Changing voting patterns slow down declaration of seats


Most people who read this website would have a reasonably good sense of the formal process of counting ballots: the counting of a primary vote, a two-candidate-preferred count, and the full distribution of preferences, where each candidate is excluded one-at-a-time until there are only two left. The AEC eventually will publish all of this data for every seat, even where a candidate has won the seat on primary votes, making preference flows redundant.

But all of this data remains a work in progress for now, and will be for some time to come. Yet the Electoral Commission has another job: to declare the winners for each electorate, so that MPs can take their seats in parliament.

The AEC announced today that the declaration process will take longer because increased votes for candidates outside of the top two. In this post I will map out how the declaration process has been radically changed by the increasing minor party and independent vote.

The declaration process doesn’t need to wait for all of the data to be finalised. Rather the AEC is free to declare a seat once it is mathematically impossible for the result to change.

If a candidate has a majority of the primary vote, and there isn’t enough votes left over for that outcome to change, you can declare.

Likewise, if the winner on the two-candidate-preferred count is clear, and it’s not mathematically possible for another candidate to make the top two in the formal distribution of preferences, it’s possible to declare the result prior to the formal distribution.

But if it is possible for another candidate (say, the third-placed candidate) to gain enough votes to overtake the second-placed candidate, that could mean that the two-candidate-preferred count is irrelevant, and the outcome could be different. In those cases, a formal distribution must be conducted prior to declaration.

Using an example from @caracal_freedom on Twitter, if the primary votes are:

  • LIB – 49
  • ALP – 23
  • GRN – 13
  • ON – 8
  • UAP – 7

It is theoretically possible that ON and UAP preferences could push the Greens into second place, and then elect the Greens on Labor preferences. For us analysts, we know that won’t happen, but the standard for declaring the result is higher. They need to actually conduct the full distribution and find out for sure.

Which brings me to today’s press release from the AEC:

The AEC stated that up to 75 seats may require a full distribution to declare the winner, more than double last time.

I looked at primary vote figures from the last seven elections and confirmed this figure. On current figures, almost half of all seats require a full distribution. On the latest figures, there are just 15 seats where the leading candidate has a majority of all formal votes (which is down from 18 seats earlier in the count, and around 50 seats at the last three elections).

This is a radical change from elections in the 2000s. In 2004, 89 out of 150 seats were decided on primary votes, and the top two was clear in another 57. Only in four seats was there any mathematical uncertainty about the winner at the end of the primary vote and two-candidate-preferred count. This shift has taken place gradually, but the change massively accelerated in 2022.

To compare 2022 to 2019, just 33 seats required a full distribution in 2019, which was more than double the 16 seats in 2013, but is very little compared to the ~75 we expect to require a full distribution in 2022.

I have been looking forward to getting two-party-preferred data for the 25 or so seats where the two-candidate-preferred count is not between Labor and Coalition, but I can see that this will take longer, since the AEC has a lot of work to do to make sure every seat is declared by the deadline on June 28. I'm sure they'll do their best in more challenging circumstances.

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


  1. Hi Ben,

    Why does the AEC perform a DOP even when it is unnecessary in the determination of the winner ?

  2. During Election Night, the AEC issued thirteen “TCP alerts” to its MediaFeed subscribers for the seats of:

    About a week later, the AEC re-jigged its MediaFeed to encode these seats as Mavericks, which I define as
    • Maverick seat: A seat for which the Australian Electoral Commission [AEC] has made an error in guessing in advance the two candidates in the final contest. This information is kept secret until after polling booths in the seat have closed, but is made available, under oath, to analysts at a rehearsal run on the Thursday night before the election. Mavericks emerge during Election Night counting and usually there are usually only 1 or 2. In 2022, there were 13 of them.

    As it happened, only two of the Mavericks were Teal seats: Wannon and Bradfield. That is to say, the AEC pretty much got it right in predicting the TCP contestants in the Teal seats. In the past the TCP contestants would have been based on the previous Federal Election Results.

    But, the AEC also has categories of seats they call “Classical” and “non-Classical”, which I define as:
    • Classical Seat: A seat in which the “final contest” boils down, after distribution of preferences from other candidates, to a contest between a candidate of “Coalition” (LNC) and a candidate of the Australian Labour Party (ALP);
    • Non-classical seat A seat in which the final contest is NOT between an LNC candidate and an ALP candidate;

    In 2019, there were 14 non-Classicals and in 2022, there were 29 … 3 having disappeared , with 14 new appearances.

    By and large the “newbies” were seats in which the Teals and/or Greens either won or were left standing at the final cut-up

    I am writing up a short paper on the consequences of all of this and how the original source of it seems to lie with Voices for Indi (V4I).

  3. Nigel McGuinness, the answer to your question is: the AEC conducts a full distribution of preferences because it is required by law to do so.

    Before 1984 it was not legally required to do so, and counts of preferences were conducted only up to the point required to determine a candidate with the absolute majority of the votes cast. The present requirement was part of the package of amendments to electoral legislation introduced by the first Hawke government.


Please enter your comment!
Please enter your name here