A three-tier framework: where bias comes from, how it's built into text, and what it does to the reader. Explore each section, then test yourself.
Political bias occurs when an outlet or journalist has an ideological alignment — left, right, centrist, libertarian, nationalist — that shapes which stories get told and how. This is the type everyone knows exists, but few know how to detect systematically. It can be explicit (an endorsement editorial) or subtle (tone, word choice, which experts are quoted, which stories lead).
Commercial bias is the influence of profit motives and advertiser pressure on news content. News organizations depend on advertising revenue and audience numbers to survive. This creates pressure to produce stories that attract attention — sometimes at the expense of depth, accuracy, or public importance. Advertisers may also shape coverage through direct pressure or through the "chilling effect" of self-censorship.
Ownership bias is related to — but distinct from — commercial bias. It's about who controls the outlet: a billionaire owner, a multinational corporation, a political party, a government, or a private foundation. Ownership shapes editorial priorities even without explicit orders being given. The owner's interests, values, and business relationships become the unspoken parameters of acceptable journalism.
Structural bias refers to systemic patterns baked into how journalism has historically operated — who gets to become a journalist, whose neighbourhoods get covered, whose voices are treated as authoritative expert sources. This isn't about any individual journalist's intent. It is the accumulated result of decades of decisions about hiring, training, coverage geography, and who is assumed to be the default audience.
Individual journalists bring their own lived experience, education, cultural background, and assumptions to every story they write. This isn't necessarily malicious — it's human. But it shapes what questions get asked, who gets called for comment, what details get noticed, and what gets treated as the "normal" baseline requiring no explanation. Every journalist has blind spots, and the best ones know it.
Journalism tends to treat existing power structures as the default normal — a baseline that requires no justification. Challenges to the status quo are automatically framed as radical, risky, extreme, or disruptive — while defending the status quo requires no justification at all. This asymmetry is built into standard journalistic convention, not just individual choice.
Selection bias is the first and most fundamental technique: which stories get covered at all. Every day there are thousands of newsworthy events; editors choose a fraction. Those choices reflect the values, interests, and assumptions of newsrooms. The stories that are never told can be as revealing as those that are — because what's absent from coverage becomes absent from public consciousness.
Bias by omission occurs when a story leaves out information that would change how a reader understands it. The story may be entirely accurate as far as it goes — but what's left out makes it misleading. This is one of the most common and hardest-to-detect techniques because it requires knowing what isn't there.
Placement is an editorial signal. Where a story appears — front page above the fold, or buried on page 11 with no photo — tells readers how important the outlet considers it. Stories can be covered while simultaneously being marginalized through placement. Placement decisions are often where ownership and commercial pressures exert their most direct influence.
Framing refers to how a subject, group, or event is described and contextualized. Labelling is the specific words used to identify people and groups. Neither is neutral — every framing and label carries assumptions, history, and connotation. The choice to describe the same person as a "freedom fighter," a "militant," or a "terrorist" is a political choice, even if it's not recognized as one.
Loaded language refers to words that carry emotional charge or political connotation beyond their dictionary meaning. These words prime the reader to feel a certain way before they engage with the facts. Unlike framing — which shapes the overall narrative structure — loaded language operates at the level of individual word choices that trigger associations, emotions, and assumptions.
Images are processed faster and remembered longer than text. The photo an outlet chooses to accompany a story — which moment it captures, how it's cropped, what angle it's taken from — tells a story independently of the words. Image selection can make the same event look threatening or benign, make a crowd appear large or small, make a person appear sympathetic or dangerous.
Research consistently shows that most people only read headlines — they don't click through to the body. This makes headline construction one of the most powerful bias techniques available. A headline can misrepresent a nuanced story, create false impressions through questions or passive voice, and reach millions of readers who will never encounter the correcting detail buried in paragraph seven.
Bias doesn't announce itself. It is built into the structure of language, the choice of images, the ordering of paragraphs, and the questions that are never asked. These are the craft-level tools through which the sources of bias in Tier 1 become the reader effects in Tier 3.
Gatekeeping is the institutional layer above individual journalists. Editors, producers, and — increasingly — algorithms decide what reaches the public. A story can be reported, written, and filed — and still be killed before publication. Source bias is about who journalists actually call: the same think tanks, government spokespeople, and university professors, over and over. Communities without media-trained spokespeople get quoted less, or not at all.
Sensationalism is a reader effect — it describes what happens when bias techniques successfully trigger emotional response over critical analysis. Fear, outrage, and shock replace nuanced evaluation. Sensationalism often distorts a reader's understanding of actual risk and frequency. Rare but dramatic events feel common; common but undramatic problems become invisible. It connects to both commercial pressures (engagement drives revenue) and algorithm design (outrage drives sharing).
Confirmation bias is a reader effect — the tendency to accept, share, and remember information that confirms existing beliefs while subjecting contradicting information to much harsher scrutiny. Understanding this is not just about critiquing media — it's about critiquing our own reading habits. No matter how critical a reader you believe yourself to be, this cognitive tendency affects you. The research on this is unambiguous.
Agenda setting theory, first articulated by Maxwell McCombs and Donald Shaw in 1972, makes a precise and important claim: media doesn't tell us what to think — but it powerfully shapes what we think about. The issues that receive sustained coverage become the issues the public considers important. Those that receive none seem unimportant, regardless of their actual significance.
Algorithmic bias is a form of media bias that lives entirely outside the newsroom. Platforms like TikTok, Instagram, YouTube, and X determine what gets amplified through recommendation systems optimized for engagement — not accuracy, balance, or public interest. The result is a media environment where the most emotionally provocative content is systematically boosted, and where each user's information diet progressively narrows into a personalized filter bubble.
Six real-style headlines. For each one, identify the primary technique or source of bias at work. There are often multiple — identify the most dominant one and explain your reasoning.