Lighting conditions and injury severity in large truck lane departure crashes: evidence from a partially constrained modeling approach

Objectives Lighting conditions are a critical factor influencing crash risk and injury outcomes for large truck drivers, particularly in lane departure crashes. This study examines how reported lighting conditions influence driver injury severity in single-vehicle lane departure crashes involving large trucks. Specifically, it evaluates whether the effects of key contributing factors remain stable across daylight and dark conditions or vary systematically.Methods Police-reported crash data from Washington State (2018–2022) were analyzed. A mixed logit model with heterogeneity in means and variances was estimated to account for unobserved heterogeneity. A partially constrained modeling approach was applied to evaluate whether the effects of contributing factors differ between daylight and dark conditions. Likelihood ratio tests were conducted to assess parameter transferability across lighting environments. An out-of-sample prediction analysis was also performed to examine aggregate predictive implications.Results Likelihood ratio tests indicated statistically significant differences in injury severity determinants across lighting conditions. Only three variables such as driver age (56–65 years), overturn, and guardrail face impacts, exhibited consistent effects across both conditions. Several factors, including seatbelt use, speeding, fatigue-related driving, and alcohol impairment, showed varying effects. Speeding was associated with a higher likelihood of minor injury during daylight but with an increased likelihood of severe injury under dark conditions. Out-of-sample prediction indicated only small differences in predicted injury-severity probabilities when daylight-condition parameters were applied to dark-condition crashes, with Krinsky–Robb uncertainty intervals including zero for all outcomes.Conclusions The results indicate that reported lighting conditions are associated with different factor–severity relationships in large-truck lane-departure crashes and that selected parameter estimates are not fully transferable across lighting environments. Safety strategies for large-truck lane departure crashes should consider lighting-condition-specific associations while recognizing the observational nature of the crash data.