On this page
Fact-check a story by identifying what is quoted, what is attributed, and what is reported without a source. Find the original document for each factual claim. Read it in full to see whether the headline's interpretation matches what the source actually says. Verify numbers against their original source. Know which claims you cannot verify alone.
- A news story contains multiple types of claims, direct quotes, attributed statements, unattributed facts, and analysis, and each type requires different verification.
- Well-sourced stories provide direct links or clear citations; their absence is a red flag that the claim may be harder to verify than it appears.
- A single sentence from a source can be misleading if surrounding context contradicts it; always read the original source in full, not just the excerpt.
- About 84% of misleading posts included errors like jumping to conclusions, using faulty logic, or making claims without evidence.
- Some claims, forecasts, specialized expertise, private communications, cannot be fact-checked alone and require either waiting for events to unfold or seeking expert review.
When you encounter a news story that surprises you or contradicts something you thought you knew, your instinct is often to check whether it's true. But 'checking' usually means reading another news outlet's take on the same story, which is not fact-checking at all, it's just finding agreement. Real fact-checking means finding the original source: the document, statement, or data the story is reporting on, and comparing what that source actually says to what the news outlet claims it means.
This guide teaches the specific steps a research analyst uses when reading primary sources. You will learn to identify which claims in a story are verifiable, locate the original source for each one, and spot the gap between evidence and conclusion. You will also learn which types of claims you cannot verify alone, so you know when to stop and seek expert help.
What you'll be able to do
After working through this guide, you will be able to read a news story and sort its claims into categories: direct quotes you can verify by finding the original speaker, attributed statements you can check against a source's actual words, reported facts you can confirm with independent verification, and analysis or opinion you can evaluate for logical consistency. You will know how to search for the original document or statement rather than relying on the reporter's summary. You will be able to read that source in full and spot where the headline's interpretation goes beyond what the evidence actually supports. You will recognize the difference between a factual error, the story misquoted the source or cited the wrong number, and a judgment call, where the story interpreted the data differently than you would. And you will know which types of claims require expertise or access you do not have, so you can decide whether to seek additional verification or accept the limits of what you can check yourself.
Separate the claim from its source
A news story makes multiple types of claims, and each type requires different verification. Learning to distinguish between them is the first step in fact-checking. A single story might contain a direct quote from a named person, an attributed statement from an unnamed source, a reported fact with no source mentioned at all, and the reporter's own analysis. Each one has a different standard for verification, and mixing them up is a common reason fact-checking fails.
Identify what is quoted directly
A direct quote appears in quotation marks and is attributed to a specific person. It is a claim you can verify by finding the original speaker or the document the quote comes from. If a story says 'The CEO said, "Revenue will double next year,"' you can verify that claim by finding a recording, transcript, or written statement from the CEO containing those exact words. The quote is only as reliable as the source it comes from, but the words themselves are verifiable. If you cannot find the quote in the original source, or if the surrounding context changes its meaning, you have found a genuine error.
Identify what is attributed to a source
An attributed statement is a claim presented as coming from a named or unnamed source, usually introduced with 'according to', 'said', 'reported', or 'claimed'. Examples: 'According to the Department of Labor, unemployment fell last month' or 'A spokesperson for the company said the product is safe.' You can verify an attributed claim by finding that source's original statement. If the attribution is to a named organization, search for an official press release or report from that organization. If the attribution is to an unnamed source ('sources familiar with the matter said'), verification is harder, you would need to find corroborating evidence from other sources or from the person themselves speaking on the record.
Identify what is reported as fact without attribution
Some claims appear as simple statements of fact with no source mentioned at all: 'The building was constructed in 1987' or 'The study involved 500 participants.' These unattributed facts require independent verification. You cannot simply trust the reporter's word. You must find a document, database, or other source that confirms the claim. This is where many readers go wrong: they assume that if a reporter states something as fact, it must be true. In reality, unattributed claims are often the ones most worth checking, because they carry no visible source to verify against.
Identify what is analysis or opinion
Analysis and opinion cannot be 'fact-checked' in the same way claims can. When a reporter writes 'This policy will likely increase inflation' or 'The company's strategy is risky,' those are interpretations, not verifiable facts. You cannot fact-check an opinion by finding a source that contradicts it. Instead, you evaluate analysis by asking whether the evidence the reporter cites actually supports the conclusion, and whether the reasoning is sound. A reporter might cite accurate facts but draw a conclusion those facts do not support. That is not a factual error; it is a logical error. Recognizing the difference is essential to fair fact-checking.
Find the original source for each factual claim
Once you have identified which claims in the story are factual and verifiable, your next step is to locate the original source for each one. This is where most readers stop, because finding a source takes time and effort. But this is also where fact-checking actually happens. A story can sound convincing because it is well-written and confident. The original source is where you discover whether that confidence is justified.
Check whether the story links to its sources
Well-sourced stories often provide direct links to the documents or statements they cite. A link to a government report, a court filing, a press release, or a peer-reviewed study is a signal that the reporter expects you to verify the claim yourself. The absence of links is a red flag. If a story makes a specific factual claim but does not link to the source, ask yourself why. Sometimes the source is too old or obscure to link to, or the outlet's format does not allow links. But often, the absence of a link means the reporter did not want readers checking the source, either because the claim does not hold up under scrutiny or because the reporter was careless. Either way, you should be more skeptical. When you do not see a link, search for the source yourself.
Search for the original document or statement
If the story does not link to its source, search for it by name, date, or organization. If the story cites a government report, search the agency's website. If it quotes a company, search for the company's press releases or investor relations page. If it references a study, search for the study by author name or publication title. Be specific in your search. Instead of searching 'unemployment report', search 'Bureau of Labor Statistics unemployment report January 2024'. Instead of searching 'CEO statement', search the company name and the CEO's name. The more specific you are, the more likely you are to find the actual source rather than news coverage of the source. If you find multiple versions of a document, check the dates and verify you have the most recent one.
Verify the source is what the story claims it is
Finding a document with the right title is not enough. You must confirm it is the actual source the story is reporting on, not a different version or a misidentified document. Check the date. If the story says 'The latest report shows', verify that the document you found is indeed the latest version, not an older one. Check the author or organization. A report titled 'Economic Outlook' from the Federal Reserve is different from a report with the same title from a think tank. Check the scope. A study about Americans' health habits is not the same as a study about Europeans' health habits, even if both are titled 'Health and Lifestyle'. Misidentifying a source is a common mistake, and it can make a false claim sound true.
Check whether the evidence actually supports the headline
This is where most people's fact-checking fails. They find a source, skim it, see that it mentions the topic the story is about, and assume the story is accurate. But accuracy is not the same as support. A source can be accurate and still not support the headline's interpretation. The headline might emphasize one aspect of the data while downplaying another. It might present a correlation as causation. It might cite a finding from a small or outdated study as though it were definitive. Reading the original source in full is the only way to catch these gaps.
Read the original source in full, not just the excerpt
A single sentence from a source can be misleading if the surrounding context contradicts it. A researcher might say 'Our findings suggest a possible link between X and Y' in a study's conclusion, but earlier in the same study acknowledge that the sample size was small or the methodology had limitations. A company might say 'We are optimistic about Q3' in a press release, but later in the same document disclose risks that could prevent that outcome. The news story might cite only the optimistic quote, making the company sound more confident than it actually is. Read the entire source, not just the section the story quoted. Pay attention to caveats, limitations, and contradictions. If the source itself is long, at least read the introduction, the methodology or scope section, and the conclusion. These sections usually contain the most important context.
Check the date and scope of the data
Verify that the data is current and covers the right time period and population. A story might cite a study from 2015 as though it describes the current situation, when in fact the landscape has changed significantly. A story might cite a survey of people in one country as though it applies globally. A story might cite a study of a specific group, say, college-educated women in urban areas, as though it describes the entire population. These are not always errors; sometimes old data is the only data available, and sometimes a limited study is still worth reporting. But you need to know the limitations so you can evaluate the claim fairly. Ask yourself: Is this data recent enough for the story's purpose? Does it apply to the population the story claims it does? Would different data, covering a different time period or population, lead to a different conclusion?
Look for what the source does not say
Absence of a claim in the source is evidence the story added interpretation that the source did not make. If the story says 'The study proves that X causes Y' but the study itself only shows a correlation, the story has gone beyond what the source supports. If the story says 'Experts agree that Z is true' but the source you find only quotes one expert, the story has overstated the consensus. If the story says 'This is the first time this has happened' but the source does not make that claim, the story has added a superlative the source did not include. These are not always major errors, but they are gaps between evidence and interpretation. Spotting them is part of fact-checking.
Verify numbers and statistics
Numbers are the easiest claims to fact-check and the easiest to get wrong. A statistic should be traceable to a specific study, report, or dataset. If a story says 'Experts say', without naming the expert or citing the study, you cannot verify the number. If a story cites a number without saying where it comes from, search for it. If you find the same number in multiple sources, that is a good sign, but trace it back to the original source anyway, because multiple outlets might all be citing the same flawed study or misquoting the same source.
Find where the number comes from
A statistic should be traceable to a specific study, report, or dataset, not just to 'experts say' or 'research shows'. If the story cites a number, look for the source. The story might say 'A new study found that 60% of Americans...' In that case, search for the study by the organization that conducted it, the date it was released, or the topic it covered. If you find the study, verify that the 60% figure appears in it and that the story is describing it accurately. If the story says 'According to the Census Bureau, the median income is $X', search the Census Bureau's website for that specific figure. If you cannot find the number in the source the story cites, you have found an error.
Check the methodology or sample size
A number is only as reliable as the method used to produce it. A survey of 50 people is not the same as a survey of 5,000. A study of laboratory conditions is not the same as a study of real-world conditions. A poll conducted online is not the same as a poll conducted by phone. These differences matter. A story might cite a statistic without mentioning that it comes from a small or unrepresentative sample. You can verify this by reading the original source's methodology section. Look for the sample size, the population surveyed, the margin of error, and any limitations the researchers acknowledge. If the methodology is weak, the number is less reliable, and the story should have been more cautious in presenting it. If the story presents a number from a weak study as though it were definitive, that is a problem worth noting.
Recognize when you've found a genuine error versus a difference in interpretation
As you fact-check, you will encounter two different types of problems. One is a factual error: the story misquoted the source, cited the wrong number, or misidentified a document. The other is a difference in interpretation: the story interpreted the data differently than you would, or drew a conclusion the evidence does not fully support. These are not the same thing. A factual error is a mistake. A difference in interpretation is a judgment call. Distinguishing between them is essential to fair fact-checking. A factual error means the story is simply wrong. A difference in interpretation means the story made a choice about how to present the data, and you might have made a different choice. Both are worth noting, but they carry different weight. If you find a factual error, the story said the study involved 500 people but it actually involved 50, or the story quoted the CEO as saying X but the CEO actually said Y, you have found a genuine problem. If you find a difference in interpretation, the story emphasized one finding from a study while downplaying another, or drew a conclusion you think the data does not fully support, you have found a judgment call. You can disagree with the judgment, but it is not a factual error. Be clear about which one you have found.
Know what you cannot fact-check alone
Some claims cannot be fact-checked by a general reader, either because they require expert knowledge, because they involve private information, or because they concern future events that have not yet occurred. Recognizing these limits is important. It keeps you from wasting time trying to verify something that cannot be verified, and it tells you when to seek expert help or accept that you cannot verify a claim yourself.
Claims about future events or predictions
A forecast cannot be fact-checked until the event occurs. If a story says 'Economists predict that inflation will rise next year', you cannot verify that prediction today. You can only evaluate whether the reasoning is sound and whether the economists citing the prediction have a track record of accuracy. You can check whether the prediction is based on current data or outdated assumptions. You can look for other economists who disagree. But you cannot prove the prediction wrong until next year arrives and you can see what actually happened. If a story presents a prediction as though it were fact, that is a red flag. Be skeptical of headlines that treat forecasts as certainties.
Claims requiring specialized expertise
Some claims require expert knowledge to evaluate. A story about medical efficacy, engineering feasibility, or legal interpretation might be accurate in its facts but wrong in its conclusions, and you might not have the expertise to tell the difference. If a story says 'A new drug is effective for treating X', you can verify that a study was conducted and that the study found a positive result. But you might not be able to evaluate whether the study's methodology was sound, whether the result is clinically meaningful, or whether the drug is safe. In these cases, seek expert review. Find an expert in the field who is not connected to the story and ask them to evaluate the claim. This is not giving up on fact-checking; it is recognizing the limits of what you can verify alone.
Claims about private communications or internal decisions
Stories often include claims about what someone said in a private meeting or thought internally: 'The CEO decided to cut costs because she was worried about competition' or 'In a private email, the official said the policy was flawed.' These claims cannot be verified unless the person confirms them publicly or the communication is made public. If a story attributes a private thought or statement to someone without that person confirming it, you cannot fact-check the claim. You can only evaluate whether the source reporting it is credible and whether the claim is consistent with other public information. Be cautious of stories that rely heavily on anonymous sources or attributed private communications. They might be accurate, but they are harder to verify.
Do this automatically
Manually sorting claims by type, direct quotes, attributed statements, unattributed facts, and opinion, across multiple stories per week requires hours of careful reading and cross-referencing that compounds across newsrooms and fact-checking organizations. Software can automate the initial categorization of these claim types, allowing human reviewers to focus their time on verification rather than classification. IMBeacon.ai runs this kind of analysis end to end and cites every figure back to its original source, across as many subjects as you need, with Mimir attaching a citation to every figure. It will not make the decision for you. It gathers and cites the evidence; the judgment stays yours, and deliberately so. Every plan includes it, including the free one.
Start your first report and run it on something you already know well.
This article is general information and research commentary. It is not professional advice, and it does not account for your particular circumstances.
