This website is dedicated to a probabilistic model of First Preference votes for each of 1126 candidates across 150 electorates in the 2025 House of Representatives elections, as well as a forecast of the party composition of the next parliament. Its purpose is to measure and describe the uncertainty inherent in Australian election outcomes before an election takes place, and to describe which sets of outcomes are more likely than others. As such, it is worth paying more attention to how far the confidence bands stretch (the bulky sections of the violin plots) rather that being pre-occupied with the average estimates. The website is set out in such a way to place emphasis on understanding probabilities and distributions of outcomes, rather than on ‘making predictions’.
This First Preference model is the work of Dania Freidgeim, a Melbournian Masters student and elections researcher. It was born out of an aspiration to develop a nice mathematical framework for calculating the effects of the 2024 Redistributions on seat margins, which led to exploring different permutations of candidate changes, how preferences flow between them, and eventually to the much more statistical task of simulating polling errors and producing distributions of votes! Antony Green‘s years – or should I say decades – of commentary and analysis of redistribution changes was where the idea of developing a model sprouted from. I am grateful for the assistance of William Bowe, Kevin Bonham and Ben Raue in sharing various historical polling and redistribution data that proved invaluable in strengthening the model’s database, as well as election forecaster aeforecasts, who has an excellent model of their own (which was a great inspiration in the later stages of this project).
The author is not associated with any political party, and the model content on this website is for information purposes only.
Please reach out if you have any questions about the model, the methodology being used, or Australian elections in general! You can email me at daniel.freidgeim@gmail.com. I also intend to publish the model’s full code on the website in due time, once I can get it tidied up a bit.