Contributing factors to the severity of crash injury and vehicle damage involving Japanese lightweight vehicles (K-car): Considering unobserved heterogeneity.

Introduction: This study investigates the safety performance of lightweight vehicles by analyzing the factors influencing injury severity and vehicle damage in Japan. Method: Likelihood ratio tests were initially employed to validate the temporal stability of the crash data. Subsequently, random parameter bivariate probit models with heterogeneity in means were estimated to account for unobserved heterogeneity and correlations between crash outcomes. Results: An out-of-sample prediction analysis reveals a quantifiable structural safety gap for lightweight vehicles, which is often masked by the risk-compensatory behaviors of older drivers. The model demonstrates a significant negative correlation between the injury severities of the two drivers involved, indicating a “zero-sum” energy transfer effect, while the weak injury-damage correlation suggests a “decoupling” effect attributable to modern passive safety designs. Key findings indicate that injury severity is predominantly exacerbated by environmental factors like adverse weather and slippery surfaces, whereas vehicle damage is driven by geometric incompatibilities with road features like curbs. Conclusions and Practical Applications: As the first study to apply this advanced modeling framework to lightweight vehicle safety, our findings offer evidence-based insights for optimizing emergency medical protocols, structural reinforcement standards, and driver assistance systems.