How to read city news without being misled by the number

Start with the sentence the number can support

A city headline often contains a true number and an interpretation that reaches beyond it. Before sharing it, rewrite the claim with its measured subject, geography, period and unit. If those details will not fit because they are unavailable, the uncertainty belongs in the story. This is our reporting checklist, not a claim that any particular publisher has made an error.

Statistics Canada describes data literacy as the ability to derive meaningful information from data, including reading, analysing, interpreting and communicating it. For a reader, that begins by opening the source table. A chart caption or search excerpt is helpful navigation, but the table’s definitions and notes determine what was actually measured.

The publication date is not the observation date

An article published this week may discuss last year’s conditions. A database checked today may return the same observation it returned last month. Record both the observation period and the retrieval date. A newly published revision to an old period is a change in the evidence, not necessarily a new change happening in the city now.

The World Bank’s World Development Indicators sources-and-methods guidance explains that reporting delays, different methods and incomplete coverage complicate comparisons. An international table is therefore a beginning for investigation, even when its columns look consistent. Check the indicator metadata and whether the countries actually have observations for the same year.

Match the boundary and denominator

OECD functional urban areas extend beyond city boundaries to include places linked through workplace commuting. A municipality, commuting region and country can all be described conversationally as a city’s setting; statistically they are different units. Pairing the numerator from one boundary with the population of another can manufacture a misleading rate.

Use a hypothetical example to see the issue: 1,000 recorded events divided by 100,000 residents produces 1,000 events per 100,000 residents. Divide the same count by a wider population of 200,000 and the rate becomes 500. Nothing about the events changed. The denominator did. Before comparing rates, establish whether the recording practice, population and coverage align.

Percent is not percentage points

In another invented example, a rate moving from 4% to 5% rises by one percentage point. Its relative increase is 25%, because the increase of one is divided by the starting value of four. Both calculations are valid, but they convey different magnitudes. Include the starting and ending values so readers can inspect the arithmetic.

Also distinguish a count from a rate, an average from a median, and an index value from a currency price. A city can gain jobs and still have a changing employment rate when its population changes. Without a denominator and a definition, the headline does not tell you which situation occurred.

Make a useful correction possible

Save the source URL, table identifier, filters, observation period and retrieval date alongside any calculation. State which version was used if the source publishes revisions. If the evidence changes, explain whether the change is a correction, a revised historical estimate or a new observation. A trustworthy city intelligence article gives its reader enough information to reproduce the claim and enough context to understand what it cannot establish.