Mapping out a majority early vote


Now that all of the votes have been counted, I wanted to look at the various methods of casting a ballot, which methods have become more popular, and how the two-party-preferred vote varies depending on when someone casts their ballot.

For the first time in modern Australian history, a majority of votes were cast before election day, with COVID-19 accelerating a long-term trend away from election day voting.

There was also a narrowing of the gap between the Labor election day two-party-preferred vote and the early two-party-preferred vote. Labor won the election day vote in 2016 and 2019 and yet did not win the election. This year Labor has won the early vote for the first time since at least 2001.

Let’s start by looking at all of the different vote types I identified in 2022, and how the total vote is split between the different types.

Vote type Timing 2019 share 2022 share Change in share
Ordinary Election day 53.67 44.73 -8.94
Ordinary pre-poll Early 28.52 32.90 4.38
Postal Early 8.51 14.61 6.10
Declaration Pre-poll Early 4.17 3.62 -0.55
Absent Election day 4.05 3.16 -0.89
Telephone Covid Unknown 0.00 0.51 0.51
Provisional Unknown 0.32 0.31 -0.01
Remote Early 0.15 0.10 -0.05
Hospitals Unknown 0.58 0.02 -0.56
Telephone Unknown 0.01 0.02 0.01
Mobile Unknown 0.00 0.01 0.01

The biggest increase was in postal votes. Postal voting has been largely stable around 8-9% since 2013, but has now jumped to almost 15%. There was also an increase of almost 4% in the pre-poll vote.

It’s also worth noting that telephone voting for those isolating at home due to COVID-19 made up about one in every 200 votes.

The AEC doesn’t make it super easy to identify votes by these categories. They have a “vote type” table for each seat, but this collapses a majority of these categories into “ordinary”. Only postal votes, provisional votes, absent votes (out of area election day votes) and declaration pre-poll votes (out of area pre-poll votes). To identify hospital, remote, mobile and in-area pre-poll votes you need to match the polling places spreadsheet to results files, since that spreadsheet contains a “Polling Place Type” from 1 to 5. Telephone votes (both Covid and the pre-existing type) are included as pre-poll booths, even though they were open on election day.

For most of this blog post I have simplified the categories down to the early vote and the election day vote. You can see here how I have categorised each of these vote types. About 0.9% of the total vote was cast using a type of vote which could have been cast on the day or beforehand, and we don’t know for sure.

Next up, let’s look at how the timing of the vote varied by state or territory.

State Early Election day Unknown
NSW 48.40 50.91 0.69
VIC 54.12 45.03 0.86
QLD 57.49 41.88 0.63
WA 48.63 49.64 1.72
SA 43.27 55.53 1.20
TAS 37.95 61.18 0.88
ACT 49.29 49.78 0.93
NT 67.35 31.64 1.02
Australia 51.29 47.89 0.87

The Northern Territory has by far the highest proportion of early votes, with more than two thirds of votes cast early. This is mostly driven by a high rate of voters using remote booths in Lingiari. If you check out my pre-election blog post which includes a map of the early vote at the 2019 election, Lingiari was 72.9% early in 2019, and this time was 74.8% early, so not much of a change. The other NT seat, Solomon, ranks eleventh in the country. High, but not off the scale.

Queensland and Victoria are the only states with a majority voting early, but the early vote wins those states by a decent margin.

The election day win for NSW was very narrow, while in Western Australia neither the early vote or the election day vote racked up a majority.

South Australia and Tasmania, however, had much lower rates of early voting, around 38-43%. They tended to rank down the bottom in 2019, so this isn’t (just) a result of those states being less affected by COVID-19 than some others.

Next up, this graph shows the shift at the national level from election day to the early vote since 2001, with 2022 finally producing an early vote majority.

While it is a common theory that COVID-19 accelerated the shift away from election day voting, and this was certainly true of state elections held during the height of the pandemic, this chart doesn't show much evidence of an acceleration of the previous trend. The election day vote declined by 9.3% (of the total vote) in 2013, 9.5% in 2019, and 9.8% in 2022.

Now, how did these people actually vote? For simplicity I'm going to stick to the two-party-preferred vote, and look at it for Labor (although you can always determine the Coalition 2PP by subtracting the Labor 2PP from 100, so all trends are identical if in reverse).

Labor won the election day vote in 2016 and 2019, but lost the election. That would have been remarkable in past decades, but the declining importance of election day has made such an outcome possible. This time, they won the early vote narrowly. The gap between the election day and early vote was larger than ever before in 2019, with Labor polling 5.1% better on election day than before election day. That gap has narrowed to 3.7% in 2022. The Labor swing was largest on the postal vote, with the swing on pre-poll better than on election day, so Labor tended to pick up bigger swings with voting methods they have traditionally done worse with.

Finally, I've mapped out the share of the vote cast early per seat.

The early vote made up a majority of all votes in 86 seats. The election day vote held the majority in 58 seats, and in the other 7 neither reached a majority.

Mostly the trends on the map are not surprising, with Victoria and Queensland generally having a higher early vote. I thought it was interesting that some of the seats with particularly high early votes compared to their neighbours were the most competitive seats. The only seats in Sydney with a majority early vote were Fowler and Wentworth, while Gilmore had a very high early vote.

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


  1. @Ben – Thanks for your work over this election. Why aren’t the high early voting rates in QLD and VIC surprising?

    The hospital voting almost vanished to avoid the risk of exposing vulnerable patients to Covid-19, according to the ABC.

    I remember on 2019’s election night, a channel (I can’t remember which one) said Labor and Bill Shorten would get a higher 2PP result and would win based on exit polling. They weren’t wrong. Then the QLD and TAS results came in and then the pre-poll and postal votes came in and the rest is history. We may expect the death of exit polling soon, unless someone wants to do it over two weeks and for postal votes too.

    The AEC may reform the voting process to cut staffing costs by reducing election day polling places whilst increasing mobile voting and increasing pre-polling hours, if the trend to voting early continues.

  2. I think the full suite of vote types are probably in the “Detailed Verbose” Media Feed files, the zips for which are at
    Even the ZIPs are huge and the XMLs are too large for my Win7 machine to deal with.

    As I understand it from Christine Joy, Digital Director at AEC, the Commission will hold a review and may tweak the Feeds to cope with the changes that occurred in 2022

  3. It seems inevitable that postal voting will increasingly become the norm, kicked on by the covid experience. As a Western Australian over the age of 70, I am able to apply to become a “General Postal voter” (i.e. I receive ballots by mail automatically) for State elections, but nor Federal Elections, since I don’t meet the other criteria. The fact is that the convenience of postal voting is substantial (and carbon efficient!) but that the criteria are illogically limiting. There is no reason why anyone wishing to do so should not be able to apply for a postal vot, particularly in times when the vulnerable are seeking to minimise physical exposure.
    I have to say that once you’ve voted by post, you don’t ever want to go back! I can cook my own democracy suasage, thanks!!

Comments are closed.