Using the multivariate spatio-temporal Bayesian model to analyze traffic crashes by severity
DOI: 10.1016/j.amar.2018.02.001
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 addresses the challenge of unobserved heterogeneity in traffic crash frequency modeling, specifically regarding spatial, temporal, and crash-type correlations. The authors argue that analyzing multiple crash severities simultaneously requires multivariate spatio-temporal models to account for these complex dependencies, which univariate models often neglect. The research aims to accurately identify the long-term effects of economic and weather factors on crash frequencies in Iowa and to explore spatial and temporal correlations across fatal, major injury, and minor injury crashes. Additionally, the study seeks to improve the identification of high-risk counties for safety funding allocation by comparing crude crash rates with Bayesian posterior expected ranks. The researchers analyzed yearly county-level crash data from Iowa’s 99 counties between 2006 and 2015. They employed a multivariate spatio-temporal Bayesian hierarchical model, implemented in OpenBUGS, which incorporated multivariate spatial (Besag-York-Mollie), multivariate temporal (first-order random walk), and multivariate spatio-temporal interaction structures. Covariates included vehicle miles traveled (VMT), unemployment rate, per capita income, and weather indicators (rainfall, snowfall, and days with minimum temperatures exceeding 32°F). Four model variations were compared using Deviance Information Criteria (DIC) to determine the optimal structure for capturing correlations across injury severities. The results indicated that the fully multivariate spatio-temporal model (S MBYM T MRW1) provided the best fit, significantly outperforming models with univariate spatial or temporal components. Income and weather indicators were found to have no significant effects on crash frequencies when controlling for VMT and unemployment rate. Both spatial and temporal effects were significant and played nearly identical roles across all three crash types. Spatial analysis revealed that counties in north and southwest Iowa tended to have fewer crashes than other regions, while fatal and major injury crashes exhibited significant positive spatial autocorrelation. Temporally, all crash types showed a general descending trend from 2006 to 2015. Crucially, the crash types were significantly positively correlated in space but not in time. The study also found that ranking counties by the posterior expected rank of the predicted crash cost rate provided a more accurate representation of underlying traffic safety status than rankings based on crude crash cost rates, particularly for low-count fatal crashes. The significance of this work lies in demonstrating the superiority of multivariate spatio-temporal Bayesian models over univariate alternatives for analyzing traffic safety data. By accounting for correlations across crash types in both space and time, the model reduces bias and improves the identification of high-risk areas. The findings suggest that policymakers should prioritize safety interventions in central Iowa counties and rely on Bayesian posterior expected ranks rather than raw data for resource allocation, as this method better captures the true risk profile, especially for rare but severe fatal crashes.
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 | 2026-06-20 |
| archive | success | semantic_scholar | — | — | 6 | 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 | 2026-06-20 |
| 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.
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