Original analysis built on federal and state crash datasets: who is being hurt, where, when, and what is actually changing. Every report states its source, its method, and its limits.
This is the research arm of the Plainclaim. We analyze public crash, injury, and enforcement data and publish the results with the underlying source, the extraction date, the method, and the known limitations attached. Reports are free to cite with attribution. We do not sell data or publish sponsored findings.
Each report is dated, versioned, and revised when the underlying dataset is updated. Corrections are logged rather than quietly edited.
Fatality and injury trends over time, nationally and by state, with the population and vehicle-miles adjustments that make year-on-year comparison meaningful.
Where crashes concentrate: corridors, intersections, counties, and metro areas, mapped against the road and traffic characteristics that explain the pattern.
How risk shifts across the year, the week, and the hours of the day, including holiday periods and the weeks around daylight saving transitions.
Open recalls affecting large vehicle populations, completion rates, and the safety systems most often implicated in defect investigations.
Pedestrian, cyclist, motorcyclist, and older-driver outcomes, which have moved differently from overall crash trends over the last decade.
Coverage minimums, uninsured motorist rates, and the cost side of crashes as reported by regulators and industry research bodies.
We work from public datasets. Each report names the specific dataset, the release version, and the date it was extracted, because these files are revised after publication.
| Source | What it covers | Known limits |
|---|---|---|
| NHTSA FARS | Fatal crashes on US public roads | Fatalities only; final files are released with a lag |
| NHTSA CRSS | Sample of police-reported crashes | A sample, so estimates carry confidence intervals |
| State DOT crash files | State-level police-reported crashes | Definitions and thresholds differ between states |
| CDC injury data | Injury and mortality by cause | Aggregated; limited crash-specific detail |
| IIHS / HLDI | Vehicle-level safety and loss research | Vehicle and insurance focused rather than population-wide |
| NAIC and state regulators | Coverage, complaints, and market data | Reporting periods and categories vary by state |
These are the errors we most often see repeated in press coverage of crash data, including of our own reports.
A large state will always record more crashes than a small one. Rates per capita, per licensed driver, or per vehicle-mile are the only comparisons that mean anything.
Crashes below a state’s reporting threshold, and many crashes with no injury, never enter the data at all. Under-reporting is uneven across states and crash types.
Annual crash figures move for reasons that have nothing to do with safety, including weather, fuel prices, and how much people drove. Multi-year series are what show direction.
A corridor with more crashes may simply carry more traffic. Exposure has to be controlled for before any causal claim is reasonable.
Early releases are revised, sometimes substantially, as investigations close. We label provisional data and republish when the final file lands.
Agencies change how injuries are classified and what counts as a reportable crash. A step in a chart is often a definition change rather than a real-world shift.
Each state guide carries its own crash figures, reporting thresholds, and agency contacts, drawn from that state’s own files alongside the federal datasets.
Yes. Reports are free to cite and quote with attribution and a link to the original report page. Charts may be reproduced unaltered with the source credit intact. If you need the underlying extract for a specific figure, ask and we will supply it where the license of the source permits.
Public datasets published by federal and state agencies and by non-profit research bodies. Every report names the specific dataset, the release version, and the date we extracted it, because these files are revised after publication.
Whenever the underlying dataset is revised, and on a scheduled review cycle regardless. Each report shows its publication date, its last revision date, and a change log describing what was altered and why.
No. We do not publish sponsored findings, paid studies, or content commissioned to reach a particular conclusion. Our commercial relationships are described on our disclosures page.
Usually because of definitions, time period, or population base. A fatality count and an injury-crash count are different measures, states set different reporting thresholds, and provisional figures differ from final ones. Our method note states exactly which we used.
Often, yes, where the underlying source supports that level of detail. State-level analysis appears on the individual state guides. For narrower requests, contact us and tell us the geography and time period you need.
These are the primary sources our reports draw on. Individual reports cite the specific dataset, version, and extraction date rather than these top-level pages.
Spotted something wrong or out of date? We correct errors and log the change. See our correction policy and editorial standards.
Journalists, researchers, and publishers can request the underlying extract for any published figure, or ask us to comment on crash data. We respond to media requests first.
