Statistics can make a news report sound conclusive even when the underlying evidence is limited, preliminary, or easy to misread. This guide gives you a repeatable way to check data-based claims: trace the original source, identify what was measured, test the denominator and time period, inspect the methodology, and record what remains uncertain before you cite or share the story.
Overview
A statistic in a news article is not the same thing as a fact detached from context. It may be an official count, an estimate, a survey result, a projection, a rate, a percentage, or a preliminary figure. Each type answers a different question and carries different limitations.
The fastest reliable check is to move from the article to the underlying source. Look for a linked report, data table, methodology note, survey questionnaire, government release, court record, academic paper, or organization that collected the data. A secondary article can help explain the finding, but it should not be your only evidence when the claim matters.
Before accepting a number, ask five basic questions:
- What is being counted or measured?
- Who or what is included?
- What time period does the number cover?
- What is the denominator or comparison point?
- How was the number produced, and could it change?
This process is useful for breaking news, local government coverage, public safety reports, election reporting, economic updates, health stories, and trending claims on social media. It also helps distinguish a legitimate disagreement between sources from a misleading comparison.
Know what kind of number you are reading
| Type of figure | What it means | What to check |
|---|---|---|
| Official count | A recorded total produced by an organization or public body. | Coverage, reporting delays, duplicate records, and whether the count is final. |
| Estimate | An informed approximation based on incomplete data or a statistical method. | Model inputs, assumptions, uncertainty, and whether the estimate may be revised. |
| Survey result | An answer from a sample used to describe a larger population. | Sample size, selection method, wording, response rate, weighting, and margin of error. |
| Projection | A forward-looking result based on stated assumptions or scenarios. | Forecast period, assumptions, alternative scenarios, and what would cause the result to change. |
| Rate | A count expressed relative to a population, exposure, or opportunity. | The denominator, unit of measurement, geographic area, and time period. |
| Percentage | A proportion expressed out of 100. | What the percentage is a percentage of, rounding, and the underlying totals. |
| Preliminary figure | An early result released before all records or checks are complete. | Revision policy, release date, missing reports, and the expected finalization process. |
How to estimate
When a news report uses a rate, percentage, or comparison, recreate the basic calculation if the source provides enough information. This does not require advanced mathematics. It requires matching the numerator and denominator.
- Copy the exact claim. Record the wording, number, unit, date range, location, and comparison. “Up by 20%” is incomplete unless you know what increased and compared with what.
- Find the original release or table. Follow links in the article, then search for the title, publication date, or distinctive phrase if the link is missing. Prefer the data producer’s own documentation over an unsourced graphic or repost.
- Identify the numerator. This is the event, people, dollars, cases, responses, or other quantity being discussed. Check whether it represents unique records or repeated events.
- Identify the denominator. A percentage may refer to all residents, respondents, households, eligible voters, reported incidents, or another group. Changing the denominator can change the meaning without changing the numerator.
- Match the time periods. Compare like with like: the same months, season, year, reporting cycle, or number of days. A partial period should not be casually compared with a complete one.
- Recalculate the change. For a percentage change, use: (new value − old value) ÷ old value × 100. For a rate, use: events ÷ population or exposure × stated multiplier. Keep the source’s multiplier visible.
- Read the methodology. Look for collection methods, exclusions, definitions, weighting, imputation, quality checks, and revision notes.
- Compare cautiously. If another source gives a different number, check whether it uses a different definition, geography, date, or stage of data collection before calling one source wrong.
A large percentage change can come from a small underlying number, while a small percentage change can represent a substantial number of people. Whenever possible, report both the percentage and the underlying totals.
Inputs and assumptions
Use this worksheet for a news statistic you may cite, publish, or share. Save a copy with the article or draft so another reader can repeat the check.
| Claim as written | Copy the exact sentence, headline, post, or chart label. |
|---|---|
| Original source | Record the publisher, document or table title, URL, and publication date. |
| Date range | Note the start and end dates, release date, and whether the figure is preliminary. |
| Population or coverage | Specify the people, places, records, or cases included and excluded. |
| Numerator | Write the count or measured quantity being discussed. |
| Denominator | Write the total against which the numerator is compared. |
| Methodology | Describe whether the figure is a count, sample, model, projection, administrative record, or other method. |
| Margin of error or uncertainty | Record the stated margin of error, confidence interval, range, or warning about uncertainty. If none is provided, write “not stated.” |
| Definitions | Copy important terms as the source defines them; do not assume everyday meanings apply. |
| Missing context | List relevant comparisons, exclusions, revisions, seasonal effects, or limitations not made clear in the article. |
For surveys, check how participants were selected, how many completed the survey, whether responses were weighted, and the exact question wording. A margin of error describes sampling uncertainty under stated conditions; it does not automatically account for every possible source of error, such as nonresponse or unclear wording.
For estimates and projections, separate observed data from assumptions. An estimate may describe the present or past using incomplete information. A projection describes a possible future outcome. Neither should be presented as a confirmed count.
Worked examples
Example 1: Checking a percentage increase
Suppose a report says incidents “rose by 50%.” The source table lists 20 incidents in one period and 30 in the next. The calculation is (30 − 20) ÷ 20 × 100 = 50%. The claim is arithmetically consistent, but the review is not finished. You still need to check whether both periods cover the same number of days, whether the definition of an incident changed, and whether the later figure is preliminary.
Example 2: Checking a survey claim
A headline says that 60% of residents support a proposal. The worksheet should ask: 60% of which residents? If the source surveyed a sample, the claim may more accurately mean 60% of respondents, with an estimate for the broader population. Check the sample design, question wording, field dates, response count, weighting, and stated uncertainty before treating the result as a precise measure of every resident’s view.
Example 3: Reconciling two counts
One report lists 100 cases while another lists 85. Do not assume the difference proves an error. One source may count reports received, while the other counts verified cases. They may use different dates, locations, or update schedules. Write down each source’s definition and coverage. The honest conclusion may be that the figures are not directly comparable.
When to recalculate
Revisit a statistic whenever the underlying data or explanation changes. This is especially important when a story is developing quickly and early figures are likely to be updated.
- A source posts a revision: Replace the old value, preserve the original for comparison, and note the revision date and reason if provided.
- The reporting period closes: Recheck a preliminary total after late records, audits, or verification may have been incorporated.
- A definition changes: Avoid combining figures from before and after the change unless the source supplies a comparable series.
- A new survey or release appears: Compare field dates and methodology, not only the headline numbers.
- The claim starts trending: Trace it back to its earliest accessible source and inspect whether a chart, caption, or post removed important context.
- You are republishing or citing it later: Reopen the source link, confirm that the table and definitions remain current, and add an “as of” date.
For an editorial workflow, set a reminder to review important data links and definitions at least twice a year, and sooner when a major source changes its methodology or reporting standard. Keep the completed worksheet with the published item. If uncertainty cannot be resolved, say so plainly: identify the source, explain the limitation, and avoid stronger wording than the evidence supports.
Before sharing a data-based news claim, pause long enough to answer: What exactly is measured? Compared with what? During which dates? Using whose method? If you can show the source, calculation, assumptions, and remaining uncertainty, your statistic is much more likely to inform readers rather than mislead them.
