Methodology & Data Sources

Primary Data Source

All data on PlainRoadSafety comes from the NHTSA Fatality Analysis Reporting System (FARS), a nationwide census of fatal motor vehicle traffic crashes maintained by the National Highway Traffic Safety Administration, part of the U.S. Department of Transportation. NHTSA collects FARS data from all 50 states, the District of Columbia, and Puerto Rico. FARS is a complete census, not a sample, of all crashes involving at least one fatality on public roadways in the US. Because it is a census rather than a sample, there is no sampling error in FARS data.

County population and per-capita rates. County road-death rates per 100,000 residents pair 2023 NHTSA FARS deaths with U.S. Census Bureau Population Estimates Program data (Vintage 2023). To keep per-capita rates statistically stable, they are shown only for counties with at least 10,000 residents and a 2023 FARS record; other counties are presented as raw counts.

Corpus placement (where each entity sits)

State detail pages rank every state by fatality rate per 100,000 residents (highest first), the same ORDER BY as deadliest states. County detail pages rank rated counties (those with a published per-capita rate) by the same rate metric, highest first, and also show the within-state ordinal among rated peers. Ties break on slug for a stable listing.

The according-to lead on each detail page restates that ordinal so a reader sees where the entity sits before the raw record fields. Ranks are recomputed live from the database.

Data Coverage

• Time period: 2015–2023 (9 years of continuous fatality data)
• Geographic scope: 51 states (including DC) and 3,137 counties across the United States
• Metrics: Total fatalities, fatality rates per 100K population and per 100M vehicle miles traveled (VMT), contributing cause breakdowns, annual trends, and county-level geographic analysis

Processing Pipeline

We download annual FARS data files from NHTSA and process them as follows:

• Fatality counts are aggregated by state and county across the 9-year period
• Annual totals enable year-over-year trend calculation and trend direction classification (improving/stable/worsening)
• Cause categories are derived from FARS flag variables: alcohol impairment (DRUNK_DR), speeding (SP_CRASH), pedestrian (PBTYPE), cyclist, unrestrained occupants (RESTUSE), distracted driving (DISTRACT), weather-related (WEATHERI), nighttime (LGT_COND)
• County population-normalized rates pair 2023 FARS fatalities with Census PEP Vintage 2023 population estimates
• VMT rates use FHWA Annual Highway Statistics for vehicle miles traveled by state

Important Notes on Cause Categories

Cause categories are not mutually exclusive. A single fatal crash may involve multiple contributing factors, for example, a crash where the driver was both alcohol-impaired and speeding counts in both categories. This means the sum of cause category percentages will exceed 100%. This reflects NHTSA's methodology, which records all relevant factors, not a single primary cause.

Fatality Rate Calculation

Per 100,000 population: 2023 fatalities ÷ 2023 population estimate × 100,000. Allows comparison between states and counties of different sizes.

Per 100 million VMT: Annual fatalities ÷ (annual VMT ÷ 100,000,000). Accounts for differences in driving volume. States with more total miles driven are expected to have more total fatalities; VMT rates normalize for this.

Data Collection Method

FARS data is collected by state-level FARS analysts (typically housed within state transportation departments or highway safety offices) who review police crash reports, death certificates, state vehicle registration files, coroner reports, hospital records, and emergency medical service reports for every fatal motor vehicle crash in their jurisdiction. This multi-source approach ensures comprehensive and consistent data quality across states. Each crash, vehicle, and person involved is coded into the FARS dataset with standardized variables, enabling national comparisons. Because FARS is a census (not a sample), it captures every qualifying fatal crash, there is no sampling error.

Limitations

• FARS records fatalities occurring within 30 days of a crash due to crash injuries, not at-scene deaths only.
• The most recent FARS year may lag the current calendar year by 18-24 months due to NHTSA data compilation timelines.
• County-level VMT data is not available from FHWA; county rates use population normalization only.
• Rural/urban classification follows NHTSA's definitions, which may differ from Census urban area boundaries.
• Non-fatal injuries are not captured in FARS. The dataset only includes crashes involving at least one fatality.
• Contributing factor flags (alcohol, speeding, distraction) are based on investigating officer determinations, which may vary in accuracy across jurisdictions.

Not Affiliated

PlainRoadSafety is not affiliated with NHTSA, FHWA, or any government agency. This site is for informational and educational purposes only.