A comprehensive joint econometric model of motor vehicle crashes arising from multiple sources of risk
DOI: 10.1016/j.amar.2018.03.002
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the limitations of traditional single-equation regression models in motor vehicle crash analysis, which assume that crashes arise from a single, linear additive source of risk. The authors argue that this approach fails to capture the unobserved heterogeneity inherent in crash data, as crashes at a specific location often stem from multiple simultaneous and interdependent sources, such as driver behavior, roadway design, and spatial environmental factors. To resolve this, the study proposes a novel theoretical framework and methodological approach that decomposes total crash counts into distinct constituent risk sources. The researchers developed a joint econometric model incorporating random parameters and instrumental variables to empirically test this hypothesis. The model assumes that total crash counts are generated by three specific latent risk sources: engineering factors (roadway geometry and operations), unobserved spatial factors (environmental and locational attributes), and driver behavioral factors. To handle endogeneity between these sources—such as the relationship between roadway design and speeding—the model employs a simultaneous equation approach with instrumental variables. The methodology was applied to a dataset of 521 road segments on state-controlled roads in Queensland, Australia, covering four years of crash data (2010–2013). The model estimates the proportion of crashes attributable to each risk source, allowing for the possibility that some sources contribute zero crashes at specific sites. The results demonstrate that the multiple risk source model significantly outperforms the traditional single risk source model in terms of prediction ability and goodness-of-fit measures, specifically mean absolute deviance and mean squared predictive error. Unlike traditional models that only predict total crash counts, the proposed model successfully predicts the proportion of crashes contributed by each specific risk source. This improved fit suggests that the complexity of crash occurrence is better explained by multiple equation linear predictors rather than a single additive function. The significance of this work lies in its challenge to the dominant "Swiss cheese" model of crash causation, offering a more nuanced understanding of how different risk factors interact to produce crashes. By decomposing crash counts into engineering, spatial, and behavioral components, the model provides a more accurate representation of the data-generating process. This approach allows for better identification of specific risk contributors at individual sites, potentially leading to more targeted safety interventions. The study concludes that while further research is needed to test repeatability across other datasets and explore additional risk sources, the multiple risk source framework offers a superior theoretical and empirical tool for analyzing motor vehicle crash causation.
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 | OpenAlex-citations | — | — | 1 | 2026-06-19 |
| archive | success | unpaywall | — | — | 2 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| enrich | success | semantic_scholar | — | — | 2 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| 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.
- induced exposure
- incidence prevalence
- telematics crash prediction
- pre crash contributing factors
- causation analyses
- motorcycle crash typology
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
- Theoretical Contribution: computational model, theory or model