Preference flows – dive into the map


The Australian Electoral Commission finalised the results of the 2019 federal election last week (now available at and this included the publication of data showing how primary votes for each candidate flowed on a two-candidate-preferred basis, as well as two-party-preferred flows for each party at a state and national level.

At a national level, the most interesting figures are preference flows for the Greens, Clive Palmer’s United Australia Party and Pauline Hanson’s One Nation. These three parties polled between 3.1% and 10.4%, and were the only minor parties to crack 1% in the House of Representatives.

82.2% of Greens preferences flowed to Labor on a two-party-preferred basis, up 0.3% percent compared to 2016. Both UAP and One Nation voters’ preferences flowed to the Coalition at a rate of about 65%. This was a big shift for both of these parties. Just over 50% of One Nation preferences flowed to the Coalition in 2016, while voters for the previous incarnation of the Palmer United Party only preferenced the Coalition over Labor 54% of the time in 2013.

There was a lot of conjecture during the previous term of government about how best to allocate these voters’ preferences in polling calculations. Typically pollsters rely on how preferences flowed at the previous election to allocate primary votes for minor parties and independents in determining the two-party-preferred figure. This is easy enough to do for Greens voters, who are reasonably consistent in their preference flows. But it’s hard when a new party like United Australia emerges. There was also reason to suspect that One Nation’s preferences would flow much more strongly to the Coalition than they did in 2016. These concerns led to some pollsters tinkering with their formulas.

In the end they were right to be concerned, as these preferences did flow more strongly in the past. I won’t try and explain it in full, but in the case of the UAP this may partly be explained by Clive Palmer’s campaign being much more right-wing in 2019 compared to his centrist positioning in 2013, which may have attracted a different kind of voter this time around. These two parties contributed to a significantly higher share of the preference flow for the Coalition than in 2016.

Below the fold I have put together a map showing the preference flow from the Greens to Labor in each seat, as well as a detailed booth map showing the preference flows for each of these three smaller parties for each booth where the data exists.

First, the seat map. Probably the most interesting pattern is that the Greens preferences flow much more strongly to Labor in inner-urban areas, as well as more broadly across Melbourne and Tasmania and the south-eastern corner of New South Wales.

I find this second map more interesting, and I think there’s a lot more to view on this map. By default it shows Greens flows, but can be toggled to show UAP or One Nation preference flows to the Coalition.

The numbers are colour-coded according to how strong the preferences flowed to either Labor or the Coalition, with the numbers representing the preference flow to Labor (Greens) or Coalition (UAP/One Nation). Numbers are scaled to represent the relative number of raw votes cast for that minor party at that booth (although scales do not match between each layer, since each party had a different total vote).

There’s a lot to explore, but in particular I find it interesting how Greens preferences flow much more strongly to Labor in the parts of Sydney where they have more of a presence and a higher vote. This may have some relationship to the size of the vote, but I also suspect it is influenced by the relative scarcity of Greens how-to-votes being handed out in Western Sydney.

There’s a lot more in the data of interest, please post in the comments what you find most interesting.

Elsewhere Kevin Bonham has posted his analysis of the final election results including the preference flows.

I’m working on a few other data projects which will be posted over the next few weeks but I will definitely return to the final election results with more posts soon enough.

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


  1. Excellent work, thank you. You have also alerted me to the full download of two-party preference flow per polling booth.
    The National preferences in the WA seats of Durack and O’Connor show a pattern of 90%+ flow to Liberal in the small farming boxes, 80% in larger rural towns and considerable leakage in “non-traditional” boxes such as the Eastern Goldfields and the Pilbara.

  2. On the preference flows, I think you’re looking at part ignorance across the nation, part whimsy in regional areas (they don’t have a hope of winning anyway so why not ‘do it for a laugh’) and tactical voting in SA (they’re probably preferencing CA above Labor on fears that *other* CA preferences could flow to the Coalition should they not get the 1st or 2nd spot). Sydney and Melbourne are different; I think a lot of those weird flows can be explained by minority concerns. The Greens are very visible as defenders of immigrants and minorities, but those same minorities often have conservative social values, which the Liberals are much more accommodating to. So Labor’s both too left and not left enough for them. There’s also the possibility that these people are either less educated or less aware of Australian politics in general and their vote doesn’t really make any reasonable sense. Some would argue that’s the case for any Greenie that preferences the Libs.

  3. Buttered Scone quite true both Liberals and ALP are too right on some issues and too left on others. This is why my vote is dependent upon who is standing and my preferences are rarely finalised till last few days of campaign. It s always easier to decide who will be last ie the fascists and the communists in their modern day fancy dress uniforms.

  4. Do you have a spreadsheet of the booth by booth results? It’d be great to look over that data, as I can’t easily differentiate all the booths in my seat (Hunter)

Comments are closed.