Geometric Design and Truck Crash Frequency on a Nigerian Federal Highway: Evidence from a Negative Binomial Modelling Approach

Abstract

Truck-involved crashes on Nigerian federal highways remain a persistent safety concern due to the operating characteristics of heavy goods vehicles and the geometric quality of high-speed corridors. This study examines the relationship between highway geometric design and truck crash frequency using the Agbor–Ekpoma Road as a longitudinal case study. A six-year crash dataset obtained from the Federal Road Safety Corps and the Nigerian Police Force was analysed using Poisson and Negative Binomial regression models. Traffic exposure was normalised annually using section length, traffic volume, and truck proportion, while field surveys documented key geometric characteristics. Field measurements revealed systematic deviations from design standards. Shoulder widths ranged from 1.30 m to 2.70 m, below the recommended 3.00 m minimum, while stopping sight distance at critical chainages fell to 38 m, representing less than 21% of the required value for the prevailing design speed. Poisson estimation exhibited substantial over-dispersion (mean 0.66; variance 1.34), supporting the use of a Negative Binomial specification. The Negative Binomial model achieved improved goodness-of-fit, with a scaled deviance-to-degrees-of-freedom ratio of 0.8140. Results indicate that truck crash frequency increases with traffic exposure and heavy vehicle proportion. Lane width was not statistically significant once unobserved heterogeneity was addressed, suggesting behavioural mediation rather than a direct geometric effect. Reduced crash frequency on sharper horizontal curves is consistent with compensatory driver behaviour, while vertical grades and restricted sight distance remain significant risk factors. The findings emphasise geometric consistency and visual guidance as key components of truck safety on high-speed freight corridors.