基于均值异质性随机参数Logit模型的城市道路事故驾驶员受伤严重程度研究

驾驶员事故严重程度诱因分析对减少伤亡事故具有重要意义，以往研究假定影响变量为固定参数容易导致参数估计及研究推论出现偏差，据此本文基于均值异质性的随机参数Logit模型深入研究城市道路事故驾驶员受伤严重程度。使用2015-2019年发生在贵阳市的道路交通事故数据，综合考虑驾驶员、车辆、道路、环境特征等潜在影响因素，同时利用平均边际效应量化各显著变量对事故严重程度的影响。结果表明，均值异质性随机参数Logit模型具有更好的拟合优度；女性、老年人、酒后驾驶、车辆无安全气囊、能见度低于50 m、夜间无路灯照明等因素显著增加了驾驶员受伤害严重程度；随机参数Logit模型的效用函数中，夜间有路灯照明和事故发生在夜间为随机变量，且夜间有路灯照明与柔性护栏和车辆无安全气囊相关；事故发生在夜间与路侧行道树相关。

Examination of Driver Injury Severity in Urban Crashes:A Random Parameters Logit Model with Heterogeneity in Means Approach

Analyzing the factors that are closely related to the driver injury severity in the accident is of great significance to reduce casualties. However, most existing literatures assume that the significant parameters are fixed, which may lead to biased and inconsistent parameter estimates, even erroneous inferences. This paper examines the driver injury severity in urban crashes based on the random parameters Logit model with heterogeneity in means. Using the road crash data of Guiyang city from 2015 to 2019, this study investigates the potential impact factors in terms of drivers, vehicles, roads, and environment characteristics. The average marginal effect is used to quantify the impact of each significant variable on the severity of the crash. The results show that the random parameters Logit model with heterogeneity in means has superior statistical performance. There are some factors that increase the driver's injury severity, for example, women driver, elderly driver, drunk driving, vehicles without airbags, visibility below 50 meters, and no streetlights at night. In the Logit model, streetlights at night and crashes at night are random variables. The streetlights at night variable is co- related with the flexible guardrail and the absence of airbags. Crashes occurring at night variable is co-related with the road segment with roadside trees.