Explainer: Interpreting Polling When Two Major Parties Become Three
For the first time in Australia’s federal political history, a minor party has attracted more than 20% support in national opinion polling.
In Fox & Hedgehog’s latest national poll conducted between 5-6 January 2026, One Nation recorded a historic 21% of the primary vote, the Coalition 25%, and Labor 29%. The combined major party vote has fallen to just 54 per cent, placing it at a record low levels for Australian federal politics.
Two-party-preferred (TPP), the uniquely Australian metric used to describe electoral competition between Labor and the Coalition, was not designed for this kind of political landscape.
In this explainer, we delve into what effect this has on the traditional TPP, and how to interpret the impact of a rising third party on Australian federal elections.
What TPP was built for
TPP exists because Australia combines compulsory preferential voting with a political system that, for most of the past century, was dominated by two large parties.
Under those conditions, the final count in almost every House of Representatives electorate came down to Labor versus the Coalition. Aggregating those contests into a national TPP figure produced a useful summary of the election as a whole.
For decades, this assumption held.
As recently as 1984, every single seat in the House of Representatives was a so-called “classic” contest between Labor and the Coalition. Minor parties and independents existed, but they rarely displaced either major party from the final count.
The rise of non-classic contests
That world has been steadily eroding.
As more voters have shifted away from the major parties, there has been a sharp increase in the number of electorates where the final contest does not involve both Labor and the Coalition.
In 2019, there were 15 non-classic contests.
In 2022, there were 27.
By 2025, that number had risen to 37, a quarter of all seats.
This was primarily a consequence of the record 34 per cent of Australians voting for a candidate outside the two major parties at the last election.
Despite this, TPP still described the national outcome in 2025. Labor won 55% of the TPP and secured 94 of the 150 seats in the House of Representatives. It was a clear and decisive victory.
So what has changed since then?
When TPP stops describing the election voters are actually voting in
TPP works best when two conditions are met:
Labor and the Coalition reliably finish in the final two across most electorates; and
Preference flows to those two parties are reasonably stable across seats.
When those conditions hold, averaging seat-level contests into a national TPP figure is an effective way to capture how minor party preferences shape overall electoral competitiveness.
The problem now is that which parties make it to the final two can vary widely from seat to seat.
With One Nation polling above 20% nationally, it is no longer just a preference source that generally benefits the Coalition. It is competing directly for second place in many electorates and, in some cases, for first.
Once a third party regularly displaces either Labor or the Coalition from the final count, a single national TPP figure stops reflecting the election Australians are actually voting in. It becomes a hypothetical construct rather than a description of real contests.
Australia has been here once before.
The 1998 Queensland State Election
At the 1998 Queensland state election, One Nation polled 23% of the vote and finished first or second in 34 of the state’s 89 seats.
Because Labor and the Coalition were no longer reliably the final two candidates, electoral analysts chose not emphasise the TPP result after the election. This is because the notional TPP figure was mathematically fragile and a poor predictor of outcomes.
That decision was not controversial. It reflected a simple reality; when the structure of the contest changes, the metrics must change with it.
The current federal polling environment shows similar characteristics.
In Queensland in 1998, Labor polled 39% of primaries, and the Coalition over 31%, finishing sixteen and eight points ahead of One Nation respectively. In January 2026, the gap between Labor and One Nation is just eight points, and only four between the Coalition and One Nation.
The next election is not scheduled for another two years, and many things can change between now and 2028. This current third-party surge could turn out to be a temporary anomaly. But if current polling reflected the results of the next election, widespread three-way contests would not just be possible, but likely.
The decline of the major party vote
It is important to be clear about what is driving this change.
The instability of TPP is not caused by preferential voting. Australia has used preferential voting for over a century.
It is caused by the collapse in the combined major party vote. As that vote share falls, the number of electorates where the identity of the final two candidates is uncertain increases.
When that uncertainty becomes widespread, national averages stop behaving predictably.
Introducing a Three-Party-Preferred (3PP)
In this environment, Fox & Hedgehog has chosen to publish a national ‘Three-Party-Preferred’ (3PP), designed to act as a measure of which parties are most likely to appear in the final two across a large number of seats.
This is done using respondent allocated preferences. Voters for Greens, independents and other minor parties are asked which of Labor, the Coalition or One Nation they would prefer, if forced to choose between those three options. Their answers are then combined with primary votes for the three largest parties.
Once those preferences are allocated, our 5-6 January 2026 poll produces the following 3PP:
Labor: 46%
Coalition: 29%
One Nation: 25%
These figures are not a projection of seats, and they do not necessarily predict who will form government.
They answer a simpler question: which parties are most likely to finish in the top two across the country?
On these numbers, Labor would appear in the final two in the vast majority of electorates. Who Labor faces in individual seats would be more evenly divided between the Coalition and One Nation, with the Coalition retaining a modest advantage.
Why there is no single TPP right now
Once three parties are viable, there is no longer a single TPP that can summarise the whole election.
Different electorates will produce different final match-ups depending on which candidate is excluded first. This makes the question “so who is winning at the moment?” more complicated to predict.
That is why Fox & Hedgehog also reports multiple TPP scenarios, which we call “TPP Match-ups”, based on respondent allocated preferences between each pairing of the largest parties, and using the same distribution logic as the 3PP measure. The national TPP Match-up figures from our 5-6 January 2026 poll are as follows:
Labor vs Coalition: 53 to 47
Labor vs One Nation: 56 to 44
Coalition vs One Nation: 63 to 37
What these numbers show is that while the Coalition is more exposed to finishing third, it gains a significant preference advantage if it reaches a final count against One Nation, rather than Labor. Labor also benefits in a match-up against One Nation rather than the Coalition, but to a lesser extent.
Each of these contests can, and likely will, exist simultaneously across the country. Publishing only one implies a uniformity in the Australian electorate that is not currently present.
In inner metropolitan seats, Labor versus the Coalition remains likely. In outer suburban seats, Labor versus One Nation becomes more probable. In some regional and rural seats, Coalition versus One Nation contests are possible.
Why preference exclusion order now matters as much as national averages
Under compulsory preferential voting, the order in which candidates are excluded determines which preferences are distributed, and to whom.
When three large parties are competitive, small differences in primary vote rankings can produce very different outcomes, even when overall vote shares are similar.
This is why preference exclusion order now matters more than any single national average.
Examples: How similar votes can produce different outcomes
The examples below use simplified, illustrative scenarios to show how exclusion order, rather than overall support, determines outcomes.
Example 1: Labor vs Coalition
In this example, Labor finishes first, the Coalition finishes second, and One Nation finishes third.
One Nation is excluded, and our data says most of their preferences would flow to the Coalition. If the three parties all have similar primary votes, the Coalition most likely overtakes Labor and wins the seat, despite Labor leading on first preferences.
This is a classic contest, but it only occurs because One Nation finishes third.
Example 2: Labor vs One Nation
Assume the overall vote shares are similar to Example 1, but One Nation finishes second and the Coalition finishes third.
The Coalition is excluded, and according to our data its preferences split slightly more evenly between One Nation and Labor. One Nation likely still overtakes Labor, but by a narrower margin than the Coalition did in Example 1.
Despite the combined conservative vote being similar, the outcome is different because the identity of the excluded party has changed.
Example 3: Coalition vs One Nation
In this scenario, One Nation finishes first, the Coalition finishes second, and Labor finishes third and is excluded.
Our data suggests Labor preferences flow heavily to the Coalition, allowing the Coalition to overtake One Nation and win the seat, even though One Nation led on first preferences.
Here, One Nation performs worse in the final count than in Example 2, despite having a higher initial vote, purely because preferences behave differently when One Nation is against the Coalition rather than Labor.
These examples illustrate the core issue. The same electorate, with similar vote shares, can produce different winners depending on who is excluded first.
That is why a single national TPP figure cannot currently describe the election as a whole.
What this means for interpreting polls in 2026
In a genuine three-party environment, no single headline number can do all the work.
3PP shows who is structurally competitive;
Party-by-party TPP Match-ups show how different contests behave; and
Seat outcomes depend on local primary vote rankings, not national averages.
The traditional Labor vs Coalition TPP remains relevant, but in Australia’s current political moment, it has become one piece in a much larger story.
If Australian politics remains fragmented, we believe polling should try to describe that complexity honestly, even if it makes the story harder to distill into a single headline number.