Start with the claim as you found it
A statistic often arrives without the details that give it meaning. You might see a percentage in a presentation, a chart shared online or an article that says a study “found” something. Before deciding whether the number is trustworthy, save the exact wording of the claim. Note the page or document where you saw it, who published it and any link or citation provided.
Look closely at what the sentence appears to measure. Is it describing people, households, businesses or survey responses? Does it refer to a particular place, age group or time period? Words such as “most,” “average” and “increased” can hide important choices. Write down the claim’s implied question in your own words: “What share of which group did what, and when?” This gives you something specific to verify.
Follow the citation trail, not just the search results
First, check whether the page links to a report, data table or methodology note. If it cites another article instead, open that article and check its source. Keep following the trail until you reach the material that produced the figure, rather than a page that merely repeats it. A search for the exact phrase in quotation marks can help when no link is supplied. Searching a distinctive part of the claim alongside the named organization or report title may also help.
Do not assume that the first page using the number is its original source. News stories, summaries and graphics may all describe the same underlying work. If several pages cite one another in a circle, look for clues such as a report title, chart label, survey name or table number. Those details are often more useful than the rounded figure itself.
Sometimes there is no accessible original. A figure may come from a private dataset, an unpublished presentation or a report that has been removed. In that case, identify the earliest source you can inspect and describe the limit honestly. An unsupported attribution should not become more convincing simply because it has been repeated.
Find the number inside the source
Opening a report is not the same as verifying a statistic. Search within it for the quoted number, the key terms in the claim and nearby table or figure labels. A quoted percentage might be rounded, so check adjacent figures as well. If the report has a separate appendix or downloadable table, the relevant detail may be there rather than in the summary.
Once you locate a likely match, read the heading, notes and surrounding text. Check whether the figure describes all participants or only a subgroup. Look for the time period, geographic scope, sample or data source, and the precise wording of any survey question. Also check whether the quoted number was directly reported or calculated from other figures. These distinctions can explain why a headline sounds broader than the evidence beneath it.
If you cannot reproduce a calculation from the available information, you can still report what the source says. Just avoid presenting your interpretation as though you independently verified the arithmetic.
Compare the original finding with the retelling
Put the claim and the source’s wording side by side. Has a statement about surveyed respondents become a statement about everyone? Has “associated with” become “caused by”? Was a change in percentage points described as a percent change? Did a figure for one year or region lose its label? A number can be copied accurately while the sentence around it changes its meaning.
Pay attention to what counts as the “original source.” For a survey result, it may be the organization’s report together with its methods and question wording. For a calculation, it may be a chart based on a separate dataset. You do not always need to inspect every raw record, but you should know whether you are citing an organization’s interpretation, its published data or someone else’s analysis of that data.
If the retelling overstates the finding, rewrite it more narrowly. For example, replace a sweeping claim about a population with a description of what respondents in the cited survey reported. If the source does not support the claim at all, leave the statistic out.
Keep a short source note for future use
A small research note can save you from repeating the whole search later. Record the statistic as published, the exact location where you found it, the report or table title, the relevant population and time period, and any caveat that changes how it should be read. Save enough information to find the source again, including a link when one is available.
When writing, cite the source you actually checked. If you relied on a summary because the underlying material was unavailable, say so rather than implying you inspected the original data. Tracing a statistic is not about achieving perfect certainty; it is about making a clear distinction between what a source shows, what someone else says it shows and what remains unknown.
