-out is Lingiari, where the early vote consists of many remote booths, which don’t vote the way pre-poll and postal do. But the votes cast there are cast before election day, so they count.
There are 14 other seats where Labor does better on the early vote, but by much smaller margins. The second-highest is Grey, where Labor does 1.5% better on the early vote. Grey also has some remote communities, but not on the same scale as Lingiari.
I didn’t want to focus entirely on 2PP, so this next chart shows the 2CP for the independent or minor party in the 27 non-classic seats. This has not been adjusted for the redistribution.
Minor parties and independents generally overperform on the 2CP on election day then lose ground on the early vote. The only exceptions were Clark and Fowler, both seats where the independent defeated Labor in the 2CP, but they didn’t overperform on the early vote by much.
Finally I have two maps. The first one shows the Labor 2PP difference between election day and the early vote per seat, and the second one lets you toggle between the election day 2PP result and the early vote 2PP result.
Overall Labor appeared to overperform on election day the most in Queensland and South Australia in 2022, and also overperformed in “Provincial” seats.
Ben, off topic, but I got into an argument with another chap below the line of the Guardian article you wrote the other day about marginal seats.
Essentially I said the official ATL vote for the Senate was 1 to 6. He referred to an AEC August 2024 document “Ballot paper formality guidelines” that inferred a Senate vote ATL with less that 6 numbers was still valid (page 14).
What is your understanding of the legality of a senate vote with less than 6 numbers ATL and would a single vote ATL still be considered a formal vote?
Thanks
The minimum you need for a vote to be formal is a simple ‘1’.
The AEC instructions are always “at least 1 to 6”. They justify saying that because that is the rule written in legislation but there is also a savings provision that clarifies that anything with a ‘1’ counts, and is formal until there is a gap in the numbering or duplication. So 1 2 3 4 5 also counts.
Ben, would that also be the same for the ACT where ballot paper instructions say number a minimum of 5 preferences but there is a savings provision that allows a single ‘1’ vote to count as well