Raw Aus Pol

← Media bias

They decide what you think about

The biggest bias isn't a slanted headline. It's which stories run at all, and how much space each gets: weeks on a minor scandal while something worth far more to the public barely rates a mention. Repetition does the same job. The Herald Sun's good news for Victorian Labor came as one-off headlines, while its bad-news storylines (the Suburban Rail Loop, lockdowns, the "red shirts") came back month after month (see the study).

A year of attacks, then a fortnight of balance

Most voters have decided before the final week: in a 2025 post-election survey by JWS Research, 26% decided in the last week and 16% in the last couple of days (source). So coverage in the months before a campaign does more to shape a vote than balance in its final weeks. We compared Victorian politics headlines before and during the 2026 campaign.

Headlines unfavourableTo Labor, beforeTo Labor, final weeksTo the Coalition, beforeTo the Coalition, final weeks
Australian Financial Review64%64%5%18%
Herald Sun62%64%6%5%
The Age55%43%4%13%

Before: 1 Nov 2025 to 31 Aug 2026, the 1st and 15th of each month. Final weeks: 1 Oct to 19 Nov 2026, every third day. 433 headlines from the library's newspaper archive. One rater, not blind to the party; a second rater agreed on 70% for tone towards Labor and 93% for tone towards the Coalition. The campaign itself creates Coalition policy to cover, so this shows the timing, not the motive.

Does The Australian set the agenda?

The claim is that a story runs in The Australian and the next day it's everywhere, even in progressive outlets. We checked every political front-page exclusive in The Australian on the 1st and 15th of each month, November 2023 to October 2026, and looked for the same story elsewhere within three days. As a comparison, we did the same for The Age and the Sydney Morning Herald.

Front-page exclusivesFollowed by anyone in 3 daysFollowed by the ABC, Nine or Guardian
The Australian (65)12%6%
SMH and The Age (54)24%15%

Follow-ups were found by keyword searches of the library's newspaper archive and the Guardian's daily headline lists, so both groups are probably under-counted; the comparison between them is more reliable than the totals. TV, Sky News and talkback radio weren't checked. One rater.

The evidence: what got covered

Bias isn't only how a story is framed. It's also which stories run at all. In 1963 the political scientist Bernard Cohen wrote that the press may rarely tell people what to think, but is "stunningly successful in telling its readers what to think about". friendlyjordies makes this the heart of his live show Cancels the Media, from 45 minutes in. The idea goes back to Walter Lippmann's 1922 book Public Opinion (background).

It has been tested. During the 1968 US presidential race, Maxwell McCombs and Donald Shaw asked undecided voters in Chapel Hill, North Carolina, which issues mattered most. Their ranking almost exactly matched how much coverage each issue got (the 1972 paper).

His Australian example is 2019, when the campaign ran on tax. The "death tax" Labor was supposedly planning didn't exist. AAP FactCheck found the Prime Minister fanning it, and First Draft traced how it spread. But knocking a scare down still means talking about tax, which is the point. Our 2025 sample has the echo: The Age's "Labor challenges last-minute scare on super tax".

He also says that if you read The Australian today, you'll know what everyone else covers tomorrow. That's called intermedia agenda-setting. We can't test it yet, because The Australian blocks our tools.

What's at stake in the stories they run

Every story sorted by what it's really about. Accountability is how public money, assets and power are used or misused: spending blowouts, assets sold off cheap, contracts, grants rorts, donations, tax. Policy is what governments actually do or promise. The contest is polls, leadership and party fights. Conduct is a politician's personal behaviour or a small scandal with little public money in it. Stories were sorted without the sorter seeing which outlet ran them, and four sorters given the same 50 stories agreed on 45.

What each outlet covered

The share of each outlet's political stories on each topic. Each story gets one topic, using keyword rules you can check below. A policy beats the contest: a pledge for new schools counts as education, not the campaign. Highlighted cells are where an outlet gave a topic at least half again its share across all outlets.

Election time vs the rest of the year

Using every sample since 2021, regardless of the period picked above. Each period has up to 40 political stories per outlet: the first ones each outlet published on five set days a week apart (the order its own listing gives), in August 2021, July 2022, July 2023, July 2024 and August 2025, and in the 2022 federal, 2022 Victorian and 2025 federal campaigns. Each cell shows an outlet's top three topics, with the share of its stories on the contest itself (leaders, polls and party fights) alongside.

Who the stories are about

The share of each outlet's stories mainly about each party. Labor is in government federally and in Victoria, so it gets the most coverage everywhere. What's worth looking at is the gaps between outlets.

The keyword rules, and their limits

Each sample is the first political stories an outlet published on our sample days, so it reflects what the outlet chose to run. We look for these words in the headline first, then in our summary of the story. The first topic that matches wins, and "Leaders, polls and party fights" only applies when no policy topic does. Keywords are crude, and a few stories will be misfiled, but anyone can check the rules. The Telegraph and Herald Sun are headline-only.