Driver crash risk factors and prevalence evaluation using naturalistic driving data
archive: archived pipeline: cataloged
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the need for accurate evaluation of driver crash risk factors by utilizing naturalistic driving (ND) data, which captures real-world driver behavior and consequences. Previous studies often relied on surrogate measures like near-crashes or police reports, which may lack statistical power or accuracy. This study leverages the Second Strategic Highway Research Program Naturalistic Driving Study (SHRP 2 NDS), a large-scale dataset comprising over 35 million miles of continuous driving data from more than 3,500 participants across six U.S. sites. The primary objective is to perform the first direct analysis of causal factors using only 905 injurious and property-damage crash events, allowing for a definitive assessment of risk without relying on non-crash surrogates. The methodology employed a case-cohort approach to evaluate time-variant risk factors. Crash exposure was extracted from short video windows (20 seconds for impairment/error, 6 seconds for distraction) preceding the crash onset. These were contrasted with 19,732 control driving segments selected via two-staged stratified random sampling to represent normal driving conditions. A mixed-effect random logistic model was used to estimate odds ratios (ORs) and prevalence, controlling for driver-specific correlations. The analysis categorized risk factors into observable impairment (drug/alcohol, fatigue, emotion), driver performance error (e.g., right-of-way violations), momentary judgment error (e.g., speeding, aggressive driving), and observable distraction (e.g., handheld device use). Key findings indicate that driver-related factors were present in 87.7% of crashes, with 73.7% involving some type of error and 68.3% involving distraction. Driver performance error carried the highest overall risk, increasing crash likelihood by 18.2 times compared to model driving. Specific high-risk subcategories included right-of-way errors (OR 936.1) and sudden/improper braking (OR 247.8), though these were rare. Observable impairment, particularly drug/alcohol influence, increased crash risk by 35.9 times. Distraction was highly prevalent, occurring in 51.93% of normal driving segments, and doubled overall crash risk (OR 2.0). Handheld electronic devices posed significant danger; dialing on a handheld cell phone increased risk by 12.2 times, while texting increased it by 6.1 times. The study estimates that 36% of U.S. crashes could be avoided if distraction were eliminated. The significance of this research lies in its provision of direct, evidence-based risk estimates for transportation policy, vehicle design, and driver education. By demonstrating that driver error and distraction are the dominant causes of crashes, the paper shifts focus from vehicle or roadway factors to human behavior. The high prevalence of distraction and the specific high-risk nature of handheld device use underscore the critical need for interventions targeting in-vehicle technology and driver attention. This large-scale, crash-only analysis provides a robust foundation for understanding modern crash causation, moving beyond previous estimates that relied on less precise data sources.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | — | — | — | 1 | 2026-05-28 |
| archive | success | manual_pmc_pow_fetch | — | — | 39 | 2026-08-22 |
| extract | success | cached | — | — | 4 | 2026-08-23 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | success | crossref | — | — | 2 | 2026-06-04 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.8-27b-gittensor | summ-v5 | 2 | 2026-08-23 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
Summary generated by qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- naturalistic crash near crash
- mobile phones
- telematics crash prediction
- manual
- temporal
- pre crash contributing factors
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: observational prevalence, crash risk outcomes
- Methodological Resource: dataset resource