Harnessing ambient sensing & naturalistic driving systems to understand links between driving volatility and crash propensity in school zones – A generalized hierarchical mixed logit framework
DOI: 10.1016/j.trc.2020.01.028
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study investigates the relationship between microscopic driving volatility and crash propensity in school zones, addressing a critical gap in transportation safety research. While previous studies relied on police reports or surveys, this work utilizes high-resolution naturalistic driving data to examine how instantaneous driving decisions—specifically variations in acceleration, deceleration, and vehicular jerk—precede safety-critical events. The research aims to quantify "event-based volatility" to determine if erratic driving behaviors serve as leading indicators for crashes and near-crashes in vulnerable areas with high pedestrian activity. The methodology leverages data from the Second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study, analyzing over 41,000 driving events (including crashes, near-crashes, and baseline normal driving) involving approximately 3,400 participants. The dataset comprises over 9.4 million temporal samples of vehicle kinematics. To isolate "intentional volatility" from reactive maneuvers, the authors applied a dynamic data censoring scheme that excludes driver reactions immediately preceding a crash. Eight volatility measures were derived using the coefficient of variation for longitudinal and lateral acceleration and vehicular jerk. These metrics were linked to event-specific characteristics, driver history, and health factors using a generalized hierarchical mixed logit framework. This statistical approach simultaneously accounts for both random heterogeneity (individual differences in parameter estimates) and scale heterogeneity (variations in the error term’s dispersion), offering a more flexible model than traditional multinomial or random parameter logit models. The results indicate that drivers exhibited significantly greater intentional volatility prior to safety-critical events in both school and non-school zones. Specifically, increased volatility in positive and negative vehicular jerk in both longitudinal and lateral directions raised the probability of unsafe outcomes. A one-unit increase in intentional volatility associated with positive longitudinal jerk increased crash probability by 0.0528 units, while the effect of negative longitudinal jerk (braking) was nearly double that magnitude. Methodologically, the Hierarchical Generalized Mixed Logit model provided the best fit, revealing that substantial heterogeneity persists due to pure scale effects even after accounting for random heterogeneity. This underscores the importance of distinguishing between these two sources of unobserved variation in safety modeling. The study demonstrates the value of big data analytics and observational designs in understanding extreme driving behaviors at vulnerable locations. By identifying specific volatility metrics that predict crash propensity, the findings support the development of personalized, proactive behavioral countermeasures for school zones. The results suggest that monitoring real-time driving volatility could enable early warnings and alerts, potentially reducing unsafe outcomes by addressing erratic driving patterns before they result in collisions.
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 | Crossref | — | — | 1 | 2026-06-19 |
| archive | success | unpaywall | — | — | 2 | 2026-06-26 |
| extract | success | pdftotext | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-26 |
| chunk | success | chunk | — | — | 1 | 2026-06-26 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-26 |
| enrich | success | semantic_scholar | — | — | 4 | 2026-06-26 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-26 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- naturalistic crash near crash
- sex gender
- induced exposure
- incidence prevalence
- telematics crash prediction
- urban rural setting
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: crash risk outcomes
- Methodological Resource: dataset resource
- Theoretical Contribution: computational model