Types of bias
Every category unslant analyzes, organized in groups, with a short explanation of each.
To see how these categories become a 0–10 score, read the Methodology.
Language
Word choices that push the reader toward a conclusion.
Analyzed in: news and articles, academic papers, and claims
- Loaded / emotional language
- Words with strong emotional or judgmental weight instead of neutral terms.
- Labels and characterization
- Labels that carry a built-in judgment (e.g., 'regime' instead of 'government').
- Euphemisms and dysphemisms
- Softening something for one side (euphemism) or harshening it for another (dysphemism).
- Asymmetric attribution verbs
- 'Stated' for one side and 'claimed' or 'admitted' for the other.
- Intensifiers and generalizations
- 'All', 'always', 'never', 'no one' used without support.
Framing and structure
How the story is built: who is the victim, who is to blame, what comes first.
Analyzed in: news and articles, academic papers, and claims
- Framing (story frame)
- The chosen narrative defines victims, culprits and 'the problem' before the facts do.
- Headline vs. body
- A sensational headline, or one that the body does not support.
- Emphasis and placement
- What appears first gets attention; what is pushed to the end disappears.
- Spin
- Neutral facts presented in a favorable or unfavorable light.
- Visual framing
- Photos, illustrations or captions that steer how the subject is read.
Sources
Who gets heard, who doesn't, and how quotes are used.
Analyzed in: news and articles, academic papers, and claims
- Source selection and balance
- Only one side is heard, or one side gets far more space.
- Anonymous or vague sources
- 'Experts say', 'analysts believe', with no name or reference.
- Selective or out-of-context quotes
- Quotes trimmed in a way that changes their meaning.
- Source credentials
- Opinions presented as expert views when they are not.
Omission and selection
What was left out that would change how the story is understood.
Analyzed in: news and articles, academic papers, and claims
- Relevant context omitted
- Missing context that would change the interpretation (history, comparison, causes).
- Missing perspectives
- Affected groups or relevant viewpoints that do not appear.
- Selective facts
- Only facts that favor one narrative are presented.
Data and statistics
Numbers that appear in the text used in misleading ways or that don't support the conclusion. Missing data counts under "Omission".
Analyzed in: news and articles, academic papers, and claims
- Misleading numbers
- Numbers without a baseline, or absolute figures where relative ones matter (and vice versa).
- Correlation as causation
- Two things that happen together presented as cause and effect.
- Cherry-picked timeframe
- A hand-picked time window that shows the desired trend.
- Unsourced data
- Numbers or charts with no indication of where they came from.
- Misleading chart or table
- Charts and tables that distort the data or say something different from the text.
Fact vs. opinion
When opinion is dressed up as fact, or a news story editorializes.
Analyzed in: news and articles, academic papers, and claims
- Opinion presented as fact
- Value judgments written as if they were objective findings.
- Editorializing in news
- The reporter inserts opinion into a story meant to inform.
Emotional appeal and sensationalism
Devices meant to provoke fear, anger or urgency rather than inform.
Analyzed in: news and articles, academic papers, and claims
- Clickbait
- Headlines or lines built to drive clicks through mystery or exaggeration.
- Fear and outrage
- Text built to provoke fear or outrage.
- Artificial urgency
- 'Now or never', dramatic deadlines without basis.
Logical fallacies
Arguments with flawed reasoning.
Analyzed in: news and articles, academic papers, and claims
- Straw man
- Distorting the other side's position to make it easier to attack.
- False dilemma
- Presenting only two options when more exist.
- Ad hominem
- Attacking the person instead of the argument.
- Slippery slope
- Suggesting one step will inevitably lead to disaster.
- Appeal to authority
- Using someone's prestige in place of evidence.
- False equivalence
- Treating positions with very different evidence as equal.
Interests and conflicts
Disguised advertising and conflicts of interest visible in the text.
Analyzed in: news and articles and academic papers
- Disguised sponsored content
- News-like text that promotes a product, company or cause.
- Conflicts of interest
- Interests of the author or outlet that may shape the text.
Representation
Stereotypes about groups of people.
Analyzed in: news and articles, academic papers, and claims
- Gender stereotypes
- Roles or traits attributed to someone because of gender.
- Racial or ethnic stereotypes
- Stereotyped associations with race or ethnicity.
- Class stereotypes
- Stereotyped portrayals of the poor, rich or workers.
- Regional stereotypes
- Prejudice about regions, states or cities.
- Religious stereotypes
- Generalizations about followers of a religion.
- Nationality stereotypes
- Generalizations about nations or immigrants.
Reasoning biases
Mental shortcuts that lead to unsupported conclusions, often shaped by what a person reads and watches.
Analyzed in: claims
- Hasty generalization
- Drawing a general rule from a few cases.
- Anecdote as proof
- Using a personal or second-hand story as if it proved a trend.
- Availability bias
- Believing something is common because it shows up a lot in the news or on social media.
- Confirmation bias
- Considering only what confirms what one already believes.
- Unfalsifiable claim
- Phrased so that no evidence could confirm or refute it.
- Slogan or ready-made narrative
- Repeats a stock phrase that circulates in a group, outlet or network, without examining it.
Academic paper issues
Typical problems in scientific research. Only shown when the text is a paper.
Analyzed in: academic papers
- Funding and conflicts of interest
- Who funded the study and any declared conflicts.
- Sampling bias / small sample
- A sample too small or unrepresentative for the conclusions drawn.
- Conclusions overreach results
- Abstract or conclusion claim more than the data show.
- Causality from observational data
- Cause-and-effect conclusions from a study that only observes associations.
- Signs of p-hacking or selective analysis
- Many comparisons, with only the 'significant' ones highlighted.
- Selective citation / confirmation bias
- Only studies that agree with the hypothesis are cited.
- Unacknowledged limitations
- Important weaknesses of the study go unmentioned.