How we talk about road deaths — and why it matters
At Road Safety News, we believe language shapes how the world understands road danger. Words can make deaths seem inevitable — or preventable. They can quietly blame victims — or accurately describe what happened. This guide sets out how we report on road trauma, and we invite journalists, campaigners and communicators everywhere to use it.
1. Crash, not accident
We say: crash, collision, accident-free alternatives such as “was struck by” or “was hit by” We avoid: accident
The word accident implies an event that was unpredictable and unpreventable — an act of fate. Decades of research tell us the opposite: road crashes follow predictable patterns and have identifiable causes, from speeding and impairment to poor road design. Describing them as accidents subtly tells readers that nothing could have been done.
This is not a fringe position. Major news organisations, transport authorities and public health bodies have moved away from “accident” in favour of “crash” or “collision”. We follow that standard in every article we publish.
Exception: We retain “accident” only in direct quotations, official names (e.g. journal titles or agency names that include the word), and legal terms of art.
2. Report agency accurately
Passive constructions and vehicle-centred phrasing can erase the human decisions involved in a crash.
We avoid: “A pedestrian was killed when a car mounted the pavement.” We prefer: “A driver struck and killed a pedestrian after the car left the road.”
Cars do not act on their own (automated driving systems aside — see Section 7). Where the facts are known, name the actor: a driver, a rider, an operator. Where facts are not yet established, say so plainly: “Police have not yet determined the circumstances.”
This is about accuracy, not blame. Attributing agency correctly does not mean pre-judging fault — it means describing events as they occurred, pending investigation.
3. Avoid victim-blaming framing
Coverage of pedestrian and cyclist deaths often carries subtle framing that shifts responsibility onto the person who was killed.
Watch for and avoid:
- Leading with what the victim was wearing (“The cyclist, who was not wearing high-visibility clothing…”) unless it is established as legally or causally relevant and attributed to an official finding.
- “Darted out”, “stepped into the path of”, “came from nowhere” — phrases that assume the victim’s movement caused the crash before any investigation has concluded.
- Mentioning helmet use in a crash where the injuries were not survivable regardless, or where helmet use is not legally required, without context.
- Counting only “innocent victims” — all road deaths are counted and reported equally.
Our test: Would this detail appear if the roles were reversed? If a detail about the victim would not be reported about the driver, question why it is being included.
4. People first
We describe people, not categories.
We prefer: “a man walking”, “a woman cycling”, “a child crossing the street”, “people who use wheelchairs” We use carefully: “pedestrian”, “cyclist” — acceptable in headlines and data reporting, but in narrative reporting we humanise where possible.
We avoid language that reduces people to obstacles (“a vehicle collided with a pedestrian”) and we never describe deaths in terms that trivialise them (“wiped out”, “mowed down” in casual usage — though “struck” and “hit” are accurate and acceptable).
When reporting on victims, we follow standard ethical practice: we name people only when names are officially released or provided by families, we respect family wishes, and we avoid gratuitous detail about injuries.
5. Numbers with meaning
Road safety is a data-driven field, and we hold our data reporting to a high standard.
- Always give the rate, not just the count. “China has more road deaths than Norway” is meaningless without population context. We use deaths per 100,000 population as the standard comparator, and deaths per billion vehicle-kilometres where exposure data allow.
- Name the source and year. WHO, IRTAD/ITF, national statistics offices. Figures without attribution do not run.
- Flag definitional differences. Countries define a “road death” differently (death at scene vs. within 30 days). When comparing, we note whether figures are adjusted.
- Trend over snapshot. A single bad year is weather; a five-year trend is climate. We contextualise annual figures against trends wherever possible.
- “Fell by” vs. “is falling”. We do not present one year of decline as a trend, and we do not present a plateau as progress.
6. Words we choose
| Instead of | We use | Why |
|---|---|---|
| Accident | Crash, collision | Crashes are predictable and preventable |
| Road toll | Road deaths | “Toll” frames deaths as an acceptable price |
| Car accident victim | Person killed in a crash | Person-first, accurate |
| Jaywalking | Crossing outside a crosswalk (with context) | “Jaywalking” is a loaded term with a contested history; laws vary widely by country |
| Speed “camera trap” | Speed camera / safety camera | “Trap” implies enforcement is a trick |
| Drunk driver “lost control” | Driver crashed after drinking | “Lost control” obscures the cause |
| Fatal accident hotspot | High-risk location / black spot | Precision; “black spot” is the accepted engineering term |
| Died after a collision with a lorry | Was struck and killed by a lorry driver / a lorry | Restores agency where facts support it |
7. Automated vehicles
As driving automation grows, precision matters more, not less.
- We use the SAE levels (0–5) where relevant and avoid the marketing term “self-driving” for systems that require driver supervision (Levels 1–2).
- “Driver-assistance system” describes Levels 1–2; “automated driving system” describes Levels 3+.
- In crashes involving automation, we report what is known about the system’s engagement status and do not assume either the technology or the human was at fault before investigations conclude.
8. Suicide, trauma and sensitive reporting
- We do not speculate that a crash was intentional self-harm. Where authorities confirm it, we follow established safe-reporting guidelines for suicide: no method detail beyond what is essential, no sensational framing, and signposting to support services.
- We do not publish graphic crash imagery. Photographs of wreckage may be used with editorial justification; images showing victims are not used.
- We take particular care in reporting deaths of children, and we do not doorstep bereaved families.
9. Campaign language and neutrality
We report on advocacy; we are not a campaign. That said, our editorial position — grounded in evidence — is that road deaths are preventable and that the Safe System approach represents the international consensus on how to prevent them. We state this openly rather than performing false balance between evidence and its absence.
- We do not give equal weight to positions that contradict established crash science (for example, claims that seat belts or motorcycle helmets are ineffective).
- We do distinguish between evidence questions (settled by data) and policy questions (legitimately contested — for example, how to balance enforcement and engineering investment), and we cover policy debates fairly from all sides.
10. Quick checklist for every article
Before publishing, every Road Safety News article passes this test:
Would the language pass unchanged if the roles of those involved were swapped?
Have we written crash or collision, not accident?
Where facts are known, is the actor named (driver, rider) rather than only the vehicle?
Is any detail about the victim relevant, attributed and free of blame-shifting?
Are all figures sourced, dated and given as rates where comparison is implied?
Have we avoided speculation about cause, intent or fault ahead of official findings?
