Use three checks: pin down exactly what was said, trace the statistic to its original source and make sure it measures what the speaker claims, then test the conclusion against context and alternative explanations. A published number can be accurate while the argument built around it is misleading.
1. Capture the exact claim
Record the speaker, date, venue and full wording. Find the original speech, transcript, video, press release or post rather than relying on a clipped repost or someone else’s summary. A phrase taken out of its surrounding remarks can appear to make a different claim.
As an Amazon Associate I earn from qualifying purchases.
FactCheck.org describes checking transcripts, videos, websites, press releases and social-media statements. It says, as a description of its own process, “The burden is on the person or organization making the claim to provide the evidence to support it.” That is a useful standard for evaluating a public assertion, not a universal legal rule. FactCheck.org’s process is U.S.-based.
Separate fact from judgment
Some parts of a statement are verifiable: a date, a quoted figure, or whether a particular event occurred. Other words require a standard that the speaker may not have stated. “Unemployment is high” is incomplete until you know: high compared with what—another period, another country, a target, or the speaker’s preferred level?
#1 Best Overall
“Wages are too low” combines a measurable question about wage levels with a value judgment about what counts as enough. Data can show how wages changed or compare with prices; it cannot, by itself, decide what level is fair or adequate.
2. Trace the number to its original source
Identify the statistic, dataset, report or calculation behind the claim. Then follow citations back to the organization that produced the underlying information, rather than stopping at a news article or a fact-check verdict. For U.S. economic and employment data, FactCheck.org points readers to the Bureau of Economic Analysis (BEA) and Bureau of Labor Statistics (BLS). For UK labor-market claims, the House of Commons Library identifies the Office for National Statistics (ONS) as an example of a reputable source. Use the corresponding official statistical agency or a well-documented primary source for the country involved.
Official statistics are primary evidence of what an agency measured, not automatic proof of a political interpretation. Read the release and its definitions. If a politician supplies no source, look for a reproducible calculation or ask what data and method they used before treating the figure as established.
3. Test whether the statistic supports the wording
The Office for Statistics Regulation (OSR) offers five practical questions for statistics used in election campaigns. Ask whether:
- The figure seems wildly unlikely or too good to be true.
- The conclusion logically follows from the statistic quoted.
- The source is publicly available.
- Context or background has been left out.
- A comparison is like-for-like.
These checks come from UK statistical-regulation guidance; they are useful more broadly as questions, but the relevant data authority depends on the jurisdiction. The OSR’s guidance on navigating election statistics was published on 25 March 2026 for elections in England, Scotland and Wales that May. The campaign date has passed; the checklist remains applicable.
Match the measure and comparison
For each number, check its definition, dates covered, geography, population and unit. Ask whether it is nominal or adjusted for inflation, where that distinction matters. Make sure the speaker compares equivalent periods and the same kind of measure. A monthly figure and an annual average, for example, do not answer the same question.
Consider “unemployment is high.” First identify which unemployment measure and population the speaker means. Then choose an explicit benchmark: a prior period, a comparable country or a stated target. Without that benchmark, the statistic may be correct but the word “high” remains unsupported or dependent on an unstated standard.
Free tools Windows power users keep installed
One-click scans. No signup required.
Check the release date and revisions
Economic estimates can change as more source data arrives or methods are updated. Note the release vintage—the version available when the speaker made the claim—and check the agency’s revision notes before comparing figures from different releases. The BEA’s Information Quality Guidelines describe its publication of sources, methods, assumptions and planned updates, as well as revisions and studies of estimate accuracy over time. This is U.S. guidance; check the equivalent documentation from the responsible agency elsewhere.
Best Value
4. Distinguish a trend from a cause
“The economy improved during this administration” describes timing; “the administration caused the improvement” makes a causal claim. The first can be checked against suitable measures and dates. The second requires evidence that connects a policy or decision to the outcome, not simply that one came before the other.
Economic outcomes usually have multiple contributing factors. For a causal argument, look for evidence supporting the proposed mechanism, the assumptions behind the analysis and plausible alternative explanations. A before-and-after comparison or a correlation may be relevant, but neither alone establishes that a policy caused the change. If credible analyses reach different conclusions, explain what each measures and assumes rather than treating disagreement as proof that every source is unreliable.
5. Corroborate the evidence and state a calibrated conclusion
Compare the original data or document with independent evidence. When possible, link readers directly to the underlying data table, statistical release or legal document, and explain any important limitations. The House of Commons Library, describing Full Fact’s approach, says its fact checks aim to use two sources for a central claim unless only one is relevant, and prefer primary data tables and legal documents. Full Fact’s rationale is: “We’d prefer you not to. Anything we publish is attributed and linked to its source, so you can make your own mind up about it.”
Recommended Free Tools
Your conclusion should say what the evidence supports and why. Depending on the case, a claim may be accurate, inaccurate, unsupported, or misleading because it omits context or draws a conclusion the data do not establish. If sources appear to conflict, compare their definitions and coverage, periods and release vintages, geography and population, methods, and whether each figure is observed, estimated or forecast. Also distinguish a descriptive claim from a comparison or a causal claim: two sources may answer different questions rather than contradict each other.
A quick check before sharing a claim
- Save the full original statement, including who said it, when and where.
- Separate checkable facts from words that express a judgment or leave a benchmark unstated.
- Find the original statistic or document and identify its producer.
- Match the measure, dates, geography, population and comparison to the claim’s wording.
- Read the method and revision notes, and check whether the figure was observed, estimated or forecast.
- Look for independent corroboration and consider whether the evidence shows causation or only timing or association.
- Explain what the evidence establishes, what remains uncertain and why.
For broader advice on evaluating context and source reliability, the European Commission’s guide to finding information you can trust recommends official sources for official data and independent fact-checkers as a starting point. It also warns that context can be stripped away and media altered.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
