# Glossary drafts for Getting It Right Proposed launch copy and editorial rules. All examples are invented for teaching. Definitions describe the content or reasoning; they do not establish anyone's intention. These drafts require the site's normal subject-matter and accessibility review before publication. ## Made-up material Slug: made-up-material Axis: content Suggested claim link: How we checked this material **Plain-language explanation:** Sometimes a quotation, document, result, or event is presented as real even though it was invented. The important question is what supports the material: an original recording, a genuine document, or a record that can be checked. **Example:** A post attributes a statement to a town's annual report. The complete report contains no such statement, and an authenticated record establishes that the supposed quotation was invented. **Try this:** Follow the reference to the original. Does it contain the quoted words or reported finding? **Limit:** Failing to find something does not prove it was invented. A missing record, an unfamiliar source, satire, fiction, and clearly labeled illustrations need different treatment. Use “not verified” if that is all the evidence establishes. **Editorial criterion:** Positive evidence of fabrication, or a sufficiently complete record that directly establishes the alleged item could not be what it claims. Identify exactly which material is fabricated. Do not infer intent of later sharers. **Related:** Wrong context; altered media. ## Wrong context Slug: wrong-context Axis: content Suggested claim link: Why the date or setting matters **Plain-language explanation:** A real image, quotation, or fact can support the wrong story when it is attached to the wrong date, place, person, or situation. **Example:** A photograph taken during a river flood five years ago is described as a photograph of yesterday's flood in a different town. The image may be real, but it does not document the event claimed. **Try this:** Check the earliest reliable source you can find. Compare its date, caption, and setting with the current claim. **Limit:** The earliest copy you find is not necessarily the original. Reusing an older image as a clearly labeled illustration is different from presenting it as current evidence. **Editorial criterion:** Establish the authentic context and the conflicting presented context, with source links. Explain how the mismatch changes the claim. **Related:** Missing context; altered media. ## Missing context Slug: missing-context Axis: content Suggested claim link: What changes when we add the context **Plain-language explanation:** An accurate detail can give an inaccurate impression when information needed to understand it is left out. The missing information must change a meaningful part of the conclusion. **Example:** A headline says a library cut its opening hours this month but leaves out that the building was closed for scheduled repairs during part of the month. That matters if the headline implies a permanent reduction in service. **Try this:** Ask which conclusion you are drawing, then check whether the omitted information changes it. **Limit:** Every summary leaves things out. A detail is not essential just because it is interesting or favors a different argument. **Editorial criterion:** State the likely inference, provide the omitted fact, and show the material change. Distinguish an explicit false statement from an implication requiring interpretation. **Related:** Wrong context; misleading numbers. ## Selective evidence Slug: selective-evidence Axis: content Alternative familiar term: cherry-picking Suggested claim link: What the wider evidence shows **Plain-language explanation:** A conclusion can look stronger than it is when it uses only the examples, studies, or dates that favor it. Look at how the evidence was chosen and what relevant evidence was left out. **Example:** A chart claims that a shop's sales are steadily rising by showing just three successful months from a year in which most months declined. **Try this:** Ask why these dates or examples were selected. Look for the full period or a review that explains how evidence was included. **Limit:** Choosing a relevant subset can be legitimate. One high-quality study may deserve more weight than several weak studies. Counting favorable and unfavorable studies is not enough. **Editorial criterion:** Define the relevant comparison set, explain why exclusions matter, and check alternative legitimate selection rules. “Selective” describes the presented evidence, not proof of deliberate concealment. **Related:** Missing context; misleading numbers. ## Misleading numbers Slug: misleading-numbers Axis: content Suggested claim link: Put the numbers on the same basis **Plain-language explanation:** A number can be correct while the comparison around it is unfair or unclear. Check what is being counted, out of how many, over what period, and using which units. **Example:** Town A reports 100 lost bicycles and Town B reports 50. If A has 20,000 residents and B has 5,000, those totals alone do not show that an individual resident faces a higher rate in A. Using population as a rough denominator gives 5 versus 10 reports per 1,000 residents; bicycle ownership would be a more directly relevant denominator if available. **Try this:** Ask “out of how many?” and “compared with what?” Distinguish counts, rates, relative changes, and percentage-point changes. **Limit:** Rates do not answer every question. Total reports can matter for workload. A population denominator can be inappropriate when exposure differs. Statistical uncertainty still matters after a comparison is normalized. **Editorial criterion:** Name the specific error: denominator, units, baseline, time window, aggregation, or uncertainty. Show a corrected comparison where the data permit one. Do not label a graph misleading merely because its axis does not start at zero; assess chart type and actual inference. **Related:** Selective evidence; unsupported causal claims. ## Unsupported causal claims Slug: unsupported-causal-claims Axis: content / inference Suggested claim link: Does this show what caused the change? **Plain-language explanation:** Two things happening together does not, by itself, show that one caused the other. Another factor, reverse causation, chance, or the way the data were collected may explain the pattern. **Example:** People who carry umbrellas are more likely to get wet than people who do not. Rain affects both umbrella use and getting wet; the comparison alone does not show that umbrellas cause wetness. **Try this:** Ask how the study separated the proposed cause from other explanations. What comparison would help distinguish them? **Limit:** A causal claim is not necessarily false because this evidence does not establish it. Randomized experiments can help, and strong observational designs can also support causal conclusions. No single design makes every conclusion certain. **Editorial criterion:** Identify the claimed cause and effect, the design's limits, and plausible alternatives that the evidence does not rule out. Use “not established by this evidence,” rather than converting uncertainty into a finding of no effect. **Related:** Misleading numbers; selective evidence. ## Altered media Slug: altered-media Axis: content Launch status: glossary entry available; add a distinct tag only when visual cases warrant it Suggested claim link: What changed in this image or recording **Plain-language explanation:** Editing or generating an image, audio clip, or video can change what it appears to document. The key question is whether the version presented supports the claim being made about a real person or event. **Example:** A recording is edited to join separate words into a sentence the speaker never said. **Try this:** Compare with a verified original or longer recording. Look for a credible account of how the material was produced. **Limit:** Cropping, subtitles, compression, color correction, and clearly labeled illustration can be legitimate. AI-generated does not automatically mean misleading, and missing provenance does not prove a fake. Automated detection alone is not a sufficient verdict. **Editorial criterion:** Establish the relevant alteration or synthetic origin and how the presentation changes meaning. When the media are authentic but the caption is wrong, use wrong context instead. ## Confirmation bias Slug: confirmation-bias Axis: cognition Use: learning link only, never a claim-verdict badge Suggested link: A check we can all use **Plain-language explanation:** We can give evidence an easier hearing when it fits what we already think, while demanding more from evidence that challenges us. This can affect what we search for, notice, remember, or accept. **Example:** While deciding whether a new route is faster, someone remembers the quick journeys and discounts the slow ones without recording either consistently. **Try this:** Before checking, write down what result would change your mind. Apply the same evidence standard to the answer you prefer and the answer you do not. **Limit:** Disagreement does not demonstrate bias. Existing beliefs can be well supported, and new evidence can be weak. This concept explains a possible process; it does not tell us whether a particular claim is true or why a particular person believes it. **Editorial criterion:** Link as a general self-check, including for the site's own editors. Never “you believe this because of confirmation bias.” Background: [Nickerson's review](https://journals.sagepub.com/doi/abs/10.1037/1089-2680.2.2.175). ## Repetition and apparent agreement Slug: repetition-and-apparent-agreement Axis: distribution, with a separate cognitive connection Use: learning page; case tag only with documented provenance **Plain-language explanation:** Seeing a claim in several places does not always mean that several sources independently checked it. They may all be repeating the same original report. Familiarity can also make a statement feel easier to accept. **Example:** Five websites repeat one unverified press release. There are five copies but only one underlying source. **Try this:** Trace the sources backward. Do they offer independent records or only link to each other? **Limit:** Repetition can spread accurate information too. It is neither proof of falsity nor proof of a coordinated campaign. **Editorial criterion:** Show the source chain rather than guessing it. Separate evidence of copying from a hypothesis about its psychological effect on any particular reader. ## Misinformation, disinformation, and lies Slug: misinformation-disinformation-and-lies Axis: definitions and intent Use: explainer, not routine badges **Plain-language explanation:** On this site, misinformation means false or materially misleading factual information, without assuming why it was shared. Disinformation means information deliberately used to deceive. Calling a statement a lie also makes a claim about the speaker: that they knew it was false and presented it as true. We can often check a claim more confidently than we can know the speaker's intention. When intention is not established, we describe the problem in the information and show the evidence. **Example:** Someone forwards an old photograph believing it shows a current event. The photograph can mislead even if the person sharing it is sincerely mistaken. **Limit:** Definitions vary between researchers. Uncertainty, an untested possibility, disagreement about values, satire, and honest revision are not automatically misinformation. A person can knowingly quote a false statement to correct it without endorsing it. **Editorial criterion:** The site's broad definition of disinformation is an explicit convention. Publish an intent attribution only with identified actors, direct supporting evidence, attribution where appropriate, and editorial review. Do not use motive labels to fill gaps in factual analysis.