Using transport data in a story: check who and what the numbers represent

A transport dataset begins with a purpose. It may record bookings, payments, vehicle movements, inspections or a survey of travel behaviour. Those records can be valuable for other questions, but they do not automatically describe every journey or every person in a city. The first task is to understand how the data came into existence.

Read the collection process

Ask who collected the information, why it was needed and what triggers an entry. A booking record may describe a request that was later cancelled. A vehicle location may show where a device reported from without explaining whether a passenger was travelling. A survey answer describes what a participant reported under a particular question.

Keep the original definitions with the data. Canada’s official guidance on data quality connects interpretability with understanding why and how data was collected. Its accuracy guidance also calls for information on methods, coverage and assumptions. These principles help a reporter decide whether a dataset fits the claim being considered.

Look for people and places outside the frame

A service’s own records usually describe interaction with that service. They may leave out people who could not book, did not have a suitable payment method or used another mode. A neighbourhood with few recorded trips might have low demand, limited availability, a difficult pickup point or several of these conditions together.

Sparrow’s Public Transport page considers connections around the journey. That perspective suggests useful questions for data: does the record show the walk to a pickup, the wait, a missed connection or the return trip? Where these stages are absent, avoid treating the recorded vehicle segment as a complete account of access.

Check whether two figures are comparable

A comparison needs more than matching column names. Check geography, time zones, reporting windows, vehicle categories and treatment of cancelled or incomplete work. A series can change because the service changed, because reporting improved, or because the definition changed. Record the point at which a new method was introduced.

Inspect missing values and duplicates before drawing a trend. Ask whether identifiers remain stable when a vehicle, account or operator changes. Separate a corrected historical record from a newly delivered trip. Where a dataset has several versions, use a named version and publication date so another reader can locate the same evidence.

Explain limits in the story itself

Put the most important limitation beside the conclusion. If the data covers one platform or a small area, describe that coverage in ordinary language. If it cannot answer why a change happened, make that uncertainty visible and identify the additional evidence needed. Interviews and field observation can help interpret a pattern, although they also have their own limits.

The Press & Media background collection supports questions about mobility operations and evidence. A strong data request names the population, period and decision the information needs to inform. That makes it easier to obtain a useful answer and harder to turn a partial record into an unsupported general claim.