Abstract: In order to study the factors affecting the injury severity of the elderly in collision at intersections and explore the heterogeneity unobserved in the collision accidents, the injury severity model of the elderly in the intersection collision accident was constructed based on the latent category Logit model. Taking the severity of accident injury as the dependent variable, 12 independent variables were selected from five aspects such as people, vehicle, road, built environment, and road environment. Latent category Logit model was used to construct accident injury severity model, and parameter estimation and pseudo-elastic analysis were carried out. The research results demonstrate that motorcycles, darkness and no light at night, the top / bottom of the mountain, and the speed of 50~60 km/h have heterogeneous effects on the severity of accident injuries. In the built environment, there is no shopping center in the 300 m buffer zone of the intersection, which is negatively correlated with minor injury accidents.

Key words: traffic engineering; traffic safety; injury severity of accident; the elderly; built environment; latent category Logit

摘要： 为研究交叉口老年人碰撞事故伤害严重程度的影响因素，探寻碰撞事故中未观测到的异质性，基于潜类别Logit模型构建了交叉口老年人碰撞事故伤害严重程度模型。以事故伤害严重程度为因变量，从人、车、路、建成环境、道路环境这5个方面选取了12个自变量，运用潜类别Logit模型构建了事故伤害严重程度模型，并进行参数估计和伪弹性分析。研究结果表明：摩托车、夜间黑暗无灯光、山顶/山底、速度50~60 km/h这4个自变量对事故伤害严重程度有异质性影响；建成环境中交叉口300 m缓冲区内不存在购物中心与轻伤事故呈现负相关。

关键词: 交通工程；交通安全；事故伤害严重程度；老年人；建成环境；潜类别Logit