Sustainable Cities and Communities: an Application of Statistical Models to Analyse Public Transport-Involved Crash Severity in Dhaka

Aims: This study investigates the factors influencing the severity of public transportation–involved crashes in Dhaka, Bangladesh, where research on public transport safety remains limited despite growing global concern. Method: Drawing on crash severity modeling theories, the study applies statistical approaches to examine relationships between crash characteristics, environmental conditions, and severity outcomes. A dataset of 498 public transportation–related crashes in Dhaka was analyzed using four approaches: the Multinomial Logit Model, Ordered Logit Model, Ordered Probit Model, and Partial Proportional Odds Model. Model comparisons were performed to identify the best-performing method and the factors significantly affecting severity. Results: The Partial Proportional Odds Model outperformed other models in explaining crash severity. All models consistently revealed that pedestrians are the most vulnerable group. Crashes occurring on straight, flat road segments were more likely to result in severe outcomes compared to curved segments. Road surface conditions and collision timing also significantly influenced severity. Conclusion: These findings provide an evidence base for designing targeted interventions to enhance public transport safety and support Sustainable Development Goal 11 on “Sustainable Cities and Communities,” by promoting safer, more inclusive, and sustainable urban mobility in Dhaka. This study is among the first to model crash severity of public transportation in a developing country, combining methodological insights with policy relevance.