A Six-Pass Method for Checking Online Claims Before You Share

A polished page, a confident narrator, and thousands of reposts can make an online claim feel settled before anyone has checked it. None of those signals establishes that the claim is true. A better response is procedural: identify exactly what is being asserted, leave the page to investigate the source, trace the evidence, recover missing context, inspect the media, and state only the conclusion the evidence supports. The six pass method below is designed for everyday claims in posts, videos, screenshots, articles, and forwarded messages. It is not a substitute for a specialist investigation, and it does not guarantee certainty. Its purpose is to slow a claim down long enough to separate what is observable from what is merely persuasive. Pass one: write the claim in a form that can be checked Start by reducing the post to one precise sentence. Separate the central factual assertion from commentary, prediction, humor, and moral judgment. Record who or what is said to have done something, where it allegedly happened, and when. If the claim uses quantities, preserve the unit, population, and time period rather than shortening it to a dramatic percentage. Then ask what evidence would actually resolve that sentence. A claim about a law calls for the enacted text or an official record, not a reaction video. A claim about a study calls for the paper, methods, and data limits, not only a press release. A quotation calls for the full recording, transcript, or original document. A breaking event claim may not yet have enough evidence for a firm judgment. This step prevents a common verification failure: researching a nearby topic while leaving the original assertion untouched. It also exposes unfalsifiable language such as "everyone knows" or "they do not want you to see this." If the claim cannot be stated clearly, do not upgrade it into a fact by sharing it. Pass two: leave the page and investigate the source laterally Do not spend the first several minutes studying a source's own About page, design, or testimonials. Open new tabs and see how independent, knowledgeable sources describe the person, outlet, organization, or domain. The Digital Inquiry Group's Civic Online Reasoning curriculum calls this lateral reading: moving away from the original page to investigate who is behind it and whether other sources support its standing. Search the exact organization or author name with terms tied to the relevant field. Look for ownership, funding, editorial controls, recognized expertise, prior work, and documented conflicts. Distinguish a primary source with direct access to the event from a commentator repeating someone else's account. A source can have relevant access without being neutral, and a neutral sounding source can lack access entirely. Use click restraint when search results appear. Scan result snippets and domains before choosing a page, and do not assume the top result is the strongest evidence. Advertising, search optimization, popularity, and personalization can affect what appears prominently. Prefer sources whose authority matches the claim: an election authority for official results, a court docket for a filing, or the research institution and paper for study details. Pass three: follow the evidence chain to its origin Treat every citation as a route, not a decoration. Open it. Check whether it contains the fact attributed to it and whether the wording, date, and scope match. If an article cites another report, continue until you reach the earliest available record: the dataset, filing, full interview, official notice, original image, or peer reviewed paper. Evaluate evidence independently of the person presenting it. Is the document complete? Is the sample appropriate to the conclusion? Does a chart begin at a misleading baseline or combine unlike categories? Does the source disclose uncertainty? Could the apparent confirmation be multiple outlets repeating the same unverified post? The News Literacy Project's RumorGuard framework asks five useful questions about authenticity, source, evidence, context, and reasoning. Apply all five. A genuine document can be used to support a false interpretation; a credible outlet can make an error; real evidence can be too weak for the conclusion; and valid premises can still be connected by faulty reasoning. When access to the underlying evidence is blocked, label that limitation. "The post links to a report I could not inspect" is more accurate than treating the citation as verified. Screenshots of headlines, charts, or messages are leads to investigate, not self authenticating records. Pass four: restore time, place, sequence, and scale Many misleading posts use authentic material in the wrong context. Search distinctive wording from a quotation, try earlier date ranges, and find the oldest traceable appearance of an image or video. Compare captions across credible reports. Check weather, architecture, language, uniforms, shadows, landmarks, and event schedules only as supporting clues; do not force an identification from one visual detail. Read before and after the quoted passage. For video, locate the longer clip and inspect what happens immediately before and after the viral segment. For numerical claims, identify the denominator and comparison period. A large relative increase from a small baseline may have a different practical meaning from the headline. A national average may not describe a local situation. Also check whether the claim has expired. Policies, software behavior, product specifications, officeholders, and public health guidance change. An accurate page from an earlier date can become misleading when presented as current. Preserve publication and update dates in your notes so that later revisions do not silently change the evidence you evaluated. Pass five: inspect the media without treating tools as oracles For an image, search for earlier or higher resolution copies and compare crops. For a video, examine several clear frames rather than one thumbnail. Listen for abrupt audio changes, but remember that compression, editing, translation, and platform processing can create artifacts. Detection tools may help prioritize attention, but a score from an automated detector is not proof that media is authentic or generated. If Content Credentials are present, inspect what they actually assert: the signer, creation or editing actions, included ingredients, and whether the credential validates. The C2PA explainer is clear that a Content Credential records provenance and can make attached assertions tamper evident; it does not decide whether the depicted event or the publisher's interpretation is true. The absence of credentials is also not proof of deception because adoption is optional and platforms may remove metadata. Keep media analysis connected to the claim. A synthetic profile image does not automatically falsify every statement from the account, while an untouched photograph does not prove its caption. Provenance, content, and context are separate questions that must converge. Pass six: seek independent confirmation and grade the conclusion Look for confirmation that does not trace back to the same post, press release, anonymous account, or syndication feed. Independence matters more than the number of links. For consequential claims, seek sources with different access paths: an official record, direct reporting, a subject matter expert who can explain the evidence, or a transparent fact check that links its work. End with a graded conclusion rather than a binary reflex. Useful outcomes include supported, contradicted, missing context, unverified, or not currently knowable. Write one sentence explaining why and name the strongest evidence and largest limitation. If material facts conflict, do not average them into certainty. Match your next action to the stakes. A harmless trivia claim may need little effort. A claim that could affect health, money, reputation, safety, or civic participation deserves stronger evidence and possibly qualified advice. Do not amplify a harmful rumor merely to ask whether it is true; searching exact wording or sharing a verification source can avoid giving the original more reach. Keep a compact verification record Save the exact claim, original URL, capture time, key sources, and your conclusion. Note inaccessible evidence and unresolved conflicts. This record makes it possible to revise your judgment when better information appears and helps another reader reproduce the check. The method is successful even when the answer is "not enough evidence." Media literacy is not the performance of doubting everything. It is the discipline of assigning confidence in proportion to traceable evidence, relevant expertise, authentic context, and sound reasoning. Sources Civic Online Reasoning https://cor.inquirygroup.org/ Lateral Reading vs. Vertical Reading https://cor.inquirygroup.org/curriculum/lessons/lateral vs vertical reading/?cuid=teaching lateral reading Evaluate credibility using the RumorGuard 5 Factors https://newslit.org/news and research/rumorguard 5 factors/ C2PA and Content Credentials Explainer https://spec.c2pa.org/specifications/specifications/2.3/explainer/Explainer.html Cover image credit Cover image by Slippery slope302 , made available under CC BY 4.0 . WIKIVISE cropped and converted the source image.